Gene duplications in individuals lead to copy number variation (CNV) of some FcγRs in the population. Surprisingly, only CD16a, CD16b, and CD32c of the FcγRs exhibited CNV in a sample population of 600 subjects (32). CNVs have been correlated to autoimmune disorders as well as variations in surface expression levels. CNV of CD16b is correlated to surface expression on neutrophils and implicated in SLE susceptibility (33, 34), as well as other autoimmune disorders (35, 36). Furthermore, CD16a CNV appears to be functionally significant since increased surface expression positively correlated with increasing CD16a gene number (ranging from one to three copies) (32, 35). A CD16a indel has been shown to increase surface expression as well (37).

FcγR amount at the cell surface varies by cell type and receptor identity (Figure 2). On neutrophils, there are an estimated 100,000–300,000 surface exposed CD16b molecules and 10,000–40,000 CD32a molecules (38, 39). The predominant monocyte subtype at roughly 80% of the pool, “classical” monocytes, does not express CD16a. “Non-classical” monocytes express CD16a at a level of roughly 10,000 CD16a molecules per cell but upon differentiation into macrophages express 40,000 CD16a molecules per cell while CD32 remained the same at ~10,000 molecules per cell (40). Another study found macrophages express 5–10 fold higher CD64, CD32a, and CD32b while CD16a expression was comparable to non-classical monocytes. M2c macrophages expressed overall higher levels of FcγRs than M1 macrophages with the following order of expression: CD32a, CD32b > CD64 > CD16a (41). A high number of CD16a molecules are expressed on CD16+ NK cells (100,000-250,000) (42).

Expression levels also vary based on the cell status. Following activation, innate immune cells can induce expression of FcγRs (23, 25, 29, 30, 35). There is also evidence of receptor downregulation upon activation. Downregulation mechanisms include both decreases in expression as well as shedding FcγR from the cell surface following metalloproteinase cleavage. CD32a is shed from Langerhans cells and also expressed as a soluble form (43). CD32b is shed upon activation of B-cells (44). CD16a and CD16b are likewise shed upon activation of NK cells and neutrophils at a known cleavage site by the metalloprotease ADAM17 (45–48). Intriguingly, sCD16b is relatively abundant in serum (~5 nM) (49) and levels vary based on the immune state of the individual (50). Surprisingly, CD64 is the only human FcγR in which a soluble, serum-borne form has not been reported. This may be explained by the presence of a third extracellular CD64 domain in place of the cleavage site found in CD32s and CD16s (Figure 2).

Soluble FcγR forms modulate immune responses. Soluble CD16b binds myeloid cells, NK cells, subsets of T cells, B cells, and monocytes through complement receptor 3 (CR3 or Mac-1 or αM β2, comprised of CD11b/CD18) and complement receptor 4 (CR4 or αx β2, comprised of CD11c/CD18). These interactions cause the release of IL6 and IL8 by monocytes and indicate a potential role for soluble CD16b in inflammation (51). Shedding of CD16a from NK cells allows disengagement of the immune synapse from the target cell and the subsequent ability to kill again. One study demonstrated that repeated engagement by CD16a depleted perforin, however, shedding of CD16a allowed perforin replenishment upon subsequent activation by another activating receptor, Natural killer group 2 member D (NKG2D), which recognizes ligands not normally expressed on healthy tissue (52). Thus, it appears that the act of shedding of CD16 can allow disengagement of the foreign particle which would be crucial for the immune cell's survival and preservation of potential future cytolytic activity. Though shed receptors are proinflammatory and recruit immune cells as discussed above, a complete picture of the mechanisms of regulating surface expression upon immune activation is not currently available.

The correct presentation of FcγRs on the cell membrane is essential for proper immune cell function. ADCC can destroy virally infected cells and cancer cells, and is thus a target for monoclonal antibody (mAb) therapies (53). ADCP is also an important mechanism in mAb therapy targeting malignant cells (14). ADCC and ADCP are dependent on the ability of low to moderate affinity FcγRs to cluster on fluid plasma membranes for activation to occur (54) (Figure 1). Equally important is the regulation of these receptors when no activation signal is present.

