Multiple regions of human Fc gamma RII (CD32) contribute to the binding of IgG.

The low affinity receptor for IgG, FcγRII (CD32), has a wide distribution on hematopoietic cells where it is responsible for a diverse range of cellular responses crucial for immune regulation and resistance to infection. FcγRII is a member of the immunoglobulin superfamily, containing an extracellular region of two Ig-like domains. The IgG binding site of human FcγRII has been localized to an 8-amino acid segment of the second extracellular domain, Asn154-Ser161. In this study, evidence is presented to suggest that domain 1 and two additional regions of domain 2 also contribute to the binding of IgG by FcγRII. Chimeric receptors generated by exchanging the extracellular domains and segments of domain 2 between FcγRII and the structurally related FcεRI α chain were used to demonstrate that substitution of domain 1 in its entirety or the domain 2 regions encompassing residues Ser109-Val116 and Ser130-Thr135 resulted in a loss of the ability of these receptors to bind hIgG1 in dimeric form. Site-directed mutagenesis performed on individual residues within and flanking the Ser109-Val116 and Ser130-Thr135 domain 2 segments indicated that substitution of Lys113, Pro114, Leu115, Val116, Phe129, and His131 profoundly decreased the binding of hIgG1, whereas substitution of Asp133 and Pro134 increased binding. These findings suggest that not only is domain 1 contributing to the affinity of IgG binding by FcγRII but, importantly, that the domain 2 regions Ser109-Val116 and Phe129-Thr135 also play key roles in the binding of hIgG1. The location of these binding regions on a molecular model of the entire extracellular region of FcγRII indicates that they comprise loops that are juxtaposed in domain 2 at the interface with domain 1, with the putative crucial binding residues forming a hydrophobic pocket surrounded by a wall of predominantly aromatic and basic residues.

Cell surface receptors for the Fc portion of IgG (Fc␥R) are expressed on most hematopoietic cells, and through the binding of IgG they play a key role in homeostasis of the immune system and host protection against infection. Three structurally related but functionally distinct classes of Fc␥R have been defined: Fc␥RI, Fc␥RII, and Fc␥RIII (1)(2)(3). Fc␥RII is a low affinity receptor for IgG that binds only IgG immune complexes and is expressed on a diverse range of cells such as monocytes, macrophages, neutrophils, eosinophils, platelets, and B cells (1)(2)(3). Fc␥RII is involved in a number of immune responses including antibody-dependent cell-mediated cytotoxicity, clearance of immune complexes, release of inflammatory mediators, and regulation of antibody production (1)(2)(3)(4)(5)(6).
The extracellular region of Fc␥RII comprises two Ig-like disulfide-bonded extracellular domains that are related to the Ig superfamily proteins and are most closely related to the C2 set of Ig domains (7)(8)(9)(10)(11)(12). The two Ig-like domain extracellular region of Fc␥RII is structurally conserved in all of the Ig superfamily leukocyte FcRs (including Fc␥RI, Fc␥RIII, Fc⑀RI, and Fc␣RI) and presumably represents an Ig-interactive motif (13)(14)(15)(16)(17). The elucidation of the molecular basis of FcR-Ig interactions is fundamental for understanding the mechanisms by which these receptors mediate biological functions such as activation of inflammatory cells, induction of cytokine release, and destruction of pathogens. In previous studies we utilized chimeric Fc receptors to identify the IgG binding region of human Fc␥RII (18,19). Chimeric Fc␥RII/Fc⑀RI ␣ chain receptors were used to demonstrate that the second extracellular domain of Fc␥RII was responsible for the binding of IgG, with a single direct binding region located between residues Asn 154 and Ser 161 . Site-directed mutagenesis of the Asn 154 -Ser 161 region identified 5 residues as playing crucial roles in the binding of human and mouse IgG1 by Fc␥RII: Ile 155 , Gly 156 , Leu 159 , Phe 160 , and Ser 161 (20).
