Dimeric FcγR Ectodomains as Probes of the Fc Receptor Function of Anti-Influenza Virus IgG

Ab-dependent cellular cytotoxicity, phagocytosis, and Ag presentation are key mechanisms of action of Abs arising in vaccine or naturally acquired immunity, as well of therapeutic mAbs. Cells expressing the low-affinity FcγRs (FcγRII or CD32 and FcγRIII or CD16) are activated for these functions when receptors are aggregated following the binding of IgG-opsonized targets. Despite the diversity of the Fc receptor proteins, IgG ligands, and potential responding cell types, the induction of all FcγR-mediated responses by opsonized targets requires the presentation of multiple Fc regions in close proximity to each other. We demonstrated that such “near-neighbor” Fc regions can be detected using defined recombinant soluble (rs) dimeric low-affinity ectodomains (rsFcγR) that have an absolute binding requirement for the simultaneous engagement of two IgG Fc regions. Like cell surface–expressed FcγRs, the binding of dimeric rsFcγR ectodomains to Ab immune complexes was affected by Ab subclass, presentation, opsonization density, Fc fucosylation, or mutation. The activation of an NK cell line and primary NK cells by human IgG-opsonized influenza A hemagglutinin correlated with dimeric rsFcγRIIIa binding activity but not with Ab titer. Furthermore, the dimeric rsFcγR binding assay sensitively detected greater Fc receptor activity to pandemic H1N1 hemagglutinin after the swine influenza pandemic of 2009 in pooled human polyclonal IgG. Thus these dimeric rsFcγR ectodomains are validated, defined probes that should prove valuable in measuring the immune-activating capacity of IgG Abs elicited by infection or vaccination or experimentally derived IgG and its variants.

T he Fc-mediated effector functions of Abs are being increasingly appreciated as making key contributions to Ab efficacy against infectious agents. Fc-mediated function is essential for the full protective activity of Abs to many viruses (1-7), including HIV (8)(9)(10), simian HIV (11), HSV type 2 (4), and influenza (6). Moreover, a broadly neutralizing Ab (6F12) directed against the influenza A hemagglutinin (HA) stalk region required FcgR engagement for full protective activity (12), making the stalk region a promising vaccine target for neutralization (13) and Fc receptor (FcR)-mediated protection (14,15). Thus, although HA inhibition by neutralizing Ab is the standard measure of a protective response to vaccination against influenza (16), Ab-dependent cellular phagocytosis (17), Ab-dependent cellular cytotoxicity (7,18,19), and other FcgR-dependent functions also contribute to immunity and form part of the optimal vaccine response (14,20). The capacity of IgG to facilitate FcgR-mediated Ag presentation (21) and induce dendritic cell activation (22) provides another mechanism for Fc-dependent effects in immunity to some infectious agents (23) and immunity more generally (24). Taken together, effector functions triggered by the engagement of FcRs are key contributors to the efficacy of Ab-mediated protection in natural immunity, vaccine responses, and treatments with therapeutic Abs (25,26).
Despite being fundamental to Ab-mediated immunity, FcRmediated effector functions are difficult to define and measure, in part because of the different receptors and the plethora of cell types and responses that may be generated. Despite this diversity, FcR activation of cells occurs when Abs aggregated by complexing with Ag present a cluster of Fc regions that cross-link FcRs and trigger subsequent downstream signaling.
Structural studies of the interaction of the human IgG1 Ab Fc region with its various cellular receptors [FcgRI (27,28), FcgRIIa (29,30), FcgRIIb (31), and FcgRIIIa (32,33)], in combination with biosensor studies, defined the atomic basis of the 1:1 interactions of single FcR ectodomains with an IgG1-Fc (34,35). However, the physiological interaction of IgG immune complexes (ICs) and FcgR requires avid binding of the complex through the display of multiple Fc regions of "near-neighbor" IgGs to engage and cluster multiple FcgRs on the cell surface (36). Hence, differences in the opsonization of targets by different IgGs influence interactions with FcgRs, chiefly by the density, size (37,38), and topology of presentation of the Fc regions. Although not easily predicted, these effects may profoundly affect cellular functions (39). Thus, how Ag-bound Abs are presented for sensing by FcgRs underpins effector functions, such as Ab-dependent cellular cytotoxicity and Ab-dependent cellular phagocytosis, but is not simple to quantitate experimentally. Cell-based assays for such key cellular responses (40)(41)(42)(43)(44)(45) are difficult to standardize, so there is a need for an assay that simply evaluates the capacity of an immune complex (IC) to present Fcs for avid binding by FcgRs.
