Rheumatoid arthritis (RA) is an autoimmune disease affecting ∼0.5–1% of the adult population (1). The main characteristic of RA is erosive inflammation in the peripheral joints. However, the early presence of rheumatoid factors (RF) (2) and anti–citrullinated protein antibodies (3, 4) demonstrates that the disease process starts several years before clinical onset. The stage at which the autoimmune response spreads to cartilage antigens, such as type II collagen (CII) and type XI collagen (CXI), in the peripheral joints needs to be evaluated, as it is possible that this stage marks the clinical onset and relapses of the disease (5, 6). The notion that autoreactive type II major histocompatibility complex (MHC)–restricted T cells have an important function in the pathogenesis of RA is supported by their presence in inflamed synovial tissue as well as by the linkage between RA and specific class II MHC genes. In humans, this linkage is predominantly to 3 allelic subtypes of the DR4 haplotype: HLA–DRB1*0401 (Dw4), DRB1*0404 (Dw14), and DRB1*0405 (Dw15) (7–9). However, understanding the pathogenic role of these T cells has been difficult, and therefore animal models, offering genetic homogeneity and control over environmental factors, have been valuable tools.
The most commonly used model in mice is collagen-induced arthritis (CIA), induced through immunization with CII (10), which restricts the adaptive immune response of importance in arthritis. An alternative model in the rat is pristane-induced arthritis (PIA) (11). Pristane is an immunogenic but non-antigenic hydrocarbon and a highly active adjuvant component of the mineral oil in Freund's adjuvant (12). Rats injected with pristane develop a severe and chronic polyarthritis that shares several features with human RA, e.g., symmetry of inflamed joints, increased levels of RF, an early elevated acute-phase response, and an erosive disease state (13). PIA is different from both classic adjuvant-induced arthritis and oil-induced arthritis in that it has a relapsing–remitting chronic disease course (11). The facts that PIA has a chronic phase and that it is induced without use of exogenously administered proteins make this model particularly suitable for identifying autoantigens. A number of quantitative trait loci (QTLs) across the genome have been shown to regulate arthritis incidence and severity in PIA and a majority of these operate via mechanisms similar to those identified in CIA, indicating that some common pathways are critical for arthritis development in both models (14, 15). In addition, PIA has been associated with the MHC region, but a direct association with and involvement of the class II MHC genes has not yet been demonstrated (16, 17).
We have recently shown that pristane injected rats develop immunity to the ubiquitously expressed RNA binding protein heterogeneous nuclear RNP A2 (hnRNP-A2 [RA33]), which also was found to be substantially up-regulated in inflamed synovial tissue in both PIA and RA (18). This demonstrated that pristane triggers adaptive responses to self antigens, but whether this also includes cartilage-derived antigens was not yet known. In the present investigation, therefore, the immune responses to 2 cartilage-specific proteins with known arthritogenicity in the rat, CII and CXI, were studied during the acute and chronic phases of PIA and analyzed for class II MHC association in MHC-congenic DA rats. During the acute stage of PIA, both DA and MHC-congenic DA.1F rats exhibited increased but weak T cell and antibody responses to CII. At the chronic stage, the immune response to CII was down-regulated in DA rats, whereas DA.1F rats developed a strong response to CXI. A specific CXI immune response was also observed in patients with recently diagnosed RA.
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
In the present study, we demonstrated a class II MHC association with immune response to the joint-specific cartilage protein CXI, which contributes to the development of chronic arthritis in rats injected with pristane. The response to CXI likely plays a pathogenic role, as immunization with CXI leads to arthritis with the same type of class II MHC association.
Pristane-induced arthritis in the DA rat is a self-perpetuating chronic disease that is independent of exogenously administered antigens. The mechanisms involved in the development of PIA are unknown, but the rapid expansion of polyclonal CD4+ T cells in draining lymph nodes (29) suggests that pristane, possibly through engagement of innate components (30), promotes reactivity with a broad spectrum of autoantigens. As in RA, neither the source nor the identity of the vast majority of these autoantigens has been identified. However, knowing how T cell reactivity develops in response to autologous proteins in PIA may benefit our understanding of the T cell pathology in this model as well as in RA. In recent years there has been increasing interest in immune responses to non–tissue specific antigens in RA, such as α-enolase and fibrinogen (31, 32). Although the frequency and specificity of these responses may suggest that they contribute to RA, a direct link to chronic inflammation of the joints has been difficult to establish.
