Immunoglobulin E-binding Site in Fc, Receptor (Fc,RII/CD23

T H E JOURNAL OF BIOLOGICAL CHEMISTRY 8 1992 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 267, No. 1, Issue of January 5 , Printed m U.S A. pp. 185-191, 1992

Immunoglobulin E-binding Site in Fc, Receptor (Fc,RII/CD23) Identified by Homolog-scanning Mutagenesis*

(Received for publication, June 18, 1991)

Bernhard BettlerSP, Gemma Texido, Silvio Raggini, Doris Ruegg, and Hans Hofstettery From the Department of Biotechnology, Pharmaceutical Division, Ciba-Geigy Ltd., CH-4002 Basel, Switzerland, the $Molecular Neurobiology Laboratory, The Salk Znstitute for Biological Studies, San Diego, California 92186-5800 and 1Ciba-Geigy Ltd., 665 Takarazukn, Japan

The IgE-binding site of the human low-affinity receptor for IgE (Fc,RII/CD23) has previously been mapped to the extracellular domain between amino acid residues 160 and 287. We now have investigated which conformational epitope within this domain spec- ifies the receptor-ligand interaction. The analysis of homolog-scanning mutants expressed in mammalian cells demonstrates that amino acid side chains that affect IgE binding are located in two discontinuous segments, between residues 165-190 and 224-256. The overall structure of the chimeric binding domains, as probed with 11 conformation-sensitive monoclonal antibodies, is generally not distorted, except by replacement of residues 165-183. In this region, disruption of binding function appears to be caused by global conformational constraints on the binding site. Substi- tution and deletion mutants demonstrate that six out of eight extracellular cysteines, Cysle3, C Y S ~ ~ ~ , Cys"', CysZs9, C Y S ' ~ ~ , and Cyszsz, are necessary for IgE binding and are most likely involved in intramolecular disulfide bridges. We show that the Fc,RII domain delineated by CyslB3 and Cyszez encodes all the structural information required to form the IgE-binding site.

The human low-affinity receptor for IgE (hFc,RII)' has been implicated in the development of allergy (Delespesse et al., 1989, 1991) as well as in other isotype-specific processes, such as IgE-dependent cytotoxicity against parasites (Capron et al., 1987) and IgE-dependent antigen focusing (Kehry and Yamashita, 1989). hFc,RII is strongly expressed on B lymphocytes, but is also detected on inflammatory cells (mono- cytes, macrophages, eosinophils, and platelets) and in some disease states, on T lymphocytes.

In order to explore the pharmaceutical potential of the hFc,RII, we and others (Kikutani et al., 1986; Ludin et al., 1987; Ikuta et al., 1987) have isolated its cDNA. The receptor was found to be identical with the B-cell differentiation antigen CD23, a protein of 321 amino acid residues having an inverse membrane orientation. The extracellular region between amino acid residues 163 and 282 has sequence similarity

* 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 U.S.C. Section 1734 solely to indicate this fact.

3 To whom correspondence should be addressed. Tel.: 619-453- 4100 (ext. 176) or 619-453-9313 (nightline); Fax: 619-450-2172.

The abbreviations used are: hFc,RII, human Fc,RII; mFc,RII, murine Fc,RII; mAb, monoclonal antibody(ies); AGPR2, asialogly- coprotein receptor 2; CHO, Chinese hamster ovary; leccams, lectin cell adhesion molecules.

with C-type animal lectins, a family of calcium-dependent carbohydrate-binding proteins (Drickamer, 1988). The ecto- domain includes a total of 8 cysteine residues (Fig. 1). Four of these, Cys'", Cys2", C ~ S ' ~ ~ , and Cys"', are conserved in all C-type lectins known to date, including the recently characterized leccams or selectins (MEL14/LAM-1, ELAM-1, PDA- GEM/GMP140/CD62), a class of cellular adhesion molecules (Stoolman, 1989; Coombe and Rider, 1989; Osborn, 1990; Brandley et al., 1990), as well as NKR-P1, a signal transduc- tion molecule on natural killer cells (Giorda et al., 1990). Besides these four highly conserved cysteines, there is a pair of cysteines corresponding in hFc,RII to C ~ S ' ~ ~ and C Y S ' ~ ~ which is conserved in some, but not all, members of the lectin family (Suter et al., 1987; Patthy, 1988). Two additional cysteines, Cys16" and CysZRR, are uniquely found in hFc,RII and the mFc,RII (Bettler et al., 1989a; Gollnick et al., 1990).

Several in vitro studies in the human system have demonstrated that (i) mAb directed against hFc,RII are able to specifically inhibit the production of IgE and (ii) different IgE-binding factors derived from hFc,RII can modulate the synthesis of IgE (Delespesse et al., 1989, 1991). All anti- hFc,RII mAb known to interfere with IgE synthesis are directed against the IgE-binding domain (Bettler et al., 1989b), suggesting that the regulation of IgE synthesis by hFc,RII is dependent on the interaction between receptor and ligand. Consequently, the ligand binding site of hFc,RII represents a possible target for preventing the excessive synthesis of IgE by allergic patients.

We have started a mutational analysis with the aim of identifying the structural features that specify the binding of hFc,RII to IgE. Deletion mapping experiments have localized the IgE-binding site of hFc,RII in the COOH-terminal ecto- domain between amino acid residues 160 and 287 (Bettler et al., 198913). This domain is almost identical with the region between amino acid residues 163 and 282 that has homology with C-type lectins. Surprisingly, hFc,RII does not bind to the carbohydrate moieties of IgE, despite the striking con- gruency between IgE-binding domain and lectin homology domain (Vercelli et al., 1989).

