Characterization of a novel and spontaneous mouse model of inflammatory arthritis

Rheumatoid arthritis (RA) is a systemic and chronic disease. The most characteristic symptoms are severe synovial inflammation, cartilage and/or bone destruction and bony nodule formation in the diarthrodial joints, but extraarticular manifestations are also prevalent and include pericarditis and vasculitis as well as pulmonary complications such as pleuritis and pulmonary fibrosis [1‐3]. The worldwide prevalence of RA among adults is estimated to be between 0.3% and 2%, with more women affected than men [4‐6]. Although the prevalence of RA does increase with age, the majority of adults with RA are diagnosed between the ages of 30 and 50 years. A juvenile form of disease, with onset before 16 years of age, does exist, and while differences in classification and diagnosis make estimates difficult, the prevalence of RA among children likely ranges between 0.007% and 0.4% worldwide [7]. RA is a debilitating disease associated with increased mortality and a decrease in survival by 3 to 10 years [8]. It is also a major cause of inability to work, with a little over one-third of RA patients reporting an inability to work within five years of diagnosis [9].

Unlike many other forms of arthritis, RA is autoimmune in nature. Autoantibodies are found in the vast majority of patients. While B-cell autoantigens include collagen type II, citrullinated proteins and glucose-6-phosphate isomerase, the classic example is rheumatoid factor (RF), an autoantibody specific for the Fc portion of immunoglobulin G (IgG) [10]. RF is found in 70% to 80% of adult RA patients and has been used as a diagnostic indicator for the past five decades [11]. While B cells have been successful targets for therapy in at least a subset of patients [10, 12, 13], the focus has shifted more toward T cells as the primary lymphocyte driving disease [11, 14‐17]. Drugs that limit T-cell activation help to ameliorate disease, and many of the genes associated with an increased susceptibility to RA are involved in T-cell function [11, 18]. Though RA was once described as "Th1-driven," evidence is mounting that Th17 cells are the primary T-helper cell subset promoting disease [16, 17, 19].

While lymphocytes account for the autoreactivity in RA, the inflammation itself results from a plethora of cytokines that recruit peripheral immune cells, promote synovial proliferation and induce osteoclast maturation to increase bone resorption [11]. Many proinflammatory cytokines, such as TNF-α and IL-6, have been the targets of therapeutic biologic drugs that successfully limit inflammation [20]. Secreted primarily by activated macrophages, TNF-α is believed to have a particularly central role, as it induces the release of many other cytokines and chemokines and activates many of the cell types present in the inflamed synovium, including polymorphonuclear cells, endothelial cells, chondrocytes and osteoclasts [21]. IL-6 is similarly known for its pleiotropic functions [22]. It is the primary cytokine responsible for the acute phase response observed in RA, contributing to systemic inflammation [21], and strongly influences lymphocyte maturation, promoting both plasma cell and Th17 differentiation [23, 24].

Mouse models of RA have been critical for understanding the underlying genetic and cellular mechanisms of pathogenesis [25‐27]. Both induced and spontaneous models have been developed. In the most common experimentally induced models, injection of adjuvant and various joint matrix components, such as collagen type II or proteoglycan aggrecan, causes autoimmune arthritis in susceptible inbred mouse strains [28, 29]. Spontaneous models can be subdivided by their mode of origination: (1) development of autoimmune-prone strains by selective mixing of previously existing inbred strains (for example, the MRL/lpr strain [30, 31]), (2) targeted gene manipulation (for example, T-cell receptor (TCR)-transgenic K/BxN model [32], TNF-α overexpression models [33], IL-1Ra-knockout [34] and gp130^Y759F-induced mutant [35]) and (3) identification of spontaneous mutants from breeding colonies (for example, SKG mice with a point mutation in Zap-70 [36]). While these models replicate many aspects of human RA, none of them mimic the disease completely. In fact, it is unlikely that RA itself is a single disease, but instead may be subdivided into various subsets (for example, RF-seropositive and RF-seronegative) with distinct etiologies. Therefore, each mouse model is potentially important not only for reinforcing the existence of common or even universal mechanisms of disease but also for revealing unique aspects not yet reflected in the other models.