In an attempt to recapitulate the avid sensing of ICs by cell surface FcgRs, our approach used a single, biotin-tagged polypeptide containing two FcgR ectodomains where the two Fc-binding modules were linked by the FcgRIIa membrane proximal stalk region. In this approach, the selective binding of these dimeric FcgR ectodomains to ICs is dependent on the presentation of pairs of "near-neighbor" Fc regions by the IC to bind both ectodomain modules of the receptor dimer. These dimeric recombinant soluble (rs)FcgR probes recapitulated many properties of the cell surface receptors, including selectivity for binding IgG subclasses and the binding affinities and specificities of the allelic forms of FcgRIIIa and FcgRIIa. Furthermore, the afucosyl form of IgG1 was more active in dimeric rsFcgRIIIa binding than the normally glycosylated mAb. The binding of the dimeric rsFcgRIIIa probe correlated with NK activation by ICs. This assay is useful for evaluating the functional activity of IgGs binding different Ags and epitopes and in different forms, including mutants, subclasses, and glycoforms.  (46)] was transfected using Lipofectamine 2000 into HEK293EBNA cells and Expi293F cells (Thermo Fisher Scientific, Scoresby, Australia). Transfectants were selected with 100 mg/ml hygromycin and maintained in 50 mg/ml hygromycin. Expression using Expi293F-BirA cells followed the manufacturer's instructions for Expi293F cells. All hexahistidine-tagged proteins were purified by affinity chromatography on a TALON Superflow (BD Biosciences), as described previously (47). RsFcgRIIa H131 ectodomain monomer and dimer expression constructs. All PCR reactions were performed using polymerase Pwo (Roche) or AccuPrime Pfx polymerase (Thermo Fisher). All other DNAmodifying enzymes were from New England Biolabs. The construction of a vector encoding monomeric rsFcgRIIa with a hexahistidine and biotin ligase target tag used amplification from the cDNA encoding FcgRIIa (clone Hu3.0, which includes a unique BamHI site) (48) with the product ligated to a codon-optimized sequence encoding a hexahistidine tag and biotin ligase target sequence (GenScript USA) so that the C-terminal sequence was GPGSSSHHHHHHPGGGLNDIFEAQKIEWHE, with the underlined residue corresponding to Gly 175 of FcgRIIa (i.e., Gly 211 of the precursor, RefSeq NP_001129691). To make a dimerized receptor construct, the clone Hu3.0 was amplified (using the equivalent of primers GTAGCTC-CCCCAAAGGCTG and GGGTGAAGAGCTGCCCATG (GGG corresponding to the antisense for the codon of FcgRIIa Pro 179 [i.e., Pro 215 of the precursor]); the blunt product was ligated together with T4 ligase, and a correctly orientated tandem dimer of the receptor ectodomain was subcloned. Using standard molecular biology, a BamHI fragment from this construct was subcloned into the unique BamHI site in the monomeric vector to produce a vector containing the endogenous FcgIIa leader, the dimeric ectodomain sequence described above, and the C-terminal GPGSSSHHHHHHPGGGL-NDIFEAQKIEWHE tag. The monomeric or dimeric rsFcgRIIa proteins were expressed from pAPEX-3p-X-DEST (pBAR424) or pCR3-DEST, as described previously (49,50), and using HEK293EBNA-BirA cells or Expi293F-BirA cells, respectively. RsFcgRIIIa monomer and dimer expression constructs. Codonoptimized sequences encoding the ectodomains of rsFcgRIIIa Val 158 and Phe 158 with the N-terminal sequence MVLSLLYLLTALPGISTEDLPKAVVFL and the C-terminal sequence QGPSMGSSSPGPGSSSHHHHHHPGGGLNDIFEAQKI-EWHE (underlined Q corresponds to residue Gln 172 , or Gln 191 of the precursor, RefSeq NP_000560) were synthesized by GeneArt (Invitrogen) and transposed by BP clonase reaction into pENTR1A (Invitrogen). The native protein leader sequences in these constructs were replaced with MVLSLLYLLTALPGIST and validated using Signal P (51). PCR with the primer pairs 59-AGCGAGGACCTGCCTAAGGCCGT-39 and 59-CGGA-GAACTAGAGCCCATGCTG-39 and 59-CGGAGAACTAGAGCCCATG-CTG-39 and 59-GGGCCTGGCAGCTCCTCTC-39 generated an ectodomain and ectodomain in vector product, respectively, which were ligated to generate the tagged dimeric rsFcgRIIIa ectodomain sequences. The underlined residues of QGPSMGSSSPSE are encoded by the linking sequence between the tandem ectodomains; Q corresponds to residue Gln 172 of the first ectodomain, and E corresponds to residue Glu 1 of the tandem ectodomain. After sequence validation, LR clonase reactions (Invitrogen) with pCR3-DEST generated expression vectors encoding monomeric and dimeric forms of rsFcgRIIIa Val 158 and Phe 158 proteins. These recombinant receptors were then expressed in Expi293F-BirA cells as described above. RsFcgRIIa Arg 131 monomer and dimer expression constructs. Similarly to the rsFcgRIIIa monomer and dimer expression constructs, a codonoptimized sequence encoding an N-terminal leader sequence MVLSLLYLL-TALPGILSA and C-terminal sequence GSSSPGSSSHHHHHHPGGGLNDI-FEAQKIEWHE, with the underlined residue corresponding to Gly 175 of FcgRIIa, was used to flank the optimized codons for mature FcgRIIa Arg 131 codons 1-175. Using