Springer Semin Immunopathol(1988) 10:215-230

Springer Seminars in Immunopathology © Springer-Verlag 1988

Rheumatoid Factors in Immune Complexes of Patients with Rheumatoid Arthritis Mart Mannik, Francis A. Nardella, and Eric H. Sasso Division of Rheumatology,Universityof Washington, Seattle, WA 98/95, USA

Introduction

Rheumatoid factors (RFs) are autoantibodies directed to the Fc region of immunglobulin G and they exist among all classes of immunoglobulins. Rheumatoid factors belonging to the major classes of immunoglobulins are designated IgG rheumatoid factors (IgG RFs), IgM rheumatoid factors (IgM RFs), and IgA rheumatoid factors (IgA RFs). Other autoantibodies to IgG are directed to areas other than the Fc region of the IgG molecules -- these autoantibodies are not termed rheumatoid factors. For example, autoantibodies have been described that do not react with intact IgG but bind to IgG molecules that have been cleaved by proteases. The best known autoantibodies in this category are the antibodies directed to the site on IgG exposed by cleavage with pepsin. The prevalence of these antibodies is increased in patients with rheumatoid arthritis (RA) and in other inflammatory conditions. Antibodies to idiotypes are directed to the variable regions of immunoglobulin polypeptide chains and combine both with the intact molecules and with Fab and F(ab')2 fragments. A discussion of autoantiidiotypic antibodies is beyond the scope of this article. The pathogenic mechanisms of RA involve both humoral and cellular immune mechanisms. Humoral immunity participates in pathogenic mechanisms through formation of immune complexes. RFs are prevalent in patients with RA as well as in many other diseases and form a variety of immune complexes as will be reviewed below. Furthermore, RFs are common constituents of immune complexes isolated from the serum or synovial fluid of patients with RA and exist in the superficial layers of articular cartilage in these patients. An important but unanswered question is whether the RFs or immune complexes formed by these autoantibodies differ between patients with RA and individuals who have RFs but have no synovitis.

Offprint requests to:

M. Mannik

216

M. Mannik et al.

Characteristics of RFs and Nature of Immune Complexes Formed by RFs Rheumatoid factors are autoantibodies with specificities directed to the Fc region of IgG. As a consequence, immune complexes are formed in vivo. RFs also bind to a variety of IgG molecules from other species. The common tests for RFs employed in the clinical setting detect primarily IgM RFs because of their high efficiency in agglutination reactions and high avidity for aggregated IgG.

IgM Rheumatoid Factors IgM RFs, like all other IgM molecules, are composed of five subunits, each subunit consisting of two heavy and two light polypeptide chains. On a structural basis they should have a valence of ten. Studies by ultracentrifugation, however, have shown that the valence of IgM RFs for IgG is five. Yet, all Fab fragments prepared from IgM RF bound to IgG [5, 59]. Thus, when one combining site of an IgM RF subunit reacts with IgG, the other combining site is hindered from binding to another IgG molecule or Fc fragment. The basis of this finding has not been fully explained, but it may relate to the size of bound antigen, because IgM antibodies to large antigenic molecules have a valence of five but IgM antibodies to small antigenic molecules, for example haptens, have a valence of ten. The association constant between IgM RF and monomeric IgG is around 105 liter/mole [38]. The valence of monomeric IgG for binding to IgM RF is one. Each of the subunits interacts potentially with one IgG molecule without co-operativity. Thus, maximally five IgG molecules bind to each IgM RF. The complexes of IgM RF with one or more IgG molecules were recognized first by ultracentrifugation in serums of patients with RA as the so called 22S complexes [12]. These complexes do not activate complement. The Fab fragments from IgM RFs bound to monomeric IgG and to aggregated IgG with comparable association constants, but intact IgM RFs bound to aggregated IgG about 106 times more effectively than the Fab fragments from IgM RF [8, 39]. These findings indicate that aggregated IgG has not developed new antigenic determinants that bind preferentially with IgM RFs as compared to monomeric IgG. Therefore, these results clearly show that the enhanced reactivity of IgM RFs with aggregated IgG results from the polivalency of this ligand. When IgM RFs bind to aggregated IgG, they activate complement readily [60]. The antigenic specificities of IgM RFs for various regions and subclasses of human IgG have been examined with different approaches. Experiments using hemagglutination inhibition systems showed that some specificities of IgM RF were directed to genetic markers on human IgG molecules, whereas other specificities of IgM RFs were directed to nongenetic markers [37]. Among the latter is the so called Ga antigenic determinant, present on IgG1, IgG2, and IgG4 molecules. Another approach has been to examine the direct binding of isolated human IgM RFs to IgG fragments, IgG subclasses and chemically modified human IgG to indicate the involved amino acids [50]. All 12 monoclonal and 11 polyclonal IgM RFs bound to Fc fragments in this study. The monoclonal IgM RFs, isolated from patients with essential mixed cryoglobulinemia or from patients with other

