Introduction

Rheumatoid arthritis (RA) is an autoimmune disease leading to synovial inflammation and destruction of cartilage and bone. RA, like many autoimmune diseases, is proposed to occur in genetically at risk individuals, in response to a trigger or triggers. Viral or bacterial infections have long been considered as putative triggers for this disease. Periodontitis (PD), is also a chronic inflammatory condition initiated by bacterial accumulation on the interface between the teeth and gingiva and modified by genetic, environments (eg smoking) and the host inflammatory reaction. The similarities between RA and PD were first recognized by Snyderman and McCarty and since then considerable evidence has accumulated to support the concept of a strong interrelationship between these two inflammatory conditions.

The association between RA and periodontitis (PD) and the relationship of smoking to both conditions has become contextually relevant with the knowledge that immune responses to citrullinated peptides are associated RA with and likely to be pathogenic. Smoking is known to increase levels of citrullination via increases in the enzyme peptidylarginine deiminase (PAD). The opportunistic infection that characterizes PD has become particularly relevant in this context. Porphyromonas gingivalis ( P. gingivalis ), the major pathogen in PD is the only bacterium known to express a PAD enzyme (PPAD), and has been reported to be significantly associated with RA . The role of PPAD in generating citrullinated bacterial and host proteins, the potential for these to interact with the host immune system in genetically at-risk individuals, and the implications for treatment will be the main focus of this review.

The link between rheumatoid arthritis and periodontitis

Rheumatoid arthritis (RA) is an autoimmune inflammatory arthritis characterised by evidence of serological autoimmunity and persistent synovial inflammation. It has a prevalence of approximately 1% with a peak incidence in pre-menopausal women. Inflammation of the synovial lining of joints results in destruction of cartilage and erosion of bone, which, if unchecked, will lead to loss of function, deformity and chronic pain. Periodontitis (PD) is the commonest described oral disease with highly variable estimated prevalence rates of up to 50–60% in most populations . Opportunistic infection with bacteria within the sub-gingival biofilm is associated with periodontal destruction, which can lead to tooth loss and importantly create a significant systemic inflammatory burden. Virulence of the colonizing bacteria as well as the intensity of the host immune response is thought to contribute to the severity of the disease .

The increasing recognition of an epidemiological association between PD and RA must be considered in the context of the very high prevalence of PD in particular, the chronic nature of both conditions and shared genetic and environmental risk factors . Evidence supporting a causal link has emerged from subsequent understanding of the pathogenic mechanisms underpinning both diseases, which will be discussed in later in this review. Nevertheless a range of epidemiological data examining presence and severity of both diseases as well as temporal relationships between the onset of RA and PD has driven deeper inquiry into pathogenic links. Although PD is common, patients with longstanding active RA were shown to have increased incidence of PD compared with healthy non-RA subjects . Correspondingly RA prevalence is increased in patients with PD . Indeed an increase in all systemic conditions including RA was reported in PD patients In a study examining shared risk factors for PD and RA, Mikuls et al., recently confirmed higher prevalence of PD in 277 RA patients compared with 330 osteoarthritis controls using standardised periodontal examination .

The recent OSARA study examined oral health over time in outpatients with rheumatoid arthritis and found that 94% of RA patients had PD. In 46 percent of RA patients PD was described as severe . Odds ratio’s of having both conditions are variably estimated as being between 2:1 and 8:1 depending on the criteria used to define periodontitis. These odds ratios must be considered in the context that RA has low but stable prevalence estimates whereas periodontitis prevalence estimates are highly divergent. However it is compelling that in the large US population study NHANES III, more than 50% of RA patients were edentulous .

The link between rheumatoid arthritis and periodontitis

Rheumatoid arthritis (RA) is an autoimmune inflammatory arthritis characterised by evidence of serological autoimmunity and persistent synovial inflammation. It has a prevalence of approximately 1% with a peak incidence in pre-menopausal women. Inflammation of the synovial lining of joints results in destruction of cartilage and erosion of bone, which, if unchecked, will lead to loss of function, deformity and chronic pain. Periodontitis (PD) is the commonest described oral disease with highly variable estimated prevalence rates of up to 50–60% in most populations . Opportunistic infection with bacteria within the sub-gingival biofilm is associated with periodontal destruction, which can lead to tooth loss and importantly create a significant systemic inflammatory burden. Virulence of the colonizing bacteria as well as the intensity of the host immune response is thought to contribute to the severity of the disease .

