Streptococci and Actinomyces induce antibodies which cross react with epithelial antigens in periodontitis

Abstract

Perturbation of epithelial structure is a prominent but poorly understood feature of the immunopathological response to bacterial antigens which characterizes the destructive lesion of periodontitis. Western analysis of sera from 22 patients with periodontitis detected multiple antigens in extracts of epithelial cells whereas sera from 12 periodontally healthy subjects displayed only trace reaction with epithelial antigens. To investigate a possible relationship between the bacterial flora adjacent to diseased sites and the presence of antibodies reactive with epithelium, subgingival plaque samples were taken from deep periodontal pockets and cultured anaerobically. Gram positive bacteria containing antigens cross-reactive with epithelial cells were reproducibly isolated by probing membrane colony-lifts with affinity-isolated (epithelium-specific) antibodies and identified by 16S rDNA sequence homology as streptococci (S. mitis, S. constellatus and two S. intermedius strains) and Actinomyces (A. georgiae, and A. sp. oral clone). Conversely, when serum from patients with periodontitis was absorbed with the captured bacterial species the number of epithelial antigens recognized was specifically reduced. It was concluded that development of cross-reactive antibodies related to these organisms may contribute to perturbation of the epithelial attachment to the tooth and the progression of periodontitis. These autoreactive antibodies could also be a contributing factor in other diseases affecting epithelia.

Introduction

Destructive periodontitis is considered to represent an immunopathological response to a complex microflora that develops adjacent to the affected tissues. A hallmark of the progressing lesion is the failure of the epithelial attachment to the tooth and the migration of epithelial components down the tooth root to form a cleft or pocket creating a favourable environment for an abundant anaerobic microbial flora. We have previously reported perturbation of the structural integrity and functional differentiation of the pocket epithelium in relation to inflammatory changes associated with destructive periodontitis [1].

Serum antibodies reacting with tissue components, including type I collagen and epithelial desmosomal proteins, have been reported to occur with greater frequency and in higher levels in periodontitis patients than in periodontally healthy subjects [2,3]. In this context, contribution of autoimmunity to the immunopathology of destructive periodontitis has been proposed [4,5].

The bacterial aetiology of periodontal disease is complex, with a variety of organisms responsible for the initiation and progression of disease. Although some 500 bacterial species have been detected in the human oral cavity, and over 400 species in the subgingival plaque [6], only a limited number have been implicated as periodontal pathogens. Many of these organisms may also be present in periodontally healthy individuals and can exist in commensal harmony with the host.

For this study, a cell line (CCL-4) derived from the normal oral mucosa of a laboratory rat was selected as a source of epithelial antigens [7]. This cell line is well characterized and has been utilized as a model in investigation of microbial pathogenicity related to gingival inflammation [8]. In the context of the reported findings, use of the CCL-4 cell line overcame the potential for contamination by adherent bacterial antigens in primary oral epithelial cultures. The majority of established human epithelial lines were contaminated by HeLa cells while many human oral carcinoma-derived lines express viral antigens or are insufficiently characterized [9], rendering them unsuitable for this purpose.

Of clinical importance is the failure of the epithelial attachment to the tooth which signals progression of the lesion from stable nondestructive marginal gingivitis. The determinants of the host–parasite interaction leading to altered epithelial behaviour remain unknown. On this basis, the objective of the study was to investigate the relationship between potential autoreactive antibodies, oral epithelial antigens and subgingival plaque bacteria. Findings are reported which indicate that bacterial plaque from diseased sites contains Gram positive organisms with antigens that display cross-reactivity with epithelial components.

