Microbial pathways to subvert host immunity generate citrullinated neoantigens targeted in rheumatoid arthritis

Abstract

The specific association between antibodies to citrullinated proteins and rheumatoid arthritis (RA) has centered interest on understanding why citrullinated proteins become immunogenic in this disease, which is believed to inform the origins of autoimmunity in RA. Since citrullination is a physiologic post-translational modification (PTM), one theory is that conditions promoting abnormal citrullination are initiators of self-reactive immune responses to citrullinated proteins in RA. Foremost candidates that dysregulate the normal balance of citrullination are microbial agents, which can exploit citrullination as an effector mechanism to subvert host antimicrobial activities and maximize their progeny. Here, we will use the host-pathogen interface as a unifying model to link microbe-induced citrullination and the loss of immunological tolerance to citrullinated antigens in RA.

Introduction

The interaction between the human host and microbial species is a dynamic process shaped by co-evolution over millennia in which a pathogen employs ways to evade or neutralize responses, while the host engages effector pathways to prevent or eliminate infection. For the host, this generally requires a compromise between effective antimicrobial immunity (increasing the chance of elimination of a pathogen) and risk of autoimmunity (increasing the chance of damage to self). Depending on selective evolutionary pressures encountered by human populations over time, this balance can shift to allow for higher risk of autoimmunity if this resulted in a net survival benefit. As such, host-microbe interactions and autoimmunity are often intricately linked.

It is hypothesized that certain HLA class II alleles have survived infectious bottlenecks — episodes of uncontrolled severe infection during which only a small percentage of the population survived— since individuals harboring these specific HLA alleles were able to mount an effective immune response. One of such examples selected for by evolutionary pressures is HLA-DR4 (encoded by the DRB1*0401 allele), a promiscuous HLA with ability to bind a large pool of microbial antigens [1]. Conversely, the HLA-DRB1*0401 allele (and a group of other HLA-DRB1 alleles that encode for a shared epitope protein motif) predispose to autoimmunity, specifically the development of RA [2], a disease thought to be driven by the production of anti-citrullinated protein antibodies (ACPAs) [3,4]. The mechanistic basis by which HLA-DRB1 shared epitope alleles confer this risk has not been conclusively established but is thought to result from either increased affinity for arthritogenic citrullinated self-antigens (peripheral hypothesis) and/or influence on thymic selection of autoreactive T cell receptors (TCRs) (central hypothesis) [5–7]. HLA-DRB1 shared epitope alleles are found in the majority of patients with ACPA-positive RA but are not sufficient to develop the disease. As such, the presence of factors that uniquely facilitate the generation of citrullinated autoantigens in the context of genetic risk may be required to break immunological tolerance.

Pathogens —in their co-evolution with humans— have developed a plethora of mechanisms to subvert host immunity. While genetic risk alleles (e.g., HLA-DRB1 shared epitope) provide a framework to understand susceptibility on a population and evolutionary level, environmental and stochastic factors need to be invoked to explain why citrullinated proteins become immunogenic in individuals at risk of developing RA [8]. Identification of such environmental factors and their unique interaction with the immune system is believed to provide answers to the origin of autoimmunity in RA. Since citrullination also occurs as a PTM during physiological processes (e.g., keratinocyte differentiation and cell activation, among many others), environmental factors promoting abnormal, dysregulated citrullination of non-tolerized proteins are often hypothesized to underlie the loss of immunological tolerance and autoimmunity in RA [8].

