The ability of immunoglobulin and immune complexes, including autoantibodies, to provoke inflammation stems from the interaction of the Fc portion of antibody molecules with one or both of two major effector pathways: Fc receptors and the complement system. The relative contributions of these two pathways vary among different disease states and experimental systems 123. A more detailed understanding of the mechanisms by which autoantibodies engage Fc receptors and complement to provoke pathology in a specific target tissue can permit a more tailored therapeutic intervention.

Fcγ receptors (FcγRs) recognize immunoglobulin G (IgG) and transduce either activating or inhibitory intracellular signals. In the mouse, the activating FcγRs include FcγRI, FcγRIII, and FcγRIV. The activating FcγRs share a common cytoplasmic signaling chain called FcRγ (encoded by the Fcer1g gene) responsible for signal transduction. Mice also express the inhibitory receptor FcγRIIB, whose cytoplasmic tail contains an inhibitory signaling motif. The outcome of an interaction of an FcγR-expressing cell with an IgG-containing immune complex depends on the relative expression levels of the various activating and inhibitory FcγRs and also the IgG subtype (for which the various FcγRs have differing affinities) 4.

The complement system is activated by three primary pathways (classic, alternative, and mannose-binding-lectin), each consisting of a series of serine proteases. These three activation pathways converge at complement component C3. Cleavage of complement C3 produces a C5 convertase. These events result in the generation of anaphylatoxins (for example, C3a and C5a) and formation of the membrane attack complex (C5b-9), whose main functions are to recruit inflammatory cells and to mediate cellular lysis, respectively (Figure 1) 56.

Several studies have pointed to the existence of additional, C3-independent mechanisms by which C5 can be activated to drive inflammatory responses (Figure 1). More than two decades ago, investigators described the existence of C5-C9-dependent immune hemolysis occurring in a C3-independent fashion 78. More recently, studies of IgG-triggered acute lung injury revealed that, in C3-deficient mice, thrombin can act as a C5 convertase to generate C5a and mediate pathology 9. Similar crosstalk between the complement system and coagulation systems has been identified in other model systems, including antiphospholipid antibody-induced and lipopolysaccharide (LPS)-induced fetal loss in mice 1011. An elegant in vitro study has recently confirmed that multiple serine proteases in the coagulation and fibrinolysis systems can cleave C3 and C5 to produce C3a and C5a 12. Interplay of FcγRs and the complement system also occurs, and several studies of IgG-initiated pathology have highlighted the existence of a C5a generation pathway that is triggered by activating FcγRs 2131415. Here, we investigated a possible contribution of C3-independent mechanisms of C5 activation in a mouse model of autoantibody-mediated arthritis.

K/BxN T-cell receptor (TCR) transgenic mice spontaneously develop inflammatory arthritis due to combined T- and B-cell recognition of the self-antigen glucose-6-phosphate isomerase (GPI) and production of high-titer anti-GPI IgG autoantibodies 1617. Arthritis can also be provoked by injecting serum from K/BxN mice into normal mice 18. Importantly, the requirements for complement and activating FcγRs differ between the K/BxN TCR transgenic mice and its derivative, the serum transfer model, likely reflecting the several-fold higher concentration of anti-GPI autoantibodies in the spontaneous genetic model. The development of serum-transferred arthritis depends on both activating FcγRs and the alternative pathway of complement activation. Specifically, mice with targeted or naturally occurring mutations in the genes encoding factor B (of the alternative pathway), C3, C5, the C5a receptor (C5aR), and FcRγ were protected from developing serum-transferred arthritis 192021. In contrast, we have shown that K/BxN TCR transgenic mice lacking FcRγ developed spontaneous arthritis equivalently to controls but that C5-deficient K/BxN mice developed less severe arthritis than controls 22. In addition, treating K/BxN mice with anti-C5 monoclonal antibody reduced their arthritis severity 22. Those findings led us to investigate which of the upstream C5-activation pathways drives arthritis in K/BxN mice.