Proper activation of FcγRs following Fc engagement by macrophages requires clustering of FcγRs and the displacement of inhibitory receptors. In one study utilizing murine RAW 264.7 cells, segregation of CD45, a phosphatase responsible for dephosphorylating ITAMs, is dependent on antigen distance from the target membrane (55) (Figure 1). It appears that if the antibody is >10 nm from the target surface, there is a substantially impaired ADCP response. This phenomenon is due to the location of the epitope; epitopes closer to the surface exclude the inhibitory CD45 molecule (which stands ~22 nm tall vs. FcγR-IgG complex = 11.5 nm) from the immune synapse. Interestingly, a follow-up study that focused on FDA-approved mAbs found the targets were small surface proteins (< 10 nm in height) suggesting there may be a requirement for mAb epitopes to be located close to the surface for therapeutic efficacy. CD45 was also excluded from the immune synapse in activated human T cells (56). Another study concerning inhibitory module segregation on human macrophages demonstrates CD64, but not CD32a, and inhibitory signal regulating protein α (SIRPα), in conjunction with CD47 (a receptor that inhibits macrophage phagocytosis), are clustered on quiescent cells but upon activation segregate in a process regulated by spleen tyrosine kinase (SYK)-dependent actin cytoskeleton reorganization (57). Recently, FcγR diffusion has been shown to be inhibited by the CD44 transmembrane protein which is immobilized by linearized actin filaments via ezrin/radixen/moesin (ERM) and binds hyaluronan in the glycocalyx (58). This study used primary human macrophages as well as murine cell lines and murine models, utilizing single particle tracking found CD44 and hyaluronan decreased the diffusion rate of FcγRs, while also sterically blocking the binding of FcγRs to immune complexes (Figure 1).

Receptor clustering overwhelms constitutive inhibition as described previously, allowing phosphorylation of the ITAM. ITAMs are phosphorylated via SYK, Src family kinase (SFKs) or ζ-chain-associated protein kinase 70 (ZAP-70) for downstream activation of phosphoinositide-3-kinase (PI3K), NF-κB, extracellular signal regulated kinase (ERK), phosphatidyl inositol 4-phosphate 5-kinase γ (PIP5Kγ), GTPases and other SRC-family kinases (53, 54, 59, 60). Along with FcγR clustering, actin polymerization and depolymerization is equally important for phagocytosis in RAW 264.7 macrophages by creating lammellipodium/pseudopods. These protrusions are controlled by Rac GTPase and lipid composition (54, 59) (Figure 1). Clustering has also been observed on the plasma membrane of murine derived macrophages using total internal reflection microscopy (TIRF) of a lipid bilayer supporting IgG (61). The FcγR microcluster appears on the macrophage pseudopod edge and is subsequently transported to a synapse-like structure thereby recruiting SYK and production of PtdIns(3–5)P3 coordinated with lamellar actin polymerization. Another study on quiescent human macrophages found lateral diffusion of FcγRs is regulated by tonic activity of SYK causing actin cytoskeleton organization to increase the likelihood of FcγRs to be pre-clustered upon finding a pathogen (62).This study further described differential FcγR mobility upon activation. FcγRs at the periphery of the actin-rich pseudopod were more mobile than those already immobilized by binding of IgG-rich regions. The authors explained that this mobility difference is controlled by SYK-mediated regulation of the actin-cytoskeleton which would increase the likelihood of FcγRs to engage more IgG molecules at the leading edge of the lamellipodium/pseudopod and not waste time diffusing into already IgG-dependent, FcγR-immobilized, actin-rich rich regions of plasma membrane. Mobility of FcγRs was described earlier to be decreased at the trailing end of polarized macrophages by CD44 that was bound to linear actin and connected to hyaluronan (58). It was also found in this study that on the leading edge of polarized macrophages, the side that encounters opsonized material, Arp2/3-driven actin branching predominates, initiated by phosphotidlyinosotide (3–5)-trisphosphate production, and increased FcγR mobility allowing for more efficient clustering at the immune synapse. When Arp2/3-driven actin branching predominates, it was found CD44 is more mobile allowing greater FcγR mobility (Figure 1).