Materials and methods

Clinical data and serum samples

Blood samples were obtained with informed consent from 22 adult chronic periodontitis patients (14 males, 8 females, age 35–68 years, mean 48 years). Fourteen patients (7 of whom were tobacco smokers) had generalized and 8 patients had localized chronic periodontitis, with no history of systemic disease or of medication use within the past 6 months and no record of periodontal therapy including subgingival scaling, root planing or relevant surgery, in the previous year. All patients showed a diagnosis of moderate to severe generalized or localized periodontitis with loss of attachment (5–11 mm), formation of periodontal pockets with probing depths of 4–11 mm and resorption of alveolar bone. As a control group, blood was also collected from 12 race, age and gender matched periodontally healthy subjects (7 males, 5 females, 27–64 years of age, average 46 years), all non-smoking and with no use of anti-microbial mouthwashes within the past 6 months; characterized by no clinical signs of gingival inflammation, probing depths of 3 mm or less, no sites with significant attachment loss, no evidence of radiographic bone loss and no history of systemic disease. Sera were separated from clotted blood by centrifugation at 1000 g for 20 min and stored at – 70°C until used.

Epithelial cells

For use in this study, the epithelial cell line (CCL-4) derived from normal oral mucosa in a laboratory rat [7], displayed typical epithelial morphology in culture, such as coherent monolayer sheets and stratification of epithelial cells. The cells were cultured in Dulbecco's modified Eagle's medium (DMEM) and Ham's F-12 nutrient mixture (1 : 1) (Sigma, Sydney, Australia), 15 mm HEPES (Sigma) and supplemented with 10% foetal bovine serum (CSL Limited, Victoria, Australia) and penicillin/streptomycin (100 IU/ml, ICN Biomedicals, Inc., Sydney, Australia) at 37°C with 5% CO₂.

CCL-4 cell antigens

The CCL-4 cells were subcultured every 2–3 days and harvested at a multilayer confluence (10⁷−10⁸ cells/3–4 mg total protein) in 175 cm² flasks (Sarstedt, Inc., Newton, NC, USA). Briefly, cells were placed on ice, washed with ice cold Dulbecco's phosphate buffered saline (DPBS), containing a proteinase inhibitor cocktail (Sigma-P8340) and harvested by gentle scraping. After centrifugation, in preparation for Western blot analysis, cells were resuspended in 2 ml PBS, then boiled in sodium dodecyl sulphate (SDS) sample buffer with 1% 2-mercapthoethanol for 10 min. For dot blot analysis, resuspended cells in 0·1 m Tris-buffered saline (TBS, pH 7·5) containing 3% Triton X-100 (AJAX chemicals, Sydney, Australia), were gently vortexed and placed on ice for one hour. The samples were stored at −20°C for up to 2 months.

Western immunoblot for patient and control subject IgGs against CCL-4 antigens

CCL-4 antigens were separated by standard SDS-PAGE mini gels (5% stacking gel, 12% separating gel). Fractionated CCL-4 cell proteins were transferred onto nitrocellulose membranes (Bio-Rad, Sydney, Australia), and blocked with 3% bovine serum albumin (BSA, Sigma) in TBS overnight. Blotted CCL-4 antigens were incubated with titrations of patient sera (n = 22) and clinically healthy subject sera (n = 12). The secondary antibody employed was alkaline phosphatase (AP) conjugated γ-chain specific goat anti-human IgG (Sigma) diluted 1 : 30 000 in 0·05% Tween 20/TBS. IgG antibody binding antigen was visualized with AP substrate (cat 170–6432, Bio-Rad). Triplicate blots were performed for each experiment.

Affinity-isolation of antibodies from CCL-4 (CCL-4-specific antibodies)

The validity of the affinity-isolation procedure was established in this study. CCL-4 cells were passaged into tissue culture plates (24-well-Sarstedt) and incubated at 37°C, 5% CO₂ for 2 days until the cells were 70–80% confluent. CCL-4 cell layers were fixed with precooled methanol for 3 min and blocked with 1% BSA (Sigma) in PBS overnight at 4°C. Patient sera, plus one blank control and one reference serum were diluted empirically with 0·05% Tween 20/TBS, added into 24-well plates and then incubated for one hour at room temperature. Unbound antibodies were removed by washing 3 times with 0·05% Tween 20/TBS. Bound antibodies were eluted in 0·05 m glycine-HCl buffer (pH 2·3) and rapidly neutralized with 1 m Tris (pH 8·0) and 10% foetal calf serum (FCS). Affinity-isolated (CCL-4-specific) antibodies were prepared from all patient sera and used immediately in further analysis.