The increasing recognition of an epidemiological association between PD and RA must be considered in the context of the very high prevalence of PD in particular, the chronic nature of both conditions and shared genetic and environmental risk factors . Evidence supporting a causal link has emerged from subsequent understanding of the pathogenic mechanisms underpinning both diseases, which will be discussed in later in this review. Nevertheless a range of epidemiological data examining presence and severity of both diseases as well as temporal relationships between the onset of RA and PD has driven deeper inquiry into pathogenic links. Although PD is common, patients with longstanding active RA were shown to have increased incidence of PD compared with healthy non-RA subjects . Correspondingly RA prevalence is increased in patients with PD . Indeed an increase in all systemic conditions including RA was reported in PD patients In a study examining shared risk factors for PD and RA, Mikuls et al., recently confirmed higher prevalence of PD in 277 RA patients compared with 330 osteoarthritis controls using standardised periodontal examination .

The recent OSARA study examined oral health over time in outpatients with rheumatoid arthritis and found that 94% of RA patients had PD. In 46 percent of RA patients PD was described as severe . Odds ratio’s of having both conditions are variably estimated as being between 2:1 and 8:1 depending on the criteria used to define periodontitis. These odds ratios must be considered in the context that RA has low but stable prevalence estimates whereas periodontitis prevalence estimates are highly divergent. However it is compelling that in the large US population study NHANES III, more than 50% of RA patients were edentulous .

Evidence supporting the association between periodontitis and RA

In addition to epidemiological data that supports the association between RA and PD, the shared environmental associations as well as the underlying biology and local pathological outcomes is striking. More than 100 genetic susceptibility loci have now been identified for RA . The association of a shared DRB1 locus with RA has been documented over many decades and has recently come into sharp focus in the context of its relationship with seropositive RA specifically Anti-Citrullinated Peptide Antibody (ACPA)-positive RA. HLADR4 has also been associated with rapidly progressive PD . A higher frequency of the shared epitope in PD patients was observed compared with controls . A recent candidate–gene association study genotyped a very large case–control sample of aggressive PD. Of 47 genes with genome wide significance in RA and SLE, two genetic variants IRF5 and PRDM1 were identified as shared susceptibility factors between RA and PD .

Environmental risks such as smoking are strongly associated with both PD and RA. In a study of approximately 4500 patients, including three prospective cohorts, a clear dose–response effect of smoking on the risk of developing RA was identified . The epidemiological relationship between RA and smoking which has been previously identified in large population studies, including the Nurses Health Study, became a focus of intense interest when smoking was identified as a stimulus to citrullination of peptides via the induction of Peptidyl Arginine Deiminase (PAD) enzyme. Citrullination, the process of enzymatic conversion of arginine to citrulline, is a physiological process that occurs in the context of inflammation and cell death. Citrullination via PAD activation requires the mobilization of free intracellular calcium in mammalian systems. Such citrullinated peptides could drive ACPA responses. The subsequent epidemiological delineation of smoking as a risk factor for ACPA positive RA rather than ACPA negative RA and the accumulating evidence that ACPA positivity is genetically restricted has created a cogent link between specific genes and environmental triggers in driving disease. These observations have equally formed the basis of exciting translational studies, which identify the structural interaction with citrullinated peptides, and the HLADRB1 (shared epitope)-encoded structurally modified antigen-binding pocket . In an elegant proof of concept study, mice, which overexpress the human shared epitope, develop florid arthritis in response to citrullinated fibrinogen but not in response to unmodified fibrinogen . This molecular and in vivo evidence that specific genetic risk can predispose to more effective binding of citrullinated peptides and drive more intense T-cell responses to these citrullinated peptides will be particularly relevant to understanding possible causal links between PD and RA, discussed later in this review.

