Identification of the high affinity receptor binding region in human immunoglobulin E.

We have investigated the capacity of N- and C-terminally truncated and chimeric human (h) IgE-derived peptides to inhibit the binding of 125I-labeled hIgE, and to engage cell lines expressing high and low affinity receptors (Fc-epsilon-RI/II). The peptide sequence Pro343-Ser353 of the hC-epsilon-3 domain is common to all h-epsilon-chain peptides that recognize hFc-epsilon-RI. This region in IgE is homologous to the A loop in C-gamma-2 that engages the rat neonatal IgG receptor. Optimum Fc-epsilon-RI occupancy by hIgE occurs at pH 6.4, with a second peak at 7.4. N- or C-terminal truncation has little effect on the association rate of the ligands with this receptor. Dissociation markedly increases following C-terminal deletion, and hFc-epsilon-RI occupancy at pH 6.4 is diminished. His residue(s) in the C-terminal region of the epsilon-chain may thus contribute to the high affinity of interaction. Grafting the homologus rat epsilon-chain sequence into hIgE maintains hFc-epsilon-RI interaction without conferring binding to rat Fc-epsilon-RI. hFc-epsilon-RII interaction is lost, suggesting that these residues also contribute to hFc-epsilon RII binding. h-epsilon-chain peptides comprising only this sequence do not block hIgE/hFc-epsilon-RI interaction or engage the receptor. Therefore, sequences N- or C-terminal to this core peptide provide structures necessary for receptor recognition.

Antibodies of the immunoglobulin (Ig)E isotype sensitize target cells expressing the class-specific Fc receptors for antigen-induced mediator release, by binding through residues located in the Fc portion of the molecule (1,2). The potent pharmacologically active substances that are released in response to this stimulus cause the clinical symptoms of allergy.
Strategies that block the initial sensitization of target cells with antigen-specific IgE have been explored following the demonstration that human (h) 1 myeloma IgE-derived Fc⑀ fragments generated by proteolytic cleavage with papain (1, 2), which produces peptides comprising h⑀-chain residues 1-226 and 227-547, can competitively inhibit the binding of IgE to cells expressing high affinity receptor (Fc⑀RI) (1,2), whereas cleavage products of pepsin digestion, which generates ⑀ fragments spanning residues 1-338, 339 -349, and 350 -547 (3) do not inhibit binding. This observation initiated the quest for progressively smaller peptides as potential IgE antagonists (reviewed in Ref. 4). In early studies, the inhibition of passive cutaneous anaphylaxis in human skin was used to assess the Fc⑀RI-blocking activity of proteolytic fragments or recombinant IgE-derived peptides expressed in Escherichia coli (1,5,6). This led to the proposal that sequences N-and C-terminal to Val 336 contribute structures necessary for Fc⑀RI interaction (6). More recent studies aimed at the identification of the receptor binding site(s) employed chimeric human/mouse IgE antibodies, ⑀/␥ chimeras, site-specific mutagenesis, anti-IgE antibodies, or IgE-derived peptides (7)(8)(9)(10)(11)(12)(13)(14)(15). They indicate that the site(s) in IgE that interact(s) with the Fc receptors depend(s) on structures associated with residues located in the C⑀3 domain, although C⑀4 involvement has also been invoked (11,12). Furthermore, it has been suggested that IgE/Fc⑀RI interaction is mediated primarily by electrostatic interactions (14) and dependent on the entire C⑀3 in its native conformation (10), while the C⑀4 domains are essential for the maintenance of the active conformation of the C⑀3 domain (7,16). Our earlier investigations showed that while IgE/Fc⑀RII interaction is critically dependent on C⑀4 or its homologue C␥3 (16), it is possible to delete the entire C⑀4 domain and more than 60% of residues in C⑀3 and still maintain the Fc⑀RI-blocking capacity of the recombinant ⑀-chain fragment (6,17).
