High-Affinity IgE Recognition of a Conformational of the Major Respiratory Phl p 2 As Revealed by X-Ray Crystallography This information is current as of September 28, 2021. Sivaraman Padavattan, Sabine Flicker, Tilman Schirmer, Christoph Madritsch, Stefanie Randow, Gerald Reese, Stefan Vieths, Christian Lupinek, Christof Ebner, Rudolf Valenta and Zora Markovic-Housley

J Immunol 2009; 182:2141-2151; ; Downloaded from doi: 10.4049/jimmunol.0803018 http://www.jimmunol.org/content/182/4/2141 http://www.jimmunol.org/ References This article cites 62 articles, 12 of which you can access for free at: http://www.jimmunol.org/content/182/4/2141.full#ref-list-1

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists by guest on September 28, 2021

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2009 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

High-Affinity IgE Recognition of a Conformational Epitope of the Major Respiratory Allergen Phlp2AsRevealed by X-Ray Crystallography1

Sivaraman Padavattan,2* Sabine Flicker,2† Tilman Schirmer,* Christoph Madritsch,† Stefanie Randow,‡ Gerald Reese,‡ Stefan Vieths,‡ Christian Lupinek,† Christof Ebner,§ Rudolf Valenta,3,4†¶ and Zora Markovic-Housley3*

We report the three-dimensional structure of the complex between the major respiratory grass pollen allergen Phl p 2 and its specific human IgE-derived Fab. The Phl p 2-specific human IgE Fab has been isolated from a combinatorial library constructed from of a pollen allergic patient. When the variable domains of the IgE Fab were grafted onto human IgG1, the resulting Ab (huMab2) inhibited strongly the binding of allergic patients’ IgE to Phlp2aswell as allergen-induced Downloaded from degranulation. Analysis of the binding of the allergen to the Ab by surface plasmon resonance yielded a very low dissociation ؋ ؊10 ␧ ؍ constant (KD 1.1 10 M), which is similar to that between IgE and Fc RI. The structure of the Phl p 2/IgE Fab complex was determined by x-ray crystallography to 1.9 Å resolution revealing a conformational epitope (876 Å2) comprised of the planar surface of the four-stranded anti-parallel ␤-sheet of Phl p 2. The IgE-defined dominant epitope is discontinuous and formed by 21 residues located mostly within the ␤ strands. Of the 21 residues, 9 interact directly with 5 of the 6 CDRs (L1, L3, H1, H2, H3) of the IgE Fab predominantly by hydrogen bonding and van der Waals interactions. Our results indicate that IgE Abs recognize http://www.jimmunol.org/ conformational with high affinity and provide a structural basis for the highly efficient effector cell activation by allergen/ IgE immune complexes. The Journal of Immunology, 2009, 182: 2141–2151.

he hallmark of type I diseases, e.g., al- series of inflammatory processes through the binding to and cross- lergic asthma, rhinitis, skin inflammation, food , linking of the high-affinity receptor for IgE (Fc␧RI) on mast cells, T anaphylactic shock, which affect more than 25% of the , , and professional APCs such as dendritic population, is the formation of IgE Abs against per se harmless cells (4–7). In contrast, IgE-allergen immune complexes can reg-

Ags, i.e., (1). IgE Abs represent the least abundant class ulate IgE production and activation via the low-affinity re- by guest on September 28, 2021 of immunoglobulins occurring at approximately 1000-fold lower ceptor for IgE (Fc␧RII, i.e., CD23) (8–11). Allergen recognition concentrations in serum and other body fluids compared with IgG also directly stimulates the IgE production in IgEϩ B cells leading and other Ig classes (2, 3). Nevertheless, minute amounts of im- to increased levels of serum IgE in patients after allergen contact mune complexes consisting of IgE Abs and allergens can trigger a (12, 13). Interestingly, mucosal contact with tiny allergen amounts *Division of Structural Biology, Biozentrum, University of Basel, Basel, Switzerland; strongly boosts allergen-specific IgE Ab production but only †Division of Immunopathology, Department of Pathophysiology, Center for Physi- weakly induces rises of allergen-specific IgG or IgA production ology and Pathophysiology, Medical University of Vienna, Vienna, Austria; ‡Paul Ehrlich Institut, Department of Allergology, Langen, Germany; §Allergie-Ambulato- (13). rium Reumannplatz, Vienna, Austria; ¶Christian Doppler Laboratory for Allergy Re- Moreover, extremely low concentrations (ng/ml) of allergens search, Department of Pathophysiology, Center for Physiology and Pathophysiology, are sufficient to induce rapid and strong inflammatory responses Medical University of Vienna, 1090 Vienna, Austria through degranulation of mast cells and basophils as well as Received for publication September 12, 2008. Accepted for publication November 28, 2008. through the activation of allergen-specific T cells (6, 14, 15). One The costs of publication of this article were defrayed in part by the payment of page prerequisite for this efficient activation of immune cells is a high- charges. This article must therefore be hereby marked advertisement in accordance affinity binding of allergen-IgE immune complexes to Fc␧RI on with 18 U.S.C. Section 1734 solely to indicate this fact. effector cells. In fact, IgE binds primarily with its C␧2 domain with 1 This project was supported by the Swiss National Foundation Grant 31-116804 (to extremely high affinity (K ϳ 10Ϫ9 M) to the ␣-chain of Fc␧RI Z.M.-H.), by Grant 813003 of the Austrian Research Promotion Agency and BIOMAY, D Vienna, Austria to (S.F.), by Grant F1815 of the Austrian Science Fund (to R.V.), by a (16). research grant from BIOMAY, Vienna, Austria, and the Christian Doppler Research Blood and body fluids of allergic patients contain relatively high Association, Vienna, Austria. levels of allergen-specific IgG Abs (17). However, only after a Coordinates and structure factors have been deposited in the Data Bank with more than 100-fold increase of allergen-specific IgG during aller- accession numbers 2vxq and r2vxqsf, respectively. gen-specific immune therapy are these IgG Abs able to compete 2 S.P. and S.F. contributed equally to this study. with allergen-specific IgE Abs and to prevent mast cell degran- 3 R.V. and Z.M.-H. contributed equally to this study. ulation and T cell activation (18). One possibility for the effi- 4 Address correspondence and reprint requests to Prof. Rudolf Valenta, Christian cient recognition of allergen by IgE Abs is that IgE recognizes Doppler Laboratory for Allergy Research, Division of Immunopathology, Department of Pathophysiology, Center for Physiology and Pathophysiology, Vienna General different epitopes on allergens than IgG Abs. Evidence for dif- Hospital, Medical University of Vienna, Waehringer Guertel 18-20, A-1090 Vienna, ferent epitope recognition comes from studies Austria. E-mail: [email protected] performed with allergen fragments and the analysis of the bind- Copyright © 2009 by The American Association of Immunologists, Inc. 0022-1767/09/$2.00 ing specificities of defined allergen-specific human IgG Abs www.jimmunol.org/cgi/doi/10.4049/jimmunol.0803018 2142 X-RAY CRYSTAL STRUCTURE OF AN ALLERGEN/IgE COMPLEX

0.3nM 40 0.6nM 1.25nM 2.5nM 5.0nM FIGURE 1. Sensor chip-based 10nM studies of the interaction between rPhl 30 20nM p 2 and huMab2. rPhl p 2 was injected at 2-fold increasing concentrations from 0.3 to 20 nM (curves bottom to 20 top) into the flow cell containing im- mobilized huMab2 rPhl p 2-specific RU Ab. Recorded curves (colored) and 10 calculated curves (black), which repre- sent a fitting of the response data to a 1:1 interaction model were superim- 0 posed onto each other. The signal inten- sity (RU) is shown on the y-axis whereas the x-axis displays the time (s). -10 0 500 1000 1500 2000 2500 Time [s] Downloaded from

(19, 20). Studies performed with allergen-specific serum IgE vs Determination of the affinity and kinetics of the interaction serum IgG report controversial results regarding possible between rPhl p 2 and huMab2 differences between the binding strength of allergen-specific The surface of a CM5 sensor chip (BIACore AB) was activated by the injec- http://www.jimmunol.org/ IgE and IgG Abs (21, 22). tion of a 1:1 mixture of EDC (1-ethyl-3-(3-dimethylaminopropyl)carbodiimide Many three-dimensional structures of important allergens have hydrochloride) and NHS at a flow rate of 5 ␮l/min for 7 min. The purified been solved but only few allergen-Ab complexes have been char- huMab2 (c ϭ 10 ␮g/ml) which had been diluted in 10 mM sodium acetate (pH acterized so far (23–26). We have previously isolated human IgE Fabs specific for major respiratory grass pollen allergens from a Table I. Inhibition of patients’ IgE binding to rPhlp2bytherPhl p combinatorial library constructed from lymphocytes of a pollen a allergic patient (27–29). 2-specific

