Molecular Determinants for Antibody Binding on Group 1 House Dust Mite Allergens*

Background: A unique, cross-reacting monoclonal antibody binds both Der f 1 and Der p 1. Results: A common epitope present on both Der f 1 and Der p 1 was identified and mutated. Conclusion: Mutagenesis and antibody binding analysis allowed identification of IgE antibody binding sites. Significance: The obtained data will lead to the production of hypoallergens with low IgE antibody binding capacity. House dust mites produce potent allergens, Der p 1 and Der f 1, that cause allergic sensitization and asthma. Der p 1 and Der f 1 are cysteine proteases that elicit IgE responses in 80% of mite-allergic subjects and have proinflammatory properties. Their antigenic structure is unknown. Here, we present crystal structures of natural Der p 1 and Der f 1 in complex with a monoclonal antibody, 4C1, which binds to a unique cross-reactive epitope on both allergens associated with IgE recognition. The 4C1 epitope is formed by almost identical amino acid sequences and contact residues. Mutations of the contact residues abrogate mAb 4C1 binding and reduce IgE antibody binding. These surface-exposed residues are molecular targets that can be exploited for development of recombinant allergen vaccines.

Most mite allergic patients (Ͼ80%) have IgE antibodies against the Group 1 mite allergens, Der p 1 and Der f 1 (6,7). The Group 1 allergens are cysteine proteases, and their proteolytic activity contributes to allergenicity. Der p 1 is responsible for disruption of tight junctions in lung epithelium and cleavage of CD23 and CD25 receptors (8). Cleavage of these receptors favors a Th2 response and induction of release of proinflammatory cytokines from bronchial epithelial cells, mast cells, and basophils. The resulting increase in IgE antibody synthesis (9) and inflammation of lung epithelium may explain why mite allergens are strongly associated with asthma (10). Less is known about the proteolytic activity of Der f 1 on proinflammatory responses, but it has been reported that Der f 1 reduces the barrier function of the skin (11).
Der p 1 and Der f 1 share 81% sequence identity, so it is not surprising that human IgE antibody and T cell responses to the Group 1 allergens are highly cross-reactive (7,(12)(13)(14). Despite the high amino acid sequence identity between Group 1 allergens, most monoclonal antibodies raised against either Der p 1 or Der f 1 are species-specific (7,12). In this respect, the anti-Der f 1 monoclonal antibody 4C1 (12) is highly unusual because it also strongly binds Der p 1. Moreover, mAb 4C1 partially inhibits IgE antibody binding to Der p 1, suggesting that the epitopes for mAb 4C1 and human IgE antibodies overlap (12,14).
Here, we present structural analyses of natural Der f 1 and Der p 1 in complex with a Fab fragment of mAb 4C1 and the identification of functional residues involved in mAb and IgE antibody interaction with the allergens. A cross-reactive epitope between Der p 1 and Der f 1 was defined, and mutagenesis of residues in the epitope reduced IgE antibody binding. Identification of IgE binding epitopes provides a strategy for the design of modified allergen molecules for use in recombinant vaccines for the treatment of dust mite allergy. at 4°C. After incubation, the solution was concentrated using an Amicon Ultra concentrator (Millipore) with a 10,000-Da molecular mass cutoff and purified on a Superdex 200 column attached to an ÄKTA FPLC system (GE Healthcare). Slightly different solutions were used during gel filtration and for protein storage. A solution composed of 20 mM Tris-HCl, 150 mM NaCl, pH 7.4 was used for gel filtration of Der f 1-4C1 complex, whereas Der p 1-4C1 complex was purified using 10 mM Tris-HCl, 50 mM NaCl at pH 7.5. After gel filtration, fractions containing Der f 1-4C1 and Der p 1-4C1 were concentrated to 7 and 9 mg/ml, respectively. The 4C1 Fab fragment used for crystallization of the antibody fragment alone was also purified on Superdex 200 using the same buffer as for Der p 1-4C1 complex and concentrated to 9 mg/ml.