We recently reported that K/BxN TCR transgenic mice genetically lacking C5 or treated with anti-C5 monoclonal antibodies developed less severe arthritis than control animals 22. To determine which of the upstream complement activation pathways were involved in arthritogenesis, we first investigated complement component C3, the convergence point for the three primary complement activation pathways and the component immediately upstream of C5. We bred K/BxN mice with a targeted deletion in the C3 gene. In agreement with a recent report, we observed that C3-deficient K/BxN mice developed arthritis equivalently to littermate controls (Figure 2) 29. This finding suggested that a non-traditional, C3-independent pathway of C5 activation was at work in these mice.

Several reports have demonstrated that engagement of activating FcγRs can trigger C5a production, even in the absence of C3 2131415. We therefore hypothesized that, in C3-deficient K/BxN mice, the absence of activating FcγRs would impair C5 production and lead to less severe arthritis. To test this hypothesis, we bred K/BxN mice genetically lacking both C3 and FcRγ, the cytoplasmic signaling chain shared by the activating FcγRs. Surprisingly, the C3/FcRγ-double-deficient K/BxN mice developed arthritis of equivalent severity to controls (Figure 3a). As expected, immunofluorescent staining of joint sections from the C3/FcRγ-deficient K/BxN mice revealed IgG deposition similar to controls, but no C3 (Additional file 1a). Treatment of the C3/FcRγ-deficient K/BxN mice with anti-C5 monoclonal antibody reduced their arthritis severity (Figure 3a). Considered together, these findings indicate that, in the genetic absence of C3, arthritis in K/BxN mice can be mediated by an FcRγ-independent C5 activation pathway.

Consistent with our prior report that arthritis in K/BxN mice depends on C5 but not FcRγ 22, we found that K/BxN mice lacking both C5 and FcRγ developed less severe arthritis than controls (Figure 3b). In the ankles of these C5/FcRγ-deficient K/BxN mice, deposition of both IgG and C3 was detectable, suggesting that C3 deposition can still occur in the absence of C5 and despite the less severe inflammatory response in these joints (Additional file 1b). The fact that the C5/FcRγ-deficient mice developed some arthritis, albeit attenuated, suggests that other minor pathogenic effector mechanisms are still operational in these mice; candidate pathways include recognition of IgG immune complexes by upstream complement components and their receptors (for example, C1qR(P)/CD93 30), the pro-inflammatory activity of C3a 6, and pathogenic effector T cells.

We performed additional studies to determine how C5 deficiency interferes with arthritogenesis in K/BxN mice. We found that the production of anti-GPI autoantibodies was unimpaired in the C3-deficient, C3/FcRγ-deficient, and C5/FcRγ-deficient K/BxN mice (Figure 4). C5a was also recently shown to drive T helper 17 (Th17) cell differentiation in another murine arthritis model 31. Th17 cells are known to be involved in the pathogenesis of arthritis in the K/BxN system 3233. We found, however, that T-cell activation as determined by high expression of CD44 and low expression of CD62L (Figure 5a) and the frequency of CD4⁺ T cells producing the cytokines IFNγ or IL-17 (Figure 5b) were unaltered in the C5/FcRγ-deficient K/BxN mice relative to controls. Collectively, our findings indicate that the T cell- and B cell-dependent events culminating in autoantibody production and pro-inflammatory cytokine production are not impacted by deficiency of C3, C5, and/or FcRγ and suggest that C5 drives arthritis in K/BxN mice via downstream effector mechanisms. Given the requirement of the C5aR but not the terminal complement components (C6-C9) for serum-transferred arthritis 20, it is most logical that it is the interaction of C5a with C5aR that drives arthritis in K/BxN mice, a hypothesis that could be tested by generating C5aR-deficient K/BxN mice. Next, we investigated how C5 was being activated in the absence of C3 and activating FcγRs.

Multiple lines of evidence point to molecular communications between the coagulation and complement cascades, both of which contain numerous serine proteases. Administration of the anticoagulant heparin blocked antiphospholipid antibody-induced complement activation and pregnancy loss in a mouse model 10. Similarly, LPS increased the expression of the membrane-bound prothrombinase FGL2, which can cleave C5 and which also has been linked mechanistically to LPS-induced abortions in mice 1134. In mice lacking C3, the expression of prothrombin was upregulated, and treatment of mice with anti-thrombin III blocked IgG-induced C5-dependent lung injury 9. We verified that the expression level of prothrombin was increased in the liver of C3-deficient mice relative to control mice (Additional file 2). In keeping with this finding, treatment of C3-deficient/FcRγ-deficient K/BxN mice with the thrombin inhibitor argatroban led to reduced arthritis severity, and discontinuation of the argatroban was temporally associated with the re-emergence of arthritis (Figure 6). We are currently investigating longer-acting compounds to determine whether thrombin inhibition acts primarily to reduce arthritis severity or to delay the onset of arthritis in this model. Thus, the generation of C5 in the absence of C3 and activating FcγRs in K/BxN mice appears to be mediated by thrombin or related serine proteases of the coagulation cascade.

