IgM, FcμRs, and Malarial Immune Evasion

Immunoglobulin M is the oldest and only isotype of Ab that is expressed by all species of vertebrates. Unlike most other isotypes, there are two classes of IgM: an Ag-specific (or immune) IgM class that is produced in response to exposure to specific pathogens and a low-affinity, more broadly reactive (so-called “natural” or “nonimmune”) IgM class that is synthesized without prior exposure to any particular Ag. This latter form, apparently designed to recognize common motifs of widespread foreign invaders as part of the first line of host defense, is likely responsible for the conservation of this isotype among distantly related vertebrates (1). Moreover, probably befitting their common ancestry, IgM molecules from all species studied to date are higher-order multimeric complexes (predominantly pentamers in human plasma), a physical property that endows the molecule with extraordinary functions, including multivalent binding to Ags and receptors (2–4).

Natural IgM produced by peritoneal B-1 cells is an important mediator of innate immunity, controlling the dissemination of viruses and bacteria (3, 4). A subset of these B-1 cells can also contribute to the production of immune IgM and provide long-term T cell-independent immunity against reinfection (5), although immune IgM can also be produced in a T cell-dependent manner by conventional B-2 cells in the spleen and lymph nodes (6). IgM constitutes ∼10% of human plasma Ig, and levels of IgM are ∼25% greater in females than in males (7, 8), a finding that led to the notion that males are significantly more susceptible to infections than are females (9, 10). Because selective IgM deficiency is rare in humans, an understanding of the importance for IgM in immunity has come from studies using secretory IgM-deficient (IgM^−/−) mice (3, 4, 11). These studies clearly show that serum IgM primes the ensuing IgG response and that IgM is particularly effective at neutralizing and agglutinating blood pathogens, particularly those that reside within cells, by inhibiting pathogen transfer from cell to cell (11). Such findings show that IgM can have a profound impact on protective immunity and, in terms of possible mechanisms, point to the involvement of interactions between the Fc domain in IgM and effector molecules, such as complement and FcRs, in mediating at least some of this immunity.

As with IgM, malarial parasites of the genus Plasmodium have been found in all classes of terrestrial vertebrates, including reptiles and birds. Thus, they likely are a long-standing vertebrate pathogen against which a vigorous immune response involving IgM would be commonly expected. Indeed, parasite-specific IgM was shown to limit parasite replication and prime memory cell generation, and it is a more potent adjuvant than Bordetella pertussis in experiments with murine malarias (12, 13). Natural IgM is also an endogenous adjuvant for vaccine-induced protective CD8⁺ T cell responses against intracellular parasites (14). Given these considerations, it is somewhat surprising that so little work has been published on the role of IgM in human malaria. Natural IgM is known to bind to the surface of Plasmodium falciparum-infected erythrocytes via the Fc, and this was shown to correlate with severe malaria in laboratory strains and field isolates (15, 16). Only some P. falciparum strains bind natural IgM, and this property is seen in parasites with specific virulence-associated adhesion phenotypes, including rosetting (15), and chondroitin sulfate A (CSA)-binding linked to placental malaria infection (17, 18). P. falciparum strains showing other common adhesion phenotypes, including CD36 and ICAM-1 binding, do not seem to bind natural IgM (15). Therefore, although nonimmune IgM binding is only shown by a subset of P. falciparum isolates, it is linked to the most serious clinical effects of malaria. A greater understanding of the role of nonimmune IgM in these host–parasite interactions has the potential to contribute new insights and interventions against life-threatening disease.

To evade Fc-mediated destruction, pathogens have evolved to produce Fc-binding proteins; those expressed by bacteria and viruses for IgG or IgA have been intensively studied (19–21). These proteins help pathogens to avoid host immune responses by preventing pathogen-specific Abs from interacting with host FcRs and, therefore, interfere with effector functions of Ab, such as phagocytosis and complement activation (19–21). The existence of IgM Fcμ-binding proteins from pathogens is less well documented than for IgG and IgA. This might be because of difficulties in differentiating low-affinity F(ab)′₂-mediated pathogen binding by natural IgM Abs from bona fide Fc–receptor interactions. Nonetheless, IgM-binding proteins have been described for several protozoa, including Toxoplasma gondii (22) and pathogenic species of Trypanosomatidae (23). Recently, we provided the first detailed molecular characterization of an IgM Fc-binding protein from the malaria parasite P. falciparum (16).