Subgingival plaque samples and growth conditions

Sixteen subgingival plaque samples were taken with paper points from advanced periodontal pockets (>6 mm probing depth) of 8-adult periodontitis patients at diagnosis. Sterile paper points were inserted into the base of pockets for 20 s for each of two sites per patient and then placed into microcentrifuge vials containing 100 µl reduced transport fluid (RTF) medium and kept on ice. The plaque samples were transferred to an anaerobic chamber (85% N₂, 5% CO₂, and 10%H₂) for primary culture. Each sample was dispersed by vortexing for 10 s to break up bacterial aggregates. Tenfold serial dilutions (10⁻¹−10⁻³) were prepared in RTF; 100 µl of the dilutions were plated onto CDC anaerobic blood agar containing 5% defibrinated sheep blood (BBL, Baltimore Biological Laboratories, Cockeysville, MD, USA) and cultured at 37°C. Processing was completed within 24 h.

Isolation of bacterial colonies from plaque

Nylon membranes (Boehringer Mannheim, Germany) were used to lift colonies from plated plaque samples cultured anaerobically. Briefly, 2 nylon membranes with lifted bacterial colonies from one plate were prepared at the same time. Membranes were dried and fixed in precooled methanol for 3 min and then blocked in 5% skim milk powder in TBS overnight. Two nylon membranes were probed separately with a selected patient serum and matched affinity-isolated (CCL-4-specific) antibodies for immunodetection. Each plaque sample was tested with at least 5 different patient sera. Colonies reacting with affinity-isolated (CCL-4-specific) antibodies were picked and inoculated into CDC anaerobic broth, supplemented with haemin (5 µg/ml, Sigma), menadione (5 µg/ml, Sigma) and 2% horse serum (Amyl Media Pty Ltd, Victoria, Australia), and CDC blood agar plates for subculture under both anaerobic and microaerophilic conditions. Colony purity was checked routinely by Gram-stain with slides fixed in the anaerobic chamber [10]. Growth curves for target organisms were established under anaerobic conditions. Bacterial numbers were estimated by reference to the standard curve determined by absorbance at OD 600 nm.

Identification of bacteria

Only the positive colonies reacting both with serum IgG and affinity-isolated (CCL-4-specific) antibodies were identified by colonial morphology, Gram staining and conventional biochemical tests (RapID ANA II System, REMEL Inc., Lenexa, KS, USA), and confirmed by 16S rDNA PCR.

16S rDNA PCR provided a reliable and rapid method to distinguish these species. Bacteria were grown in CDC broth under anaerobic conditions. DNA was isolated from cells (10⁸−10⁹) resuspended in 200 µl 10 mm phosphate buffer (pH 6·7) containing 400 µg lysozyme (Boehringer Mannheim), 200 U mutanolysin (Sigma) and 400 µg proteinase K (Qiagen Pty Ltd, Clifton hill, Victoria, Australia). Treated cells were lysed by adding 1% SDS. Following the addition of 200 µg RNase (Sigma), samples were incubated at 37°C for a further 10 min, and purified using QIAamp DNA Mini Kit (Qiagen) according to the Manufacturer's instructions. DNA concentrations were measured at A₂₆₀ and used as a template to generate PCR fragments based on universal primers recognizing 16S rDNA sequences as described previously [11]. PCR products were purified by Wizard PCR preps DNA purification system (Promega Corporation, Annandale, Australia) and detected by electrophoresis of a 5-µl sample for 30 min at 5 V/cm in 2% agarose in Tris-acetate-EDTA (TAE) buffer, stained with ethidium bromide (0·5 µg/ml) and visualized under UV illumination. Purified PCR products (50–100 ng) were sequenced using standard dye-terminator chemistry (Prince Alfred Macromolecular Analysis Centre, University of Sydney, NSW, Australia).

