The human low affinity Fcgamma receptors IIa, IIb, and III bind IgG with fast kinetics and distinct thermodynamic properties.

Fcgamma receptors (FcgammaRs) are expressed on all immunologically active cells. They bind the Fc portion of IgG, thereby triggering a range of immunological functions. We have used surface plasmon resonance to analyze the kinetic and thermodynamic properties of the interactions between the ectodomains of human low affinity FcgammaRs (FcgammaRIIa, FcgammaRIIb, and FcgammaRIIIb-NA2) and IgG1 or the Fc fragment of IgG1. All three receptors bind Fc or IgG with similarly low affinities (K(D) approximately 0.6-2.5 microm) and fast kinetics, suggesting that FcgammaR-mediated recognition of aggregated IgG and IgG-coated particles or cells is mechanistically similar to cell-cell recognition. Interestingly, the Fc receptors exhibit distinct thermodynamic properties. Whereas the binding of the FcgammaRIIa and FcgammaRIIb to Fc is driven by favorable entropic and enthalpic changes, the binding of FcgammaRIII is characterized by highly unfavorable entropic changes. Although the structural bases for these differences remain to be determined, they suggest that the molecular events coupled to the binding differ among the low affinity FcgammaRs.


Fc␥ receptors (Fc␥Rs) are expressed on all immunologically active cells. They bind the Fc portion of IgG, thereby triggering a range of immunological functions.
We have used surface plasmon resonance to analyze the kinetic and thermodynamic properties of the interactions between the ectodomains of human low affinity Fc␥Rs (Fc␥RIIa, Fc␥RIIb, and Fc␥RIIIb-NA2) and IgG1 or the Fc fragment of IgG1. All three receptors bind Fc or IgG with similarly low affinities (K D ϳ0.6 -2.5 M) and fast kinetics, suggesting that Fc␥R-mediated recognition of aggregated IgG and IgG-coated particles or cells is mechanistically similar to cell-cell recognition. Interestingly, the Fc receptors exhibit distinct thermodynamic properties. Whereas the binding of the Fc␥RIIa and Fc␥RIIb to Fc is driven by favorable entropic and enthalpic changes, the binding of Fc␥RIII is characterized by highly unfavorable entropic changes. Although the structural bases for these differences remain to be determined, they suggest that the molecular events coupled to the binding differ among the low affinity Fc␥Rs.
Fc␥ receptors (Fc␥Rs) 1 are membrane glycoproteins that are expressed on all immunologically active cells. They bind the Fc portion of the main serum immunoglobulin IgG, and this binding triggers a wide range of events. These include the activation of B cells, endocytosis of antibody/antigen (immune) complexes, phagocytosis of antibody-coated particles or cells, and antibody-dependent cellular cytotoxity (1). Upon aggregation of Fc␥Rs (e.g. by ligation of aggregated IgG or IgG-coated particles or cells) their cytoplasmic domains are phosphorylated by tyrosine kinases, which recruit SH2-domain-containing cytoplasmic signaling molecules. Fc␥Rs can also appear as soluble molecules (sFc␥Rs) in serum and other body fluids. Although sFc␥Rs may be involved in immunological disorders, their role in normal immune responses is questionable. Nevertheless, these soluble receptor forms may be useful as a prognostic and staging tool in diseases such as systemic lupus erythematosus (2), rheumatoid arthritis (3), multiple myeloma (4), and human immunodeficiency virus infection (5). It is notable that several viruses, including human immunodeficiency virus (6), dengue (7), measles (8) and Ebola (9), employ Fc␥R to the disadvantage of the host, e.g. by antibody-mediated enhancement of infection (10), underlining the central role of these receptors in the immune response.
