Extracellular domains of the bradykinin B2 receptor involved in ligand binding and agonist sensing defined by anti-peptide antibodies.

Many of the physiological functions of bradykinin are mediated via the B2 receptor. Little is known about binding sites for bradykinin on the receptor. Therefore, antisera against peptides derived from the putative extracellular domains of the B2 receptor were raised. The antibodies strongly reacted with their corresponding antigens and cross-reacted both with the denatured and the native B2 receptor. Affinity-purified antibodies to the various extracellular domains were used to probe the contact sites between the receptor and its agonist, bradykinin or its antagonist HOE140. Antibodies to extracellular domain 3 (second loop) efficiently interfered, in a concentration-dependent manner, with agonist and antagonist binding and vice versa. Antibodies to extracellular domain 4 (third loop) blocked binding of the agonist but not of the antagonist, whereas antibodies to extracellular domains 1 and 2 or to intracellular domains failed to block ligand binding. Antibodies to ectodomain 3 competed with agonistic anti-idiotypic antibodies for B2 receptor binding. Further, affinity-purified antibodies to the amino-terminal portion of extracellular domain 3 transiently increased intracellular free Ca concentration and thus are agonists. The Ca signal was specifically blocked by the B2 antagonist HOE140. By contrast, antibodies to the carboxyl-terminal segment of extracellular domain 4 failed to trigger Ca release. The specific effects of antibodies to the amino-terminal portion of extracellular domain 3 suggest that this portion of the B2 receptor may be involved in ligand binding and in agonist function.

Many of the physiological functions of bradykinin are mediated via the B2 receptor. Little is known about binding sites for bradykinin on the receptor. Therefore, antisera against peptides derived from the putative extracellular domains of the B2 receptor were raised. The antibodies strongly reacted with their corresponding antigens and cross-reacted both with the denatured and the native B2 receptor. Affinity-purified antibodies to the various extracellular domains were used to probe the contact sites between the receptor and its agonist, bradykinin or its antagonist HOE140. Antibodies to extracellular domain 3 (second loop) efficiently interfered, in a concentration-dependent manner, with agonist and antagonist binding and vice versa. Antibodies to extracellular domain 4 (third loop) blocked binding of the agonist but not of the antagonist, whereas antibodies to extracellular domains 1 and 2 or to intracellular domains failed to block ligand binding. Antibodies to ectodomain 3 competed with agonistic anti-idiotypic antibodies for B2 receptor binding. Further, affinity-purified antibodies to the amino-terminal portion of extracellular domain 3 transiently increased intracellular free Ca 2؉ concentration and thus are agonists. The Ca 2؉ signal was specifically blocked by the B2 antagonist HOE140. By contrast, antibodies to the carboxyl-terminal segment of extracellular domain 4 failed to trigger Ca 2؉ release. The specific effects of antibodies to the amino-terminal portion of extracellular domain 3 suggest that this portion of the B2 receptor may be involved in ligand binding and in agonist function.
Physiological and pathophysiological processes are mediated by kinins and their receptors. Kinins are liberated by proteolytic cleavage of the precursor proteins kininogens (1); they decrease blood pressure, induce pain and inflammation, contract smooth muscles, and regulate ion fluxes (2). Receptors for kinins are classified pharmacologically into two major sub-types, B1 and B2 (3). The B1 receptors are triggered by carboxyl-terminally truncated kinins such as [des-Arg 10 ]kallidin, whereas bradykinin is the agonist of B2 receptors. Molecular cloning has revealed the primary structures of the B1 (4) and the B2 receptors (5) and classified them as members of the G-protein-coupled receptor family that are thought to contain seven membrane spanning ␣-helices.
The signaling pathways of the B2 receptors have been explored in some detail. The bradykinin B2 receptor is preferentially coupled to G proteins of the G␣ q subtype (6), which activate the phospholipase C-mediated cascade. This results in the hydrolysis of inositol-containing lipids, the generation of inositol phosphates, and the transient rise of the intracellular free Ca 2ϩ concentration (7). The initial increase of intracellular Ca 2ϩ is followed by Ca 2ϩ extrusion, which counteracts Ca 2ϩ influx, thereby regulating total cell calcium (8). B2-mediated release of diacylglycerol, another hydrolysis product of phospholipase C, results in the translocation of specific protein kinase C isoforms (9). The B2 receptor is also coupled to the phospholipase A2 pathway, which releases the prostaglandin precursor, arachidonic acid (10).