Based on our demonstration of the parallel nature of the inter-⑀-chain disulfide bonds in hIgE (18), we developed a structural model that predicts that an exposed and probably flexible segment connects the globular portions of the C⑀2 and C⑀3 domains (18,19). Subsequently, Gould et al. (20,21) claimed that the N-terminal 11 residues in C⑀3, which are included in this segment, are essential for Fc⑀RI binding. This proposal relied on studies where the biological activity of recombinant ⑀-chain fragments was tested by blocking the binding of ragweed-specific IgE to mast cells in the skin of the senior investigator conducting the study (6). When the fallibility of the passive cutaneous anaphylaxis reaction in assessing the biological activity of recombinant IgE-derived fragments emerged (17,22), we re-assessed the biological activity of these and additional truncated fragments using our recently developed receptor binding assay, which allowed us to study direct binding of IgE-derived ligands to rat (r) basophilic leukemia cells (RBL-2/2/C) transfected with the ␣-chain of hFc⑀RI (23).
In the present investigation we describe the capacity of a series of overlapping N-and C-terminally truncated and chimeric ⑀/␥-chain derived fragments, expressed as glutathione S-transferase (GST) fusion proteins in E. coli to bind directly to and block the binding of hIgE to RBL-2/2/C cells. We show that the peptide sequence spanning amino acid residues Pro 343 -Ser 353 is common to all recombinant ⑀-chain fragments capable of binding to Fc⑀RI. Deletion of this sequence is associated with a complete loss of receptor recognition, confirming earlier observation by others that grafting the homologous sequence from IgG1 into hIgE reduces Fc⑀RI binding by 97% (14). Replacing this sequence in hIgE by the homologous rat sequence maintains binding to hFc⑀RI, but there is a loss of hFc⑀RII interaction, confirming earlier observations by others that rodent IgE recognizes only hFc⑀RI but not hFc⑀RII (10). Since recombinant GST⅐⑀-chain fusion proteins containing this sequence do not block IgE/Fc⑀RI␣ interaction, we conclude that sequences N-or C-terminal to this core peptide are essential for the provision of additional structural scaffolding in order to generate a receptor binding conformation. Viewed in the context of our model structure for IgE (18,19), this core peptide has been computed to form a loop proximal to the interface between the C⑀3/4 domains that is homologous to the site in rodent IgG involved in the binding to the groove formed by the ␣1 and ␣2 domains of the neonatal Fc␥Rn (24). Interestingly, as for IgG/Fc␥Rn interaction, we also observe two pH optima at pH ϳ6.4 and 7.4 for hIgE/Fc⑀RI␣ interaction. While N-or C-terminal truncation has little effect on the association rate, deletion of C-terminal sequences increases the rate of dissociation several hundred-fold and reduces receptor occupancy at pH 6.4. The slow dissociation of IgE from Fc⑀RI␣ therefore may be due, at least in part, to the stabilization of the interaction by His residues in the C-terminal region of the ligand.

EXPERIMENTAL PROCEDURES
Gene Constructs and Site-specific Mutagenesis-The numbering scheme for h⑀-chain amino acid residues used in previous publications (6, 16 -18) has been maintained. Polymerase chain reaction (PCR) was used to amplify ⑀-chain fragments comprising the entire Fc region (residues 226 -547), the C␥3 from mouse IgG2a, the C⑀2 domain (residues 226 -329), and the C⑀4 domain (residues 440 -547). N-terminal deletions of the Fc region were prepared starting at amino acid residue positions 326, 330, 340, 342, 343, 344, 345, 350, and 355 and terminating at residue 547. C-terminal deletions were prepared starting at amino acid residue 226 and terminating at residues 361, 357, 354, 353, 352, 345, and 340. The DNA products were purified by agarose gel electrophoresis, digested with appropriate restriction enzymes, and subcloned into the bacterial expression plasmids pGEX-3X and pGEX-KG, which direct the synthesis of foreign polypeptides in E. coli as fusions with the 26-kDa GST (26). Cloning the recombinant ⑀-chain fragments in frame to the 3Ј end of the GST gene facilitates the production of large amounts of fusion protein (ϳ500 mg/liter). In addition to a versatile multiple cloning site, the vectors have been engineered so that the GST carrier can be cleaved off by digestion with coagulation factor Xa or thrombin. The initial screening for receptor-blocking activity was carried out with partially purified GST⅐⑀-chain fusion peptides. Following affinity purification on rabbit anti-GST affinity columns and GST removal with thrombin, ⑀-chains showed identical receptor-blocking capacities when compared with GST⅐⑀-chain fusion peptides. Therefore, this step was eliminated, and all assays described in this study were carried out with affinity-purified GST⅐⑀-chain fusion peptides. Short GST fusion peptides comprising ⑀-chain residues 338 -359 and 340 -357 were also generated. Site-specific mutagenesis was performed by overlap extension PCR (25). Bacterial strains used as host for transformation were JM109 or MC1061.