Interestingly PAD-2 and PAD-4, citrullinated proteins and ACPA have been identified in inflamed human periodontal tissues . This could provide another source of extra-articular citrullination and serve as a link between PD and RA. The role of periodontal citrullination and RA is still in its formative stages and any link to RA and shared HLA epitopes will be important to establish.

Periodontitis, P. gingivalis and citrullination

An additional pathway by which proteins may be citrullinated in periodontal tissues is through PPAD produced by P gingivalis . The relationship between the infection which characterises PD and its relationship to ACPA positivity, RA development and/or exacerbation is the principal focus of this review and will therefore be discussed in detail in this section.

Periodontitis and bacterial infection

In periodontitis predominantly gram-negative anaerobic bacteria are located within a bacterial biofilm on the tooth surface in a sub-gingival location. Bacterial and host factors drive local inflammation and tissue destruction. Osteoclast activation results in alveolar bone loss in a process, which closely resembles cytokine-driven osteoclast activation, and bone erosion in RA. Among the micro flora associated with PD, one of the best studied is P. gingivalis .

Periodontitis and systemic inflammation

Pathogenic bacteria are able to participate in a complex interaction with resident bacteria. These pathogenic bacteria are protected from host defense by virtue of this co-operation, the biofilm, and sub-gingival location . P. gingivalis has been shown to promote dysbiosis by down-regulating complement and TLR signaling . Another study demonstrated differential suppression of macrophage subsets by P. gingivalis favouring persistent carriage . Notwithstanding this protection, chronic bacterial infection in PD leads to the release of bacterial toxins that can stimulate the immune system . Evidence for activation of inflammatory cytokines, integrins, complement and prominent local activation of neutrophils is well established . Given this evidence of chronic immune activation, it is not surprising that chronic PD has recently been associated with elevations in high sensitivity CRP (hs-CRP), both locally and systemically . In this way the chronic bacterial burden itself and associated immune response may mimic the effect of adjuvant, which boosts immune responses in a range of animal models of inflammatory disease most notably experimental arthritis. Over 400 bacterial species have been identified within the periodontal microbiota and recent reports indicate this is likely to increase as more sophisticated microbiome techniques are used to probe for as yet unidentified periodontal pathogens . The possible contribution of these to augmentation of immune responses relevant to RA is beyond the scope of this review. P. gingivalis has been the subject of intense investigation in the last decade for reasons relating to its unique properties discussed in the section that follows.

P. gingivalis infection, PPAD and citrullination

The virulence of P. gingivalis has been attributed to the production of proteolytic enzymes known as gingipains, which contribute to local tissue destruction and direct apoptosis of gingival cells, as well as overriding host defense mechanisms including complement. Of key relevance to this review, P. gingivalis remains the only known bacterium which can produce the enzyme Peptidyl Arginine Deiminase (PAD) . This enzyme was subsequently named P. gingivalis peptidyl arginine deiminase (PPAD). Following this observation, it was proposed that PPAD could be a link between PD, citrullination and RA . This enzyme, like human PAD, converts arginine to citrulline resulting in post-translational structural protein modification. This structural modification is believed to generate a range of neo-antigens including citrullinated filaggrin, vimentin, collagen and enolase. Although P. gingivalis PAD (PPAD) is not fully homologous to human PAD, it has been clearly shown to citrullinate both host and bacterial peptides . In strong support of this, the titre of antibodies to P. gingivalis has been shown to positively correlate with circulating levels of ACPA in RA .

PPAD is distinct from mammalian PAD in having no requirement for calcium flux to activate enzyme activity . Gingipains co-operate with PPAD by generating arginine-containing residues through proteolytic cleavage. PPAD unlike human PAD can deiminate free arginine as well as C-terminal arginine residues . Underscoring this co-operation is evidence that P. gingivalis mutants lacking arginine-gingipain proteolytic cleavage function cannot facilitate citrullination in spite of fully functional PAD . Evidence that PPAD citrullinates peptides generated by gingipain degradation of fibrinogen and alpha-enolase is of particular interest. This is in the context of increasing evidence that antibodies to citrullinated enolase peptides (CEP) have higher specificity in RA than other ACPA species .