However, despite the direct demonstration of only a single region involved in the binding of IgG, there is compelling evidence to suggest that other regions of Fc␥RII contribute to binding. A genetic polymorphism of human Fc␥RIIa, the so called "responder/non-responder" system, results in an amino acid substitution in domain 2 at residue 131 (Arg 3 His), which has been shown to influence the binding of mouse IgG1 and human IgG2 (21)(22)(23). Similarly, in the mouse a genetic polymorphism of Fc␥RII, identified as differences at residues 116 and 161, defines the epitope of the anti-Ly17.2 mAb 1 that blocks the binding of IgG to this receptor (24,25). Our previous molecular modeling studies of Fc␥RII domain 2 (wherein the Asn 154 -Ser 161 binding region was located to an exposed loop region; the F/G loop) suggest that these functionally important amino acid changes are situated in the B/C and CЈ/E loops (containing residues 116 and 131, respectively), which are juxtaposed to the F/G loop (contains residue 161) at or near the interface with domain 1 (20). Furthermore, the studies using chimeric Fc␥RII/Fc⑀RI receptors have identified three regions in the structurally homologous receptor, Fc⑀RI, capable of directly binding IgE: residues 87-128, 130 -135, and 154 -161, which encompass the B/C, CЈ/E, and F/G loops respectively (1,18,19). Taken together, these findings suggest that the B/C and CЈ/E loops of Fc␥RII may in addition to the F/G loop also play a role in the binding of IgG by Fc␥RII. Also of interest is that while the role of domain 2 of Fc␥RII in Ig binding has been clearly defined, a role for domain 1 of Fc␥RII has not been determined. However, domain 1 of Fc⑀RI, although demonstrated to not have a direct role in IgE binding, has been shown to play an important role in high affinity binding (18,26) possibly by maintaining the structural integrity of the receptor or by providing additional contact sites. Since Fc␥RII is structurally related to Fc⑀RI, domain 1 of Fc␥RII may also play a similar role.
The possibility that domain 1 and the B/C or CЈ/E loop regions of domain 2 also contribute to the binding of IgG1 by Fc␥RII is addressed herein, using both chimeric receptor and site-directed mutagenesis strategies.

Generation of Chimeric Fc␥RII/Fc⑀RI and Mutant Fc␥RII Receptor cDNA Expression Constructs-Chimeric
Fc␥RII/Fc⑀RI ␣ chain or mutant Fc␥RII cDNAs were constructed by splice overlap extension (SOE) PCR (27) using the Fc␥RIIa NR cDNA (8) as template. SOE PCR was performed as follows. Two PCRs were used to amplify the Fc␥RII-Fc⑀RI or Fc␥RII fragments to be spliced together. The reactions were performed on 100 ng of the Fc␥RIIa NR cDNA in the presence of 500 ng of each oligonucleotide primer, 1.25 mM dNTPs, 50 mM KCl, 10 mM Tris-Cl, pH 8.3, and 1.5 mM MgCl 2 using 2.5 units of Taq polymerase (Amplitaq, Perkin-Elmer) for 25 amplification cycles. A third PCR reaction was performed to splice the two fragments and amplify the spliced product and included 100 ng of each fragment (purified by size fractionation through an agarose gel) (28) with the appropriate oligonucleotide primers under the PCR conditions above.
Sequences derived from Fc⑀RI ␣ chain are underlined, Fc␥RII is not underlined, and nonhomologous sequences including restriction enzyme sites used in cloning of the PCR products are in boldface type. Nucleotide positions refer to the previously published Fc␥RIIa and Fc⑀RI ␣ chain cDNA sequences (8,16).
Chimeric and mutant receptor cDNA expression constructs were produced by subcloning the cDNAs into the eukaryotic expression vector pKC3 (29). Each cDNA was engineered in the PCRs to have an EcoRI site at its 5Ј end (the 5Ј-flanking oligonucleotide primer NR1 containing an EcoRI recognition site) and a SalI site at the 3Ј end (the 3Ј-flanking oligonucleotide primer EG5, containing a SalI recognition site), which enabled the cDNAs to be cloned into the EcoRI and SalI sites of pKC3. The nucleotide sequence integrities of the chimeric cDNAs were determined by dideoxynucleotide chain termination sequencing (30) using Sequenase TM (U.S. Biochemical Corp.) as described (31).