In this report, we describe a new mouse line that spontaneously develops chronic inflammatory arthritis most evident in the large distal joints. Called the IIJ (inherited inflamed joints) strain, it was derived from arthritic (AR) animals discovered in a breeding colony at Northwestern University. One male AR mouse of predominantly SJL/J background was used to begin the line, and, after 13 generations of inbreeding, the phenotype remains stable. RA incidence in the entire colony is approximately 33%, with females developing arthritis more often than males. Joint histology confirmed cartilage and bone erosion, synovial proliferation and robust leukocyte infiltration. In addition to the inheritance pattern and histopathology, we describe the clinical and immunological properties of disease.

In this report, we describe a new strain of mouse, which we have named the IIJ strain, that spontaneously develops chronic inflammatory, and possibly autoimmune, arthritis that shares many similarities with human RA and other mouse models of arthritis. Though visible swelling was most evident in the large distal joints of IIJ mice, histological hallmarks of arthritis, including bone and cartilage erosion, invasive synovial hyperproliferation and immune cell infiltration were observed in all joints of the hindlimbs and forelimbs (Figure 1). Overall, the histopathology was similar to that described in previously published mouse models of autoimmune arthritis [32, 34‐36]. In addition, the predominantly neutrophilic and lymphocytic infiltration into the inflamed IIJ joints (Figure 4) parallels the high number of neutrophils and T cells present in the inflamed synovial fluid of RA patients [48, 49]. Also similar to RA, inflammation was not restricted solely to the joints of IIJ mice, but was systemic in nature. Elevation of serum proinflammatory cytokines (Figure 5) and histological signs of inflammation, ranging from mild to severe, were observed in various organs (Table 3). Extraarticular manifestations of disease have also been noted in IL-1R antagonist-deficient [50] and SKG mice [36]. Finally, the IIJ mice also share serological similarities with RA and other mouse models. Compared to NAR littermates, AR IIJ mice had mildly elevated serum Ig, which is due, at least partially, to elevated IgG1, a classic Th2-induced isotype (Figure 7). RA patients also have slightly increased serum Ig levels [51], and researchers who have studied several spontaneous mouse models of arthritis, including the K/BxN [32], gp130^Y759F [35] and IL-1R antagonist deficiency models [34], have reported increases in IgG1 specifically. Importantly, autoantibodies were also detected in our study, as many IIJ mice were positive for RF and antibodies specific to joint components (Figure 7).

The arthritis incidence in IIJ mice also displayed the sex bias common to many complex autoimmune diseases, such as RA, multiple sclerosis and systemic lupus erythematosus (SLE) [52]. RA incidence was higher in females than in males (38% vs. 28%; P < 0.001) (Table 1), though ultimately clinical severity and mean day of onset did not differ significantly. The sex bias appeared to be specific for the arthritis phenotype, since the incidence of typhlocolitis was similar between male and female IIJ mice (Table 4). As most models reach 100% incidence in both sexes, no other spontaneous mouse model of arthritis has found such a sex bias, although researchers who have studied both the SKG [36] and gp130^Y759F models [35] have noted more severe arthritis in females. A female bias in incidence was observed in collagen-induced arthritis in humanized HLA-DR4-transgenic mice [53] and was attributed to both hyperactive B cells and HLA-DR4-restricted antigen presentation in female mice and increased numbers of Tregs and B-regulatory cells in male mice [54]. The reason for the sex bias in arthritis incidence in IIJ mice remains unknown.

It remains unclear whether arthritis in the IIJ mice is truly autoimmune. The similarities to RA and other mouse models of autoimmune arthritis outlined above would strongly argue in favor of this scenario. The relative expansion of the B-cell population, elevated serum Ig and presence of autoantibodies are also strongly suggestive that this is true. Nevertheless, all AR IIJ mice did not have levels of serum autoantibodies that exceeded those of NAR littermates (Figure 7), and the increase might simply be a side effect of chronic inflammation. Some murine models of arthritis have shown elevated levels of Ig and autoantibodies, but the arthritis was lymphocyte-independent. For example, joint inflammation still occurs in the N-ethyl-N-nitrosourea-induced mutants Ali5 and Ali18, even when crossed to Rag1^-/- backgrounds [55, 56]. Finally, transfer of sera from AR IIJ mice failed to induce disease in SJL/J hosts, suggesting that antibodies alone are not sufficient to cause disease. It is possible that not enough serum was transferred and/or that it was not transferred by the appropriate method (for example, intravenous rather than intraperitoneal injection). Nevertheless, the same transfer protocol successfully transfers disease in antibody-dependent arthritis models such as the K/BxN model [57].