Immune Complexesand RheumatoidFactors in RheumatoidArthritis

217

B cell neoplasms, failed to bind to IgG fragments that lacked either the C3'2 or the C73 domain. These results indicated that the C3,2-C3,3 domain interface is an important binding site for IgM RFs. Among the polyclonal IgM RFs isolates from seven patients showed low level binding to pFc' fragmentsi which are composed of the C3,3 domain. These findings indicated that subtle differences exist between monoclonal and polyclonal IgM RFs. The same panels of IgM RFs were also examined for binding to individual, purified human IgG myeloma proteins representing the four IgG subclasses. The polyclonal IgM RFs from patients with RA had the common pattern of binding best to IgG1, IgG2, and IgG4 and least to IgG3. The monoclonal IgM RFs, however, demonstrated more varied patterns, including two with a clear IgG1, IgG2, IgG4 binding pattern, similar to the Ga specificity. Some monoclonal IgM RFs bound to all subclasses of IgG. Of note was that the binding of monoclonal IgM RFs to IgG3 was correlated with binding to IgG on which histidines were modified, indicative that these IgM RFs bind to a site not involving histidine residues. Finally, the binding of most IgM RFs to IgG was inhibited by fragment D of Staphylococcal protein A (SPA). Crystallographic studies have shown that SPA binds t0IgG in the C3,2-C3/3 domain interface area that involves tyrosyl and histidyl residues; which are also involved in binding ofIgG RF to IgG [35]. SPA binds to IgG1, IgG2, and IgG41 but not IgG3, because the latter lacks histidine 435 from the area that binds SPA. These observations collectively indicate that many IgM RFs interact with a site on IgG which is the same as, or overlaps with the site that binds SPA. Some IgM RFs, however, bind to other sites on IgG since their binding to IgG1, IgG2, and !gG4 is not inhibited by fragment D or even intact SPA. : Several studies have shown that Fc binding proteins induced by herpes type-1 virus and Fc binding proteins from Streptococci groups A , C and G bind to a similar area of IgG in the C3,2-C3,3 domain inerfacel [20, 36; 54,: 58, 65]. These findings, along with the data on SPA cited above, suggested that a similarity exists between the IgG binding sites of IgM RFs and the Viral and bacterial Fc binding proteins. Indeed, chicken antibodies to SPA and their Fab fragments bound polyclonal and monoclonal IgM RFs and IgG RFs [41]. These results raised the possibility that an internal image autoantiidiotype relationship exists between RFs and microbial Fc binding proteins. By this mechanism an immune response to the appropriate site of the microbial Fc binding Proteins could initiate the synthesis of RFs. A number of investigations have demonstrated the presence of RF-producing cells in the synovium of patients with RA. Therefore it is of interest if the IgM RFs produced in this location posses specificities different from IgM RFs in circulation. Robbins et al. [47], by examining inhibition of hemolytic activity of serum RFs and inhibition of hemolytic plaque-forming cells from the synovium, concluded that considerable differences were present. The serum IgM RFs were least inhibited by myeloma IgG3 proteins, i. e., they had the IgG 1, IgG2, IgG4 binding pattern. In contrast, the synovial tissue hemolytic plaque formation was best inhibited by IgG3 and IgG1. Furthermore, the data indicated that the avidity between IgG3 and IgM RF produced locally was higher than between these antibodies and other IgG subclasses. These findings were corroborated by a solidphase enzyme immunoassay, in which the synovial cell IgM RFs bound better

218

M. Mannik et al.

to IgG3 than to IgG1 [48]. These IgM RFs also bound to the pFc' fragment of IgG3, indicating reaction with antigenic determinants on the C3,3 domain. These findings along with the indicated higher production of IgG3 in synovium [14] suggested to the authors that the high avidity IgM RFs produced in the synovium combine locally with IgG3 and thus may have a significant pathogenic role in RA and that these antibodies do not appear in the circulation to any significant degree. This possibility becomes even more intriguing in view of the findings of Watson et al. [64] that the antibodies to type II collagen in patients with RA belong predominantly to IgG3 subclass. Thus, the development of IgG3 antibodies to type II collagen may contribute to the local production of IgM RFs in the synovium. These antibodies in turn can combine with the IgG3 antibodies bound to articular cartilage and thereby enhance the pathogenic events leading to destruction of cartilage. Clearly, more detailed studies are needed to ascertain the possible relationships between IgM RF specificities and the disease process in patients with RA.

IgG Rheumatoid Factors The existence of IgG RFs was established by Kunkel et al. [21]. They identified in serum of patients with RA the so called intermediate complexes, which upon analytical ultracentrifugation sediment between the 19S and 7S components of normal serum. These intermediate complexes dissociated into 7S material at low pH and reformed upon neutralization. The complexes contained IgG with RF activity and were thought to result from interaction of IgG RF and normal IgG. Schrohenloher [51] demonstrated that the F(ab')2 fragments derived from IgG RF were able to bind IgG. IgG RFs are unique antibodies in that they can form immune complexes by self-association without the presence of separate antigenic molecules. This occurs when the IgG RF specificity is directed to antigenic determinants present on the Fc region of the same molecule. As a consequence, two such antibodies form a cyclic dimer and then undergo further concentration-dependent polymerization. The self-association of IgG RFs was identified by studying the intermediate complexes from patients with RA [43, 44]. These complexes were isolated by gel filtration and consisted of IgG molecules. Upon sedimentation equilibrium ultracentrifugation these IgG molecules were mainly in dimeric form, and these dimers underwent concentration-dependent further polymerization. The Fab and F(ab')2 fragments of these isolates did not dimerize or polymerize, indicating that idiotype-antiidiotype reactions were not involved in self-association. Over 90% of the F(ab')z fragments in these preparations bound to IgG, showing that nearly all of the IgG molecules present had RF activity. On the basis of these findings the model of self-association was proposed. In this model two IgG RFs form a cyclic dimer, where one antibody combining site of each molecule binds to the Fc region of the other I gG RF molecule. This reaction is stabilized by the formation of two antigen-antibody bonds. The second antibody combining site remains free on both IgG RF molecules in the dimer. The availability of antigenic determinants on the Fc regions and the free antibody combining sites then account for the observed concentration-dependent further polymerization of the dimers. This polymerization of dimers had an association constant of about 105 liters/mole, which

