Rheumatoid arthritis and the complement system

Abstract

Complement activation contributes to a pathological process in a number of autoimmune and inflammatory diseases, including rheumatoid arthritis (RA). In this review we summarize current knowledge of complement contribution to RA, based on clinical observations in patients and in vivo animal models, as well as on experiments in vitro aiming at elucidation of underlying molecular mechanisms. There is strong evidence that both the classical and the alternative pathways of complement are pathologically activated during RA as well as in animal models for RA. The classical pathway can be initiated by several triggers present in the inflamed joint such as deposited autoantibodies, dying cells, and exposed cartilage proteins such as fibromodulin. B cells producing autoantibodies, which in turn form immune complexes, contribute to RA pathogenesis partly via activation of complement. It appears that anaphylatoxin C5a is the main product of complement activation responsible for tissue damage in RA although deposition of membrane attack complex as well as opsonization with fragments of C3b are also important. Success of complement inhibition in the experimental models described so far encourages novel therapeutic approaches to the treatment of human RA.

• Complement inhibitors
• complement system
• rheumatoid arthritis

Abbreviations
CII	=	type II collagen
C5aR	=	C5a receptor
CAIA	=	collagen‐antibody‐induced arthritis
CIA	=	collagen‐induced arthritis
CR1 (2, 3)	=	complement receptor 1 (2, 3)
CVF	=	cobra venom factor
FH	=	factor H
FM	=	fibromodulin
G6PI	=	glucose 6 phosphoisomerase
IC	=	immune complex
IgG	=	immunoglobulin G
IL‐1	=	interleukin 1
LRR	=	leucine‐rich repeat (protein family)
MHC	=	major histocompatibility complex
MAC	=	membrane attack complex
MBL	=	mannan‐binding lectin
RA	=	rheumatoid arthritis
RF	=	rheumatoid factor
TNF‐α	=	tumor necrosis factor

The complement system

The human complement system was first described more than a century ago as a bactericidal aid to antibodies, and most of its components have been known for over 30 years. This has allowed thorough investigation of various physiological functions of complement. The complement system is a part of innate immunity that is able to remove pathogens without need for previous exposure. Human complement not only protects against invading pathogens due to its opsonic, inflammatory, and lytic activities but also contributes to the regulation of other biological systems, particularly the adaptive immune response 1–3.

The complement cascade can proceed by the classical, the alternative, and the lectin pathways. The classical pathway is initiated by several mechanisms such as by clustered antibodies bound to their targets, the lectin pathway begins due to recognition of certain saccharides by mannan‐binding lectin (MBL) or ficolins, and the alternative pathway is started by autoactivation of unstable complement factor C3 and its subsequent deposition on activating pathogen surfaces. Central to complement activation is the formation of enzymatic complexes such as C3‐ and C5‐convertases following cleavage of proenzymes leading to their activation and concomitant release of chemoattractant anaphylatoxin fragments. Upon activation of C5, the C5a fragment is released and acts as chemoattractant for neutrophils, monocytes, macrophages, and eosinophils, induces production of cytokines and other proinflammatory mediators, and enhances vascular permeability 4. Similar but less pronounced effects are mediated by C3a 5. The other product of C5 activation, C5b, initiates assembly of membrane attack complex (MAC), which can further contribute to inflammation due to its ability to lyse cells and also to induce drastic changes in cells when present at sublytic concentrations 6–8. Complement factors involved in the complement cascade are controlled by a number of inhibitors, which are described in more detail below. A schematic illustration of events in the complement system is presented in Figure 1.

Figure 1. Scheme of the complement system and its inhibitors. Three pathways by which the human complement system can be activated and their physiological effects: 1) clearance of apoptotic cells, 2) opsonization of pathogens and immune complexes for phagocytosis, 3) release of anaphylatoxins and lysis. Furthermore, sites of action of soluble (italics) and membrane‐bound (regular font) complement inhibitors are indicated.

Major progress has been made regarding understanding of various functions of complement from the pathology of conditions with deficiency of various components of the complement cascade. These are often associated with recurrent infections with encapsulated bacteria and immune complex diseases such as systemic lupus erythematosus and glomerulonephritis 9. However, deficiencies and defects in complement inhibitors also cause diseases such as atypical hemolytic uremic syndrome 10,11, membranoproliferative glomerulonephritis 12 and age‐related macular degeneration 12,13. In general, the complement system is a double‐edged sword. Misguided or excessive complement activation contributes to pathogenesis of glomerulonephritis 14, ischemia/reperfusion injury 15, rheumatoid arthritis (RA) 16,17, multiple sclerosis 18, and many more acute and chronic inflammatory diseases.