Monoclonal Antibodies and Ig Reagents-The anti-Fc␥RII mAb 8.2 was produced in this laboratory (32). The mouse IgE anti-TNBS mAb (TIB142) was produced from a hybridoma cell line obtained from the American Type Culture Collection (Rockville, MD); the mouse IgG1 anti-TNBS mAb (A3) was produced from a hybridoma cell line, which was a gift of Dr. A. Lopez (33). Human IgG1 myeloma protein was purified from the serum of a myeloma patient as described (34). Human IgG1 oligomers were prepared by chemical cross-linking using S-acetylmercaptosuccinic anhydride (Sigma) and N-succinimidyl 3-(2pyridyldithio)propionate (SPDP) (Pierce) as follows: hIgG1 myeloma protein (5 mg at 10 mg/ml) in phosphate buffer (0.01 M sodium phosphate, pH 7.5, 0.15 M NaCl) was treated with a 5-fold molar excess of SPDP in dioxine for 30 min. Excess reagents were removed by dialysis into phosphate-buffered saline, pH 7.0, 2 mM EDTA. The S-acetylmercaptosuccinic anhydride-modified hIgG1 was treated with 200 l of hydroxylamine (1 mM in phosphate-buffered saline, pH 7.0) for 30 min and then mixed with SPDP-modified hIgG1 (1:1 molar ratio) and incubated for a further hour. The reaction was terminated with N-ethylmaleimide (Sigma) added to a final concentration of 6.6 mM (35). All reactions were performed at room temperature. Dimeric hIgG1 was purified from monomeric and oligomeric hIgG1 by size fractionation chromatography on Sephacryl S-300 HR (Pharmacia Biotech Inc.).
Immune Complex Binding-The binding of immune complexes by COS-7 cells following transfection with chimeric or mutant receptor cDNAs was determined using two approaches: erythrocyte-antibody rosetting or direct binding of dimeric hIgG1. For erythrocyte-antibody rosetting, COS-7 cell monolayers transfected with FcR expression constructs were incubated with antibody-sensitized erythrocytes (EA complexes), prepared by coating sheep red blood cells with trinitrobenzene sulfonate (Fluka Chemika, Switzerland) and then sensitizing these cells with mouse IgG1 or IgE anti-trinitrobenzene sulfonate mAb (37). Two ml of 2% EAs (v/v) were added per 5-cm 2 dish of transfected cells and incubated for 5 min at 37°C. Plates were then centrifuged at 500 ϫ g for 3 min and placed on ice for 30 min. Unbound EAs were removed by washing with L-15 medium modified with glutamine (Flow Laboratories) and containing 0.5% bovine serum albumin. For direct binding of dimeric hIgG1, COS-7 cells transfected with FcR expression constructs were harvested, washed in phosphate-buffered saline, 0.5% bovine serum albumin, and resuspended at 10 7 cells/ml in L-15 medium, 0.5% bovine serum albumin. Cells in 50-l aliquots were incubated with 50-l serial dilutions of 125 I-dimeric hIgG1 for 120 min at 4°C. 125 I-Dimeric Ig was prepared by the chloramine-T method as described (38) and shown to compete equally with unlabeled dimeric Ig in binding to Fc receptor expressing COS-7 cells. Cell bound 125 I-dimeric IgG1 was determined following centrifugation of cells through a 3:2 (v/v) mixture of dibutyl phthalate and dioctyl phthalate oils (Fluka Chemika), and cell bound 125 I-dimer was determined. Nonspecific dimer binding was determined by assaying on mock transfected cells and subtracted from total binding to give specific dimeric IgG1 bound. Levels of cell surface Fc␥RII expression were determined using the anti-Fc␥RII mAb 8.2, shown to bind distantly to the binding site (32) and used to correct for variable cell surface receptor expression between the mutant Fc␥RII COS-7 cell transfectants. The binding of mAb 8.2 was determined in a direct binding assay as described for the human IgG1 dimer binding assays.