Fc␥Rs belong to the immunoglobulin (Ig) superfamily and have been subdivided into three classes. Fc␥RI (CD64) has a high affinity for IgG (K d Ϸ10 Ϫ8 M) and has three Ig-like domains in the extracellular region. In contrast, Fc␥RII (CD32) and Fc␥RIII (CD16) have lower affinities (K d Ϸ10 Ϫ6 M) and only two extracellular Ig-like domains. There are three Fc␥RII genes (A, B, and C) and two Fc␥RIII genes (A and B), the products of which show distinct patterns of expression. The ectodomains of Fc␥RIIb and Fc␥RIIc are identical but distinct from Fc␥RIIa. Fc␥RIIIa and Fc␥RIIIb have a transmembrane and a glycosylphosphatidylinositol anchor, respectively, as well as 6 amino acid differences in their ectodomains. Fc␥RIIIb exists as the two alleles NA1 and NA2, which vary in their extracellular domains at three amino acid positions, resulting in two more potential N-glycosylation sites for the NA2 form. Although all low affinity IgG receptors bind to aggregated IgG or IgG-coated cells or particles, their different membrane anchors and cytoplasmic portions confer distinct functional properties.
Low affinity IgG receptors that transduce activating signals typically either have immunoreceptor tyrosine-based activation motifs in their cytoplasmic domains (Fc␥RIIa or Fc␥RIIc) or, as with Fc␥RIIIa, are associated with molecules possessing immunoreceptor tyrosine-based activation motifs. In contrast, the receptor that transduces inhibitory signals (Fc␥RIIb) has an immunoreceptor tyrosine-based inhibitory motif in its cytoplasmic domain, much like the killer cell Ig-like receptors (KIRs) identified on natural killer (NK) cells. Finally, Fc␥Rs bind with different affinities to the var-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  ious IgG subclasses, a property that may allow a fine tuning of the immune response (11).
Crystal structures have been solved for the low affinity IgG receptors Fc␥RIIa (12), Fc␥RIIb (13), and Fc␥RIII (14), as well as the high affinity IgE receptor Fc⑀RI␣ (15). As might be expected from their sequence similarity, these structures are very similar. One notable feature is that the Ig domains in these receptors are structurally related to Ig domains in KIRs (16 -18) (Fig. 1A), which also have a unique topology intermediate between the I-set and C2-set Ig domains (Fig.  1B), as also pointed out by Dennis et al. (19). Although sequence similarity between the Fc receptors and KIRs is not high, this structural similarity suggests that they share a common ancestral receptor. A second similarity is the acute angle between the long axes of the membrane-distal and membrane-proximal domains (Fig. 1A). Nevertheless there are notable differences. First, the arrangement between the two domains differs substantially (Fig. 1A). Second, the recent solution of the structure of the crystallized complexes KIR2DL2/HLA-Cw3 (20) and Fc␥RIIIb/Fc (14) reveals distinct ligand binding sites. While the KIR molecule uses loop residues from both of its Ig-like domains to bind the major histocompatibility complex (MHC) class I molecule, the Fc␥R only uses loop and strand residues from domain 2 and the interdomain linker for interacting with IgG.
Despite recent advances in understanding the structural basis of Fc␥R/Fc binding, the basic thermodynamic and kinetic properties of these interactions are not well characterized. These data could help address several important questions in relation to Fc␥R function including the requirements for, and mechanism of, triggering through low affinity Fc␥Rs, the mechanism of detachment of Fc␥R-bearing cells from immune com-plexes or IgG-coated cells or particles, and the requirement for significant molecular rearrangements upon binding. Here we report the kinetic and thermodynamic characteristics of IgG and Fc binding to soluble forms of all human low affinity Fc␥Rs (Fc␥RIIa, Fc␥RIIb, and Fc␥RIII).

Production and Purification of Recombinant Fc␥ Receptors-
The extracellular regions of the Fc␥R receptors, Fc␥RIIa, Fc␥RIIb, Fc␥RIIb-His 6 (Fc␥RIIb with a C-terminal hexahistidine tag), and Fc␥RIIIb-NA2, all having two Ig-like domains, were expressed in Escherichia coli as inclusion bodies, refolded, and purified by hIgG affinity chromatography and gel filtration, as described (21), resulting in sFc␥Rs. The purified sFc␥Rs are monomeric and specifically bind to human IgG (22).