In the human NK92 cell line, transduced to express CD16a, a study showed β2 integrins mediate the dynamics of FcγR receptor microclusters in a protein-tyrosine kinase 2 (Pyk2)-dependent manner, controlling the rate of target cell destruction by ADCC (63). β2 integrins bind ICAM-1 on the target cell allowing adhesion and signal transduction through Pyk2 for actin remodeling and the subsequent enhancement of FcγR mobility. Furthermore, sites of granule release are surrounded by clusters of CD16a and release points are devoid of actin. Human NK cell lytic granules also converge at the surface in a dynein and integrin-signal dependent manner which aids spatial targeting of the weaponized molecules to limit off-target damage (64). Surprisingly, CD16a is essential for ADCC of human CD16+ monocytes and upon CD16a engagement, β2 integrins are activated along with TNFα secretion thereby indicating that non-classical monocytes (CD16+) are the sole monocyte class capable of ADCC (65).

During the early stages of phagocytosis by RAW 264.7 cells, direct contact between FcγRs and IgG is increased by greater IgG density on particles, and increased IgG density results in an increased level of early signals. However, late stage signals are “all or nothing,” not concentration dependent, and regulated by PI3K concentration in the phagocytic membranes (66). In this study, low IgG density decreased the amount of opsonized particles but not the rate of phagosome formation and low IgG density particles that did result in phagocytosis recruited the same amounts of late stage signaling molecules (PIP3, Protein kinase C ε type, p85 subunit) and actin. Overall it appears that FcγRs control the initial binding process essential for scanning the foreign particle and initial activation by binding IgG and later stages of commitment to destruction of the particle are controlled by both IgG density and membrane lipid composition.

On murine and human macrophages, receptor clustering upon activation is consistent with a change in the heterogeneity of the membrane lipid composition to a highly ordered phagosomal membrane that is heavily enriched in sphingolipids and ceramide but lacking cholesterol (67) (Figure 1). The authors state that lipid remodeling mediates F-actin remodeling and the biophysical characteristics of the phagosomal membrane are essential for phagocytosis. On human B cells, a polymorphism of the inhibitory receptor CD32b (Ile232Thr) located in the middle of the transmembrane domain, is described to decrease inhibitory function (68). This mutation was shown to result in aberrant localization to a sphingolipid and cholesterol rich region in contrast to the Ile232 wild-type. Aberrant localization is not surprising considering the introduction of a polar residue into the transmembrane domain (69). Furthermore, the ability of CD32b to inhibit B cell receptor (BCR)-mediated PIP3 production, AKT, phospholipase C-γ-2 (PLCγ2) activation and calcium mobilization was impaired in cells expressing the CD32b Thr232 allotype as compared to Ile232. The authors indicate the FcγR locus was associated with SLE and this polymorphism may promote disease. Thus, it appears lipid composition is important for FcγR-mediated mechanisms.

The unique construction of CD16b indicates the potential for a different activation mechanism for neutrophils. Neutrophils predominantly express CD16b with 10-fold less CD32a. CD32a signal transduction is well described and thought to be the canonical FcγR signal transduction via phosphorylation of ITAMs and subsequent SYK recruitment (70). However, CD16b contains a GPI anchor and does not have a polypeptide transmembrane domain nor is it known to associate with a signaling coreceptor, therefore, it is unclear how CD16b promotes signaling in neutrophils (Figure 2). CD16b plays a role in the initial binding of immune complexes in concert with β2 integrins (71). Currently there are conflicting studies suggesting that CD16b can transduce a signal on its own (70, 72, 73), or it transduces a signal by acting with CD32a (74). A recent study found CD16b cross-linking increased IL-10 and TNFα expression, phosphorylated SHP-2 in a lipid-raft mediated manner and inhibited apoptosis in neutrophils. Lipid composition certainly may be an important part of CD16b signal transduction in mechanisms similar to those discussed previously for macrophage phagocytosis and CD32b on B-cells, however the role of lipids in neutrophil activation is not understood (75–81). Interestingly proteinase 3 (PR3), CD16b, cytochrome b558, and NADPH oxidase co-immunoprecipitate on lipid rafts and PR3 and CD16b colocalize in confocal imaging suggesting these may interact in a lipid raft (75). Other findings suggest CD16b signals in conjunction with CR3 via lectin-like interactions (82), leading to neutrophil respiratory bursts (72). The function of GPI-linked CD16b remains undefined despite the high abundance of CD16b in the body and critical roles in mAb therapies (83).