Generation of Fc␥RII Domain 1-Domain 2 Model Structure-Molecular modeling of the extracellular region of hFc␥RIIa (domains 1 and 2) was performed using the Homology and Discover modules of the InsightII software package of Biosym Technologies, using the crystal structure of domains 1 and 2 of CD4 (Brookhaven protein data base file pdb2cd4.ent) essentially as described previously for domain 2 (20). Sequence alignments were used to determine the location of ␤-sheets, with other regions defined as loops. Since the N-terminal A-strand of Fc␥RII-D1 is longer than that of CD4-1, the Bence-Jones protein REI (pdb2rei.ent) V domain was chosen as a template for the first 7 residues after superimposition of REI on CD4-1. A search of the Brookhaven protein crystallographic data base was then carried out using the Loop Search command to find suitable loop templates for the remaining pieces (see below). In some cases, this required a re-evaluation of the structurally conserved residues of ␤-sheets. In two cases, the A/B loop of domain 1 and the E/F loop of domain 2, the coordinates were assigned directly from the equivalent loops in the CD4 template and hence are called "designated loops." After construction of the two disulfide bonds and elimination of severe atomic overlaps ("bumps"), the structure was minimized using the Discover module to a maximum r.m.s. derivative of 0.0001 using 2000 steepest descents and up to 25,000 conjugate gradients, with the backbone atoms of structurally conserved regions fixed. The operation was repeated with no fixed atoms. The final structure was checked for poor , , and angles and residues of high energy. The loops used in the model are detailed as follows. Domain 1: A/B, (EDS) modeled to GDT from 2cd4 (designated loop); B/C loop (SPESD) modeled to PGTSN of 2mev, starting at residue I169 (including the previous structurally conserved region), deviation 1. 25

Chimeric Receptors Identify Multiple Regions of Fc␥RII Involved in IgG
Binding-In order to determine the roles of domain 1 (residues 1-86) and the B/C (residues 109 -116) or CЈ/E (residues 130 -135) loop regions of domain 2 in the binding of IgG by Fc␥RII, chimeric receptors were generated whereby each of these regions in Fc␥RII were replaced with the equivalent regions of the Fc⑀RI ␣ chain. Chimeric receptor cDNAs were constructed by SOE PCR, subcloned into the eukaryotic expression vector pKC3, and transiently transfected into COS-7 cells. The binding of IgG immune complexes to the chimeric receptors was determined by both EA rosetting and the binding of dimeric hIgG1. The distinction between the two assays lies in the nature of the immune complexes; EAs comprise large multivalent immune complexes capable of binding with high avidity to Fc␥RII and were used to qualitatively assess Ig binding of the chimeric receptors, whereas dimeric Igs represent the smallest complexes able to bind Fc␥RII with readily detectable affinity and were used in the quantitation of Ig binding.
The substitution of the Fc␥RII domain 1 with that of the Fc⑀RI ␣ chain produced a receptor (designated D1⑀D2␥), which as expected retained the capacity to bind the multivalent IgG-EA complexes, as did the wild-type Fc␥RII (Fig. 1a). However, in contrast to the wild-type receptor the D1⑀D2␥ chimeric FcR did not bind dimeric-hIgG1 at any concentration (Fig. 2). This suggests that domain 1 is necessary for optimal Ig binding as demonstrated by the binding of highly substituted but not small dimeric complexes.
The previous analysis of genetic polymorphisms of Fc␥RII (21)(22)(23)(24)(25) in conjunction with our molecular modeling studies described above (20), suggest that the region around residue 114 (human equivalent of polymorphic residue 116 in mouse After transfection into COS-7 cells, this receptor was clearly able to bind Ig in the form of multivalent immune complexes, i.e. erythrocytes highly sensitized with IgG (IgG-EA) (Fig. 1b). By contrast, this receptor was unable to bind dimeric hIgG1 at any concentration, implying that the B/C loop is essential for optimal Ig binding (Fig. 2). Similarly, the region surrounding residue 131 responsible for the responder/nonresponder phenotype of Fc␥RIIa, i.e. the CЈ/E loop (Ser 130 , Arg/His 131 , Leu 132 , Asp 133 , Pro 134 , Thr 135 ) was replaced with the equivalent Fc⑀RI ␣ chain sequence (Trp 130 , Tyr 131 , Glu 132 , Asn 133 , His 134 , Asn 135 ), generating a chimeric receptor (␥130 -135⑀) that upon transfection into COS-7 cells was able to bind IgG-EA (Fig. 1c) but not dimeric IgG1 (Fig. 2). As expected COS-7 cells transfected with an expressible form of the Fc⑀RI ␣ chain (18) did not bind hIgG1 dimers or IgG-EA (Figs. 1d and 2). Thus the ability of the chimeric Fc␥RII containing B/C or CЈ/E domain 2 substitutions to bind the highly sensitized EA complexes but not dimeric hIgG1 suggests that these receptors bind IgG less avidly than wild-type Fc␥RII and clearly indicates that the B/C and CЈ/E regions also make a contribution to the binding of IgG by Fc␥RII.