For the construction of the chimeric h/r IgE molecule we employed the ⑀-chain expression plasmids pSV-V NP h ⑀ /r ⑀ (18). A construct where the sequences known to be essential for hFc⑀R1 interaction had been replaced by the homologous rat sequence encoding residues 341-356 was also generated by overlap extension PCR (25). The template for PCR was a 3.4-kilobase pair IgE C⑀1-4 genomic DNA cassette cloned into the BamHI site in pUC19 (pH⑀). A 719-base pair fragment coding essentially for C⑀2-3 was generated by PCR. This involved two rounds of PCR and four primers, two external (5Ј BglII, CGTGAAGATCTTACAGTCGTC; 3Ј NcoI, CCTGCCCATGGCT-CACCG) and two internal primers (5Ј h-r16, CCTCGACCTGTATGAA-AATGGGACTCCCAAACTTACCTGTCTGGTGGTGGACCTG; 3Ј h-r16, CCATTTTCATACAGGTCGAGGGGACTGGGTGGGATTAGGTAGG-CGCTCACCCCTCT).
For each PCR, reaction mixtures contained 200 ng of template, 2 g of each primer, 1 mM dNTPs, 10 mM Tris-HCl, pH 8.3, 50 mM KCl, 1.5 mM MgCl 2 in 100 l, and following a hot start 1 unit of Taq polymerase was added. An initial denaturation cycle at 96°C for 6 min, 64°C for 2 min, 72°C for 1 min 30 s was followed by 30 cycles at 94°C for 1 min 30 s, 64°C for 1 min 30 s, 72°C for 1 min 30 s. The resultant 719-base pair fragment was cloned into pH⑀ using BglII and NcoI sites to give a chimeric C⑀1-4 cassette, which was subcloned, using the BamHI sites, into the mammalian expression vector pSV-V NP (18). The orientation of this cassette was checked by PCR.
The identity of all gene constructs was confirmed by sequencing the DNA of both strands.
Gene Expression-E. coli strains transformed with the expression plasmids were grown overnight at 37°C, and the overnight culture was diluted 100-fold into LB broth containing 100 g/ml ampicillin and grown to an absorbance of 0.4 at 600 nm at 37°C. The inducer isopropyl-1-thio-␤-D-galactopyranoside (Sigma) was added to a final concentration of 0.1 mM, and the cultures were grown under constant shaking at 37°C for 4 h. Bacterial cells were harvested by centrifugation at 5,000 ϫ g for 15 min, and the pellets were frozen at Ϫ70°C until purification of the recombinant proteins. Freezing and subsequent thawing of the bacterial pellets were essential to obtain effective solubilization of the recombinant proteins, which are expressed as insoluble inclusion bodies.