The deletion mapping study indicated that maintenance of the native conformation of hFc,RII is critical for the binding function and that a further molecular dissection of the IgE- binding site would require a mutational strategy that avoids extensive disruption of the protein conformation, An experimental strategy which is likely to preserve folding of the analyzed protein is homolog-scanning mutagenesis (Cun- ningham et al., 1989). This approach is based on the assump- tion that homologous proteins adapt similar three-dimensional structures. In this case, the substitution of short segments by the corresponding segments from a homologous protein is expected to be accomodated without extensive

185

186 Binding Site for Immunoglobulin E in Fc, Receptor

disruption in the overall conformation. Taking advantage of the structural similarity between

hFc,RII and C-type lectins, we now have constructed a set of homolog-scanning mutants. The analysis of these mutants discloses that the IgE-binding site is complex, with residues likely to contact the ligand at two distinct locations. We also identified the cysteine residues that are critical for the binding function. We propose that the discontinuous determinants of the IgE-binding site integrate into a structural framework stabilized by intramolecular disulfide bridges involving 6 of the 8 extracellular cysteine residues.

EXPERIMENTAL PROCEDURES

Construction of Mutant hFc,RII Cysteine substitution mutants as well as mutants that have short

exchanges with homologous human AGPR2 (Spiess and Lodish, 1985) sequences were constructed by oligonucleotide-directed in uitro mutagenesis (Amersham Corp. kit). Mutant hFc,RII in which the entire lectin core domain was precisely replaced with the corresponding domain from the AGPR2 or the mFc,RII were constructed by polymerase chain reaction according to the strategy of gene splicing by overlap extension (Horton et al., 1989). Mutant cDNA were confirmed by double-strand sequencing with cDNA specific oligonucleotide primers and subsequently inserted into the CHO cell expression vector pCAL5mbDhfr (Bettler et al., 1989b) and the COS cell expression vector pBJ5 (Takebe et al., 1988).

Oligonucleotides used for in uitro mutagenesis reactions are listed below and grouped according to the type of mutation introduced.

Homolog-scanning Mutagenesis-Making use of the redundancy of the genetic code, we have designed most mutagenesis primers such as to introduce diagnostic restriction enzyme cleavage sites, thus allow- ing rapid screening for the mutant genotype. These restriction sites are indicated between parentheses.

HS-165>173: 5”GCC GAA GTA GTA GCA GGA TCC CTG GTG CTCCACCCAGTTCACAGGGCACGTGTT GCA-3’ (BamHI)

HS-176>183: 5”CGG GCA TGG ACC CAA GCT TTA CCG CTA TGTGAAAACCAGTAGCACTTCCGTTGG- 3‘ (HindIII)

HS-185>190 5”TTC CAT GTC GTC ACA GTA CTT TTC AGC CTC AGC CCA CTG CTT GGT GCC-3’ (RsaI)

HS-192>197: 5’-GTG GAT GCT GAC CAG GTG AGC GTT CTC CAG CTG ACA GGC ATA CCG GGC-3’ (PuuII)

HS-200>204: 5”GAA GTC CTG CTC CTC CCA GCT GTT GAT CACGAC CAG CTG CCC TTC-3’ (PuuII)

HS-208>212: 5’-GGT GTG GCT GGC ATG CTG CAC GAT GAA

HS-214>219: 5’-CCG AAG GCC AAT CCA GGT GTT GAA AGG

HSD-224>233: 5’-CTC CCA TCC ACC CAT TTC CAG GAT CCA

CTT CTG CTC CTC CGG GCT-3’ (-)

ATT CGT ATG CTT GGT CAG GAA-3‘ (HinfI)

TCA GAA TCT GTA AGG CCA ATC CAG GAG-

HS-224>233: 5‘-CTC CCA TCC ACC CAT TTC CAC TCT CCG G A T C C A T C A G A A T C T G T A A G G C C A A T C CAG GAG-3’ (BamHI)

HSI-238>243: 5’-CCT GGA GCC CAG TTT TTA TAG TTG TGA

3’ (BarnHI)

CGA TAG TCG GTA CCA TCC ACC CAG ATA- 3’ (KpnI)

HS-238>243: 5’-CCT GGA GCC CAG TTT TTA TAA CGA TAG

HSI-251>256: 5’-CAT CAT CAC GCA GTC CTC GGA TCC GCC GAGCTCGTGGCCGTGCCAGTTGTCGGG

TCG GTA CCA TCC ACC CAG ATA-3’ (KpnI)

CTC CCC TGG AGC CCA-3’ (BamHI/SacI) HS-251>256: 5”CAT CAT CAC GCA GTC CTC GTG GCC GTG

CCAGTTGTCGGGCTCCCCTGGAGCCCA- 3‘ (-)

HS-261>265: 5’-GTC GGT CCA GCG ACC GTC AGG CTG TAC TTC CAC GCA GTC CTC GCC-3’ ( R d )

HSD-274>279: 5’-CCG GTC GCA CAC CCA CCG GTA CAC CTG CAG GCA GAA GGC GTC GGT-3’ (PstI)

HS-274>279: 5”CCG GTC GCA CAC CCA CCG GCC GTA CAC CTG CAG GCA GAA GGC GTC GGT-3’ (PstI)

Serine substitution mutagenesis is as follows.