Surface Plasmon Resonance (SPR)-Human IgG1 (hIgG1, kindly provided by R. E. Schmidt and T. Witte) was obtained by plasmapheresis from a myeloma patient, and its Fc fragment (hFc1) was produced by plasmin digestion. Both proteins were purified by standard methods, including protein A and size-exclusion chromatography. SPR experiments were performed using a BIAcore 2000 (BIAcore AB, St. Albans, United Kingdom). All experiments were performed with HBS-EP (10 mM Hepes, pH 7.4, 150 mM NaCl, 3.4 mM EDTA, and 0.005% surfactant P20) as running buffer using CM5 sensor chips and HBS-P (no EDTA) when using nickel-charged nitriloacetic acid (Ni 2ϩ -NTA) sensor chips (BIAcore AB). Direct coupling of human IgG1 (hIgG1), human Fc1 (hFc1), or mouse IgG2a (mIgG2a) on CM5 sensor chips (BIAcore AB) was performed using the standard amine coupling kit (BIAcore AB). For coupling hIgG1, hFc1 and mIgG2a were injected at 5 g⅐ml Ϫ1 in 10 mM sodium acetate, pH 5.5. Fc␥RIIb-His 6 was immobilized onto, and eluted from, a Ni 2ϩ -NTA sensor chip (BIAcore AB) using the protocol recommended by BIAcore AB.
All affinity, kinetic, and thermodynamic parameters were measured as described previously (23). Briefly, affinity constants were derived by Scatchard analysis and nonlinear curve fitting of the Langmuir binding isotherm to the data. Kinetic constants were derived using the curvefitting facility of the BIAevaluation program (version 3.0, BIAcore). Equations derived from the simple 1:1 Langmuir binding model (A ϩ B 7 AB) were used. Other curve fitting was performed in Origin (Micro-Cal, version 3). Thermodynamic data were obtained by fitting the affinity data obtained at several temperatures to the nonlinear form of the van't Hoff equation (24).
T is the temperature in Kelvin (K); T 0 is an arbitrary reference temperature (e.g. 298.15 K); ⌬G°is the standard free energy of binding at T (kcal⅐mol Ϫ1 ); ⌬H T 0 is the enthalpy change upon binding at T 0 (kcal⅐mol Ϫ1 ); ⌬S°T 0 is the standard state entropy change upon binding at T 0 (kcal⅐mol Ϫ1 ); ⌬C p is the specific heat capacity (kcal⅐mol Ϫ1 ⅐K Ϫ1 ); ⌬G°was calculated from the K d ; ⌬G°ϭ R ϫ T ϫ ln(K d /C), where R is 1.987 ϫ 10 Ϫ3 kcal⅐mol Ϫ1 K Ϫ1 ; K d is expressed as mol⅐L Ϫ1 ; and C is 1 mol⅐L Ϫ1 (therefore making K d /C dimensionless, as required).

Affinity and Specificity of sFc␥R Binding to IgG and Fc
Fragment-Soluble forms of Fc␥Rs (sFc␥RIIa, sFc␥RIIb, sFc␥RIIb-His 6 , and sFc␥RIII) were expressed in E. coli and refolded in vitro (see "Experimental Procedures"). Affinity chromatography with the native ligand hIgG1 bound to the matrix indicated that the proteins were biologically active. sFc␥RIIa (25), sFc␥RIIb (13), and sFc␥RIII (14) were used successfully for x-ray crystallographic structure determinations, indicating that they are uniformly and correctly folded. All sFc␥Rs migrated as monomers on nonreducing SDS-polyacrylamide gels and size-exclusion columns (data not shown).
Binding of Fc␥R constructs was analyzed on a BIAcore instrument by surface plasmon resonance, which measures the changes in refractive index near a sensor surface (26). Each sFc␥R was injected ( Fig. 2A, solid bar) simultaneously through four flow cells containing sensor surfaces to which human IgG1 (hIgG1), its Fc fragment (hFc1), mouse IgG2a (mIgG2a), or bovine serum albumin (BSA) had been directly immobilized via primary amines. A "background" response (measured in re- sponse units) is seen in the negative control BSA flow cell, which is a consequence of the high concentration, and therefore high refractive index, of injected sFc␥R samples. However, a greater response was seen with injection over hIgG1, hFc1, and mIgG2a, indicating specific binding (Fig. 2).
