The discovery of high titers of rheumatoid factor (RF) in sera from patients with rheumatoid arthritis (RA) (1) was the beginning of a new era of molecular research on the disease (2, 3). This research not only represented a new scientific approach to RA, but also contributed to deeper understanding of B cell immunology, in particular, the role of autoantibodies in various diseases. Increased RF titer was incorporated as one of the criteria for the classification of RA, and RF was for a long time the only serum antibody that was helpful in the diagnosis. The sensitivity and specificity of RF for identification of patients with RA were, however, relatively low compared with more recently discovered autoantibodies, such as antibodies to citrullinated proteins (4). Nevertheless, it was anticipated that RF might play a pathogenic role and be a key factor in the regulation of the disease (5, 6).
It was observed that RF of both the IgG and the IgM isotypes, found in the joints of patients with active arthritis (7–9), was produced by plasma B cells within synovial germinal centers in inflamed tissue (10, 11). Importantly, it could be shown that RF in many cases was produced by B cells that had undergone an antigen-driven somatic mutation process, as demonstrated both in humans (12) and in sophisticated experimental models of the disease in animals (13). RF predominantly recognizes epitopes located in the Fc region of the IgG molecule, especially the CH2 and CH3 domains (14, 15), but also human β2-microglobulin (16, 17). However, molecular studies have not identified the T cell– and B cell–recognizing antigens that drive the process.
RF is produced also in healthy individuals and likely has important physiologic functional roles, such as clearance and transportation of immune complexes by crosslinking of IgG in complex with antigen (18–21) and enhancement of antigen capture for antigen-presenting cells. The possible pathogenic effects of overproduced RF might be related to such functions, but this has been difficult to demonstrate directly in RA or in animal models.
Several studies show, however, that RF is a predisposing factor in the development of RA (22–24). Increased levels of RF in symptom-free individuals greatly increase the risk of acquiring RA (25), and high titers predict continuing severity of radiographic damage in inflammatory polyarthritis (26). It has also been found that levels of RF were increased in relatives of RA patients (27), providing evidence of independent genetic control of RF production. Thus, identification of the genes controlling RF production could be helpful in understanding their role in the pathogenesis of RA. However, although this approach seems simple and reasonable, it has met with severe problems. RA is a complex disease affecting a large portion of the population and is influenced by both unknown environmental factors and many genes (28–30). The genetic influence, especially involving the class II major histocompatibility complex (MHC) DR4 haplotype (31, 32), is illustrated by studies demonstrating higher rates of disease concordance in monozygotic twins (33). Involvement of non-MHC regions has been suggested by the results of genome-wide scanning (34, 35), but studies are complicated by genetic heterogeneity, variable penetrance, and poorly defined disease phenotypes.
An alternative approach that avoids many of the difficulties in human genetic studies is to use animal models, which provide better control of genetic heterogeneity and environmental influences. Experimental animal models such as collagen-induced arthritis (CIA) (36), proteoglycan-induced arthritis (37), and pristane-induced arthritis (PIA) (38) have been used to identify several disease loci, among which many are shared between strains and between disease models (39–43). The question of relevance to the human disease process is certainly an issue with animal studies, but their use can be justified if similar pathogenic pathways are used in the different species. RF production is possibly an indicator of some important pathways of RA, but unfortunately the production of similar types of RF has not been reported to occur in animal models. In the most commonly used model, CIA in mice, RFs are produced as self-associating IgG factors detectable only with diffusion in gel enzyme-linked immunosorbent assays (ELISAs) (44). We have now found that IgM- as well as IgG-RF production, measured by a direct ELISA method as used for clinical assays, is readily detectable in certain rat strains, e.g., the E3 and DA strains, and occurs during the development of PIA. This has enabled us to investigate the genetic control of RF production and to compare this with the genetic control of arthritis.
- Top of page
- MATERIALS AND METHODS
In this study we investigated the roles of antibody production in pristane-induced arthritis. Blood samples were obtained at 5 different time points and the sera were analyzed, focusing on rheumatoid factor production and total immunoglobulin levels. A cohort of 153 F2 animals of either the pristane-susceptible DA strain or the E3 strain, which is completely resistant to PIA, was investigated using genome-wide linkage techniques with 288 informative microsatellites. We identified 13 different regions controlling antibody responses, of which 5 loci colocalized with previously defined arthritis loci (PIA loci). The E3 × DA cross has been studied thoroughly, and several disease linkages have been described (Pia1–Pia9) (40, 52). Shared PIA loci for antibody responses were found for Pia4, Pia5, Pia6, Pia7, and Pia8. Linkages to previously defined nonarthritis loci were also detected, including the EAE loci Eae7 and Eae8 (54) and the Apr1 locus, an acute-phase protein response locus in the E3 × DA cross (53). Associations with disease loci were established for total immunoglobulin as well as for RF and isotype responses. Total IgG and IgM shared chromosome regions with Pia5, Pia7, and Pia8. RF associated with chromosomes 4 (Pia5), 11, and 16. Isotype responses colocalized with Pia4, Pia5, Pia6, Cia2, Apr1, Eae7, and Eae8.