Der p 1 and Der f 1 mutants of the mAb 4C1 epitope were expressed in Pichia pastoris. Cells were grown in 1-liter Buffered Glycerol-complex Medium culture for 24 h. They were concentrated 5 times by centrifuging the culture at 3000 ϫ g for 5 min and resuspended in 200 ml of Buffered Methanol-complex Medium for methanol-induced expression of the allergens. Two of the four Der f 1 mutants (R157A and D199A) were successfully expressed as confirmed by mass spectrometry. Der f 1 mutants were purified by two steps, HPLC cation exchange chromatography and HPLC hydrophobic interaction chromatography, resulting in rDer f 1 mature forms, due to acidic conditions used during purification. Three of the four pro-rDer p 1 mutants were expressed (R156A, Y185V, and D198A). The allergens with the mutations R17A or R18A were not expressed. Pro-rDer p 1 mutants were purified from culture medium by affinity chromatography using mAb 5H8 and basic elution conditions. The antibody binding inhibition assays were performed with the three pro-rDer p 1 mutants due to the following advantages: (a) reduction of the number of epitopes involved due to the presence of the proregion that blocks IgE antibody binding sites and (b) the simpler purification method.
Rational Design of Mutagenesis-The rational design of sitedirected mutagenesis of the mAb 4C1 epitope in Der p 1 or Der f 1 was based on the crystal structures of Der p 1 or Der f 1 in complex with the Fab fragments of the mAb 4C1 (Protein Data Bank accession codes 3RVW and 3RVX for Der p 1 and 3RVV for Der f 1).
Site-directed Mutagenesis of mAb 4C1 Epitopes on Der p 1 and Der f 1-Four single amino acid mutants of the 4C1 mAb epitope were designed. The mutations R18A, R157A, Y186D, and D199A were originally performed on the DNA template that encodes the recombinant pro-Der f 1-N53Q deglycosylated mutant. Subsequently, mutations in equivalent positions in pro-rDer p 1-N52Q (R17T, R156A, Y185V, and D198A) were also performed. The difference in amino acid numbering between Der p 1 and Der f 1 is due to a deletion in Der p 1 of a serine present in Der f 1 at position 8. The templates for mutagenesis were DNA encoding the pro-Der p 1.0105-N52Q (deglycosylated) and pro-Der f 1.0107-N52Q allergens inserted into the yeast P. pastoris expression vectors pPICZ␣C and pPICZ␣B, respectively, for methanol-inducible expression of the allergens. The Der f 1 isoform is the Der f 1.0107 variant from the original Dilworth clone (P16311, which has an Asp at position 184). The Der f 1.0107 variant has a Val instead at position 184. The Der f 1.0107 sequence is like Der f 1.0101 except for Arg 103 (instead of Gln 103 in Der f 1.0101). Additionally, Asn 53 was mutated to Gln for deglycosylation purposes. Site-directed mutagenesis was performed using QuikChange TM (Stratagene). The sequence of the mutated DNA was confirmed before linearization and transformation into the P. pastoris strain KM71.
Sera from Mite-allergic Patients-The sera from allergic patients were obtained from PlasmaLab International (Everett, WA), which operates in full compliance with Food and Drug Administration regulations. An informed donor's consent was obtained from each individual prior to the first donation. Sera were from mite-allergic patients sensitized to Der f 1 (n ϭ 15; 16 Ϯ 20 IU/ml Der f 1-specific IgE antibodies; range, 0.9 -75 IU/ml; measured by multiplex array technology) and Der p 1 (n ϭ 21; 159 Ϯ 267 IU/ml Der p 1-specific IgE antibodies; range, 31-1072 IU/ml).
Crystallization-Crystallization was performed at 293 K using the hanging drop vapor diffusion method. The protein solution was mixed with the well solution in a 1:1 ratio. Tracking and analysis of the crystallization experiments were performed with the XTALDB crystallization system (16,17). Crystallization and cryocooling conditions are summarized in supplemental Table S1.
