Regulation of humoral immunity by Complement

Introduction

The paradigm first suggested by Janeway over 2 decades ago, that innate immunity was essential for regulating adaptive responses, focused attention on how pathogens were recognized and taken-up by dendritic cells for presentation to T cells (Janeway, 1989). Later, Fearon and Locksley (Fearon and Locksley, 1996) proposed that B cells were equally reliant on innate recognition for acquisition of cognate antigens and that the complement system played a critical role in signaling B cell activation. More recently, in vivo imaging studies tracking B cell capture of lymph borne antigens in real time have again provided a major shift in our understanding of how the innate system regulates humoral immunity. For example, specialized macrophages lining the sub-capsular sinus of peripheral lymph nodes (pLNs) have been shown to capture both particulate (Carrasco and Batista, 2007; Junt et al., 2007) and soluble immune complexes via complement receptors (Phan et al., 2007) and display them to cognate B cells in the underlying B cell follicles. Alternatively, small protein antigens have been found to drain directly from the sub-capsular sinus into the follicles through collagen-rich conduits enveloped by fibroblast reticular cells (FRC) (Bajenoff and Germain, 2009; Roozendaal et al., 2009). While these pathways provide insight into how memory B cells can directly sample antigen within lymph nodes; they leave open the question how B cell antigens are retained and sampled over longer periods. An important clue to answering this question has come from the recent report by Cyster and colleagues as they made the striking observation that naïve B cells are highly efficient at taking up and transporting immune complexes bound to complement receptor (CR2) to B cell follicles within peripheral lymph nodes (pLN) where they are transferred to follicular dendritic cells (FDC) for long term retention (Phan et al., 2009). These exciting new findings combined with the recent solving of the co-crystal structure of CD21in complex with the complement fragment C3d (van den Elsen and Isenman, 2011) suggest a new perspective in how the complement system regulates access of antigen by B cells. In this review, we will discuss these recent findings and how they might lead to possible therapies to “tune-down” aberrant B cell responses in the case of autoimmunity.

Complement receptor CR2 in humoral immunity

Starting from an initial observation ~40 years ago that depleting complement component C3 from the serum of an experimental animal results in an impaired antibody response to T-dependent antigen (Pepys, 1972), there has been a substantial body of research directed at understanding the molecular basis for this complement link to the regulation of the adaptive humoral immune response. The various milestones in this quest have been documented in several past reviews (Carroll, 2000; Carroll, 2004; Nielsen and Leslie, 2002; Rickert, 2005; Roozendaal and Carroll, 2007). It is now clear that at the heart of complement’s effect on adaptive humoral immunity is the interaction between complement receptor 2 (CR2, CD21), which is primarily present on B cells and follicular dendritic cells (FDC) in humans and mice, with an antigen-bound fragment of C3 in which the C3d moiety is fully accessible (Fig. 1).

Before dealing with the models that have emerged for the role of the CR2:C3d interaction in adaptive immunity, it is important to appreciate how the antigen gets “tagged” with the appropriate C3 split product via the classical or lectin complement pathways. These processes are dependent upon antigen recognition by an activator of the classical or lectin pathways, such as natural repertoire IgM antibody (Boes et al., 1998) (Fischer et al., 1996), C-type lectins such as SIGN-R1 (Kang et al.,2006), or mannan binding lectin (Degn et al., 2007; Takahashi et al., 2008). When associated with antigen, IgM and SIGN-R1 engage the C1q subcomponent of the proenzyme form of C1 and this results in C1 activation. The now activated form of the C1s subcomponent of C1 cleaves the next component of the classical pathway, C4, into C4a and C4b, with a portion of the C4b becoming covalently bound to the antigen (see below). There it binds C2, which in turn is also cleaved by activated C1s into the active serine protease fragment C2a. The resulting antigen-attached enzyme, C4b2a, is the classical pathway C3 convertase and it cleaves many molecules of C3 into C3a and C3b (Fig. 1). Like C4b, a portion of the nascent C3b molecules become covalently bound to the antigen. In the case of antigen recognition by mannan binding lectin (MBL), the MBL-associated serine proteases (MASPs) become activated, and like C1s, activated MASP2 can cleave C4 and C2, thus yielding an antigen-bound C3 convertase identical to the one formed by via C1 activation.