Purification of Recombinant GST⅐⑀-chain Fusion Proteins from E. coli Cell Pellets-This was carried out using procedures described for the purification of recombinant ⑀-chain fragments expressed in E. coli (6). Frozen cell pellets were defrosted on ice before homogenization (5-fold pellet volume) in 0.05 M Tris-HCl buffer, pH 7.9, containing 2 mM EDTA, 0.1 mM dithiothreitol, 1 mM ␤-mercaptoethanol, 0.25 M NaCl, 0.1% sodium deoxycholate, 25 g/ml phenylmethylsulfonyl fluoride, 5% glycerol. The homogenates were dispersed by sonication before the addition of 100 g/ml lysozyme and 20 g/ml DNase I. Homogenates were kept on a rotary shaker for 12-15 h at 4°C before centrifugation at 10,000 ϫ g for 10 min. The pellets were washed twice in a 20-fold pellet volume of 0.05 M Tris-HCl buffer, pH 7.9, containing 1 mM EDTA, 0.1 M NaCl, 25 g/ml phenylmethylsulfonyl fluoride. Inclusion bodies from cell pellets were solubilized in 0.05 M Tris-HCl buffer, pH 7.9, containing 8 M urea, 1 mM EDTA, 0.1 M NaCl, 25 g/ml phenylmethylsulfonyl fluoride and dialyzed for 12 h against a 200-fold volume of the same buffer omitting urea but with the addition of 0.1 mM dithiothreitol and 1 mM ␤-mercaptoethanol. Insoluble materials were removed by centrifugation, and 30 -75% of recombinant ⑀-chain peptides were found in the supernatant fraction. Affinity purification from this fraction was carried out using a rabbit anti-GST antiserum coupled to Sepharose 4B. The chimeric h/r antibody was purified from cell culture supernatants using NP-specific affinity columns and analyzed by polyacrylamide gel electrophoresis (PAGE) and immunoblotting (6,18).
Ligand Binding Studies and Cell Culture-hIgE V NP (18) was iodinated as described previously, and the conditions for ligand binding and cell culture have been published (23). Affinity-purified GST⅐⑀-chain fusion peptides were iodinated at 0 -4°C in 0.4 M phosphate pH 7.4/7.5 using 4.4 Ci of Na 125 I and 150 -300 g of peptide in tubes coated with 40 g of IODO-GEN (Pierce). Following a 15-min incubation period, the reactions were terminated by removing fluid from the coated tubes. Each preparation was fractionated on a 140-ml Sephacryl S-200 column (Pharmacia), pre-equilibrated with binding buffer (phosphate-buffered saline, 0.2% BSA, pH 7.4), which effects the separation of dimers and monomers. Following ␥-counting of collected fractions, peak fractions were pooled, aliquoted, and stored at Ϫ70°C. Specific activity ranged from 5.8 -15.5 Ci/g.
The conditions for maintenance of RBL-2H3 cell lines transfected with the ␣-chain of hFc⑀RI (RBL-2/2/C) have been described (23). RBL-2/2/C clones were plated into 48-well plates at an initial plating density of 10 5 cells/well and incubated with 10 Ϫ6 M dexamethazone for 24 h at 37°C. In preliminary experiments, RBL-2/2/C cells were incubated with increasing concentrations of 125 I-labeled ligands (0.1-7.5 g/ml) to determine the minimum saturation concentrations for Fc⑀RI binding. The proportion of molecules capable of binding to Fc⑀RI was 78 -91% for Nonspecific binding was determined using a 50 -100-fold molar excess of nonlabeled hIgE over 125 I-hIgE, and the same amount of GST was used as a negative control. The binding of recombinant proteins to Fc⑀RI␣ was determined indirectly, after correcting for nonspecific binding (7-17%), by calculating the percentage of inhibition of 125 I-IgE binding to cells. To measure the inhibition (IC 50 ) of 125 I-IgE binding to RBL-2/2/C or the 8866 lymphoblastoid cell line (18) by native and chimeric h/r IgE V NP and recombinant ⑀-chain fragments, cells were preincubated with increasing concentrations (10 Ϫ12 -10 Ϫ5 M) of each of the unlabeled peptides in 125 l of binding buffer or, as control, binding buffer alone at 22°C for 1 h, before the addition of 50 l of binding buffer containing 125 I-hIgE (1 nM). After 45 min, the cells were washed twice with 0.5 ml of ice-cold binding buffer and lysed with 0.5 ml of lysis buffer (0.5 M NaOH, 1% Triton X-100). Samples (0.25 ml) were removed and counted for 5 min on a LKB1277 ␥-counter.