Although the amino acid sequence of the B2 receptor has been deduced from its cDNA and its transmembrane topology has been predicted from the corresponding hydropathy plots, the specific role of the extracellular domains in ligand binding and in signal transduction is unknown. To address this question, we have raised antibodies against peptides derived from the ectodomains of the B2 receptor and used them to probe for the function(s) of the corresponding structures. Our data show that extracellular domain 3 is involved in ligand binding and may play an essential role in communicating the agonist signal through the receptor.
Triticum vulgaris, N-acetylglucosamine, and fluorescein isothiocyanate-conjugated goat anti-rabbit immunoglobulin were from Sigma; rhodamine-conjugated donkey anti-rabbit immunoglobulin was from Jackson; Centricon 30 filters were from Amicon; keyhole limpet hemocyanin and fura-2/AM were from Calbiochem; polyvinylidene difluoride sheets were from Millipore; Affi-Gel 10 and nonfat dry milk was from Bio-Rad; MaxiSorb titer plates were from Nunc; HOE140 and 3-(4hydroxyphenyl-propyl)-HOE140 (HPP-HOE140) was from Hoechst; and gelatin was from Merck. All other chemicals were of analytical grade.
Cell Transfection and Infection with Recombinant Baculovirus-CHO cells were transfected with the rat B2 receptor cDNA (rB2CHO12/4) using the Lipofectin transfection method as described (8,9). The pVL1392 vector (kindly provided by Dr. H. Reilä nder, Frankfurt, Germany) contained the human B2 receptor cDNA such that transcription was directed to the predicted initiation site (11). Sf9 cells (2 ϫ 10 6 /ml) were infected with the wild type or the recombinant baculovirus at a multiplicity of infection of 2-5. Cells were harvested 48 -72 h after infection (12).
Cell Culture-Human foreskin fibroblasts, HF-15 (13) were grown to confluency in Dulbecco's modified Eagle's medium containing 10% (v/v) fetal calf serum for 2-3 weeks and used at passages 10 -15. Chinese hamster ovary (CHO) cells were grown in Ham's F12 medium and Sf9 cells in TC100 medium, each containing 10% fetal calf serum. The epithelial carcinoma cell line A431 was maintained in RPMI 1640 medium supplemented with 10% fetal calf serum, 1 mM sodium pyruvate, 100 units/ml penicillin and 0.1 mg/ml streptomycin. HF-15, CHO, and A431 cells were kept in a humidified 5% CO 2 , 95% air atmosphere at 37°C, and Sf9 cells were kept in a humidified air atmosphere at 27°C.
Extraction of Membrane Proteins by Triton X114 -CHO cells were washed twice with ice-cold PBS and harvested in 0.5% (v/v) of Triton X-114 in PBS containing protease inhibitors; the yield was approximately 1 mg of protein/0.5 ml of Triton X-114. After centrifugation (3 min, 4°C, 13,000 rpm) the pellet debris was discarded. The supernatant was heated to 30°C for 4 min and centrifuged (3 min, 24°C, 3,000 rpm) to cause phase separation (14). The supernatant was discarded, and the Triton X-114 phase was dissolved in ice-cold PBS. After another phase separation step, the Triton X-114 phase was diluted with 2 volumes of SDS sample buffer (15) and applied to polyacrylamide gel electrophoresis (PAGE).
Radioiodination of Peptides and Antibodies-Peptides or antibodies (1 g each) dissolved in 100 l of PBS were incubated with 2 mCi of carrier-free Na-[ 125 I] on a solid phase of Iodogen (100 g/tube) for 10 min (16). Unreacted iodine was separated by gel filtration over Sephadex G50 colums or by anion exchange chromatography over Dowex-1.
Synthesis of Peptides and Production of Anti-peptide Antibodies-Peptides derived from the rat or human B2 receptor sequence ( Fig. 1) were synthesized by solid phase peptide synthesis using the Fmoc (N-(9-fluorenyl)methyloxycarbonyl) or the t-Boc (t-butyloxycarbonyl) chemistry (Table I). Peptides purified by high performance liquid chromatography were routinely analyzed by Edman degradation and electrospray mass spectrometry. Peptides were covalently coupled to the carrier protein, keyhole limpet hemocyanin, by maleimidocaproyl Nhydroxysuccinimide (17). Rabbits were immunized with the conjugates (18). Peptide MLN33 was used for immunization without prior coupling to a carrier protein. Antisera were tested for antigen specificity and cross-reactivity with homologous human or rat peptides by the indirect enzyme-linked immunosorbent assay (ELISA) (19) using microtiter plates (MaxiSorb, Nunc) coated with 2 g/ml of the peptide or 0.5 g/ml of the conjugate.