As alluded to above, the transfer of the homologous proteins C3 and C4 from the fluid phase to a target surface involves a covalent binding reaction. In this transacylation reaction a surface nucleophile, (hydroxyl group in the case of C3, hydroxyl and amino groups in the case of human C4) attacks an activated intramolecular thioester bond formed between the side chains of cysteine and glutamine residues in the sequence CGEQ located within the C3d and C4d subfragments of these proteins (Fig. 1). These fragments correspond closely to the respective structurally-defined thioester domains (TED). Subsequent degradation of the antigen-bound C3b to iC3b and C3dg by the complement regulatory enzyme factor I, and further protease trimming of C3dg to C3d by a non-complement protease, will leave these C3 secondary and tertiary degradation fragments still covalently attached to the antigen (Fig. 1). It is known from structural work that relative to native C3, the TED domain is in a completely different conformational environment in the C3b fragment, and then again in iC3b (Alcorlo et al., 2011; Janssen et al., 2006; Nishida et al., 2006). The equivalence of the affinities for CR2 of the C3 split products iC3b, C3dg, and C3d (Clemenza and Isenman, 2000; Kalli et al., 1991; Nagar et al., 1998) not only indicates that C3d contains the CR2 binding site, but also that the site is equally accessible to CR2 in the precursor degradation fragments iC3b and C3dg.

There are three distinct modalities that have emerged through which the interaction of CR2 with antigen opsonized with C3d affects antibody-mediated adaptive immunity. The first is depicted in Fig 2a, and occurs at the surface of a B cell. It involves the co-ligation of the B cell antigen receptor (BCR) to CR2, where both the primary receptor, i.e. the BCR, and the co-receptor, i.e. CR2, are each individually associated with cell signaling platforms. Co-ligation results in augmented signaling as a result of the redistribution of the entire receptor-co-receptor complex to lipid rafts (Cherukuri et al., 2001). In the presence of large doses of antigen, there is sufficient cross-linking of the BCR complex so that the signaling provided via the associated Igα;Igβ platform is sufficient to induce activation and clonal expansion of the antigen-specific B cell, without the need for complement mediated co-ligation of the CR2; CD19; CD81 co-receptor complex. However, under conditions of limiting antigen, as would likely be the case upon initial encounter with a microbial pathogen, the augmented signaling occurring as a result of CR2-co-receptor ligation effectively lowers the threshold dose of antigen required to trigger B cell clonal expansion.

The second role for the CR2:C3d interaction, as depicted in Fig 2b, involves FDCs within germinal centers where, as a result of trapping C3d-opsonized antigen via CR2, they can present the antigen to naïve or previously antigen-engaged B cells during the processes of affinity maturation, isotype switching, and the generation of effector and memory B cells (Fang et al., 1998). These CR2-mediated effects at the surface of B cells and FDCs are collectively referred to as the “molecular adjuvant” effects of complement and have been shown in various experimental systems in which antigen is fused to multiple copies of C3d to lower the threshold dose of antigen required for an antibody response in the absence of other “traditional” adjuvants from two to four orders of magnitude (Barrault et al., 2005; Dempsey et al., 1996; Wang et al., 2004a; Wang et al., 2004b). However, these same CR2:C3d-mediated molecular adjuvant effects, which are beneficial in the host’s immune response to foreign antigen, can also amplify an antibody-mediated autoimmune response in the case of individuals having B cells reactive against self-antigens. This possibility is discussed in more detail below.

The third modality through which complement interfaces with adaptive humoral immunity involves the delivery of antigen to FDC and this topic will be dealt with separately in a subsequent section.

