Based on their affinity for monomeric IgG, FcγRs can be divided into the high-affinity FcγRI and the low-affinity FcγRII and FcγRIII (Figure 1). Signaling by FcγR is mediated by immunoreceptor tyrosine-based activating (ITAM) or inhibitory (ITIM) motifs that are present either in the cytoplasmic tail of the receptor itself or in non-covalently associated signaling adaptor proteins, such as the common γ-chain (FcRγ; see below). Aggregation of activating FcγR, i.e., those containing or associated with ITAMs, by binding of multivalent ligands, such as an opsonized pathogen or cancer cell, results in the phosphorylation of ITAM tyrosine residues by Src family protein tyrosine kinases (PTKs), and ultimately leads to activation of cellular responses (10). Aggregation of inhibitory FcγR, i.e., those containing ITIMs, also results in phosphorylation of tyrosine residues by Src family PTKs. In contrast to ITAMs, phosphorylated ITIMs serve as binding sites for phosphotyrosine phosphatases (PTPs), which dephosphorylate other proteins resulting in inhibition of activating pathways (11).

FcγRI (CD64) is a 72 kDa protein that has three extracellular (EC) Ig-like domains, involved in binding of IgG, a transmembrane (TM) domain and a short intracellular (IC) domain of 61 amino acids. The TM domain associates with the FcRγ-chain, an adaptor protein containing an ITAM, to induce signaling and maintain stable expression (12). FcγRI is constitutively expressed by monocytes, macrophages, and dendritic cells and its expression can be induced on neutrophils by stimulation with IFN-γ and/or G-CSF (13, 14). Although there are three genes with various transcripts (15, 16), it is generally believed that only one, the FCGR1A transcript, results in the expression of the classical FcγRIa (CD64).

FcγRII (CD32) is actually a collection of three highly homologous proteins, known as FcγRIIa, -b, and -c that all have a molecular mass of ~40 kDa. Their genes are located in one gene cluster at chromosome 1q23.3 (Figure 2). In contrast to FcγRI, the FcγRII proteins have only two IgG binding EC domains. On the other hand, the much larger intracellular domains of FcγRIIa, -b, and -c harbor intrinsic signaling motifs. In contrast to the activating FcγRIIa and -c, FcγRIIb contains an immunoreceptor tyrosine-based inhibitory motif (ITIM) (17). As no other FcγR contains or associates with proteins containing ITIMs, FcγRIIb is the only inhibitory FcγR (18).

FcγRIIa is the most widely expressed isoform of FcγRII and is found on monocytes, macrophages, dendritic cells, neutrophils and platelets.

FcγRIIb is highly expressed on B-cells, where it constitutes the only surface-expressed FcγR. FcγRIIb is also expressed, albeit at much lower levels, on a subset of monocytes, on macrophages, and on dendritic cells. Expression of FcγRIIb can also be detected on neutrophils and NK cells, but only in individuals with certain genotypes [Ref. (4), Tsang-a-Sjoe et al., submitted].

FcγRIIc has long been considered not to be expressed at all, as its gene (FCGR2C) was thought to be a pseudogene (19, 20), and therefore, relatively little was known about the expression pattern of this receptor. In 1998, FcγRIIc was first found on NK cells of individuals with a particular haplotype of the receptor (21), but we now know that – apart from NK cells – it can also be expressed on neutrophils and monocytes in the individuals with the appropriate genotype (4, 5) (Figure 3A). As this activating FcγRIIc is expressed on circulating monocytes of some individuals, it may be expected that expression also occurs on (monocyte-derived) macrophages of these same individuals, but this has not been reported to date. Here, we show for the first time that monocyte-derived macrophages do indeed express FcγRIIc, at least when cultured in the presence of M-CSF (M2 phenotype), as shown by flowcytometry stainings with MoAb 2B6, which recognizes both FcγRIIb and FcγRIIc (Figure 3C). Although these data are difficult to interpret because of the (varying) presence of FcγRIIb on these cells, we can assume the mean difference in MFI between FCGR2C-ORF and FCGR2C-Stop donors to derive from FcγRIIc. Expression of FcγRIIc specifically was confirmed by qPCR of FCGR2C mRNA and a specific immunoprecipitation using a combination of MoAb 2B6 and a polyclonal antibody that binds FcγRIIc but not FcγRIIb (Figure 3C). Monocyte-derived macrophages differentiated with GM-CSF (M1 phenotype) do not seem to express FcγRIIc based on flowcytometry, although also in these cells, low levels of FCGR2C mRNA could be detected, therefore expression of FcγRIIc cannot be ruled out and may occur in later stages of differentiation of M1 macrophages. Recently, it was proposed that FcγRIIc can also be expressed on B-cells (22). However, evidence of expression of this receptor on B-cells on our own cohort of healthy donors was not confirmed on B-cells, irrespective their naïve or memory phenotype (Figure 3B; data not shown).