Immune Complexesand Rheumatoid Factors in Rheumatoid Arthritis

219

is similar to the association constant of binding one Fab arm of IgG RF to one IgG molecule. This relationship implies that in the presence of large excess of normal IgG further polymerization is inhibited by mass action. The valence of IgG molecules for interaction with IgG RF is one [34]. These studies were conducted by examining the interaction of Fab fragments prepared from IgG RF with intact IgG and with Fc fragments form normal IgG and Fc fragments from IgG RF. The valence of all Fc fragments was two for interaction with the Fab fragments from IgG RF, as expected from the presence of both carboxyterminal halves of 7 chains in these fragments. Furthermore, the association constants for binding to Fc fragments from normal IgG and from IgG RF were comparable, indicating that unique features of the Fc regions of IgG RFs neither exist nor contribute to self-association of these molecules. In contrast, the valence of intact human and rabbit IgG for binding IgG RF was one, even though the valence of the isolated Fc fragments from the same molecules was two. These experimental results indicate that when a Fab fragment or an intact IgG RF binds to one antigenic determinant on the Fc region of an intact IgG molecule then the second antigenic determinant on the Fc region becomes unavailable by steric hindrance. This may result from movement of the Fab arms as the first antigenic site on the Fc region is engaged by IgG RF, positioning the opposite Fab arm on the IgG molecule so that the second site on the Fc region is blocked from interaction with IgG RF. In the process Of concentration-dependent polymerization of IgG RF dimers the second antigenic determinant must be available on IgG RF molecules. If this were not the case, the self-association ofIgG RFs would terminate at dimer formation. In fact, experimental data have shown that further polymerization of IgG RF dimers occurs [44]. Therefore, as the cyclic dimer is formed between two IgG RFs, sufficient conformational changes must occur with the ring closure so that the unoccupied antigenic determinants on Fc regions remain available for further polymerization. The described features of self-association of IgG RFs have important impfications in relation to pathogenic processes in patients with RA. Several investigations have shown that in the synovial tissue a substantial number of plasma dells are engaged in the synthesis of IgG RFs. This would lead to high local concentrations oflgG RFs in a relative paucity of normal IgG. In this setting IgG RFs would undergo self-association. These immune complexes can activate complement [4, 49]. Thus, the self-association of IgG RFs in tissues can represent a unique pathogenic process in patients with RA. This possibility is supported by several observations. First, IgG-containing immune complexes were shown to exist in synovial fluid of patients with RA [66, 67]. Second, the detection of IgG RF in synovial tissue and synovial tissue eluates was enhanced by pepsin digestion, which would free the antibody combining sites of self-associated IgG RFs [32]. Third, the content of some synovial tissue plasma cells activated added human complement [33]. This finding indicated that the IgG RFs synthesized by the cells self-associated already in the cytoplasm and were therefore able to activate complement. When the IgG RFs enter circulation, further polymerization beyond the dimer would be terminated. Two factors contribute toward this. First, as discussed above, the valence of normal IgG for interaction with IgG RF is one. Thus, when each

220

M. Mannik et al.

of the free antibody combining sites of the self-associated IgG RF dimer interact with normal, monomeric IgG, further self-association is terminated. Second, as already pointed out, the association constants for further polymerization of IgG RF dimers and for interactions of one free IgG RF antibody site with normal IgG are about 105 liters/mole. Therefore, in circulation the interaction of self-associated IgG RF dimers with the abundant normal IgG is favored by mass action and further polymerization is terminated at a state of !gG tetramer. It is well documented that these immune complexes can exist harmlessly in relatively large amounts in circulation, though they have been found to cause the hyperviscosity syndrome [45]. The discussed mechanisms for self-association exist when the specificity of IgG RFs is directed to antigenic determinants that are present and available on the Fc regions for cyclic dimer formation. If the specificity is directed to antigenic determinants absent from the Fc region of the same IgG RF molecule, then these IgG RFs can only react with normal IgG or with IgG RFs that carry these antigenic determinants. Such complexes, however, would differ substantially from the immune complexes formed by self-associating IgG RFs. The maximum size attainable with IgG RF and normal IgG would be an IgG trimer, an immune complex that has low if any pathogenic potential. For self-association Of IgG RFs to occur, the relationships between the antigenic specificity of these antibodies and their IgG subclass becomes important. For two self-associating IgG RFs, purified from patients with RA and abundant intermedate complexes, the antigenic specificity was localized to the site on IgG which binds SPA, involving the C72-C73 domain interface [35]. In these studies the IgG RFs did not bind to fragments of IgG molecules that lacked the intact C72-C73 domain interface, the IgG RFs interacted weakly with IgG3, pH titrations indicated the involvement of tyrosines and histidines at the active sites, chemical modifications of tyrosines or histidines on the antigen decreased the reactivity with IgG RFs, and the small monovalent fragment D of SPA blocked the binding of IgG RF to IgG. This information collectively with the known crystallographic structure of IgG and the crystallographically identified site on IgG that binds SPA permitted the identification of one site on IgG that is involved in selfassociation. This site is present on IgG1, IgG2 and IgG4 and, therefore, should permit self-association between IgG RFs of these different IgG subclasses. IgG3 molecules are unlikely to enter self-association via this antigenic specificity. The subclass of IgG RFs has been examined in patients with RA. Shakib and Stanworth [55] found that IgG RFs that bound to rabbit IgG were most commonly IgG1 RFs and IgG2 RFs both in serum and synovial fluid of patients with RA, using polyclonal antisera to the IgG subclasses. Of note was that IgG3 RFs were detected in 4 % of tested serums and in 43 % of tested synovial fluids. A high prevalence of IgG1 RFs and IgG4 RFs was found in a recent study employing monoclonal antibodies specific to IgG subclasses [6]. It has been pointed out that IgG4 subclass antibodies become prominent during chronic antigenic stimulation. This subclass of IgG is relatively ineffective in complement activation when bound with antigen, but IgG4 molecules bind to mast cells and may be involved in release of histamine and other cytokines. All of these points may be relevant to the pathogenic processes in RA.

Immune Complexesand RheumatoidFactors in RheumatoidArthritis

221

The specificity of IgG RFs for IgG subclasses was studied by Elson et al. [10] using as antigens in the enzyme-linked immunosorbent assay (ELISA) system Fc fragments of IgG1 and IgG2, and intact IgG3 and IgG4 myeloma proteins. The binding of IgG RFs was detected with a monoclonal mouse antibody directed to the C3'1 domain, which did not bind to the subclass proteins used as antigens for RFs. These investigators found that in 64 serums from RA patients the IgG RFs bound the best to IgG2, then IgG1 and IgG4 and to a much lesser degree to IgG3. Furthermore, a statistically significant correlation was present between the specificity of IgG RFs and IgM RFs to the same test antigens. They found no relationships between disease characteristics and the specificities of IgM RFs and IgG RFs. In a small number of patients the specificity of binding of IgG4 RFs to IgG1, IgG2, and IgG3 myeloma proteins was examined. The highest binding was also found to IgG2 [6]. The overall pattern of the specificity of IgG RFs in this study approximated the IgG1, 2, 4 pattern already discussed for IgM RFs and the two IgG RFs studied in detail. This specificity is similar to the Ga binding site, now equated to the binding site on IgG which also interacts with SPA and other bacterial Fc binding proteins. This specificity, as noted above, permits self-association of IgG RFs. The observed highest comparative binding to IgG2 has not been fully explained.