Complement C3 in T cell immunity

While the focus of this review is primarily on B cell immunity, complement also participates in activation of T cells and several excellent reviews have reported on the topic ((Kemper and Atkinson, 2007b) (Dunkelberger and Song, 2010; Sacks and Zhou, 2012)). Using a murine model of influenza infection, Kopf and colleagues first identified an essential role for complement C3 in the T cell response to the virus. Mice deficient in C3 fail to clear the virus efficiently relative to WT controls and their adaptive response was characterized by an impairment in both the B cell and T cell compartment. Examination of the CD4⁺ and CD8⁺ T cell response at early time points show a marked decrease in T cell proliferation and release of cytokines such as interferon gamma. Later studies confirmed the sensitivity of C3 deficient mice to influenza infection showing a high incidence of mortality that correlates with impaired B cell memory (Fernandez Gonzalez et al., 2008). Interestingly, mice deficient in CD21 and CD35 (Cr2^-/-) appear to respond to infectious virus similar to WT controls suggesting the T cell defect is not mediated through this pathway (Kopf et al., 2002). The results are in agreement with earlier studies by Birgitta Heyman et al. where they found that blockade of CD21 receptors leads to an impaired antibody response without limiting the T helper (Th) cells (Gustavsson et al., 1995). A possible explanation for the impaired CD4⁺ Th1 response to influenza response is suggested by a recent study by (Pekkarinen et al., 2011). They found that the CD4⁺ Th1 cell response of C3 deficient mice is reduced relative to WT controls following immunization with ovalbumin (OVA) in Freund’s adjuvant and this correlates with a decrease in expression of interleukin-12 (IL-12) and the Th1 cell lineage transcription factor T-bet, together with a deviation to Th2 cell response. They proposed that the defect could be due to the either the cytokine environment or affects of antigen presentation by dendritic cells (DC). For example, DC can produce complement components locally which may modulate their cytokine secretion (Sacks and Zhou, 2012). One relevant pathway is the C5a receptor (C5aR) which is known to affect toll-like receptor (TLR) signaling, especially IL-12, and is linked to Th17 cell and T regulatory (Treg)cell differentiation (Kemper and Atkinson, 2007b; Weaver et al., 2010). In human T cells, the complement receptor CD46 may be a major mediator of the effects of C3(Kemper and Atkinson, 2007a). Earlier studies have shown that cross-linking of CD3 and CD46 on T cells induces a T regulatory phenotype (Kemper et al., 2003). In their more recent report, the authors show that activated CD4⁺ T cells produce ligands for CD46, i.e. C3b and C4b, that result in IL-10 production and induction of a regulatory phenotype(Cardone et al., 2010). Thus, the CD46 receptor which was first identified as a regulator of C3b and C4b to protect host cells from activated complement has potent effects on the cytokine production of T cells and a strong influence on the overall T helper cell phenotype.

The CR2:C3d binding interface

As can be seen in the cartoon depiction in Fig. 2, depending on an alternative splice site used, the extracellular region of CR2 consists of 15 or 16 complement control protein (CCP) domains, but the C3d binding site is entirely contained within the two N-terminal-most CCP domains (Kalli et al., 1991). Moreover, both of these domains must be present to observe C3d binding (Carel et al., 1990; Lowell et al., 1989). When the structure of C3d became available (Nagar et al., 1998) it was noted that one of its prominent features was an acidic residue-lined depression that was remote from a more convex face of the molecule harboring the covalent attachment site. Based on its accessible location, the known ionic strength sensitivity of the CR2:C3d interaction, and the potential for charge complementarity in binding to prominent positively-charged patches on one face of the CR2(CCP1-2) molecule, as inferred at that time from a homology model of these domains, the acidic residue-lined depression of C3d was proposed to be a major site of interaction with CR2(CCP1-2) (Nagar et al., 1998). Subsequently, an alanine scan mutational analysis of fully exposed residues within the candidate site was published (Clemenza and Isenman, 2000). This study not only provided support for the hypothesis, but further identified two distinct subclusters of residues located at opposite ends of the acidic pocket whose mutation lead to profound defects in CR2 binding. Shortly thereafter, an alternative model for CR2 binding was suggested by a co-crystal structure of CR2(CCP1-2):C3d which, as depicted by the red CR2(CCP1-2) molecule in figure 3a, shows no interaction within the mutationally-defined contact areas on the concave surface, but rather shows CR2 binding via CCP2 only to a side face of C3d (Szakonyi et al., 2001). From the outset, this structure was discordant with biochemical data on several fronts, including the strong indication that both CCP1 and CCP2 would mediate the contact (Carel et al., 1990; Lowell et al., 1989). This structure has not been supported by subsequent studies, for example one showing that the mutation of several basic residues in CCP1, which made no contact with C3d in the 2001 co-crystal structure, had a major effect on binding. (Hannan et al., 2005). Experiments showing that the concentration of zinc ion used in the crystallization of the complex actually abolishes CR2:C3d binding in solution (Isenman et al., 2010) have made it likely that the 2001 structure represented a non-physiologic interaction promoted by the zinc in the crystallization buffer.