Here, we focus on bacterial and viral pathobionts as environmental agents that can confer abnormal citrullination during infection. Importantly, these pathogens promote dysregulated or abnormal protein citrullination as a mechanism to subvert antimicrobial immune responses of the host and thereby to maximize their progeny. Microbial mechanisms utilizing citrullination to subvert host immunity are variable and include i) expression of microbial citrullinating enzymes (bacterial peptidylarginine deiminases), ii) the over-expression of host citrullinating enzymes (human peptidylarginine deiminases, PADs), and iii) the over-activation of human PADs both in target and immune cells [9–11]. These processes, which lead to the generation of abnormally citrullinated self-proteins, can reprogram cellular pathways to favor viral replication, reduce the production of reactive oxygen species (ROS), decrease the antibacterial effects of human effector proteins such as histones, or simply kill innate immune cells [9–14]. We herein propose that pathogens evolved to exploit the increased citrullination events that occur during infection or heightened immune states to improve their fitness and survival. The interplay of genetically susceptible hosts and pathobionts that share co-optation of citrullination as effector pathway will be used as paradigm to explain the loss of tolerance to citrullinated antigens observed in RA.

Protein citrullination

Citrulline is generated in proteins by deimination of arginine residues mediated by the PAD enzymes. In mammals, PADs include a family of five related isozymes (PAD1, 2, 3, 4 and 6) [15]. Among these enzymes, only PAD1–4 have citrullinating activity [16]. PAD1–4 are calcium-dependent enzymes, which are specific for peptidylarginine (i.e., cannot citrullinated free L-arginine) and can only modify residues within polypeptide chains but not at their termini (i.e., they are endo-deiminases) [17,18]. Since arginine is positively charged, the change to a neutral residue (citrulline) reduces the net charge of the protein depending on the number of modified arginines, producing structural changes that can lead to either gain or —much more likely— loss of protein function [19–25]. To date, there is no evidence that citrullinated residues can be reverted to arginine and therefore, citrullination is considered an irreversible process. Thus, citrullination must be tightly regulated to avoid excessive citrullination of physiologic targets and any non-physiologic substrates [26].

Since calcium binding induces conformational changes that generate the active form of the enzyme [27–30], calcium is considered an important regulator of PAD activity [18]. Mechanisms that control calcium-dependent PAD activation, however, are not fully understood. A paradox in the study of PADs is that these enzymes require micromolar to millimolar amounts of calcium to achieve full activation in cells and in vitro [18,31]. Intracellular activation of PADs under these high calcium concentrations is however only observed during unique conditions, which also lead to cell death, such as keratinocyte differentiation, cytolysis induced by host or microbial pore-forming proteins, and artificially induced by calcium ionophores [13,31–36]. In keratinocytes, calcium-induced differentiation, citrullination, and cell death is a physiological process leading to skin keratinization and moisturizing [32]. Citrullination induced by pore-forming proteins, however, has only been associated with pathological conditions, such as autoimmunity and infection [13]. By inducing target cell lysis and calcium-mediated hyperactivation of PAD enzymes, pore-forming proteins —such as perforin, the terminal complement complex, and bacterial toxins— can induce dysregulated cellular citrullination in a process we termed leukotoxic hypercitrullination (LTH) [13]. During this process, uncontrolled PAD activation drives exuberant citrullination of an extensive number of substrates, termed hypercitrullination, which has been mechanistically linked to the abnormal accumulation of citrullinated proteins found in the RA joint and in septic foci [13,31,33].

For organisms to exploit citrullination as a mechanism to control cellular functions, overactivation of PADs would be an undesired outcome. It is therefore unlikely that calcium is the only regulator of PAD activity in cells. Indeed, citrullination of specific protein targets is observed under physiologic conditions where intracellular calcium does not exceed nanomolar concentrations [37–42], which —in theory— are insufficient for PAD activity. It is possible that these suboptimal calcium concentrations induce a PAD conformation that selects for only high-efficiency substrates, thereby limiting abnormal citrullination events. In this context, it is noteworthy that proteins whose cellular function is affected by citrullination (e.g., NF-kB, E2F-1, GATA3 and RORγt) are found in complex with PADs [39,41,42]. The presence of pre-formed substrate-enzyme complexes likely aids PAD efficiency and specificity upon activation with nanomolar changes in intracellular calcium during cell activation. Lastly, intracellular protein cofactors may be responsible for modulating calcium sensitivity and specificity of PAD enzymes, but such binding partners have not been identified.