The 16S rDNA gene sequences from each bacterium were searched against the NR nucleic database on WebANGIS (Australia National Genomic Information Service) by FastA program and further compared with Gap program. The accession numbers of the 16S rDNA gene sequences were obtained from the EMBL Data Library as shown in Table 1.

Table 1.

Identification of cross-reactive bacteria isolated from plaque samples

	Analysed 16S rDNA sequences
Species	Strain*/clone	Accession no.	Identity (%)†	Frequencies of detection from 16 plaque samples Anaerobes, n (%)
Streptococcus mitis	NCTC12261	D38482	100·000	2 (12·5)
Streptococcus constellatus	206	AF104677	99·774	3 (18·6)
Streptococcus intermedius	ATCC27335	AF104671	99·111	4 (25·0)
Streptococcus intermedius	488	AF104673	98·670	3 (18·6)
Actinomyces georgiae	DSM 6843	X80413	97·539	1 (6·0)
Actinomyces sp. oral clone	BL008	AF385553	98·658	1 (6·0)

^*NCTC, National Collection of Type Culture, UK; ATCC, American Type Culture Collection, USA; DSM, Deutsch Sammlung von Mikroorganismen und Zellkulturen, Braunschweig, Germany.

^†These strains/clones showed the greatest homology to the isolated species.

Bacterial cell antigens

Each bacterial culture was collected in late exponential phase and washed with ice-cold PBS in the presence of a proteinase inhibitor cocktail (Sigma) by centrifugation prior to protein extraction. Components were extracted by sonication on ice, and then boiled in reduced SDS sample buffer.

Western immunoblot for patient IgGs and affinity-isolated (CCL-4-specific) IgGs against oral bacterial antigens

Antigens from the identified bacterial strains recognized by affinity-isolated (CCL-4-specific) antibodies were separated by 12% polyacrylamide mini gels, respectively, then transferred onto nitrocellulose membranes (Bio-Rad) and blocked with 3% BSA (Sigma) in TBS overnight. The membranes were immunodetected by titrated patient sera (n = 22) and matched affinity-isolated (CCL-4-specific) antibodies, and then detected as described above. Triplicate blots were performed for each experiment.

Sera absorption

Sera from 22 periodontitis patients were absorbed individually to each of the identified bacteria, or a combination of all of the identified bacteria; and other bacteria: Lactobacillus acidophilus ATCC 4356 and Lactobacillus casei ssp. rhamnosus ATCC 7469 as controls. Briefly, bacterial cells were harvested at late exponential phase, washed in PBS by centrifugation, resuspended in 0·05% Tween 20/TBS to approximately 10⁸ bacteria per ml (OD at 600 nm = 1·0) and then aliquoted, each bacterial suspension or pooled bacterial suspension to 22 vials, respectively, for individual patient sera diluted empirically. The suspensions were mixed and shaken overnight at 4°C. The cells were removed by centrifugation at 13 000 g for 5 min, and the supernatants used for immunoblots.

Immunoblots against CCL-4 antigens following absorption with identified bacteria

For Western immunoblot, CCL-4 antigens were separated by 12% polyacrylamide mini gels, then transferred onto nitrocellulose membranes (Bio-Rad) and blocked with 3% BSA (Sigma) in TBS overnight. For dot blot, CCL-4 antigens diluted in TBS to a concentration of 1 µg/spot, were applied to pre-wetted nitrocellulose membrane (0·45 µm, 9 × 12 cm size, Bio-Rad) and filtered by a vacuum microfiltration apparatus (96-well format, Bio-Rad). Blotted antigens were blocked with 1% BSA/TBS. All membranes were immunodetected by patient sera depleted by absorption with the identified bacteria or by control bacteria, and compared with nonabsorbed sera, respectively. Bound antibodies were detected as described above. Triplicate blots were performed for each experiment.