Considering the rapidly activated and potentially destructive character of the complement cascade, there is an obvious need for inhibitors for protection of own tissues of the host. Complement inhibitors are either soluble or membrane‐bound and act at various levels of the complement cascades. There are four principal kinds of complement inhibitors—the first one is a proteinase inhibitor, C1 inhibitor, which belongs to the serpin family 19. The second type is serum carboxypeptidase N, which proteolytically inactivates anaphylatoxins generated from C5 and C3 20. The third is CD59 (protectin), a glycosylphosphatidylinositol (GPI)‐anchored cell surface inhibitor of MAC 21. Finally a number of inhibitors act on C3‐ and C5‐convertases either by increased dissociation of enzyme complexes (acceleration of decay) or by promoting enzymatic degradation of C3b or C4b mediated by the serine proteinase factor I 21,22. The latter inhibitors are composed almost exclusively of 4–35 complement control protein (CCP) domains 23.

All these CCP‐containing complement inhibitors are encoded by genes on chromosome 1. The gene cluster is referred to as the Regulators of Complement Activation (RCA) cluster 24. C4b‐binding protein is the major soluble inhibitor of the classical and lectin pathways, whereas factor H (FH) inhibits the alternative route 25. In recent years, a whole family of FH‐like and related molecules has been identified, but it is still unclear if the major role of these molecules is to inhibit complement 26,27. Membrane‐bound complement inhibitors belonging to RCA include complement receptors 1 (CR1) and 2 (CR2), membrane cofactor protein (CD46), and decay‐accelerating factor (CD55).

Several complement inhibitors have already gained recognition as potential therapeutics due to their cardioprotective roles 28, anti‐inflammatory actions 29, and their potential to counteract paroxysmal nocturnal hemoglobinuria 30 or hyper‐acute rejection during xenotransplantation 31. However, although complement inhibitors have therapeutic potential, a major obstacle is that our knowledge of the intermolecular interactions involving complement regulatory proteins is limited.

Key messages

• Clinical observations in patients and in vivo animal models show that activation of the complement system contributes to a pathological process in rheumatoid arthritis.

• The arthritic joint contains several compounds like deposited autoantibodies, immune complexes, dying cells or exposed cartilage proteins capable of activating complement and thus mediating local inflammation mainly in a anaphylatoxin‐dependent manner.

• Complement inhibition in experimental models of rheumatoid arthritis results in improved clinical score and reverses even established disease.

Rheumatoid arthritis

RA is a chronic inflammatory disease affecting approximately 1% of the adult human population with three times more females than males affected 32. RA is associated with persistent inflammatory polyarticular synovitis affecting mainly peripheral joints. In the early stages of disease there is proliferation and edema of synovial lining cells. The synovial lining of the affected joint is infiltrated by large numbers of T and B cells, macrophages, and granulocytes. Normally thin synovium becomes thicker resulting in swollen and tender joints. As the disease progresses, the synovium invades the joint cartilage and bone to form the ‘pannus’ 33. This causes irreversible joint destruction through bone resorption by osteoclasts and breakdown of bone, in particular beneath the cartilage. Release of many proteolytic enzymes such as metalloproteinases, aggrecanases, and cathepsins leads to degradation of a number of matrix constituents including proteoglycans in both bone and cartilage 34. Furthermore, there is neovascularization of the synovial lining surrounding the joint as well as of the pannus 35. As results of these processes, the joint becomes deformed, unstable, inflamed, and painful causing major disability of the patient. Cartilage as well as bone are affected, with destruction of normal architecture and function 36.