PCR, the DNA encoding the ectodomain module (30) (PDB ID code: 3RY5) was joined to itself in tandem, such that the encoded junction of the duplicated ectodomain modules was ITVQVPSMGSSSPAAPPK (the linking peptide sequence between the modules is underlined); the first module retained the sequence for the leader, and the second module retained the sequence for the C-terminal tag. Expression of this pCR3-based construct used Expi293F-BirA cells, as described above. Ab expression constructs. The production of a chimeric IgG1 comprising a mouse leader and VH sequence (from TIB142 anti-trinitrophenyl [TNP]; American Type Culture Collection) joined to a human IgG1 C region sequence was described previously (49). Chimeric anti-TNP k L chain consisting of TIB142VL and human constant k was produced from a codon-opsonized construct synthesized by Bioneer Pacific (Kew, Australia). An IgG4 chimeric H chain comprising TIB142 leader and VH and IgG4 constant (Bioneer Pacific) sequence was codon optimized, synthesized, and transferred into pCR3. Likewise, the TNP-specific mouse VH sequence was joined to a human IgG2 genomic C region sequence (accession no. J00230) using standard molecular biology techniques, and the chimeric Ab sequence was subcloned into pCR3. FcgR-binding assays using dimeric rsFcgR-biotin. IgG capture reagent, F(ab9) 2 goat anti-human IgG, F(ab9) 2 (10 mg/ml), TNP-BSA (20 mg/ml), or influenza A HA (1 mg/ml) was prepared in PBS and adsorbed (50 ml/ well) to plates (MaxiSorp; Nunc). For the analysis of patient samples or IvIg, wells (typically three) were directly coated with 5 mg/ml IvIg (INTRAGAM P), and signals from these wells were used to normalize the FcR activity of the test samples. Coated wells were subsequently blocked with PBS containing 1 mM EDTA and 1% (w/v) BSA (Fraction V; Sigma-Aldrich); 1% HSA was used if subsequent testing was of patient or IvIg samples. IgG samples (typically diluted from a starting concentration of 1-5 mg/ml) were incubated with the IgG capture reagent, TNP-BSA, or Ag-coated wells for 1 h at 37˚C. Plates were washed five times with PBS containing 0.05% Tween-20. The Ab-bound plates were incubated with 0.2 mg/ml purified dimeric rsFcgRIIa-biotin or 0.1 mg/ml purified dimeric rsFcgRIIIa-biotin, in PBS diluent containing 1 mM EDTA, 0.05% Tween-20, and 1% (w/v) BSA for 1 h at 37˚C. After five cycles of filling and emptying with wash buffer, High Sensitivity Streptavidin-HRP (Thermo Fisher), 1/10,000 in diluent buffer, was added for 1 h at 37˚C, followed by 8-10 cycles of filling and emptying with wash buffer and development with TMB Single Solution (Thermo Fisher). The reaction was stopped by addition of an equal volume of 1 M HCl, and absorbance at 450 nm (A450nm) was determined immediately. Delayed determination of absorbance can produce an apparent prozone effect artifact because precipitation of the colorimetric product occurs at high concentration. Bound IgG was measured using polyclonal rabbit anti-human IgG-HRP (Agilent Technologies-Dako, 1/10,000 dilution; Sigma-Aldrich, 1/20,000), and plates were blocked using HSA depleted of IgG by protein G chromatography (GE Life Technologies) or 1% FCS in PBS. NK cell-activation assay. NK cell activation was measured by HA:anti-HA IC-dependent induction of intracellular IFN-g and cell surface CD107a, as previously described (6,7,18,(52)(53)(54). Briefly, 96-well ELISA plates (Nunc, Rochester, NY) were coated with 600 ng of purified HA protein overnight at 4˚C in PBS. The wells were washed five times with PBS and incubated with heat-inactivated (56˚C for 1 h) sera/plasma or IvIg for 2 h at 37˚C. Plates were washed seven times with PBS, and 10 6 PBMCs were added to each well. Healthy donor PBMCs were obtained from buffy packs provided by the Australian Red Cross. PBMCs were isolated by Ficoll-Paque PLUS (GE Healthcare, Madison, WI), washed with RPMI 1640 supplemented with 10% FCS, penicillin, streptomycin, and L-glutamine (Life Technologies, Grand Island, NY), frozen in FCS containing 10% DMSO, and stored in liquid nitrogen. Thawed PBMCs were washed twice with RPMI 1640 supplemented with 10% FCS, penicillin, streptomycin, and L-glutamine before addition to each well. Anti-human CD107a allophycocyanin-H7 Ab (clone H4B4; BD Biosciences, San Jose, CA), 5 mg/ml brefeldin A (Sigma-Aldrich), and 5 mg/ml monensin (Golgi Stop; BD Biosciences) were added to the cells and incubated for 5 h at 37˚C with 5% CO 2 . PBMCs were then incubated with 1 mM EDTA to minimize cell adherence to the plates, anti-human CD3 PerCP (clone SP34-2), and antihuman CD56 allophycocyanin (clone B159; both from BD Biosciences) for 30 min at room temperature in the dark. Cells were fixed with 1% formaldehyde (Sigma-Aldrich) for 10 min and permeabilized with FACS Permeabilizing Solution 2 (BD Biosciences) for 10 min. PBMCs were then incubated at room temperature for 1 h with IFN-g Alexa Fluor 700 (clone B27; BD Biosciences) in the dark. Finally, cells were fixed with 1% formaldehyde and acquired on an LSR II flow cytometer (BD Biosciences).