Data analysis

The semiquantification of dot blots was measured using densitometric scanning (Scion Image software, Scion Corporation, Frederick, MD). A value of P < 0·05 considered significant was assessed by Kruskal–Wallis test analysis of variance (GraphPad software, San Diego, CA, USA).

Results

Patient and control subject IgG responses to CCL-4 epithelial components

Western blots displayed distinct immunoreactive characteristics for individual patient sera with CCL-4 antigens (Fig. 1); however, there existed a common recognition of a 30-kD band. Only trace recognition of epithelial antigens in periodontally healthy sera was observed including common recognition of the 30 kD band (data not shown). Profiles for reactivity of individual sera were entirely reproducible over three consecutive experiments using separately prepared antigenic extracts in each case. These data suggested that detectable levels of serum antibodies reactive with epithelial components were present in relation to existing inflammatory periodontal disease.

Fig 1.

Recognition patterns for epithelial antigens. Western immunoblots exhibit distinct immunoreactive characteristics for individual patient sera IgG (n = 22) against CCL-4 antigens. Lane (i.e. patient No.) 1, 7, 8, 13, 14, 16, 17, 20 and 22 showed strong reactive bands ranging from 25 to 90 kD; remaining patients showed only trace to moderate recognition of antigens other than the 30 kD band recognized by all patients. Only weak recognition of the 30 kD band was observed in healthy subjects (data not shown). There were no clear differences between 14 generalized (lane 1–7, 9, 11, 12, 16, 19, 20 and 22) and 8 locialized periodontitis subjects (lane 8, 10, 13–15, 17, 18 and 21), or between smokers (lane 1, 2, 4, 6, 9, 16 and 22) and nonsmokers.

Isolation of bacteria with cross-reactivity to epithelial antigens from subgingival plaque samples

Nylon membranes used to lift colonies from plated plaque samples grown anaerobically were probed with serum from a highly reactive patient (Fig. 1, lane 7; Fig. 2a) and with matched affinity-isolated (CCL-4-specific) antibodies (Fig. 2b). Only a small number of colonies (1–5% of the total) were recognized by affinity-isolated (CCL-4-specific) IgG antibodies. This test was repeated using different patient sera and similar results were noted. Affinity-isolated antibodies could not be prepared from periodontally healthy subject sera, in accordance with the lack of reactivity observed for CCL-4 epithelial antigens.

Fig 2.

Detection of bacteria with cross-reactivity to epithelial antigens in dental plaque. Anti-IgG immunodetection following colony-lift techniques. Colonies were lifted onto 2 sheets of nylon membrane from one plated plaque sample. Probed with a serum from patient No.7. Probed with a matched affinity-isolated (CCL-4-specific) antibody from patient No.7. A small number of colonies were recognized.

Identification of colonies with cross-reactivity for epithelial antigens

All positive colonies which reacted both with patient serum IgG and affinity-isolated (CCL-4-specific) IgG were isolated from the plaque samples, and identity confirmed by 16S rDNA PCR (Table 1). The six bacterial strains were identified and ranged from 6 to 25% frequency of detection in 16 plaque samples (Table 1).

Cross-reactivity between affinity-isolated (CCL-4-specific) antibodies and streptococci/Actinomyces antigens

Heterogeneous profiles for the 6 bacteria were identified by serum IgG from each of the 22 patients (Fig. 3a,c,e,g,i,k), and matched affinity-isolated (CCL-4-specific) IgGs (Fig. 3b,d,f,h,j,l). Figure 3(k) shows a nonspecific ladder-like (polysaccharide) pattern of reactivity to Actinomyces spp. with close homology to oral clone (BL008); however, after affinity-isolation (CCL-4-specific) of antibody, the ladder-like pattern of reactivity was not shown while a specific reactive band at 72 kD was observed (Fig. 3l). This could be attributed to the concentrating effect of affinity isolation. Patterns of reactivity for affinity-isolated (CCL-4-specific) IgGs varied for the different bacteria; however, cross-reacting antigens at 27, 30, 35, 38 and 50 kD among two or more bacterial strains were noted indicating the potential for common antigens. In contrast, sera IgGs from periodontally healthy subjects showed low levels of recognition of antigens from this group of organisms (data not shown).