IgA Rheumatoid Factors IgA RFs have been recognized in patients with RA. IgA molecules in humans exist in monomeric or polymeric form. In selected patients with the systemic sicca syndrome the IgA RFs existed in polymeric form [9]. These polymeric IgA RFs participated in the formation of intermediate complexes by interaction with normal IgG. A subsequent study demonstrated that monomeric IgA RFs may not be detected adequately by solid-phase binding assays because of low affinity [52]. This problem was overcome by adding mouse monoclonal antibodies to human o~chains to serum fractions containing the monomeric IgA RFs. These antibodies formed soluble immune complexes with the monomeric IgA RFs, rendering them polymeric, and enhancing their binding in the solid-phase assay. The application of this method to the study of serums from 42 patients with RA showed that 29 of 42 patients had monomeric IgA RFs and all 42 patients showed increased levels of polymeric IgA RFs [53]. Quantitatively the letter exceeded the monomeric IgA RFs. These studies show that circulating immune complexes may form with either monomeric or polymeric IgA RFs, the latter would obviously be larger than the former. The specificity of IgA RF has been examined in detail only in two patients [1, 2]. In both patients the specificity was in the IgG1, IgG2, IgG4 binding pattern discussed above for IgM RFs. Furthermore, one of these proteins bound well to Fc fragments but not to an isolated C72 domain or to pFc' fragments (C73 domain). These results indicated that the binding required intact domains and may involve the C3,2-C3,3 interface [2].

222

M. Mannik et al.

Presence of Immune Complexes in Patients with RA

A large number of studies have examined the presence of immune complexes in the serum and synovial fluid of patients with RA with different methods for detection of immune complexes. A review of these studies is not possible in view of limited space and previous reviews have covered some of this material [22, 61].

Serum and Synovial Fluid Immune Complexes in Patients with RA A number of different biological test systems have been employed to detect immune complexes in serum and synovial fluid of patients with RA. Several studies have indicated that not all tests detect immune complexes to comparable degree in patients with RA. The Raji cell test is less frequently positive than the solid or fluid phase C lq binding tests. The generalization is true that the positive test results are more likely to be found in patients with severe disease than in those with mild disease. Similarly, positive test results are encountered more often in patients with extra-articular disease manifestations than in those with the disease affecting joints only [31, 46]. Furthermore, the level and prevalence of immune complexes are higher in synovial fluid than in serum. These conclusions were based on a study using the Clq fluid-phase binding assay, an assay using monoclonal rheumatoid factor, and the Raji cell assay to compare the level and presence of immune complexes in serum and synovial fluid specimens of 82 patients with RA [13]. The tests for immune complexes using antigen-unrelated methods will obviously not contribute to the diagnosis of RA, but the positivity of these tests appears to parallel the disease activity. At this time no evidence is on hand to show that these tests should replace bedside observations and sound clinical judgement. These tests, however, have contributed new knowledge towards the understanding of the pathogenic mechanisms of the disease.

Immune Deposits in Synovial Tissue, Synovial Phagocytic Cells and Surface Layers of Articular Cartilage IgG, IgM, and complement components were detected by immunofluorescence microscopy between cells in the synovium of patients with RA, within the phagolysosomes of phagocytic cells in the synovial fluid, and in the cells lining the synovial cavity. These findings have been reviewed elsewhere [27] and suggest that immune complexes are present in interstitial spaces. Some evidence indicated that rheumatoid factors are present in these deposits. Since RFs are synthesized by plasma cells in the synovium, the increased local concentration of these antibodies could lead to local immune precipitation and complement activation. The production of immunoglobulins, rheumatoid factors and other antibodies by the synovial tissue has been reviewed elsewhere [11, 57]. The presence of other antigenic molecules in these interstitial immune deposits, however, has not been ruled out. The same holds true for the immune deposits that have been identified in synovial fluid neutrophils and in monocytes along the synovial lining. Phagocytosis of large complexes can occur from the synovial fluid, the interstitial spaces or the cartilage surface.

Immune Complexes and Rheumatoid Factors in Rheumatoid Arthritis

223

The presence of immune deposits in articular cartilage of patients with RA was first detected by immunofluorescence microscopy [7]. Immune complexes were assumed to be present when at least two immunoglobulins and C3 components were present. By this criterion 92% of 37 patients with classical RA had immune deposits in the cartilage. IgG and C3 were the most frequent constituents, followed by IgM and IgA. These materials were present in the superficial 100 ~tm of the cartilage. These immune deposits were also visualized by immunoelectron microscopy, and the results suggested phagocytosis of the deposits by neutrophils [40]. Furthermore, once the pannus covered the cartilage with immune deposits, underneath the pannus the deposits were not detectable by immunofluorescence microscopy or electron microscopy. Macrophage-like cells were present in this area and may have ingested and degraded the immune deposits [56]. In order to identify the specificity of antibodies in immune deposits in articular cartilage, Jasin [19] extracted immunoglobulins from cartilage after degradation of cartilage by collagenase. These specimens were obtained from 16 patients with RA, 11 with osteoarthritis and 6 normal controls. The RA cartilage yielded the highest amounts of IgM (3.6-+ 1.6 ~tg/g of cartilage) and IgG (11.3 +-3.4 ~tg/g of cartilage), and IgM RFs and IgG RFs were recovered only from cartilage from patients with RA. Antibodies to native or denatured type II collagen were recovered from some cartilage specimens from patients with RA or osteoarthritis, but higher amounts of these antibodies were present in patients with RA.