The pathogenesis of RA is not yet fully understood. RA has a clear genetic predisposition and part of the disease development is contributed by HLA‐DR alleles within the major histocompatibility complex (MHC) 37, implicating an important role for T cell autoimmunity. However, RA is a polygenic disease with a complex genetics, and it is likely that many genes in the genome will be found to contribute to the pathogenesis 38. In monozygotic twins the RA concordance rate ranges from 12% to 15%. When compared to disease incidence in the general population, this high value predicts that genetic predisposition could be a dominant factor in RA susceptibility. However, for dizygotic twins and non‐twin siblings the concordance rate was 2%–4% implying that there are multiple genes contributing to the genetic predisposition 39. Also environmental factors like exposure to tobacco smoke, mineral oil, or silica dust have been shown to increase the risk of developing RA 40.

Evidence for complement involvement in RA from patients

RA is considered to be an autoimmune disease since the presence of autoantibodies to a variety of antigens has been reported for patients. The classical serological marker of RA is rheumatoid factor (RF), an autoantibody that reacts with Fc parts of immunoglobulin G (IgG) 41. Detection of RF is one of the diagnostic criteria of RA but has relatively low specificity as it can be detected in healthy individuals as well as in chronic infections and other inflammatory conditions. Importantly, the titers of RF have been shown to predict the development of RA 42. However, despite intense investigations there is no strong evidence that RF has any pathogenic role. More recently, it has been found that antibodies specific for protein epitopes containing a citrullinated arginine side chain have a remarkably high sensitivity and specificity to RA. In fact, these antibodies can be detected many years before the clinical onset of RA 43. It is, however, unclear whether such antibodies are a result of or cause inflammation. There is a number of other autoantibody responses reported with lower degrees of sensitivity 44. Antibodies specific for type II collagen (CII) can be detected in 10%–50% of RA patients and has clearly been shown to be pathogenic as such antibodies induce arthritis in mice. Antibodies to glucose 6 phosphoisomerase (G6PI) may occur in patients with very severe arthritis. Importantly, both CII and G6PI are specifically targeted in the joints by antibodies and these antibodies trigger arthritis after binding to the joint cartilage. Taken together, pathogenesis of RA involves a wide range of antibodies reacting with antigens in the joints and with the potential ability to form immune complexes (IC) within cartilage and in the synovial pannus tissue. IC activate complement, become opsonized by early complement components and finally taken up by phagocytes carrying the complement receptors 45. Indeed, elevated levels of complement activation products and increased consumption of C3 and C4 can be detected in synovial fluids of patients suffering from RA 46–52. Therefore complement activation by IC in the arthritic joint is likely to be a mediator, which attracts effector cells (Figure 2). Activation of infiltrated macrophages, mast cells, fibroblasts, and granulocytes by proinflammatory cytokines leads to their degranulation and subsequent detrimental release of proteolytic enzymes with ensuing joint erosion 34,53,54. Furthermore, in vitro studies show that sublytic concentrations of MAC can stimulate collagenase production in synovial fibroblasts 55. Also C3b and its degradation product iC3b can activate complement receptors CR1 and CR3 normally present on synovial fluid neutrophils. These receptors are upregulated in RA as well as in other forms of arthritis 56. Complement activation in RA is likely to cause local inflammation due to release of small anaphylatoxins C3a and C5a. A study with 41 participating RA patients showed a 7‐fold higher concentration of C3a in synovial fluid compared to patients with degenerative joint disease or traumatic arthritis 51. Another study performed on synovial fluid from 22 patients demonstrated C5a concentration sufficient to induce two of the characteristic features of the acute inflammatory phases of RA: neutrophil accumulation and leakage of plasma proteins from microvessels 52. Also the C5a receptor (C5aR) was found on synovial macrophages and on synovial fibroblasts from RA patients. Analysis of the relationship between C5aR expression and clinical data showed significant correlation with the number of swollen joints 57.

Figure 2. Schematic diagram of possible contributions of the complement system to development of rheumatoid arthritis(RA) in a joint. Complement can be activated in several ways in the RA joint. Examples are deposited autoantibodies and immune complexes (IC), apoptotic and necrotic cells, certain cartilage molecules such as fibromodulin (FM) exposed upon initial cartilage damage. Activation of complement leads to a release of proinflammatory anaphylatoxin C5a that attracts and activates neutrophils and macrophages. These cells in turn release proteases and cytokines attracting T and B lymphocytes and other inflammatory cells, further fuelling the process of inflammation. As a result of complement activation membrane attack complex (MAC) is formed and has a plethora of effects on cells even at sublytic concentrations.