Western Blotting and Immunoprinting-Proteins were resolved by SDS-PAGE and transferred to polyvinylidene difluoride sheets using semidry blotting (20). The sheets were treated with 50 mM Tris, 0.2 M NaCl, pH 7.4 (buffer A), containing 5% (w/v) of nonfat dry milk and 0.1% (w/v) of Tween 20 for 1 h. Antisera were diluted 1:1000 in buffer A containing 2% (w/v) of bovine serum albumin. After 30 min of incubation at 37°C the polyvinylidene difluoride sheets were washed five times for 15 min each with buffer A and incubated for 30 min with peroxidase-labeled F(abЈ) 2 fragments of goat anti-rabbit antibody (Sigma, 1:5000). After extensive washing, bound antibody was visualized using the ECL chemiluminescence detection kit (Amersham).
Purification of Anti-peptide Antibodies by Affinity Chromatography-Peptides were covalently coupled to Affi-Gel 10 (5 mg/ml of gel) according to the manufacturer's instructions (Bio-Rad). The antiserum (5 ml/ml of gel) was applied and incubated under gentle agitation for 12 h at 4°C. The affinity matrix was washed three times with PBS, and the bound antibodies were eluted with 0.2 M glycine, pH 2.5, and immediately neutralized with 1 M KOH. Antibodies were desalted and concentrated using a Centricon filtration unit, exclusion limit 30,000 Da. The purity and specificity of the antibodies were analyzed by SDS-PAGE and enzyme-linked immunosorbent assay, respectively.
Lectin Affinity Chromatography of B2 Receptor-WGA was covalently coupled to Affi-Gel 10 (10 mg/ml of gel). B2 receptors from HF-15 cell membranes were solubilized with 4 mM CHAPS in 20 mM PIPES, pH 6.8. The solution was diluted with an equal volume of 20 mM PIPES, pH 6.8, adjusted to 1 M NaCl, 100 mM MnCl 2 , 100 mM CaCl 2 , and incubated for 4 h at 4°C with the WGA affinity matrix. After extensive washing, bound proteins were eluted by a 30-min incubation with an equal volume of 20 mM PIPES, 1 M NaCl, 100 mM MnCl 2 , 100 mM CaCl 2 , 0.5 M N-acetyl glucosamine, pH 6.8. Proteins were desalted and precipitated by 80% (v/v) acetone (21) and recovered by centrifugation. The protein pellet was dissolved in 2% (w/v) SDS, 5 mM EDTA, 5% Anti-HOE140 (AS 255) a Peptides are identified by their first three amino terminal residues using the one-letter code, followed by the total number of residues constituting the peptide.
Immunoaffinity Chromatography of the B2 Receptor-Affinity-purified domain-specific antibodies were covalently bound to Affi-Gel 10 (15 mg/ml gel). Membranes of Sf9 cells infected with baculovirus encoding the human B2 cDNA (100 pmols of B2 receptor/20 mg of total membrane protein) were solubilized with 2% (w/v) sodium deoxycholate in PBS including 1 mM phenylmethanesulfonyl fluoride, 1 g/ml E64, and 2 M leupeptin. The deoxycholate was diluted to 0.1% (w/v) by the addition of 20 mM HEPES, pH 7.4, containing 150 mM NaCl, 1 mM EDTA (buffer B). Then 10% (v/v) glycerol and 0.1% (w/v) Triton X-100 were added, and the solution was applied to the immunoaffinity matrix for an overnight incubation. The affinity matrix was extensively washed with buffer B, and bound proteins were eluted with 0.2 M glycine, pH 2.5 supplemented with 10% (v/v) 1,4-dioxane. The eluted protein fraction was neutralized with 1 M Tris, pH 8.0, and concentrated by Centricon filtration (exclusion limit 30,000 Da). The purity of the enriched B2 receptor was assessed by SDS-PAGE and silver staining. For NH 2 -terminal sequencing, proteins from three experiments were pooled, applied to a ProSpin sample preparation cartridge, and sequenced on a 477 A protein sequencer equipped with an on-line 120A PTH Analyzer (Applied Biosystems).
