HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Mackay, G. A. Articles by Gould, H. J. Search for Related Content PubMed PubMed Citation Articles by Mackay, G. A. Articles by Gould, H. J.

Mutagenesis Within Human FcRI Differentially Affects Human and Murine IgE Binding¹

Graham A. Mackay^2,*, Mark D. Hulett^3,, Justin P. D. Cook^*, Halina M. Trist, Alistair J. Henry^4,*, James M. McDonnell, Andrew J. Beavil^*, Rebecca L. Beavil^*, Brian J. Sutton^*, P. Mark Hogarth and Hannah J. Gould^5,*

^* Randall Centre, New Hunt’s House, King’s College London, Guy’s Campus, London, United Kingdom; Helen M. Schutt Laboratory for Immunology, Austin Research Institute, Austin Repatriation Medical Center, Heidelberg, Victoria, Australia; and Laboratory of Molecular Biophysics, Department of Biochemistry, University of Oxford, Oxford, United Kingdom

	Abstract

Top Abstract Introduction Materials and Methods Results Discussion References

 
Soluble fragments of the -chain of FcRI, the high-affinity receptor for IgE, compete with membrane-bound receptors for IgE and may thus provide a means to combat allergic responses. Mutagenesis within FcRI is used in this study, in conjunction with the crystal structure of the FcRI/IgE complex, to define the relative importance of specific residues within human FcRI for IgE binding. We have also compared the effects of these mutants on binding to both human and mouse IgE, with a view to evaluating the mouse as an appropriate model for the analysis of future agents designed to mimic the human FcRI and attenuate allergic disease. Three residues within the C-C' region of the FcRI 2 domain and two residues within the 2 proximal loops of the 1 domain were selected for mutagenesis and tested in binding assays with human and mouse IgE. All three 2 mutations (K117D, W130A, and Y131A) reduced the affinity of human IgE binding to different extents, but K117D had a far more pronounced effect on mouse IgE binding, and although Y131A had little effect, W130A modestly enhanced binding to mouse IgE. The mutations in 1 (R15A and F17A) diminished binding to both human and mouse IgE, with these effects most likely caused by disruption of the 1/2 interface. Our results demonstrate that the effects of mutations in human FcRI on mouse IgE binding, and hence the inhibitory properties of human receptor-based peptides assayed in rodent models of allergy, may not necessarily reflect their activity in a human IgE-based system.

	Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

 
The interaction between allergen-specific IgE and its high-affinity receptor, FcRI, is key to the onset and propagation of allergic disease (1). The tight association between these molecules enables multivalent allergen to readily cross-link and activate inflammatory cells bearing FcRI with the consequent secretion of numerous inflammatory and immunomodulatory molecules (2).

FcRI consists of three subunits: , which has two domains (1 and 2) and provides the binding site for IgE, , and homodimeric (3). These latter subunits are responsible for transducing the initial cross-linking stimuli into the cellular functional responses of immediate mediator secretion, synthesis of proinflammatory arachidonic acid metabolites, and the transcriptional regulation of numerous genes (4, 5). Some studies also suggest a role for the extracellular region of the -chains in the effective high-affinity membrane display of FcRI (6), a feature paralleled in the homologous Fc receptor, FcRI (7).

Interruption of the first step of this proallergic pathway, and blocking formation of the complex between IgE and FcRI, should prevent inflammatory cell activation and alleviate allergic symptoms. This has been borne out by the clinical success of a humanized anti-IgE Ab (8) that interacts with nonreceptor-associated IgE to prevent binding to FcRI, although other mechanisms such as the inhibition of IgE synthesis and down-regulation of cellular FcRI levels are thought to contribute to the efficacy of the Ab (8, 9). We have chosen to explore another means of producing a similar effect by investigating the potential of soluble forms of FcRI, or rationally designed peptides derived from the receptor, to bind and prevent IgE association with cell-expressed FcRI (10). Such an approach would potentially have the added benefit of inactivating the anti-FcRI autoantibodies that are clearly involved in the etiology of chronic idiopathic urticaria and perhaps other immunological disorders (11, 12).

A soluble form of FcRI (sFcRI)⁶ has been shown in numerous in vivo and in vitro studies to block IgE association with membrane-bound FcRI (13, 14, 15, 16) and, moreover, to displace existing endogenously bound IgE (13). We have partially duplicated these effects using a cyclized 13-mer peptide derived from the C-C' loop of 2 (16). This peptide was able to inhibit both human IgE binding to immobilized sFcRI, and the sensitization of rat basophilic leukemia (RBL) cells with mouse IgE. This latter approach will undoubtedly be further facilitated by the recently determined crystal structure of the complex between FcRI and an IgE fragment that unambiguously identifies the contact residues in both proteins (17).

The crystal structures of both FcRI (18) and the FcRI/IgE complex (17) have conclusively validated significant earlier studies. Domain exchange studies (19, 20, 21), together with phage display and soluble expression of single FcRI domains (22, 23, 24), had demonstrated the essential requirement of the 2 domain for IgE binding, and indeed the crystal structure has shown that the regions of FcRI that contact the C3 domains of IgE lie solely within 2. However, it was also clear from these earlier studies that 1 was required for optimal Ab binding. This can now be rationalized in light of the extensive interdomain contact seen in FcRI (18) and homologous FcR (25, 26, 27). At a more detailed level, potential IgE binding residues had been identified in loop and strand exchange (19, 20, 21), peptide synthesis (16), and site-directed mutagenesis studies (28, 29, 30, 31) within the 2 domain of FcRI. These studies had implicated several regions and highlighted specific residues within 2 as important in complex formation, including the C-C', C'-E, and F-G loops. Involvement of these regions, and many of the specific residues identified in these earlier studies, has now been confirmed by the crystal structure of the FcRI/IgE complex (17).