IgM binding by P. falciparum-infected erythrocytes occurs via the parasite variant Ag, P. falciparum erythrocyte membrane protein one (PfEMP1), found on the surface of infected erythrocytes (15, 16). PfEMP1 variants are encoded by var genes, and each parasite contains 50–60 var genes in its genome (24, 25), with only one variant expressed on the infected erythrocyte surface at a time (26). The var gene repertoires of different P. falciparum isolates have very little overlap, resulting in extensive diversity among different parasite isolates (27). PfEMP1 molecules are composed of Duffy binding-like (DBL) domains classified into six types (α, β, γ, δ, ε, and X) and cysteine-rich interdomain region domains classified into three types (α, β, and γ) (28). Individual var genes differ from each other by the number and type of these domains. A number of different domains from specific PfEMP1 variants expressed by IgM-binding–infected erythrocytes of different parasite strains were shown to bind nonimmune IgM (Table I), including our identification of IgM binding by DBL4β from the PfEMP1 variant var1 in the TM284 isolate (16). It has not been possible to define a specific sequence motif within these various domains that is responsible for the ability to bind nonimmune IgM.

Using recombinant domain-swapped Abs, IgM mutants, and blocking mAbs, we defined the region of the IgM molecule bound by PfEMP1 (16). In particular, these experiments showed that PfEMP1 binds to the Cμ4 domain of polymeric IgM (16), at a region conserved with those on IgG and IgA recognized by Fc-binding proteins from bacteria and viruses (Fig. 1) (19–21). The binding site on IgM was also shared by five distinct parasite strains (two rosetting and three CSA binding), suggesting that diverse parasite isolates expressing distinct PfEMP1 variants all bind to the same (or similar) site on the human IgM molecule. Recent work from our laboratories has shown that the laboratory isolate 3D7, which was panned on CSA to express the PfEMP1 variant var2csa (33), also interacts with IgM via the Cμ4 domain. In this case, the DBL5ε domain in var2csa binds IgM (A. Salanti, S.B. Ditlev, and R.J. Pleass, unpublished data). The binding of IgM to recombinant 3D7 DBL5ε was not inhibited by soluble CSA, implying that the CSA-binding sites on the DBL domains do not overlap with those used by IgM. Recombinant var2csa single domains were shown to bind promiscuously to a number of glycosylated receptors that the native var2csa does not bind (33). However, using full-length recombinant FCR3 var2csa, we verified that IgM interacts with the complete protein, providing validation for the results with the individual domains (A. Salanti, S.B. Ditlev, and R.J. Pleass, unpublished data).

The use of a similar IgM-binding site by distinct PfEMP1 variants suggests the possibility of a similar structural mechanism of binding to IgM. To determine whether there is a common IgM interaction site on these diverse P. falciparum DBL domains, we modeled the previously described IgM-binding DBL domains (Table I) onto known DBL structures and tried to dock these molecules into a recent model of human IgM (34). This model of IgM, based on the known bent structure of IgE and supported by direct cryo-atomic force microscopy images, shows IgM to be a mushroom-shaped molecule, with a central region formed by the Cμ3/Cμ4 domains protruding out of the plane formed by the Cμ2/Fab domains. The residues Pro³⁹⁴–Pro³⁹⁷ and Pro⁴⁴⁴–Val⁴⁴⁷ in the Cμ4 domain, previously implicated in PfEMP1 binding, are localized near the junction between two monomers (16) and, therefore, can form the binding pocket for PfEMP1, while explaining the requirement for a polymeric structure.

Although it might be expected that there are many ways by which such large proteins could possibly interact, a number of observations place strict constraints on potentially viable models, including the narrowness of the binding pocket (with a width of a little more than only two helices), the central disposition of the IgM-binding domains within the full PfEMP1 sequence (which constrains the possible locations of the termini of the IgM-binding domain), the markedly limited solvent accessibility of these domains within the native PfEMP1 (33, 35, 36), and the localization of ligand-binding residues in other DBL domains to the loops in subdomain 2 (35, 36).