Fig 3.

Western immunoblots with patient sera against antigens of 6 isolated plaque bacteria. Immunoreactive patterns of total patient serum IgG are represented in panels a, c, e, g, i, and k(n = 22). Immuno-cross reactive patterns of matched affinity-isolated (CCL-4-specific) IgG are represented in panels b, d, f, h, j, and l(n = 22). Reference strains with closest homology to the test organisms are shown in parentheses. (a, b) Streptococcus mitis (NCTC 12261) cell antigens. (c, d) Streptococcus constellatus (206) cell antigens. (e, f) Streptococcus intermedius (ATCC 27335) cell antigens. (g, h) Streptococcus intermedius (488) cell antigens. (i, j) Actinomyces georgiae (DSM 6843) cell antigens. (k, l) Actinomyces sp. oral clone (BL005) cell antigens.

Removal of cross-reactive antibodies by serum absorption to Streptococcus/Actinomyces spp

To confirm the existence of cross-reactivity of serum antibodies between epithelial components and bacterial antigens, sera were absorbed with the four streptococcal and two Actinomyces strains. This partially removed cross-reactivity with epithelial antigens as demonstrated by analysis of semiquantitative dot blots (P < 0·05, Kruskal–Wallis test, Fig. 4a). When the effect of absorption by all six bacteria was assessed by Western blot, antiepithelial reactivity was either absorbed or substantially reduced for all patients (Fig. 4b). The reduction of cross-reactivity for serum IgG was specific for the captured organisms with only minor effect detected following preincubation with other Gram positive bacteria: either Lactobacillus acidophilus ATCC 4356 (Fig. 4c) or Lactobacillus casei ssp. rhamnosus ATCC 7469 (data not shown).

Fig 4.

Demonstration by immunoblot of a removal of epithelial cross-reactive antibodies. Reference strains with closest homology are shown in parentheses. (a) Mean density in dot blot analysis of IgG reactivity with CCL-4 antigens by patient sera (n = 22) non-absorbed (A) or absorbed by each identified bacterium (B-G), respectively. An overall P-value < 0·05 (Kruskal–Wallis test), was consider significant among these groups. (b) Absorption of titrations of patient sera (n = 22) by pooled cross-reactive bacteria was performed by Western analysis. The figure shows the result for maximum serum concentration (1: 20) with the indication of removal or reduction of cross-reactive bands compared to nonabsorbed controls (data not shown). Further dilution of sera resulted in abolition of all cross-reactive bands. (c) Patient sera absorption (n = 22) with Lactobacillus acidophilus ATCC 4356 did not result in detectable removal of reactivity for any patient when compared with nonabsorbed profiles run in parallel (data not shown). This finding was reproduced over three consecutive experiments. The apparent overall reduction of staining intensity compared with the experiments described in Fig. 1 can be explained by the reduction in titre of low levels of polyclonal auto-reactive antibodies following prolonged storage (18 months) of the serum samples over the time course of this study. Identical findings were observed for Lactobacillus casei ssp. rhamnosus ATCC 4356 (data not shown).

Discussion

In this study, recognition of streptococcal and Actinomyces antigens by affinity-isolated (epithelial-specific) antibodies from periodontitis subjects and the capacity of these antigens to conversely absorb serum antibodies reactive with epithelial components were indicative of true immunological cross-reactivity. The almost complete removal of antibodies reactive with epithelium by the summed effect of the consortium of bacteria identified implied that it is unlikely that unrelated or uncultured organisms also contributed to this effect. Specific association with the disease state was demonstrated by the lack of reactivity in sera from periodontally healthy subjects.