The affinity of sFc␥Rs binding to IgG and Fc was measured by equilibrium binding analysis on the BIAcore. A range of concentrations of sFc␥Rs were injected through flow cells with immobilized human IgG1, its Fc fragment, or BSA as a control. The binding response at each concentration was calculated by subtracting the equilibrium response measured in the control flow cell from the response in each flow cell. Conventional (Fig.  3) and Scatchard (Fig. 3, inset) plots of these binding data indicate that the interaction conforms to a simple 1:1 Langmuir binding model (see "Discussion"). The results of several experiments are summarized in Table I. All sFc␥Rs showed similar affinity constants (K d ϭ 0.6 -1.7 M) measured at 25°C (298.15 K). A similar affinity was measured in the reverse orientation, using hFc1 in solution and sFc␥RIIb-His 6 immobilized to a Ni 2ϩ -NTA sensor chip via its C-terminal hexahistidine tag (Table I). The affinity of sFc␥RIIa for hIgG1, hFc1 fragment, and mIgG2a is slightly higher than that of sFc␥RIIb and sFc␥RIII. For all sFc␥Rs, the affinity toward hIgG1 was quite similar to that of its Fc fragment, indicating that Fab fragment does not contribute to sFc␥R binding. Although sFc␥RIIa and IIb bind with similar affinity to mIgG2a, sFc␥RIII bound with a much lower affinity in this experiment, as reported previously by Chesla et al. (27), indicating that the ligand specificity of sFc␥RIII is different from that of sFc␥RIIa and IIb.
Binding Kinetics-All sFc␥Rs bound their ligands with typical association and fast dissociation kinetics similar to those of other receptor/ligand interactions (Table IV). Association and dissociation rate constants were obtained by direct nonlinear curve fitting to data obtained by SPR, as illustrated for sFc␥RIIb and sFc␥RIII binding to the Fc fragment of hIgG1 (Fig. 4). Global fitting of mono-exponential rate equations derived from the simple 1:1 Langmuir binding model produced reasonable fits to the data (Fig. 4), yielding the rate constants shown in Table II. The rate constants did not change significantly when the level of immobilized IgG and Fc varied 2-fold (Table II), indicating that binding was not substantially affected by mass transport or rebinding artifacts. The excellent agreement between the kinetically derived K d (Table II) and the K d determined by equilibrium binding (Table I) is further evidence that these kinetic rate constants are correct. Furthermore, similar rate constants were measured in the reverse orientation, with sFc␥RIIb-His immobilized and Fc fragment in  solution (Table II). The confirmation of both affinity (Table I) and kinetic measurements in both orientations indicates that binding is not substantially influenced by the coupling method and that the affinity and kinetic measurements are unlikely to be influenced by the presence of multivalent aggregates or errors in determination of active protein concentrations. Because recent studies have shown that the kinetics of TCR (T cell receptor) binding to peptide-MHC are strongly temperature-dependent (28), we measured the temperature dependence of the sFc␥R/hFc1 interactions. Arrhenius plots yielded an activation energy for dissociation of 7-16 kcal⅐mol Ϫ1 for the measured sFc␥R/hFc1 interaction (data not shown). These values are comparable with that obtained for sKIR2DL3/HLA-Cw7-DS11 dissociation (13 kcal⅐mol Ϫ1 ) (23) and far lower than the values (ϳ30 kcal⅐mol Ϫ1 ) measured for TCR/peptide-MHC dissociation (28) (see "Discussion").