There are clear differences between the signaling and antibody-binding affinity of soluble and membrane-anchored FcγR forms. However, less is known about the effects of the specific FcγR membrane anchors on affinity and cell activation. All FcγRs are localized to the membrane by a transmembrane polypeptide moiety or a glycosylphosphatidylinositol (GPI) moiety (CD16b only) (Figure 2). A micropipette adhesion assay demonstrated CD16a attached to microspheres via a GPI anchor bound roughly 5-fold tighter to IgG1-coated red blood cells (RBCs) than CD16a tethered by a transmembrane domain (84, 85). Interestingly, it also appears IgG1-coated spheres treated with phosphoinositide phospholipase C (PIPLC) to remove the diacylglycerol moiety bound to GPI-linked CD16a with 12-fold less affinity. These authors observed a 60-fold decrease when the GPI-anchor was completely removed. A CD16b-GPI construct showed 2-fold decrease of affinity upon PIPLC treatment and an 11-fold decrease following removal of the GPI-anchor. The authors hypothesized that enhancement of binding affinity associated with the GPI anchor may be due to an allosteric effect on CD16, changing the structure to bind IgG more effectively; such an allosteric mechanism was observed with other GPI-anchored proteins (80). Further studies will be required to fully elucidate how the GPI-anchor affects CD16b and how specific aspects of the membrane anchor confers distinct properties in vivo.

The analysis of N-glycan composition from FcγRs provides a characteristic profile of a protein (112). Glycomics analysis of CD16a on circulating NK cells from three healthy donors revealed a surprising abundance of under-processed forms (~45% hybrid and oligomannose-type N-glycans). CD16a is N-glycosylated at five sites (Figure 2). The remainder of the N-glycans were primarily complex type, biantennary N-glycan structures with a high degree of sialylation (78%) and fucosylation (89%) (109). The under-processed forms do not likely originate from unprocessed CD16a in the ER because all of the observed hybrid forms were sialylated, a modification that occurs in the late Golgi compartments (113) (Figure 3). Moreover, the presence of oligomannose type N-glycans on CD16a from almost all recombinant sources suggests that restricted processing is a conserved feature (108–110). N-glycans at Asn38 and Asn74 were not observed using this glycomics approach to study NK cell CD16a; perhaps these large glycans ionize too poorly to be observed in a derivatized form, but robust ionization of the peptide provides measurable signals for CD16b N38 and N74 glycopeptides (111).

Recombinant expression has thus far failed to generate CD16 with glycan profiles matching those measured for CD16a or CD16b from primary cells. CD16a is the most heavily studied FcγR due to its role in ADCC and the associated therapeutic applications. Glycomics characterization of soluble extracellular domain of CD16a (sCD16a) from HEK293, NS0, and CHO cell lines showed stark differences when compared to CD16a from NK cells, including a high abundance (over 90% compared to 55% in NK cells) of biantennary and triantennary complex type N-glycans with low levels of sialylation (108–110). Moreover, each recombinant system has the potential to synthesize unique N-glycan structures that are not commonly found on native human proteins, such as LacDiNAc (GlcNAc-GalNAc) from HEK293 cells, α-Gal epitopes (αGal-βGal-βGlcNAc), terminal N-acetylglycolylneuraminic acid in NS0 cells and only α-2,3 linked sialic acids in CHO cells (106, 114). These terminal modifications can potentially alter the binding affinity to IgG in an unexpected and undesirable manner.