Interestingly, while millimolar amounts of extracellular calcium are sufficient for efficient PAD activation, PAD activity is not detected in extracellular fluids in vitro unless reducing agents are exogenously added [43]. This finding suggest that PADs released or secreted from dying or activated cells are inactivated through oxidation [44], which may be a mechanism to avoid uncontrolled hypercitrullination in the extracellular space. Citrullinated extracellular proteins, however, can be detected in healthy human serum and synovial fluid at low levels, which are significantly enhanced in RA [19], suggesting that additional mechanisms are involved in controlling extracellular PAD activity and that such mechanisms are altered in RA [26].

Citrullination is dysregulated in RA

The PAD enzymes are highly conserved and share 50–55% sequence identity with each other [27], with different PADs having distinct substrate preferences and expression patterns in specific tissues [45]. Using mass spectrometry analysis, a recent study identified 209 proteins as targets of citrullination across 26 of 30 human tissues [46]. Together, this set of proteins constitute the healthy citrullinome. Interestingly, this study showed a drastic variation in the amount and type of citrullinated proteins among different tissues. The highest levels of protein citrullination were found in the brain and lung, which also contained the largest spectrum of citrullinated proteins (~60 proteins per tissue). In more than half of the tissues, however, less than 10 proteins were identified as targets of citrullination, and surprisingly, citrullination was not detected in the spleen despite having prominent PAD4 expression [46], suggesting that citrullination in immune cells is primarily important during inflammation. Importantly, the majority of proteins were only found in a single tissue in their citrullinated form, implying that citrullination of unique proteins is organ specific [46]. While it is expected that additional studies may increase the variety of proteins within the healthy citrullinome, these initial findings strongly suggest that citrullination under steady state conditions is tightly regulated, and that protein citrullination may regulate or be regulated by distinct factors in different tissues [46]. Functionally, the majority of citrullinated proteins across normal tissues are associated with broad categories of cytoskeleton organization and protein synthesis [46] (Figure 1).

Figure 1.

Network analysis of enriched pathways in healthy, rheumatoid arthritis (RA), cytomegalovirus (CMV) and Aggregatibacter actinomycetemcomitans (Aa) citrullinomes. Statistically enriched terms on each citrullinome were first identified with the multi-gene-list meta-analysis tool from Metascape [69] using the Gene ontology (GO) molecular function, GO biological process, and Reactome gene set collections (see details in Supplementary materials). An accumulative hypergeometric p-value < 0.01 and enrichment factor > 1.5 were used to select the most relevant terms. Then, the subset of representative terms (n=119) was converted into a network layout. Each term is represented by a circle node, where its size is proportional to the number of the input genes that fall under that term. Each node is displayed as a pie graph, where each sector is proportional to the number of proteins originated from a citrullinome. Color code for pie sector represents a citrullinome. Terms with a similarity score > 0.3 are linked by an edge (the thickness of the edge represents the similarity score). The network was built with Cytoscape [70] with a “force-directed” layout and with edge bundled for clarity. One term from each cluster is selected to have its term description shown as a label. Data from published citrullinomes were used for analysis (Supplementary File 1) [10,19,31,33,46–49]. The healthy citrullinome includes the analysis of the following tissues: adrenal gland, appendix, bone marrow, brain, colon, duodenum, endometrium, esophagus, fallopian tube, fat, gall bladder, heart, kidney, liver, lung, lymph node, ovary, pancreas, placenta, prostate, rectum, salivary gland, small intestine, smooth muscle, spleen, stomach, testis, thyroid, tonsil, urinary bladder [46]. The RA citrullinome includes a compilation of proteins identified in synovial tissue, the fluid fraction of synovial fluid, and the cellular fraction of synovial fluid [19,31,47–49].