Thermodynamic Analysis-The enthalpy change (⌬H) that accompanies Fc␥R binding to IgG or Fc was estimated by van't Hoff analysis, which involves measuring the dependence of affinity on temperature (Fig. 5). Because the enthalpic (⌬H vH ) and entropic (T⌬S) changes vary with temperature the nonlinear form of the van't Hoff equation was used (see "Experimental Procedures"). For sFc␥RIIa and IIb, favorable enthalpic (⌬H vH ϳ Ϫ5.3 to Ϫ6.4 kcal⅐M Ϫ1 ) and entropic (ϪT⌬S°ϳ Ϫ1.9 to Ϫ2.7 kcal⅐M Ϫ1 ) changes contribute to the binding energy (⌬G°ϳ Ϫ7.9 to Ϫ8.5 kcal⅐M Ϫ1 ) at 25°C. In contrast, for sFc␥RIII/IgG or sFc␥RIII/Fc interactions, large unfavorable entropic changes (ϪT⌬S°ϳ 6.9 -7.4 kcal⅐M Ϫ1 ) were compensated by even larger favorable enthalpic changes (⌬H vH ϳ Ϫ14.9 to Ϫ15.4 kcal⅐M Ϫ1 ) (Table III). Thus despite having a similar binding affinity and similar kinetics, the sFc␥RIII/Fc interaction has very different thermodynamic characteristics. The Fc␥Rs also differ with respect to the heat capacity (⌬C p ). This is derived from the nonlinear van't Hoff fit and is a measure of the dependence of the enthalpic and entropic changes on temperature. ⌬C p for the sFc␥RIII/Fc interaction is at the low end of the range of values (42) measured for other protein/protein interactions (Ϫ0.333 Ϯ 0.2 kcal⅐mol Ϫ1 ⅐K Ϫ1 ), whereas the ⌬C p values measured for sFc␥RIIa and sFc␥RIIb are in the middle of the range (Table  III). DISCUSSION In this paper we report the use of SPR to measure the kinetic and thermodynamic properties of interactions between the low affinity Fc␥Rs and IgG or Fc. Although some of these properties have been measured before, no previous study has undertaken a comprehensive kinetic and thermodynamic analysis of sFc␥R/ ligand interactions.
All three sFc␥Rs bound with similar affinities (K d ϳ 0.6 -2.5 M). These were at the high affinity end of the range of values reported for other interactions that mediate cell-cell recognition (Table IV). Our results agree well with affinities measured independently using SPR (K d ϭ 1.  (50). This consistency between independent studies using different techniques suggests that these affinity measurements are correct.
All three Fc␥Rs bound to Fc or IgG with typical association rate constants (k on ϭ 130,000 -540,000 M Ϫ1 ⅐s Ϫ1 ) and dissociated with fast dissociation rate constants (k off Ͼ 0.3 s Ϫ1 ). This indicates that interactions between monomeric IgG and the low affinity Fc␥Rs are relatively unstable, and that only multivalent interactions will persist, providing an explanation for the specificity of the low affinity Fc␥Rs for aggregated or cellassociated IgG. Fast dissociation rate constants facilitate the reversal of multivalent Fc␥R/IgG interactions where this is necessary. This may be important in situations where Fc␥Renhanced activity is inhibited because of the engagement of the inhibitory Fc␥RIIb on respective cells. In common with many leukocyte receptors, signal transduction by engaged Fc␥Rs involves tyrosine phosphorylation of their cytoplasmic tails and   (Table IV) suggests that this first step of the signaling mechanisms is conserved. Thermodynamics-Despite similarities in binding affinities and kinetics, thermodynamic analysis unexpectedly revealed major differences between Fc␥RII (a and b) and Fc␥RIII. Binding by sFc␥RIIa and sFc␥RIIb was characterized by favorable entropic and enthalpic changes, which is typical of protein/ protein interactions including cell-surface receptor/ligand interactions (Fig. 5B). In contrast, sFc␥RIII binding was accompanied by large, unfavorable entropic changes compensated for by even larger favorable enthalpic changes (Fig. 5). Large, unfavorable entropic changes have been reported for TCR/peptide-MHC (28,32) and gp120/CD4 (33) interactions. Major causes of unfavorable entropic changes upon binding are a reduction in conformational flexibility upon binding or the trapping of water molecules at the binding interface. For gp120/CD4 and TCR/peptide-MHC interactions, binding was also characterized by slow k on values and (for TCR/peptide-MHC) strongly temperature-dependent kinetics, consistent with a requirement for conformational adjustments, which need some time to fit and reach the proper binding state. In support of this, structural studies have shown that binding is accompanied by conformational adjustments. In contrast, sFc␥RIII/Fc interactions did not exhibit slow k on values or strongly temperature-dependent kinetics (E a diss ϭ 7-16 kcal/ mol for Fc␥Rs). This suggests that unfavorable entropic changes upon binding might be the result of the retention of water molecules at the binding interface or a reduction in conformational flexibility. The last explanation is supported by the crystal structure of the sFc␥RIII/hFc1 complex, where structural changes upon complex formation are observed (14).