We have selected three residues within the 2 domain of FcRI for mutagenesis and expression as a soluble construct, to dissect out their relative importance to the formation and maintenance of the high-affinity complex. We previously reported that the mutation K117D in 2 greatly reduces binding of human IgE (29), and we now extend these studies with two additional mutants, W130A and Y131A, and use a wider range of in vitro assay methods. We also explore the possibility of disrupting the interface between 1 and 2 as a way to reduce FcRI affinity for IgE, beginning with independent alanine mutations at R15 and F17. Crystallographic (18) and modeling studies (32) reveal that these amino acids in the 2 proximal A-strand of 1 interact directly with residues of 2.

Previous studies from our laboratories have demonstrated that development of a soluble receptor with (modestly) enhanced affinity for IgE compared with wild-type (WT) FcRI is achievable through single point mutagenesis (29, 30). Such a soluble receptor fragment would presumably out-compete endogenous cell surface-expressed FcRI, and would consequently be more efficacious as a therapeutic agent than the native protein. Mouse models of allergic disease are well suited to testing the in vivo therapeutic potential of such higher-affinity receptor fragments or receptor-based synthetic peptide antagonists. Therefore, we investigated whether mutations in human sFcRI, which are known to affect human IgE binding, have a similar effect upon binding to mouse IgE. Thus, results from the present study have implications for the design of novel FcRI-based therapeutics and for the screening methodologies used in subsequent assessment of their potential therapeutic utility.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

 
Cloning and site-directed mutagenesis within FcRI

Mammalian. The cloning of the soluble WT FcRI fragment (residues Val¹ to Lys¹⁷⁶) and the K117D mutant for mammalian expression in the mouse myeloma cell line NS0 has been described in a previous paper (29).

Pichia pastoris. A shorter construct comprising residues Val¹ to Ala¹⁷² was generated as previously described (14). WT sFcRI and mutants W130A and Y131A were generated by PCR from pKC3 templates that already contained the mutants but in a membrane-expressible form (30) using the primers HT-11 (forward) 5'-GGCGTGGAATTCGTCCCTCAGAAACC-3' and GM-3 (reverse) 5'-GTACTTGAATTCCTAAGCTTTTATTACAGTAATGTTGGGGG-3', which included EcoRI restriction sites (shown in bold) and stop codon (shown in italics). For the generation of R15A and F17A mutations, we used the splice overlap extension method (33) and the following primers alongside HT-11 and GM-3: R15A (forward) 5'-CCATGGAATGCAATATTTAAAGG-3', (reverse) 5'-CCTTTAAATATTGCATTCCATGG-3', F17A (forward) 5'-GAATAGAATAGCAAAAGGAGAG-3', and (reverse) 5'-CTCTCCTTTTGCTATTCTATTC-3'; mutated sequences are underlined.

Constructs were generated and amplified using standard PCR methods and they were then cloned into the EcoRI site of the Pichia pastoris expression vector pPIC9 (Invitrogen, Groningen, The Netherlands) in frame with the Saccharomyces cerevisiae -mating factor prepropeptide, which targets the expressed sFcRI to the secretory pathway (34). This peptide is later cleaved by the proteases KEX2 and STE13, although a residual peptide of Tyr-Val-Glu-Phe, the latter two residues formed from the introduced EcoRI site, is retained at the N terminus of all the sFcRI products expressed in P. pastoris. Integrity and orientation of the constructs were confirmed by PCR-based automated sequencing (ABI Prism 377; Applied Biosystems, Foster City, CA) of both strands using the pPIC9 primers (forward) 5'-CCAACAGCACAAATAACG-3' and (reverse) 5'-GACATCCTCTTGATTAG-3', which map to regions 100 bp on either side of the multiple cloning site.

Expression and purification of the sFcRI proteins

Mammalian WT and K117D were expressed and purified as previously described (29). Expression of the P. pastoris clones largely followed the manufacturer’s guidelines (Invitrogen). Briefly, 10 µg of purified plasmid DNA (Qiagen, Valencia, CA) was electroporated (0.2-cm cuvette, 2500 V, 25 µF, 400 Ohms, 10 ms) into the P. pastoris strain SMD1168. To this was added 1 ml of sorbitol (1 M; Sigma, Poole, Dorset, U.K.) and cells were spread on histidine-free plates and left at 30°C for 2–3 days by which time resistant colonies were clearly visible. Approximately 100 colonies were picked and spread in parallel on either dextrose- or methanol-containing plates and they were left for an additional 2–3 days to grow. Around 5–10% of the clones were slow growing on methanol plates, indicating that in these clones homologous recombination of the pPIC9 vector and insert had occurred around the AOX1 yeast gene. These slow-growing colonies were selected for expansion and small-scale sFcRI production analysis.

Clones were expanded overnight at 30°C with constant shaking in growth medium (buffered glycerol-complex medium (BMGY; 10 ml) consisting of 1% yeast extract, 2% peptone, 1.34% yeast nitrogen base (all from Difco, Cowley, Oxfordshire, U.K.), 100 mM potassium phosphate (pH 6.0), 4 x 10^-5 M biotin, and 1% glycerol (all from Sigma-Aldrich, St. Louis, MO). Cells were then pelleted (3000 x g for 5 min) and resuspended in buffered methanol-complex medium growth medium (3 ml) where 0.5% methanol replaced the glycerol. Cultures were induced for 3 days with methanol being added on a daily basis to compensate for evaporation. Cells were then sedimented and the media was assayed for sFcRI by ELISA.