FcγRIII (CD16), similarly to FcγRII, actually represents a collection of two genes, expressed within the gene cluster at chromosome 1q23.3, each encoding proteins with two EC Ig-like domains (Figures 1 and 2). Due to differences in glycosylation, their molecular masses are in the range of 50–80 kDa. FcγRIIIa is similar to FcγRI in its TM and IC domains. In monocytes and macrophages, this receptor associates with the FcRγ-chain, while in NK cells it associates with the CD3 ζ-chain to induce signaling (26–28). In contrast to FcγRI, association with these adaptor proteins is not only essential for maintaining stable expression but also for targeting the receptor to the cell membrane (28).

FcγRIIIb is a GPI-anchored protein, expressed only on neutrophils and eosinophils. As it does not have a TM domain, it cannot associate with FcRγ or the ζ-chain. Nonetheless, FcγRIIIb has been suggested to induce signaling, although the exact mechanism(s) is still unclear (29, 30), and current believe is that it functions mainly as a decoy receptor (31).

Recently, it has been proposed that a totally different class of receptors, the C-type lectins, may also bind the Fc-region of IVIg (32), and such receptors were termed “type II Fc receptors.” In humans, DC-SIGN was proposed to bind IgG with a sialic acid sugar moiety at the Asn297 N-linked glycosylation site of IgG. However, this interaction could not be reproduced by another group (33), which found no binding at all of IgG-Fc regions to DC-SIGN, regardless of the sialylation status. Therefore, with evidence for the interaction of IgG-Fc and DC-SIGN being currently marginal and not broadly supported in the literature, in our opinion, it remains to be seen whether DC-SIGN should indeed be classified as a true IgG-Fc receptor.

The genes encoding the classical FcγRs are highly polymorphic and functionally relevant genetic variations have been described for all low-affinity FcγRs (Table 4).

In FCGR2A, encoding for FcγRIIa, a single nucleotide polymorphism (SNP) was first noticed, which results in either a histidine or an arginine at position 131 (H131R) in the IgG binding domain (EC2) (34). FcγRIIa-H131 has a higher binding affinity for IgG1 and especially IgG2, as compared to FcγRIIa-R131, but binding to IgG3 and IgG4 is similar for both variants (38). Functionally, mononuclear cells from FcγRIIa-131HH individuals produce more IL-1beta when stimulated with IgG2 than FcγRIIa-131HR and -131RR individuals (39). Similarly, neutrophils from individuals homozygous for H131 (FcγRIIa-131HH) have been shown to have increased phagocytosis and degranulation in response to serum-opsonized bacteria and increased rosette formation and phagocytosis in presence of IgG3 anti-D sensitized erythrocytes when compared to FcγRIIa-131RR individuals (36, 40).

FCGR2B also exists in two allelic variants, encoding for FcγRIIb containing either an isoleucine or a threonine at position 232 in the TM domain (35). As this SNP (I232T) does not affect the IgG-binding EC domains, it has no influence on the binding affinity. However, its localization at the TM domain results in differences in downstream signaling and subsequent inhibition of FcγRI signaling in macrophages and BCR signaling in B-cells. In particular, I232 provides stronger inhibitory signaling than T232, and this is caused by the exclusion from lipid rafts of FcγRIIb-T232 (41, 42). As FcγRIIb is the only inhibitory FcγR, it has a central role in the regulation of immune responses. The loss-of-function FcγRIIb-T232 has been linked to susceptibility and/or severity of several auto-immune diseases, particularly SLE (43–45), and also in rheumatoid arthritis (RA) (46) and ITP (47).

Inter-individual variation in FcγRIIb is also found in expression patterns and levels. Similar to the I232T SNP, the important immune-regulatory role for FcγRIIb is also reflected in the observations of aberrant expression levels of FcγRIIb in SLE, RA, ITP, and chronic inflammatory demyelinating polyneuropathy (7, 48–51). As a result of a deletion in the FCGR locus that includes FCGR2C, FCGR3B and is called CNR1, FcγRIIb can surprisingly also be expressed on the surface of NK cells, where it is capable to inhibit killing of target cells in ADCC (4). Expression of FcγRIIb in other cells is hardly affected by this deletion. Furthermore, two SNPs in the proximal promoter of FCGR2B and FCGR2C, a guanine or cytosine at position −386 and a thymine or adenine at position −120, form four haplotypes of which one (−386G, −120A; 2B.3) has never been found in any individual thus far. In case of FCGR2B, the wildtype promoter (−386G, −120T; 2B.1) has a lower transcriptional activity than one of the other haplotypes (−386C, −120A; 2B.4) [Ref. (51); Tsang-a-Sjoe et al., submitted].