Data Collection, Structure Determination, Refinement, and Validation-Data collection was performed at 19-BM and 19-ID beamlines of the Structural Biology Center (18) at the Advanced Photon Source. Data were collected at 100 K using 0.979-Å wavelength and were processed with HKL-2000 (19). Data collection statistics are reported in Table 1.
All structures were solved using HKL-3000 (20) in combination with MOLREP (21). For Der f 1-4C1 complex, structures of Der f 1 (Protein Data Bank code 3D6S) and Fab fragment of a monoclonal antibody (Protein Data Bank code 1MLB) were used as starting models. In the case of Der p 1-4C1 complex, the structure of recombinant Der p 1 (Protein Data Bank code 3F5V) and the structure of 4C1 (from the Der f 1-4C1 complex) were used as starting models. Similarly, both crystal forms of the Fab fragment of 4C1 were determined using the model of the antibody derived from the Der f 1-4C1 structure. Structures were refined using HKL-3000, REFMAC (22), COOT (23), and CCP4 programs (24). In the final stages, refinement was performed with addition of TLS groups defined with the TLMSD server (25). Validation of the structures was performed using MOLPROBITY (26) and ADIT (27). According to MOLPROBITY, there were no outliers on the Ramachandran plot. Details of refinement as well as Protein Data Bank accession codes are summarized in Table 1 Computational Methods-Superposition of structures or their fragments was performed using LSQ (28) as implemented in COOT (23). Figures were prepared using PyMOL (29).
ELISA to Measure Group 1 Mite Recombinant Mutants and Dose-Response Curves-Recombinant Der p 1 allergens were measured by ELISA using either the mAb 5H8 or 10B9 as the capture antibody (2 g/ml) and biotinylated mAb 4C1 or 5H8 (1:1000 dilution), respectively, for detection. Recombinant Der f 1 allergens were measured with mAb 6A8 as the capture antibody (2 g/ml) and biotinylated 4C1 for detection (1:1000 dilution). A mite extract standard containing natural Der p 1 or Der f 1 was used starting at 250 ng/ml.
The mAb 4C1 epitope mutants were compared with the allergens with wild epitope (pro-rDer p 1, rDer p 1, and natural Der p 1) by performing dose-response experiments. The mAb 5H8 was used as the capture antibody. The antibodies used for detection were either biotinylated mAb 4C1 or a rabbit polyclonal IgG antibody raised against D. pteronyssinus extract (anti-Dpt polyclonal antibody) that has antibodies against natural Der p 1. To be able to assess the relative antibody binding affinities of the mutants compared with the wild type allergens, the biotinylated and polyclonal capture antibodies were first titrated to select the concentration to be used in the dose-response experiments. This concentration was the lowest possible, closest to the K d of the allergen-antibody interaction while still providing a window of antibody binding activity (A 405 nm ϭ 1-2). The dilutions for the biotinylated 4C1 mAb and the anti-Dpt polyclonal antibody were 1:50,000 (241 nM) and 1:100,000, respectively.
Multiplex Fluorescent Array Assay for Assessing Direct IgE Antibody Binding-Monoclonal antibody 5H8 (20 g) was coupled to different Luminex carboxylated fluorescent microsphere bead sets (Luminex Corp., Austin, TX), and the multiplex fluorescent array was performed as described (34,35). The mAb-coupled beads were added along with the Group 1 mite allergens at a concentration (500 ng/ml) above the top of the dose-response curves to ensure that antibodies in the beads were bound to allergen. The mutants of the epitope for the mAb 4C1 were tested using beads coated with the mAb 5H8. Sera (1:5 dilution) or the anti-Dpt polyclonal IgG antibody (1:1000 dilution) were added to wells, mixed with beads, and incubated. Biotin-labeled goat anti-human IgE (Kirkegaard and Perry Laboratories, Gaithersburg, MD) or biotin-labeled goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories, West Grove, PA), both at 1:1000 dilution, were added to the wells that had been incubated with sera or anti-mite extract rabbit polyclonal antibody, respectively. Biotinylated 4C1 mAbs were added to the wells where a mAb sandwich assay was performed. In the final step, streptavidin-phycoerythrin (4 g/ml) was added to all wells and mixed. Absorbance was measured in a Bio-Plex fluorescent suspension array reader (Bio-Rad). Mutants of the 4C1 mAb were compared with pro-Der p 1-N52Q by paired Student's t test. p values lower than 0.05 were considered significant.