Screening for expressing clones and scale-up production

An ELISA was developed for assaying production levels of sFcRI. Anti-FcRI 2-specific mAb 15.1 (a kind gift of Dr. J.-P. Kinet, Harvard Medical School, Boston, MA) was coupled to 96-well plates (Maxisorp; Life Technologies, Paisley, U.K.) at 2 µg/ml for 1 h at 37°C. Following blocking, dilutions of the supernatants were added and left for an additional 1 h at 37°C. The plate was washed extensively before the addition of the anti-FcRI 1-specific mAb 3B4 conjugated with HRP (2 µg/ml). Following incubation with tetramethylbenzidine (Sigma-Aldrich), substrate OD₄₅₀ values were taken and compared against a standard curve constructed using mammalian WT sFcRI (mWT). Typical expression levels were 2–4 µg/ml.

High-expressing clones were selected for scale-up production following the manufacturer’s guidelines (Invitrogen). A single yeast colony was used to inoculate 10 ml of BMGY medium and was grown overnight. This culture was used to seed a larger 2-L BMGY volume contained in a 5-L conical flask, which was again incubated at 30°C overnight. Yeast cells were harvested by centrifugation (3000 x g for 10 min), the supernatant was discarded, and cells were resuspended in 250 ml of buffered methanol-complex medium. Again, induction was allowed to continue for 3 days with daily addition of methanol to 0.5%. Yeast cells were sedimented as described above, and the supernatant containing sFcRI was further clarified by another centrifugation (8,000 x g for 30 min) followed by filtration to 0.45 µm.

Purification of sFcRI

Purification of mammalian expressed material was as reported previously (29). A similar method using the anti-FcRI mAb 3B4 was used for purification of sFcRI proteins expressed in P. pastoris (pWT) with some modifications. Briefly, the filtered yeast supernatant was partially purified by 66% ammonium sulfate precipitation for 30 min on ice followed by centrifugation (8,000 x g for 30 min). The protein pellet was reconstituted in 50 ml of PBS and was then extensively dialyzed against PBS. The protein solution was supplemented with BSA (0.5%) and leupeptin (2 µg/ml; both from Sigma-Aldrich) and was then filtered to 0.45 µm before recirculation on a 3B4-conjugated Sepharose column. After extensive washing with PBS, sFcRI was eluted from the column using 0.1 M glycine (pH 2.5). The eluate was neutralized with Tris (1 M), was concentrated to 1 ml, and was then dialyzed into PBS. Proteins were quantified by evaluating absorbance at 280 nm using an extinction coefficient for sFcRI of 2.56 = 1 mg/ml. This value was adapted accordingly in receptor mutants where replacement of aromatic residues occurred. Protein purity was determined by 12% SDS-PAGE (35) under nonreducing conditions and staining with Coomassie brilliant blue and Western blotting using both Abs 3B4 and 15.1.

Human and mouse IgE and their specific Ags

Human anti-4-hydroxy-3-nitro-5-iodo-phenylacetyl (NIP)-IgE (36) was obtained from the cell line JW8 (European Cell Culture Collection no. 87080706). Supernatants from the cell line were recirculated over an IgG4-Fc-(sFcRI)₂ (37) Sepharose column. Following extensive washing with PBS, NIP-IgE was eluted with glycine (0.1 M, pH 2.5), neutralized with Tris (1 M), concentrated, and dialyzed against PBS. In degranulation assays using NIP-IgE, the antigenic stimulus used was NIP-BSA, which was prepared by conjugating BSA with NIP-cap-O-succinimide (Cambridge Research Biochemicals, Northwich, Cheshire, U.K.). Mouse anti-DNP-IgE and DNP-human serum albumin (HSA) were both purchased from Sigma-Aldrich and were used without further purification.

CD spectroscopy

Circular dichroism (CD) measurements were performed on a Jobin-Yvon CD6 spectrometer (Longjumeau, France). WT and mutant proteins were analyzed in circular quartz cells (path length of 0.05 cm) at concentration ranges between 250 and 650 µg/ml in 140 mM sodium perchlorate and 10 mM sodium phosphate (pH 7.2) at 20°C. Measurements represent an average of four repeated scans in steps of 0.2 nm with an integration time of 1 s. Buffer-only controls were used as blanks and were subtracted from all measurements. The spectrophotometer was calibrated for wavelength and ellipticity () using d-10-camphor-sulfonic acid. The units of are M^-1 cm^-1/mean residue weight.

SPR analysis of binding

All surface plasmon resonance (SPR) measurements were conducted using a BIAcore 1000 (Pharmacia Biacore, Stevenage, Hertfordshire, U.K.). Extensive discussion of the SPR methodology and data analysis relating to FcRI and IgE interactions has been previously described (29, 38). Briefly, purified sFcRI was coupled to a BIAcore CM5 sensor chip using an aldehyde coupling technique, which uses protein carbohydrate to form stable linkage between the chip and protein. sFcRI immobilized through the more standard amine coupling methodologies was found to bind IgE poorly (data not shown). Runs were conducted in HEPES-buffered saline (10 mM HEPES, 150 mM NaCl, 3.4 mM EDTA, and 0.005% surfactant P-20; Pharmacia Biacore) at a flow rate of 5 µl/min, with 750-1000 resonance units of immobilized protein. We have previously tested a range of immobilization levels with this flow rate and have shown that mass transport effects are only detectable at higher receptor densities (J. M. McDonnell, unpublished data).