Reagents
The NK cell line NK-92 (55) expressing human FcgRIIIa Val 158 (GFP-CD16 [176V] NK-92) was used to perform some of the NK cell-activation assays and was kindly provided by Dr. Kerry Campbell (Institute for Cancer Research, Philadelphia, PA). ELISA plate coating was performed as described above. However, following PBS washing, an additional blocking step was performed with PBS containing 5% BSA (Sigma-Aldrich) and 0.1% Tween-20 (U-CyTech) for 2 h at 37˚C. Once blocked, plates were washed with PBS and incubated with heat-inactivated sera/plasma or IvIg for 2 h at 37˚C. Plates were washed with PBS, and 2 3 10 5 GFP-CD16 (176V) NK-92 cells were added to each well and incubated at 37˚C with 5% CO 2 for 5 h. Antihuman CD107a allophycocyanin (clone A4H3; BD Biosciences) and 1 mM EDTA were added to the cells for 30 min at room temperature in the dark. The GFP-CD16 (176V) NK-92 cells were washed twice with PBS, fixed with 1% formaldehyde, and acquired on an LSR II flow cytometer. Analysis was performed using FlowJo X software version 10.0.7r2 (TreeStar, Ashland, OR). Data and statistical analysis. Statistical analysis was performed with GraphPad Prism version 6.05 (GraphPad, San Diego, CA). Binding data (Figs. 1-4) were fitted using Prism software, to log(agonist) versus response (variable slope, constraining bottom value = 0, the top value was allowed to vary freely). The 95% confidence interval (C.I.) for the individual fitted values for EC 50 are indicated graphically for the representative curves in Fig. 1E and in the insets in Fig. 2B-E. Horizontal bars in Fig. 2B-E representing EC 20 2 EC 80 were calculated similarly. Cumulative data shown as EC 50 with error bars are presented as mean 6 95% C.I. in Figs. 2F, 3K, and 3L. Curve fitting for some low-affinity interactions, with IgG2 and IgG4, were ambiguous, and EC 50 values were only used from data fitting at r 2 . 0.97. When binding was undetectable or too weak to enable adequate fitting, the EC 50 value was not calculated.
Binding to wild-type (WT) versus nonfucosylated IgG in Fig. 2F and, likewise, binding to dilutions of pre-2009 versus 2010 preparations of IvIg were evaluated by the Mann-Whitney unpaired t test. Data in Fig. 4 from five experiments were analyzed together by normalizing all OD values to the A450nm value for 250 mg/ml IvIg batch 1848-opsonized H3 HA bound by HRP-conjugated anti-IgG (normalized point denoted by x in Fig. 4). Similarly in Fig. 5A, 5B, and 5D, A450nm values were normalized using the A450nm value for 250 mg/ml IvIg batch 1848-opsonized H1 HA bound by dimeric rsFcgRIIIa Val 158 . For Figs. 6 and 7, A450nm values were normalized to receptor binding to directly coated IvIg (5 mg/ml), and the pairs of assays were fitted by linear regression; data in Fig. 5D were fitted to log(agonist) versus response. Correlations were assessed by nonparametric Spearman analysis.

Characterization of dimeric rsFcgRs using model human IgG1 ICs
The universal requirement for "near-neighbor" clustering of lowaffinity FcRs by appropriately arrayed Fc portions in ICs underpins proinflammatory Ab-dependent effector functions. Because cell surface FcgRII and FcgRIII are low-affinity receptors for IgG that avidly bind ICs, we engineered genetic homodimers of their ectodomains, using a flexible linking sequence from the membrane proximal stalk, to generate defined probes that avidly bind to "near-neighbor" Ab pairs in ICs.
The biotin-labeled genetically fused dimeric ectodomains of FcgRIIa His 131 , FcgRIIIa Val 158 , and FcgRIIIa Phe 158 were produced in cells expressing BirA ligase in the endoplasmic reticulum (46). The activities of these purified human dimeric rsFcgRs were initially characterized by testing their binding activities to different forms of model ICs formed with human IgG1 and TNP-BSA Ag or by capture with anti-F(ab9) 2 . Subnanomolar concentrations of the dimeric rsFcgRIIa and dimeric rsFcgRIIIa had detectable binding to both forms of IgG ICs, whereas the monomeric forms of the proteins had .1000-fold lower binding activity (Fig. 1). The weak binding of the receptor monomers (Fig. 1A) indicates that the avid binding of the receptor dimers requires pairs of Ab Fc regions that are presented with a proximity to each other that allows the simultaneous binding of the two receptor modules making up the dimeric rsFcgR. It is such "near-neighbor" IgGs that are required for activation of cells via engaging and clustering FcRs.