Affinity Cross-linking of the B2 Receptor-B2 agonist or antagonist was cross-linked to the B2 receptor as described previously (22) with minor modifications. B2 receptors of HF-15 cells were enriched by WGA affinity chromatography, and the eluted proteins were desalted by dialysis prior to ligand binding and cross-linking with 1 mM difluorodinitrobenzene. Cross-linking to recombinant B2 receptors of CHO (1.5 pmol/mg of protein) and Sf9 cells (4 -5 pmol/mg of protein) was performed on intact cells without prior enrichment of receptor protein.
Competition Studies with Radiolabeled Ligands-Membranes or confluent HF-15 cells on 24-well plates in 0.5 ml of RPMI 1640 including protease inhibitors and buffered with 20 mM Na ϩ -HEPES, pH 7.4 (binding buffer) were incubated with [ 125 I]HPP-HOE140 (0.5 nM, specific activity 1367 Ci/mmol) or with [ 3 H]bradykinin (2 nM, specific activity 98 Ci/mmol) in the presence of increasing concentrations of affinity-purified antibodies (5 ϫ 10 Ϫ11 M to 1 ϫ 10 Ϫ5 M). After 2 h of incubation at 4°C, the cells were washed three times with ice-cold medium. The cells were dissolved in 1% (w/v) NaOH, and radioactivity was determined.
Competition Studies with Iodinated Antibodies-Confluent HF-15 cells on 24-well plates were washed twice with binding buffer (see above). Then 0.5 ml of binding buffer was added to each well. For competition studies cells were incubated at 4°C with 125 I-labeled immunoselected antibodies (1 ϫ 10 Ϫ8 M; specific activity 0.02 Ci/mg) in the presence or absence of 1 ϫ 10 Ϫ5 M bradykinin or HOE 140. After 2 h of incubation at 4°C, the cells were washed three times with ice-cold medium and dissolved in 1% (w/v) NaOH, and radioactivity was determined.
Immunofluorescence Studies of A431 Cells-The human epithelial cell line, A431 was grown on glass coverslips for 48 h. Three h before the experiment, the medium was replaced by RPMI 1640 supplemented with 0.5% (v/v) fetal calf serum. Prior to immunofluorescence, cells were washed three times with 60 mM PIPES, 25 mM HEPES, 10 mM EDTA, 2 mM Mg(CH 3 COO) 2 , pH 6.9, and fixed for 30 min with 3% (w/v) paraformaldehyde in the same buffer adjusted to pH 7.5. Excess paraformaldehyde was quenched by the addition of 50 mM NH 4 Cl in PBS, pH 7.4; this was followed by 30 min of incubation with PBS, pH 7.4, containing 0.3% (w/v) gelatin. The cells were treated for 1 h at room temperature with anti-peptide antisera, 1:100 in 0.3% gelatin/PBS. The first antibody was detected using a rhodamine-coupled donkey antirabbit immunoglobulin, 1:100 in 0.3% gelatin/PBS. Controls included antisera preincubated for 2 h with 20 M of their respective antigens. The coverslips were embedded in Moviol ® and viewed with an Orthoplan microscope (Leitz).
Measurement of Changes in Intracellular Free Ca 2ϩ Concentration-Intracellullar free Ca 2ϩ concentration, [Ca 2ϩ ] i , of HF-15 cells was determined by fura-2/AM as described previously (8) with minor modifications. Confluent HF-15 cells grown on 10-mm-diameter glass coverslips were washed twice with minimum essential medium buffered with 20 mM Na ϩ -HEPES, pH 7.4 (HMEM), and incubated with 2 M fura-2/AM in HMEM containing 0.04% (w/v) pluronic F-127. After a 45-min incubation at 30°C, the cells were washed twice and incubated in HMEM for another 30 min to allow for complete deesterification of fura-2/AM. For determination of changes in [Ca 2ϩ ] i , the coverslips were mounted in a holder at an angle of 45°and put into a thermostatted quartz cuvette, and fluorescence at 510 nm was determined. The excitation wavelength alternated between 340 and 380 nm in intervals of 600 ms. Changes in [Ca 2ϩ ] i are given as the ratio of 340 and 380 nm.