Human or mouse IgE (15–250 nM) was injected, and binding and then dissociation of bound material was monitored before the surface was regenerated by three successive injections (60 s) of glycine (0.1 M, pH 2.5). Data were analyzed using BIAevaluation (Version 3.0; Pharmacia Biacore) with global monophasic fitting. In previous studies, we have shown that a biphasic analysis of this interaction is justified, and can provide information about binding mechanism (29, 38). However, because monophasic K_a values show the same trends (38), to simplify comparison of the effects of mutations in the mouse and human systems we report only these values in this study. Nonspecific binding, generated by passing IgE over a flow cell without any protein coupled, was subtracted from experimental data before kinetic evaluation.

Degranulation assays

Primary basophil cells. Human basophils (purity 1%) were prepared by dextran sedimentation and washed twice in a PIPES-buffered saline (25 mM PIPES, 110 mM NaCl, 5 mM KCl, and 0.1% glucose, pH 7.3) containing 0.003% HSA but without Ca²⁺ and Mg²⁺ (39). Cells were then incubated with various concentrations of sFcRI and human anti-NIP-IgE (5 nM) for 1 h at 37°C. The cells were washed and resuspended in buffer containing both 1 mM Ca²⁺ and Mg²⁺ and they were then challenged with NIP-BSA (10 ng/ml) for an additional 40 min (37°C) and histamine release was measured using an automated fluorometric analyzer. The release of histamine was expressed as a percentage of total cellular histamine. Spontaneous histamine release (occurring in the absence of NIP-BSA) was typically 3% and was subtracted from all other values. Control histamine release (obtained in activated samples without the presence of sFcRI) was in the range of 30–50% depending on the blood donor, and inhibition of secretion data are expressed as a percentage of this control release.

RBL cells. RBL-2H3 cells were cultured in Eagle’s MEM (EMEM) with 10% FBS, L-glutamine (2 mM), penicillin (10 U/ml), and streptomycin (1 µg/ml; all from Life Technologies). On the day before experimentation, RBL cells were removed from the culture flasks using Versene (Life Technologies) and were seeded into 96-well plates (7 x 10⁴ cells/well) and left overnight to attach. Wells were washed with EMEM and were then incubated (1 h at 37°C) with varying concentrations of sFcRI and mouse anti-DNP-IgE (0.1 nM) made up in EMEM but with 2% FBS. Cells were washed with PIPES buffer (25 mM PIPES, 120 mM NaCl, 5 mM KCl, 5.6 mM glucose, 1 mM CaCl₂, 0.4 mM MgCl₂, 0.1% BSA, pH 7.3) and were then activated by adding DNP-HSA (20 ng/ml; 40 min; 37°C) in the above PIPES buffer (200 µl). Samples (150 µl) were removed and analyzed for -hexosaminidase release as previously described (16). Results are expressed as for the basophil experiments described above with spontaneous -hexosaminidase release being typically 1–2%, and control release being in the range of 20–25%.

Molecular modeling

Structures of FcRI (18) and its complex with the C3 and C4 domains of human IgE Fc (17) were viewed using INSIGHT II (Molecular Simulations, Waltham, MA). A homology model of the corresponding Fc fragment of mouse IgE, based upon the structure of the human IgE Fc in the complex with FcRI, was generated using SWISS-MODEL (40).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

 
Expression of sFRI in mammalian and yeast systems

We have previously expressed and purified both WT and mutant sFcRI from a mammalian expression system (mouse myeloma cell line NS0) generating clones that secrete in the range of 2–5 mg protein/L culture supernatant (29). To facilitate the more rapid production of sFcRI mutants for this and future studies, we have now used the yeast P. pastoris for recombinant protein expression. Using this system, we obtained expression levels similar to those using NS0 cells, and affinity purification was almost as straightforward as the mammalian system, with no indication of degradation of either product or purifying Ab. Products of WT sFcRI produced in NS0 cells (mWT) and in P. pastoris (pWT) were compared as a means of testing the suitability of the latter system for our analysis. The mutant K117D, which was generated in NS0 cells and included in a previous study (29), and four new soluble mutants produced in P. pastoris (W130A, Y131A, R15A, and F17A) form the basis of the present study.

As shown by SDS gel electrophoresis in Fig. 1A, following affinity purification using the anti-FcRI Ab 3B4, WT and mutant sFcRI produced in both mammalian and yeast were monodisperse and equivalent in apparent molecular mass. The yeast products exhibited a narrower range, most likely reflecting their modification with carbohydrate chains of a more homogeneous length and/or composition. There are seven N-glycosylation sites within human FcRI and, thus, the apparent molecular mass is increased from 21 kDa (core polypeptide) to 40–60 kDa (mature glycoprotein). The gel staining patterns (Fig. 1A) were confirmed by Western blotting (Fig. 1B) using mAb 15.1 (mWT, K117D, pWT, W130A, and Y131A) and mAb 3B4 (pWT, R15A, and F17A), reinforcing the general absence of aggregated or degraded material and demonstrating that epitopes within the specific domains where the mutations had occurred were not disrupted by the introduction of these single residue changes.

View larger version (71K): [in this window] [in a new window]   FIGURE 1. SDS-PAGE analysis of purified sFcRI. A, Five micrograms of each purified sample was loaded per lane and the gel was stained with Coomassie blue. Lane 1, mWT; lane 2, K117D; lane 3, pWT; lane 4, W130A; lane 5, Y131A; lane 6, R15A; lane 7, F17A. The molecular mass markers (Rainbow high-m.w.; Amersham) are shown in kilodaltons. B, Corresponding Western blot. Three hundred nanograms of each purified sFcRI was electrophoresed and transferred to nitrocellulose, and samples 1–5 were probed with the 2-specific mAb 15.1, and samples 6 and 7 were probed with the 1-specific mAb 3B4.