The appearance of a high percentage of DN T cells in the peripheral lymphoid organs does indicate that the T-cell compartment in the AR IIJ mice is significantly disrupted (Figure 8). DN T cells are characteristic of SLE in humans as well as in murine models of SLE, such as the MRL/lpr mouse, a strain that also gets arthritis [58, 59]. Since the peripheral DN T cells in MRL/lpr mice are B220⁺ [59] and those in IIJ mice are B220^- (data not shown), it is unlikely that these mouse populations are identical. The origin of the DN T cells in IIJ mice remains unclear. It does not appear to caused by an expansion of NK T cells, as DN T cells lack the pan-NK cell marker DX5 (Figure 8G), and their high expression of CD3 and the TCR suggests that they are mature. The fact that CD4⁺ T cells decrease in number as DN T cells increase does suggest that the DN T cells might be CD4⁺ cells that have downregulated CD4. While downregulation of TCR coreceptors by viruses have been well documented [60], it has also been proposed as a natural mechanism of peripheral tolerance [61, 62]. Autoreactive T-cell clones can downregulate CD4 upon long-term stimulation in vitro [63], and self-specific TCR-transgenic mice can downregulate coreceptors to control autoimmunity [64]. Nevertheless, the fact that the IIJ DN T cells largely lack an activated and/or memory phenotype (Figure 8B) argues against coreceptor downregulation due to chronic activation. The existence of a DN T-cell-regulatory population to control self-reactivity in humans and mice has been proposed [65, 66], and the expanded DN T-cell population in mice bearing the homozygous Fas^lpr or FasL^gld mutations has been found to have regulatory properties [67]. The ontogeny and functional ability of the IIJ DN T cells are currently under investigation.

The gene mutations and background alleles that might cause or modify the arthritis phenotype in the IIJ strain have yet to be identified. It is unlikely that the phenotype is related to either the 5B6 TCR transgenes or the Thy1.1 allele that were present in the line from which the IIJ strain was derived. A low percentage of peripheral blood T cells in the IIJ mice expressed Vβ6, the variable region used in the 5B6-transgenic TCRβ chain, and none of the IIJ mice (> 1,000 animals) have shown clinical or histological signs of EAE (for example, loss of muscle tonicity, hindlimb paralysis or inflammation in the central nervous system). Also, the arthritis occurs regardless of Thy1 status, as we have identified AR and NAR IIJ mice that are Thy1.1⁺Thy1.2^-, Thy1.1^-Thy1.2⁺ and Thy1.1⁺Thy1.2⁺ (data not shown). Both SJL/J and FVB background alleles likely contribute to arthritis development. The strain from which the IIJ mice were derived was predominantly SJL/J in background. Statistically, however, with a minimum of only six generations of backcrossing to SJL/J after generating the 5B6-transgenic mice on the FVB background, a small percentage of FVB alleles must be present in the IIJ strain. The fact that multiple attempts to backcross the AR IIJ mice further to SJL/J only decreased arthritis incidence also argues in favor of FVB-influencing alleles. Since backcrossing AR IIJ mice to FVB inbred mice also resulted in low AR incidence, SJL/J alleles must also be important.

The pedigree does indicate that the arthritis phenotype is a complex, non-Mendelian trait and, together with the female bias in incidence, is clearly sex-influenced. The incidence of 32% in the F1 generation after crossing the "founder" AR male with the SJL/J females argues for a dominant phenotype (Figure 2). Nevertheless, the fact that crossing two NAR IIJ mice yielded a similar incidence of arthritis suggests that the trait is also incompletely penetrant, since the NAR parents must harbor the disease-promoting alleles but do not display the trait. While it is also possible that a combination of multiple alleles must be inherited to develop the disease, the relative consistency of incidence regardless of the phenotype of the parents argues against simple segregation of multiple alleles. Furthermore, at the time of the writing of this article, the IIJ mice had been intercrossed for 14 generations and the overall incidence had not increased, despite the fact that most loci should be fixed after that much inbreeding. Therefore, while genes are undoubtedly influencing the phenotype, other factors, such as microbial stimulation of the innate immune system, or even stochastic events, such as the formation of the T- and B-cell repertoires through random rearrangements in the TCR and Ig genes, might also be involved.