Selection of Peptides for
Immunizations-To raise antisera that cross-react with four predicted extracellular domains (EDs) 2 of the B2 receptor, we selected six segments from ED1 through ED4 of the B2 receptor so that the peptides included a cysteine residue if available (Table I). The peptides covered the entire sequence of the putative ectodomains of the B2 receptor; there were two overlaps of 1 and 3 residues between the peptides selected from ED3 and ED4, respectively (Fig. 1). Two  Table I. additional peptides chosen from the putative intracellular domains (IDs), ID2 and ID4, served as controls. The peptides were covalently coupled to a carrier protein, keyhole limpet hemocyanin, and used for immunization except for MLN33, which was directly used without prior conjugation. The resultant antisera recognized their cognate antigens as verified by the indirect enzyme-linked immunosorbent assay, and crossreacted with the sequences from the human or the rat B2 receptor, respectively (not shown). A single peptide, designated CWN12, which is derived from ED4 and covers the center portion of peptide SGC18, failed to produce a significant titer of specific antibodies (not shown).
Specificity of the Anti-B2 Antisera-To assess the specificity of the antisera, we used membranes from CHO cells and Sf9 cells that overexpress the rat or the human B2 receptor for Western blotting and immunoprinting. The results are exemplified for the antiserum to the first ectodomain, ED1. In Triton X-114-extracted membranes of recombinant CHO cells expressing the rat B2 cDNA, anti-ED1 detected a single 69 Ϯ 3-kDa protein (Fig. 2B, lane 1). No specific staining was found in nontransfected cells, Fig. 2B, lane 2. As control B2 receptor from CHO cells that was labeled by HOE140 was detected using anti-HOE140 antiserum (Fig. 2B, lane 3). Binding of the antibodies to the B2 receptor was suppressed when cross-linking was performed in the presence of a 1,000-fold molar excess of bradykinin (Fig. 2B, lane 4). We further tested the specificity of the anti-ED1 antiserum using CHAPS-extracted membranes of Sf9 cells infected with recombinant baculovirus encoding the human B2 cDNA, and detected three protein bands of 38, 41, and 45 Ϯ 5 kDa (Fig. 2C, lane 1). Proteins with a molecular mass of approximately 75-80 kDa are likely to be dimerized B2 receptors that aggregated probably during sample preparation of Sf9 cell membranes expressing high amounts of B2 receptor (22). Membranes from mock-infected Sf9 cells did not show any specific bands (Fig. 2C, lane 2). As a control HOE140 was cross-linked to the B2 receptor followed by detection with anti-HOE140 antiserum. Proteins of similar molecular weight were stained, Fig. 2C, lane 3. This staining was suppressed by a 1,000-fold molar excess of bradykinin (Fig. 2C, lane 4). The differences in the apparent molecular masses of recombinant B2 receptors from Sf9 cells and of B2 receptors of HF-15 fibroblasts and the occurrence of multiple immunoreactive bands in Sf9 cells are likely caused by incomplete glycosylation, characteristic for glycoproteins expressed in Sf9 cells (23).
Affinity Purification and Amino-terminal Sequence Analysis of the B2 Receptor-Is the protein identified by immunostaining with anti-peptide antisera the authentic B2 receptor? We enriched the receptor protein from Sf9 membranes by immunoaffinity chromatography using a mixture of immunoselected anti-peptide antibodies to the extracellular domains. Edman degradation of the protein revealed an amino-terminal sequence of Met-Leu-Asn-Val-Thr-Xaa-Gln-Gly-Xaa-Thr-Leu-Asn-Gly-Thr-Phe-Ala-Xaa-Ser, where Xaa stands for an unidentified residue. This sequence is identical with the human B2 receptor sequence starting at the third in-frame initiator codon (11); note that the construct in baculovirus had been engineered such that only the most 3Ј-located initiator codon was available. These data demonstrate that the affinity-purified anti-peptide antibodies selectively enrich B2 receptor from solubilized Sf9 membranes.

Fluorescence-activated Cell Sorting (FACS) Analysis of Native B2 Receptors on HF-15
Fibroblasts-To test the reactivity of the antibodies with native B2 receptor we performed FACS analysis of HF-15 cells that were stained by antisera to the various extracellular domains ED1 to ED4 (Fig. 3A). Antisera to extracellular domains ED1 to ED4 bound to B2 receptors of intact HF-15 cells (Fig. 3A, I-VI) as demonstrated by increased fluorescence intensity in comparison with preimmune serum (Fig. 3A, VII), suggesting that the various anti-peptide antibodies cross-react to similar extents with the B2 receptor. No specific staining was observed with antisera to intracellular domain ID2 or ID4, exemplified for anti-ID2 (Fig. 3A, VIII). This finding is in agreement with the hypothetical model of the B2 receptor (cf. Fig. 1).
