Washington University School of Medicine, St. Louis, Missouri, USA
First published December 1, 2003 - More info
The antiphospholipid syndrome is characterized clinically by fetal loss and thrombosis and serologically by the presence of autoantibodies to lipid-binding proteins. In a model of this procoagulant condition in which these antibodies are injected into pregnant mice, fetal loss was prevented by blocking of complement activation. Specifically, interaction of complement component 5a (C5a) with its receptor is necessary for thrombosis of placental vasculature (see the related article beginning on page 1644). Inhibition of complement activation may have a therapeutic role in this disease.
In the first autoimmune condition that was recognized as such over a century ago, antibody and complement were shown to lyse human red blood cells in a rare form of hemolytic anemia known as paroxysmal cold hemoglobinuria (1). Impressively, these same two factors had been found to mediate lysis of bacterial pathogens just three years earlier in 1892. While lysis was the end point in those early studies because it was both pathologically meaningful and easy to monitor, we now realize that it is the phlogistic and opsonic activities of complement that account for much of its physiologic and pathologic importance. The antiphospholipid syndrome (APS) is an autoimmune disease in which autoantibodies are synthesized to lipids and lipid-binding proteins. These antibodies bind to antigens expressed by endothelial cells and produce a clinical syndrome featuring thrombosis. Fetal wastage is caused by placental infarction. In this issue of the JCI, Girardi et al. report that activation of the complement cascade by the immune complexes is required for disease development in a mouse model of APS (2).
Rheumatoid arthritis. In one model, that of a spontaneously arising inflammatory arthritis, autoantibodies reactive to the ubiquitously expressed intracellular protein glucose-6-phosphate isomerase are synthesized (3–5). These IgG autoantibodies bind to this autoantigen in the joint and trigger a destructive arthritis. Additionally, an acute transient form of synovitis can be produced by the passive transfer of antibody alone. While antibody-mediated activation of the classical pathway is a time-honored concept, antibody can also serve as a site that is protected from plasma and membrane inhibitors of complement activation. In particular, these regulators of complement activation are designed to block amplification of the alternative pathway (AP) on self tissue (6). To produce arthritis, antibody binding at a site relatively deficient in complement inhibitors allows “firing” of the AP (5). Complement component 5a (C5a), interacting with its G-coupled receptor, is absolutely required to recruit neutrophils and to activate nearby cells (5). Based on depletion studies and knockouts, neutrophils, mast cells, and Fc receptors are also necessary in this model of rheumatoid arthritis (RA) (5, 7).
Antiphospholipid syndrome. In the second mouse model of autoimmunity, antiphospholipid antibodies derived from patients are injected into pregnant mice. Remarkably, these antibodies react with lipid-binding proteins on endothelial cells to trigger complement activation, thrombosis, and fetal death secondary to placental insufficiency (8–11). These pathologic antibodies attach to lipid-rich antigens on the developing placental vasculature and then bind complement (12–15). The complement system in turn damages the endothelial cells, leading to a procoagulant state and undesirable thrombosis. In this issue of the JCI, Girardi et al. (2) convincingly demonstrate that both the classical and alternative pathways are required to generate sufficient C5 cleavage to cause fetal loss. The released C5a binds to its receptor on neutrophils and endothelial cells. In this model, neutrophils are again required but Fcγ receptors are not. Simply put, a procoagulant state is set up, clots form, placental infarction ensues, and embryos die.
The complement-activation scheme is different in the two models. In the APS model, the classical pathway triggers the process by depositing a relatively small amount of C3b on the target, and then this C3b serves as a nidus for the AP’s self-amplifying feedback loop (Figure 1). The AP deposits the majority of the C3b and cleaves most of the C5. Another particularly intriguing aspect of this model is the indication that neutrophils enhance the complement-activation process. In the RA model, the AP alone can provide the necessary C3b and subsequent C5 activation (5). As noted, plasma and membrane complement regulators are particularly effective at blocking the feedback loop of the AP (16, 17) but are less able to control antibody-mediated classical-pathway activation. Thus, this potentially lethal amplification process occurs routinely and robustly against foreign materials (which in most cases lack regulators). This line of reasoning accounts for why large amounts of autoantibodies that activate the classical pathway are so damaging, yet low levels of autoantibodies or autoreactive lectins are so well tolerated.
C3b can be deposited on a target by immune complexes that commonly engage the classical pathway, by lectins such as mannose-binding protein that bind to sugars, or through the continuous low-grade turnover of the AP. The deposited C3b serves as a nidus for amplification of C3b deposition via the positive feedback loop. Control of this amplification process is critical to prevent undesirable injury to host cells. The presence of antibodies or lectins on a membrane at a site relatively lacking in membrane regulators is conducive to excessive complement activation on self tissue.
Consequently, for excessive and unwanted complement activation to occur on self tissue, something is likely to be amiss relative to the host’s complement inhibitors (see Requirements for complement activation on self tissue in autoimmunity). This point is clearly illustrated in the arthritis model, where the evidence points to antibody binding and complement activation taking place at a site relatively deficient in regulators (5). Furthermore, the autoantibodies themselves serve as a protected site for complement activation (6). Genetic deficiencies of complement inhibitors reinforce these points. For example, in paroxysmal nocturnal hemoglobinuria, red blood cells are lysed because of an acquired clonal stem cell inactivation of the enzyme required to anchor two regulators of complement on the erythrocyte membrane (18). In atypical (nondiarrheal) hemolytic uremic syndrome, heterozygous deficiency of one of the complement inhibitors factor H or CD46 predisposes to a thrombotic microangiopathy because of excessive complement activation (17, 19). The take-home lessons from these reports are that complement activation on vessel walls must be rigorously controlled and that even heterozygous deficiency of an inhibitor predisposes to thrombosis if a vessel wall is under siege. A reasonable hypothesis, therefore, to explain clot formation in APS is that the developing placental vasculature is relatively deficient in complement-inhibitor function.
Are these pathologic situations in the mouse similar to the human diseases RA and APS? At the minimum, these studies identify signaling pathways and mechanisms associated with tissue destruction that should be further investigated in individuals suffering from RA or APS. At the maximum, they represent almost exactly what takes place in RA and APS. In the former case, one would argue that we have yet to define the autoantibody and autoantigen combination. In the latter case, one would say that the same pathologic process occurs in the human placenta and on other vascular endothelial cells to cause venous and arterial thrombosis.
What, then, are the implications for the treatment of these conditions with a complement inhibitor? The trial of an mAb that inhibits C5 was efficacious in RA but not to the same extent as inhibition of TNF (20). For APS, treatment has focused primarily on anticoagulation (21). Complement inhibition would seem to be an attractive therapeutic target in some of these patients. In particular, C5a and its receptor should now become prime targets for inhibition. Girardi and colleagues have substantially advanced our understanding of the APS model by showing us factors required to mediate the associated tissue injury, and this, in turn, may have therapeutic implications (2).
See the related article beginning on page 1644.
Conflict of interest: The author has declared that no conflict of interest exists.
Nonstandard abbreviations used: antiphospholipid syndrome (APS); alternative pathway (AP); complement component 5a (C5a); rheumatoid arthritis (RA).