Rodent models of arthritis: relevance for human disease

Rheumatoid arthritis (RA) is a chronic disabling disease affecting around 1% of the population. Although much has been learnt in recent years about the mediators that drive the pathology in RA, there is still a lack of knowledge of the underlying causes of the disease. It is for this reason, together with the need for more effective and less toxic remedies, that animal models of arthritis are being studied. Rodent models are used in a wide variety of different studies, including the testing of novel therapies, the identification of proinflammatory mediators, the analysis of genetic susceptibility factors and in the search for markers of disease progression. The first model of RA to be described was adjuvant arthritis, which can be induced in rats by a single injection of Freund's adjuvant, containing Mycobacterium tuberculosis [1]. Arthritis develops at around 10–45 days after injection and generally subsides after a month. The chief pathological features of adjuvant arthritis include oedema, infiltration into the joint of mononuclear and polymorphonuclear cells, pannus formation, periostitis and erosion of cartilage and bone. Although a link between immunity to 65-kD heat shock proteins (hsp) and the induction of adjuvant arthritis is suspected [2], no single mycobacterial immunogen has been identified which is responsible for the arthritogenic response in this model [3]. Rather, the induction of adjuvant arthritis has been attributed to a mycobacterial cell wall component, muramyl dipeptide, which is immunostimulatory but does not evoke a specific immune response [4]. In addition, a number of adjuvants which lack immunogenic properties have been shown to induce arthritis in susceptible strains of rats, including avridine [5], Freund's incomplete adjuvant and pristane [3]. An arthritis, bearing many similarities to RA, is also observed in mice following administration of pristane [6]. Doubts as to the immunological nature of these diseases were dispelled by the findings that (i) anti-T cell treatments prevent the induction of arthritis [3,7], (ii) susceptibility is influenced by genes within the MHC [8,9], and (iii) arthritis can be adoptively transferred by T cells [10,11]. The mechanism of arthritis induction following immunization with adjuvants is unknown, but one possibility is that following immunization there is an increase in the activity of antigen-presenting cells (APC) [12], leading to the presentation to autoreactive T cells of a hitherto unrecognized or ‘sequestered’ endogenous antigen. The possibility that human RA could also be triggered by exposure to environmental factors with adjuvant-like activity has been highlighted by recent studies in which it was found that arthritis could be induced in DA rats by percutaneous exposure to adjuvant oils [13] or even a mineral oil-containing cosmetic product [14]. Antigen-induced arthritis (AIA) is seen in mice, rats and rabbits following intra-articular injection of protein antigen (e.g. methylated bovine serum albumin) into the knee joints of animals that have been previously immunized with the same antigen [15,16]. The histopathological appearance of AIA bears similarities to RA, including synovial lining layer hyperplasia, perivascular infiltration with lymphocytes and plasma cells, lymphoid follicles, pannus and cartilage erosions. However, unlike RA, AIA is a monoarticular disease that affects only injected joints. Susceptibility to AIA is not MHC class II-restricted and this makes the model useful for studies involving transgenic and gene knock-out mice. For example, Busso et al. recently studied the evolution of AIA in urokinase gene knock-out mice in comparison with wild-type mice and were able to demonstrate a role for fibrin in the maintenance of chronic inflammation [17]. A single i.p. injection of an aqueous suspension of sonicated streptococcal cell wall (SCW) has been shown to cause chronic arthritis in rats and mice. Pathological changes of relevance to RA include infiltration of polymorphonuclear cells, CD4+ T cells and macrophages, hyperplasia of the synovial lining layer, pannus formation and erosion of cartilage and bone. Susceptibility to SCW-induced arthritis varies between strains. For example, Lewis (LEW/N) rats develop severe chronic disease, whereas histocompatible Fischer (F344/N) rats develop mild arthritis that rapidly subsides. This difference in susceptibility between the two strains has been attributed to a defect in the synthesis of corticotropin-releasing factor in the Lewis rat leading to suboptimal activation of the hypothalamic–pituitary–adrenal axis [18,19]. Collagen-induced arthritis (CIA) has been described in rats, mice and primates