Composition of Serum and Synovial Fluid Immune Complexes in Patients with Rheumatoid Arthritis As discussed in a preceding section, immune complexes are abundant in serum and synovial fluid of patients with rheumatoid arthritis. Several studies have indicated that the severity of the disease is related to the abundance of immune complexes. Thus, it is of interest to establish the composition of the immune complexes to determine if they contain only self-antigens, or if exogenous antigens are present as well. Initial immunochemical analysis of immune complexes from the synovial fluid of patients with RA identified only immunoglobulins and complement components [66, 67]. In subsequent studies, immune complexes were isolated by sucrose-polyethylene glycol gradient ultracentrifugation and then analyzed by two-dimensional polyacrylamide gel electrophoresis under dissociating conditions [24, 25]. IgM, IgG, Clq and C3 breakdown products were present in all isolates. IgA, C4, factor B, activated Cls and Clr were found in a large number of specimens. Unidentified material accounted for less than four percent of protein and varied from specimen to specimen. These findings in conjunction with the demonstration of self-associating IgG RFs in serum immune complexes [44] and in five of nine studied synovial fluids [28] argue that self-associating IgG RFs and IgM RFs are the major constituents of immune complexes in both serum and synovial fluid of patients with RA. This possibility was further supported by a study where polyethylene glycolprecipitated immune complexes were fractionated by gel filtration and IgM RFs were detected in the largest fraction and IgG RFs were present in all three gel filtration fractions [30]. In another study immune complexes containing C3d were

224

M. Mannik et al.

isolated from synovial fluid of patients with RA with polyethylene glycol precipitation and anti-C3d affinity chromatography [3]. In these isolates IgG and IgM were abundant but rheumatoid factor activity was demonstrated only in a small proportion of the specimens. On the basis of these findings the authors concluded that the immunoglobulin aggregates were produced by systems other than RFs. These aggregates, however, were not characterized further and the sensitivity for detection of RFs in the isolates was not defined. The possibilities for aggregation of irnmunogtobulins by nonimmunological means will be considered at the end of this section. More sensitive tests have been employed recently in the characterization of the constituents of immune complexes in patients with RA. Inman et al. [16] immunized rabbits with immune complexes isolated from patients with RA by polyethylene glycol precipitation and binding to Staphylococcal protein A. Aliquots of these complexes were then separated by sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and charaterized following transfer to nitrocellulose paper. The polypeptide chains identified in the immune complexes belonged to IgM, IgG, IgA and Clq. Four unidentified polypeptide chains that were recognized by the produced antisera had molecular weights of 26,000; 39,000; 40,000; and 43,000. Of note was that the specific antisera to fibronectin and C3 used did not react with these components. The specificity of the antibodies to C3, however, was not defined. In continuation of these studies Inman et al. [17] specifically looked for the presence of the viral capsid antigen of the Epstein Barr virus in the isolated immune complexes from patients with RA. They discovered an unexpected cross-reactivity of Fc fragments of human IgG and the viral capsid antigen. In another study serum immune complexes from 25 patients with RA were isolated after polyethylene glycol precipitation with anti-C 1q affinity chromatography [29]. Thereafter these isolates were separated by SDS-PAGE and stained with Coomassie brilliant blue stain or a highly sensitive silver stain. The constituents were identified by Western blot analysis with a number of antisera specific to human serum proteins. The most frequently found serum proteins or their polypeptide chains were albumin, IgM, Clq, IgG and C3, but polypeptide chain~ of C3, Cls, C4, IgA, fibrinogen and C reactive protein were also identified. IgE, C9 and o~2-macroglobulin or their polypeptide chains were absent. Of note was that a 48,000 molecular weight polypeptide chain was present in the immune complexes of six of 14 tested isolates from patients with RA and extra-articular features of the disease. This substance could not be identified with antibodies to human serum or with a number of specific antibodies to individual serum proteins, raising the possibility that it may be an exogenous antigen. It is important to note, however, that this polypeptide chain represented a relatively small proportion of all polypeptide chains present in the immune complexes, as revealed by staining of the gels with a sensitive silver stain. Thus, all studies on the composition of immune complexes in the serum or synovial fluid of patients with RA have disclosed the presence of immunoglobulins and complement components with only a minor amount of yet unidentified proteins. These findings are consistent with a view that rheumatoid factors and their antigens represent the major constituents of immune complexes in both the serum and synovial fluid of patients with RA.

Immune Complexes and Rheumatoid Factors in Rheumatoid Arthritis

225

Before concluding this section, one must also consider the alternative possibility that the complement activating imrnunoglobulin polymers arise by mechanisms other than immune complex formation. Several possibilities have been proposed in published works. Jasin [ 18] showed that incubation of human IgG with purified myeloperoxidase, hydrogen peroxide, and an appropriate hydrogen donor (e. g., catechol) caused formation of aggregated IgG by covalent bond formation. These aggregated IgG molecules activated complement and gave positive results in tests for immune complexes. Lunec et al. [23] showed that ultra-violet irradiation and exposure to activated polymorphonuclear leukocytes (PMN) resulted in free-radical-induced damage to IgG. This was evident by development of changes in flourescence emission upon excitation by light of appropriate wavelength and formation of aggregates of theses molecules. The nature of the formed bonds during aggregation was not defined in these studies. Furthermore, Parekh et al. [42] showed that galactosylation of IgG is decreased both in patients with osteoarthritis and in patients with RA, but the decrease is higher in patients with RA. On the basis of these results the authors proposed that the absence of galctose in crucial positions in the carbohydrate chains of IgG might make these molecules more "sticky". This would then lead to aggregation of IgG molecules by mechanisms other than antigen-antibody bond formation. Clearly, several mechanisms exist for nonspecific aggregation of IgG. Evidence is not on hand to implicate one of these mechanisms for polymerization of IgG in patients with RA. Some of the described mechanisms for aggregation lead to covalent bond formation. Studies of the immune complexes in serum and synovial fluid samples have shown that the polymerized IgG can be dissociated and reformed by change of conditions known to disrupt antigen-antibody bonds. In addition, variations in the antigen-antibody specificity of RF interactions with IgG have been delineated that at this time cannot be explained by nonspecific polymer formation. Thus, even though mechanisms have been documented for aggregation of IgG by biological systems, available evidence does not permit the conclusion that these mechanisms are operative in patients with RA. The composition of immune deposits in the interstitial spaces of synovial tissue, within phagocytic cells of the inflamed synovium and in the superficial layers of articular cartilage has not been fully established. The presence of rheumatoid factors in some of these deposits has been shown as discussed above. The presence of IgM RFs, IgG RFs and antibodies to type II collagen in the superficial layers of articular cartilage was shown by Jasin [19] as reviewed in the preceding section.