This is the first report of the three-dimensional structure of the ␣ ␣ Phlp5 huMab2 % r Phl p r Phl p % by guest on September 28, 2021 complex between an IgE-derived Fab with a major respiratory al- Patient ab (OD) (OD) Inhibition 5 (OD) 2 (OD) Inhibition lergen, the grass pollen allergen Phl p 2, which is recognized by 1 1.510 0.634 58.0 n.d. n.d. n.d. more than 200 million allergic patients (30). The Fab consisted of 2 0.938 0.350 62.7 n.d. n.d. n.d. an authentic IgE H chain fragment and a randomly combined L 3 0.624 0.187 70.0 n.d. n.d. n.d. chain obtained from the same allergic patient and therefore possi- 4 1.486 0.775 47.8 n.d. n.d. n.d. 5 1.609 0.755 53.1 n.d. n.d. n.d. bly is not identical with the structure of a complete human IgE. 6 1.087 0.155 85.7 1.150 0.157 86.3 However, the Ab strongly inhibited the binding of allergic patients 7 0.694 0.118 83.0 0.756 0.162 78.6 8 0.737 0.247 66.5 0.783 0.119 84.8 IgEtoPhlp2aswell as allergen-induced basophil degranulation. 9 0.677 0.297 56.1 0.744 0.248 66.7 The three-dimensional structure of the complex reveals that the 10 0.450 0.293 34.9 0.465 0.173 62.8 11 0.463 0.374 19.2 0.467 0.187 60.0 epitope defined by the allergen-specific Fab, involves several seg- 12 0.755 0.237 68.6 0.772 0.209 72.9 ments of discontinuous amino acids that are part of a solvent ex- 13 0.440 0.106 75.9 0.457 0.119 74.0 posed anti-parallel ␤-sheet of the allergen. In the context of ex- 14 0.871 0.643 26.2 0.923 0.174 81.1 15 0.747 0.066 91.2 0.750 0.102 86.4 perimental evidence gained through the molecular analysis of 16 2.390 0.954 60.1 2.157 0.375 82.6 major allergens, including Phl p 2 (31, 32) our results suggest that 17 0.455 0.082 82.0 0.355 0.093 73.8 18 0.763 0.209 72.6 0.736 0.217 70.5 IgE Abs from allergic patients recognize conformational epitopes 19 0.638 0.299 53.1 0.656 0.101 84.6 on allergens. This finding has implications for the understanding of 20 1.057 0.143 86.5 1.051 0.178 83.1 allergen-specific IgE responses and the design of specific immu- 21 0.925 0.258 72.1 0.948 0.328 65.4 22 0.645 0.195 69.8 0.584 0.116 80.1 notherapy strategies for IgE-mediated . 23 0.512 0.126 75.4 0.483 0.128 73.5 24 0.567 0.201 64.6 0.551 0.143 74.1 25 0.734 0.160 78.3 0.721 0.104 85.6 26 0.826 0.369 55.3 0.578 0.124 78.5 Materials and Methods 27 0.265 0.050 81.3 0.197 0.072 63.5 28 0.505 0.154 69.5 0.404 0.068 83.2 Allergen, synthetic peptides, sera, huMab2, rabbit immune sera 29 0.373 0.146 60.9 0.326 0.131 60.0

Purified recombinant Phl p 2 and Phl p 1 were purchased from BIOMAY. Mean: 64.8 75.5 The Phl p 2–derived peptides (peptide 1, aa 1–33; peptide 2, aa 34–65; Non-allergic 0.042 0.051 0.042 0.081 peptide 3, aa 66–96) comprising the complete allergen were synthesized on individual the Applied Biosystems peptide synthesizer model 433A. Sera from grass a pollen-allergic patients, who had been characterized by clinical history, OD values corresponding to bound Phl p 2-specific IgE are displayed for 29 skin prick testing, and grass pollen extract-specific IgE detection, were allergic patients’ sera and the serum from a non-allergic individual. Phlp2was preincubated with huMab2, an -matched Phl p 5-specific (Phl p 5 provided by Christof Ebner. HuMab2 and Phl p 1-specific IgG1 Abs were ab), Phl p 2-specific rabbit antibodies (r␣Phl p 2), or Phl p 5-specific rabbit antibodies constructed, expressed, and purified as described (28, 29). Rabbit anti-sera (r␣Phl p 5). The percentages of inhibition of IgE binding to rPhl p 2 are displayed for were obtained by immunization of a rabbit with purified recombinant al- the Phl p 2-specific antibody and the Phl p 2-specific rabbit antibodies. Mean per- lergens using CFA and IFA, respectively (Charles River Laboratories). centages of inhibition are shown in the bottom line. n.d., not done. The Journal of Immunology 2143

FIGURE 2. Inhibition of basophil de- granulation with huMab2. RBL-30/25 were passively sensitized with sera of grass pollen Downloaded from allergic patients (A–D) and were stimulated with increasing concentrations of rPhl p 2 (x-axes:1 ng/ml–1 ␮g/ml) preincubated with rPhl p 2-specific Abs (huMab2), rPhl p 2-specific rabbit Abs (r␣Phl p 2) and, for control purposes, with rPhl p 1-specific Abs http://www.jimmunol.org/ (Phl p 1 ab) and rPhl p 5-specific rabbit Abs (r␣Phl p 5). ␤-Hexosaminidase release is ex- pressed as percentage of total ␤-hexosamini- dase release on the y-axes. by guest on September 28, 2021

5) was injected at 5 ␮l/min into flow cell 2. The surface of the sensor chip was Degranulation experiments with humanized rat basophil then deactivated by the injection of 1 M ethanolamine-HCl (pH 8.5) at a flow leukemia cells rate of 5 ␮l/min for 7 min. Using the same protocol, a purified monoclonal Phl p 1-specific Ab (c ϭ 10 ␮g/ml) was immobilized in flow cell 1, which served The RBL-30/25 cell clone expressing the human ␣-chain of the Fc␧RI was as reference cell. The final immobilization levels were 3044.4 resonance units established by cotransfection of RBL-2H3 cells with three plasmids en- (RU)5 (Phl p 2-specific Ab, flow cell 2), and 3086.5 RU (Phl p 1-specific Ab, coding the human ␣-, ␤-, and ␥-chains of Fc␧RI (33). The RBL-30/25 cells flow cell 1), respectively. Calculated Rmax for Phl p 2 was 426 RU. rPhl were maintained in MEM (Invitrogen) supplemented with 5% FCS (In- p 2 was diluted in HBS-EP (0.01M HEPES, 0.15M NaCl, 3 mM EDTA, vitrogen), and 1% L-glutamine (Biochrom). RBL-2H3 cells were plated in 0.005% (v/v) surfactant P20, pH 7.4) in 2-fold increasing concentra- 96-well tissue culture plates (1.5 ϫ 106/ml; Nunc) and incubated at 37°C tions (starting concentration, 0.3 nM; top concentration, 20 nM) and in 5% CO for 2 h. Passive sensitization of the cells was performed by ␮ 2 tested for the binding to the immobilized Abs at a flow rate of 30 l/min incubating the cells with serum IgE from grass pollen allergic patients for 8 min. Dissociation of rPhl p 2 was investigated by the injection of (diluted 1/16 in MEM) overnight. The cell layer was then washed three ␮ HBS-EP at 30 l/min for 30 min. Regeneration of the sensor chip times with 1ϫ Tyrode buffer (137 mmol/L NaCl, 2.7 mmol/L KCl, 0.5 surface was performed by injection of 10 mM glycine-HCl, pH 1.5 (30 mmol/L MgCl , 1.8 mmol/L CaCl , 0.4 mmol/L NaH PO , 5.6 mmol/L ␮l/min, 30 s). The complete procedure was done at 25°C. Kinetic and 2 2 2 4 D-glucose, and 0.1% w/v BSA, pH 7.2; Sigma-Aldrich) to remove unbound affinity constants were calculated with BIAEvaluation 3.2 (BIACore IgE Abs. The cross-linking of the Fc␧RI-bound IgE and subsequent de- AB) using a 1:1 interaction model. granulation of RBL cells was induced by adding rPhl p 2 (10-fold dilutions between 1 ng/ml and 1 ␮g/ml) in 1ϫ Tyrode buffer plus 50% deuterium 5 Abbreviations used in this paper: RU, resonance units; PEG, polyethylene glycol; oxide to the cells and incubation for1hat37°C. Before rPhlp2was CEP, conformational epitope prediction; BLG, ␤-lactoglobulin. incubated with the passively sensitized cells, it was either preincubated 2144 X-RAY CRYSTAL STRUCTURE OF AN ALLERGEN/IgE COMPLEX overnight at 4°C with 1) the huMab2 diluted 1/20 in 1ϫ Tyrode buffer (c ϭ Table II. Crystallographic statisticsa 30 ␮g/ml); 2) the rPhl p 1-specific Abs diluted 1/20 in 1ϫ Tyrode buffer (c ϭ 30 ␮g/ml); 3) rPhl p 2-specific rabbit Abs diluted 1/5 in 1ϫ Tyrode Data collection buffer (ϭ20% rabbitЈs serum); or 4) rPhl p 5-specific rabbit Abs diluted 1/5 in 1ϫ Tyrode buffer (ϭ20% rabbitЈs serum). The release of ␤-hexosamini- Space group P42212 dase was analyzed by incubating culture supernatants with 1.3 mg/ml 4-ni- Unit cell dimensions ␤ trophenyl-N-acetyl- -D-glucosaminide (Sigma-Aldrich) in citrate buffer a, b, c (Å) 105.4, 105.4, 110.2 (0.1 mol/L, pH 4.5) for1hat37°C. The reaction was stopped by the a, ␤, ␥ (°) 90.0, 90.0, 90.0 addition of glycine buffer (0.2 mol/L glycine, pH 10.7), and the optical X-ray source SLS-PX densities were measured at an extinction wavelength of 405 nm to the Detector type PILATUS emission wavelength of 620 nm by using a Spectra Max 340Pc (Molecular Wavelength (Å) 1.000 ␤ Devices). The specific release was quantified by measuring -hexosaminidase Resolution range (Å) 27.6–1.9 (2.0–1.9)a ␤ activity and expressed as a percentage of the total -hexosaminidase content No. of total observation 269374 (27541) that was obtained by lysing the cells by the addition of 1% Triton X-100 No. of unique observation 48723 (6352) (Merck). For measurements of spontaneous release and possible unspecific Completeness (%) 98.4 (89.5) effects, RBL cells were incubated with cell culture medium, allergen alone, or Multiplicity 5.5 (4.3) ␤ Abs alone. Nonspecific release was never higher than 5% of the total -hex- I/␴(I) 17.9 (4.5) osaminidase release. The specific release was corrected for any spontaneous R b (%) 7.0 (24.4) release, and the resulting release curves were fitted by Sigma Plot. sym Refinement Total number of reflections 46,212 ELISA competition experiments c Rcryst/Rfree (%) 17.6/22.5