Dimeric rsFcgR binding detects the modified FcgR-binding activity of Fc variants
The dimeric rsFcgR assays were further validated by analyzing the interactions of activating and inactivating variants of IgG1. Mutation of the lower hinge of IgG1, LL(234-235)AA (LALA mutant), greatly diminishes FcgR binding activity and FcR mediatedfunction (56,57). The loss of binding activity was recapitulated in the dimeric rsFcgR assays, with dimeric rsFcgRIIa and dimeric rsFcgRIIIa binding only weakly to IgG1-LALA ICs (Fig. 2B-E). In contrast, enhanced binding of dimeric rsFcgRIIIa to afucosyl IgG1 (58) was demonstrated with the receptor dimer assay (Fig. 2D, 2E). Both the higher-affinity (Val 158 , p = 0.008) and the lower-affinity (Phe 158 , p = 0.016) allelic forms of the receptor had significantly increased affinity to the nonfucosyl IgG ( Fig. 2D-F). For dimeric rsFcgRIIIa Val 158 , the EC 50 decreased from 118 ng/ml (86-150, 95% C.I.) with WT IgG to 43 ng/ml (21-65, 95% C.I.) with nonfucosyl IgG; likewise for the Phe 158 allele, the EC 50 decreased from 275 ng/ml (89-460, 95% C.I.) to 97 ng/ml (55-139, 95% C.I.). Thus, the dimeric rsFcgR assays are useful for discriminating variants of IgG with enhanced or diminished FcR activity. IC formation measured by determining the bound IgG with antihuman IgG had a broad response profile, with the 20-80% response occurring over an 18-fold increase in IgG concentration (i.e., Fig. 2A, open bar; anti-IgG, EC 20 -EC 80 for WT IgG1 = 8-151 ng/ml). In contrast, a distinctive feature of the dimeric rsFcgR binding activity was a narrow response profile, with the 20-80% response occurring over a 2-3-fold increase in IgG concentration (i.e., Fig. 2B, open bar; FcgRIIa H131, EC 20 -EC 80 ; WT IgG1 = 150-360 ng/ml). The steeper character of these dimeric rsFcgR response curves can also be described by the numerical equivalent of the Hill constant (Table I). Although anti-IgG binding is typically described by a coefficient near or ,1, the binding curves in Fig. 2 are typical of dimeric rsFcgR profiles for binding to IgG1 ICs with coefficients, in this case, ranging from 2.3 to 4.1 ( Table I). The steeper receptorbinding curves reflect the requirement for closely placed "nearneighbor" IgG-Fcs for dimeric rsFcgR binding, which produces a response over a narrow range of IgG concentration that would not be obvious from measuring Ab binding or titer (e.g., compare Fig. 2A with Fig. 2B-E).

Dimeric rsFcgR binding is determined by IgG subclass and presentation of the Fc
The effect of IgG subclass on dimeric rsFcgR binding was investigated using two established methods of forming model ICs (37,38). First, the capture of IgG by the F(ab9) 2 anti-human F(ab9) 2 to form ICs (Fig. 3A) results in the presentation of Fc regions of the IgGs in varied orientations for dimeric rsFcgR binding (Fig. 3C, 3E, 3G, 3I). Second, TNP hapten-specific rIgG and TNP-BSA form ICs (Fig. 3B) in which all IgGs are oriented by the same variable domain:hapten interaction and thus, display a more uniform presentation of Fcs for dimeric rsFcgR binding (Fig. 3D, 3F, 3H, 3J).
The nature of the IC also influenced dimeric rsFcgR binding. The binding differed most markedly for the weakest FcR interactions observed. For example, dimeric rsFcgRIIa His 131 binding activity was just detectable with ICs formed with anti-human F(ab9) 2 at 5 mg/ml IgG4 (Fig. 3C), whereas when binding to IC of IgG4 formed with the Ag TNP-BSA, the signal was ∼6-fold higher at 5 mg/ml IgG4 (Fig. 3D, 3L, EC 50 ∼ 2 mg/ml). For this lowest-affinity FcR interaction, Ab presentation in these different forms of ICs profoundly affects receptor binding. For the higheraffinity interactions, differences between the two methods of IC formation for FcgR binding were less apparent. For example, the dimeric rsFcgRIIa His 131 bound similarly to the anti-human F(ab9) 2 :IgG1 and IgG2 ICs (EC 50 = 400 and 300 ng/ml respectively, Fig. 3C, 3K) and to the TNP-BSA:IgG1 and IgG2 ICs (EC 50 = 280 and 190 ng/ml, respectively, Fig. 3D, 3L).
The dimeric rsFcgRIIa Arg 131 bound anti-human F(ab9) 2 ICs with the ranking IgG3 . IgG1 .. IgG2 ∼ IgG4 (Fig. 3E); a similar hierarchy (IgG1 . IgG2 ∼ IgG4) was apparent with anti-TNP ICs (Fig. 3F, Table II). Thus, the allelic forms of dimeric rsFcgRIIa recapitulate the IgG subclass binding behavior of their cellular counterparts, and their weak binding to IgG4 is influenced by Fc presentation in different forms of ICs.
Analysis of the allelic variants of dimeric rsFcgRIIIa proteins showed, as expected, that the dimeric rsFcgRIIIa Val 158 had greater binding activity to IgG1 IC than the lower-affinity dimeric rsFcgRIIIa Phe 158 (EC 50 Table II) were largely comparable to the binding of ICs of the different IgG subclasses to cell surface-expressed FcgRIIIa (34,37,38).