Fine Structure Analysis of the B/C and CЈ/E loops of Fc␥RII Domain 2-The contribution of individual amino acids of the B/C and CЈ/E loop regions of Fc␥RII to the binding of IgG was determined using a point mutagenesis strategy whereby residues in both the B/C (residues 113-117) and CЈ/E (residues 129 -134) loops were replaced with alanine. cDNAs encoding the mutant receptors were generated using SOE PCR and subcloned into the eukaryotic expression vector pKC3. The resultant expression constructs were transiently transfected into COS-7 cells, and the Ig binding capacity of the mutant receptors was determined by assessing the binding of dimeric hIgG1. The levels of cell membrane expression of the mutant Fc␥RII on the COS-7 cell transfectants were determined using the anti-Fc␥RII mAb 8.2 (shown to detect an epitope distant from the binding site) and were comparable with the expression levels of the wild-type receptor (see Fig. 3 legend). The relative capacities of the mutant receptors to bind hIgG1 were determined using the direct binding assay following correction for variation in cell surface expression levels and expressed as percentage of wild-type Fc␥RII binding.
The replacement of the B/C loop residues (Lys 113 , Pro 114 , Leu 115 , Val 116 ) in turn with Ala in each case resulted in diminished hIgG1 dimer binding (Fig. 3). The most dramatic effect was seen on substitution of Lys 113 or Leu 115 , which exhibited only 15.9 Ϯ 3.4% (mean Ϯ S.D.) and 20.6 Ϯ 4.0% binding compared with wild-type Fc␥RII. The replacement of Pro 114 or Val 116 with Ala had a lesser effect, these receptors displaying 53.5 Ϯ 13.5% and 73.5 Ϯ 7.9% wild-type binding respectively. It is interesting to note that the individual replacement of these amino acids did not result in the complete abolition of dimer binding seen in chimera ␥109 -116⑀. These results suggest that each of these residues in the B/C loop contribute to the binding of IgG by Fc␥RII either as direct contact residues or indirectly by maintaining the correct conformation of the binding site. The same approach was used to analyze the role of individual amino acids within the CЈ/E loop (Phe 129 , Ser 130 , Arg/His 131 , Leu 132 , Asp 133 , Pro 134 ). In contrast to that observed for resi-dues of the B/C loop, mutation of individual residues of the CЈ/E loop resulted in both loss and enhancement of IgG binding. Substitution of Phe 129 and Arg/His 131 dramatically decreased hIgG1 dimer binding by approximately 90 and 80%, respectively, to 8.2 Ϯ 4.4 and 21.9 Ϯ 3.9 compared with that seen for wild-type Fc␥RII (Fig. 3). Interestingly, replacement of residues Asp 133 and Pro 134 increased binding to 113.5 Ϯ 8.8% and 133.5 Ϯ 3.2% of the wild-type receptor. The substitution of Ser 130 or Leu 132 had no significant effect on the binding of hIgG1 dimers, since these mutants exhibited binding comparable with that seen for wild-type Fc␥RII (Fig. 3). These findings suggest that Phe 129 and Arg/His 131 may play an important role in the binding of hIgG1, and the observation that the substitution of Asp 133 and Pro 134 increase binding also suggests an important role for these residues, which appears distinct from Phe 129 and Arg/His 131 . Again, a distinction between a possible direct binding role or contribution to structural integrity of the receptor cannot be made; however, these findings clearly identify both the B/C and CЈ/E loops as playing a role in the binding of IgG by Fc␥RII.
Site-directed mutagenesis was also performed on 3 residues of the CЈ/C loop, a region predicted to be distant from the putative binding region, i.e. the B/C, CЈ/E, and F/G loop regions. The substitution of residues Asn 123 , Gly 124 , and Lys 125 had no effect on the binding of hIgG1 dimer, since each of these mutants exhibited similar binding to the wild-type receptor (data not shown).

FIG. 2. Human IgG1 dimer binding of chimeric Fc receptors.
