While antibodies against CCPs are considered to be a specific diagnostic marker for RA [39
], there is evidence that citrullinated proteins are also generated in tissues affected by other inflammatory diseases including MS. Here we utilised protein arrays to characterise the evolution of autoantibody responses in CIA and EAE. We observe targeting of the native inducing autoantigens and a limited set of citrullinated epitopes in pre-disease and acute disease samples, and expansion of B cell responses to target multiple citrullinated epitopes in both chronic CIA and EAE. The native counterparts of the targeted citrullinated proteins were not targeted by the autoantibody responses in these models.
Our findings extend the results of others who described antibody reactivity to CCP in CIA [14
] and the induction of EAE with citrulline-substituted MBP peptides [18
]. In CIA we observed development of ACPA responses pre-boosting (Figures ), which is five to eight days before the onset of arthritis and consistent with the timing of development of anti-CCP antibody responses in CIA reported by others [14
]. Following development of arthritis, we observed significant expansion of the ACPA responses in both acute and chronic arthritis (Figure ) and EAE (Figure ). It is likely that the pre-arthritis ACPA responses observed in CIA arise due to the local and systemic inflammation that results from immunisation with CFA, and could in part parallel the ACPA responses observed in pre-arthritis samples derived from individuals that subsequently developed RA.
Intramolecular spreading is the expansion of autoantibody responses to target additional epitopes within a polypeptide, while intermolecular spreading is expansion to epitopes on other polypeptides. Our results demonstrate extensive intramolecular and intermolecular spreading of autoantibody responses in both CIA and EAE. It is unclear whether the epitope spreading of ACPA responses observed in acute and chronic CIA and EAE are responsible for the progression of disease or if they arise secondary to joint inflammation. It is possible that joint inflammation results in the generation of citrullinated epitopes, that then act as neoantigens to induce expansion of ACPA responses, that in turn perpetuate arthritis.
To further characterise citrullinated proteins present in the joint and brain tissue under attack in CIA and EAE, we performed immunoblotting and mass spectrometry analysis. We identified multiple previously undescribed citrullinated proteins in joint tissue derived from mice with CIA and brain tissue derived from mice with EAE (Tables and ). The smear observed in lanes containing the CIA joint protein lysates was most probably due to heavy glycosylation of multiple synovial proteins, and the smearing limited our ability to excise discrete bands for mass spectrometry analysis. Autoantibody targeting of citrullinated antigens in established and advanced, but not acute, disease suggests that as previously hypothesised inflammation of target tissues could result in the aberrant citrullination of multiple proteins, thereby generating neoantigens that provoke autoreactive B cell responses.
Epitope spreading of autoreactive T and B cell responses has been previously described in the EAE model [20
], whereas few researchers have described expansion of autoreactive B cell responses in animal models for RA [42
]. Our synovial protein array analysis provides further evidence of and insights into epitope spreading in CIA (Figure ). Our results indicate that epitope spreading in CIA includes development of autoantibody responses targeting citrullinated epitopes at the pre-boost (pre-arthritis) timepoint, and that there is significant expansion of autoantibody responses to target a panel of citrullinated polypeptides in both acute and chronic arthritis. Such anti-citrulline antibody responses could result in more severe arthritis [14
Epitope spreading of autoantibody responses has been demonstrated to be associated with the development and progression of autoimmune diabetes [44
], multiple sclerosis [45
] and systemic lupus erythematosus [46
], and it remains to be determined if epitope spreading of anti-citrulline autoantibody responses is also associated with the development and/or clinical progression of RA. The purpose of these studies was to characterise the development and evolution of ACPA responses in rodent models of RA and MS, to gain insights into the aetiology and potential pathogenic role of such responses in human RA and MS. Although anti-citrulline responses pre-date clinical arthritis by years in many RA patients, there are several lines of evidence suggesting that such responses can evolve over time in a subset of RA patients: a subset of new-onset RA patients are CCP negative at the time of diagnosis and subsequently develop anti-CCP antibodies [47
]; data have suggested that anti-CCP responses evolve both in terms of isotype usage [47
] as well as epitope specificity (Robinson laboratory, unpublished data). As a result, the characterisation of the development and expansion of anti-citrulline responses in rodent models of RA and MS could provide insights into a process that may also occur in human RA. It will be important to further characterise the evolution of ACPA responses in human RA, and to determine if the evolution and/or expansion of ACPA responses is associated with the persistence and severity of RA.