Possible Mechanismsfor Entrapment of Immune Deposits in Articular Cartilage The presence of immune deposits in articular cartilage, containing RFs and antibodies to type II collagen, and the accumulation of abundant plasma cells and other cells in the synovial tissue are the features of RA that distinguish patients with this disease from other patients with RFs in circulation. The processes that lead to these pathological features of RA are not adequately understood at this time. Several observations, however, need to be considered that may offer some insights into these processes.

226

M. Mannik et al.

Immune deposits were frequently seen in articular cartilage in patients with RA and with lesser frequency in patients with osteoarthritis [7]. Antibodies to type II collagen were recovered from cartilage of patients with RA in larger amounts than from patients with osteoarthritis, but RFs were recovered only from cartilage of patients with RA [19]. Furthermore, the circulating antibodies to type II collagen were predominantly IgG3 subclass [64] and IgM RFs produced by synovial cells were directed more often to IgG3 than IgM RFs in circulation [47, 48]. Therefore, the possibility exists that in RA the development of IgG3 antibodies to type II collagen leads to their deposition in articular cartilage and this stimulates in genetically programmed individuals local RF synthesis with specificity to IgG3 subclass. These RFs then could enhance the inflammation that leads to the destruction so characteristic of RA. Alternatively, the presence of antibodies to type II collagen or even to other collagens or other components of articular cartilage and this immune response may not be related to the synthesis of RFs. If this is true, then other explanations are needed for the deposition of RFs in articular cartilage. The IgG molecules synthesized by rheumatoid synovium were electrophoretically restricted and tended to be cationic in character [15]. The articular cartilage is rich in fixed negative charges owing to the presence of glycosaminoglycans. A number of studies in experimental models have clearly shown that cationic antigens, antibodies and immune complexes can accumulate in articular cartilage by charge-charge interactions, first demonstrated in mice by van den Berg et al. [62]. Highly cationic IgG penetrated articular cartilage in vitro but cationic ferritin (molecular weight 440,000) did not, indicating a molecular size limitation for penetration of cartilage [63]. Immune complexes containing cationic antibodies bound only to the surface of cartilage and unaltered IgG was not able to penetrate cartilage in vitro, even when an antigen was planted deep into the cartilage matrix [68]. These studies suggest that cationic IgG RFs may be attracted to cartilage by charge-charge interactions. These autoantibodies may thus become locally concentrated as compared to synovial fluid and then undergo self-association by mechanisms already discussed. In turn, these aggregates may become a favored substrate for IgM RFs. In an analogous situation, local concentration of immune complexes by chargecharge interactions has been shown to contribute to immune complex formation at the glomerular basement membrane [26]. Charge-charge interactions, therefore, may contribute to a variety of known and yet to be discovered processes involved in the attraction of antibodies to the articular cartilage. Future inquiries are likely to clarify these processes and enhance our understanding of RA.

Conclusions

Substantial evidence indicates that humoral immunity through antigen-antibody complex formation contributes to the pathogenesis of RA. Several studies have examined the composition of immune complexes present in the synovial fluid of patients with RA. The constituents of these immune complexes have been immunoglobulins and complement components. Only a few polypeptides in these immune complexes have not been identified. These unidentified components ac-

Immune Complexes and Rheumatoid Factors in Rheumatoid Arthritis

227

count for only a few percent of total proteins present. These results suggest that rheumatoid factors and their antigens are quantitatively the important ingredients of immune complexes in synovial fluid of patients with RA. The immune deposits in the superficial layers of articular cartilage from patients with RA contain RFs and antibodies to type II collagen. IgG RFs are unique autoantibodies in that they can form immune complexes with pathogenic potential by self-association and without the presence of separate antigen molecules. Self-association becomes possible when the antigenic specificity of IgG RFs is directed to antigenic determinants present on the Fc region of the same molecule. Self-association of IgG RFs is enhanced in a milieu of high concentration of these antibodies and in relative paucity of normal IgG--i. e., conditions that exist in the synovial tissue where IgG RFs are synthesized. Relatively little attention has been devoted to antigenic specificity of RFs in relation to disease. The self-associating IgG RFs illustrate how the antibody specificity and presence of antigenic determinants can contribute to formation of unique immune complexes. One of the major antigenic determinants for IgM RFs is in the C72-C73 domain interface and involves some of the same amino acids that constitute the site for binding SPA. The specificity for self-associating IgG RFs is directed to the same antigenic determinant. Fc binding proteins from Streptococci and the Fc binding proteins induced by herpes type 1 virus bind to the same or overlapping areas of human IgG. The reasons for this similarity of binding to IgG between microbial Fc binding proteins and RFs is not apparent. This relationship, however, may stimulate the synthesis of some RFs by an internal image autoantiidiotype mechanism. Acknowledgements. This work was supported in part by research grant AM12849 from the National Institute of Arthritis, Musculoskeletal and Skin Diseases and the Townsend-Henderson Fund.

References 1. Abraham GN, Clark RA, Vaughan JH (1972) Characterization of an IgA rheumatoid factor: binding properties and reactivity with the subclasses of human "yG globulin. Immunochemistry 9:301 2. Abraham GN, Welch E, Trieshmann HW (1978) Human triclonal anti-IgG gammaopathy II. Determination of the antigenic specificity patterns of the IgG, IgA and IgM autoantibodies for the subclasses of IgG. Immunology 35:437 3. Bedwell AE, Elson C J, Carter SD, Dieppe PA, Hutton CW, Czudek R (1987) Isolation and analysis of complement activating aggregates from synovial fluid of patients with rheumatoid arthritis using monoelonal anti-C3d antibodies. Ann Rheum Dis 46:55 4. Brown PB, Nardella FA, Mannik M (1982) Human complement activation by self-associated IgG rheumatoid factors. Arthritis Rheum 25:1101 5. Chavin SI, Franklin EC (1969) Studies on antigen-binding activity of macroglobulin antibody subunits and their enzymatic fragments. J Biol Chem 244:1345 6. Cohen PL, Cheek RL, Hadler JA, Yount WJ, Eisenberg RA (1987) The subclass distribution of human IgG rheumatoid factor. J Immunol 139:1466 7. Cooke TD, Hurd ER, Jasin HE, Bienenstock J, Ziff M (1975) Identification of immunoglobulins and complement in rheumatoid articular colleagenous tissues. Arthritis Rheum 18:541 8. Eisenberg R (1976) The specificity and polyvalency of binding of a monoclonal rheumatoid factor. Immunochemistry 13:355 9. Elkon KB, Delacroix DL, Gharavi AE, Vaerman JP, Hughes GRV (1982) Immunoglobulin A and polymeric IgA rheumatoid factors in systemic sicca syndrome: partial characterization. J Immunol 129:576