Redistribution of B2 Receptor Detected by Antibodies to Extracellular Domains-The successful binding of anti-peptide antisera to cellular B2 receptors allowed us to examine the fate of the B2 receptor after its activation by an agonist. In these studies we applied a mixture of the various anti-ED antibodies. HF-15 cells were preincubated at 37°C for 60 min in the absence (Fig. 3B, I) or presence of 1 M bradykinin (Fig. 3B,  III). Pretreatment of the cells by the B2 agonist drastically reduced the antibody binding to B2 receptors (Fig. 3B, III). Thus, after agonist treatment the antigenic epitopes are no longer available for antibody binding. Preincubation with 1 M of the antagonist, HOE140, did not change the overall fluorescence intensity (Fig. 3B, IV). Together these data suggest that our anti-peptide antibodies readily recognize the extracellular domain(s) of the B2 receptor and that, like other G-proteincoupled receptors, sequestration and/or internalization modifies the accessibility of extracellular domains for antibody detection.
Immunofluorescence of A431 Cells-To immunovisualize the B2 receptor on cells other than fibroblasts, the epidermoid carcinoma cell line A431 was chosen. A strong immunostaining of the plasma membrane of fixed, nonpermeabilized cells was observed with a mixture of antibodies against ED1 to ED4 (Fig.  4a). A specific staining of the outer rims of the cells was also observed when the individual antisera against extracellular domains, ED1, ED2, ED3 N , and ED4 C , were used ( Fig. 4e to  4h). The cells exhibit a punctuated labeling, which may be due to receptor clustering and/or high receptor density in pseudopodia and microvilli of A431 cells; this latter notion was confirmed by electron microscopy (not shown). The presence of the cognate antigen for each antibody abrogated the specific immunostaining (b). No staining was seen with preimmune serum (c), with antiserum to an unrelated peptide (d), or with an antiserum to intracellular domain, ID2 (i). Hence the antipeptide antibodies are useful to probe for the B2 receptor on the surface of various cell types (Figs. 3 and 4).
Blockade of Bradykinin Binding by Anti-ED Antibodies-We asked whether our antibodies interfered with the binding of bradykinin to the B2 receptor. HF-15 cell membranes were preincubated for 2 h at 4°C with 250 nM affinity-purified antibodies to the various extracellular domains. This was followed by the addition of 2 nM [ 3 H]bradykinin and further incubation for 60 min at 4°C. The unbound radioligand was separated by filtration through GF/C glass filters, and the filter-bound radioactivity was determined. Controls were done in the absence (total binding) or presence of 2 M of the unlabeled ligand, bradykinin (Fig. 5A, columns 1 and 2). Out of six antisera tested only antibodies against the amino-terminal portion of extracellular domain 3, ED3 N , and to the carboxylterminal segment of extracellular domain 4, ED4 C , interfered with bradykinin binding to the B2 receptor (Fig. 5A, columns 5  and 8). Antibodies to other segments of the extracellular domains (Fig. 5A, columns 3, 4, 6, and 7) and to the intracellular domains (not shown) had no effect on [ 3 H]bradykinin binding. These data suggest that domains ED3 and ED4 of the B2 receptor might be critically involved in ligand binding.
Concentration-dependent Displacement of Radioligand by Anti-ED Antibodies-To further analyze the involvement of extracellular domains 3 and 4 in agonist and antagonist binding, we tested whether the effect of the antibodies is dose-de- pendent. Intact HF-15 cells were incubated for 2 h at 4°C with the radioligand in the presence of increasing concentrations, 100 pM to 10 M, of the affinity-purified antibodies (Fig. 5, B or  C). Cell-bound radioactivity was determined. Antibodies to ED3 N effectively blocked binding of [ 3 H]bradykinin (Fig. 5B) and of [ 125 I]HPP-HOE140 (Fig. 5C) (Fig. 5B); however, the antibodies did not block the binding of [ 125 I]HPP-HOE140 (Fig. 5C). Other antibodies such as anti-ED3 C did not interfere with the binding of the ligands (Fig. 5, B and C). We conclude that at least part of the contact site for bradykinin and/or for HOE140 may be near or within the amino-terminal portion of ED3. The carboxyl-terminal portion of ED4 may contribute to the receptor binding of the agonist but not of the antagonist.