Under these strict constraints, we were able to identify only a single plausible DBL–IgM model (Fig. 2A). In this structure, the DBL domain, orientated roughly perpendicular to the Cμ2/Fab plane, binds to the side of the central IgM stem, in between two IgM monomers that project radially (Fig. 2A, Supplemental Video 1). The predominant contacts to the IgM molecule are mediated by loops within subdomain 2, which directly contact the opposing residues of neighboring IgM monomers (Table II), including those between the Pro⁴⁴⁴–Val⁴⁴⁷ loop in IgM (16). Additional contacts away from this region involve a small helix near the DBL C terminus (residues Gln⁴²⁹–Glu⁴³⁷) and the other Pro³⁹⁴–Pro³⁹⁷ loop of IgM (16). Although this model satisfies all of the aforementioned constraints, what was particularly striking about this disposition of the DBL domain was the presence and number of its charged residues that are immediately adjacent to oppositely charged residues on the IgM surface (Fig. 2B). Although these residues are not strictly conserved between different IgM-binding domains, a similar set of charged residues is found in all of the characterized IgM-binding DBL domains (Table II). Moreover, inspection of these regions in the known structures of members of the DBL family, which are from proteins that do not bind IgM (e.g., EBA-175, and PKα-DBL) or have been shown not to bind IgM (e.g., DBL3X), reveals an absence of charged residues similar to this set (data not shown).

Such a binding mechanism involving similarly charged, but not strictly conserved, residues from DBL domains of different subclasses, which, moreover, are not shared by all members of any single subclass, suggests that there was not an ancestral IgM-binding DBL domain from which these domains are derived. Instead, IgM binding may be an emergent property of certain DBL domains that resulted from some other common trait of these domains. Interestingly, the most characterized IgM-binding domains (Table II) are located proximal to the erythrocyte plasma membrane. As a result, the possible orientations of the IgM molecule, when bound to PfEMP1, would presumably be severely limited, owing to the presence of the nearby erythrocyte membrane (compared with a more membrane-distal binding). Hence, it may be important for the function of these IgM Fc-binding proteins to not just interact with IgM, but also to have the IgM molecule, in essence, distinctly oriented with respect to the red cell membrane. This may be necessary to prevent or perturb the access of IgM to host effector proteins, such as complement C1q, or, perhaps more intriguingly, to actually promote the binding of IgM to host receptors (Fig. 3).

It is unknown why P. falciparum has evolved an IgM Fcμ-binding protein. In fact, this interaction may be part of a more general set of interactions that have evolved between P. falciparum and IgM, including a recently suggested inhibitory role of IgM as a novel component of the tetraspanin network on B cells (40), and in hepatocyte invasion by sporozoites (41). However, in this brief review, we focus on the possible functional consequences of the PfEMP1–IgM interaction. As summarized in Fig. 3, these include interference with immunological signaling and clearance mechanisms, blocking the binding of specific Abs, or as a way of enhancing infected erythrocyte sequestration in the microvasculature or placenta.

The recent identification of a bona fide high-affinity FcμR for IgM expressed by CD19⁺ B cells (including memory B cells), CD4⁺/CD8⁺ T cells, and CD56⁺/CD3⁻ NK cells may provide the target for interference by malaria parasite DBL domains (42). Although the function of FcμR in these cells is not understood, ectopic expression of this receptor was found to affect cell survival and proliferation, and a role in B cell activation has been suggested (42). Infection with P. falciparum malaria is known to induce polyclonal B cell activation, leading to a marked hypergammaglobulinemia and elevated titers of autoantibodies (43). Therefore, binding by parasite PfEMP1 proteins to IgM may interfere with the interaction of IgM with FcμR on B cells in a way that favors the induction of proliferative responses by the parasite. The benefit for the parasite of such a rampant B cell activation is not understood.

Intriguingly, increased Ag retention by marginal zone B cells and germinal center formation also depend on CR1 (CD35) and CR2 (CD21), the latter being responsible for EBV entry into B cells (44). EBV-infected B cells occasionally undergo malignant transformation in vivo, giving rise to Burkitt’s lymphoma, a tumor that commonly occurs in areas of Africa hyperendemic for malaria (45, 46). Therefore, the ability of PfEMP1 to interfere with signals induced by FcμR might promote cellular proliferation and predispose the B cell to become malignant upon infection by EBV. Because children affected by Burkitt’s lymphoma often present with chronic malaria, it would be interesting to determine whether parasite isolates derived from these children also bind IgM.