SS-Serl60 5’-CGT GTT GCT CAC AAA GC-3’ SS-Serl63: 5’-CTT TTC AGG GCT CGT GTT GC-3’ SS-Ser174 5’-CGA AGT AGT AGC TCT TCC G-3’ SS-Serl91: 5’-CCA TGT CGT CAG AGG CAT ACC-3’ SS-Ser259 5’-GCA TCA TCA CGC TGT CCT CG-3’ SS-Ser273: 5’-GCT TAC GGT CGC TGA AGG C-3’ SS-Ser282: 5’-GCC GGT CGC TCA CCC AGG-3’ SS-Ser288: 5’-CGG CGT GCT TGT GGC CAG C-3’

The serine double-substitution mutants ST-Ser160/288, ST-Serl63/ 174, ST-Ser191/282 and ST-Ser259/273 were constructed by double mutagenesis with combinations of the oligonucleotides described above.

Domain Swap Mutants-The oligonucleotides listed below were used for polymerase chain reaction in order to construct the lectin core replacement mutants DS-hFcfRII/AGPR2 and DS-hFc,RII/ mFc,RII. The cDNA inserts of plasmid pFc,RIIm (Bettler et al., 1989a) and pA34 (Spiess and Lodish, 1985) were used as templates. The annealing temperature used was 55 “C, polymeration was at 72 “C. The entire cDNA sequence of the mutants produced by polymerase chain reaction was determined.

DS-hFc,RII/ AGPR2: 5’-GGG CTG CAG GTC GAC CCC AAC ACA CTA

GGA-3‘ 5”GGG GGT GGG CAG GCG GCC GTC AGG GTC

TGG-3’

c c c - 3 ‘

GCG-3‘

3‘

CAC-3’

5”GGT CAC TGC AGG CAT ACC GGG CGT GGA

5”CCG GTA TGC CTG CAG TGA CCT GCA AGG

5“CCA GCC GGT CAC ACA CCC ATG CAT CCA AGT-

5”ATG GGT GTG TGA CCG GCT GGC CAC ATG

DS-hFc.RII/ mFc,RII: 5’-GGG CTG CAG GTC GAC CCC AAC ACA CTA

GGA-3‘ 5”GGG GGT GGG CAG GCG GCC GTC AGG GTC

TGG-3’ 5”CCG GTA TGC CTG CCA GCT GGA GAA CGC

ACA-3‘

ccc -3 ‘

CAC-3‘

5”CCA GCT GGC AGG CAT ACC GGG CGT GGA

5”CTG GGT GTG TGA CCG GCT GGC CAC ATG

5”CCA GCC GGT CAC ACA CCC AGC GGT ACA CCT-3‘

Deletion Mutagenesis-The deletion mutant D134-163/X287 was constructed by combining cDNA elements of the two mutants D134- 163 and X287 described previously (Bettler et al., 1989b).

Stable and Transient Transfectants of Mutant hFcJ3II in CHO and COS Cells

Dihydrofolate reductase- CHO cells (DUKX-B1) were stably transfected with mutant cDNA by the lipofection procedure (GIBCO; 4 pg of plasmid DNA/5 X lo5 cells). After selection for the dihydrofolate reductase+ phenotype, at least 100 independent colonies were pooled. COS cells were transiently transfected using DEAE-dextran. Cell surface staining of hFc,RII with mAb and rosetting with IgE- coated latex beads were performed as described (Bettler et al., 1989b). The mAb used in this study are described in Bettler et al. (1989b).

Scatchard Analysis COS cells were assayed for IgE binding 70-75 h after transfection

with plasmid DNA. The cells were washed with Dulbecco’s phosphate- buffered saline and resuspended in Dulbecco’s modified Eagle’s medium (GIBCO) containing 20% fetal calf serum (GIBCO). Aliquot8 (100 p1) of the cell suspension (lo7 cells/ml) were incubated with varying concentrations of lZ5I-IgE (4 X lo4 cpm/pg) in a total volume of 200 pl. All aliquots were made in duplicate. The mixture was incubated at 4 ‘C for 2 h and centrifuged at 10,000 rpm in Beckman Microfuge tubes containing 200 pl of 16% parafin, 84% silicone oil (DC550, Serva) for 5 min. The tubes were cut to separate the cell pellet from the cell-free medium, and both fractions were counted in a y-counter. The difference in counts/min between duplicate samples

Binding Site for Immunoglobulin E in Fc, Receptor 187

was always less than 5%. Nonspecific binding to the transfected cells was measured in the presence of a 60-fold excess of unlabeled IgE. Association constants (K.) were determined by Scatchard analysis (Scatchard, 1949).

RESULTS

Design and Analysis of hFc,RII Mutants-Deletion mutagenesis has previously demonstrated that the IgE-binding site of hFc,RII is confined to the domain between amino acid residues 160 and 287. In order to precisely characterize the elements within this domain that specify binding to IgE, we have constructed and analyzed the mutants described in Fig. 1 and Table I.

We have used human AGPR2 (Spiess and Lodish, 1985) sequence information for the design of homolog-scanning mutants. At the time this study was initiated, the AGPR2 was the protein with the highest degree of similarity to hFc,RII, sharing 40% of sequence identity in its lectin domain (residues 177-300) with the hFc,RII lectin domain (residues 163-282). All 6 cysteine residues found in this segment of hFc,RII are conserved in AGPR2, suggesting a similar three- dimensional folding of the lectin domain in these two proteins. Therefore, we systematically substituted segments (5-11 amino acid residues) divergent between hFc,RII and AGPR2 throughout this domain. This approach led to a set of 12 chimeric hFc,RII, collectively having altered 68 residues between C y P 3 and CysZ8* (Fig. l). Four mutants, HSD-224>233, HSI-230243, HSI-251>256, and HSD-274>279, have, due to the sequence divergence between hFc,RII and AGPR2, small insertions or deletions in the chimeric protein as compared with wild-type hFc,RII. To each of these mutants we have constructed an additional homolog-scannig mutant that is adjusted in the length of the replacement region (mutants HS-224>233, HS-230243, HS-251>256 and HS-274>279). Two domain swap mutants contain precise replacements of the hFc,RII lectin core domain (residues 191-282) w&h the corresponding AGPR2 or mFc,RII sequence (mutants DS- hFc,RII/AGPR2 and DS-hFc,RII/mFc,RII).