These demonstrated biological propensities of PPAD become very meaningful in the context of studies, which examine the temporal relationship between periodontitis and RA in humans and the possible causal relationships in animal models.

Patients with new onset RA have been observed to have a higher prevalence of PD; out of keeping with their young age and smoking status . Evidence that P. gingivalis infection precedes RA in humans has been complemented by animal studies supporting a causal relationship. For example, mice with pre-existing periodontitis developed earlier and more severe arthritis . P. gingivalis infection appeared to sensitise animals to early arthritis development. The developing arthritis in these animals was characterised by more dramatic bone and cartilage destruction and a more rapid clinical course. Of critical importance, only viable P. gingivalis inoculation produced this effect. Other Porphyromonal pathogens such as P intermedia were not able to mediate this exacerbation of arthritis. This animal data reconnects with observations in humans in which circulating antibodies to P. gingivalis , but not other Porphyromonal species, correlated with ACPA positivity and titre . Mice infected with wild-type P. gingivalis in the context of collagen-induced experimental arthritis develop antibodies to citrullinated peptides including enolase . Recombinant human α-enolase and P. gingivalis enolase, either citrullinated or uncitrullinated, were used to immunise DR4-IE-transgenic mice and control mice. Mice immunised with either citrullinated P. gingivalis enolase or human enolase developed arthritis. This study has important and broad implications given that it demonstrates that the human shared epitope mediates increased MHC class II binding of citrullinated peptides of both human and bacterial origin. It additionally demonstrates that the specific immune response that ensues is arthritogenic.

P. gingivalis oral infection has been shown to increase the release of a range of cytokines including RA-relevant monokines IL-1, IL-6 and TNF . In the same study P. gingivalis increased Th17 responses and IL-23 production.

However, a critical finding that a PPAD deficient strain of P. gingivalis lacks pathogenic potential in terms of both PD and RA, indicates that the citrullination function of P. gingivalis is of paramount importance . In this study the genetically modified PAD-deficient strain of P. gingivalis was associated with reduced development of PD, and importantly decreased clinical arthritis, bone erosion and circulating ACPA levels compared with animals inoculated with wild-type P. gingivalis .

Auto-citrullination of PPAD and associations with RA

An additional layer of complexity relates to the recognition that the PPAD enzyme itself can be citrullinated in a process referred to as auto-citrullination . The Cit-PPAD, which is generated may then, itself, act as a neo-antigen. Evidence for immunogenicity of PPAD is emerging from both human and animal studies. Circulating anti-PPAD antibodies are reported in RA . Evidence of elevated levels of anti-PPAD in RA serum also point to the capacity of a bacterial protein to break tolerance. In contrast to this, in a recent study , patients with RA and healthy controls were assayed for IgG antibodies to citrullinated recombinant PPAD (rPPAD) and unmodified rPPAD. Anti-PPAD antibodies did not correlate with ACPA and were decreased in RA patients with PD, leading the authors to propose that anti-PPAD antibodies may have a protective role for the development of PD in patients with RA. Whilst both PD and P. gingivalis constitute an inflammatory burden and may shape auto-reactivity in RA, the overwhelming evidence converging from human and animal studies supports a specific role for the unique PPAD enzymatic function of P. gingivalis . Gingival PPAD activity, in conjunction with the constitutive citrullination occurring as a result of the gingival inflammation, provide a continuous and evolving range of citrullinated host and possibly bacterial peptides leading to progressive loss of tolerance initially locally and then systemically. Physiological citrullination in the synovium in the context of cell death, inflammation or repair, may then direct this primed immune system to attack joint tissues.