FcμR is also constitutively expressed on CD4⁺ and CD8⁺ T cells, although its function on these cells awaits elucidation (42). The receptor may allow the T cell to interact optimally with the IgM BCR or IgM immune complexes on B cells, facilitating immunological synapse formation between the two cell types and thereby enhancing B cell activation and memory induction. It may also trigger IgM-dependent cell-mediated cytotoxicity by CD8⁺ T cells, already known to play an important role in immunity against liver-stage parasites (47). The ability of PfEMP1 to bind IgM could significantly interfere with these FcμR functions and provide a significant advantage to a parasite under attack from parasite-specific or natural IgM. These possibilities need to be investigated for IgM and malaria.

Fcα/μR, a human receptor for IgM (and IgA) closely related to the polymeric IgR in its ligand-binding domain, has also been described (48–50). Expressed predominantly on follicular dendritic cells (FDCs), Fcα/μR is believed to function in Ag presentation and B cell selection in the germinal center response (42, 51). Recent work in Fcα/μR knockout mice suggests that the receptor negatively regulates T-independent Ag (common in malaria) retention by FDCs, culminating in the suppression of humoral immune responses against T-independent Ags (52). The receptor can also mediate endocytosis of IgM-coated bacteria and immune complexes (48–50). Therefore, IgM binding by PfEMP1 may allow infected erythrocytes to interfere with these functions of Fcα/μR. We recently showed that the binding of IgM to Fcα/μR is dependent on the pentameric structure of IgM and that mAbs that inhibit binding to PfEMP1 also prevent binding of IgM to Fcα/μR (53), suggesting that the binding sites for the two ligands lie close to each other on IgM. By using domain-swapped Abs, we demonstrated that binding involved the Cμ3 and Cμ4 domains in the IgM polymer, and similar results were recently observed with FcμR (42). However, the 80-kDa secretory component that shares homologous binding sites with Fcα/μR does not block the binding of IgM to DBL4β, implying that IgM can accommodate the simultaneous binding of both ligands (16, 53). It is known that the interaction with Fcα/μR requires unique amino acid contacts in IgM, because domain-swap Abs containing only Cμ4 bind PfEMP1 but do not bind Fcα/μR. Furthermore, mouse IgM does not bind PfEMP1, yet it is capable of binding human Fcα/μR and FcμR, supporting our notion that unique contacts are involved in the binding of human IgM to PfEMP1 or Fcα/μR and FcμR. Given the overlapping nature of the binding sites, it could be argued that this region is important for human IgM function and that it is beneficial for the infected erythrocyte to manipulate the receptor’s access to this region.

The ability of IgM to agglutinate and neutralize pathogens so effectively is largely due to its polymeric structure that endows the molecule with increased avidity. This unique structure also makes IgM very effective at activating the classical complement cascade. A single molecule of IgM can trigger complement activation, leading to the lysis of a single erythrocyte, an equivalent task requiring 1000 IgG molecules (44, 54).

Therefore, binding of PfEMP1 at the interface of the IgM Cμ4 domain offers an elegant explanation for our finding that C1q binding did not interfere with the DBL domain interaction (16), because the residues involved in C1q binding are found in the Cμ3 domain on the flat side of the IgM molecule (Fig. 4) (34). However, IgM binding by PfEMP1 might interfere with complement activation by inducing conformational changes in the central core of IgM that lock out the C1q-binding sites on IgM. Alternatively, PfEMP1 binding may force IgM to be oriented with its flat side directly facing the red cell membrane, making the C1q-binding sites inaccessible to C1q.

If these hypotheses are correct, C1q would rarely be found bound to infected erythrocytes. Although no published studies have reported the presence of C1q on infected erythrocytes, parasite-specific IgM and other complement factors, including C3, C4, and C9, have been described on infected erythrocytes (55). However, these can all be deposited via the alternative or lectin pathways in a manner independent of C1q or IgM. Although we know that DBL binding can occur to IgM opsonized onto haptenated erythrocytes, we still do not know whether PfEMP1 binding can occur after engagement of an infected erythrocyte by parasite-specific IgM and what effect this might have on subsequent C1q docking. Human IgM mAbs against other knob-associated Ag(s) or non-IgM–binding DBL domains would clearly be useful to address these questions.