The ability of the human rPhl p 2-specific Ab to inhibit the binding of grass Protein atoms 3976 Downloaded from pollen allergic patients‘ IgE Abs to rPhl p 2 was studied by ELISA com- Water molecules 524w petition experiments. In brief, ELISA plates (Nunc-Maxisorp) were coated Average B factor (Å2) with rPhl p 2 (100 ␮l; c ϭ 1 ␮g/ml) in 0.1 M sodium bicarbonate (pH 9.6) Phl p 2 26.8 for2hat37°C. The plates were washed twice with PBST and blocked with Fab L chain 25.4 PBST containing 1% w/v BSA at 37°C for 3 h. After blocking, the fol- Fab H chain 24.3 lowing preincubations were performed in duplicates overnight at 4°C: 1) Solvent 35.0 incubation with purified huMab2 diluted 1/100 in PBST/0.5% w/v BSA Root mean square deviation (c ϭ 6 ␮g/ml); 2) incubation with purified rPhl p 1-specific Abs diluted from ideal values http://www.jimmunol.org/ 1/100 in PBST/0.5% w/v BSA (c ϭ 6 ␮g/ml); 3) incubation with rPhl p Bond lengths (Å) 0.008 2-specific rabbit Abs diluted 1/1000 in PBST/0.5% w/v BSA (ϭ0.1% rab- Bond angles (°) 1.25 bitЈs serum); 4) incubation with rPhl p 5-specific rabbit Abs diluted 1/1000 Root mean square deviation ⌬B in PBST/0.5% w/v BSA (ϭ0.1% rabbitЈs serum). After washing five times of bonded atoms (Å2) with PBST, the plates were incubated in duplicates with sera from the Main chain 4.6 pollen allergic patients diluted 1/5 in PBST/0.5% w/v BSA and, for control Side chain 7.1 purposes, with serum from a nonallergic individual overnight at 4°C. Ramachandran plot (non-Gly Bound IgE Abs were detected with mouse monoclonal anti-human IgE Abs or Pro), residues in region (%) (BD Pharmingen) and with a HRP-linked sheep anti-mouse IgG (GE Most favored region 90.5 Healthcare). OD (OD at 405 nm; reference filter, 490 nm) was determined Additionally allowed region 8.8 by guest on September 28, 2021 with an ELISA reader (Spectramax PLUS; Molecular Devices). The results Generously allowed region 0.5 represent means of duplicate determinations with variations of less Disallowed region 0.2 than 10%. a Number in parentheses are statistics for the data in the highest resolution shell. b R ϭ⌺⌺I h i Ϫ I h ⌺⌺I h i Production of huMab2 Fab by papain digestion and Phl p 2/ sym ( ) ( ( ))/ ( )i, observed intensity in the th data set and (I(h)), mean intensity of reflection h over all measurements of I(h). c Fab complex formation Rcryst is the conventional R factor and Rfree is the R factor calculated with 5% of the data that were not used in refinement. Purified Phl p 2-specific IgG1 Ab was dialyzed against 30 mM HEPES buffer (pH 6.8) containing 1 mM EDTA. Papain was added (1:33 w/w ratio) in the presence of 10 mM freshly prepared cysteine, and digestion was conducted at 37°C for 4 h. The reaction was stopped by the addition of the specific papain inhibitor E-64 (N-[N-(L-3-trans-carboxirane-2-car- P42212 using program MOSFLM (34). There is one Phl p 2/Fab complex ϭ bonyl)-leucyl]-agmatine; Roche) in large excess. The digestion mixture per asymmetric unit, resulting in a solvent content of 50% (Vm 2.48 was dialyzed against 30 mM HEPES buffer (pH 6.8) loaded on a Mono S Å3/Da) (35). cation exchange column (Amersham Biosciences) and eluted with a salt gradient 0–500 mM NaCl. The Fab was eluted at 350 mM NaCl and its Structure determination and analysis purity confirmed by SDS-PAGE under both reducing and non-reducing conditions. The purified Fabs were mixed with Phlp2in1:1.2 molar ratio The structure of the Phl p 2/Fab complex was determined by molecular and incubated at 23°C for 60 min. The Phl p 2/Fab complex was separated replacement using the program MOLREP (36). In the first step, the posi- from excess Phlp2byusing a Superdex S-75 16/60 gel filtration column tions and orientation of the variable, V, and constant, C, Fab domains were (Amersham Biosciences). The protein concentration of the Phl p 2/Fab determined using the structure of the Fab domain with Protein Data Bank complex was calculated from the absorbance measured at 280 nm, assum- (PDB) code 1U6A (37). To avoid difficulties related to the variability in the ing an extinction coefficient of 0.7 mg/mlϪ1 cmϪ1. relative orientations of the V and C domains, the search procedure for the two domains was performed separately. After placing the V and C do- Crystallization and data collection mains, the position of the Phl p 2 was found by using the structure of the major grass pollen allergen Phl p 2 (PDB code: 1WHO) as a search model.

For the crystallization experiments, the purified Phl p 2/Fab complex in 30 Rigid body refinement with five bodies (Phl p 2, VL,CL,VH, and CH) with mM HEPES (pH 6.8) containing 150 mM NaCl was concentrated to 14.5 REFMAC (36) gave an initial R/Rfree of 43.3/43.8%. Manual adjustment of mg mlϪ1. The large crystals were grown by the hanging-drop vapor dif- the model and replacement of the model amino acid sequences with that of fusion method within 2–3 days under the following conditions: equal vol- Fab were performed with program O (38). This was followed by restrained umes (1.0 ␮l) of the Phl p 2/Fab complex and precipitant 20% polyethylene maximum-likelihood refinement with REFMAC and the addition of water glycol (PEG) 3350, 0.2 M NaF (condition 1, PEG/ion screen kit) were molecules by the program ARP, resulting in the convergence of R/Rfree mixed and equilibrated over the latter solution at 20°C. Before data col- value to 17.7/22.6% at 1.9 Å resolution (36). The stereochemistry of the lection, crystals were soaked briefly in the cryoprotective solution (precip- refined structure was validated with program PROCHECK (39), which itant solution plus 15% PEG 400 and 150 mM NaCl) and the diffraction showed that only 0.2% residues are in disallowed regions of a Ramachan- data were collected to a 1.9 Å resolution at SLS (Swiss Light Source), dran plot. The final model comprises residues V4–P94 of Phl p 2, E1– using PILATUS detector (␭ ϭ 0.800 Å). All measurements were per- G213 of the L chain and V1–P213 of the H chain of the Fab, as well as 524 formed at 100 K. The images were indexed and integrated in space group water molecules. For Phl p 2, the electron density is missing for three The Journal of Immunology 2145

FIGURE 3. Structure of the Phl p 2/huMab2 Fab complex. A, In this ribbon representation Phlp2isshown in magenta and H and L chains of Fab are colored yel- low and green, respectively. Epitope residues Downloaded from ofPhlp2incontact with Fab variable do- mains are shown as stick model. The ␤ strands of Phl p 2 are numbered 1–8 and the polypeptide chain termini are labeled with N and C for all three polypeptides. B, Location of the Phl p 2 epitope on the surface of the

4-stranded ␤-sheet. The epitope residues http://www.jimmunol.org/ forming hydrogen bonds with huMab2 Fab are shown in magenta as stick models. Res- idues that are at least partially buried in the interface (pink) are mainly interacting with Fab either by apolar contacts (Table III) or water-mediated hydrogen bonds (Table IV). by guest on September 28, 2021