Radiolabeled dimeric human IgG1 was titrated on COS-7 cells transfected with wild-type Fc␥RII (f), an expressible form of the Fc⑀RI ␣ chain (Ⅺ), or the following chimeric receptor cDNAs: D1⑀D2␥ (q), ␥109 -116⑀ (E), ␥130 -135⑀ ( ). All of the chimeras were expressed on the cell surface as determined by EA rosetting, outlined in the Fig. 1 legend. ular modeling was used to generate a homology model of domains 1 and 2 of Fc␥RIIa using the crystal structure of CD4 domains 1 and 2 as a template (Fig. 4) as described under "Materials and Methods." The two domains of Fc␥RII are structurally related, both belonging to the truncated C2 set of the Ig superfamily, comprising 7␤ strands (A, B, C, CЈ, E, F, G) forming two antiparallel ␤-sheets of strands ABECЈ and CFG, respectively. The modeling of the extracellular region of Fc␥RII suggests that the regions implicated in the binding of IgG, i.e. the B/C, CЈ/E, and F/G loops of domain 2, are juxtaposed at the interface with domain 1. Based on this model together with the mutagenesis data, the topology of the binding region can be best described as a hydrophobic patch surrounded on three sides by a "wall" of predominantly aromatic and basic residues. The hydrophobic patch consists of Pro 114 , Leu 115 , Val 116 , Ile 155 , and Gly 156 contributed by the B/C and F/G loops. All loops contribute to the wall including Lys 113 and other residues in the B/C loop, Phe 129 and Arg 131 in the CЈ strand and CЈ/E loop, and Leu 159 and Phe 160 in the F/G loop (Fig. 4). DISCUSSION The studies described herein provide evidence to suggest that the interaction of IgG with human Fc␥RII involves multiple regions juxtaposed in the receptor. Previously, we have described the localization of a single region of Fc␥RII capable of directly binding IgG situated in the second extracellular domain between residues Asn 154 and Ser 161 (20). Of the entire extracellular region, only the 154 -161 segment was demonstrated to directly bind IgG, since placement of only this region in the corresponding region of the human Fc⑀RI ␣ chain, imparted IgG binding function to the IgE receptor Fc⑀RI. Moreover, replacement of this region in Fc␥RII with that of Fc⑀RI␣ resulted in the total loss of IgG binding including large complexes, implying that residues Asn 154 -Ser 161 comprise the key IgG1 interactive site of Fc␥RII. However, the generation of further chimeric Fc␥RII/Fc⑀RI␣ receptors as described herein indicates that two additional regions of Fc␥RII domain 2 also influence the binding of IgG by Fc␥RII. The replacement of the regions encompassing Ser 109 -Val 116 (B/C loop) and Ser 130 -Thr 135 (CЈ/E loop) of Fc␥RII with the equivalent regions of the Fc⑀RI ␣ chain, produced receptors that, despite containing the putative binding site (Asn 154 -Ser 161 ) and retaining the ability to bind large complexes (IgG-EA), lost the capacity to bind small complexes (dimeric hIgG1). Indeed, site-directed mutagenesis performed on residues of the B/C and CЈ/E regions identified a number of amino acids that appear to play crucial roles in hIgG1 binding by Fc␥RII. The replacement of Lys 113 , Pro 114 , Leu 115 , and Val 116 of the B/C loop and Phe 129 and Arg/His 131 of the CЈ/E loop with alanine all resulted in diminished hIgG1 binding. Furthermore, the substitution of Asp 133 and Pro 134 of the CЈ/E loop increased hIgG1 binding. Therefore, these findings provide strong evidence to suggest that the B/C and CЈ/E loops of Fc␥RII, in addition to the F/G loop, also contribute to the binding of IgG.