In case of FCGR2C, only the wildtype and one other promoter haplotype (−386C, −120T; 2B.2) are found. Moreover, the 2B.2 haplotype is linked to another polymorphism in FCGR2C (5). This other polymorphism, a SNP in exon 3, determines whether or not individuals can express FcγRIIc at all. This C > T mutation results in either an open-reading frame (FCGR2C-ORF, allele frequency ~10–15% in Caucasians) or a stop codon (FCGR2C-Stop) (5). Although expression on NK cells is low, it has been shown to be capable of inducing killing of target cells in a redirected ADCC assay (5). Classically, ORF/Stop genotyping of individuals is done based on this SNP alone. However, we have recently found that some individuals carry splice site mutations in intron 7 that introduce novel stop codons, leading to a loss of FcγRIIc expression (4), and genotyping FCGR2C should include these novel mutations to provide an accurate prediction for FcγRIIc expression.

The FcγRIIIa-encoding FCGR3A gene contains a SNP that results in either a valine or a phenylalanine at position 158 (V158F), located in the second EC domain (52). FcγRIIIa-V158 has a higher binding affinity for all human IgG classes compared to FcγRIIIa-F158 (38). In ADCC assays, NK cells from FcγRIIIa-V158 donors show increased killing of target cells that are opsonized with sub-saturating levels of Rituximab (53).

FcγRIIIb-encoding FCGR3B gene exists in three polymorphic variant proteins, NA1, NA2, and SH, which are also known as HNA-1a, -1b, and -1c, respectively (54, 55). FcγRIIIb-NA1 and -NA2 nucleotide sequences differ at five positions [G > C at nucleotide (nt) 141, C > T at nt 147, A > G at nt 227, G > A at nt 277, and G > A at nt 349], with four predicted amino acid differences (R36S, N65S, D82N, and V106I for NA1 and NA2, respectively). As a consequence, the NA2 variant has two additional N-linked glycosylation sites, compared to NA1. The SH variant is identical to NA2 at the five positions that distinguish NA1 from NA2, but differs from both variants at one additional position (C > A at nt 266), resulting in an A78D amino acid change that predicts a change in the tertiary structure of the protein. Additional complexity is added by the discovery of rare individuals carrying other mutations within this gene or different combinations of these nucleotide polymorphisms (37, 56), indicating that the NA1/NA2/SH typing is incomplete. While the binding affinities for IgG1 and IgG3 appear similar between the three variants (38), neutrophils from FcγRIIIb-NA1NA1 individuals bind and phagocytize IgG-opsonized bacteria and red blood cells more efficiently than those from FcγRIIIb-NA1NA2 and -NA2NA2 individuals (40, 57).

Besides being polymorphic, some of the low-affinity FCGR genes are subject to gene copy number variation (CNV). Although several large-scale studies on CNV have suggested that human FCGR2A and FCGR2B are candidate genes for CNV (58–61), our group has shown previously that this is not the case. In fact, CNV in the FCGR locus is restricted to FCGR2C, FCGR3A, and FCGR3B (24). It occurs in three different combinations: FCGR3A/FCGR2C (two possibilities with slightly different borders to the CNV region), and FCGR2C/FCGR3B (Figure 2).

Copy number variation translates into differences in expression levels of FcγRIIc (in case of FCGR2C-ORF), FcγRIIIa, and FcγRIIIb, with more gene copies leading to a higher receptor expression (and vice versa) (21, 62, 63). In case of FcγRIIIa, the level of expression on NK cells is, at least for 1 versus 2 copies, related to the level of killing of target cells in (redirected) ADCC assays (24). Increased expression of FcγRIIIb leads to higher binding and uptake of immune complexes (ICs) (64).

As is the case with polymorphic variants, CNV in FCGR genes is associated with several auto-immune diseases. Our group has previously shown that FCGR2C-ORF predisposes for ITP. The SNP in exon 3 causing an open-reading frame instead of a stop allele of FCGR2C, results in the expression of FcγRIIc and thus behaves as if it were CNV of FCGR2C-ORF. However, individuals can have an increased CNV at this locus of three FCGR2C-Stop alleles without increased risk since only the ORF allele was shown to predispose to ITP (5).

Although we could not find an association with various disease cohorts (24), an increased copy number of FCGR3A has been observed in anti-glomerular basement membrane antibody disease (anti-GBM disease) (65).