The recent characterization of secreted Staphylococcus aureus virulence proteins, extracellular fibrinogen-binding protein (Efb) and Staphylococcal binder of immunoglobulin (Sbi), which have constituent domains, Efb-C and Sbi-IV, respectively, that, like CR2, bind weakly to C3b, but strongly to iC3b and C3dg (Burman et al., 2008; Hammel et al., 2007) refocused attention on the acidic residue-lined depression of C3d as being a major contact site for CR2(CCP1-2). Both Sbi-IV (Burman et al., 2008) and Efb-C (Ricklin et al., 2008) were shown to compete with CR2 for binding to C3dg. Moreover, a co-crystal structure of the Efb-C:C3d complex showed the 3-helix bundle Efb-C molecule binding on the concave face of C3d, and with several of its helix 2 residues having side chains making specific contact with residues in the acidic residue-lined depression of C3d. These results prompted not only the use of the collection of concave surface C3d mutants to map the interaction site of Sbi-IV, which by this point had also been determined to have a 3-helix bundle fold (Upadhyay et al., 2008), but also to reexamine using biophysical methodology, as compared to the rosette assays of binding used in the 2000 study (Clemenza and Isenman, 2000), the CR2(CCP1-2) binding capacity of an expanded collection of C3dg mutants. The results of these studies reconfirmed the original findings regarding the importance of residues in the C3d acidic pocket for CR2 binding, as well as showing that like Efb-C, this region of was also utilized by Sbi-IV for its binding (Isenman et al., 2010).

Co-crystal structure of the CR2(CCP1-2):C3d complex

Using crystallization conditions that avoid the use of nonphysiologic salts, new diffracting co-crystals of the CR2(CCP1-2):C3d complex have been obtained and yield the model depicted in Fig 3a and in which it is the magenta colored CR2(CCP1-2) that is the C3d-binding entity (van den Elsen and Isenman, 2011). The V-shaped arrangement of the constituent domains of CR2(CCP1-2) both interact with the acidic pocket on the concave surface of C3d and there is excellent shape complementarity of the surfaces. In the electrostatic surface potential rendering of the C3d molecule shown in Fig. 3b, it can be seen how there are several basic side chains of lysine and arginine residues of CR2(CCP1-2) that protrude down into the negatively-charged acidic pocket in a manner that produces the charge complementarity expected based on the long known ionic strength sensitivity of this binding interaction. Many of the interactions observed at the interface, be they charged, polar, or apolar, conformed to mutagenesis data obtained previously (Clemenza and Isenman, 2000; Hannan et al., 2005; Isenman et al., 2010), or to mutagenesis experiments prompted by the new structure (van den Elsen and Isenman, 2011). Finally, the substantially overlapping contact areas on C3d observed among CR2(CCP1-2), Efb-C, and Sbi-IV in their respective co-crystal structures (Clark et al., 2011; Hammel et al., 2007; van den Elsen and Isenman, 2011) readily explains the competition of the staphylococcal immune evasion molecules for CR2 binding to C3d. Because of the nanomolar range affinity of Efb for C3d, compared with the μM range affinity of Sbi for C3d (Hammel et al., 2007; Upadhyay et al., 2008), the interaction of Efb with any C3d molecules that become covalently attached to the Staphylococcus aureus microbemay indeed be a physiologically relevant element through which this human pathogen evades antibody-mediated adaptive immunity (Ricklin et al., 2008).

Delivery of antigen to follicular dendritic cells

The finding that naive B cells are a major transporter of complement-coated immune complexes in the B cell follicles and that both uptake and transfer of the complexes to FDC are dependent for the most part on CR2 and C3d, provide a third important pathway by which the complement system regulates humoral immunity (Fig 4). Using multi-photon intravital imaging in anesthetized mice, Phan et al (Phan et al., 2009) have found that soluble immune complexes that form in the afferent lymphatics and activate complement are captured by sinus lining macrophages, possibly via FcγR and complement CR3 (CD11b; CD18), and are shuttled to naïve B cells in the underlying follicles. Complement-coated complexes are efficiently transferred to follicular B cells in a CR2-dependent manner. While the model system used by Phan et al relied on rabbit IgG – phycoerythrin immune complexes to activate complement, in a more physiologic setting of vaccination where preexisting antibody is likely absent, it seems probable that complement is activated within the lymphatics by other innate recognition proteins such as those noted above, i.e. natural IgM, lectin proteins or pentraxins like C-reactive protein. The finding that the cell surface C-type lectin SIGN-R1, expressed by sinus lining macrophages activates C1q and the classical pathway of complement on binding of Streptococcus pneumonia (Kang et al., 2006) provides another mechanism for formation of C3d-tagged complexes. Whether this pathway leads to B cell uptake of C3d-tagged bacteria has not been determined in their model. Since LN resident dendritic cells also express SIGN-R1; and they have been reported to bind inactivated influenza via SIGN-R1 and transport virus into the follicles(Gonzalez et al., 2010), it is possible that they also transport S. pneumonia into the B cell compartment. It will be interesting to learn in the future whether the resident DC transfer C3-coated particles, such as influenza or bacteria, to naïve follicular B cells for the actual delivery to FDC or whether they deliver the complexes directly.