228

M. Mannik et al.

10. Elson CJ, Carter SD, Scott DGI, Bacon PA, Lowe J (1985) A new assay for IgG rheumatoid factor activity and its use to analyze rheumatoid factor activity with human isotypes. Rheumatol Int 5:175 11. Fehr K (1982) Autoimmune reactions and rheumatoid arthritis. Eur J Rheumatol Inflamm 5: 439 12. Franklin EC, Holman HR, Mfiller-Eberhard HJ, Kunkel HG (1957) An unusual protein component of high molecular weight in the serum of certain patients with rheumatoid arthritis. J Exp Med 105:425 13. Halla JT, Volanakis JE, Schrohenloher RE (1979) Immune complexes in rheumatoid arthritis sera and synovial fluids. A comparison of three methods. Arthritis Rheum 22:440 14. Hoffman WL, Goldberg MS, Smiley JD (1982) IgG-3 subclass production by rheumatoid synovial tissue cultures. J Clin Invest 69:136 15. Hoffman WL, Douglas RR, Smiley JD (1984) Synthesis of electrophoretically restricted IgG by cultured rheumatoid synovium. Arthritis Rheum 27:976 16. Inman RD, Hamilton NC, Redecha PB, Hochhauser DM (1986) Electrophoretic transfer blotting analysis of immune complexes in rheumatoid arthritis. Clin Exp Immunol 63:32 17. Inman RD, Chin B, Hamilton NC (1987) Analysis of immune complexes in rheumatoid arthritis for Epstein-Barr virus antigens reveals cross-reactivity of viral capsid antigen and human IgG. J Immunol 138:407 18. Jasin HE (1983) Generation of IgG aggregates by the myeloperoxidase-hydrogen peroxide system. J Immunol 130:1918 19. Jasin HE (1985) Autoantibody specificities of immune complexes sequestered in articular cartilage of patients with rheumatoid arthritis and osteoarthritis. Arthritis Rheum 28:241 20. Johansson PJH, Schr6der AK, Nardella FA, Mannik M, Christensen P (1986) Interaction between herpes simplex type I-induced Fc receptor and human and rabbit immunoglobulin G (IgG) domains. Immunology 58:251 21. Kunkel HG, Mfiller-Eberhard HJ, Fudenberg HH, Tomasi TB (1961) Gamma globulin complexes in rheumatoid arthritis and certain other conditions. J Clin Invest 40:117 22. Lambert PH, Casali P (1978) Immune complexes and the rheumatic diseases. Clin Rheum Dis 4:617 23. Lunec J, Blake DR, McCleary SJ, Brailsford S, Bacon PA (1985) Self-perpetuating mechanisms of immunoglobulin G aggregation in rheumatoid inflammation. J Clin Invest 76:2084 24. Male D, Roitt IM (1981) Molecular analysis of complement-fixing rheumatoid synovial fluid immune complexes. Clin Exp Immunol 46:521 25. Male D, Roitt IM, Hay FC (1980) Analysis of immune complexes in synovial effusions of patients with rheumatoid arthritis. Clin Exp Irmamnol 39:297 26. Mannik M (1987) Mechanisms of tissue deposition of immune complexes. J Rheumatol [Suppl 13] 14:35 27. Mannik M, Nardella FA (1981) Self-associating IgG rheumatoid factors. In: Shiokawa Y, Abe T, Yamauchi Y (eds) New horizons in rheumatoid arthritis. Exerpta Medica, Amsterdam, pp 124-131 28. Mannik M, Nardella FA (1985) IgG rheumatoid factors and self-association of these antibodies. Clin Rheum Dis 11:551 29. Melsom RD, Smith PR, Maini RN (1987) Demonstration of an unidentified 48-kD polypeptide in circulating immune complexes in rheumatoid arthritis. Ann Rheum Dis 46:104 30. Mochan E, Passon TJ Jr (1986) Properties and reactivity of immune complexes in rheumatoid synovial fluid. Rheumatol Int 6:103 31. Mumford PA, Horsfall AC, Maini RN (1982) The frequency of circulating immune complexes in rheumatoid arthritis and systemic lupus erythematosus. Rheumatol Int 1:181 32. Munthe E, Natvig JB (1972) Immunoglobulin classes, subclasses and complexes of IgG rheumatoid factor in rheumatoid plasma cells. Clin Exp Immunol 12:55 33. Munthe E, Natvig JB (1972) Complement-fixing intracellular complexes oflgG rheumatoid plasma cells. Scand J Immunol 1:217 34. Nardella FA, Teller DC, Mannik M (1981) Studies on the antigenic determinants in the selfassociation of IgG rheumatoid factor. J Exp Med 154:112 35. Nardella FA, Teller DC, Barber CV, Mannik M (1985) IgG rheumatoid factors and Staphylococcal protein A bind to a common molecular site on IgG. J Exp Med 162:1811