N-terminal residues whereas for Fab only weak electron density is ob- gorithm is based on using the accessibility of amino acids in an explicit served for residues L147–L154 of the L chain and for residues H131–H134 manner (41). DiscoTope predicts only one conformational epitope and it of the H chain. assigns the probability of being a part of an epitope to each protein residue. The buried surface area was calculated with the program PISA (protein interfaces, surfaces and assemblies) European Bioinformatics Institute Results (http://www.ebi.ac.uk/msd-srv/prot_int/pistart.html) (40) using a probe ra- dius of 1.4 Å. The Ag residues are defined as interfacial if the buried Specificity and high-affinity binding of huMab2 to Phl p 2 2 surface area of a residue upon complex formation is larger than 1 Å . The interaction of huMab2 with its corresponding allergen was Interactions across the Phl p 2-Fab interface were identified using the pro- gram CONTACT (36). assessed by surface plasmon resonance-based assays (Fig. 1). The binding of 2-fold increasing concentrations of rPhl p 2 (0.3–20 Epitope prediction nM) was recorded for the flow cell containing immobilized The knowledge of epitopes of a given allergen is of great importance huMab2 and the reference flow cell. The Ab shows a fast onset of for the development of improved diagnostics and specific allergy vaccines. binding and almost no dissociation. Association (kon) and dissoci- ϭ ϫ 5 Presently, two methods are available for the prediction of B cell epitopes ation (koff) rate constants were determined to be kon 2.96 10 and are based on the knowledge of the three-dimensional structure of an Ag Ϫ1 Ϫ1 ϭ ϫ Ϫ5 Ϫ1 M s and koff 3.38 10 s resulting in a very low dis- alone, conformational epitope prediction (CEP) (41) and DiscoTope (42). ϭ ϫ Ϫ10 Both methods predict the Ag’s surface area that is in contact with Ab sociation constant of KD 1.14 10 M. (epitope). The CEP predicts the antigenic determinants, the conformational huMab2 cross-reacted with homologous group 2 allergens from (CE) and sequential (SE) B cell epitopes which do not overlap; CEP al- several grasses but did not bind to rPhl p 1 (28). Also huMab2 did 2146 X-RAY CRYSTAL STRUCTURE OF AN ALLERGEN/IgE COMPLEX not bind to synthetic Phl p 2-derived peptides (peptide 1, aa 1–33; Table III. Phl p 2/Fab interactions: direct contactsa peptide 2, 34–65; peptide 3, 66–96) or to a recombinant Phl p 2 mosaic protein with altered fold consisting of reassembled Phl p 2 Hydrogen Distance peptides 1-3-2 (32). bonds Fab Phl p 2 (Å) huMab2 inhibits allergic patients‘ IgE binding to the allergen CDR-L1 Gln L27 OE1 Glu A40 OE2 3.1 Tyr L32 OH Glu A32 OE2 2.5 ELISA plate-bound rPhl p 2 was preincubated with huMab2 to CDR-L3 Ser L91 O Arg A67 NH1 2.7 study its effect on IgE binding using the sera from 29 grass pollen Ser L91 O Arg A67 NH2 2.9 Leu L92 O Trp A41 N 2.9 allergic patients and, for control purposes, serum from a non- Ala L94 N Asp A39 O 2.7 allergic individual. huMab2 strongly inhibited IgE binding to rPhl Tyr L96 OH Trp A41 NE1 3.1 p 2 (mean inhibition: 64.8%) (Table I). Preincubation of rPhl p 2 CDR-H1 Gly H32 O Val A77 N 2.9 with an Ab specific for an immunologically unrelated allergen (Phl Gly H32 O Asn A76 OD1 3.0 p 5) had no influence on patients‘ IgE binding to Phl p 2 (Table I). Tyr H34 OH ASP A79 OD1 2.4 Tyr H34 OH Arg A34 NH2 3.1 Next, we compared the inhibition obtained with huMab2 with the CDR-H2 Tyr H59 OH Arg A34 NH2 3.2 inhibition obtained with polyclonal Phl p 2-specific rabbit Abs. Tyr H59 OH Asp A79 OD2 2.5 Polyclonal rabbit anti-Phl p 2 Abs yielded only a slightly better CDR-H3 Gly H101 O Arg A67 NH1 2.9 inhibition of IgE reactivity (mean inhibition 75.5%) than huMab2. Apolar interactions CDR-H1 Gly H32 CA Phe A78 CE1 3.7 Phl p 5-specific rabbit Abs did not show any inhibition (Table I). Gly H32 CA Phe A78 CZ 3.6 Tyr H34 CZ Val A77 CG2 3.6 Downloaded from huMab2 inhibits allergen-induced degranulation of basophils Tyr H34 CE1 Val A77 CG2 3.7 The effect of huMab2 on basophil degranulation induced by rPhl p Tyr H34 CE2 Asp A79 CG 3.5 2 was investigated using the humanized rat basophilic cell line CDR-H2 Tyr H53 CD2 Asp A79 CB 3.6 CDR-H3 Gly H101 CA Val A77 CG1 3.6 RBL 30/25. Cells were passively sensitized with sera from grass Gly H101 CA Val A77 CB 3.6 pollen allergic patients and challenged with Phl p 2 (Fig. 2). Tyr H102 CZ Arg A67 CG 3.7 huMab2 completely suppressed degranulation of RBL 30/25 cells Tyr H102 CZ Arg A67 CD 3.5 http://www.jimmunol.org/ for the whole range of allergen concentrations tested (1 ng–1 ␮g/ Tyr H102 CE2 Arg A67 CD 3.7 Tyr H102 CE2 Arg A67 CZ 3.5 ml), as shown in four representative patients. Similar results were Tyr H102 CD1 Lys A75 CD 3.7 obtained for 11 additional patients (data not shown). The addition a The cutoff distances for polar and apolar contacts are 3.25 Å and 3.75 Å, of a Phl p 1-specific Ab did not inhibit Phl p 2-induced degranu- respectively. lation at all. When rPhl p 2 was preincubated with polyclonal Phl p 2-specific rabbit Abs the release of mediators was also sup- pressed but to a lower extent than that achieved with huMab2. Rabbit Abs specific for the unrelated allergen Phl p 5 did not ␤ 7 (M74, K75, N76, V77, F78). Additional epitope residues D39, by guest on September 28, 2021 inhibit the mediator release. E40, W41, G73, D79, D80 are located in the loops joining the Crystal structure of the Phl p 2/huMab2 Fab complex strands and the C-terminal residue P94 (Fig. 3B). The Ag-com- bining site of the Fab is moderately concave with protuberances The crystal structure of the (1:1) complex of Phl p 2 with a and depressions formed by the side chains, which are complemen- huMab2 Fab was determined to 1.9 Å resolution by molecular tary to those of the Ag thus enabling a tight fit between the two replacement (Table II). It was possible to place one Phl p 2/Fab molecules (Figs. 3A and 4C). The shape correlation parameter S complex per crystallographic asymmetric unit. The final model, c for the complex, in the absence of water molecules, is 0.71, which refined to 1.9 Å resolution, is characterized by an R factor of is slightly higher than the observed mean S value of 0.64–0.68 for 17.7% (R , 22.6%) and the structure is well ordered with corre- c free Ag-Ab complexes (45), indicating good shape complementarity. sponding excellent electron density and good stereochemistry (Ta- The solvent accessible surface area of Phl p 2 which becomes ble II). The overall structure of the Phl p 2/huMab2-Fab complex buried upon complex formation is 855 Å2 which accounts for is shown in Fig. 3A.ThePhlp2isacompressed anti-parallel 15.4% of the total accessible molecular surface (5561 Å2). The barrel built by two four-stranded ␤-sheets sandwiched against each corresponding buried area in the Fab is 891 Å2 whereby the buried other, a folding typical for immunoglobulins. In the complex, the area of the H chain (544 Å2) is larger than that of the L chain (347 allergen is directed toward the center of the V /V dimer, as usu- H L Å2), as is usually observed in immune-complexes. ally observed in Ag/Ab complexes. The Fab elbow angles between the variable and constant domains of L and H chains are 139° and 140°, respectively; values which are frequently found for Fab that Mode of the Phl p 2/huMab2 Fab interactions have a ␬ L chain (26, 43). The superposition of free (44) and The allergen-Ab interactions involves 14 hydrogen bonds, 89 van complexed Phl p 2 results in a root mean square deviation of 0.56 der Waals interactions, 13 apolar contacts, and no salt bridges were Å for 94 C␣ atoms, indicating that the overall conformation of Phl found below 3.8 Å. Of the 21 epitope residues, 9 form hydrogen p 2 has not changed significantly upon complex formation. bonds with residues from five CDRs (L1, L3, H1, H2, H3), (Table III, Fig. 4A), and 12 are in apolar contacts with residues of H1, H2, The Phl p 2/huMab2-Fab interface H3 (Table III). No interactions were observed with residues from The interface between Phl p 2 and huMab2 Fab is extensive and CDR L2, which is the farthest from the bound epitope (Fig. 3A), as involves 21 allergen residues and 25 Fab residues, together with 8 frequently observed in Ag/Ab complexes (46). In addition, there water molecules. The Ab recognizes a discontinuous epitope are 8 water molecules in the complex interface (Table IV) that which is the four-stranded ␤-sheet composed of strands ␤3, ␤4, form 19 hydrogen bonds that bridge Phl p 2 residues with residues ␤6, and ␤7 (Fig. 3B). Most of the epitope residues (67%) are from all six CDRs and one residue from the framework region located within four non-contiguous segments belonging to strands FWR-L2 (Y49) thus improving the fit between the two surfaces ␤3 (E30, E32, D34, H36), ␤4 (A43), ␤6 (N65, R67, F68, L69) and (Fig. 4B). The Journal of Immunology 2147

FIGURE 4. The Phl p 2/huMab2 Fab complex interface. A, Stereo view of the hy- drogen bonding interactions between Phl p 2

and huMab2 Fab shown as dashed lines (ma- Downloaded from genta). Color code as in Fig. 3A. B, Stereo view of water-mediated hydrogen bonding interactions across the Phl p 2/huMab2 Fab interface. Water molecules are shown as spheres (cyan) and hydrogen bonds as dashed lines (red). Color code as in Fig. 3A.