ELISA Inhibition of IgE Antibody Binding by Pro-rDer p 1-4C1 mAb Epitope Mutants-Microplates were coated with mAb 5H8 (10 g/ml) in 50 mM carbonate-bicarbonate buffer, pH 9.6. The mutants tested for inhibition of IgE antibody binding were pro-rDer p 1 R156A, Y185V, and D198A (at 0, 0.1, 1, 10, and 100 g/ml concentrations). These mutants were added followed by addition of sera (1:5 dilution). The sera were a pool of the four sera from mite-allergic patients that showed the largest decrease of IgE antibody binding to the 4C1 mAb epitope mutants, indicating the importance of this site for IgE antibody binding in these patients. Incubation was performed at room temperature for 3 h. Affinity-purified peroxidase-labeled goat anti-human antibody IgE (Kirkegaard and Perry Laboratories) was used for detection at 1:1000. Plates were developed using 2,2Ј-azino-bis(3-ethylbenzothiazoline-6-sulphonic acid) and hydrogen peroxide as substrates for the peroxidase, and absorbance was read at 405 nm.
Determination of Metal Type in Der f 1 Structure-The presence of Ca 2ϩ in Der f 1 was confirmed by an additional diffraction experiment. A new data set was collected using x-ray radiation with significantly longer wavelength (1.771 Å). The experiment was performed at 19-BM beamline at the Advanced Photon Source, and the resulting data were used to generate an anomalous map confirming the presence of calcium.
Isothermal Titration Calorimetry-Isothermal titration calorimetry measurements (supplemental Table S2 and Fig. 2) were performed at 25°C using an iTC200 isothermal titration calorimeter (MicroCal). Protein preparations were dialyzed against 50 mM Tris, 50 mM NaCl, pH 7.5, 5 mM iodoacetamide buffer overnight at 4°C. The concentrations of Der p 1, Der f 1, and 4C1 were 30, 40, and 350 M, respectively. The first injection of 4C1 solution had a volume of 0.5 l followed by 19 2.0-l injections at 200-s intervals. The experiment was performed in high gain mode with the syringe rotating at 700 rpm. The thermodynamic parameters, stoichiometry, and dissociation constants (K D ) were calculated by Origin 7.0 software (MicroCal) using a one-binding site model. To determine the heat corresponding to antibody dilution, a control experiment was performed in which the 4C1 antibody was injected into buffer alone. The heat of the antibody dilution was used to correct the heat of the binding reaction between the Der p 1/Der f 1 and 4C1.

RESULTS
Der p 1 and Der f 1 in Complex with 4C1-Complexes of mAb 4C1 Fab fragments and allergens purified from natural sources were isolated and crystallized ( Fig. 1 and Table 1). The structures of two crystal forms of the Der p 1 complex and one Der f 1 complex as well as two Fab structures of 4C1 were determined. In all of the 4C1 complexes, the allergen was monomeric. A dimeric form of the mature recombinant Der p 1 was reported previously (36). However, it was subsequently shown that the dimeric structure is unlikely to be physiologically relevant (15). The high quality of the electron density maps allowed the unambiguous determination of the sequences within the binding epitopes and made possible the identification of allelic variants (37). Der f 1.0101 and Der p 1.0105 were the variants present in the mAb 4C1 complexes. The Der p 1 structure presented here is the first structure of this allergen purified from its natural source.