The contribution of each of the 8 extracellular cysteine residues to the protein structure was evaluated with 12 single and double cysteine-to-serine replacement mutants.

Finally, based on the results obtained, and as an extension

of our previous study, we have designed the deletion mutant A134-163/X287 that defines the minimal domain required for IgE binding.

Mutant and wild-type hFc,RII were stably expressed in CHO cells, as well as transiently expressed in COS cells. Correct protein trafficking to the cell surface was analyzed by staining the cells with mAb3-5, a mAb which recognizes a continuous epitope in hFc,RII between residue 302 and the COOH terminus (Bettler et al., 198913). Cell surface staining with mAb3-5 demonstrates that the level of membrane expression was similar for all mutant receptors and compa- rable with wild-type hFc,RII (data not shown). Cells expressing mutant protein were analyzed by rosette formation with IgE-coated latex beads (Table I and Fig. 2). As a control, no IgE binding was observed when cells were transfected with plasmid Dl-298 (Bettler et al., 1989b) lacking most of the hFc,RII cDNA insert. Furthermore, IgE-rosette formation could always be inhibited by excess IgE (1 mg/ml) or anti- hFc,RII mAb135 (15 pg/ml), whereas IgG and IgA had no influence (data not shown). Binding of IgE observed by rosette formation was confirmed by Scatchard analysis (Table I and Fig. 3; Scatchard, 1949). The KO determined for the association between IgE and recombinant wild-type hFc,RII agrees well with published values (Conrad, 1990). No IgE binding could be measured in the Scatchard analysis for all rosette- negative mutant hFc,RII.

IgE Binding Is Impaired by Homologous Replacement of Amino Acid Residues within Two Discontinuous Segments- Six hFc,RII proteins with homologous sequence substitutions, mutants HS-192>197, HS-200>204, HS-208>212, HS- 214>219, HS-261>265, and HSD-274>279, retained the ability to form IgE rosettes (Fig. 2 and Table I). In general, these chimeric hFc,RII had affinities for IgE in the range of wild- type receptor, with exception of mutant HS-192>197 that showed a lower KO value. Since homologous sequence substitution between residues 192-219 and 261-279 does not affect IgE binding, these two segments of hFc,RII are not likely to contain elements of the recognition site.

In contrast, IgE binding is impaired in six mutants (HS-

238>243, and HSI-251>256) that involve replacement of res- 165>173, HS-176>183, HS-185>190, HSD-224>233, HSI-

CP TM EC

FIG. 1. Schematic representation of mutant and wild-type hFc,RII. A scheme of wild-type receptor is given at the top. The black box depicts the transmembrane ( 2 " ) region: the NH2 terminus is exposed to the cytoplasm (CP) and the COOH terminus is extracellular ( E C ) . The open box resresents the domain homologous with animal lectins. Cysteine residues are numbered and indicated by a bold letter (C). Mutant hFc,RII having homologous sequence exchanges with AGPR2 (Spiess and Lodish, 1985) are listed below. The sequence of hFc,RII between C Y S ' ~ and CyszBB and the sequence of AGPR2 between C Y S ' ~ ~ and Cys3O0 is shown. Amino acid residues identical between hFc,RII and AGPR2 are specified by vertical lines. Segments exchanged in the homolog-scanning mutants are boxed. Mutants HS-224>233, HS-238>243, HS-251>256, and HS-274>279 were designed such as to create proteins that are optimized for sequence identity with hFc,RII. The sequence of the segments adjusted for length is given below the sequence of AGPR2. The nomenclature of mutant hFc,RII is outlined in Table I.

188 Binding Site for Immunoglobulin E in Fc, Receptor TABLE I

Effect of mutations in hFc,RII on binding to IgE and anti-hFc,RII mAb Mutants were stably expressed in CHO cells for IgE rosette formation and transiently expressed in COS cells

for the Scatchard analysis. For the IgE rosetting test the symbols are: +, mutants forming IgE rosettes; -, mutants not forming IgE rosettes. Association constants (KO) in the Scatchard analysis are in M". For the analysis of mAb binding, the symbols before the slash refer to immunoblotting experiments (Fig. 4), symbols after the slash indicate results obtained by cell surface staining. The following symbols were used +, mutant receptor binding to mAb; -, mutant not binding to mAb. Nomenclature of mutants: HS, homolog scanning; HSI, homolog scanning with insertion of amino acid residue($; HSD, homolog scanning with deletion of amino acid residue(s); DS, domain swap; SS, substitution of single cysteine residues; ST, substitution of 2 cysteine residues; D, internal deletion; X, COOH-terminal deletion. Mutated residues are numbered from the amino terminus of hFc.RII.

hFc.RII Mutant ICE Rosettes K. (IeE) MHM6 3-5 7-9 14 25 30 45 63 64 135 168

Wild-type HS-165>173 HS-176>183 HS-185>190 HS-192>197 HS-200>204 HS-208>212 HS-214>219 HSD-224>233 HS-224>233 HSI-238>243 HS-238>243 HSI-251>256 HS-251>256 HS-261>265 HSD-274>279 HS-274>279 DS-hFc,RII/mFc,RII