A number of other studies have provided evidence to support the proposed IgG binding roles of the B/C and CЈ/E loop regions of Fc␥RII. Studies of genetic polymorphisms of mouse and human Fc␥RII have implicated residues 114, 131, and 159 in the binding of IgG by human Fc␥RII. These residues are located in the B/C (residue 114), CЈ/E (131), and F/G (159) loops, respectively. The Ly-17 polymorphism of mouse Fc␥RII has been described at the molecular level as two allelic variants (Ly17.1 and Ly17.2) that differ only at residues 116 and 161 (the equivalent of residues 114 and 159 in the human). Monoclonal antibodies specific for Ly17.2 inhibit the binding of IgG to the receptor, implying that residues 116 and/or 161 (and therefore their human equivalents) are involved in binding themselves or closely situated to residues crucial in the interaction of Fc␥RII with IgG (24,25). Furthermore, the high responder/low responder polymorphism of hFc␥RIIa results in an amino acid substitution at residue 131, which has been shown to influence the binding of mIgG1 and hIgG2 (21-23). The findings described herein also indicate that the nature of the residue at 131 plays a role in the binding of hIgG1, since replacement with alanine results in almost complete loss in bind-  ing of this isotype to Fc␥RII. Thus, although the F/G loop of Fc␥RII is clearly a major region involved in the direct interaction with IgG, as demonstrated by the fact that only this region has been definitively shown to directly bind IgG (20), residue 131 also appears to play a binding role. However, the question of whether residue 131 is directly participating in IgG binding or providing a secondary or indirect influence remains to be answered.
The mutagenesis data clearly implicate a number of distinct regions within Fc␥RII in the interaction with IgG complexes as described above. The spatial relationship of these regions, i.e. residues 109 -116 (B/C loop), 129 -135 (CЈ/E loop), and 154 -161 (F/G loop) is postulated in our model of Fc␥RII (Fig. 4). This model suggests that these regions are juxtaposed to each other in domain 2 at the interface with domain 1 and form a hydrophobic pocket surrounded by a wall of additional residues. The data supporting this model include the following. 1) Mutagenesis of the hydrophobic residues Ile 155 , Gly 156 , Pro 114 , Leu 115 almost completely abolishes binding of dimeric hIgG1 complexes. 2) Substitution of residues that may contribute the wall (Lys 113 in the B/C loop, Phe 129 and Arg 131 in the CЈ/E loop, and Leu 159 and Phe 160 in the F/G loop) also modify binding of immune complexes.
3) It may also be expected that such a wall would be accessible to anti-FcR antibodies. Indeed several anti-Fc␥RII monoclonal antibodies detect epitopes in the B/C, CЈ/E, and F/G loops. For example, the epitope detected by the antihuman Fc␥RII antibody 41H16 (39) is dependent on residue 131 of the CЈ/E loop, and the Ly-17 epitope of mouse Fc␥RII is dependent on residues that equate to residues 114 and 159 in human Fc␥RII (25) that are located in the B/C and F/G loops, respectively. 4) The studies described herein demonstrate that domain 1 of Fc␥RII, although it does not appear to play a direct role in IgG binding, does play an important role in the affinity of IgG binding by Fc␥RII. This is suggested since replacement of domain 1 of Fc␥RII with domain 1 of Fc⑀RI reduced the capacity of Fc␥RII to bind IgG, as shown by the failure of this receptor to bind dimeric hIgG1. These data imply that the IgG binding role of domain 1 is likely to be an influence on receptor conformation, stabilizing the structure of domain 2 to enable efficient IgG binding by Fc␥RII. Again this proposal is consistent with the molecular modeling, which suggests the localization of the IgG binding site of Fc␥RII to loop regions in domain 2 at the interface with domain 1. The binding site would therefore be in close proximity to domain 1 and as such predicted to be influenced in conformation, presumably by the loop and strand regions at the "bottom" of domain 1. These regions include the G strand and the A/B and E/F loops, which may Residues implicated in IgG1 binding are indicated as described above. The computer model of Fc␥RII domain 1domain 2 was generated by molecular modeling based on the structure of the related CD4 domains 1 and 2 as described under "Materials and Methods." therefore interact with the "active" binding region of domain 2.