The 4C1 binding epitope is located relatively far from the cysteine protease active site as well as from the metal binding site (15,36,38,39). This was unexpected as the largest patch of conserved surface area in both Der f 1 and Der p 1 included the active site of the enzymes (15). The buried surface area for the Der f 1-4C1 interface was 790 Å 2 , whereas for Der p 1-4C1, the equivalent contact areas were 770 and 750 Å 2 in "long" and "short" crystal forms (Protein Data Bank codes 3RVX and 3RVW), respectively (see Table 1). In all cases, contacts formed by the heavy chain of the antibody constituted over 70% of the interface area. The third complementarity-determining region (CDR), 4 which protrudes into a concave fragment of the allergen surfaces, was responsible for most of the protein-protein interactions ( Fig. 1d and Table 2). The elbow angle, defined by the relative orientation of the structural constant (C L and C H 1) and variable (V L and V H ) domains, for the 4C1 Fab fragments in complex with allergens was similar to the observed angle in the case of the P2 1 2 1 2 1 form of the Fab with values of 146°and 139°i n the short and long forms of the Der p 1-4C1, respectively.
Metal Identification in Natural Der f 1 and Der p 1-Both Der f 1 and Der p 1 bind a calcium ion, not Mg 2ϩ , as was reported for Der p 1 (36). The identity of the ion was confirmed 4 The abbreviation used is: CDR, complementarity-determining region. by anomalous difference map analysis. However, the coordination sphere of the metal in the structures reported here is somewhat different in comparison with coordination reported previously (15,39) as one of the Ca 2ϩ -binding water molecules was replaced by a molecule of ethylene glycol, which was used as a cryoprotectant.
Structures of 4C1 Fab Fragment-The isolated Fab fragment of 4C1 was crystallized in two orthorhombic crystal forms ( Table 1). The Fab fragments differed significantly with respect to the elbow angles (supplemental Fig. S1). They were 143°and 167°for P2 1 2 1 2 1 and C222 1 forms (Protein Data Bank codes 3RVT and 3RVU), respectively. Such elbow angle values were in agreement with values observed for antibodies containing a light chain (40). The conformations of the polypeptide chains forming constant and variable domains and CDRs were very similar (supplemental Figs. S2 and S3). Within the CDRs, the biggest differences were observed for residues 102-106 in CDR3 of the heavy chain. For example, Tyr 104 had a completely different conformation in the two crystal forms. Superimposi-tion of corresponding V L and V H domains (C␣ atoms) from both crystal forms gave root mean square deviation values of 0.3 Å. The CDRs were well ordered in both structures despite the fact that they did not interact with an antigen and formed limited crystal contacts.
Structural Comparison of Complexed and Uncomplexed Antibody Paratope and Allergen Epitope-The overall conformation of the antibody and the conformations of the CDRs were similar in the uncomplexed and complexed crystal forms (supplemental Figs. S2 and S3), suggesting that CDRs of mAb 4C1 do not undergo significant rearrangements upon allergen binding. The antibody residues that formed the most extensive   (Fig. 2). They formed a central "rigid" surface to which the antibody binds, and Arg 157 worked as an anchor to form additional interactions that stabilize the complex. Residue Arg 157 participated in several H-bonds and interacted with each CDR from the light chain. The mAb 4C1 binding epitopes on Der f 1 and Der p 1 were very similar but have some distinct features (Table 2 and supplemental Figs. S2 and S3). The largest conformational differences between uncomplexed and complexed allergens were observed within Der p 1 where antibody binding caused changes that affected the region formed by residues 181-183. Residue 181 differs in Der f 1 (Thr) and Der p 1 (Ala). After complex formation, Thr 181 present in Der f 1 formed a hydrogen bond with Glu 106 from the heavy chain of the antibody. In Der p 1, formation of the H-bond was impossible, and the alanine was pushed away from the antibody Glu 106 . The change in the alanine position was propagated to Gln 182 , which shifted ϳ3 Å (C␣ position) and significantly changed its side chain conformation.