This array of citrullinated proteins importantly contrasts with the inflamed rheumatoid joint, in which both the type and the magnitude of citrullinated proteins are significantly amplified [19,31,47–49]. The study of protein citrullination in RA has included the analysis of the three major inflammatory components in the RA joint, namely, the synovium, and the cellular and the fluid fractions of synovial fluid (SF) [19,31,47–49]. Together, more than 250 citrullinated proteins have been identified in the RA joint, which comprise the RA citrullinome. The identify of citrullinated proteins in the RA joint includes proteins found across healthy tissues, although the vast majority are proteins not detected in the healthy citrullinome (Figure 2). This finding is not surprising as citrullination varies by tissue and is highly influenced by inflammation. Indeed, the spectrum of citrullinated proteins in RA reflects hypercitrullination of cellular and soluble components enriched in the distinct inflammatory compartments in the rheumatoid joint. For instance, the fluid fraction of RA SF contains elevated levels of citrullinated soluble extracellular proteins, such as components of the complement and coagulation cascades [19,48,49] (Figure 1). The cellular fraction of RA SF (dominated by neutrophils) and the synovium (containing synovial and inflammatory cells) are the source of intracellular citrullinated proteins [19,31,47,48], which are associated with categories of cytoskeleton organization and cellular stress (Figure 1). In particular, it is intriguing that citrullination in RA, but not in the healthy citrullinome, targets proteins found in metabolic pathways (Figure 1), suggesting that these proteins became accessible to citrullination under inflammatory conditions.

Figure 2.

Circos plot of citrullinated proteins from healthy, RA, Aggregatibacter actinomycetemcomitans (Aa) and human cytomegalovirus (CMV) citrullinomes. Purple curves link identical proteins. The inner circle represents each citrullinome, where proteins are arranged along the arc. Proteins present on multiple lists are colored in dark orange, and unique proteins to a citrullinome are shown in light orange. The Circos plot was done using Metascape [69] and citrullinated proteins from published citrullinomes (Supplementary File 1) [10,19,31,33,46–49]. The healthy citrullinome includes the analysis of the following tissues: adrenal gland, appendix, bone marrow, brain, colon, duodenum, endometrium, esophagus, fallopian tube, fat, gall bladder, heart, kidney, liver, lung, lymph node, ovary, pancreas, placenta, prostate, rectum, salivary gland, small intestine, smooth muscle, spleen, stomach, testis, thyroid, tonsil, urinary bladder [46]. The RA citrullinome includes a compilation of proteins identified in synovial tissue, the fluid fraction of synovial fluid, and the cellular fraction of synovial fluid [19,31,47–49].

Together, the RA joint amasses an unparalleled amount and spectrum of citrullinated proteins not been identified in any other single tissue or organ [31], which has been mechanistically linked to LTH [13]. These findings support that citrullination is highly dysregulated in RA, and suggest that dysregulated PAD activation is targeting proteins within pathways that are amplified in the RA joint. Whereas cell death induced by host pore-forming pathways (i.e. complement and perforin) may represent a primary source of citrullinated proteins in patients with established RA [31], environmental factors that trigger PAD dysregulation and hypercitrullination, such as microbial agents, remain primary candidates to initiate an immune response to citrullinated proteins in the context of genetic predisposition.