Immune responses carbamylated peptides: anti-CarP Ab and RA

An emerging literature points to evidence of immune responses to other post-translationally modified peptides that may predate ACPA responses. Carbamylation is the non-enzymatic modification of molecules containing primary amine or thiol groups. Anti-Carbamylated protein antibodies (Anti-CarP) have been described in patients with RA but also in those with arthralgia or ‘pre-arthritis’ . In this context, anti-CarP antibodies are associated both with increased risk of developing RA, and increased risk of erosive RA. Anti-CarP antibodies have been detected in 20% of ACPA negative RA patients . In contrast to ACPA, the lack of association of Anti-CarP Ab with HLA-DRB1 alleles, other known risk genes or smoking point to distinct biological mechanisms underlying their formation. A striking observation is that mice with CIA do not develop reliable ACPA responses, except after inoculation with P. gingivalis . In a recent study, however, in a mouse CIA model, mice developed anti-CarP antibodies before the onset of arthritis . As arthritis developed the anti-CarP Ab response intensified. This study suggests that breaking tolerance to Carbamylated peptides is perhaps an earlier and easier step than loss of tolerance to citrullinated peptides. Whilst P. gingivalis , through gingipain activity, is able to mediate proteolysis, there are no direct links as yet published between PD, P. gingivalis and carbamylation. However, a recent publication has proposed a role for carbamylation within inflamed periodontal tissues in the RA/PD axis and this was substantiated by immunohistochemical demonstration of carbamylated proteins within inflamed human periodontal tissues.

Dysbiosis and dysregulated immunity: chicken or egg?

Much of the above review aims to highlight increasing evidence, which supports a causal or aetiopathogenic relationship between chronic gingival inflammation and P. gingivalis infection with RA. However, this must be balanced by examining the complexity surrounding the interaction between dysregulated microbiota and the immune system. A return to what is known about shared genetic risks in PD and RA may be relevant. Shared susceptibility genes in PD and RA are critically involved in shaping antigen recognition, tight binding and hence subsequent T-cell responses. It is therefore conceivable, and indeed has been suggested, that these same factors may shape the host inflammatory and immune response, which occurs as gingivitis is developing. The progression of periodontitis and the survival of particular bacteria such as P. gingivalis may therefore be favoured by certain host immune factors. The evolution and composition of resident microbial populations in all mucosal surfaces is thought to be critically important in triggering a range of immune and inflammatory diseases . Whilst the initial focus has related to the consequences of host immune responses to this microbiota, there is increasing interest in the role of local mucosal immunity and dysregulation of this in driving a particular microbiota composition which favours chronic inflammation or immune activation.

As with RA, many of the hundreds of risk genes for inflammatory bowel disease relate to immune function . Genetic factors which lead to altered innate immune surveillance may predispose to increased mucosal translocation of bacteria. In mice the NOD2 genotype was shown to lead to instability in the composition of microbiota that could promote colon inflammation. Genotype dependent disease could be transferred via maternally transmitted microbiota to NOD2 deficient and wild type hosts . Variation in host susceptibility to P. gingivalis has been shown in mouse models . In this study the Balb-CJ mouse strain was highly susceptible to PD development after P. gingivalis inoculation compared with five other strains. In humans, genetic variants associated with neutrophil function were associated with more aggressive periodontitis. Fc Gamma receptor and IL-6 polymorphisms have been associated with increased odds of identifying pathogenic bacteria including P. gingivalis in PD patients . Class II HLA polymorphisms in the DQ locus have been described, which are associated with susceptibility to, as well as protection from, PD . In another study, certain HLA variants were negatively associated with colonisation by PD-related microorganisms .

Whilst HLA association studies in the last decade have identified relationships between HLA polymorphisms and PD, the hypothesis that upregulated immune responses to local bacteria have driven more aggressive PD, has been favoured. However, evolving theories especially from examination of gut dysbiosis, suggest that genetically determined unbalanced, or inefficient local mucosal immunity, along with environmental factors, may favour the persistence of pathogenic bacteria. Known genetic associations of PD may, therefore, be mechanistically linked via predisposition to the persistence of pathogenic bacteria including P. gingivalis . The precise ways in which these genetic risks may modify oral mucosal immunity to support infection and persistence of pathogenic bacteria will have important implications for both PD and RA.

Enormous progress in the understanding of the pathogenesis of RA and PD has, and will continue to have impact on treatment of both diseases, as well as the new goal of disease prediction and prevention. The management of rheumatoid arthritis has evolved rapidly in the last three decades and the treatment approaches directly reflect an understanding of the underpinning biology. In the section that follows, the implications for treatment of RA, which are raised by specific aspects of this review, will be discussed in more detail.