It is notable that, in this study (Fig. 3) and in other studies (37,38), differences occur in FcgR binding to ICs made with anti-F(ab9) 2 or TNP-BSA Ag (Table II). Using the assay to evaluate FcgRIIIa interactions with anti-TNP ICs revealed, similarly to FcgRIIa, that the weakest interactions between dimeric rsFcgRIIIa and ligand were influenced the most by the manner of formation of the IC. The interactions of ICs of IgG2 and IgG4 with the higher-affinity (Fig. 3G, 3H) and lower-affinity (Fig. 3I,  3J) alleles of the dimeric rsFcgRIIIa were undetectable with the anti-TNP-formed ICs (Fig. 3H, 3J), but binding was measurable for ICs of F(ab9) 2 anti-human F(ab9) 2 , with IgG2 binding the dimeric rsFcgRIIIa Val 158 (Fig. 3G, EC 50 = 620, 95% C.I. = 520-740 ng/ml). Clearly, the binding, or nonbinding, to IgG subclasses by FcgRIIa (compare IgG4 IC binding in Fig. 3C and Fig. 3D) and FcgRIIIa (compare IgG2 IC binding in Fig. 3H and Fig. 3G) can depend on the manner of incorporation of IgG into the IC.
In summary, the dimeric rsFcgRs demonstrate binding equivalent to cell surface FcgRs with regard to the hierarchy of binding to IgG subclasses, e.g., binding to IgG3 . other subclasses, (see figure 2A in Ref. 37); the subclass specificity of the polymorphic forms of FcgR (e.g., binding of IgG2 by FcgRIIa His 131 . FcgRIIa Arg 131 ); the expected differences in binding strength of the polymorphic forms of FcgR (e.g., FcgRIIIa Val 158 . FcgRIIIa Phe 158 ); and the appropriately altered binding of LALA hinge mutant and nonfucosyl-variant IgG, b12. In addition, how the Fc is presented influences dimeric rsFcgR binding, especially for low-affinity interactions, in particular IgG2 and IgG4 [e.g., FcgRIIa His 131 with IgG4 ICs (Fig. 3C, 3D) and FcgRIIIa with IgG2 (Fig. 3G-J)]. The use of dimeric rsFcgRs in the analysis of anti-influenza A immunity: dimeric rsFcgR binding to serum IgG-opsonized influenza HA There is increasing evidence that Fc-mediated functions are important in clearance of influenza infections (6,12), but typical assays to measure such responses through activation of cells or killing of target cells remain cumbersome. The dimeric rsFcgRbinding assay was evaluated in the context of Abs to this common viral pathogen. Dimeric rsFcgR-binding activity to ICs formed between influenza A HA and human IgG Abs was assessed using IvIg and HA from the 2009 H1N1 pandemic virus and from an H3N2 virus. Because IvIg is prepared from the sera of thousands of individuals, it is composed of the IgG Ab repertoire at a population level and opsonizes all epitopes of HA for which specific IgG molecules exist in the population. Like the model ICs (Figs. 2,  3), H3N2 HA opsonized with IvIg (Fig. 4) showed a steeper dimeric rsFcgR-binding profile (Fig. 4B, 4C) compared with the broader profile for the detection of opsonizing IgG (anti-IgG, EC 20 -EC 80 = 20-275 mg/ml, Fig. 4A). The estimated EC 50 values for dimeric rsFcgR binding (rsFcgRIIa His 131 EC 50 H3 ∼ 370 mg/ml; FcgRIIIa Val 158 EC 50 H3 ∼ 250 mg/ml) were several fold above the EC 50 for IgG binding (anti-IgG EC 50 H3 = 74 mg/ml), a feature also consistent with the model ICs. Enumeration of the EC 50 values for dimeric rsFcgR binding could only be approximated by curve fitting because the binding did not reach saturation (r 2 = 0.88-0.92, Fig. 4B, 4C). The same trends were apparent in dimeric rsFcgRIIa and dimeric rsFcgRIIIa binding to ICs formed by IvIg opsonization of A (H1N1)pdm09 HA, although the levels of opsonization and dimeric rsFcgR binding were less than for the HA from H3N2 A/Perth/16/ 2009 (Fig. 4B, 4C). For input concentrations of IvIg , 100 mg/ml, opsonization of HA by IgG, although well detected by anti-IgG, is sparse and therefore, bound IgGs cannot simultaneously engage the two FcgR-binding modules of the dimeric rsFcgRIIa or dimeric rsFcgRIIIa. The relatively steep reaction profiles of the dimeric rsFcgR-binding curves dictate that, rather than determining end point titer, a meaningful measure of the FcR activity for comparing different sera is the dimeric rsFcgR-binding signal at a fixed Ab concentration (or plasma/serum dilution) within the EC 20 -EC 80 range of the assay.  Binding to the opsonized H1 and H3 Ags is shown for anti-IgG HRP (A), dimeric rsFcgRIIa His 131 (B), and dimeric rsFcgRIIIa Val 158 (C). The OD value for anti-IgG binding at 250 mg/ml of IvIg binding to H3 ( x ) was used to normalize ODs for each independent experiment (n = 5). A clipped 95% C.I. error bar is denoted by the pound sign ( # ). functional activity toward virally infected cells also increases. Furthermore, NK cell-activating capacity of HA Abs trended toward an increase in pooled IgG preparations after the 2009 H1N1 influenza pandemic (54). Therefore, we assessed whether the dimeric rsFcgR assay was capable of detecting subtle changes in the FcR activity of HA Abs at the population level, as represented in pooled IgG (IvIg), after the 2009 H1N1 pandemic. Thus, IvIg was used to opsonize H1N1 pandemic HA, and dimeric rsFcgRIIIa binding was measured. IvIg preparations during 2010 showed higher dimeric rsFcgRIIIa-binding activity, suggesting that the FcR-activating capacity of Abs specific for the HA of influenza virus A(H1N1)pdm09 was increased subsequent to the emergence of the pandemic in 2009 (Fig. 5A). Interestingly, this increased dimeric rsFcgRIIIa activity appeared to be transient, peaking and declining in 2010. When the dimeric rsFcgRIIIa activity was compared pre-2009 with 2010, the postpandemic increase in receptor activity was significant (p = 0.0003) over the three Ab concentrations tested (Fig. 5B). Consistent with the dimeric rsFcgR assay being a measure of a correlate of immunity, the peak in FcgR activity in 2010 was mirrored by a contemporaneous peak in HA-inhibition titer (Fig. 5C). The correlation between these differing, but contemporaneous, Ab functions (Fig. 5D, p , 0.0001) suggests a biological relevance for the dimeric rsFcgR assay and lends support to the idea of coordination of Ab functions in immunity.