Despite small structural differences between the mAb 4C1 epitopes in Der p 1 and Der f 1, isothermal titration calorimetry experiments revealed that both Der f 1 and Der p 1 each formed 1:1 complexes with the Fab fragment of 4C1 with similar thermodynamic parameters and dissociation constants (K d around 18 nM) ( Table 3 and supplemental Fig. S4). Contacts between Group 1 allergens and 4C1 were also mediated by structurally conserved water molecules, and their presence improves the fit of the interacting macromolecular surfaces. Analysis of all mac-romolecular complexes reported here revealed that water molecules created bridging interactions among Tyr 186 , Ser 180 (Asn 179 ), and Glu 106 (CDR H3) and among Asp 199 and both Val 101 and Arg 103 (CDR H3) in both Der f 1 and Der p 1. In the case of the Der f 1-4C1 complex, an additional water molecule was bound by Asp 16 from the allergen and Asp 32 from the heavy chain of the antibody (CDR H1).
Allergen-Antibody Interfaces-The mAb 4C1 epitope structures (Protein Data Bank codes 3RVV and 3RVW, Der f 1-4C1 and Der p 1-4C1 short, respectively) were compared with epitopes from 14 allergen-antibody complexes available in the Protein Data Bank (44) determined by x-ray crystallography. Our analysis shows a difference in amino acid composition between interacting surfaces of allergens and antibodies ( Fig. 3 and supplemental Figs. S5-S7). Sequences of CDRs were biased toward some amino acids (41,45) with special emphasis toward Arg, Asp, Asn, Gly, Ser, Thr, Trp, and Tyr with Tyr having a dominant role (46 -48). Tyr and Arg residues are especially suitable for molecular recognition as they are able to interact with other residues through hydrogen bonds as well as hydrophobic and interactions. Analysis of 16 available allergenantibody complexes showed that Tyr and Ser in the antibodies play the most important role in interactions with allergens followed by Arg, Asn, and Asp. There were no dominant residues in epitope composition, but polar amino acids were the most abundant as described for protein interfaces (49,50). Interestingly, the mAb 4C1 binding epitope differed somewhat from other allergen epitopes as its amino acid composition was very similar to the composition of CDRs from antibodies with Tyr and Arg residues dominating their surface.
Analysis of H-bond networks on the interfaces of allergenantibody complexes ( Fig. 3a and supplemental Fig. S5) revealed that the following pairs of residues (listed with the residue from the allergen first) were involved in these interactions most often: Arg-Asp, Arg-Tyr, Asp-Arg, Asp-Tyr, Glu-Arg, Glu-Ser, and Gln-Ser. However, for lysozyme complexes, H-bond networks were different, and the following pairs of residues were responsible for interactions: Arg-Glu, Arg-Tyr, Asn-Gln, Asp-Ser, and Lys-Asp.
Design of Site-directed Mutagenesis-Four amino acids of the mAb 4C1 epitope were selected for site-directed mutagenesis and expression as single mutants of pro-rDer p 1 (Arg 17 , Arg 156 , Tyr 185 , and Asp 198 following Der p 1 numbering). Two arginines (Arg 17 and Arg 156 ) were in a marginal site of the 4C1 epitope. Arg 17 was part of an ␣-helix and interacted with Tyr 104 from CDR H3 and Tyr 54 from CDR H2 (supplemental Table  S2). Arg 17 also participated in a cation-interaction with Tyr 102 from the heavy chain (3.3 Å). Arg 156 participated in four hydrogen bonds with the three CDRs of the light chain of the antibody. Two were with side chains (Asp 92 ), and two were with main chain atoms of Tyr 32 and Arg 50 from the antibody. Two  additional amino acids more centrally located in the epitope, Tyr 185 and Asp 198 , were also mutated. Tyr 185 participated in hydrophobic interactions with the heavy chain and residue Tyr 104 from CDR H3. Asp 198 interacted through one hydrogen bond with the CDR H3 (Arg 103 ) of the antibody. From the four designed mutations, three resulted in the correct expression of Der p 1 mutants as proenzymes (R156A, Y185V, and D198A), and two resulted in mature rDer f 1 mutants (R157A and D199A). Functional Proof of Relevance of Amino Acids in Epitope-ELISA dose-response curves showed almost complete reduction of mAb 4C1 binding to the Der p 1 mutants (Fig. 4a), confirming the importance of the selected residues in the allergen-antibody interactions. The presence of the proregion did not affect the binding of either the capture mAb 5H8 or the detector mAb 4C1 because pro-rDer p 1 and rDer p 1 had overlapping dose-response curves similar to that of natural Der p 1.