View larger version (13K): [in this window] [in a new window]   FIGURE 2. Loss of mAb 15.1 binding in sFcRI mutants R15A and F17A as determined by Western blot. Duplicate 300-ng samples of sFcRI pWT (lane 1), R15A (lane 2), and F17A (lane 3) were analyzed by SDS-PAGE (12%), transferred to nitrocellulose, and probed with either the 1-specific mAb 3B4 or the 2-specific mAb 15.1.

An advantage of the expression of FcRI as a soluble protein is that it permits assessment of changes in secondary structure induced by mutagenesis using CD. Spectra obtained for both yeast- and mammalian-derived sFcRI (data not shown) exhibited features similar to those previously described (16, 29, 41). The highly characteristic positive dichroism peak at 230 nm is seen for all seven species and probably reflects the high content of aromatic residues and the two intrachain disulfide bonds. Equally, the characteristic feature of the sheet structure of Ig and Ig-like domains, the negative peak 215 nm, is present in all spectra (data not shown).

Slight differences in the size and, to a lesser extent, position of the positive and negative peaks were observed between the mWT, pWT, and mutant FcRI proteins. These differences between the proteins may be due to the short N-terminal peptide (Tyr-Val-Glu-Phe) that is added to the P. pastoris material as part of the expression procedure, and/or may reflect glycosylation differences. Moreover, replacement of aromatic residues occurs in several of the mutants and this will be expected to affect the CD spectra. As might be predicted from the FcRI crystal structure, we conclude that there is no major unfolding of the structure induced by any of the point mutations, although we cannot exclude minor perturbations that could influence binding activity.

Binding kinetics using SPR

Following immobilization of sFcRI proteins, the binding kinetics of both human and mouse IgE were measured using SPR. Representative sensorgrams of typical experiments are shown in Fig. 3. The traces show the binding of various concentrations of IgE (12.5–250 nM) to immobilized sFcRI in real time, where 150–500 s represents the association phase, and 500–800 s represents the dissociation phase.

View larger version (34K): [in this window] [in a new window]   FIGURE 3. Binding of human and mouse IgE to WT and mutant sFcRI as determined by SPR. Representative sensorgrams are displayed from five concentrations of IgE (15–250 nM), and global monophasic fitting was used to generate the kinetic values shown in Table I.

Comparison of WT FcRI proteins

Results shown in Fig. 3, rows 1 and 2, demonstrate that both mWT and pWT sFcRI interact with human and mouse IgE in a very similar manner. This is confirmed by kinetic analysis of the data, which shows that both mWT and pWT sFcRI have closely matched values for both association and dissociation rate constants (k_a and k_d, respectively) and resultant affinity constant (K_a; Table I). This indicates that neither the addition of the N-terminal leader peptide (Tyr-Val-Glu-Phe) introduced by the yeast expression strategy, the C-terminal truncation of the yeast product by four residues compared with the mammalian protein, nor the differences in glycosylation between P. pastoris and NS0 cells has any discernable effect on the affinity for either human or mouse IgE. The lack of any effect of differential glycosylation upon binding is not surprising because bacterially expressed receptor or mammalian sFcRI clones grown in the presence of tunicamycin still bind IgE with high affinity, demonstrating that sugar moieties within sFcRI are not directly involved in IgE binding (22, 42, 43).

View this table: [in this window] [in a new window]   Table I. Summary of kinetic data for the interactions of WT and mutant sFcRI with human or mouse IgE as determined by SPR1

Effect of mutations within the 2 domain of FcRI

Previous mutagenesis studies, largely in cell-based assays, and the structure determination of the FcRI/IgE complex (17) have demonstrated the importance of the K117, W130, and Y131 residues in the interaction with human IgE. Using sFcRI mutants with native-like structures as assessed by CD, we wished to determine the individual contribution of each of these residues to the affinity of interaction with both human and mouse IgE. In our previous work, we identified K117D as a mutation that dramatically reduced the binding of sFcRI to human IgE (29). In the present study, we have confirmed this result, demonstrating that this mutation reduces the affinity for human IgE 200-fold, principally due to an enhanced dissociation rate (Fig. 3 and Table I). Moreover, it appears that this mutation has an even greater effect on the binding of mouse IgE, as the K117D/mouse IgE interaction was so weak that no meaningful kinetic parameters could be determined (Fig. 3, row 3, and Table I). The other two mutations within 2 behave quite differently. W130A decreased the affinity for human IgE 25-fold but actually increased, by 5-fold, the affinity for mouse IgE (Fig. 3, row 4, and Table I). Y131A had an even greater effect than W130A, reducing the binding affinity for human IgE 300-fold, with the dramatically increased dissociation rate (>500-fold) as the principal factor. But again, this decreased affinity was not reflected in the binding of mouse IgE, for which only a comparatively small (6-fold) increase in dissociation rate constant, and similar decrease in affinity, was measured (Fig. 3, row 5, and Table I). Thus, we conclude that mutations in human sFcRI do not have similar effects on binding to human and mouse IgE.

Effect of mutations within the 1 domain of FcRI

Both R15 and F17 have been highlighted as residues that make interdomain contacts important for receptor stability (18, 32). Mutants R15A and F17A produced 4- and 2-fold reductions, respectively, in the total amount of both human and mouse IgE that was able to bind within the association period compared with pWT sFcRI (Fig. 3, rows 5 and 6). However, neither the rates of association nor dissociation were greatly affected, leading to K_a values not very different from pWT sFcRI (Table I). These results suggest that a different mechanism is responsible for the ablation of binding in the 1 mutants compared with the 2 mutants, and for this reason we do not include the R15A and F17A mutants in the affinity ranking described below.