Displacement of 125 I-Labeled Anti-ED Antibodies by B2 Ligands-Anti-ED antibodies interfere with radioligand binding to the B2 receptor. This may either indicate that they are competitive inhibitors for the binding site or that the antibodies act allosterically by inducing and/or stabilizing a receptor conformation unable to bind the ligand. To discriminate between these possibilities competition binding between the unlabeled ligands, bradykinin and HOE140, and the 125 I-labeled antibodies anti-ED3 N and anti-ED4 C was done with HF-15 fibroblasts. The binding of 10 nM 125 I-labeled anti-ED3 N to the receptor was reduced by 80% in the presence of 10 M bradykinin, and it was abolished by 10 M HOE140 (Fig. 6A). At the same concentration the cognate peptide, KDY13, completely displaced radiolabeled anti-ED3 N ; the B1 receptor agonist, [des-Arg 9 ]bradykinin, had no effect (Fig. 6A). In the case of 125 I- labeled anti-ED4 C antibodies, no inhibition of binding was seen in the presence of 10 M bradykinin or HOE140, whereas displacement was observed by the same concentration of the cognate peptide, SGC18, Fig. 6B. We conclude that antibodies to ED3 N , but not antibodies to ED4 C , are competitive with B2 receptor ligands. Anti-ED4 C may cause an allosteric alteration and stabilize a conformation of the receptor unable to bind agonists.
Displacement of Anti-idiotypic Antibodies by Antibodies to ED3 N -To further address the interaction between anti-ED3 N and the kinin receptor we applied anti-idiotypic antibodies that had been raised against the idiotype, monoclonal antibody MBK3 to bradykinin (24). These antibodies have previously been shown to bind to and stimulate in an agonist-like manner the human or mouse B2 receptor (24). Competition experiments demonstrate that bradykinin and HOE140 interfere with 125 I-labeled anti-idiotypic antibodies for receptor binding (Fig. 6C). Antibodies to ED3 N displaced the radiolabeled antiidiotypes, although not completely, whereas antibodies to ED1, ED2, and ED4 C had no effect (anti-ED4 C is shown in Fig. 6C). These findings indicate that the majority of the anti-idiotypic antibodies are likely to bind to ED3 N and that interaction sites for bradykinin, HOE140, anti-idiotypic antibodies, and anti-ED3 N antibodies with the external portion of the receptor are mutually overlapping.
Agonist-like Effects of Antibodies to ED3 N -Our finding that antibodies to ED3 N interfere with the receptor binding of bradykinin and anti-idiotypic antibodies prompted us to ask whether anti-ED3 N itself is an agonist. Therefore, we measured intracellular free Ca 2ϩ in HF-15 fibroblasts treated with anti-ED3 N . At a concentration of 250 nM anti-ED3 N transiently increased [Ca 2ϩ ] i in an agonist-like manner, Fig. 7A. This effect is mediated by the B2 receptor because a 10-fold molar excess of the B2 antagonist, HOE140, prevented the Ca 2ϩ transient (Fig. 7B). Antibodies to the distal portion of the same domain, ED3 C , or to other ectodomains such as ED4 C were without effect (Fig. 7, C and D). Hence polyclonal antibodies to the amino-terminal portion of ectodomain 3 are agonists. DISCUSSION In these studies antibodies directed to putative extracellular domains of the bradykinin B2 receptor were prepared. These antibodies were used to map extracellular domains involved in ligand binding. This approach of ligand binding site mapping is complementary to the site-directed mutagenesis studies (25)(26)(27). The antibody approach reduces the commonly voiced concern about site-directed mutagenesis, that the mutation changes the receptor structure and binding ability without being located at the ligand binding site.
Our experiments show that the amino-terminal portion of ectodomain 3, ED3 N , is involved in agonist binding and sensing because (i) antibodies to this segment competed with bradykinin for binding to the B2 receptor, (ii) bradykinin almost completely abolished the binding of radiolabeled anti-ED3 N to B2 receptors, and (iii) anti-ED3 N antibodies were agonists. A direct contact between the ED3 N segment and bradykinin is uncertain; however, the mutual competition of ED3 N antibodies and bradykinin suggests such a possibility. The ED3 N region is also involved in binding of the antagonist, HOE140, because (i) anti-ED3 N blocked HOE140 binding to B2 receptors, (ii) HOE140 completely abolished the binding of radiolabeled anti-ED3 N to B2 receptors, and (iii) HOE140 nearly completely blocked anti-ED3 N -induced cytosolic Ca 2ϩ increase, i.e. anti-ED3 N agonism.