Equivalent dose-response curves were observed for polyclonal IgG antibody binding to pro-rDer p 1 and pro-rDer p 1 mutants presented by the mAb 5H8. This confirms an overall folding of the mutants similar to that of the wild type because the conformational epitope(s) required for mAb capture and for binding of the polyclonal antibodies was preserved. A wider window of absorbance for the allergens without the proregion (natural and recombinant Der p 1) than for the allergens with the proregion (pro-rDer p 1 and the three pro-rDer p 1 mutants) was also observed (Fig. 4b). This indicates that the polyclonal antibody recognizes some epitopes under the proregion of the allergen as was reported previously for IgE (51). The effects of mutations on IgE antibody binding varied by patient. A reduction of IgE antibody binding from 20 up to 85% versus the wild type recombinant allergen was observed for 11 of 21 sera and mutation D198A. The strongest effect on reduction of IgE binding occurred for four sera and each of the single mutations (with up to 85% reduction in antibody binding) (Fig. 4c). Most of the sera tested for Der p 1 (18 of 21) and Der f 1 (14 of 15) also showed reduction from 20 up to 95% in IgE antibody binding versus the natural allergen (supplemental Fig. S8). The mutations that led to lowest IgE antibody binding for these four sera were R156A and Y185V.
A pool of four sera with IgE that had the strongest reduction of binding to the mutations was prepared to perform inhibition experiments. The mutants inhibited the binding of IgE antibodies to pro-rDer p 1, but the inhibition curves were displaced 5-10-fold to the right compared with that of pro-rDer p 1. These results indicate a weaker binding of the mutants to IgE antibodies compared with the pro-rDer p 1. The order of the mutants regarding IgE antibody binding capacity was R156A Ͻ Y185V Ͻ D198A (Fig. 4d).

DISCUSSION
Der f 1 and Der p 1 are the first cysteine proteases from clan CA for which metal binding was reported (15,36,38,39). The function of the calcium is still unknown. Its role is probably not critical for overall protein conformation as calcium-stripped Der f 1 has a structure very similar to the structure of protein containing the metal ion (15). The calcium ion is distal to the catalytic site of the enzyme, excluding any direct role in catalysis. However, both the active site and the metal binding site are located on the opposite ends of the water-filled cavity, and it is possible that the presence of Ca 2ϩ has some indirect effect on the active site. In the structures reported here, the cryoprotectant ethylene glycol replaced one of the water molecules from the Ca 2ϩ coordination sphere, suggesting that the calcium ion and surrounding residues are involved in binding of some ligand(s). Interactions with endogenous ligands were reported for other cysteine proteases; for example, glycosaminoglycans were shown to influence functions of certain cathepsins (52). Analysis of the sequence conservation (supplemental Table S2) shows that residues coordinating Ca 2ϩ ion (Asp 57 , Leu 58 , Glu 60 , and Glu 92 ) in both Der f 1 and Der p 1 are conserved in Eur m 1, but Asp 57 is replaced by Glu in Blo t 1. Calcium-binding residues are conserved in Pso o 1, but Asp 57 and Glu 92 are replaced by Gly and Ser residues, respectively, in Sar s 1. Der f 1 and Der p 1 are the only cysteine proteases from clan CA for which experimental data on metal binding are available. For other proteins, like Blo t 1, Pso o 1, and Sar s 1, such data are not available.
The analysis of molecular interactions between mAb 4C1 and the dust mite cysteine protease allergens Der p 1 and Der f 1 provides a structural basis for cross-reactivity between homologous allergens from different species (53). Residues forming the major part of the epitope are not only conserved in terms of amino acid sequence, but they share similar conformations in the complexed and uncomplexed allergens. Moreover, the epitope region is not affected by sequence polymorphisms (37) (Fig. 5), which in combination with its "rigidity" (Fig. 2) may make it an easy target for a cross-reactive antibody. Surprisingly, the mAb 4C1 epitope is located away from the largest conserved region between Der f 1 and Der p 1 at the vicinity of the active sites. This also suggests that antibody-bound allergens are still catalytically active in keeping with a previous study (54). The epitope is also relatively far from the Ca 2ϩ binding site. Superimposition of the allergen-antibody complexes with the structure of pro-rDer p 1 (38) and the antibody binding results show that the proregion of the enzyme did not block the 4C1 epitope. Therefore, pro-rDer p 1 is expected to interact with 4C1 Fab fragment in the same way that it interacts with the mature enzyme.