The pH optimum for the binding of 125 I-labeled hIgE and the GST⅐⑀chain fragments to Fc⑀RI was determined by incubating RBL-2/2/C cells (23) in 48-well plates with 100 l of 50 mM phosphate-buffered saline containing 0.2% BSA (pH range 5.9 -8.1) for 10 min at 37°C before adding 50 l of 2 g/ml 125 I-hIgE or 0.7 g/ml of the 125 I-GST⅐⑀chain fragments. Cells were incubated for 30 min, and the excess protein was removed by washing with saline containing 0.2% BSA before measurement of cell-bound label. Results were corrected for nonspecific binding.
Potassium Iodide Titrations-The generation of IgE V NP h ⑀ -(Cys 328 3 Met) has been described (18). The conformations of native and mutant (Cys 328 3 Met) recombinant IgE were investigated by comparing their intrinsic fluorescence. Solute quenching of protein fluorescence involved excitation of Trp residues at 297 nm and measurement of emission in the range 300 -450 nm. Potassium iodide was added gradually to give a quench profile for each protein. Mathematical analysis was carried out according to the Stern-Volmer law (28).

RESULTS AND DISCUSSION
In the present study, we focused on the identification of the site(s) that determine the interaction of hIgE with its cellular receptors. The strategies employed for the expression of an overlapping family of chimeric GST⅐h⑀-chain fusion proteins are outlined in Fig. 1. Panel A summarizes the receptor-binding capacities of the GST⅐h⑀-chain fusion proteins, that of a chimeric ⑀/␥ peptide, and that of a chimeric h/r IgE molecule. The assignment of biological activities is based on (i) competition and (ii) direct binding studies detailed in Fig. 2 and Table I. In Fig. 1, panels C and D show the electrophoretic mobilities of Cand N-terminally truncated recombinant GST⅐h⑀-chain fusion proteins immunoprecipitated with a rabbit anti-GST antiserum, followed by PAGE analysis under nonreducing conditions and immunoblotting with a horseradish peroxidase-labeled rabbit anti-IgE serum. As shown in Fig. 1, C-terminal truncation yields a number of ⑀-chain peptides for each construct. As judged by PAGE (Fig. 1C) and column chromatography (data not shown), approximately one-third of the peptides in each set corresponds to the full-length fusion peptide as a monomeric fragment. None of these fragments show any propensity to dimerize, although biologically inactive polymeric aggregates form at concentrations Ͼ1.3 mg/ml. A set of identical fragments is observed following analysis under reducing conditions (data not shown). Most of the smaller ⑀-chain fragments represent proteolytic cleavage fragments that are recognized by monoclonal antibodies specific for the C⑀2 domain. 2 In contrast, deletion of N-terminal sequences gives rise to two ⑀-chain fragments and their apparent molecular weight under nonreducing (Fig. 1D) and reducing conditions (data not shown) indicates that they correspond, in almost equal quantities, to monomeric and dimeric GST⅐h⑀-chain fusion proteins.
The present investigation confirms our previous observations, which show that only those peptides that contain C⑀4 or the homologous C␥3 domain can engage both Fc⑀RI and Fc⑀RII (6,16), while C-terminal truncation of the ⑀-chain results in elimination of binding to Fc⑀RII. As summarized in Fig. 1A, sequences common to all fragments capable of binding to Fc⑀R1␣ comprise residues Pro 343 -Ser 353 in the C⑀3 domain. Further deletion from either the C-or N-terminal end beyond these residues is associated with a loss of Fc⑀RI binding. As shown in Fig. 2 and Table I, GST⅐⑀-(340 -547) and GST⅐⑀-(226 -354), which comprise the core peptide, inhibit the binding of hIgE with an IC 50 in the nanomolar range. In contrast, blocking of IgE/Fc⑀RI interaction by the GST control, GST⅐⑀-(226 -340), GST⅐⑀-(355-547), and GST⅐⑀-(440 -547) is identical and cannot be detected even above micromolar concentrations. These results confirm observation by others who find that recombinant IgE-derived fragments comprising residues 355-547 do not block hIgE binding to hFc⑀RI (11) and that substitution of residues 346 -353 by the homologous sequence from IgG1 reduces binding of the chimera to background levels (14).