Complement components resulting from early events in the classical pathway activation were shown to reflect disease activity since the plasma level of C1q‐C4 complex was much higher in patients with active RA than in those with clinical remission 58. Synovial fluid of RA patients contains high concentrations of microparticles originating from apoptotic leukocytes 59, which activate the classical complement pathway, as shown by studies of Ig, C1q, C3, and C4 deposition on the microparticles 16. Noticeably, there was a positive correlation between circulating IC and C4/C4b but not C3/C3b 60. However, complement activation during RA seems not to be limited to the classical pathway as there is evidence that the concentration of Bb fragments generated during formation of the C3 convertase of the alternative pathway also increases in the synovial fluid of affected patients 48. It is still discussed which complement activation pathway is the most relevant in RA. Some data suggest that in juvenile RA the alternative pathway is the main contributor 61, and factor Bb concentrations were found to correlate better with the level of circulating IC 62. One possible explanation depends on activation of the alternative pathway by IC bearing IgA RF. Indeed, direct activation of the alternative pathway by IgA has been described 63, and high‐molecular‐weight IgA RF‐containing IC have been reported in children with polyarticular disease 64 as well as adults with RA 65. Another explanation relates to the fact that the alternative pathway can be activated directly by collagen type II 66, which is specific for cartilage and becomes exposed as a result of proteolysis during the course of disease. Furthermore, one has to remember that the alternative pathway always acts as an efficient amplification loop for the classical one at the stage of generation of C3b. Therefore the alternative pathway is crucial also for processes that were initiated via activation of the classical pathway.

Relatively little is known about contributions of the lectin pathway to RA. Among other sugar moieties of glycopeptides, MBL is able to bind agalactosylated glycoforms of IgG (IgG‐G0) and certain glycoforms of IgM, which have a high incidence of glycans terminated by GlcNAc. In RA, serum IgG‐G0 concentrations can increase significantly compared to healthy individuals 67. When sera of RA patients and healthy donors were examined for MBL‐binding proteins, IC containing both IgG and IgM‐RF were identified in patients' sera whereas only IgGs were found in controls. Once bound to MBL, these IC could consume C4 68. These results suggest that the lectin pathway may contribute to inflammation during RA. However, the clinical data obtained so far have not verified this theory, since MBL insufficiency is positively associated with clinical symptoms and radiographic signs indicating a poor prognosis 69. Also, RA patients with MBL deficiency are younger at disease onset 70. This can perhaps be explained by the fact that MBL is one of the molecules recognizing late apoptotic and necrotic cells 71. In the absence of MBL, clearance of apoptotic cells and necrotic debris in the inflamed joint may be impaired leading to exacerbation of the disease.

It is worth mentioning that the synovial membrane is especially vulnerable not only to damage mediated by attracted immune cells but also by MAC formation and lysis of synovial cells. Major complement‐mediated lysis inhibitor CD59 is underexpressed in synovial‐lining cells compared to nearby stroma or endothelium. This is reflected by histological evidence from RA patients, where C5b‐9 deposition is located mainly in the synovial membrane but not in the vessel wall 72.

Evidence from genetic and pharmacologic studies in animal models

The importance of complement for RA in animal models was first implied when it was found that rats temporarily decomplemented with cobra venom factor were resistant to collagen‐induced arthritis (CIA) 73. CIA is perhaps the most widely used model, which meets many of the criteria for diagnosis of RA 74. A well established strategy to explore the genetic causative factors and molecular mechanism for common disease is to use inbred animal models. Various genetic linkage studies have located more than 30 genomic loci potentially harboring genes predisposing to CIA and other models for human RA in mice and rats. The Cia2 and Cia9 loci were identified as the main genetic factors influencing the arthritis development in an F2 intercross between the arthritis susceptible strain C57Bl/10.Q (B10.Q) and the resistant NOD.Q strain 75. The Cia2 locus associated with CIA was also identified in (DBA/1×SWR/J) F2 crosses 76 and also associated with arthritis in the K/B×N arthritis model in a cross between B6 and NOD 77. The Cia2 locus harbors a strong candidate gene for arthritis susceptibility, the complement component C5 encoding gene Hc. CIA was also used to show that SWR mice expressing the I‐A^q susceptibility gene but naturally lacking Hc gene were resistant to the disease 78. Both the NOD and the SWR/J strains carry a 2 bp deletion in an exon near the 5' end of the Hc gene resulting in frame‐shift and termination resulting in C5 protein deficiency 79,80.