Infection with P. falciparum induces a hypergammaglobulinemia indicative of the presence of B cell mitogens (43). We have become intrigued by the possibility that these various DBL domains may interact directly with IgM⁺ B cells through the BCR, because Cμ4-specific mAbs that inhibit PfEMP1 binding to IgM were also shown to augment S-phase entry of human B cells via the IgM⁺ BCR (16, 56). IgM-mediated signaling is required for the development of a normal B cell memory response (57). Our finding that malaria parasites, known to promote polyclonal B cell activation, encode molecules specific for Cμ4 certainly suggests that binding to this region of the BCR might also be immunologically important and merits further investigation. Although most circulating infected erythrocytes will presumably be saturated in IgM, making them unable to simultaneously engage IgM⁺ B cells through the same DBL domain, other regions of PfEMP1 may contact the IgM⁺ BCR after the primary IgM contact. One candidate is the cysteine-rich interdomain region domain, which is already known to interact directly with the B cell (58).

Evidence demonstrating the ability of the BCR to cluster could support the idea that PfEMP1 may bind to the Cμ4 of the BCR and to soluble pentameric IgM via similar mechanisms. It is well known that BCR mutants missing the Cμ4 domain are unable to inhibit signaling upon Ag binding and that Cμ4 domains expressed alone on the B cell surface cluster spontaneously to activate the cell (56, 59, 60). Therefore, the DBL domain may cross-link two BCR monomers [as in the model with pentameric IgM (Figs. 2, 3)], or binding may induce conformational changes in the BCR, revealing an interface that promotes BCR clustering with subsequent uncontrolled activation. Coincidently, the BCR governs the subcellular location of TLR 9, a ligand for Plasmodium DNA that leads to hyperresponses to DNA-containing Ags (60, 61). Whether FcμR, potentially as a coreceptor at the surface of the B cell, plays a role in BCR function and memory induction and the potential role for interference by malaria parasites through IgM binding now need to be investigated, perhaps in FcμR-deficient mouse models of malaria.

Sequestration of infected erythrocytes in the microvasculature and the placenta is a cardinal feature of P. falciparum infection; it is thought to benefit the parasite by allowing infected erythrocytes to avoid being removed by the spleen (27). It is possible that IgM bound to infected erythrocytes could cross-bridge to host IgM-binding receptors on endothelial or placental cells to promote the sequestration of parasitized cells. It is unknown whether any of the IgM receptors discussed above are expressed on human endothelial cells in the microvasculature or on placental syncytiotrophoblast cells. One intriguing recent finding is that CD300LG (also known as CLM9, TREM4, or NEPMUCIN) shares 35% identity with Fcα/μR and is expressed on capillary endothelium, particularly in placental tissues (62). Although CD300LG is a functional receptor for l-selectin–mediated lymphocyte rolling (63), it was recently shown to bind IgM (62). Future experiments should investigate the expression patterns of IgM-binding proteins in the human microvasculature and placenta and address the possibility that IgM (natural or parasite-specific) bound to infected erythrocytes could cross-bridge to promote sequestration.

Rosetting and placental parasite isolates may also bind nonspecific IgM to allow masking of critical PfEMP1 domains from the destructive action of specific Abs, e.g., IgG. This “umbrella” hypothesis is supported by the observation that the domain adjacent to the IgM-binding domain in var2csa seems to be the only domain inducing highly adhesion-blocking Abs in humans. The hypothesis is readily testable using domain-specific and parasite adhesion-blocking anti-var2csa IgG (64).

The ability of diverse P. falciparum isolates to bind a conserved region of the IgM Fc suggests that this function is essential to parasite survival. Therefore, a molecular understanding of how malaria parasites bind the Fc of natural and specific IgM may lead to the development of novel inhibitors of this interaction. For example, small molecule inhibitors may prevent docking of PfEMP1 to IgM or the BCR and prevent hypergammaglobulinemia or, in the case of immunity, lead to the development of greater memory responses when coinjected with vaccines. Understanding how infected erythrocytes manipulate IgM may also lead to the development of improved adjuvants based on the remarkable properties of this fascinating molecule (12, 14). DBL-based reagents, like those already developed for bacterial protein A and G, may be used for the specific detection and purification of human IgM. As such, they promise to become valuable tools for structural and functional studies of IgM.

We thank Pongsak Khunrae and Matthew Higgins, University of Cambridge, Cambridge, U.K., for providing a model of DBL4β.

Disclosures The authors have no financial conflicts of interest.