Differences between native and recombinant CD16a processing render studies of binding affinity using recombinant material suboptimal, however, these materials still represent the best option for many in vitro studies. Furthermore, binding affinity measurements have utilized the soluble extracellular FcγR domains due to challenges associated with extracting full-length material from the membrane. Tethering CD16a to the membrane changes the N-glycosylation, likely due to differential localization within the Golgi (88, 109). Unfortunately, the N-glycosylation profile of full-length CD16a (frCD16a) expressed with HEK293 cells revealed an N-glycan profile unlike that found on NK cells (109). N-glycans from frCD16a showed less under-processed oligomannose and hybrid types (27% in frCD16a and 45% in NK cell CD16a) and the complex-type N-glycans were highly branched. Thus, cell-type specific glycosylation accounts for the dissimilar N-glycan profile on CD16a from primary and recombinant sources and impacts binding affinity measurement, as discussed below.

Recombinant FcγRs are valuable to characterize the role of N-glycosylation on IgG binding affinity, despite clear differences in N-glycan processing when compared to endogenous material. One recent study reported a 40-fold increase in affinity toward afucosylated IgG1-Fc (G0 form) when complex type N-glycans on CD16a were replaced with Man₅ N-glycans (110). This gain revealed that CD16a can bind with an affinity comparable to CD64, the “high affinity FcγR.” A comparable study demonstrated that higher amounts of larger sialylated complex type N-glycans on CD16a expressed in CHO cells correlated with lower affinity for Rituximab (108).

Of five CD16a N-glycans, only two appear essential for high affinity interactions. Mutating the protein to eliminate N-glycan addition with N45Q and N162Q substitutions reduced the affinity for IgG1-Fc (109, 115–117). However, the reported influence of N-glycan composition was primarily driven by the N-glycan at N162: only the N162Q mutation abolished the affinity gain due to Man₅ N-glycans on CD16a (110). These observations are in agreement with the fact that glycans at N45 and N162 form interactions with the CD16a polypeptide and influence protein structure (118) and glycans at these two sites showed the greatest restriction in N-glycan processing using the HEK293 and CHO systems (119). Thus, cell-type specific CD16a N-glycosylation patterns influence affinity for IgG1 and a range of potential affinities are accessible purely through modifying N-glycan processing.

CD16b is a highly similar paralogue of CD16a and only found in humans (97% sequence homology of the extracellular antibody-binding domain). However, two common CD16b alleles encode either four (NA1) or six (NA2) N-glycosylation sites (120, 121) (Figure 2). Considerable site-specific diversity in N-glycan structures was present on sCD16b obtained from 2l of pooled human serum (111). Serum sCD16b is generated by ADAM17 cleavage of cell surface CD16b upon neutrophil activation (48). Thus, sCD16b were likely membrane bound when the N-glycans were being processed. The N-glycans at each site had unique profiles ranging from smaller oligomannose type N-glycans at N45 to large complex type N-glycans with extensive elongation, sialylation, and fucosylation at N38 and N74, unlike sCD16b expressed in recombinant systems (106, 114, 122). Additionally, allele specific (NA1 and NA2) N-glycosylation profile at N162 and N45 of donor matched serum and neutrophil CD16b confirmed the observations of CD16b from pooled serum, revealing moderate variability in the abundance of the most prominent glycoforms (123). The profile of sCD16b from serum was distinct from CD16a expressed by NK cells that displayed a greater level of under-processed N-glycans (109, 111). The presence of oligomannose type N-glycans only at N45 strongly suggests under-processing of N-glycan is restricted to a single site on the protein with as many as six N-glycosylation sites (111).

The stark differences in the glycosylation profile of sCD16b from serum compared to recombinant sCD16b further emphasized the importance of cell type specific N-glycosylation. Glycomics analysis of CD16b from HEK293, NS0 and BHK revealed mainly multiantennary complex type N-glycans with a high degree of sialylation and fucosylation (106, 114, 122). The N-glycosylation profile of recombinant sCD16a and sCD16b are comparable as most of the N-glycosylation sites are shared (124). There was a minimal difference (2-fold increase) in affinity when sCD16b-Man5 binding to IgG1-Fc (G0F form) was compared to sCD16b with complex-type N-glycans (110). This was surprising considering that the extracellular antibody binding domains of CD16a and CD16b (NA2) differ at only four amino acid residues. Moreover, both CD16s are functionally distinct because CD16a-complex type has a 15-fold greater affinity for IgG-Fc than CD16b-complex type (110). The affinity and sensitivity to glycan composition for CD16b was improved to that of CD16a by mutating a single residue, Asp129, to Gly based on the CD16a sequence (124). The authors demonstrate with x-ray crystallography and molecular dynamics simulations that Asp129 buckles the CD16b backbone upon binding IgG1 Fc. Thus, buckling shifts a nearby residue, Arg155, which makes a different contact with the N162-glycan that is not observed in CD16a.