C, Tight packing between selected residues http://www.jimmunol.org/ in the Phl p 2/huMab2 Fab interface. The side chain of hydrophobic residue W41 (Phl p 2) appears sandwiched by the alkyl chains of R34 and R67 (Phl p 2) on one side and on the other by polar contacts with residues L92 and Y96 from CDR L3. The side chain of Y102 (H3) protrudes deeply into a small epitope cavity formed by the side chains of R67, L69, and K75 and main chain atoms of by guest on September 28, 2021 F68 of the allergen. The side chains are shown with their van der Waals radii. Color code as in Fig. 3A.

The Phl p 2 epitope is formed by 8 polar (E32, R34, D39, E40, and the nearby K75, located in the epitope center, extend into the

R67, K75, N76, D79) and 3 hydrophobic (W41, V77, F78) resi- VH/VL groove where they interact with polar residues from L3 and dues (Table III). The residues in the center of the epitope are fully H3 (Table III). Likewise, the side chain of Y102 (H3) protrudes buried; the highest contribution to the buried surface of Phl p 2 deeply into a small epitope cavity formed by the apolar side chains (53%) is provided by residues K75, R67, V77, D79, and W41 (all of R67, L69, and K75 and the main chain of F68; the hydroxyl fully buried except for K75 which is 90% buried). W41, V77, and group of Y102 forms two water-mediated hydrogen bonds with F78 form a continuous hydrophobic patch in the center of the E30 and E32 (Table IV, Fig. 4C). Thus, the two hydrophobic res- contact region and are surrounded mainly by the hydrophilic res- idues W41 (Phl p 2) and Y102 (H3) appear sandwiched by alkyl idues of the epitope and CDR domains (Fig. 4A). The indole ring parts of the side chains of epitope residues R34, R67, and K75. of W41 is packed against the alkyl portions of the side chains of The Ag-combining site of the Fab is formed by the side chains R34 and R67 (Table III, Fig. 4C). The side chain of the buried R67 of the predominantly polar residues from the L and H chain CDRs 2148 X-RAY CRYSTAL STRUCTURE OF AN ALLERGEN/IgE COMPLEX

Table IV. Phl p 2/Fab interactions: water-mediated contactsa the CDR H3 loop is longer in the BLG/D1 Fab complex. The L chains in both Abs are very similar with respect to the sequence Distance Water Distance and conformation whereas the H chains are very different and are Fab (Å) Molecule (Å) Phl p 2 probably responsible for the specificity of binding. The latter as- CDR-L1 Gln L27 OE1 3.2 454O 2.7 Glu A40 OE2 sumption is supported by the observation that the L chains of cer- Tyr L32 OH 3.0 367O 2.8 Trp A41 O tain allergen-specific IgE Abs can be exchanged without losing FWR-L2 Tyr L49 OH 2.7 172O 2.7 Gly A73 O Tyr L49 OH 2.7 172O 2.8 Lys A75 NZ specificity for the allergens (49). Furthermore certain animals such CDR-L2 Gln L55 OE1 2.8 99O 2.9 Lys A75 NZ as camels and lamas contain Abs that are formed by H chains only CDR-L3 Tyr L96 OH 3.2 346O 3.1 Arg A34 NH2 CDR-H1 Tyr H33 OH 2.8 23O 2.6 Lys A75 O (50). CDR-H2 Tyr H51 OH 2.9 346O 3.1 Arg A34 NH2 Tyr H59 OH 2.6 394O 2.7 Asn A65 ND2 CDR-H3 Tyr H102 OH 2.7 85O 2.9 Glu A32 OE2 Tyr H102 OH 2.7 85O 2.7 Glu A30 OE1 Comparison of allergen/IgG Fab and allergen/IgE Fab Thr H103 OG1 3.1 99O 2.9 Lys A75 NZ complexes a The cutoff distance is 3.25 Å. The structures of only four allergens in complex with specific Fab Ab fragments have so far been determined by x-ray crystallogra- (12 and 13 residues, respectively). According to the standard Ka- phy: the two complexes of the major allergens from birch pollen bat definitions (47), the CDRs of huMab2 comprise the following (Bet v 1) (24) and bee venom (Api m 2) (25) with specific murine IgG1 Fabs, and the two complexes of the major allergens from residues: CDR-L1 (R24-N34), CDR-L2 (A50-S56), CDR-L3 Downloaded from (Q89-T97), CDR-H1 (S31-S36), CDR-H2 (Y51-S66) and grass pollen (Phl p 2), this study, and bovine milk (BLG) (26) with CDR-H3 (L99-I106). Of the 17 Ab residues interacting with Phl p their respective human IgE Fabs (Fig. 5). In both IgG Fab-allergen 2, nine are tyrosines (Tables III and IV) which is in agreement with complexes the allergen epitopes have distinct protruding shapes: a ␤ the observation that aromatic residues, particularly tyrosines, form hairpin of Bet v 1 and a helix-loop-helix motif of Api m 2. most of the contacts with Ags (46, 48). Generally, the H chain Another similarity is that both epitopes have a pronounced linear character; the Api m 2 epitope consists of 12 consecutive residues

contributes more to the binding of an Ag than the L chain (46). http://www.jimmunol.org/ Within the complex, each of the H and L chains form 7 hydrogen whereas in Bet v 1, 11 of 17 epitope residues are part of a con- bonds with Ag whereas the 13 apolar contacts with Ag involve tinuous segment (42–52) that contains all the residues that form only residues from three H chain CDRs (Table III). hydrogen bonds with the Ab and covers 80% of the surface area buried upon complex formation (24). Comparison of Phl p 2/huMab2 Fab and ␤-lactoglobulin/D1 In marked contrast, the IgE binding epitopes of BLG and Phl p Fab structure 2 are distinctly discontinuous. The epitope of BLG is remarkably The structures of the two complexes of human Abs huMab2 and flat although it is moderately convex in the Phl p 2 allergen. Be- D1 (26) with their respective Ags showed that the IgE-defined cause of the high surface complementarity of the Ag-Ab interface, epitope of Phl p 2 consists of a four-stranded anti-parallel ␤-sheet the shapes of the Ag-combining sites in BLG and Phl p 2 com- by guest on September 28, 2021 formed by 21 residues whereas in ␤-lactoglobulin (BLG) a larger plexes are flat and moderately concave, respectively. Generally, part of the epitope is a four-stranded ␤-sheet composed of 16 res- there is a correlation between the overall shape of the Ag-com- idues whereas a smaller part includes an additional 11 residues bining site and the nature of Ags; deep pockets are observed with located in the loops and ␣-helix. The two IgE-defined epitopes , grooves with peptides and flat combining sites with pro- share a number of similar features: a percentage of residues located teins (51). It has been proposed that the shape of the Ag-combining within the four-stranded ␤ strands (67 vs 74% in BLG complex), site is determined mainly by the lengths of the CDRs (51, 52). In a size of the buried surface area (855 vs 890 Å2) and a number of particular, the short CDR-H3 is more commonly seen in anti- hydrogen bonds (14 vs 13). peptide and anti- Abs whereas the longer CDR-H3 loops A comparison of the Ab combining sites of the two complexes tend to favor large Ags, such as , and implies a flat (or showed differences in the number of engaged CDRs (5 vs 6 in protruding) Ag-combining site. In the Phl p 2/Fab complex BLG complex), number of aromatic residues involved (9 vs 2) and CDR-H3 is composed of eight residues and the Ag-combining site

FIGURE 5. Ribbon presentation of known struc- tures of allergens in complex with IgE or IgG Fabs. Allergens from grass pollen (Phl p 2), bovine milk (BLG), birch pollen (Bet v 1) and honey bee venom (Api m 2) are colored magenta while the H and L chains of Fab are colored yellow and green, respec- tively. The protruding shapes of IgG binding epitopes are apparent. The Journal of Immunology 2149