The role of citrullination in host-pathogen interactions

The host-microbial interaction is a dynamic process in which the pathogen attempts to evade the immune attack while the host tries to prevent and eradicate infection with minimal collateral damage to itself. Among multiple pathways activated during infection, PTMs are important determinants of microbial virulence. For instance, phosphorylation, ubiquitination, methylation, acetylation, ISGylation, SUMOylation and Neddylation, among others, are known to exert anti-viral effects by either inducing interferon (IFN) responses, destabilizing viral proteins, or targeting viral proteins to degradation [50,51]. Similarly, pathogens promote their survival by interfering with or hijacking the host’s PTM pathways, which can enhance virial replication, as well as induce inactivation and degradation of host antiviral proteins [50,51]. Citrullination is not an exception to this paradigm. By targeting transcription factors and signaling molecules, PADs are involved in regulating inflammatory responses relevant to clearance of pathogens, such as cytokine expression [39,52], polarization towards Th1, Th2, or Th17 responses [42,53], dendritic cell maturation [54], NLRP3 inflammasome assembly [55], NF-kB nuclear translocation [41], ROS production [12], and TLR4 agonism [56]. Analogously, growing evidence suggests that pathogens can use citrullination for their own advantage. Although this has not been as extensively studied as for other PTMs, we will provide several examples that highlight diverse mechanisms evolved by pathogens to gain survival benefits from citrullination.

Bacterial peptidylarginine deiminase

The first evidence suggesting that citrullination is a mechanism used by pathogens to evade host defenses originated from the study of Porphyromonas gingivalis (P. gingivalis), a periodontal pathogen associated with chronic periodontitis [57]. Somewhat unique among microbes of human pathogenic potential, this organism produces and secretes a PAD enzyme (hereafter PPAD). Different to mammalian PADs, however, PPAD citrullinates free L-arginine in addition to peptidylarginine and does not require calcium or any other metal ion for activity. Importantly, PPAD has preference for (and is potentially restricted to) citrullination of carboxy-terminal arginines (C-terminal exodeiminase) [57,58]. PPAD is considered a virulence factor, which has been proposed to subvert the immune system by at least two mechanisms. One involves the production of ammonia, a side-product of citrullination, which has a negative effect on neutrophil function [57]. The second mechanism is by abrogating the activities of host proteins and peptides, which function is dependent on C-terminal arginine residues, such as C5a [59].

Because the tempting association between RA and periodontitis, the presence of PPAD has been an attractive mechanism to link P. gingivalis, RA, and the production of citrullinated proteins as a putative mechanism to induce the production of antibodies to citrullinated proteins [9]. Although a major limitation to this model is that C-terminal citrullinated proteins are not preferential targets of ACPAs [60], the recent finding of a P. gingivalis strain encoding a mutant PPAD with endodeiminase activity renewed interest on unique P. gingivalis strains that could play a role in RA pathogenesis [61]. Whether endodeiminase PPAD has different virulent activity against host immune responses is unclear. Similarly, how and whether this endodeiminase PPAD may alter the content of proteins within the healthy citrullinome is unknown.

Viruses and the PAD enzymes

Recent studies have shown that human rhinovirus (RV) and human cytomegalovirus (CMV) can induce the expression of PAD enzymes in target cells [10,11]. While the effect of PAD expression on RV infected cells is unclear [11], the study of CMV provided initial evidence that viral-induced hypercitrullination can have a critical role on viral replication [10]. During infection of human fibroblasts, CMV induces overexpression and activation of PAD2 and PAD4, promoting citrullination of at least 40 viral and 177 host proteins. While the function of both viral and cellular proteins may be affected by citrullination, the finding that PAD inhibition blocks CMV replication further supports that citrullination is a key event for viral replication. Mechanistically, this study showed that citrullination of the IFN-induced protein with tetratricopeptide repeat 1 (IFIT1) affects its ability to bind to 5’-triphosphorylated RNA, thereby impairing the antiviral activity of this protein [10]. Viral-induced IFIT1 citrullination is therefore a potential mechanism used by CMV to evade host antiviral immunity.

Bacterial-induced leukotoxic hypercitrullination

Pore-forming toxins secreted by highly virulent bacterial strains are potent inducers of LTH [13,33]. Leukotoxin A (LtxA) secreted by the periodontal pathogen Aggregatibacter actinomycetemcomitans (Aa) and Panton–Valentine leucocidin (PVL) from Staphylococcus aureus (S. aureus) are prototype examples of LTH-inducing bacterial toxins [13,33]. These toxins subvert the immune system by inducing membranolytic cell death in leukocytes, particularly neutrophils [62,63]. Through this action, key components of antimicrobial pathways are citrullinated during LTH [33].