In contrast, a low copy number of FCGR3B has been shown to be a risk factor for SLE, even when linkage disequilibrium between FCGR3B CNV and FcγR SNPs that have previously been shown to be associated with SLE is taken into account (66–70). Similar associations have been reported for Sjögren’s syndrome (67), systemic sclerosis (71), and RA (72, 73), although other reports have shown no association with RA (67, 74).

Administration of IVIg greatly increases the total concentration of IgG in the recipients’ plasma and extracellular fluid, and with such an increase, more FcγRs may be bound by circulating non-complexed IgG, thereby saturating the FcγRs and making them less available for auto-antibodies in oligo- or polymeric complex with their (auto)antigen. The idea that especially the low-affinity FcγRs can be blocked by their monomeric ligand in vivo may at first hand seem surprising, but it has been shown in the past that also low-affinity receptors bind monomeric IgG (77), indicating that some “low-affinity” FcγRs are not so low-affinity, and maybe should be better named “medium-affinity,” especially in the case of FcγRIIa and FcγRIIIa (38). Greatly increasing the concentration of monomeric IgG above the normal plasma levels may shift the equilibrium toward a situation in which too many FcγRs are occupied for proper functioning – which may in part explain the immunomodulatory actions of IVIg under some of the conditions for which IVIg is used. Saturation of activating FcγRs was one of the first theories that was formulated to explain the working mechanism of IVIg (2), and this “classic” mechanism has for a long time been assumed as the most plausible explanation for the effect of IVIg in ITP (23, 78). Circumstantial evidence for this theory derives from observations that IgG preparations with increased affinity for FcγRs appear to have an increased effect (6, 79), and that in all diseases in which an immunomodulatory effect is wanted, high doses of IVIg are needed. Nevertheless, there is no formal proof for this concept, and although it has never been disproven, focus has shifted away from this theory as other explaining theories arose.

Over the past decade, the prevailing theory for the working mechanism of IVIg in most immunomodulatory situations has become that IVIg induces an upregulation of the inhibitory FcγRIIb on effector cells. More specifically, a fraction of IVIg, i.e., the IgGs containing a sialic acid sugar residue at the end of the N-linked glycosylation site at Asn297, would be responsible for this effect by binding to SIGNR1 (mouse), or its human ortholog DC-SIGN, inducing various signaling cascades ultimately leading to the upregulation FcγRIIb. This theory has recently been excellently reviewed in Ref. (80). However, the major problem with this theory is that many findings could not be reproduced by other research groups. For instance, we have recently found that FcγRIIb is not upregulated in human macrophages in response to IVIg, but nevertheless, these macrophages respond very well to IVIg treatment, being inhibited in phagocytosis (79). The role of FcγRIIb in ITP treatment by IVIg was also questioned in mouse studies (81). Similarly, we found that IgG-Fc sialylation was not important for the effect of IVIg on human macrophages (79), and many groups have recently published evidence that IgG-Fc sialylation of IVIg is not required for the immunomodulatory effects (82–85). As mentioned before, even the binding of sialylated IgG-Fc to DC-SIGN could not be reproduced (33). Furthermore, essentially all the evidence supporting this theory derives from murine studies, which may not be translated to the human situation, as mice and humans extensively differ in FcγR expression. Many of the murine studies describing this theory for instance use a model for arthritis, but IVIg has never proven to be a useful therapy in treating arthritic patients (86–88).

On the other hand, glycosylation may still be important, influencing the binding affinity IgG molecules to the various FcγRs. For instance, the binding affinity of FcγRIIIa is undoubtedly influenced by the level of fucosylation of the Fc-domain of IgG, a notion that may help to develop new, afucosylated IgG treatment options (89, 90). An important question will be whether the anti-inflammatory properties are directly influenced by afucosylated IgG or IgG otherwise modified in their glycosylation status.

FcRn is a receptor expressed by human endothelial cells to recycle plasma IgG, extending its half-life in the circulation (91, 92); saturating this “rescue-receptor” with a high dose of IVIg may shorten the half-life of all IgG including harmful auto-antibodies. Interestingly, for a number of diseases in which IVIg therapy is beneficial, plasmapheresis, aiming to remove pathogenic auto-antibodies by replacing the patients’ plasma with donor plasma is also a good option. This is for instance the case in Guillain Barre syndrome (93, 94). On the other hand, plasmapheresis is not effective in ITP (95, 96), and so apparently, rapid removal of auto-antibodies is not effective in ITP, suggesting that the effect of IVIg in ITP must be exerted in a different way.