To better demonstrate that immune response to cartilage proteins plays an important pathogenic role in arthritis, we determined the antigenicity of 2 cartilage-specific joint proteins, CII and CXI, in the early and later stages of PIA. Both collagens are arthritogenic in rats, as demonstrated in earlier studies addressing the MHC association with CII and CXI in LEW rats (20). In the present study we used minimal congenic fragments introgressed on an arthritis-susceptible DA background to investigate the association of 2 different class II MHC alleles with CII- and CXI-induced arthritis. The results refine our understanding of the genetic control of collagen immunity, as they reduce the associated gene region of interest to <0.23 Mb and only 8–11 genes. Sequence analysis of the 4 class II MHC genes and alignment of the peptide-binding domains further suggests that CIA susceptibility is controlled by RT1.B genes, the equivalents to HLA–DQ in humans.
The MHC association with CII- and CXI-induced arthritides appears to be similar in the LEW and DA congenic rat strains (20), although the severe chronic arthritis that developed after immunization of LEW.1F rats with CXI was not observed in DA.1F rats. Nevertheless, we found that levels of anti-CXI antibodies increased over time in rats immunized with CXI. In contrast, the levels of CII-specific antibodies significantly declined after severe arthritis had developed in animals immunized with CII. The continuous increase in CXI reactivity might be related to the architecture of the healthy, intact cartilage, in which CXI is present only in low abundance and is concealed within a core of CII fibrils, where it is protected against direct antibody recognition (33, 34). The sequential antibody response to these collagens might therefore reflect a process in which antibody recognition of CII is followed by a second step of antibodies specific to CXI. It has indeed been suggested that CII-specific antibodies can affect cartilage stability by interfering with the fibrillogenesis of the chondrocytes (35). Increasing serum reactivity with CXI, as shown in this study, might therefore represent increasing destruction of the collagen fibrils in the peripheral joints. It is important to stress, however, that the pathogenic role of these antibodies was not addressed in this study, and wild-type DA rats immunized with CXI also displayed higher levels of anti-CXI antibodies during the chronic stage of arthritis.
A major result from the current study was the characterization of collagen reactivity at various stages of PIA. This model is most likely closer to RA in terms of polyclonality and antigen complexity than is CIA, and a strong immune response to collagen was not expected. However, pristane-primed T cells did in fact react with CII when arthritic rats were challenged with the antigen on day 25 postimmunization, although the response was not limited to a specific MHC haplotype. The specificity was instead directed toward the type of antigen, as demonstrated by the comparatively stronger response to CII than to CXI in arthritic DA.1F rats, providing further evidence of a lack of CXI immunity in early arthritis.
Most importantly, we showed that CXI immunity is confined to animals with chronic arthritis. In addition, we demonstrated that the immune response to CXI in chronic PIA is clearly class II MHC associated, with an RT1f haplotype. The chronic response to CXI is consistent with the increased levels of anti-CXI antibodies at the later stage of CXIIA. An increased antibody response to CXI and a subsequent decrease in the response to CII were observed in chronic PIA. The change in immune response to CXI in chronic arthritis likely reflects a change in antigen availability, as discussed above, rather than an epitope spreading in the context of posttranslational modifications on the shared α3-chain, as no response was detected in animals challenged with purified α3 antigen. The weak response to CII and lack of specific class II MHC restriction in the CII response during early arthritis compared to the relatively strong response to CXI with a clear class II MHC association in chronic arthritis further supports this notion.
The translation of basic research findings to the clinic is not always uncomplicated, and the human relevance of data obtained in animal studies is often questioned. Comparing the autoantibody response to CXI and CII in 356 RA patients, we showed that CXI was recognized by 18% of the patients and that this response was distinct from the CII-specific response in 12% of the patients. This is in accordance with our preclinical data in PIA and highlights the utility of this antigen-independent model for identifying new autoantigens in arthritis. Studies of a larger patient cohort, preferentially with RA at a later stage, will be necessary to evaluate the MHC association with CXI seen in PIA.
In recent years there has been increasing focus on the role of ubiquitous, often modified, antigens in RA. Although these antigens provide attractive tools as biomarkers for predicting later development of RA, their possible pathogenicity remains to be demonstrated. Meanwhile, important and undoubtedly arthritogenic articular collagens, such as type XI, may have been overlooked.
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Holmdahl had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Tuncel, Carlsén, Holmdahl.
Acquisition of data. Tuncel, Haag, Yau, Lu, Burkhardt.
Analysis and interpretation of data. Tuncel, Haag, Carlsén, Yau, Lu, Burkhardt, Holmdahl.