Some bacteria, such as Borrelia burgdorferi and Staphylococcus aureus, or microbial products such as zymosan, a crude fungal extract of β-glycans, can induce arthritis in susceptible mouse strains [25‐27]. While it cannot be completely ruled out, there is little evidence that the arthritis in the IIJ mice is caused purely by an infectious agent. The mice are maintained in a barrier facility, and arthritis incidence does not cluster by cage. Also, SJL/J mice housed in the same cage as AR IIJ mice never develop arthritis, and SJL/J sentinel mice that receive portions of bedding from over 20 cages of IIJ mice housed on the same rack also never develop arthritis. The IIJ breeding colony was transported between institutions (from Northwestern University to Westmont College) in the F3 and F4 generation, but arthritis incidence did not significantly change after the move. No viral, bacterial or parasitic pathogen has been consistently found in the IIJ strain, aside from the mice with H. rodentium infection, and the AR IIJ mice that have been treated with antibiotics to clear the Helicobacter sp. have remained AR. Nevertheless, even spontaneous arthritis can be highly influenced by microbial flora. For example, the SKG strain has a high incidence of arthritis in conventional caging conditions and a very low incidence of arthritis in SPF conditions, unless they are injected with zymosan [68]. It is not known whether a similar innate immune system would trigger an increased incidence of arthritis in the IIJ mice.

While H. rodentium does not appear to cause the arthritis phenotype, it seems likely that it is responsible for or at least contributes to typhlocolitis, since H. rodentium infection is prevalent in both AR and NAR IIJ mice and treatment with antibiotics cleared the Helicobacter sp. and resolved the diarrhea and gut inflammation (Figure 9). As mentioned previously, H. rodentium was identified only 13 years ago, when it was isolated from asymptomatic laboratory mice [44]. Few studies have since investigated its pathogenicity. In experimental monoinfections, H. rodentium did not cause cecal lesions or significantly change gut gene expression in either immunocompetent A/JCr mice or SCID (severe combined immunodeficiency) mice, while, in parallel experiments, a related species (H. hepaticus) did [46]. Several studies have found that H. rodentium exacerbates gastrointestinal inflammation when coinfected with other Helicobacter spp. [45‐47], although no other Helicobacter spp. were found in the IIJ mice. If this bacterial species alone is responsible for the typhlocolitis, it might reflect a particular predisposition to proinflammatory immune responses in IIJ mice. RA as well as IBDs such as ulcerative colitis and Crohn's disease are chronic inflammatory diseases, and genome-wide association studies have identified genetic loci that increase the risk for both [69‐72]. Although several case studies have been published about RA patients who developed IBD [73, 74], IBD is not considered a typical extraarticular manifestation of RA [2, 3]. In contrast, joint inflammation occurs in 20% of patients diagnosed with IBD, though it is classified as peripheral or axial spondyloarthritis [42, 43] and is typically seronegative for RF and anticyclic citrullinated peptide antibodies [75, 76].

Overall, the data presented identify the IIJ strain as a new murine model of inflammatory, possibly autoimmune, arthritis. The IIJ strain is similar, both histologically and serologically, to RA and other murine models of autoimmune arthritis. In addition, the increased incidence of arthritis in female IIJ mice makes it a potentially important model for the study of the underlying causes of the sex bias in autoimmunity. Nevertheless, notable differences do exist between the disease in IIJ mice and human RA. For example, joint swelling in RA is typically symmetric in humans [1, 77], while it is asymmetric in many of the IIJ mice (Figure 3). Also, the accumulation of DN T cells in the periphery, which is so marked in IIJ mice (Figure 8), is a hallmark of SLE, not RA [58]. Despite these differences, the IIJ strain is a new experimental tool with which to investigate the genetic and immunological mechanisms of arthritis.