Dimeric rsFcgR binding is a predictor of NK cell activation
Next, the relationship between dimeric rsFcgR binding and cellular FcgR effector function was investigated. The ability of dimeric rsFcgR activity to correlate with cellular activation by HA-specific Abs was tested by comparison with the activation of NK-92 cells expressing FcgRIIIa Val 158 , a well-established system for measuring Ab-mediated NK activation. Plasma from individuals was used to separately opsonize HA of Perth (H3N2) and to activate NK-92-FcgRIIIa Val 158 cells and for binding dimeric rsFcgRIIIa Val 158 . The cell and dimeric rsFcgRIIIa activities correlated strongly (p , 0.0001), validating use of the dimeric rsFcgRIIIa receptor to predict cellular responses (Fig. 6). Thus, the FcR dimer assay ranks individuals on the basis of the NK cell-activating potential of their anti-HA IgG Abs. Next, the ability of the dimeric rsFcgR assay to predict the capacity of anti-HA ICs to activate fresh blood primary NK cells was assessed. The plasma from 30 individuals was used to opsonize A(H1N1)pdm09 HA, and these ICs were used to activate NK cells and bind dimeric rsFcgR (Fig. 7). NK cell activation (intracellular IFN-g and/or CD107a surface expression) correlated more strongly with binding of the dimeric rsFcgRIIIa Val 158 (p = 0.005, Fig. 7C) than the anti-HA IgG Ab titer (p = 0.027, Fig. 7A) or Ab EC 50 (p = 0.79, Fig. 7B). Thus, dimeric rsFcgRIIIa binding activity to IgG-opsonized HA correlates with IC capacity for NK activation.

Discussion
FcgRs and IgG Abs are increasingly understood to play key roles in immunity to pathogens, vaccine responses, and autoimmunity. Although many effector functions can result from FcgR activation, all are triggered when multiple Fc regions of IgGs in ICs bind and cluster receptors to initiate signaling (59,60). To identify IgG molecules that bind to Ag and are sufficiently closely spaced to bind and cluster FcgRs, we produced defined dimeric soluble forms of FcgRIIa and FcgRIIIa as mimics of neighboring, signaling FcgRs on the cell surface. The usefulness of these probes was tested using IgG ICs formed with model Ags and with human Abs to influenza HA.
A general feature of the assay was a higher threshold and steeper reaction profile for dimeric rsFcgR binding to IC than the binding of anti-IgG. This follows from the requirement for divalent binding of the dimeric rsFcgR by bridging the closely spaced Fcs of neighboring Abs. At IgG concentrations above the EC 50 for opsonization, relatively small differences in the levels of opsonizing IgG presumably achieve a critical density of presented Fc regions that result in large increases in dimeric rsFcgR-binding activity.
The use of the dimeric rsFcgR assays demonstrated expected hierarchies of IgG subclass binding IgG3 . IgG1 ∼ IgG2 .. IgG4 for FcgRIIa (His 131 ) interactions and IgG3 . IgG1 .. IgG2 ∼ IgG4 for FcgRIIa (Arg 131 ) interactions (Fig. 3, Table II), which were comparable to the reported activities of the cell surface FcgRs (34,37,38); the FcgRIIIa Phe 158 allele has weaker binding activity than the Val 158 allelic form, and both FcgRIIIa Val 158 and FcgRIIIa  Dimeric rsFcgR binding has superior correlation with donor NK activation by influenza A HA, A(H1N1)pdm09 than with anti-IgG end point titer or EC 50 . ICs were formed by reacting plasma from 30 individuals with plate-bound HA from A(H1N1) pdm09 virus. Activation of the CD56 + NK cell population of PBMCs by these ICs was measured by intracellular stain for IFN-g and/or surface expression of CD107a and was correlated with IgG end point titer (A), IgG 1/EC 50 (B), and dimeric FcgRIIIa Val 158 binding, normalized using binding in wells directly coated with IvIg (5 mg/ml) (C). Each pair of assays was correlated using nonparametric Spearman analysis and fitted by linear regression.