As shown in our model structure of hIgE-Fc (Fig. 3) (18), this sequence forms a loop that is homologous to the loop in rIgG shown to bind to the neonatal Fc␥Rn (24). A further similarity emerged when we investigated the pH dependence of the binding of hIgE to Fc⑀RI. As shown in Fig. 4 two pH optima are observed for the binding of hIgE to Fc⑀R1, and occupancy of the receptor is almost twice as high at pH 6.4 as at pH 7.4. Although the significance of this is not known, it is tempting to speculate that hIgE has evolved the lower pH optimum as a result of its physiological importance in the fight against parasitic infestations in the lumen of the intestine at acid pH.
Data summarized in Table I show that N-or C-terminal truncation has a negligible effect on the rate of association of biologically active ⑀-chain fragments with Fc⑀RI. In contrast, the rate of dissociation increases several hundred-fold following the deletion of residues from the C-terminal end, and, as shown in Fig. 3, this is associated with a concomitant decrease in receptor occupancy at pH 6.4. Taken together, these data suggest that His residues in the C-terminal region of the IgE molecule make a contribution toward the maintenance of the high affinity interaction between IgE and Fc⑀RI␣ since this is largely determined by the slow rate of dissociation of the ligand from the receptor.
Results obtained in the current study differ in one significant respect from those in our previous investigation (6), where the Fc⑀RI-blocking capacity of IgE-derived fragments was evaluated by the senior investigator, who performed passive cutaneous anaphylaxis tests in his own skin. Employing a well defined cellular assay system (23), we demonstrate here that N-terminal IgE sequences can be deleted beyond residue 340 without any significant effect on the kinetics of ligand/receptor interaction. Our data show that the essential determinant for hIgE/ Fc⑀RI recognition depends on a consecutive sequence comprising 11 amino acids computed to form a loop at the interface between the C⑀3 and C⑀4 domain (18). In accord with others (7-10), our observations exclude any direct contribution of C⑀4specific residues as proposed by Stanworth et al. (12). Our results confirm and extend those made by Nissim et al. (8 -10), who demonstrated that the receptor binding site in IgE is located in the C⑀3 domain. They differ from the claims of Hamburger (29) and Gould et al. (20,21), who propose, respectively, that residues 330 -334 and 329 -340 in the switch region between C⑀2 and C⑀3 are essential for IgE/Fc⑀RI binding. As the results of our study clearly demonstrate, these sequences can be deleted without any major influence on the kinetics of hIgE/Fc⑀RI interaction.
It is interesting to note that the active core sequence identified by us corresponds closely to the hIgE-derived peptide generated by Nio et al. (15), who report its capability to block the binding of antigen-specific IgE to cells expressing Fc⑀RI at concentrations in the mM range (15).
Although the results of our study indicate that fragments containing the C⑀2 domain show an increased susceptibility to proteolysis (Fig. 1C), the inclusion of the protease inhibitor phenylmethylsulfonyl fluoride during the isolation procedure facilitates the purification of peptides that engage Fc⑀RI/II. Using our published method, others have been unable to generate h⑀-chain fragments in E. coli that retain Fc⑀RI-binding capacity and have attributed this failure to folding problems (11). At least one other laboratory has expressed ⑀-chain fragments in E. coli which are biologically active (30).