The citrullinome induced by LtxA in human primary neutrophils comprises at least 86 proteins [33]. Of particular interest, two cytosolic components of the phagocyte NADPH oxidase (NOX2) complex [i.e. neutrophil cytosol factor 1 (NCF1) and NCF2 (also known as p47-phox and p67-phox, respectively)] and histone H3 are citrullinated during LtxA-induced LTH [33]. Interestingly, a recent study demonstrated that citrullination of p47phox and p67phox leads to dissociation of the complex (which also includes cytosolic PAD4), suppression of NADPH oxidase assembly, reduced ROS production, and impaired killing of phagocytized bacteria [12]. Several studies have similarly shown that citrullination of histone H3 reduces its bactericidal activity [14,64]. Cytoskeletal components such as actin, vimentin, moesin, lamin B, coronin, and tubulin are also citrullinated during LtxA-induced LTH [33], likely facilitating the downfall of the target cell by disrupting structural interaction of the cytoskeleton. The induction of LTH by bacterial pore-forming toxins is therefore a powerful mechanism to suppress host defenses and promote pathogen survival.

RA and microbes reshape the host citrullinome

Proteins involved in host immune defenses are not the only targets during microbial-induced citrullination. Rather, by changing normal patterns of PAD expression and bypassing mechanisms that control PAD activation, pathogens can induce hypercitrullination of a broad range of proteins, generating novel citrullinomes. Among different pathogens known to dysregulate PAD activation, the citrullinomes induced by LtxA from Aa and CMV are currently available [10,33]. Similar to the RA joint, the citrullinomes induced by Aa and CMV comprise proteins described within the healthy citrullinome, as well as unique citrullinated proteins (Figures 1 and 2).

The presence of citrullinated proteins specific for each pathogen likely reflects citrullination of cell type specific proteins and cellular pathways modulated by these pathogens (Figure 2). For instance, the collection of citrullinated proteins induced by Aa and CMV are defined by the type of target cell used in the assays – human neutrophils for LtxA-Aa and human fibroblasts for CMV [10,33]. Moreover, in the case of Aa, proteins involved in apoptosis are particular targets of citrullination likely because this pathway is activated in neutrophils, whereas citrullination of proteins involved in biological processes modulated by the virus (e.g., translation and protein degradation) is enhanced during CMV infection (Figure 1).

Nevertheless, despite that citrullinated proteins are generated under different conditions in RA, Aa and CMV, it is striking that a large number of citrullinated proteins are common among their citrullinomes, and many of these proteins are not citrullinated in healthy tissues (Figure 2). In this regard, it is intriguing that proteins involved in metabolism (i.e. precursor metabolites and energy) are common targets of citrullination in RA, Aa and CMV, but spared in healthy tissues (Figure 1), suggesting that proteins within pathways that are amplified/activated in response to cellular stress and inflammation are preferential targets during dysregulated PAD activation. Moreover, these findings supporting the notion that pathogens inducing PAD activation and hypercitrullination have the potential of generating citrullinated neoantigens found in RA [10,33].

Conclusion and outlook

While the presence of citrulline residues in proteins has been known for almost 90 years [65], our understanding of the biology of protein citrullination —how citrullination is regulated both intracellularly and extracellularly, whether citrullination is really irreversible, how citrullinated proteins are cleared, physiological and pathological substrates for citrullination and their functional consequences, exploitation of citrullination by pathogens— is still limited. Similar to other PTMs, citrullination is likely a dynamic process involved in activating and inhibiting cellular and extracellular protein functions. Importantly, although citrullination is enhanced during inflammation and infection, this process is not intrinsically pathological. Citrullination likely drives both inflammatory and anti-inflammatory effects depending on the target protein, which may promote pathogen clearance and resolution of inflammation. In contrast, conditions chronically driving hypercitrullination may confer a higher risk to autoimmunity, particularly in the setting of genetic predisposition.