SS-Serl6O SS-Serl63 SS-Ser174 SS-Serl91 SS-Ser259 SS-Ser273 SS-Ser282 SS-Ser288 ST-Ser1601288 ST-Ser1631174 ST-Ser1911282 ST-Ser2591273

DS-FcfRII/hAGPR2

D134-1631X287

7 x lo7 - - -

5 x IO6 7 x 10' 7 X 107 7 X 107 - - - - - -

2 X 107 2 X 107 - -

ND" ND ND ND ND ND ND ND ND ND ND ND

5 x 10'

-

+I+ -I- -I- +/+ +I+ +I+ +I+ +I- +/+ +I+ -I+ -I- +/+ +/- -I- -I- -I- -I- -I- +/+ -I- -I- -I- -I- -I- -I- +/+ +I+ -I- -I- -I- +/+

+I+ -I- -I- +/+ +I+ -I- -I- +/+ +I+ +I+ +I+ +I+ +I+ +I+ +I+ +I+ -I- -I- -I- +/+ -I- -I- -I- -I- -I- -I- +/+ +I+ -I- -I- -I- +/+

+I+ -I- -I- -I- -I- +/+ +I+ +I+ -I- -I- -I- +/- -I- +/+ -I- -I- -I- -I- -I- +/+ -I- -I- -I- -I- -I- -I- +/+ +I+ -I- -I- -I- +/+

+I+ +I- +/- +I+ +I+ +I+ +I+ +I+ +I+ -I- -I- -I- -I- +/+ -I- -I- -I- -I- -I- +/+ -I- -I- -I- -I- -I- -I- +/+ +I+ -I- -I- -I- +/+

+I+ -I- -I- +/+ +I+ +I+ +I+ +I+ +I+ +I+ -I- -I- -I- -I- -I- -I- -I- -I- -I- +/+ -I- -I- -I- -I- -I- -I- +/+ +I+ -I- -I- -I- +/+

+I+ -I- -I- +/+ +I+ +I+ +I+ +I+ +I+ +I+ +I+ +I+ -I- -I- -I- -I- -I- -I- -I- +/+ -I- -I- -I- -I- -I- -I- +/+ +I+ -I- -I- -I- +/+

+I+ -I- -I- +/+ +I+ +I+ +I+ +I+ +I+ +I+ +I+ +I+ +I+ +/- -I- -I- -I- -I- -I- +/+ -I- -I- -I- -I- -I- -I- +/+ +I+ -I- -I- -I- +/+

+I+ -I- -I- +/+ +I+ +I+ -I- -I- +/+ +I+ +I+ -I- +/+ +I+ -I- -I- -I- -I- -I- +/+ -I- -I- -I- -I- -I-

+/+ -I-

+/+ -I- -I- -I- +/+

176

+I+ -I- -I- -I- -I- +/+ +I+ +I+ +I+ +I+ +I+ +I+ +I+ +I+ -I- -I- -I- -I- -I- +/+ -I- -I- -I- -I- -I- -I- +/+ +I+ -I- -I- -I- +/+

ND, indicates where the association constant has not been determined.

FIG. 2. IgE binding of selected hFc,RII mutants. CHO cells stably transfected with mutant-hFc,RII cDNA were analyzed by rosette formation with IgE-coated latex beads. 1, CHO cells transfected with control vector Dl-298 (Bettler et al., 198910); 2, wild-type

261>265; 7, ST-Ser160/288; 8, D134-1631X287. The nomenclature of mutant hFc,RII is outlined in Table I.

hFc,RII; 3, HS-185>195; 4, HS-192>197; 5, HS-214>219; 6, HS-

idues within two discontinuous segments of 26 (residues 165- 190) and 33 amino acids (residues 224-256) in length. Mutants HSD-224>233, HSI-238>243, and HSI-251>256, which determine the segment from residues 224-256 as being critical for IgE binding, have insertions or deletions as compared with

wild-type hFc,RII. Mutants HS-224>233, HS-238>243, and HS251>256, which contain amino acid exchanges analogous to HSD-224>233, HSI-238>243, and HSI-251>256 but are adjusted in length within the homologous replacement region, do not bind IgE either. This demonstrates that the lack of IgE reactivity in mutants with deletions or insertions is not simply caused by the absence of appropriately spaced amino acid residues. Furthermore, the two-domain swap mutants DS-hFcfRII/AGPR2 and DS-hFc,RII/mFc,RII have also lost the ability to bind IgE. Since mFc,RII binds IgE in a strictly species-specific manner (Bettler et al., 1989a) and AGPR2 does not bind IgE, this underlines that amino acid residues within the replaced lectin core (residues 191-282), specifically residues 224-256, are critical for ligand recognition.

In summary, functional analysis of homolog-scanning mutants suggests that the residues of hFc,RII involved in the binding function reside in the segments from amino acids 185-190 and 224-256.

Test of the Structural Integrity of Homolog-scanning hFc,RII Mutants-A panel of conformation-sensitive anti- hFc,RII mAb was used to assess the structural integrity of the mutated lectin domains. We argued that binding of several of these mAb, which recognize different epitopes (see below), would indicate correct overall protein folding. Cells expressing

Binding Site for Immunoglobulin E in Fc, Receptor 189

A B

Y 10 20

IgE added (nM) 30

FIG. 3. Scatchard analysis. Binding of '251-IgE to hFc,RII expressed in COS cells. Only data for wild-type hFc,RII ( A ) and mutant D134-163/X287 ( B ) are shown. The insets depict the corresponding Scatchard analysis. Each Doint remesents the mean of dudicates. The Scatchard analysis of additional mutants is summarized in Table-I.