Based on the outcome of Fc⑀RI binding studies with chimeric and mutant hIgE molecules, Presta and co-workers (14) proposed that six amino acid residues located in three loops, C-D, E-F, and F-G, computed to form the outer ridge on the most exposed side of the C⑀3 domain, are involved in receptor binding primarily by electrostatic interactions (14). These conclusions were based on the observation that replacement of these  (25) and expressed in E. coli (6,18). The ⑀-chain expression plasmids pSV-V NP h⑀/r⑀ were employed for the construction of mutant and chimeric IgE molecules and expressed in the J558L myeloma cell line (18). Panels A and B summarize the ability of the truncated, chimeric, and mutant ⑀-chain variants to bind to Fc⑀RI␣ expressed on RBL-2H3.1 and RBL-2/2/C cells (23,27) and to Fc⑀RII expressed on the 8866 lymphoblastoid cell line (18). Initial screening for biological activity was determined by assessing the capacity of GST⅐⑀-chain fusion proteins to inhibit the binding of 125 I-labeled hIgE (1 nM) to the receptors. The degree of inhibition effected by nonbinders was identical, within limits of experimental error, to that observed with GST, which was included as a negative control (see Fig. 2). Purification of truncated recombinant GST⅐⑀-chain fusion proteins and mutant and chimeric IgE molecules was carried out as described (see "Experimental Procedures"). Ligands were labeled with 125 I for direct binding studies (see Table I). Nonbinders showed no binding above background even at concentrations above 10 Ϫ5 M. Panels C and D show GST⅐⑀-chain fusion proteins that were immunoprecipitated with a rabbit anti-GST serum, followed by SDS-PAGE (12%) separation under nonreducing conditions and immunoblotting with a horseradish peroxidase-labeled rabbit anti-human IgE serum. residues reduced the binding of variant molecules to Fc⑀RI␣ relative to native hIgE. We disagree with their conclusion in view of the fact that most of the mutations at Arg 376 (408), Ser 378 (411), Lys 380 (415), Glu 414 (452), Arg 427 (465), and Met 429 (469) (Presta et al. (14) numbering scheme in parentheses), which affect IgE/Fc⑀RI interaction to a greater or lesser extent, are invariably due to replacements by residues of opposite charge or by a Pro, changes which could cause structural rearrangements. In contrast, more conservative substitutions of these residues either have little effect or cause an apparent enhancement of binding (Ref. 14, Table I). Our own study shows that e.g. a single point mutation involving Cys 328 , which by itself is not required for either Fc⑀RI or Fc⑀RII binding (18), can have a dramatic effect on the conformation of the IgE molecule. Its substitution by Met, but not Ser, destroys binding to both receptors (18). When we compared the intrinsic fluorescence of Trp residues in the native and IgE Met 328 molecule, we found that on average native IgE has 41% of its Trp residues exposed to solvent, while IgE Met 328 was found to have only 22% of Trp residues exposed, although both molecules were recognized by a conformation-dependent monoclonal antibody directed against the C⑀2 domain. This observation shows that the substitution of a single amino acid that is not involved in receptor recognition can induce a significant deformation in structure and profoundly affect ligand/receptor association. Presta et al. (14) have also claimed that the grafting of loops C-D, E-F, and F-G and the inter-C⑀2/3 switch region into hIgG (which they refer to as IgGEL), conferred Fc⑀RI binding to hIgG 1 . Their own data, however, on the binding of variant IgE do not support this interpretation. It is important to point out that their chimeric IgGEL construct still retains the endogenous IgG 1 loop A-B sequence, which when grafted into hIgE (Ref. 14, Table I  This represents a greater reduction in activity than any other loop replacements described in their study. Data in their Fig. 3, which is interpreted by them to support their claim that the IgGEL chimera can recognize Fc⑀R1, demonstrate the opposite since they show that when CHO 3D10 cells are incubated with IgGEL at a concentration of 1 g/ml, less than 2% of cells become labeled. Since these are the criteria that they applied for the assignment of the receptor binding activity of all other chimeric and mutant IgE molecules described in their investigation (Ref. 14, Table I), we conclude that IgGEL has the same affinity for Fc⑀RI as variant 1, which is less active than any other chimera or mutant reported in their study. Although Fig.  3 of their paper (14) appears to suggest ϳ40% binding at 20 g/ml IgGEL, there is no evidence for saturation, and the slope of the line is indicative of nonspecific binding. In our opinion, the data of Presta et al. provide compelling evidence that the A-B loop in hIgE comprises the essential structural determinant for IgE/Fc⑀RI␣ interaction because the only difference between their variant 1 and native hIgE is that in variant 1 the core sequence, which we have shown to be common to all hIgE fragments that can engage hFc⑀R1␣, is replaced by the IgG1 homologue.