Further support for the involvement of the B/C and CЈ/E loops of Fc␥RII domain 2 in the binding of IgG has been provided in the cloning and subsequent Ig binding studies of rat Fc␥RIII (40), which is structurally and functionally homologous to Fc␥RII. Two rat Fc␥RIII isoforms, IIIA and IIIH, which have extensive amino acid differences in their second extracellular domains, have been shown to bind rat and mouse IgG subclasses differently. Both isoforms bind rtIgG1, rtIgG2a, and mIgG1; however, they differ in that only the IIIH form binds rtIgG2b and mIgG2b. Significantly, the amino acid differences between rat Fc␥RIIIA and IIIH isoforms are situated predominantly in the predicted B/C and CЈ/E loops of domain 2. However, it should be noted that the F/G loop regions of rat Fc␥RIIIA and IIIH are almost totally conserved, which together with the observation that both forms bind rtIgG1, rtIgG2a, and mIgG1, is consistent with the proposal that the F/G loop region is the major IgG interactive region and that the B/C and CЈ/E loop regions provide supporting binding roles. In addition, a recent mutagenesis study of human Fc␥RIII has also implicated residues in the B/C and CЈ/E loops of this receptor in the binding of IgG (41). It is also interesting to note that in this study the C/CЈ region of Fc␥RIII was suggested to play a major role in IgG binding, which is in marked contrast to our findings with Fc␥RII. Indeed, the substitution of 3 residues in the C/CЈ loop of Fc␥RII with alanine, namely Asn 123 , Gly 124 , and Lys 125 , did not have any effect on the binding of dimeric hIgG1. Therefore, these findings somewhat surprisingly suggest that Fc␥RII and Fc␥RIII, which exhibit substantial amino acid sequence conservation and similar IgG binding affinities and specificities, may interact differently with IgG.
It is interesting to note that a number of parallels are apparent in the molecular basis of the interaction of Fc␥RII with IgG and that of Fc⑀RI with IgE. The Ig binding roles of the two extracellular domains of Fc⑀RI are similar to Fc␥RII, with domain 2 responsible for the direct binding of IgE and domain 1 playing a supporting structural role (18,26,42). Furthermore, as described for Fc␥RII, we and others have also identified multiple IgE binding regions in domain 2 of Fc⑀RI. Using chimeric Fc␥RII/Fc⑀RI receptors we have demonstrated that domain 2 of Fc⑀RI contains at least three regions, each capable of directly binding IgE, since the introduction of the Fc⑀RI regions encompassed by residues Trp 87 -Lys 128 , Tyr 129 -Asp 145 , and Lys 154 -Glu 161 into the corresponding regions of Fc␥RII was found to impart IgE binding to Fc␥RII (1,18,20). A similar study using chimeric Fc␥RIII/Fc⑀RI receptors has implicated 4 regions of Fc⑀RI domain 2 in IgE binding since the regions Ser 93 -Phe 104 , Arg 111 -Glu 125 , Asp 123 -Ser 137 , and Lys 154 -Ile 167 of Fc⑀RI when replaced with the corresponding regions of Fc␥RIII resulted in the loss or reduction of IgE binding (42). Taken together, these data suggest that at least four regions of Fc⑀RI domain 2 contribute to the binding of IgE, Ser 93 -Phe 104 , Arg 111 -Glu 125 , Tyr 129 -Ser 137 , and Lys 154 -Glu 161 . Three of these regions correspond to the three regions identified herein as important in the binding of IgG by Fc␥RII, Arg 111 -Glu 125 , Tyr 129 -Ser 137 , and Lys 154 -Glu 161 , which encompass the B/C, CЈ/E, and F/G loops, respectively. In addition, studies with anti-Fc⑀RI ␣ chain mAb have indicated that the region encompassed by residues 100 -115 contains an epitope detected by mAb 15A5, which can completely block the binding of IgE to Fc⑀RI (43). Thus, these findings implicate the B/C, CЈ/E, and F/G loops juxtaposed in domain 2 at the domain 1 interface as the crucial IgE-interactive region of Fc⑀RI. Clearly, the findings described herein for Fc␥RII together with those discussed for Fc⑀RI provide evidence to suggest that the Ig-interactive regions of Fc␥RII and Fc⑀RI are conserved between the two receptors, with the do-main 1-domain 2 interface forming the Ig binding site.
In conclusion, the results presented herein demonstrate that multiple regions of hFc␥RII are involved in the binding of IgG, with three putative loop regions juxtaposed in the second extracellular domain at the domain 1 interface comprising the IgG binding site. The proposition that the functionally distinct receptor Fc⑀RI also interacts with IgE in a structurally similar fashion, in conjunction with the conserved nature of the extracellular regions of the Ig superfamily FcR, strongly suggests that this region will also comprise the key Ig-interactive site of all members of this family.