Immune Complexes and Rheumatoid Factors in Rheumatoid Arthritis

229

36. Nardella FA, Schr6der AK, Svensson M, Sj6quist, Barber CV, Christensen P (1987) T15 group A Streptococcal Fc receptor binds to the same location on IgG as Staphylococcal protein A and IgG rheumatoid factors. J Immunol 138:922 37. Natvig JB, Gaarder PI, Turner MW (1972) IgG antigens of the C3,2 and C3'3 homology regions interacting with rheumatoid factors. Clin Exp Immunol 12:177 38. Normansell DE (1970) Anti-3,-globulins in rheumatoid arthritis sera-I. Studies on the 22S complex. Immunochemistry 7:787 39. Normansell DE (1971) Anti-7-globulins in rheumatoid arthritis sera-II. The reactivity of anti3,-globulin rheumatoid factors with altered "gG-globulin. Immunochemistry 8:593 40. Ohn0 O, Cooke TD (1978) Electron microscope morphology of immunoglobulin aggregates and their interactions in rheumatoid articular collagenous tissue. Arthritis Rheum 21:516 41. Oppliger IR, Nardella FA, Stone GC, Mannik M (1987) Human rheumatoid factors bear the internal image of the Fc binding region of Staphylococcal protein A. J Exp Med 166:702 42. Parekh RB, Dwek RA, Sutton BJ, Fernandes DL, Leung A, Stanworth D, Rademacher TW, Mizuochi T, Taniguchi T, Matsuta K, Takeuchi F, Nagano Y, Miyamoto T, Kobata A (1985) Association of rheumatoid arthritis and primary osteoarthritis with changes in the glycosylation pattern of total serum IgG. Nature 316:452 43. Pope RM, Teller DC, Mannik M (1974) The molecular basis of self-association of antibodies to IgG (rheumatoid factors) in rheumatoid arthritis. Proc Natl Acad Sci USA 71:517 44. Pope RM, Teller DC, Mannik M (1975) The molecular basis of self-association of IgG-rheumatoid factors. J Immunol 115:365 45. Pope RM, Mannik M, Gilliland BC, Teller DC (1975) The hyperviscosity syndrome in rheumatoid arthritis due to intermediate complexes formed by self-association of IgG-rheumatoid factors. Arthritis Rheum 18:97 46. Reynolds WJ, Yoon SJ, Emin M, Chapman KR, Klein MH (1986) Circulating immune complexes in rheumatoid arthritis: a prospective study using five immunoassays. J Rheumatol 13:700 47. Robbins DL, Skilling J, Benisek WF, Wistar R Jr (1986) Estimation of the relative avidity of 19S IgM rheumatoid factor secreted by rheumatoid synovial cells for human IgG subclasses. Arthritis Rheum 29:722 48. Robbins DL, Benisek WL, Benjamini E, Wistar R Jr (1987) Differential reactivity of rheumatoid synovial cells and serum rheumatoid factors to human immunoglobulin subclasses 1 and 3 and their CH3 domains in rheumatoid arthritis. Arthritis Rheum 30:489 49. Sabharwal UK, Vaughan JH, Fong S, Bennett PH, Carson DA, Curd JG (1982) Activation of the classical pathway of complement by rheumatoid factors. Assessment by radioimmunoassay for C4. Arthritis Rheum 25:161 50. Sasso EH, Barber CV, Nardella FA, Yount WJ, Mannik M (1988) Antigenic specificities of human monoclonal and polyclonal IgM rheumatoid factors: the C3,2-C3,3 interface region contains the major determinants. J Immunol (in press) 51. Schrohenloher RE (1966) Characterization of the -y-globulin complexes present in certain sera having high titers of anti-globulin activity. J Clin Invest 45:501 52. Schrohenloher RE, Koopman WJ, Moldoveanu Z, Solomon A (1985) Activity of rheumatoid factors of different molecular sizes: comparison of autologous monomeric and polymeric monoclonal IgA rheumatoid factors. J Immunol 134:1469 53. Schrohenloher RE, Koopman WJ, Alarc6n GS (1986) Molecular forms of IgA rheumatoid factor in serum and synovial fluid of patients with rheumatoid arthritis. Arthritis Rheum 29: 1194 54. Schr6der AK, Nardella FA, Mannik M, Svensson M-L, Christensen P (1986)Interaction between Streptococcal IgG Fc receptors and human and rabbit IgG domains. Immunology 57:305 55. Shakib F, Stanworth DR (1978) Antigammaglobulin (rheumatoid factor) activity of human IgG subclasses. Ann Rheum Dis 37:12 56. Shiozawa S, Jasin HE, Ziff M (1980) Absence of immunoglobulins in rheumatoid cartilage-pannus junctions. Arthritis Rheum 23:816 57. Smiley JD, Hoffman WL, Moore SE, Paradies LH (1985) The humoral immune response of the rheumatoid synovium. Semin Arthritis Rheum 14:151 58. Stone GC, Bj6rck L, Barber CV, Nardella FA (1987) Streptococcal protein G and rheumatoid factors bind in the same area of the Fc fragment of human IgG. Arthritis Rheum [Abstr] 30: S 103

230

M. Mannik et al.

59. Stone JM, Metzger H (1968) Binding properties of a Waldenstr6m macroglobulin antibody. J Biol Chem 243:5977 60. Tanimoto K, Cooper NR, Johnson JS, Vaughan JH (1975) Complement fixation by rheumatoid factor. J Clin Invest 55:437 61. Theofilopoulos AN, Dixon FJ (1979) The biology and detection of immune complexes. Adv Immunol 28:89 62. van den Berg WB, van de Putte LBA, Zwarts WA, Joosten LAB (1984) Electrical charge of the antigen determines intra-articular antigen handling and chronicity of arthritis in mice. J Clin Invest 74:1850 63. van Lent PLEM, van den Berg WB, Schalwijk J, van de Putte LBA, van den Bersselaar L (1987) The impact of protein size and charge on its retention in articular cartilage. J Rheumatol 14:798 64. Watson WC, Cremer MA, Wooley PH, Townes AS (1986) Assessment of the potential pathogenicity of type II collagen autoantibodies in patients with rheumatoid arthritis. Evidence of restricted IgG3 subclass expression and activation of complement C5 to C5a. Arthritis Rheum 29: 1316 65. Wiger D, Michaelson E (1985) Binding site and subclass specificity of the herpes virus type 1-induced Fc receptor. Immunology 54:565 66. Winchester RJ (1975) Characterization of IgG complexes in patients with rheumatoid arthritis. Ann NY Acad Sci 256:73 67. Winchester RJ, Agnello V, Kunkel HG (1970) Gamma globulin complexes in synovial fluids of patients with rheumatoid arthritis. Clin Exp Immunol 6:689 68. Zatarain-Rios E, Mannik M (1987) Charge-charge interactions between articular cartilage and cationic antibodies, antigens and immune complexes. Arthritis Rheum 30:1265