FIGURE 6. Sequence alignment of Phl p 2 and its grass pollen homologs from group 2/3 (A) and group 1 (B). A, The sequence conservation of nine Phl p 2 epitope residues forming hydrogen bonds with IgE Fab (Table III) were examined; these residues are either fully conserved (yellow) or are conservatively substituted (blue). Epitope residues partially buried in the Phl p 2/IgE Fab complex but not forming hydrogen bonds are printed on a gray Downloaded from background. B, The conservation of the Phl p 2 epitope residues, involved in hydrogen bonding with Fab, among group 1 allergens; fully conserved and conservatively substituted residues are printed on yellow and blue background, respectively, whereas the remaining epitope residues are printed on gray background. The numbering of the amino acids is indicated on the top of the alignment for Phl p 2.

is moderately protruding while in the BLG/Fab complex a dis- the allergen. This can be inferred from the fact that the complete http://www.jimmunol.org/ tinctly flat Ag-combining site is seen, which correlates well with Ab huMab2, a hybrid engineered from the variable domain of the the longer CDR-H3 loops composed of 17 residues. IgE Fab and the constant domain of human IgG1, inhibited the polyclonal grass pollen allergic patients’ IgE reactivity to the Phl Structural basis for cross-reactivity p 2 allergen up to 91% and efficiently blocked allergen-induced Patients allergic to grass pollen frequently display sensitivity to basophil degranulation. This strong inhibition indicates that many various grasses, indicating that sensitization to one pollen may lead of the patients IgEs are directed to or close to the same, dominant, to multiple grass-pollen allergies. There is an extensive cross-re- epitope of Phl p 2, which is recognized by huMab2, and thus activity of allergic patients’ IgE Abs and of huMab2 with group 2 huMab2 has been considered as an IgE-blocking Ab with thera- and 3 grass pollen allergens which share a significant sequence peutic potential (28). One reason for the strong blocking of IgE by guest on September 28, 2021 identity (61%) (28, 53, 54). However, no relevant IgE cross-reac- binding to the allergen by huMab2 may be that the IgE repertoire tivity was found between group 2/3 and group 1 allergens (54, 55) of grass pollen allergic patients may be less diverse than previ- despite a considerable sequence homology between the C-terminal ously anticipated (56). Indeed, epitope mapping studies performed part of group 1 allergens and the entire sequence of group 2/3 with several respiratory allergens indicate that an allergen harbors allergens (40% identity) (53, 55). To study whether the structurally only a few distinct IgE epitopes (31). As a result of such an oli- identified IgE epitope of Phl p 2 is conserved among allergens of goclonal IgE response to allergens, it is possible that single mAbs group 2 and 3, we examined the conservation of nine epitope res- can strongly block IgE recognition of allergens. A similar obser- idues which are hydrogen bonded to the IgE Fab (Table III) in vation has also been made for a human monoclonal IgG Ab, group 2/3 allergens from six grass species. As shown in Fig. 6A, BAB1, which strongly blocks IgE binding to the major birch pol- seven epitope residues are fully conserved (E32, E40, W41, R67, len allergen Bet v 1 (57). N76, V77, D79) whereas R34 and D39 contain conservative sub- In contrast, it has been demonstrated that allergic patients con- stitutions, suggesting that the identified ␤-sheet epitope is highly tain quite high levels of allergen-specific IgG Abs that cannot conserved among group 2/3 grass pollen allergens. The alignment block IgE binding and IgE-mediated effector cell activation, de- of the Phl p 2 sequence with nine homologous sequences of the spite the fact that allergen-specific IgE Ab levels are normally 100 C-terminal domains of group 1 grass pollen allergens revealed full to 1000 times lower than allergen-specific IgG levels (17). In conservation of only three epitope residues D39, W41, and R67 fact, very high affinity of IgE binding to Phl p 2 revealed by our and conservative substitutions for residues E32 and R34, thus ex- study may provide answers to the question why allergen-spe- plaining the lack of relevant cross-reactivity between group 2/3 cific IgE Abs, despite their extremely low concentrations, are so and group 1 allergens (Fig. 6B). efficient in activating effector cells and thus inflammatory re- sponses and cannot be competed easily even by an abundant Discussion amount of allergen-specific IgG. Here we present the three-dimensional structure of a complex be- The structural analysis of the allergen/IgE- tween a major respiratory allergen, the timothy grass pollen aller- showed that the IgE Fab recognizes a conformational Phl p 2 gen Phl p 2 and a Phl p 2-specific Fab derived from an IgE Ab, epitope composed of residues grouped in four sequentially distant which was isolated from a combinatorial library constructed from segments which are located on the four-stranded ␤-sheet. When we a grass pollen allergic patients’ lymphocytes (27). The IgE Fab submitted the coordinates of Phl p 2 (PDB: 1who) to CEP (41) and consists of an authentic IgE H chain fragment that had recombined DiscoTope (42), servers that are currently used for epitope predic- randomly with a L chain from the same patient and therefore re- tion, CEP predicted a total of six antigenic determinants, five con- sembles but is not identical with a genuine human IgE Fab. Nev- formational epitopes, and one sequential epitope. One predicted ertheless, the isolated IgE Fab reacts with a major IgE epitope of epitope (CE4) was found to have a maximum overlap with the 2150 X-RAY CRYSTAL STRUCTURE OF AN ALLERGEN/IgE COMPLEX experimentally characterized binding site; that is, 18 of 21 epitope immunization with partially denatured or degraded allergens, residues were correctly predicted (86%). In contrast, DiscoTope which may occur upon linking of Ags to adjuvants. In contrast, predicted correctly only 6 of 21 residues (29%). The reason for this major respiratory allergens to which patients become sensitized are low prediction success is probably due to the fact that the epitope presented as soluble proteins to the mucosal tissues in native con- residues are a part of a flat region ␤-sheet, not protruding from the formation. Pollen allergens, for example, are eluted within minutes surface, which will negatively influence epitope prediction by Dis- from pollen as intact and soluble proteins (63). Allergens are often coTope emphasizing the need for experimental epitope data. also very resistant against denaturation by heat or other conditions A similar ␤-sheet epitope as found for Phl p 2 has been recently such as digestion in the gut in the case of food allergens (64). The observed in the structure of BLG, a major food allergen, in com- structurally identified conformational epitope of Phl p 2 and its plex with a specific human IgE Fab designated D1 (26). Also this tight binding to huMab2, closely mimicking the binding of patients Fab has been obtained by combinatorial cloning and therefore is IgE, seems to be a paradigmatic example of the high-affinity rec- not identical with a genuine human IgE Fab. But, this Fab and the ognition of conformational epitopes of allergens by IgE and ex- Fab described by us recognize epitopes that may be different from plains why the IgE system is so potent in triggering strong inflam- those epitopes observed in allergen/IgG complexes. The latter have matory reactions despite low abundance of IgE. been shown to have a protruding topology and are often continu- In conclusion, our data suggest that the tight and efficient bind- ous (24, 25). Many epitope properties revealed by the structures of ing of allergens by IgE and by IgE/Fc␧RI complexes on mast cells, Phl p 2/huMab2 Fab and BLG/D1 Fab complexes have also been basophils, eosinophils, and APCs results in formation of excep- observed in Ag/IgG complexes (46). However, the observation tionally stable immune complexes that efficiently provoke cas- that four-stranded anti-parallel ␤-sheet epitopes of Phlp2and cades of inflammatory cell activation. Ultimately, a detailed Downloaded from BLG comprise residues mainly located in the secondary structure knowledge of the allergen IgE interaction may lead to the devel- elements (67 and 74%, respectively) is a feature not observed opment of therapeutic strategies for type I hypersensitivity reac- among the recently compared 82 structures of protein-Ab com- tions based on the interference with IgE recognition of allergens. plexes (26) despite the frequent occurrence of ␤-structures in the Protein Data Bank. Acknowledgments