The N-glycosylation profiles of sCD32a and sCD32b expressed with recombinant systems were highly comparable (106, 108, 114). There are two to three N-glycosylation sites on CD32: CD32a (3), CD32b (2), and CD32c (2; 32b and 32c have identical extracellular domains) (Figure 2). Glycomics analysis of CD32a and CD32b expressed in HEK293, NS0, and CHO displayed predominantly biantennary and triantennary complex type N-glycan structures with a low degree of sialylation and varying levels of fucose (106, 108, 110, 114). Binding affinity between sCD32a and sCD32b was comparable and neither appeared sensitive to N-glycan composition as sCD32(a or b)-Man5 and sCD32(a or b)-complex type bound IgG1 Fc with similar affinities (106, 110). CD32a polymorphisms (R131 or H131) cause differences in binding to IgG subtypes, potentially changing the sensitivity of immune complexes to phagocytosis by neutrophils and monocytes (121, 125). However, N-glycan analysis on the receptor expressed in CHO cells showed no substantial difference in glycosylation pattern between the two CD32a allotypes (108). The site-specific N-glycosylation profile and native N-glycosylation profile for any CD32 is not currently available.

The high affinity FcγR, CD64, is distinct from other FcγRs because it contains an additional extracellular domain (126). Moreover, CD64 can potentially receive N-glycosylation modification at seven sites in its extracellular domain (Figure 2). A comparative glycomics analysis of the sCD64 expressed in HEK293, NS0, and CHO cell lines showed biantennary and multi-antennary complex type N-glycans with varying degrees of sialylation and fucosylation as the most abundant glycoforms (106, 108, 114). A distinct feature which was conserved across sCD64 expressed in all three cell lines was the higher abundance of oligomannose structures when compared to recombinant CD16 or CD32. It was speculated that the presence of Man₅ forms (the most abundant oligomannose N-glycan in these cell types) conferred a stabilizing effect toward IgG1 binding since the higher abundance of Man₅ forms (14.4% in NS0 and 5.2% in CHO) correlated with an increase in binding affinity to Rituximab (108). According to the authors, the increased affinity was due to the lack of core fucose on the Man₅ structure which can potentially prevent steric hindrance effects similar to that observed in fucosylated N-glycan on IgG1 (115, 127). The authors also observed that the presence of large sialylated complex type N-glycans on CD64 correlated with reduced binding affinity for Rituximab, indicating that these glycans destabilized the interaction (108). Even though N-glycan composition on CD64 can affect IgG1 affinity, the N-glycosylation profile of native CD64 and the composition of N-glycans at each site remains unknown.

N-glycosylation processing depends on the amino acid sequence and secondary structures which affect the exposure of substrate monosaccharide residues to the glycan processing enzymes (Figure 3). Presence of both the under-processed and highly-processed (tetraantennary sialylated) N-glycan structures on NK cell CD16a and recombinant sCD64 suggests site-specific glycan modification. Oligomannose structures at specific sites on sCD16a have been implicated in modulating IgG affinity; similarly, specific sites on CD64 can be involved in modulating CD64-IgG1 affinity (108, 110). Thus, a thorough analysis of site-specific N-glycosylation analysis of recombinant and endogenous FcγRs from all expressing tissues is required to fully elucidate the role of N-glycosylation pattern at specific sites in affinity modulation.

The importance of interactions between multiple monovalent ligands and multiple receptors is well known, however, the study of multivalent interactions remains challenging. Early attempts to generate multivalent immune complexes through heat aggregation of IgG produced aggregates with varied valency, immunogenicity and ill-defined sizes (131, 132). Technological advances in recent years produced immune complexes of defined size and valency which accurately represent those generated in vivo (130). Functional interrogation using defined immune complex revealed that immune complex size contributes to interactions with FcγRs.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.