The agonistic effect of anti-ED3 N antibodies demonstrates that this receptor region can assume or can be induced to assume conformation(s) that transmit the signal to the Gprotein. If one considers a two-domain model of G-proteincoupled receptors as suggested by the observation that transmembrane regions (TMs) 1-5 and TMs 6 -7 need not be covalently connected for G-protein-coupled receptors to bind and signal (28,29), then anti-ED3 N might push apart the two domains, allowing access of the G-protein to the intracellular loops. Alternatively, anti-ED3 N might stabilize the R* activated form of the receptor which is at equilibrium with the R inactive form under basal conditions (30). Autoantibodies to extracellular domains of the adrenergic and muscarinic receptors have been detected in the serum of patients with myocardial diseases or malignant hypertension (31)(32)(33). These antibodies are directed to the same extracellular loop as is anti-ED3 N , interfere with ligand binding, and are agonists. These similarities between antibodies to extracellular domains of cationic amine receptors and of a peptide receptor emphasize the common molecular mechanisms governing the action of Gprotein-coupled receptors.
A few attempts have been made to elucidate the binding site of the B2 receptor using site-directed mutagenesis (26,27). 3 Alanine substitutions of negatively charged residues were FIG. 7. Effect of anti-ED antibodies on the [Ca 2؉ ] i of human fibroblasts. HF-15 cells were loaded with fura-2/AM, and the change in the ratio of fluorescence at 340/380 nm was followed. At the time points indicated 250 nM affinity-purified antibodies to ED3 N (panels A, B), ED3 C (C), or ED4 C (D) were applied. In panel B, 2.5 M HOE140 was added 50 s prior to application of the antibody. The results were similarly obtained with three different antibody preparations, each derived from two different rabbits. made at the TM7/ED4 boundary, D286A contained in the ED4 C epitope, and at the TM6/ED4 boundary, D268A contained in the ED4 N epitope. The D268A change reduced slightly the affinity of bradykinin and did not change the affinity with the related antagonists HOE140 or NPC17761 (26). 3 However, anti-ED4 N antibodies had no effect on bradykinin or HOE140 binding. The D286A mutation had larger effects on the bradykinin affinity and small effects on antagonist binding affinity. In accordance with that observation, anti-ED4 C antibodies reduced bradykinin binding but had no effect on antagonist binding. Finally an alanine substitution, E179A, contained in the ED3 N epitope, also caused a small reduction in bradykinin affinity (26). Thus the mutagenesis and the antibody methods for probing ligand binding sites concur in their indication that the charged residues Asp 286 and Glu 179 and associated peptide regions may be involved in agonist binding.
Our anti-ED4 N antibody does not confirm the involvement of Asp 268 in bradykinin binding; however, we note that D268 is the amino-terminal residue of the ED4 N peptide, DTL12; and thus, the anti-ED4 N antibodies may not bind the Asp 268 residue as part of an extended protein chain in the same way as when it is the first residue of a peptide. The finding that anti-ED4 C inhibits bradykinin binding but is unable to inhibit HOE140 binding suggests that the binding sites for peptidic agonists and antagonists on B2 receptors do not perfectly overlap. This conclusion agrees with the suggestions, derived from site-directed mutagenesis studies, that agonists and antagonists do not bind to identical sites on the receptor (26,34). 3 These studies, which used anti-extracellular domain antibodies covering all the extracellular domains of the bradykinin receptor, demonstrate that the binding of agonists by the receptor involves extracellular regions at the top of TM4, ED3 N , and to a lesser extent at the top of TM7, ED4 C . In contrast, the binding of antagonists is only affected by antibodies directed to the top of TM4, ED3 N . Furthermore, the anti-ED3 N antibodies are agonists, suggesting that the TM4 to TM5 loop, ED3, is important for signal transduction. These studies also point to the importance of extracellular domains for binding and signal transduction in this member of the G protein-coupled receptor family and demonstrate the utility of epitope-specific antibodies in defining functionally important regions of receptors.