Whether C3-independent pathways of C5 activation contribute to pathology in wild-type mice or only in the setting of genetic C3 deficiency is important for understanding the relevance of these pathways to human diseases. For instance, heparin was effective in reducing antiphospholipid antibody-induced fetal loss in wild-type mice by blocking C5a generation 10. Similarly, treatment with the thrombin inhibitor polyethyleneglycol-hirudin (PEG-hirudin) decreased the severity of collagen-induced arthritis in mice, although this was attributed to reductions in intra-articular fibrin deposition rather than to effects on complement activation 35. In contrast, in the IgG-mediated lung injury model, the effect of the anticoagulants anti-thrombin III and hirduin in reducing lung pathology was evident only in C3-deficient animals, leading those investigators to speculate that genetic deficiency in C3 leads to upregulation of thrombin as a compensatory mechanism to allow C5 activation via a non-traditional pathway 9. Similarly, we observed no effect of argatroban on arthritis severity in C3-sufficient K/BxN mice (data not shown), but we are currently exploring longer-acting thrombin inhibitors. Thus, the contribution of the coagulation cascade to C5 activation might vary depending on the disease model. In addition, a recent report suggests that, in vitro, thrombin cleaves C5 at a site different from that cleaved by C5 convertase, leading to the generation of novel intermediates 36. Understanding whether these intermediates are also generated in vivo and, if so, how they affect inflammation will be essential next steps.

How IgG antibodies activate thrombin in the absence of activating FcγRs and C3 remains an open and important question. Since thrombin activation by IgG has been observed in multiple autoantibody-dependent models, it is not likely that the antigenic specificity is critical. It seems more likely that other IgG-interacting molecules (for example, complement C1q or the neonatal Fc receptor) could be at play. Alternatively, antibody fragments with direct prothrombinase catalytic activity have been described 37.

From a clinical perspective, monoclonal antibody reagents designed to interfere with C5 activation systemically or locally (for example, in the synovium) might be effective treatments for inflammatory arthritis (Figure 3a) 223839. Our findings suggest that agents designed to interfere with non-traditional C5 activation pathways such as the coagulation cascade might also prove beneficial for treating inflammatory arthritis in certain settings.

The key finding of this study is that autoantibody-mediated arthritis in K/BxN mice can occur via a C5 activation pathway that requires neither C3 nor activating FcγRs, the two main effector mechanisms of IgG molecules. Genetic deficiency of C3, C5, and/or FcRγ did not affect T-cell activation or autoantibody production, indicating that the pro-arthritogenic activity of C5 is mediated by its conventional effector mechanisms (likely C5a production). Our data further suggest that thrombin or related proteases of the coagulation cascade mediate C5 activation in the absence of C3 and FcRγ. Understanding how novel pathways of complement activation contribute to autoantibody-mediated arthritis and other inflammatory disorders is expected to lead to new therapeutic approaches.

DMSO: dimethyl sulfoxide; FcγR: Fc receptor for immunoglobulin G; FcRγ: the cytoplasmic signaling chain shared by activating Fc receptors for immunoglobulin G; GPI: glucose-6-phosphate isomerase; IFNγ: interferon-gamma; IgG: immunoglobulin G; IL-17: interleukin-17; LPS: lipopolysaccharide; MAC: membrane attack complex; MHC: major histocompatibility complex; NOD: non-obese diabetic; TCR: T-cell receptor; Th17: T helper 17.

The authors declare that they have no competing interests.

JLA designed and performed experiments, interpreted data, and assisted in writing the manuscript. SSH performed experiments and assisted in writing the manuscript. BAB designed and oversaw experiments, interpreted data, and wrote the manuscript. All authors read and approved the final manuscript.