Gene inactivation technology was also used to ascertain the role of complement in arthritis. Genetic deletion of C5, C3, or factor B in DBA/1 mice in each case resulted in mice being resistant to CIA despite high titers of anti‐collagen type II antibodies 81,82. C3−/− mice developed mild arthritis whereas arthritis in FB−/− mice was intermediate with regard to C3−/− and control mice. In other studies no role for the classical pathway was found, and it was postulated that only the alternative pathway operated to mediate arthritis in CIA 83. Thus, the relative importance of the classical versus the alternative pathway in CIA is not yet established. C3 and factor B knockout mice were shown to be resistant to induction of disease in the anti‐collagen antibody‐induced arthritis (CAIA) model 84. This model bypasses the complexity of CIA, which consists of both the priming and effector phase by focusing on the effector phase only as it is induced with monoclonal antibodies without the need for T cells or B cells 85. Regarding C5 component, both deficiency and systemic administration of anti‐C5 antibodies ameliorated CIA 86. The antibodies prevent cleavage of C5 and release of C5a. C5a induced chemotaxis as well as a wide array of effects such as increase in vascular permeability and cellular degranulation. Importantly, antibody against C5a prevented not only arthritis but also reversed ongoing disease when injected several days after disease onset 86.

Recently another model of RA was used to systematically assess the role of complement. K/B×N mice produce high titers of IgG1 antibodies to G6PI and develop arthritis spontaneously 87. Injection of serum from sick into healthy animals provoked arthritis with clinical and histological symptoms similar to human RA 88. No difference was induced by crossing the K/B×N mice with C1q and C4 knockout mice showing that the classical pathway is not crucial for disease progression in this model. However, progeny of K/B×N mice and knockout mice deficient in factor B, C3, C5, or C5aR were completely resistant to arthritis. It appears that this model is dependent on the alternative pathway 87. An important difference in this model, as compared with CAIA, is that the serum from K/B×N mice contains antibodies to G6PI predominantly of the IgG1 isotype, which do not activate complement, whereas CAIA is induced by monoclonal antibodies of different isotypes including IgG2a/c and IgG2b 85. Most importantly, it was observed that C6 deficiency did not influence disease progression in this model, showing that MAC is not a mediator of the tissue damage 87. Table I summarizes the evidence for involvement of complement in RA observed in animal models.

Table I. Evidence for involvement of complement in rheumatoid arthritis (RA) observed in animal models.

Animal	Type of intervention	RA model	Observed effect	References
mouse	Natural deficiency of C5	CIA	Resistance to disease.	78,81,127
mouse	Crossing C5‐deficient and sufficient strains	CIA	No arthritis in C5−/− population, lower incidence of arthritis in C5+/− than in C5 +/+ mouse.	76
mouse	Anti‐C5 antibody	CIA	Prevention of disease onset after immunization, amelioration of established disease.	86
mouse	C1q, C3, C4, factor B, C5, C6, and C5aR knockout or natural deficiency	K/B×N	No arthritis upon passive transfer of serum in C3‐ or C5‐deficient mouse, little manifestation of disease in factor B knockout, normal clinical onset in C1q, C4, and C6 knockout.	77,87,128
mouse	CR1 and CR2 knockout	K/B×N	Normal clinical onset of disease.	128
mouse	C3 knockout	CIA	Resistance to CIA and decreased collagen II‐specific IgG Ab compared to control mouse after first immunization. Second immunization resulted in lower clinical score.	82
mouse	C3 knockout	CAIA	Clinical score of disease markedly decreased. Delayed onset of disease and lower clinical score.	83
				84
mouse	Factor B knockout	CIA	Partial resistance to CIA and decreased collagen II‐specific IgG Ab compared to control mouse after the first immunization. Second immunization resulted in intermediate clinical score.	82
mouse	Factor B knockout	CAIA	Clinical score of arthritis decreased. Decreased clinical score but not as much as in C3 knockout mouse.	83,84
mouse	Injection of sCR1‐expressing arthritogenic splenocytes or fibroblasts	CIA	Inhibition of clinical symptoms, anti‐collagen II antibody production and T cell response to collagen II in vitro.	129
mouse	Cobra venom factor (CVF)‐mediated decomplementation	Bacterial cell‐wall induced arthritis	Diminished incidence and severity of disease	130
mouse	CVF‐mediated decomplementation	AA	Silenced clinical manifestation of arthritis	131
rat	C5aR antagonist or sCR1 i.a. administration	AIA	sCR1 but not C5aR antagonist caused suppressed synovium thickening, joint swelling, and inflammation.	132
rat	CVF‐mediated decomplementation or i.v. application of sCR1	CIA	Delayed disease onset.	133
rat	Systemic or local decomplementation by CVF or i.a. application of sCR1	AA	Reduced joint swelling after local but not systemic decomplementation before disease onset.	134