As our study shows, sequences N-or C-terminal to this core peptide are necessary to provide structural scaffolding for the maintenance of a receptor binding conformation since the core peptide alone cannot engage the receptor. Deletion of C-terminal, but not N-terminal, sequences diminishes receptor occupancy at pH 6.4 and increases the dissociation of the ligand from the receptor. We conclude that residues, including His, in the C-terminal domain make an important contribution toward the maintenance of the high affinity of interaction between IgE and Fc⑀RI␣.
Although human and rodent IgE are highly homologous, rodent IgE can engage only hFc⑀RI but not hFc⑀RII (10). In contrast, hIgE cannot bind to either of the rodent receptors. The structural basis for this phenomenon is unknown. In the current study we have replaced the human A loop by the rat homologue and find that the chimera still recognizes hFc⑀RI. This graft, however, does not confer binding to rFc⑀RI, indicating that conformational determinants outside the A loop in hIgE inhibit recognition of rFc⑀RI. The fact that this replacement destroys the binding of hIgE to hFc⑀RII indicates exquisite species specificity and suggests that residues in this motif contribute to the binding to both receptors. Nissim et al. (10) have demonstrated that the species specificity for recognition of Fc⑀RII by murine and hIgE is contained in the C⑀3 domain and proposed that part of the binding energy for hFc⑀RII interaction is contributed by amino acid residues between 346 and 356.
At present, limited structural information is available regarding the interaction between IgE and its receptors. Unlike IgG/Fc␥Rn interaction, where the ligand can engage two receptors, IgE molecules bind to Fc⑀RI␣ in a 1:1 stoichiometry, although bilateral symmetrical protection to proteolysis has been observed when rodent IgE is complexed to the ␣-chain (24,31,32). This has been explained in terms of a bent conformation of IgE (33), where the second ⑀-chain becomes inaccessible to an additional copy of the receptor, or to antibodies directed against epitopes in IgE that become masked following receptor engagement. There is little evidence for a beneficial role for IgE antibodies except in parasitic diseases, and such epitopes may therefore have an application as immunogens for the therapy of IgE-mediated allergies, since naturally occurring and monoclonal antibodies (33-36) have been described that block the binding of IgE to cells expressing Fc⑀R1 but do not trigger mediator release. An improved understanding of the docking of hIgE to its receptors will provide the structural information needed for the rational design of such immunogens. The identification of the binding site constitutes a major step in this direction. FIG. 4. pH profile for the binding of native IgE and recombinant IgE-derived peptides to RBL-2/2/C cells. h␣-chain-transfected RBL-2/2/C clones were distributed into 48-well plates at 10 5 cells/well and incubated with 10 Ϫ6 M dexamethazone for 24 h at 37°C. Prior to the assay cells were washed twice with 0.5 ml of saline containing 0.4% BSA and preincubated with 125 l of 50 mM phosphate-buffered saline containing 0.4% BSA (pH range 5.9 -8.1) for 10 min at 37°C before adding 50 l of 2 g of 125 I-labeled IgE (j), 0.7 g of GST⅐⑀-(340 -547) (ࡗ), and GST⅐⑀-(226 -354) (º). Cells were incubated for 30 min, after which unbound ligand was removed by washing with binding buffer before measurement of cell-bound label. Results were corrected for nonspecific binding. (Data shown represent the means of two determinations carried out in duplicate)