Fc␥RIIs and Fc␥RIII also differ significantly in the heat capacity changes (⌬Cp) associated with binding. It has been shown (34) that ⌬C p for a given interaction can be predicted from the size of the buried polar (⌬A p ) and nonpolar surface areas (⌬A np ) using the relationship shown in Equation 2.
⌬C p͑calc) ϭ ͑͑0.32 Ϯ 0.04͒⌬A np Ϫ ͑0.14 Ϯ 0.04͒⌬A p )cal⅐mol Ϫ1 K Ϫ1 (Eq. 2) Based on the crystal structure of the Fc␥RIII/Fc complex, the ⌬C p(calc) can be calculated to be Ϫ200 cal⅐mol Ϫ1 ⅐K Ϫ1 , assuming the surface areas of Fc and Fc␥RIII structures do not change upon complexation. This derived value is much higher than the   (34). ⌬S rt should be identical for all receptor-ligand interactions and has been estimated to be Ϫ50 cal⅐M Ϫ1 ⅐K Ϫ1 (34). From our present data, ⌬S other was calculated to be Ϫ87, Ϫ18, and Ϫ225 cal⅐mol Ϫ1 K Ϫ1 for sFc␥RIIa, -IIb, and -III, respectively. Because ⌬S other is believed to result from local folding and/or water-mediated interactions in the interface (see references in Ref. 35), these data suggest that sFc␥RIII binding is accompanied by a more substantial conformational change and/or retention of more water molecules in the interface than is the case for sFc␥RIIa and -IIb binding.
The recently solved structure of the sFc␥RIII/Fc fragment complex (14) revealed that binding is accompanied by conformational adjustments. More specifically there is an increase in the interdomain angle within the Fc␥RIII molecule upon binding, resulting in an opening of the interdomain cleft. It is noteworthy that Fc␥RIII has only two potential hydrogen bonds within the domain interface in comparison with the seven hydrogen bonds in the Fc␥RII isoforms (25). This suggests that Fc␥RIII exhibits greater interdomain flexibility than the Fc␥RIIs and that there is consequently a greater reduction in conformational entropy upon ligand binding. Furthermore, rotational freedom is lost for the Trp-117 side chain upon IgG binding as it rotates into the interdomain cleft, where it contacts Tyr-17. The observed thermodynamic differences between Fc␥RII and Fc␥RIII might therefore be explained by greater interdomain domain flexibility of Fc␥RIII compared with Fc␥RIIs, as the hydrophobicity of the putative Fc binding site on Fc␥Rs (FG, BC, and CЈE loops) is not significantly different among Fc␥Rs.
Ghirlando and co-workers (31) also found that the Fc␥RIII/Fc interaction is characterized by unfavorable entropic and favor-able enthalpic changes, but shifts were less than reported in this study (⌬H ϭ Ϫ10 kcal⅐mol Ϫ1 (adapted to 25°C) versus Ϫ15 kcal⅐mol Ϫ1 in this study; ϪT⌬S ϭ ϩ3 kcal⅐mol Ϫ1 versus ϩ7 kcal⅐mol Ϫ1 in this study). The discrepancy may result from the use of glycosylated Fc␥RIII by Ghirlando et al. (31), which has shown to affect IgG binding (36).