Inhibition of passive sensitization of IgE effector cells

sFcRI- and receptor-derived peptides, from a therapeutic perspective, would have to block IgE binding to cell-bound FcRI, and so we have measured the ability of the mutants to do this using human basophils and human IgE, and RBL cells and mouse IgE, to represent the human system and animal model, respectively. Mouse IgE is able to bind mouse, rat, and human FcRI, whereas human IgE can only interact with human FcRI (4). We have used the rat cell line primarily for practical reasons due to its relative ease of culture and adherence properties that facilitate multiwell plate assay. However, the FcRI extracellular domain sequence identity is much stronger between rat and mouse (71%) compared with human and mouse (51%; Ref. 44). Hence, mouse IgE/mouse FcRI binding, as it occurs in a mouse in vivo allergy model, is likely to be more closely modeled by the mouse IgE/rat FcRI system. Efficacy of the sFcRI proteins was assessed by their ability to inhibit IgE-dependent, Ag-induced cellular degranulation.

The results obtained yield the same rank order for the mutants as seen by SPR. mWT and pWT sFcRI were equally effective in attenuating the secretion of histamine release from human basophils (IC₅₀ 3 nM; Fig. 4). Y131A (IC₅₀ > 3000 nM) was clearly of lower affinity for human IgE in this assay compared with K117D (IC₅₀ = 300 nM; Fig. 4), which was in turn lower than that of W130A (IC₅₀ = 70 nM), consistent with data obtained by SPR. The fact that there is not always exact quantitative agreement (for example, the Y131A mutant is much less active in the functional assay than would be predicted from the SPR data) is not unexpected. We have previously shown that K_a values determined by SPR are about 10-fold lower than cell binding assays. This may be due to a number of factors, including the concentration conditions of the experiment and the fact that there are other components of the receptor (- and -chains) on the cell membrane. Furthermore, although the degranulation assay reflects cell binding, additional downstream events are also involved. Nevertheless, we have found a common rank order of affinity/potency for the human sFcRI WT and 2 mutant receptor protein interactions with human IgE in the two assays: mWT = pWT > W130A > K117D > Y131A (Fig. 4 and Table I).

View larger version (24K): [in this window] [in a new window]   FIGURE 4. Inhibition of passive sensitization of human basophils by WT and 2-mutated sFcRI. Basophils were sensitized with human NIP-IgE (1 µg/ml) with or without a variable concentration of sFcRI for 1 h before activation with NIP-BSA (10 ng/ml) for an additional 40 min. Released histamine was measured and expressed as a percentage of total histamine content. Inhibition of histamine secretion was then expressed as a percentage of the release occurring in the absence of any sFcRI (typically 30–50%).

View larger version (21K): [in this window] [in a new window]   FIGURE 5. Inhibition of passive sensitization of RBL cells by WT and 2-mutated sFcRI (A) and WT and 1 mutated proteins (B). RBL cells were incubated with mouse DNP-IgE (0.1 nM) in the absence or presence of a variable amount of sFcRI for 1 h before activation with DNP-HSA (20 ng/ml) for an additional 40 min. Released -hexosaminidase was quantified and expressed as a percentage of total cellular enzyme. Inhibition of mediator release was then expressed as a percentage of the release occurring in the absence of any sFcRI (typically 20–25%).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

We have examined the effects of four new, human sFcRI mutant proteins, shown in this study to be structurally native, on the binding to human and mouse IgE, and we have compared these with WT receptor and a previously described mutant K117D (29). IgE from both species was used, as we wished to determine whether certain key residues within human FcRI were common to the high-affinity complex formed with both human and mouse IgE. Evaluating the reliability of extrapolation from the mouse to the human system has direct implications for both the screening and in vivo testing of sFcRI mutants with elevated affinity, or receptor-derived peptides, as potential therapeutic agents. Point mutations were initially selected on the basis of our previously published human FcRI model (16), but their roles in interdomain contact (1 mutants) and direct interaction with IgE (2 mutants) are now definitively demonstrated in the recent crystal structures of sFcRI and the sFcRI/IgE complex (17, 18).

Both 1 mutants (R15A and F17A) reduced the ability of sFcRI to bind both human and mouse IgE, as determined by both SPR and cellular assays. However, unlike the 2 mutants (discussed below), no obvious change in the dissociation rate was observed nor was the association rate greatly affected, yet the total amount of IgE binding was greatly reduced compared with WT. Structural changes induced by mutagenesis at these residues are implied by the reduced binding of mAb 15.1, results paralleled by FACS analysis using cell membrane-expressed forms of R15A and F17A (32). This is significant, as the 2 domain-specific Ab is competitive with IgE for FcRI binding, and its epitope has recently been mapped to the C strand region, now clearly identified as a site of IgE interaction (48). Thus, these results support previous work that showed that modifications in the A strand of the 1 domain affect the overall receptor structure (21) and affect IgE binding, although R15 and F17 are not contact residues (32).

Confirmation of this interpretation comes from the sFcRI crystal structure (18), which shows that both residues are buried at the 1/2 interface. Replacement of these residues by alanine would be expected to destabilize the intradomain hydrophobic core, and to account for the observed reduction in mAb binding to 2. It may be that the relative disposition of the two domains is affected, without significantly unfolding either individually, as the CD spectra of these alanine mutants provide no evidence of loss of structure or unfolding. Thus, destabilization of the 1/2 interface in the R15A and F17A mutants may result in either the 2 contact residues assuming the correct configuration or the 1 and 2 domains displaying the appropriate relative disposition for IgE binding in only a fraction of the molecules at any one time. This could explain our SPR data, which show that the 1 domain mutants decreased the total amount of IgE binding without greatly changing the kinetics or overall affinity of the interaction. Such an explanation is also consistent with our finding that these mutations similarly affect binding to both human and mouse IgE.