The findings suggest that antibodies cross-reacting with epithelial antigens may arise by structural mimicry as a consequence of the host's immune response against subgingival plaque bacteria. In addition, cross-reactive antigens were detected at common apparent molecular weights of 27, 30, 35, 38 and 50 kD, that were shared between two or more bacterial strains within the consortium and possibly in related organisms not identified in this study.

All patients had antibodies that recognized a 30-kD epithelial antigen. Control subjects also showed trace recognition of this antigen. While the identity of this component is unknown at present it is of interest as a common factor. Based on the evidence that recognition of the 30 kD epithelial antigen is probably related to cross-reactivity with the consortium of bacteria, it is possible that in turn, a common bacterial antigen is responsible.

An alternative hypothesis is that recognition of epithelial antigens is a primary event in patients with periodontitis, with induced auto-antibodies demonstrating cross-reactivity for antigenic components from the consortium of bacteria. In this context, tobacco smoking is recognized as an important risk factor for the development of periodontitis [12]. One aspect could be that accumulating mutations within the oral epithelia facilitate auto-immunization of the subject. This could be best analysed by studies of tobacco smoke effects in germ free animals, but there appears to be a lack of supportive literature. Further, analysis of the patterns of epithelial recognition (Fig. 1) failed to support clear differences between smokers and nonsmokers within the generalized periodontitis group.

Amongst the isolated strains, S. intermedius and S. constellatus belong to the Streptococcus milleri group (S. anginosus, S. intermedius, S. constellatus). Members of this diverse group include Lancefield groups C and G, known to express protein G [13], a 30-kD receptor which binds to a common determinant on IgG and which could produce spurious apparent cross-reactivity [14]. This is unlikely to contribute to the data obtained for the S. milleri group in this study. Boiling of extracted bacterial products in SDS sample buffer for Western analysis would be expected to inactivate protein G. Further, periodontally healthy subjects showed weak recognition of bacterial extracts and the pattern of recognition by affinity-isolated antibodies was complex with reactive bands present at a range of molecular weights.

In the periodontitis lesion site, pathogenic attributes have been assigned principally to a consortium of anaerobic Gram negative bacteria, particularly Porphyromonas gingivalis [15]. In this regard, it was of interest that subjects with periodontitis had much greater recognition of the streptococci and Actinomyces identified in this study than did periodontally healthy subjects. Actinomyces and Streptococcus spp., occur in human dental plaque and may contribute to various disease states including gingivitis [16]. Streptococcus mitis is considered among the pathogenic species of viridans streptococci, associated with an extensive range of infections including meningitis, endocarditis, urinary tract infection as well as liver, lung and epidural abscesses [17, 18, 19, 20, 21, 22].

Streptococcus constellatus and S. intermedius were associated with indices of periodontal inflammation in interleukin-1 genotype(s) (IL-1 A and IL-1B) positive subjects [23]. Further, higher proportions of Streptococcus spp., particularly S. constellatus and S. intermedius have been detected in subgingival plaques of refractory periodontitis [24,25], and titres of serum antibodies to these organisms were positively correlated with disease while periodontally healthy subjects had low levels of serum antibodies reactive with these organisms [26] similar to the findings reported here. In addition, some members of the S. milleri group have been reported to be significant pathogens in a range of infections [27–30].

Antibodies reactive with epithelial components and related to the consortium of captured bacteria, were a consistent feature in the periodontitis subjects studied. It remains to be determined if the cross-reacting antibodies are pathogenic in the context of human gingival epithelium.

Streptococci and Actinomyces are common components of dental plaque and some species appear to contain antigens cross-reactive with epithelia. Molecular mimicry by these organisms could be involved in the generation of autoreactive immune responses during the development of periodontitis lesions with implications for both the pathogenesis of this disease and also for systemic manifestations of autoimmunity.