Stoichiometry of sFc␥Rs/Fc Complex-Given that the IgG molecule is dimeric, it is possible that either one or two FcR might bind a single IgG (or Fc). Both the equilibrium and the kinetic data in the present study were fitted well by the standard 1:1 binding model, which, however, cannot discriminate between one or two binding sites on the IgG with identical affinity for the low affinity Fc␥Rs. X-ray crystallography of the Fc␥RIII/Fc (14) and NMR analysis of the sFc␥RII/Fc interaction (37) both reported a 1:1 binding stoichiometry between Fc␥Rs and Fc. In the case of the sFc␥RIII/Fc interaction, binding of one sFc␥RIII molecule induced a conformational change in the Fc fragment that appeared to obstruct the binding of a second Fc␥RIII. These findings are in contrast to an earlier report, where a 2:1 stoichiometry of the sFc␥RII/hFc complex was reported using the method of equilibrium gel filtration with sFc␥R proteins expressed in insect cells (22). A repetition of these experiments with the same sFc␥RIII and hFc1 fragment used in the present study suggested that more than one Fc␥R is binding to one Fc fragment (data not shown). This result can either be interpreted as a second binding site on Fc with very low affinity or the tendency of the sFc␥R proteins to bind unspecifically to the Fc at higher concentrations, which were used in these experiments. Thus, although the bulk of the evidence supports a 1:1 stoichiometry, we cannot exclude that a second Fc␥R can bind with significantly lower affinity under certain conditions. Table III, other members of IRS, i.e. KIR and CTLA4, also show binding characteristics typical of cell-cell recognition interactions. Dennis et al. (19) noted that some Ig-like members of IRS family, namely Fc␥Rs and KIR, show a similar, distinctive structural topology (Fig. 1). Wines et al. (38) pointed out that the ligand binding sites for several members of IRS family, such as Fc␥RII, Fc␥RIII, Fc⑀RI␣, Fc␣R, KIR, and CTLA-4 all include the FG loop (Fig. 1). Based on gene localization, Kasahara (39) suggested that human chromosome 19q13.4, which includes the KIR/ILT family genes, may be paralogous to 1q23, where the Fc␥R genes are located (39). These similarities in topology, gene localization, ligand binding site, and (probably) signaling mechanism suggest that all IRS molecules evolved from a common ancestral inhibitory receptor.

Comparison with Other Inhibitory Receptor Superfamily (IRS) Members-As shown in
Interestingly, most of IRS family have both activating and inhibitory forms. Van der Merwe et al. showed that the inhib- itory receptor for CD80 (CTLA-4) has a higher affinity than the activating one (CD28) (40). The same differences were observed between inhibitory and activating counterparts of other IRS molecules such as KIRs and CD94/NKG2 (reviewed in Ref. 41).
In contrast, we show here that the activating (Fc␥RIIa and III) and inhibitory (Fc␥RIIb) Fc␥R receptors have very similar affinities. This difference between Fc␥R and other IRS molecules could be related to functional differences. Because NK cells are inactive toward normal cells and activated by tumor or virusinfected cells, it seems reasonable that inhibitory KIR and CD94/NKG2 interactions should be dominant over activating ones. However, the role of the inhibitory Fc␥RIIb in cellular function may be more complex. There may not be a direct competition between ligation of inhibitory and activating Fc␥Rs, and the balance of inhibition and activation could be either determined by the affinity and/or expression levels of Fc␥Rs on the cell surface or by a varying sensitivity on other levels of the complex signaling cascade.
Conclusion-In this report we show that all three low affinity human Fc␥Rs bind their ligand with low affinities and fast kinetics, as is typical for interactions mediating cell-cell recognition. This suggests that binding of, signaling through, and detachment of low affinity Fc␥Rs may be mechanistically similar to that of other cell-cell recognition molecules. Thermodynamic analysis revealed that binding of sFc␥RIII is accompanied by highly unfavorable entropic changes, whereas entropic changes are favorable for sFc␥RII binding. Taken together with recent structural data, our results suggest that sFc␥RIII binding is accompanied by energetically significant conformational adjustments and/or water uptake in the interface.