CIA = collagen induced arthritis; CAIA = collagen antibody‐induced arthritis; AA = antigen arthritis; AIA = antibody‐induced arthritis; i.a. = intraarticular; i.v. = intravenous.

Cartilage molecules contributing to the regulation of complement activation in RA

A well established clinical observation, pertinent to understanding of mechanisms in joint disease, is that joint replacement surgery ameliorates inflammation. It appears that molecules present in or released from cartilage may have a role in propagating this inflammation by activating complement. One candidate is fibromodulin (FM) 89.

Cartilage contains low numbers of cells separated by a predominating extracellular matrix with several distinct supramolecular assemblies. These include extremely negatively charged proteoglycans (aggrecan) entrapped in networks of collagen fibers. The latter contain many additional molecules bound at the fiber surface to provide stability and interactions with surrounding structures. Among these molecules there are members of the leucine‐rich repeat (LRR) protein family, such as decorin, biglycan, osteoadherin, FM, lumican, and chondroadherin 90. Some LRR proteins, present in several tissues, including cartilage, were found to interact with complement factors. Thus, C1q binds to decorin 91 and biglycan 91 as well as to non‐related fibronectin 92 and laminin 93. However, in none of these examples does the interaction lead to activation of the complement cascade.

FM was first described as a 59‐kDa protein with substituents of highly anionic keratan sulphate glycosaminoglycan chains. The protein interacts with collagen types I and II 94. It localizes to collagen fibers in cartilage 95. FM is thought to play an important role in collagen fiber formation as shown by the observation that FM knockout mice form abnormal collagen fibrils in e.g. tendons 96. During inflammatory joint disease proteinases, particularly of the metalloproteinase family (MMPs) degrade cartilage tissue molecules. In this process an initial event is liberation of the N‐terminal portion of FM by the actions of MMP‐13 97, while at later stages the remainder of the molecule is released.

FM interacts with the globular heads of C1q. This interaction, typical for ligands like IC, leads to the activation of complement. FM mainly activates the classical pathway of complement 89. The activation initiated by FM results in pronounced deposition of C3b and C4b but less pronounced deposition of MAC. The failure to induce MAC deposition is most probably due to binding between FM and the complement inhibitor FH. C1q and FH bind to non‐overlapping sites on FM, and none of these interactions appear to be mediated by the highly polyanionic polysaccharide keratan sulfate decorating FM 89. Taken together, it appears that FM is one of the factors fuelling joint inflammation when the initial damage to the cartilage has occurred.

Interestingly, decorin 91,98 and biglycan 91 bind C1q, but instead of triggering complement activation they inhibit the ability of C1 to activate the classical pathway. These interactions take place at physiological ionic strength and may be relevant for regulation of complement activation and C1‐mediated effects on cells in the inflamed joint 91. One putative mechanism involves the large and extended anionic glycosaminoglycan chain, which may modulate other interactions, also at a distance from its binding site, with C1q. Immunohistochemical studies revealed that decorin concentrates in the same region of articular cartilage where IC and C3 have been found 98,99. Moreover, biglycan was shown to inhibit C1q‐induced production of proinflammatory cytokines in endothelial cells 91. Summing up, there are several molecules in cartilage able to bind C1q. While one of them (FM) activates complement the others may act as local complement inhibitors, and the final outcome will depend on many local parameters, such as how fragmentation of proteoglycans during the tissue destruction will affect the binding site and possibly separate a binding site from e.g. an anionic domain modulating interactions and effects.