The 2 mutant K117D has been shown to dramatically increase the dissociation rate of human IgE from sFcRI as measured by SPR (29). We have confirmed and extended these data using both SPR and cellular assays, and we have also shown that this mutation not only affects human IgE binding, but indeed has an even more pronounced effect on mouse IgE binding. By SPR, binding to human IgE was reduced 200-fold, and this was reflected almost exactly in the human basophil assay (Fig. 4). Binding of this mutant to mouse IgE was virtually undetectable by SPR, placing an upper limit of the order of 10⁵ M^-1 on the K_a value. However, in the RBL/mouse IgE assay, a 10,000-fold reduction in efficacy was recorded (Fig. 5), which is entirely consistent with the SPR result. The crystal structure of the human sFcRI/human IgE complex confirms that K117 plays a critical role in the interface (contributing to one of the two subsites, termed "site 1"; Ref. 17), forming a salt bridge with D362 of the C3 domain. Clearly, the reversal of charge in the K117D mutant would actively disrupt this interaction. D362 is conserved in mouse IgE, as are residues in its immediate vicinity, but at positions 364 and 365, the human sequence is Ala-Pro, whereas it is Glu-Ser in mouse. This must alter the polypeptide chain backbone conformation (at least at position 365, as shown by modeling), and mean that the position of D362 is slightly different in mouse relative to human IgE. This could account for the differential effect of the K117D mutation upon binding to human and mouse IgE. More significantly, the negatively charged residue, E364, in mouse IgE (in place of the neutral A364 of human IgE), and a similar substitution of the negatively charged E332 in mouse IgE (in place of the neutral N332 of human IgE), both lie within 6 Å of the side chain of residue 117. The considerably enhanced negative charge distribution at the surface of mouse C3 in the vicinity of K117D is shown in Fig. 6, where "K" indicates the location of the lysine side chain from the receptor. Thus, the conserved aspartic acid residue 362 (red patch underneath the K in each panel of Fig. 6) is flanked by the negative charges of glutamic acid residues 332 (below) and 364 (to the right) in mouse C3. This may well account for the fact that binding of the K117D (positive to negatively charged) mutant to mouse IgE is reduced by four orders of magnitude and is virtually undetectable by SPR.

View larger version (42K): [in this window] [in a new window]   FIGURE 6. The molecular surfaces of homologous regions of human and mouse C3 domains. The former is taken from the crystal structure of the complex with human FcRI (17 ) and is centered on the point of interaction with residue K117 of the receptor (indicated as K). The mouse C3 surface is calculated from a model based upon the human C3 structure. Red indicates negative surface charge and blue indicates positive surface charge. This figure was generated by GRASP.

The human sFcRI/human IgE crystal structure rationalizes the role of these two residues, which contribute to the same subsite as K117 ("site 1"; Ref. 17). Y131 projects into a pocket in C3, hydrogen bonding to H424, and packing between A364 and R334 (Fig. 7). W130 lies alongside Y131 and makes tenuous contact with the same residues H424, A364, and R334; it is peripheral to the interaction and predominantly exposed, even in the complex. It is easy to see why the Y131A mutation has a more profound effect upon binding to human IgE than W130A because the former removes a hydrogen bond and leaves a cavity at the interface, whereas the latter merely results in the loss of a small surface area of hydrophobic interaction. However, Y131A has a much smaller effect upon binding to mouse IgE than to human IgE—a 6-fold reduction in K_a compared with a 300-fold reduction. In mouse IgE, critical contact residues for Y131, namely H424 and A364, are Asp and Glu, respectively, changing the nature of the interacting surface of C3 considerably (Fig. 7), and in particular, the surface charge (Fig. 6). These differences may account in part for the (7-fold) lower binding affinity of human FcRI for mouse IgE than human IgE (16). Certainly they can account for the fact that the Y131A mutation has a greater effect upon human IgE binding than mouse IgE binding, because for the latter, there is no hydrogen bond to lose, and the C3 pocket is less hydrophobic (Fig. 7).

View larger version (18K): [in this window] [in a new window]   FIGURE 7. A, Details of the interaction between human FcRI (blue, indicating residues Trp130 and Tyr131) and human IgE Fc C3 domain (green) taken from the crystal structure of the complex (17 ). The hydrogen bond between Tyr131 and His424 is indicated. B, The homologous interaction between human FcRI (blue) and a model of mouse IgE Fc C3 domain (red, with residue numbers following the human sequence for convenient comparison) modeled upon the structure of the human complex. The significance of each of the residues that differ between human and mouse IgE is discussed in the text.

Inspection of the mouse, rat, and human receptor sequences, together with the crystal structure of the complex, offers an explanation for the well-known observation that although both rodent IgEs bind to human receptor (albeit with reduced affinity as discussed above), the rodent receptors do not bind to human IgE (49, 50). However, this lack of cross-reactivity appears to have a different explanation for the two species. Mouse receptor, uniquely among the known sequences, has a deletion at position 135 and a proline at 134; this must substantially alter the main chain conformation in this region immediately adjacent to contact residues such as Tyr¹³¹. Although this could clearly account for the nonreactivity of mouse receptor for human IgE, the rat sequence is highly homologous to human receptor in this region. (Trp¹³⁰ is replaced by serine in rat receptor, but this is unlikely to account for the lack of binding because our W130A mutation only reduces 25-fold the affinity for human IgE, as discussed above.) However, rat receptor has a lysine at position 157, where the human and mouse sequences have the neutral glutamine. In the crystal structure of the human complex, Gln¹⁵⁷ packs between Asn³³² and Arg³³⁴ on the C3 domain of IgE. Arg³³⁴ is conserved across all known IgE sequences, and although Lys¹⁵⁷ in the rat receptor would interact unfavorably with Arg³³⁴, mouse and rat IgE have a glutamic acid at position 332 with which a salt bridge could be formed. Human IgE does not have this neutralizing negative charge, which may account for the nonreactivity.