Citrullination exists in a variety of tissues and circumstances that are not specifically related to autoimmunity, including normal development of the skin and myelin sheath [4
], as well as conditions related to inflammation [48
]. As a result, one cannot be sure to what extent the observed effects are specific for autoimmune disease or merely arise secondary to the generation of citrullinated epitopes as a result of inflammation. Further, our results suggest that local or systemic inflammation results in the generation of anti-citrullinated protein responses that react with citrullinated proteins generated in other tissues. This possibility is supported by the reactivity of human RA sera with not only citrullinated proteins generated in inflamed joint tissue but also with citrullinated filaggrin, a protein expressed in stratified epithelium but not in joints [4
It is interesting that many of the citrullinated proteins identified in inflamed CIA joint tissue and EAE brain tissue (Tables and ) are either structural proteins or enzymes involved in common cellular metabolic pathways. The originally described citrullinated proteins were structural proteins and included fibrinogen, vimentin, keratin, filaggrin and MBP. Citrullination is known to alter the structural properties of keratin and MBP, and thus citrullination may play an important role in modulating the structural properties of such proteins [4
]. In our study, immunoblotting and mass spectrometry analysis identified citrullinated tropomyosin 1a and actin in CIA joint tissue, as well as citrullinated tubulin b 2c in EAE brain tissue. To this end, perhaps other structural proteins become accessible to and substrates of PAD in inflamed tissues, such as the joints in CIA and RA, and the myelin sheath in EAE and MS.
The observation of multiple citrullinated metabolic enzymes in inflamed CIA joint and EAE brain tissue is intriguing. Citrullinated enzymes identified included ATP synthase b chain in CIA joint tissue, as well as citrullinated ATP synthase, tyrosine 3 monooxygenase, maltate dehydrogenase, glyceraldehyde-3-phosphate dehydrogenase, phosphoglycerate kinase I and others in EAE brain tissue. Mathis and Benoist demonstrated that autoantibodies targeting the ubiquitously expressed glycolytic enzyme GPI mediate inflammatory arthritis in the K/BxN model [33
]. Based on this observation, citrullination of metabolic enzymes could result in such enzymes becoming targets of the ACPA response, and perhaps such responses could result in the formation of immune complexes that would contribute to inflammatory arthritis based on mechanisms analogous to those described by Mathis and Benoist for anti-GPI antibodies. To date, these enzymes have not been described as candidate autoantigens in RA or MS, and their potential role in human disease is unclear.
The synovial and myelin microarrays contained predominantly mouse and/or human versions of the proteins and peptides, depending on their availability from commercial sources, collaborators or the sequence originally used to synthesise peptides (see Additional files 1
). Many of the polypeptide sequences are highly conserved between species, and thus most protein and peptides provide utility for screening of autoantibody reactivity across species. It is likely that use of an array composed entirely of murine proteins and sequences would further accentuate the findings and results we describe herein.
These experiments were performed using an anti-IgM/G secondary antibody, and as a result these experiments did not discriminate IgG versus IgM autoantibody reactivity. IgM and IgG antibodies have fundamentally different properties, including affinity and avidity, ability to activate the complement cascade, and ability to bind Fc receptors and thereby activate effector cells. Further, some IgM antibodies are natural antibodies, while IgG antibodies are produced through adaptive immune responses that result in B cell isotype class switching. The anti-citrullinated fibrinogen antibody that exacerbated anti-collagen antibody-induced arthritis was of the IgM isotype [14
], suggesting that anti-citrulline IgM antibodies can exacerbate arthritis in certain rodent models. An important future direction will be characterisation of the isotypes (IgM, IgG and IgG subclasses) of the initial and expanding autoantibody responses in rodent models, as well as human RA and MS.
Autoimmune responses are well established to target both native and linear epitopes present in autoantigens [20
]. It is likely that autoantibody targeting of native CII epitopes is particularly important in CIA, as evidenced by the observation that CIA is only inducible with whole CII or certain large fragments of CII. CIA is not inducible with peptides derived from CII, which differentiates it from EAE in which short peptides derived from multiple myelin antigens induce disease in multiple mouse strains. Although low level antibody responses are observed against several CII peptides (CII 194–213, CII 224–243, CII 354–373, CII 444–463 and others), the strongest anti-CII response is observed against native CII and probably reflects the importance of targeting native CII epitopes in the pathogenesis of CIA.