Several investigations implicate RF as a predisposing factor in arthritis development and severity. Although this has been debated, recent studies clearly show a high correlation between early findings of high-titer RF and later RA disease severity (22–24). The mechanism of persistent production of RF is, however, not known. Elevated RF titers can also be detected in other rheumatic disorders, after infections, and in elderly healthy individuals (56, 57). The RF production following inflammation, which depends on the continuous presence of infectious antigen, can be abolished by antibiotic treatment (58, 59). The pathologic RF involved in RA has been shown to differ from normal anti-IgG in specificity and mutation rates (60, 61). The V-gene utilization in RF is diverse, and the number of somatic mutations indicates an antigen-driven response (7, 62). However, nonmutated RF has also been described, suggesting the possibility that low-affinity RF could be produced directly from the germline repertoire (63). Complement proteins can be produced locally in the synovial tissue (64), and activation of the complement system can cause inflammation by cytokine release following ligation of Fcγ receptors on macrophages (65). Complement has, in addition, been shown to have a role in the induction of the humoral immune response against T cell–dependent antigens (66, 67).
In spite of the identification of several antibody-associated linkages to disease loci, only IgM-RF (day 35) could be shown to correlate with disease severity, expressed as maximum score sum (r = 0.297 by regression analysis). The time of maximum disease severity was found to be close to day 35 (day 32). We could not detect any correlations with RF titers early in the disease course. Investigation of the levels of antibody titers in the parental E3 and DA rats showed that the disease-resistant E3 animals developed, both initially and constantly, higher titers of RF antibodies than the DA rats. By genetically mapping the IgM-RF phenotypes, we demonstrated an E3 allele–dependent linkage to chromosome 11 (Rf1), which was highly significant and unique for the IgM-RF isotype. The facts that E3 rats are resistant to PIA and that the linkage to Rf1 is E3 dependent raise questions about the role of RF in arthritis. It is likely that the major genetic influence was not detected in this study; these loci could have remained undetected due to lower penetrance or genetic interactions. It is also possible that the genetic associations with arthritis by the detected RF loci have not been fully identified. A possible means to clarify this would be to establish congenic strains and to identify the underlying genes explaining the RF loci, and subsequently to directly investigate their role in arthritis.
It is interesting to note that the Igλ locus is located within the Rf1 QTL, and a possible explanation could be that the E3 rat expresses a unique Vλ gene used for RF production. Indeed, the RFs seem to be predominantly of the lambda type, but conclusive evidence for this hypothesis will require positional cloning of the gene. This will take some time since the region needs to be isolated in a minimal, congenic fragment as has been shown to be essential for the positional cloning of the Ncf1 gene in the Pia4 locus (55). There are indeed other genes and mechanisms that could also explain the increased RF production by the Rf1 locus and in the E3 strain. Several studies have shown the importance of RF in complement binding and clearance of immune complexes (18–21, 25, 26). One possible explanation would be a defect in complement binding or activation in the E3 rat. This would, if complement activation contributes to disease development, explain the fact that E3 rats do not develop arthritis. Furthermore, the consistently higher levels of IgM-RF and IgG-RF in these rats could indicate unregulated production secondary to the inflammation process.
Chromosome 4 displayed several interesting linkages. Associations with the Pia5 locus, identified for inflammation score, were found for RF as well as for total IgG and IgG2a isotypes. Comparison of the allelic influences on disease and antibody responses revealed different inheritance origins. Apparently, E3 genes control the immunoglobulin responses, whereas disease is influenced by DA genes. Other studies have shown linkages to chromosome 4 in autoimmune thyroiditis (68) and adjuvant arthritis (69). Interestingly, it has been shown that the homologous mouse region, on chromosome 6, harbors genes controlling antibody responses (70). Another linkage to rat chromosome 4, total IgG levels (day 100), colocalized to the Pia7 locus, originally defined from a 2-locus interaction model for clinical score (day 19) in E3 × DA and DA × DXEC crosses (52). The region also harbors major loci controlling oil-induced arthritis in the DA × LEW.1A cross (71) and CIA in DA × BN rats (72). Taken together, these findings indicate that chromosome 4 contains important genes that regulate several forms of autoimmune disease.
Another difference in inheritance pattern was seen for the Pia8 locus, previously established for maximum clinical score in the DA × DXEC cross (E3 dominant, females) (52). In the present study, total IgM (day 14) linked in a DA-additive mode. The Pia4 locus, originally described for arthritis severity in E3 × DA animals and now identified as the Ncf1 gene (55), and Pia6 (chronic arthritis) were shared for isotype responses (IgG1 and IgG2a). An early phenotype linkage for interleukin-6 (IL-6) (day14) to Pia6 has also been reported (53), indicating a possible role of IL-6 in the B cell response. The IgG1 linkage to Pia4 could indicate involvement of the complement system; in mice, small RF-like immune complexes induce an IgG1-RF response (67).
In summary, our findings indicate that antibody responses in pristane-induced arthritis are under genetic control and contribute to disease development. The cosegregation of several of the antibody-controlling genes with arthritis loci indicates a pathogenic connection. However, although E3 rats are genetically resistant to arthritis, they have higher RF levels, and some E3 genes dominantly control the RF production. Interestingly, other loci operate in the opposite direction. The identification of the responsible genes, and their interactions, for both arthritis association and RF production will be of critical importance, and the present study identifies a platform for this work. The genetic control of arthritis and RF production is likely equally as complex in humans, and there is a clear need to identify the basic biologic pathways operating, in order to explain these phenomena.