3-5

25

6 4

135

168

176

MHMB

FIG. 4. Antigenic phenotype of mutant receptors. For immunoblot analysis, solubilized crude membrane extracts were pre- pared from the B-lymphoblastoid cell line RPMI 8866 (8866) and from CHO cells producing wild-type ( WT) or mutant hFc,RII. Elec- trophoresis of samples was carried out under nonreducing conditions. Analysis is shown for mAb3-5, mAb25, mAb64, mAb135, mAb168, mAb176, and MHM6. The immunoblot results obtained with mAb7- 9, mAbl4, mAb30, mAb45, and mAb63 are summarized in Table I. The nomenclature of mutant hFc,RII is outlined in Table I.

mutant receptors were tested for reactivity with anti-hFc,RII mAb by surface staining (Table I) and by immunoblotting of membrane extracts under non-reducing conditions (Fig. 4 and Table I).

All chimeric proteins have the expected size, as shown by immunoblotting with mAb3-5 (Fig. 4). For most homolog- scanning mutants, disruption of binding to different mAb was highly selective, indicating that the changes introduced only cause a local distortion of the native conformation. However, mutants HS-165>173 and HS-176>183, which do not bind IgE (see above), appear to affect structure more globally. These two mutant proteins are recognized by mAb45 and MHM6 but none of the other conformation-sensitive mAb, indicative of major structural alterations in the IgE-binding domain (Table I). Disruption of IgE binding in these two mutant proteins could therefore result from conformational changes distal to the site of mutation. Some mutant proteins react with mAb3O and mAb45 on immunoblots but do not bind these antibodies on the cell surface (Table I). The explanation for this may be that conformational epitopes are in flux and experimental conditions can influence them or that some epitopes are not accessible at the cell surface. As with the IgE-binding site, the epitopes of conformation-sensitive mAb appear to be discontinuous. This might indicate that the receptor is folded such that the regions constituting the IgE-binding site or a given mAb epitope are juxtaposed.

FIG. 5. Schematic representation of the proposed disulfide connectivities in the IgE-binding domain of hFc,RII. This model is based on the bridging pattern determined for three C-type animal lectins. C Y S ' ~ and CysZm are not conserved in the C-type lectin family, and our data show that they are not required for folding of the IgE-binding site. We propose that they are linked via a disulfide bridge (displayed by a dashed l ine) . Mutagenesis analysis demonstrates that IgE binding is sensitive to homologous sequence substitutions in the two segments indicated by a heauy line.

Cysteine Residues Involved in Three-dimensional Folding of the IgE-binding Domuin-Serine substitution mutants were constructed that change individually all 8 extracellular cysteine residues into serine residues. This conservative amino acid exchange is expected to prevent disulfide bridging without inducing extensive local alterations. We also have replaced pairs of cysteines hypothetically involved in disulfide bridge formation (see Fig. 5 for proposed disulfide bridges) because the free SH group left after the substitution of 1 cysteine in a disulfide bridge possibly causes the formation of aberrant disulfide linkages. These double mutants are expected to reduce the probability for inappropriate disulfide linkage thus preventing scrambling of the mutant receptor (Rusk et al., 1988).

The only individual cysteine mutant recognizing IgE is SS- Ser288 (Table I). This confirms results obtained earlier, where it was found that only CysZa could be deleted without abol- ishing binding function (Bettler et al., 1989b). However, the double-cysteine mutant ST-Ser160/288 also binds IgE, demonstrating that both Cys'"' and CysZm are, when replaced together, dispensable for folding of the IgE-binding site (Fig. 2).

Test of the Structural Integrity of Cysteine to Serine Substi- tution Mutants-Six individual cysteine to serine substitution mutants, SS-Serl63, -174, -191, -259, -273, and -282, fail to

190 Binding Site for Immunoglobulin E in Fc, Receptor

bind all conformation-sensitive mAb, indicating that these mutants have lost their conformational epitopes (Table I). By contrast, two mutants, SS-Serl6O and SS-Ser288, bind to all mAb, albeit mutant SS-Ser288 consistently weaker than mutant SS-Serl6O or wild-type hFc,RII (Table I and Fig. 4). The fact that mAb binding tolerates the substitution of C ~ S ' ' ~ confirms data obtained by deleting this residue (Bettler et al., 1989b). The result with mutant SS-Ser288 is unexpected, because we found earlier by deletion analysis that CysZ8' was indispensable for a mAb reactive conformation. The serine double mutant ST-Ser160/288 reconciles the difference in the results obtained by deleting or substituting CysZrn. This mutant reacts at wild-type level with all mAb, clearly demonstrating that Cys"j0 and Cys2=, when substituted together, are both dispensable for a wild-type conformation (Table I and Fig. 4). The result obtained with the double mutant ST- Ser160/288, when compared with single-serine substitution mutants, emphasizes that uneven alteration of cysteines can affect proper folding of recombinant proteins. The three serine double mutants ST-Ser163/174, ST-Ser191/282, and ST-Ser259/273 have lost their conformational epitopes, show- ing that CYS"~, CYS'~~, Cydgl, CyP3, and CysZR2 are absolutely necessary to establish the antigenic phenotype (Table I and Fig. 4).

The data obtained with serine substitution mutants indicate that the folding of the region between Cys"j3 and CysZR2 is crucial for the three-dimensional structure of the lectin domain, whereas the flanking C ~ S ' ' ~ and CysZR8 are irrelevant in this respect.