The actual transfer of C3d-coated immune complexes from naïve B cells to FDC has not been observed in vivo, but CR2-deficient mice show impaired uptake of antigens (Roozendaal and Carroll, 2007). Although FDC express FcγRII receptors, they are not thought to be constitutively expressed and only appear on activated FDC(El Shikh et al., 2006). By contrast CR1 and CR2 are constitutively expressed on FDC (Roozendaal and Carroll, 2007). Thus, the initial transfer of C3-coated complexes from naïve B cells may induce activation of FDC leading to expression of FcγRIIb that may subsequently modulate FDC activity (FDC activation is discussed more below).

FDC are a source of B cell antigen

The clonal selection hypothesis proposes antigen is required for positive selection and affinity maturation of B cells undergoing somatic hypermutation within germinal centers (Rajewsky, 1996). Studies by Goodnow and colleagues have established the importance of antigen in maintaining germinal centers as blockade of antigen receptors led to ablation of the germinal center response(Goodnow et al., 1995). Notably, blockade of the B cell co-receptor (CD21;CD19;CD81) with a soluble form of the CR2 receptor also leads to a rapid loss of germinal centers (Fischer et al., 1998) suggesting that the overall strength of signal of the B cell receptor is critical for B cell survival. Early histology studies suggest the major source of antigen for germinal center responses are FDC. Studies of LN and spleen of immunized mice traced labeled antigens to dendritic-like cells that were later identified as FDC (Hanna and Szakal, 1968; Nossal et al., 1968). Surprisingly, it has been observed that antigen is retained for long periods and in some cases up to one year (Mandel et al., 1980). Thus, unlike macrophages and dendritic cells that degrade antigen after relatively short periods, FDC appear to retain antigens for extended periods.

How these enigmatic stromal cells retain antigen for such extensive periods but make it readily available to B cells is unknown. Ultrastructural studies identify antigen retained as membrane coated bodies, also known as ICCOSOMES, that are released and taken up by B cells within germinal center reactions (Szakal et al., 1988; Szakal et al., 1989). Alternatively, it is reported that FDC retain electron dense antigens on their surface as dense deposits that can take a filiform shape.

Despite the strong correlation between FDC retention of antigen and efficient germinal center responses, actual observation of B cell acquisition of antigen from FDC in vivo has been elusive. In a recent elegant study using intravital microscopy, Suzuki et al. has observed direct uptake of antigen from FDC by cognate B cells (Suzuki et al., 2009). In their murine model, they adoptively transferred immunoglobulin transgenic B cells specific for hen egg lysozyme (Hel) into mice immunized with Hel conjugated to phycoerythrin up to nine days earlier. Interestingly, the cognate B cells acquire not only specific antigen but also FDC membrane proteins CR2 and BP-3 (CD157, BST-1). It is known that B cells can take-up membrane antigens along with antigen from target cells by a process referred to as trogocytosis. Although the study didn’t report whether complement C3d and antibody were also taken-up, it is possible that the original C3d-coated immune complexes are retained intact via CR2 on the FDC surface. These findings are important as they establish that B cell antigens are retained by FDC and can be directly acquired from the cell surface. These elegant results further suggest that FDC are a likely source of antigen for a primary B cell response.

FDC maintain germinal centers

In addition to their role in retention of antigen, FDC are a major source of the B cell chemoattractant CXCL-13 which is required for migration of B cells into the follicles (Cyster et al., 2000). FDC differentiate from ubiquitous perivascular precursors (preFDC) that have been recently identified in an elegant report by Aguzzi and colleagues (Krautler et al., 2012). Maturation of the precursors requires signaling by members of the tumor necrosis factor and lymphotoxin alpha;beta family and presumably additional factors within lymphoid tissues (Gonzalez et al., 1998; Katakai et al., 2008; Koni et al., 1997). However, preFDC can also differentiate within non-lymphoid tissues and are often present in sites of chronic inflammation as found in autoimmune disease and infections.