following immunization with type II collagen [20–22]. The pathological changes include synovitis with infiltration of polymorphonuclear and mononuclear cells, pannus formation, erosion of bone and cartilage, and fibrosis. In mice, immunization with bovine, chick or rat type II collagen usually leads to a relatively acute form of arthritis. However, immunization with self collagen results in a more protracted disease course [23,24] and the chief determinant of chronicity in CIA is likely to be the extent to which the immune response is targeted at self collagen, as opposed to the collagen used for immunization. This is supported by the observation that transgenic DBA/1 mice, which over-express the TCR β gene from a T cell clone that recognizes mouse type II collagen, develop chronic arthritis whilst their non-transgenic littermates develop a self-limiting form of the disease [25]. Susceptibility to CIA is restricted to mouse strains bearing MHC types I-Aq and I-Ar [26,27] and this is analogous to human RA, where certain subtypes of DR4 and DR1 are strongly associated with susceptibility to the disease [28]. By scanning overlapping peptides of mouse type II collagen, Bayrak et al. were able to identify multiple epitopes recognized by T cells from collagen-immunized mice of the H-2q haplotype [29]. A comparison of these epitopes revealed a characteristic sequence motif which was found, using a computer model, to be accommodated in the binding groove of the I-Aq molecule. A finding of potential significance for human RA was that the binding motif of I-Aq was similar to the binding motifs of the subtypes of DR4 that predispose for RA, suggesting that similar epitopes of type II collagen may be involved in RA and CIA. Convincing data have not yet emerged pointing definitively to a role for collagen autoimmunity in the bulk of RA patients. This has prompted the search for other potential joint antigens which may be the target of the autoimmune response in RA. For example, a T cell-driven arthritis has been described in mice following immunization with the G1 domain of human proteoglycan aggrecan [30]. Histopathological changes include oedema, proliferative synovitis, infiltration of mononuclear cells, pannus formation and erosion of cartilage and bone. A novel model of arthritis, induced by immunization of rats with homologous cartilage oligomeric matrix protein (COMP), is described in this issue [31]. Like CIA, susceptibility to COMP-induced arthritis is controlled by genes within the MHC, with the RT1u and RT1l haplotypes showing the greatest degree of susceptibility. The arthritis is characterized by synovial hyperplasia and hypertrophy accompanied by pannus and, in some strains, joint erosion. Like type II collagen and aggrecan, COMP is a product of chondrocytes but represents a relatively minor fraction of the extracellular matrix of cartilage. The authors argue that under natural conditions a major cartilage protein, like type II collagen, would probably induce a strong degree of central and/or peripheral tolerance, and would therefore be an unlikely target for the autoimmune response [31]. Thus, in the search for potential autoantigens in RA, minor cartilage proteins, like COMP, should not be overlooked. The application of transgenic technology to the field of cytokine biology has generated a number of new models of arthritis. For example, spontaneous arthritis involving severe erosive changes has been described in transgenic mice over-expressing the human tumour necrosis factor-alpha (TNF-α) gene [32]. The occurrence of arthritis in these mice is perhaps not surprising, in the light of what is now known about the role played by TNF-α in the pathogenesis of arthritis [33]. However, a less expected finding was that treatment of TNF-α transgenic mice with an IL-1α/β blocking antibody completely prevented the development of arthritis [34]. This finding parallels studies in human RA synovial cell cultures, where blockade of TNF-α was found to diminish IL-1β production [35], and indicates that IL-1 is acting as a major downstream mediator of joint pathology. In conclusion, arthritis can be induced in rodents in a variety of ways, and all of the models described above mimic human disease in some respects. An important lesson to be learnt from our experience with animal models of arthritis is that where there is a single eliciting immunogen, disease susceptibility is likely to be influenced by MHC class II genes. The strong association between RA and genes encoding DRβ chain may therefore be regarded as evidence pointing to the existence of an autoantigen in RA. The identification of such an antigen would greatly amplify the scope for therapy based on selective modulation of the immune response.