The Minimal Domain Required for IgE and mAb Binding- The studies discussed so far indicate that the epitope of IgE on hFc,RII is influenced by homologous replacement of residues 165-190 and 224-256. Furthermore, CYS''~ and CysZR8, flanking the lectin-homology region, are dispensable for mAb and IgE binding. Moreover, it is known that sequences NHz- terminal to C ~ S ' ' ~ and COOH-terminal of ThrZa7 are not involved in interacting with IgE (Bettler et aL, 1989b). Alto- gether, this suggests that the concommitant deletion of C ~ S ' ' ~ and CysZa8 might create a functional receptor, thus determin- ing the shortest IgE reactive truncated version of hFc,RII. Therefore, mutant D134-163/X287 was constructed. This protein has extensive internal and COOH-terminal deletions, lacking residues 135-162 and the residues carboxyl-terminal of ThrZg7, thus having both C ~ S ' ' ~ and CysZrn deleted. This truncated receptor binds to IgE (Table I and Figs. 2 and 3) and all conformation-sensitive mAb (Table I and Fig. 4), demonstrating that a correctly folded receptor is produced. Thus, the domain sufficient for IgE binding corresponds to residues 163-287 and can, most likely, be narrowed down to the region flanked by C y P 3 and Cys282, the maximal homology region between hFc,RII and C-type animal lectins (Suter et al., 1987; Patthy, 1988).

DISCUSSION

The analysis of homolog-scanning hFc,RII mutants iden- tifies two segments, residues 165-190 and 224-256, compris- ing 59 amino acids in which the binding of IgE is sensitive to sequence substitutions (Table I). For a particular mutant, a decrease in binding affinity points to residues which are important for direct contact with IgE or for correct folding of the binding site. A survey of the overall conformation in each mutant protein is therefore a prerequisite for the identification of determinants crucial for the binding function. Anti- hFc,RII mAb, which recognize the protein only in its native conformation, indicate that no extensive structural alterations have been introduced in most homolog-scanning mu-

tants (Table I and Fig. 4). The preservation of the mAb epitopes in most chimeric proteins underlines the structural relatedness between C-type lectins and hFc,RII. Only two IgE-rosetting negative mutants, HS-165>173 and HS- 176>183, exhibit dramatic reductions in binding to most mAb. In these two mutants the protein conformation appears to be disrupted outside of the engineered site (residues 165-183) and may abolish function by global effects on protein folding rather than by local effects at the binding site (Table I). Residues 165-183 are located in a region with four closely spaced cysteines. The presence of AGPR2 amino acid side chains in the chimeric protein may hinder the normal intramolecular disulfide bridging, thereby obstructing proper three-dimensional folding. The homologous replacement of amino acid residues 185-190 produces a correctly folded chimeric protein which does not bind IgE, demonstrating that at least the residues adjacent to amino acids 165-183 are involved in the interaction with IgE. Therefore, we conclude that amino acid residues 185-190, together with residues 224- 256, contain discontinuous elements of the IgE-binding site. However, our study does not provide information as to whether residues 165-183 are also part of the IgE recognition site.

Our mutational analysis does not distinguish between amino acid residues that contact IgE and residues that alter the conformation in the vicinity of the binding site. However, the identification of two short segments affecting ligand binding now provides a basis to determine the amino acid side chains interacting with IgE, e.g. by alanine-scanning mutagenesis (Cunningham and Wells, 1989).

Binding of IgE by hFc,RII is dependent on 6 (CyP3, CYS'~~, Cys"', CysZ5', CYS'~~, and CysZR2) of the 8 extracellular cysteines (Table I). The loss of function resulting from substitution of these 6 cysteines is associated with a distortion of the structure probably caused by missing and by aberrantly linked disulfide bridges. Most likely, disulfide bonds play a role in bringing the discontinuous elements critical for binding into spatial proximity. A conserved pattern of intramolecular disulfide bridges has emerged from the characterization of three members of the C-type lectin family Tetranectin (Fuhl- endorf et al., 1987), Echinoidin (Giga et al., 1987) and a protein of Megabalunus rosa (Muramato and Kamiya, 1990). The conserved bridging pattern suggests that in hFc,RII disulfide bridges are formed between cysteines 163/174, 191/282, and 259/273. It is not known whether C ~ S ' ' ~ and CysZrn, which are dispensable for IgE binding and which are exclusively found in hFc,RII and mFc,RII, form a disulfide bridge. Earlier results obtained with deletion mutants suggest that they are also linked and help to stabilize the protein (Bettler et al., 1989b). We therefore have introduced a putative disulfide bridge between C ~ S ' ' ~ and CysZRR in the model of hFc,RII presented in Fig. 5.

Deglycosylated IgE binds to hFc,RII, indicating that the receptor-ligand interaction is not dependent on carbohydrates (Vercelli et al., 1989). This and our mapping of the IgE- binding site demonstrate that the ancestral lectin domain has evolved to interact with protein residues. With the recent demonstration that hFc,RII binds to a motif in the C,3 constant region domain of IgE (Vercelli et al., 1989), the interacting surfaces of IgE and hFc,RII are now characterized.

Recently, the soluble form of hFc,RII (residues 148-321) has been described to stimulate the maturation of human CD7' T-cell precursors, as well as to promote the proliferation of CD34' immature myeloid cells (Mossalayi et al., 1990a, 1990b). The striking homology of hFc,RII with leccams or selectins, a class of hematopoetic adhesion proteins recogniz-

Binding Site for Immurwglobulin E in Fc, Receptor 191

ing carbohydrate residues, suggests that these newly discov- ered hFc,RII activities might be directed by a lectin-carbohydrate interaction. Our mutants will be useful in testing if the hFc,RII domain between residues 163-282 mediates these seemingly IgE-independent activities via a lectin-type interaction.