Phe 158 showed the expected higher binding activity to afucosylated IgG1; FcgRIIa and FcgRIIIa binding to the weaker interacting subclasses, IgG2 and IgG4, was affected by the nature of the IC; dimeric rsFcgR-binding activity to IgG-opsonized Ag showed a higher threshold and steeper reaction profile than the binding of anti-IgG; as expected, receptors failed to bind to IgG1 containing the established FcR binding-inactivation LALA lower hinge mutant; and dimeric rsFcgRIIIa binding activity of ICs correlated with NK activity measured using FcgRIIIa-expressing NK-92 cells or primary NK cells and detected a transient increase in the activity of anti-H1pdm HA Abs in IvIg prepared during 2010 postpandemic H1N1 influenza. This last finding demonstrated that the dimeric rsFcgR assay was at least as capable as cell-based assays (54) in revealing changes in Ab activity.
The strength of engagement of FcgRs by IgG depends on Fcintrinsic properties, such as amino acid residues at the interaction interfaces and glycosylation, but it also depends on how an Ab reacts with Ag to form an IC or opsonized target. Interestingly, the subclasses IgG2 and IgG4, which have weaker interactions with FcgRs, exhibited different FcgRIIa-and FcgRIIIa-binding activities when oriented by capture with anti-F(ab9) 2 compared with when they were bound by TNP-BSA Ag. Both of these methods of forming ICs are described in studies that defined cellular FcgR interactions and the effect of IC size on these interactions (37,38). FcgRII and FcgRIII binding to Fc are low-affinity interactions for which the 1:1 intrinsic FcR:Fc reactions have been well described. However, the biological interactions of FcgRII and FcgRIII with ICs are multivalent and even with the divalent interactions described in this article, the net outcome of weak but avid binding reactions was profoundly affected by differences in ligand presentation in the IC. This phenomenon is important in evaluating the activities of therapeutic Abs, particularly agonist mAbs (61), developed in low FcR-binding formats, such as IgG2 or IgG4 backbones. Thus, although the Fc-intrinsic effects can be engineered by altering amino acid residues and glycosylation, the consequences of opsonization-related effects are difficult to predict because Abs are highly diverse, with ∼10 11 possible Ab: Ag interactions (62).
The dimeric rsFcgR assay described in this article has a number of advantageous features. Dimeric rsFcgRs are genetically defined dimers, have site-specific biotinylation, and, importantly, bind ligand without further complexing (e.g., to streptavidin) to provide multivalency. The intrinsic FcgRIIa and FcgRIIIa interactions with IgG are 1:1 and were thoroughly characterized by surface plasmon resonance assays (e.g., BIAcore, figure 3 in Ref. 37), but biological responses mediated by these receptors result from multivalent interactions. The assay of the dimeric receptor interaction with two Fcs described in this article is a measure of the minimal avid interaction of IgG with the low-affinity FcgRs. The dimeric rsFcgR binds to adjacent Fcs when presented by two Abs in close proximity. The correlation of receptor activity assayed off the cell membrane with activity in a cellular context, and indeed with in vivo biology, is difficult to evaluate. Nonetheless, it can be noted that other cell-free approaches, which clearly produced key insights (20,(63)(64)(65)(66), require receptor complexing or immobilization resulting in measuring higher or ill-defined valency interactions, respectively. In contrast, dimeric FcgR binding measures the spatial relationship of pairs of Abs complexed with Ag, with the effects of receptor polymorphism, Ab subclass, and glycosylation contributing to the overall measure of FcgR activity.
Furthermore, the use of FcgR dimers is informative and does not require specialized equipment. It needs only limited amounts of sample and can be used to evaluate the FcgR activities of many types of IgG Abs bound to different Ags. Central to the assay is the biotinylated dimer rsFcgR, which could be easily incorporated into powerful, parallel high-throughput bead-based assays (20,63,64,(66)(67)(68), adapting these for detecting closely spaced Ab pairs.
Evaluating the quality of IgG responses for FcgR activation will be of particular importance for understanding natural (67) and vaccine-induced immunity (6,18,20,24,69) to viruses and other infectious diseases. The detection of a transient increase in the activity of IvIg from 2010 toward H1N1 pandemic HA, which correlated with HA-inhibition titer and NK activation, indicates that the dimeric rsFcgR assay may provide a useful measure of immunity. Moreover, dimeric rsFcgR may be of value for evaluating therapeutic IgG and IgG mixtures. Therapeutic IgG Abs have revolutionized the treatment of some cancers and rheumatoid arthritis; however, the application of therapeutic Abs to infectious disease may be more challenging in situations in which strain differences and a need for sterilizing immunity may necessitate the use of Ab cocktails of increasing complexity and functionality (70,71). Indeed, such cocktails of Abs are used for treatment of rabies virus (72), toxins (73), and, most recently, Ebola (74). The FcR-activating activity of such mixtures of Abs is generally greater than for single therapeutic Abs, except in situations in which the target Ags are highly abundant (such as CD20 on B cell tumors) or have repeating epitopes. The engineered FcgR ectodomain dimers described in this article measure the content of closely spaced, and thus, receptoraggregating, pairs of "near-neighbor" IgG Abs on opsonized targets. Clearly, they are important tools for the prediction of FcgR activation by Ag-specific IgG and thereby, the evaluation of Ab responses to vaccines, infections, and therapeutic Abs.