The interaction between the two already solved IgE-allergen We thank the staff of the synchrotron beam line PX at SLS in Villigen, http://www.jimmunol.org/ immune complexes, Phl p 2/huMab2 Fab and BLG/D1 Fab, re- Switzerland. vealed a remarkably high binding affinity which differentiates them from the generally lower affinities of many IgG-allergen immune Disclosures complexes. In fact, the affinity constants of Phl p 2/huMab2 com- The authors have no financial conflict of interest. ϭ ϫ 10 Ϫ1 ϭ ϫ plex (KA 0.9 10 M ) and BLG/D2 complex (KA 0.7 9 Ϫ1 References 10 M ) are of a magnitude similar to that of the constant region 1. Kay, A. B., J. Bousquet, and P. Holt. 2008. Allergy and Allergic Disease. Black- of IgE and the corresponding high-affinity receptor, Fc␧RI (ϳ109– well Science, Oxford, U.K. 1010 MϪ1) (16). The identification of major interactions involved 2. Ishizaka, K., T. Ishizaka, and M. M. Hornbrook. 1966. Physico-chemical prop-

erties of human reaginic antibody, IV: presence of a unique immunoglobulin as by guest on September 28, 2021 in the protein-protein association process is still an area of inten- a carrier of reaginic activity. J. Immunol. 97: 75–85. sive research. In principle, protein epitopes are large enough to 3. Johansson, S. G., and H. Bennich. 1967. Immunological studies of an atypical provide a high affinity through various sets of interactions. For (myeloma) immunoglobulin. Immunology 13: 381–394. 4. Kraft, S., and J. P. Kinet. 2007. New developments in Fc␧RI regulation, function example, the binding free energy can be generated through only a and inhibition. Nat. Rev. Immunol. 7: 365–378. few strong interactions or through the sum of many contacts over 5. Gounni, A. S., B. Lamkhioued, K. Ochiai, Y. Tanaka, E. Delaporte, A. Capron, J. P. Kinet, and M. Capron. 1994. High-affinity IgE receptor on eosinophils is the entire surface. involved in defence against parasites. Nature 367: 183–186. In the Phl p 2/huMab2 Fab complex a number of hydrogen 6. Blank, U., and J. Rivera. 2004. The ins and outs of IgE-dependent mast-cell bonds and van der Waals interactions could contribute signifi- exocytosis. Trends Immunol. 25: 266–273. 7. Bischoff, S. C. 2007. Role of mast cells in allergic and nonallergic immune cantly to the binding energy by generating favorable enthalpic ef- responses: comparison of human and murine data. Nat. Rev. Immunol. 7: 93–104. fects. The high structural and chemical complementarity of the 8. Wurzburg, B. A., S. S. Tarchevskaya, and T. S. Jardetzky. 2006. Structural epitope/ surfaces in Phl p 2/huMab2 Fab complex also changes in the lectin domain of CD23, the low-affinity IgE receptor, upon cal- cium binding. Structure 14: 1049–1058. contributes to the stability of the complex and has been described 9. Hibbert, R. G., P. Teriete, G. J. Grundy, R. L. Beavil, R. Reljic, V. M. Holers, as an important determinant of specificity of interactions and bind- J. P. Hannan, B. J. Sutton, H. J. Gould, and J. M. McDonnell. 2005. The structure of human CD23 and its interactions with IgE and CD21. J. Exp. Med. 202: ing affinity (46). The complementarity is additionally enhanced by 751–760. the presence of water molecules in the Ag/Ab interface, which 10. van der Heijden, F. L., R. J. van Neerven, and M. L. Kapsenberg. 1995. Rela- mediate hydrogen bonds between the Ag and Ab. Water molecules tionship between facilitated allergen presentation and the presence of allergen- specific IgE in serum of atopic patients. Clin. Exp. Immunol. 99: 289–293. are frequently found in Ag/Ab and protein/protein interfaces (58– 11. van Neerven, R. J., E. F. Knol, A. Ejrnaes, and P. A. Wurtzen. 2006. IgE-me- 60) where they contribute to a complex stability by neutralizing the diated allergen presentation and blocking antibodies: regulation of T-cell activa- destabilizing effects of unpaired hydrogen bonding groups (61) and tion in allergy. Int. Arch. Allergy Immunol. 141: 119–129. 12. Naclerio, R. M., N. F. Adkinson, Jr., B. Moylan, F. M. Baroody, D. Proud, by improving the contact between the interacting proteins. Crys- A. Kagey-Sobotka, L. M. Lichtenstein, and R. Hamilton. 1997. Nasal provocation tallographic studies of lysozyme complexes with specific Abs Hy- with allergen induces a secondary serum IgE antibody response. J. Allergy Clin. Immunol. 100: 505–510. Hel-5 (60) and F10.6.6 (62) and D44.1 revealed seven and four 13. Niederberger, V., J. Ring, J. Rakoski, S. Jager, S. Spitzauer, P. Valent, F. Horak, water molecules buried in the binding interface where they medi- M. Kundi, and R. Valenta. 2007. drive memory IgE responses in human ate hydrogen bonding interactions between Ag and Ab. allergy via the nasal mucosa. Int. Arch. Allergy Immunol. 142: 133–144. 14. Stingl, G., and D. Maurer. 1997. IgE-mediated allergen presentation via Fc␧RI on The hydrophobic effect also contributes to the stability of the Phl -presenting cells. Int. Arch. Allergy Immunol. 113: 24–29. p 2/huMab2 Fab complex particularly in the center of the epitope, 15. Bieber, T. 1997. Fc␧RI on human epidermal Langerhans cells: an old receptor which is hydrophobic and deprived of water molecules, whereas with new structure and functions. Int. Arch. Allergy Immunol. 113: 30–34. 16. Garman, S. C., B. A. Wurzburg, S. S. Tarchevskaya, J. P. Kinet, and this effect must be weak at the epitope periphery due to the sol- T. S. Jardetzky. 2000. Structure of the Fc fragment of human IgE bound to its vation by eight water molecules. high-affinity receptor Fc␧RI␣. Nature 406: 259–266. 17. Stern, D. A., J. Riedler, D. Nowak, C. Braun-Fahrlander, I. Swoboda, N. Balic, The continuous character of previously reported IgG-binding K. W. Chen, S. Vrtala, H. Gronlund, M. van Hage, et al. 2007. Exposure to a epitopes of Bet v 1 and Api m 2 may have resulted from the mouse farming environment has allergen-specific protective effects on TH2-dependent The Journal of Immunology 2151

isotype switching in response to common inhalants. J. Allergy Clin. Immunol. 41. Kulkarni-Kale, U., S. Bhosle, and A. S. Kolaskar. 2005. CEP: a conformational 119: 351–358. epitope prediction server. Nucleic Acids Res. 33: W168–W171. 18. Flicker, S., and R. Valenta. 2003. Renaissance of the blocking antibody concept 42. Andersen, P. H., N. Morten, and O. Lund. 2006. Prediction of residues in dis- in type I allergy. Int. Arch. Allergy Immunol. 132: 13–24. continuous B-cell epitopes using 3D structures. Protein Sci. 15: 2258–2567. 19. Ball, T., W. R. Sperr, P. Valent, J. Lidholm, S. Spitzauer, C. Ebner, D. Kraft, and 43. Stanfield, R. L., A. Zemla, I. A. Wilson, and B. Rupp. 2006. Antibody elbow R. Valenta. 1999. Induction of antibody responses to new B cell epitopes indi- angles are influenced by their light chain class. J. Mol. Biol. 357: 1566–1574. cates vaccination character of allergen immunotherapy. Eur. J. Immunol. 29: 44. Fedorov, A. A., T. Ball, N. M. Mahoney, R. Valenta, and S. C. Almo. 1997. The 2026–2036. molecular basis for allergen cross-reactivity: crystal structure and IgE-epitope 20. Denepoux, S., P. B. Eibensteiner, P. Steinberger, S. Vrtala, V. Visco, A. Weyer, mapping of birch pollen profilin. Structure 5: 33–45. D. Kraft, J. Banchereau, R. Valenta, and S. Lebecque. 2000. Molecular charac- 45. Lawrence, M. C., and P. M. Colman. 1993. Shape complementarity at protein/ terization of human IgG monoclonal antibodies specific for the major birch pollen protein interfaces. J. Mol. Biol. 234: 946–950. FEBS allergen Bet v 1: anti-allergen IgG can enhance the anaphylactic reaction. 46. Davies, D. R., E. A. Padlan, and S. Sheriff. 1990. Antibody-antigen complexes. Lett. 465: 39–46. Annu. Rev. Biochem. 59: 439–473. 21. Hantusch, B., I. Scholl, C. Harwanegg, S. Krieger, W. M. Becker, S. Spitzauer, 47. Kabat, E. A., T. T. Wu, and H. Bilofsky. 1977. Unusual distributions of amino G. Boltz-Nitulescu, and E. Jensen-Jarolim. 2005. Affinity determinations of pu- acids in complementarity-determining (hypervariable) segments of heavy and rified IgE and IgG antibodies against the major pollen allergens Phl p 5a and Bet light chains of immunoglobulins and their possible roles in specificity of anti- v 1a: discrepancy between IgE and IgG binding strength. Immunol. Lett. 97: body-combining sites. J. Biol. Chem. 252: 6609–6616. 81–89. 22. Jakobsen, C. G., U. Bodtger, L. K. Poulsen, and E. L. Roggen. 2005. Vaccination 48. Padlan, E. A. 1990. On the nature of antibody combining sites: unusual structural for birch pollen allergy: comparison of the affinities of specific immunoglobulins features that may confer on these sites an enhanced capacity for binding ligands. E, G1 and G4 measured by surface plasmon resonance. Clin. Exp. Allergy 35: Proteins 7: 112–124. 193–198. 49. Laffer, S., P. Steinberger, D. Kraft, and R. Valenta. 2001. Promiscuous use of 23. von Boehmer, H., and M. C. Nussenzweig. 2004. The manipulation of : light chains by human IgE antibodies specific for three major grass pollen aller- conference on from allergy to cancer: new perspectives for therapeutic vaccina- gens. Int. Arch. Allergy Immunol. 124: 29–30. tion. EMBO Rep. 5: 765–770. 50. Muyldermans, S., C. Cambillau, and L. Wyns. 2001. Recognition of antigens by 24. Mirza, O., A. Henriksen, H. Ipsen, J. N. Larsen, M. Wissenbach, single-domain antibody fragments: the superfluous luxury of paired domains. Downloaded from M. D. Spangfort, and M. Gajhede. 2000. Dominant epitopes and allergic cross- Trends Biochem. Sci. 26: 230–235. reactivity: complex formation between a Fab fragment of a monoclonal murine 51. MacCallum, R. M., A. C. Martin, and J. M. Thornton. 1996. Antibody-antigen IgG antibody and the major allergen from birch pollen Bet v 1. J. Immunol. 165: interactions: contact analysis and binding site topography. J. Mol. Biol. 262: 331–338. 732–745. 25. Padavattan, S., T. Schirmer, M. Schmidt, C. Akdis, R. Valenta, I. Mittermann, 52. Collis, A. V., A. P. Brouwer, and A. C. Martin. 2003. Analysis of the antigen L. Soldatova, J. Slater, U. Mueller, and Z. Markovic-Housley. 2007. Identifica- combining site: correlations between length and sequence composition of the tion of a B-cell epitope of hyaluronidase, a major bee venom allergen, from its hypervariable loops and the nature of the antigen. J. Mol. Biol. 325: 337–354.