While the RA joint is an abundant source of citrullinated autoantigens to maintain the production of autoantibodies in patients with established disease, antibodies to citrullinated proteins in individuals who will develop RA are detected several years before clinical evidence of joint inflammation [8]. Therefore, the joint is unlikely to be a source of autoantigens during the pre-clinical phase of the disease. An alternative to this joint-centric hypothesis is the existence of chronic extra-articular sources of citrullinated neoantigens, which may drive the loss of immunological tolerance toward these modified proteins before joint inflammation develops and sustains disease. The finding that some microbes exploit citrullination to increase their fitness is intriguing, and provides a framework by which environmental factors participate in the development of RA. However, while many pathogens can induce limited PAD activation, mostly generating citrullination of few proteins like histone H3 [66], the focus of this model is on microbial agents that generate hypercitrullination/LTH and, in consequence, fundamentally alter the host citrullinome.

An interesting feature among conditions that can drive dysregulated PAD activation, such as RA and infection, is the massive expansion in the repertoire of citrullinated proteins beyond the healthy citrullinome. Thus, citrullinated proteins normally distributed at low quantities among different tissues are generated at high amounts in a single cell type or tissue. In parallel, citrullination can target proteins that may not be part of the healthy citrullinome. Thus, during this process of hypercitrullination, distinct citrullinomes are generated, which may have the capacity to be immunogenic.

In summary, we propose a model in which abnormal citrullinomes induced by environmental factors, particularly microbes, are a source of neoantigens that can initiate the antibody response to citrullinated proteins in susceptible individuals during pre-clinical RA. In this model, the finding that citrullinated proteins are generated during infection, which concur with the immune response against the pathogen, may also create the proper inflammatory milieu to enhance immunogenicity of microbe-induced citrullinated proteins. In the case of CMV, it is intriguing that viral proteins are also citrullinated [10], adding an extra layer of potential immunogenicity. Nevertheless, since autoantibodies in RA target protein motifs containing peptidylcitrulline rather than specific proteins [60,67], viral citrullinated proteins may not have an additional immunogenic advantage over novel citrullinated host antigens. Importantly, microbial-induced hypercitrullination as a model to initiate RA does not evoke or require a single pathogen as the cause of the disease in all patients. Instead, chronic or acute exposure to single or different pathogens able to induce the production of abnormal citrullinomes is the central model of this proposal [68]. Once an autoimmune response to citrullinated antigens has been initiated, hypercitrullination induced by host immune pathways are likely to sustain autoantigen production in established RA [31].

Highlights.

• Antibodies to citrullinated proteins are a hallmark in RA.

• While citrullinated proteins are found in tissues under physiological conditions, citrullination of unique antigens appears to be organ specific, suggesting that this process is tightly regulated, and protein citrullination may control or be controlled by distinct factors in different tissues.

• In RA, the type and magnitude of protein citrullination is amplified compared to physiological conditions. In particular, the inflamed joint in RA contains large amounts of citrullinated proteins found in only limited amounts in healthy tissues, as well as citrullinated antigens not found within the healthy citrullinome. These findings support that citrullination is dysregulated in RA.

• Microbes exploit citrullination to subvert host immune defenses, including reprograming cellular pathways to favor viral replication, reducing the production of reactive oxygen species, and decreasing antimicrobial functions of effector proteins such as C5a and histones.

• Similar to RA, pathogens bypass mechanisms that regulate PAD expression and activation, modifying the normal landscape of citrullinated proteins.

• In individuals with genetic susceptibility to develop RA, chronic exposure to abnormal citrullinomes generated by microbes may underlie the initial loss of immunological tolerance to citrullinated proteins and the development of RA.