In the presence of adjuvants and/or immune complexes, FDC become activated and increase release of B cell chemokines, but also secrete the B cell activating factor BAFF, pro-inflammatory cytokines such as IL-6, IL-10 and upregulate integrins ICAM-I and VCAM-1(Garin et al., 2010; Wu et al., 2009). Whether activation alters or promotes display of B cell antigen on their surface is not known but given the importance of antigen for the duration of the germinal center response (approximately 3 weeks) the latter seems most likely.

Earlier studies have demonstrated that treatment of mice with a soluble form of the lymphotoxin receptor leads to transient loss of mature FDC and disruption of humoral responses (Gommerman et al., 2002). In a definitive study, Wang et al. (Wang et al., 2011) have used a genetic approach to selectively ablate FDC in chimeric mice demonstrating that FDC are required for maintenance of B cell follicles and germinal centers. Their model system took advantage of CD21-specific Cre recombinase transgenic mice crossed with a conditional diphtheria toxin receptor (DTR) strain. By constructing bone marrow chimeras using WT bone marrow to reconstitute irradiated CD21-DTR mice, DTR expression is limited primarily to the CD21⁺ stromal population. Treatment of immunized chimeric mice with diphtheria toxin leads to ablation of FDC within lymphoid tissues as expected. Characterization of the mice within 48 hrs of treatment reveals loss of germinal centers and a disruption of the architecture of the B cell follicles. A limitation of the model is that only transient responses could be studied since the mice die within several days of treatment with the toxin from an apparent involvement of the central nervous system. An earlier study by Wyss-Coray has identified expression of CD21 mRNA in the nervous system in reporter mice; although the timing of expression is not known (Moriyama et al., 2011). In the future, it will be important to determine the kinetics of CD21 expression within the developing nervous system and identify potential participation in peripheral immunity.

FDC and autoimmunity

Not all antigens acquired by FDC lead to immune responses. Using an intricate maternal self-antigen transgenic model in which ovalbumin (OVA) expression is limited to the placentae, McCloskey et al has found that self-antigen released at parturition accumulates on FDC located in draining LNs and its uptake is enhanced by classical pathway complement (McCloskey et al., 2011). To test for T cell response to the self-antigen retained by FDC, carboxyfluorescein diacetate succinimidyl ester (CFSE)- labeled CD8⁺ OVA-specific T cells (OT-I) were adoptively transferred several weeks post-parturition, at a time when it was predicted that antigen would be cleared from the host with the exception of antigen retained by FDC. Analysis of harvested LNs at the optimal period has determined that antigen-specific CD8⁺ T cells were not activated and appeared to be tolerized by resident DC sampling antigen from the FDC. As a control, they demonstrated that when OT-1 cells were adoptively transferred before immunization with OVA immune complexes and adjuvant there was a robust response; by contrast adoptive transfer several weeks later results in tolerance. Their results are important for several reasons. First, they suggest resident DC periodically sample both self- and foreign- antigen displayed by FDC. Secondly, DC sampling of antigen from FDC results in a tolerogenic rather than activation signal for CD8⁺ T cells under the conditions defined in this study.

Cr2:C3d as a target for autoimmunity

Like development of a conventional B cell memory response following vaccination with foreign antigen, autoreactive B cells may undergo a similar pathway of differentiation into isotype-switched memory and effector cells. There is growing evidence that spontaneous germinal center formation may contribute to autoimmunity in mice and humans (Cappione et al., 2005; Vinuesa et al., 2009). Moreover, ectopic germinal centers, i.e. form in non-lymphoid tissues, are a hallmark of chronic autoimmunity such as in arthritis (Grammer et al., 2003). Therefore, it seems plausible that differentiating autoreactive B cells may have a similar requirement for co-receptor engagement of C3d along with B cell receptor binding of self-antigen. Thus, blockade of the CR2 co-receptor could provide a transient therapy to eliminate autoimmune germinal centers without imposing immune suppression to the naïve B cell repertoire.

The details of the interface revealed from the co-crystal structure of CR2:C3d may allow one to design a compound that would, for example, interfere with the trapping of C3d-tagged self antigens by CR2 on FDC, thereby potentially down-modulating or “tune-down” antibody-mediated autoimmunity. It is known from mutagenesis data that removing even a part of the contact interface can lead to a major binding defect and so with therapeutic target surfaces available on both sides of the interface, sometimes involving contiguous stretches of amino acid sequence, a peptide-based mimetic approach to the design of such an antagonist should be within the realm of feasibility. Therefore, it will be important in future experiments to determine if autoreactive B cells have a similar requirement as conventional B cells for CR2 signal in survival within germinal centers.