Recent data have confirmed the expression of FcγRII isoforms, including FcγRIIb (97, 98) on dendritic cells (DCs), which may help to explain the subsequent steps in which inhibition of autoantibody release by B-cells, inhibition of T-helper (Th)1 and Th17 differentiation, and enhancement of CD4⁺FoxP3⁺ regulatory T cells (Treg), helps to modulate certain unwanted (auto)inflammatory responses. IVIg may be able to reset the balance at the level of DCs, involving not only the classical IgG receptors but also non-classical lectin-like surface molecules, as has been repeatedly proposed during the last decade (33, 99). We should emphasize that the relevance of such mechanisms and non-classical IgG receptors remain to be shown in humans for the IVIg-associated effects for immunomodulation.

A recent report showed an increased expression of type I interferon response genes in ITP patients, which was rapidly reduced in patients after receiving IVIg, leading to decreased expression of FcγRIII on monocytes, thereby altering the balance between activating and inhibiting FcγRs (100). The relevance of interferons in such responses is unclear, as interferon response genes have been found in various diseases, among which is SLE, and sometimes independent of clear-cut reaction to treatment and clinical response to therapy (101, 102).

One of the first explanations for the anti-inflammatory effect of IVIg was that there are anti-idiotypic antibodies present in the IVIg that neutralize the pathogenic auto-antibodies. This theory is often claimed but to our knowledge has only been proven to effectively explain the anti-inflammatory potential of IVIg in the case of neutralizing antibodies to coagulation factor VIII, which could be inhibited by anti-idiotypic antibodies in IVIg (105).

Apart from the known microbial antigen-specific binding properties, IgG preparations also contain neutralizing and clearance-enhancing antibodies that may switch a proinflammatory trigger into an anti-inflammatory condition. This suggests that healthy individuals from which plasma is collected and pooled for therapeutic IgG preparations already contain autoreactive “natural” antibodies at low levels in their blood. The infusion of such natural antibodies into the patient may be sufficient to reset certain diseases by the cross-reactive capacity of such natural “auto”antibodies (106–110).

As mentioned above, a SNP in exon 3 of FCGR2C determines whether or not individuals can express FcγRIIc at all (5). Although expression on NK cells is low, it has been shown to be capable of inducing killing of target cells in a redirected ADCC assay (5). We could also detect FcγRIIc expression on neutrophils and monocytes in individuals with an ORF allele (Figure 3A) (4). When investigating surface FcγRIIc expression on monocyte-derived macrophages skewed to either M1 or M2 phenotype, the M-CSF-cultured cells were clearly expressing FcγRIIc (Figure 3C).

Surprisingly, a recent report also found FcγRIIc expression on B-cells (22). Upon transfection into a murine B-cell line, the co-ligation of FcγRIIc with the BCR resulted in enhanced and more sustained tyrosine phosphorylation of the key B-cells signaling components Syk and BLNK. In contrast, the engagement of FcγRIIb with the BCR and its activation caused a reduced level of Syk and BLNK phosphorylation. Antibodies generated upon immunization in this transgene mouse model were found to be enhanced, coinciding with a higher level of B-cell activation (22). In a cohort of about 300 individuals of unknown ethnicity the levels of Ab against a neoantigen (Anthrax protein) were tested (22). At the earliest time points in the vaccine study donors homozygous for the FCGR2C-ORF allele [i.e., two alleles (n = 11)] showed higher Ab levels at 4 weeks (p < 0.02) but not any longer at 8 weeks (22). The more common single-ORF donors were not included.

Although interesting, direct proof of FcγRIIc expression in human B-cells is lacking, as the protein was only specifically detected in EBV-transformed B-cell lines (22). Stainings with an antibody detecting both FcγRIIb and FcγRIIc showed a difference between Stop and ORF donors (22), suggesting some expression of FcγRIIc in primary and memory B-cells. However, we performed similar stainings in a much larger cohort and did not detect such differences. Hence, we must conclude that the expression of FcγRIIc on primary B-cells is at most marginal if present at all.

In fact, the myeloid expression on macrophages and DCs of FcγRIIc may be held responsible for the earlier peak in Ab generation in FCGR2C-ORF-positive donors. Not only the macrophage as effector mechanism in immune responses could be relevant for adaptive immunity including Ab generation but also the afferent part of adaptive responses may be involved, as may also be suggested for FcγRIIc-expressing human DCs in FCGR2C-ORF individuals. Thus, we hypothesize that FcγRIIc may be a subtle but relevant genetic factor in the fine balance between health and disease, including the way the immune response will shape the adaptive repertoire as indicated by the immunization studies mentioned above (22) (Figure 4, overview).