RA, B cells, complement, and cell death

Substantial numbers of B cells can be found within the infiltrate attracted into RA joints. B cell infiltrates may either be diffuse and lack any structural organization such as form small clusters and large follicle‐like structures 100. Most of these B cells can take part in antigen‐driven immune response and differentiate into plasma cells 101. Since the RF present in arthritic joints form IC opsonized with complement, such complexes are internalized by B cells, which in turn generate a broad spectrum of peptides enhancing the antigen‐presenting function 102. In this way a positive feedback increasing the production of high‐affinity antibodies with RF specificity could develop, thus sustaining the chronic inflammatory process in RA 100. The role of B cells in the context of autoimmune disease seems to be important not only due to Ig production but also because of their antigen‐presenting function. This was confirmed in a mouse lupus model, in which MRL/lpr mice that transgenically expressed surface immunoglobulins, but did not permit the secretion of circulating Ig, did develop nephritis 103. Moreover, another study showed that fibroblast‐like synoviocytes are able to bind B cells and prevent them from entering into apoptosis. This phenomenon was observed only for synoviocytes from RA patients but not patients with other types of arthritis 104.

Since a role of B cells in autoimmune disease was suggested, B cell depletion therapy was proposed for patients with RA 105. One of the targets of such therapy is the CD20 molecule present on more than 95% of normal or neoplastic B cells 106 but not stem cells and B cells differentiating into plasma cells 107. Very promising results of B cell depletion therapy in RA patients were achieved with a genetically engineered, chimeric anti‐CD20 monoclonal antibody—rituximab 107. A study performed on more than 100 patients subjected to a single short course of treatment with rituximab, either alone or in combination with cyclophosphamide or continuing methotrexate, provided significant improvements in disease symptoms. The mechanism suppressing arthritis is, however, not yet clarified and may involve decrease of pathogenic antibodies or may be due to other B cell functions than the production of antibodies. The mechanisms whereby the injected antibodies delete the B cells seem to be better clarified and complement appears to be involved in B cell elimination. Several observations testify to this theory: 1) After ligation of CD20 with mAb cell death is independent of caspases 108; 2) treatment with rituximab did not cure C1q‐deficient mice bearing human CD20‐positive lymphomas 109; 3) elevated expression of CD55 and CD59 was found on cells from patients not responding to rituximab 110; and 4) substantial complement consumption was observed during rituximab treatment 111.

Taken together, B cell activity in RA appears to sustain the process of acute inflammation already initiated in the joint tissue. However, depletion of reactive B cells proves to be promising in treatment, and is at least able to diminish activity at clinical onset or delay it. When mAb directed against B cells are introduced, their therapeutic potential is likely to be dependent on the decrease of production of pathogenic antibodies and the downstream dependence on complement.

Developing therapeutic strategies based on complement contribution to RA

Advances in understanding the roles of cytokines are directed to new developments in the treatments available to RA patients. Nowadays there are several approved compounds based on monoclonal antibodies or binding proteins, which target inflammatory cytokines. The anti‐tumor necrosis factor (TNF‐α) clinically approved drugs are infliximab (Remicade), etanercept (Enbrel), and adalimumab (Humira), and the interleukin 1 (IL‐1) inhibitor is anakinra (Kineret) 112. Anticytokine therapy in RA was found to reduce the symptoms of inflammation as well as slow or stop structural joint damage 113. However, although targeting of these key cytokines is very efficient in the two‐thirds of patients that react, they have inherent complications as these factors are important in immune defense and their inhibition compromises immune regulation with ensuing rare infections 114, lupus‐like syndrome, or re‐activation of chronic hepatitis B 115. TNF‐α antagonists should be avoided in patients with a preexisting demyelinating disease 115. Considering these numbers of restrictions, new drugs targeting upstream mediators of joint inflammation would provide new openings for more specific therapy.