In conclusion, we have found that mutagenesis of residues in the 1 domain of FcRI at the interface with 2 can significantly reduce the fraction of active receptor molecules. Thus, targeting of this region may be a means to inhibit IgE binding. We have also found that mutagenesis of three contact residues in the IgE binding site in the 2 domain (K117, W130, and Y131) affects human IgE binding as expected, and we show quantitatively that Y131 contributes more to the binding energy of the FcRI/IgE complex than W130. We have previously described a peptide derived from the C-C' region of FcRI that was able to inhibit both human IgE binding to immobilized sFcRI and the sensitization of RBLs with mouse IgE (16). However, this peptide did not include K117, which lies at the N-terminal end of the C strand, nor W130 or Y131, which lie at the C-terminal end of the C' strand. Thus, an extended, C-C'-cyclized peptide incorporating these contact residues is likely to have an elevated affinity for human IgE and could, therefore, be of greater potential therapeutic utility.

Most significantly, however, we have shown that substitutions in human FcRI differentially affect binding to human and mouse IgE. The K117D mutation reduces binding to mouse IgE to a far greater degree than to human IgE, and the W130A mutation even enhances binding to mouse IgE, whereas it reduces binding to human IgE. It is particularly striking that the affinity differences recorded for these mutants (by SPR) are reflected, at least qualitatively, in the results of the cell-based assays; even the 5-fold increase in affinity of the W130A mutant is detectable in the RBL assay. Thus, unless the binding data indicate comparable affinities, assaying soluble human receptor proteins, or receptor-based peptides in a mouse allergy model, may not be a reliable guide to their performance in the human system.

	Acknowledgments

 
We thank Dr. J.-P. Kinet (Harvard Medical School) for providing anti-FcRI mAb 15.1 and Prof. F. L. Pearce (University College London, London, U.K.) for use of the histamine autoanalyzer. We also thank Dr. M. Varsani and S. Yue (University College London) for help with the basophil experiments. We also acknowledge the photographic assistance of M. Simon (Randall Centre) and J. Pfeiffer (University of New Mexico, Albuquerque, NM). The facilities of the Division of Cell Pathology, University of New Mexico, are gratefully acknowledged in the preparation of this manuscript.

	Footnotes

 
¹ This work was supported by the Medical Research Council of the United Kingdom, by grants from the National Asthma Campaign (to H.J.G. and B.J.S.) and the National Health and Medical Research Council of Australia (to P.M.H.), and by a Union Chimique Belge Institute of Allergy Traveling Fellowship (to G.A.M.).

² Current address: Division of Cell Pathology, Department of Pathology, University of New Mexico, Albuquerque, NM 87131.

³ Current address: Division of Immunology and Cell Biology, John Curtin School of Medical Research, Australian National University, Canberra Australian Capital Territory, Australia.

⁴ Current address: Celltech Therapeutics, Slough, Berkshire, U.K.

⁵ Address correspondence and reprint requests to Dr. Hannah J. Gould, Randall Centre, New Hunt’s House, King’s College London, Guy’s Campus, London SE1 1UL, U.K. E-mail address: hjg{at}noah.rai.umds.ac.uk

⁶ Abbreviations used in this paper: sFcRI, soluble FcRI; RBL, rat basophilic leukemia; WT, wild type; BMGY, buffered glycerol-complex medium; mWT, WT sFcRI expressed in a mammalian system; pWT, WT sFcRI expressed in a P. pastoris system; NIP, 4-hydroxy-3-nitro-5-iodo-phenylacetyl; CD, circular dichroism; SPR, surface plasmon resonance; EMEM, Eagle’s MEM; HSA, human serum albumin.

Received for publication June 22, 2001. Accepted for publication December 10, 2001.

	References

Top Abstract Introduction Materials and Methods Results Discussion References

 

1. Sutton, B. J., H. J. Gould. 1993. The human IgE network. Nature 366:421.[Medline]
2. Costa, J. J., P. F. Weller, S. J. Galli. 1997. The cells of the allergic response: mast cells, basophils, and eosinophils. J. Am. Med. Assoc. 278:1815.[Abstract]
3. Metzger, H.. 1992. Fc receptors. Annu. Rev. Immunol. 9:457.[Medline]
4. Kinet, J.-P.. 1999. The high-affinity IgE receptor (FcRI): from physiology to pathology. Annu. Rev. Immunol. 17:931.[Medline]
5. Metcalfe, D. D., D. Baram, Y. A. Mekori. 1997. Mast cells. Physiol. Rev. 77:1033.[Abstract/Free Full Text]
6. Suzuki, K., T. Hirose, H. Matsuda, S. Hasegawa, K. Okumura, C. Ra. 1998. The Fc receptor (FcR) subunit is essential for IgE-binding activity of cell-surface expressed chimeric receptor molecules constructed from human high-affinity IgE receptor (FcRI) and FcR subunits. Mol. Immunol. 35:259.[Medline]
7. Miller, K. L., A.-M. Duchemin, C. L. Anderson. 1996. A novel role for the Fc receptor subunit: enhancement of FcR ligand affinity. J. Exp. Med. 183:2227.[Abstract/Free Full Text]
8. Heusser, C., P. Jardieu. 1997. Therapeutic potential of anti-IgE antibodies. Curr. O