Acknowledgments-Many thanks to Drs. J. Bancherau, G. Deles- pesse, D. Conrad, T. Kishimoto, and J. Gordon for gift of mAb, to Dr. M. Spiess for providing plasmid pA34 and to J. Bews and Dr. E. Kilchherr for lZ5I-IgE and unlabeled IgE. We thank Drs. G. Bilbe and E. Freed for helpful comments.

REFERENCES Bettler, B., Hofstetter, H., Rao, M., Yokoyama, W. M., Kilchherr, F.,

and Conrad, D. H. (1989a) Proc. Natl. Acad. Sci. U. S. A. 86,7566- 7570

Bettler, B., Maier, R., Ruegg, D., and Hofstetter, H. (1989b) Proc. Natl. Acad. Sci. U. S. A. 86, 7118-7122

Brandley, B. K., Swiedler, S. J., and Robbins, P. W. (1990) Cell 63,

Capron, A., Dessaint, J. P., Capron, M., Ouma, J. H., and Butter-

Conrad, D. H. (1990) Annu. Rev. Zmmunol. 8,623-645 Coombe, D. R., and Rider, C. C. (1989) Zmmunol. Today 10,289-291 Cunningham, B. C., and Wells, J. A. (1989) Science 244, 1081-1085 Cunningham, B. C., Jhurani, P., Ng, P., and Wells, J. A. (1989)

Delespesse, G., Sarfati, H., and Hofstetter, H. (1989) Zmmunol. Today

Delespesse, G., Suter, U., Mossalyayi, Bettler, B., Sarfati, M., Hof- stetter, H., Kilchherr, E., Debre, P., and Dalloul, A. (1991) Adu. Zmmunol. 49, 149-191

861-863

worth, A. E. (1987) Science 238, 1065-1072

Science 243,1330-1336

10,159-164

Drickamer, K. (1988) J. Biol. Chem. 263,9557-9560 Fuhlendorf, J., Clemmensen, I., and Magnusson, S. (1987) Biochem-

Giga, Y., Ikai, A., and Takahashi, K. (1987) J. Biol. Chem. 262, istry 26,6757-6764

6197-6203

Giorda, R., Rudert, W. A, Vavassory, C., Chambers, W. H., Hiserodt, J. C., and Trucco, M. (1990) Science 249, 1298-1300

Gollnick, S. O., Troustine, M. L., Yamashita, L. C., Kehry, M. R., and Moore, K. W. (1990) J. Zmmunol. 144,1974-1982

Horton, R. M., Hunt, H. D., Ho, S. N., Pullen, J. K., and Pease, L. R. (1989) Gene (Amst.) 77,61-68

Ikuta, K., Takami, M., Kim, C. W., Honjo, T., Miyoshi, T., Tagaya, Y., Kawabe, T., and Yodoi, J. (1987) Proc. Natl. Acad. Sci. U. S. A.

Kehry, M. R., and Yamashita, L. C. (1989) Proc. Natl. Acad. Sci

Kikutani, H., Inui, S., Sato, R., Barsumian, E. L., Owaki, H., Yama- saki, K., Kaisho, T., Uchibayashi, N., Hardy, R. R., Hirano, T., Tsunasawa, S., Sakiyama, F., Suemura, M., and Kishimoto, T. (1986) Cell 47, 657-665

Ludin, C., Hofstetter, H., Sarfati, M., Levy, C. A., Suter, U., Alaimo, D., Kilchherr, E., Frost, H., and Delespesse, G. (1987) EMBO J. 6,

Mossalayi, M. D., Arock, M., Bertho, J. M., Blanc, C., Dalloul, A. H., Hofstetter, H., Sarfati, M., Delespesse, G., and Debre, P. (1990a)

Mossalayi, M. D., Lecron, J. C., Dalloul, A. H., Sarfati, M., Bertho, J. M., Hofstetter, H., Delespesse, G., and Debre, P. (199Ob) J. Exp. Med. 171,959-964

Muramato, K., and Kamiya, H. (1990) Biochim. Biophys. Acta 1039,

Osborn, L. (1990) Cell 62,3-6 Patthy, L. (1988) J. Mol. Biol. 202,689-696 Rusk, C. M., Neeper, M. P., Kuo, L.-M., Kutny, R. M., and Robb, R.

Scatchard, G. (1949) Ann. N. Y. Acad. Sci. 51,660-672 Spiess, M., and Lodish, H. F. (1985) Proc. Natl. Acad. Sci. U. S. A.

Stoolman, L. M. (1989) Cell 56, 907-910 Suter, U., Bastos, R., and Hofstetter, H. (1987) Nucleic Acids Res.

Takebe, Y., Seiki, M., Fujisawa, J.-I., Hoy, P., Yokota, K., Arai, K.-

Vercelli, D., Helm, B., Marsh, P., Padlan, E., Geha, R. S., and Gould,

84,819-823

U. S. A. 86, 7556-7560

109-114

Blood 75, 1924-1927

52-60

J. (1988) J. Zmmunol. 140, 2249-2259

82,6465-6469

15,7295-7308

I., Yoshida, M., and Arai, N. (1988) Mol. Cell. Biol. 8, 466-472

H. (1989) Nature 338,649-651

Documents

Immunoglobulin E-binding Site in Fc, Receptor (Fc,RII/CD23