As the complement system is one of the significant cascades generating inflammation in the effector phase of RA, it is feasible that complement inhibitors will be able to ameliorate the disease. Up to date, several types of compounds interfering with the complement cascade were extensively studied and entered into clinical trials (for review see 116). One example is the soluble CR1 (sCR1) receptor derivatives. Membrane‐bound CR1 acts as cofactor for factor I‐mediated cleavage of C3b or C4b and decay‐accelerating factor for classical and alternative C3 and C5 convertases 117. The protein is expressed mainly on erythrocytes and other blood cells. Negligible quantities (up to 30 ng/mL) of soluble CR1 are found in blood 118. Since sCR1 retains its powerful complement inhibitory activity, it appeared as potential candidate for therapeutic complement regulator. Two strategies were employed to improve and target the CR1 potential to cells under complement attack. First was the decoration of sCR1 with sialyl‐Lewis^x tetrasaccharides. This modification introduced the affinity to selectins, which are overexpressed by cells at the sites of inflammation 29. Another modification concerns truncation of sCR1 to three N‐terminal CCP domains and coupling to a membrane‐targeting peptide 119. This compound, named APT070 was successfully tested in a rat model of antigen‐induced arthritis 119. TP10, a sCR1‐derived compound, was tested in patients with myocardial infarction, acute respiratory distress syndrome, and postcardiopulmonary bypass syndrome 120,121. A second‐generation of this drug named TP20 was tested successfully in animals 122. Another anticomplement drug, monoclonal antibody against C5 (eculizumab, 5G1.1), targets the C5 molecule in a way which disables its splitting into C5a and C5b, thus preventing both enhancement of inflammation by anaphylatoxin release and formation of MAC 123. Previous studies showed that mouse anti‐C5 Ab (BB5.1), which is functionally related to eculizumab, gave promising results in CIA 86. It inhibited both the onset of disease after immunization and ameliorated already established disease. Up to date, the most beneficial application of eculizumab was reported in patients suffering from paroxysmal nocturnal hemoglobinuria 30. Furthermore, the results of eculizumab's clinical phase II studies in RA patients are available 124. They revealed significant improvement in RA clinical score when the drug was given intravenously at certain intervals for 3 months. The best results were obtained for patients who exhibited high basal level of C5b‐9 in serum 125.

Additionally to those anticomplement drugs already tested clinically, there are new compounds targeting the complement system, and some of them have passed trials in animal models of arthritis. They can be grouped into several types, e.g. serine protease inhibitors (FUT‐175, BCX‐1470), C3 inhibitors (compstatin), C5 inhibitors (TS‐A12/22, K76COONa), anaphylatoxin receptor antagonists (SB290157, AcF‐[OPdChaWR], PMX‐53); for review see 126. The most promising therapeutic approaches targeting complement for RA treatment which are presently under development are listed in Table II.

Table II. Therapeutic approaches to complement inhibition in rheumatoid arthritis.

Approach/agent	Company	Stage of development	Ref/homepage
Clinical and pre‐clinical phase:
sCR1‐related compound APT070	Inflazyme Pharmaceuticals	Clinical phase I/II	http://www.inflazyme.com/files/April%207%202004.pdf
Anti‐C5 eculizumab	Alexion Pharmaceuticals	Clinical phase IIb	124,125
C5a antagonist PMX‐53	Peptech	Clinical phase II	3
Vaccination against C5a	Resistentia Pharmaceuticals	Pre‐clinical development	http://www.resistentia.se/se/researchdevelopement/c5a/
Anti‐CD59 CD59‐Prodaptin^TM	Inflazyme Pharmaceuticals	Pre‐clinical development	http://www.inflazyme.com/files/July%2027%202004.pdf
Animal trials:
Serine protease inhibitor. FUT‐175	Torii Pharmaceutical Company	Tested in rat adjuvant arthritis model	135
C3aR antagonist SB290157	–	Tested in rat adjuvant arthritis model	136
C5aR antagonist AcF‐[OPdChaWR]	–	Tested in rat antigen arthritis model	137
Hybrid Ig‐CD55 molecule	–	Tested in rat antigen arthritis model	138
Hybrid of CD59 and membrane‐address tag	–	Tested in rat antigen arthritis model	139

While there is compelling evidence for great benefits of therapeutic complement inhibition, recent advances indicate that complement components are multifunctional. Many of these components are critical for the proper function of mammalian cells and tissues. Thus, prior to clinical translation an understanding of the proper targeting, timing, duration, and potency of anticomplement agents is needed in order to avoid interfering with critical processes related to tissue repair, healing, and function. A putative, ideal therapy based on complement inhibition would act only locally in affected tissue or act systemically but keep the complement activity at the level sustaining its other functions. Systemic complement depletion can be dangerous since it is clear that the complement cascade is made up of intricate systems with widely distinct effects. Any proposed therapy must take this variety into account, particularly in the case of diseases like RA that often requires lifelong treatment.