crystal structure in complex with a specific Fab. J. Mol. Biol. 368: 742–752. 53. Ansari, A. A., T. K. Kihara, and D. G. Marsh. 1987. Immunochemical studies of http://www.jimmunol.org/ 26. Niemi, M., S. Jylha, M. L. Laukkanen, H. Soderlund, S. Makinen-Kiljunen, Lolium perenne (rye grass) pollen allergens, Lol p I, II, and III. J. Immunol. 139: J. M. Kallio, N. Hakulinen, T. Haahtela, K. Takkinen, and J. Rouvinen. 2007. 4034–4041. Molecular interactions between a recombinant IgE antibody and the ␤-lactoglob- 54. Niederberger, V., S. Laffer, R. Froschl, D. Kraft, H. Rumpold, S. Kapiotis, ulin allergen. Structure 15: 1413–1421. R. Valenta, and S. Spitzauer. 1998. IgE antibodies to recombinant pollen aller- 27. Steinberger, P., D. Kraft, and R. Valenta. 1996. Construction of a combinatorial gens (Phl p 1, Phl p 2, Phl p 5, and Bet v 2) account for a high percentage of grass IgE library from an allergic patient: isolation and characterization of human IgE pollen-specific IgE. J. Allergy Clin. Immunol. 101: 258–264. Fabs with specificity for the major timothy grass pollen allergen, Phl p 5. J. Biol. 55. Dolecek, C., S. Vrtala, S. Laffer, P. Steinberger, D. Kraft, O. Scheiner, and Chem. 271: 10967–10972. R. Valenta. 1993. Molecular characterization of Phl p II, a major timothy grass 28. Flicker, S., P. Steinberger, L. Norderhaug, W. R. Sperr, Y. Majlesi, P. Valent, (Phleum pratense) pollen allergen. FEBS Lett. 335: 299–304. D. Kraft, and R. Valenta. 2002. Conversion of grass pollen allergen-specific 56. Andreasson, U., S. Flicker, M. Lindstedt, R. Valenta, L. Greiff, M. Korsgren, human IgE into a protective IgG(1) antibody. Eur. J. Immunol. 32: 2156–2162. C. A. Borrebaeck, and M. Ohlin. 2006. The human IgE-encoding transcriptome 29. Flicker, S., P. Steinberger, T. Ball, M. T. Krauth, P. Verdino, P. Valent, S. Almo, to assess antibody repertoires and repertoire evolution. J. Mol. Biol. 362: by guest on September 28, 2021 and R. Valenta. 2006. Spatial clustering of the IgE epitopes on the major timothy 212–227. grass pollen allergen Phl p 1: importance for allergenic activity. J. Allergy Clin. 57. Visco, V., C. Dolecek, S. Denepoux, J. Le Mao, C. Guret, F. Rousset, Immunol. 117: 1336–1343. M. T. Guinnepain, D. Kraft, R. Valenta, A. Weyer, et al. 1996. Human IgG 30. Andersson, K., and J. Lidholm. 2003. Characteristics and immunobiology of monoclonal antibodies that modulate the binding of specific IgE to birch pollen grass pollen allergens. Int. Arch. Allergy Immunol. 130: 87–107. Betv1.J. Immunol. 157: 956–962. 31. Valenta, R., and D. Kraft. 2001. Recombinant allergen molecules: tools to study 58. Bhat, T. N., G. A. Bentley, G. Boulot, M. I. Greene, D. Tello, W. Dall’Acqua, effector cell activation. Immunol. Rev. 179: 119–127. H. Souchon, F. P. Schwarz, R. A. Mariuzza, and R. J. Poljak. 1994. Bound water 32. Mothes-Luksch, N., S. Stumvoll, B. Linhart, M. Focke, M. T. Krauth, molecules and conformational stabilization help mediate an antigen-antibody as- A. Hauswirth, P. Valent, P. Verdino, T. Pavkov, W. Keller, et al. 2008. Disrup- sociation. Proc. Natl. Acad. Sci. USA 91: 1089–1093. tion of allergenic activity of the major grass pollen allergen Phlp2byreassembly 59. Ysern, X., B. A. Fields, T. N. Bhat, F. A. Goldbaum, W. Dall’Acqua, as a mosaic protein. J. Immunol. 181: 4864–4873. F. P. Schwarz, R. J. Poljak, and R. A. Mariuzza. 1994. Solvent rearrangement in 33. Vogel, L., D. Luttkopf, L. Hatahet, D. Haustein, and S. Vieths. 2005. Develop- an antigen-antibody interface introduced by site-directed mutagenesis of the an- ment of a functional in vitro assay as a novel tool for the standardization of tibody combining site. J. Mol. Biol. 238: 496–500. allergen extracts in the human system. Allergy 60: 1021–1028. 34. Leslie, A. G. W. 1992. In joint CCP4 and ESF-EACBM newsletter on protein 60. Cohen, G. H., E. W. Silverton, E. A. Padlan, F. Dyda, J. A. Wibbenmeyer, crystallography no. 26. Technical report SERC Daresbury Laboratory, War- R. C. Willson, and D. R. Davies. 2005. Water molecules in the antibody-antigen rington, U.K. interface of the structure of the Fab HyHEL-5-lysozyme complex at 1.7 A res- 35. Matthews, B. W. 1968. Solvent content of protein crystals. J. Mol. Biol. 33: olution: comparison with results from isothermal titration calorimetry. Acta Crys- 491–497. tallogr. D Biol. Crystallogr. 61: 628–633. 36. CCP4. 1994. Collaborative computational project, number 4. The CCP4 suite: 61. Fersht, A. R., J. P. Shi, J. Knill-Jones, D. M. Lowe, A. J. Wilkinson, D. M. Blow, programs for protein crystallography. Acta Crystallogr. D Biol. Crystallogr. 50: P. Brick, P. Carter, M. M. Waye, and G. Winter. 1985. Hydrogen bonding and 760–763. biological specificity analysed by protein engineering. Nature 314: 235–238. 37. Wilkinson, R. A., C. Piscitelli, M. Teintze, L. A. Cavacini, M. R. Posner, and 62. Cauerhff, A., F. A. Goldbaum, and B. C. Braden. 2004. Structural mechanism for C. M. Lawrence. 2005. Structure of the Fab fragment of F105, a broadly reactive affinity maturation of an anti-lysozyme antibody. Proc. Natl. Acad. Sci. USA 101: anti-human immunodeficiency virus (HIV) antibody that recognizes the CD4 3539–3544. binding site of HIV type 1 gp120. J. Virol. 79: 13060–13069. 63. Vrtala, S., M. Grote, M. Duchene, R. van Ree, D. Kraft, O. Scheiner, and 38. Jones, T. A., J. Y. Zou, S. W. Cowan, and M. Kjeldgaard. 1991. Improved R. Valenta. 1993. Properties of tree and grass pollen allergens: reinvestigation of methods for building protein models in electron density maps and the location of the linkage between solubility and allergenicity. Int. Arch. Allergy Immunol. 102: errors in these models. Acta Crystallogr. A 47 (Pt 2): 110–119. 160–169. 39. Laskowski, R. A., M. W. MacArthur, D. S. Moss, and J. M. Thorton. 1993. 64. Bugajska-Schretter, A., M. Grote, L. Vangelista, P. Valent, W. R. Sperr, Procheck: a program to check the stereochemical quality of protein structures. H. Rumpold, A. Pastore, R. Reichelt, R. Valenta, and S. Spitzauer. 2000. Puri- J. Appl. Crystallogr. 26: 283–291. fication, biochemical, and immunological characterisation of a major food aller- 40. Krissinel, E., and K. Henrick. 2007. Inference of macromolecular assemblies gen: different recognition of the apo- and calcium-bound from crystalline state. J. Mol. Biol. 372: 774–797. forms of carp parvalbumin. Gut 46: 661–669.