Our structural analysis supports previous observations that a relatively small conserved surface patch is sufficient for antibody binding, which in the case of the Group 1 allergens results in antigenic cross-reactivity. Even proteins with low overall sequence identity could share conformational epitope(s) in which case structural analysis of such proteins may be the only way to obtain such information. Analysis of homologous protein sequences reveals that residues forming the epitope in Der f 1 and Der p1 are conserved in Eur m 1 from Euroglyphus maynei but not in Blo t 1 from the storage mite Blomia tropicalis, Pso o 1 from the ectoparasitic mite Psoroptes ovis, or Sar s 1 from the itch mite Sarcoptes scabiei (supplemental Table S3). Moreover, corresponding residues in Eur m 1 are the same as in Der f 1, which suggests that mAb 4C1 should bind Eur m 1 in a fashion similar to Der f 1. Asp 199 of Der f 1 and Der p 1, a residue important for IgE antibody binding, and Arg 18 are conserved in Blo t 1, Pso o 1, and Sar s 1. However, Arg 157 and several additional amino acids in Blo t 1 are not conserved, which may partially explain the low IgE cross-reactivity between Group 1 allergens from B. tropicalis and Dermatophagoides species (55,56).
The Group 1 allergen and allergen-antibody complex structures also provide a unique opportunity to analyze the mode of complex formation especially when the CDRs and uncomplexed epitopes are not significantly affected by crystal contacts. The allergens and antibodies from the complexes reported here do not undergo significant conformational rearrangements at the level of the epitopes and paratopes (except in the region of Thr/Ala 181 ). Therefore, their interaction is better described by the "lock-and-key" analogy rather than as an "induced fit." The epitope common to Der f 1 and Der p 1 differs in amino acid composition from epitopes observed in all of the 14 allergen-antibody complexes reported to date in the Protein Data Bank, which of course may not be representative of all epitopes. Six epitopes are present on five allergens (bee venom hyaluronidase Api m 2, cockroach Bla g 2, cow's milk ␤-lacto-globulin Bos d 5, and the birch and timothy grass pollen allergens Bet v 1 and Phl p 2, respectively), and eight are epitopes on lysozyme, a model antigen for crystallographic studies that is also a minor food allergen. Enrichment in Arg and Tyr residues makes the mAb 4C1 epitope quite distinct from epitopes that are recognized by antibodies that were not reported to be crossreactive. Two of these allergen-antibody complexes were made with Fab fragments of recombinant human IgE (57,58) in an attempt to identify the location of epitopes for IgE antibodies. Natural IgE antibodies are present in minute amounts in human sera and are polyclonal; thus, crystallization studies of natural IgE are not feasible. The approach presented here used a monoclonal antibody that inhibits IgE antibody binding as a surrogate for IgE and allowed us to perform mutagenesis and antibody binding analysis to identify IgE antibody binding sites. Analysis of IgE antibody binding to Der f 1 and Der p1 mutants showed that mAb 4C1 epitope overlaps with IgE epitope(s). For some patient sera, single mutations (Arg 157 , Tyr 186 , and Asp 199 ) were sufficient to alter IgE binding in a significant way. Mutation of Arg 157 , which participates in an extended network of interactions with mAb 4C1 (it forms H-bonds with residues from all light chain CDRs), has a prominent effect on IgE antibody binding. Both Tyr 186 and Asp 199 are located in the central rigid part of the epitope, and they interact mainly with CDR H3, which has the dominant role in binding between mAb 4C1 and allergens. Both Tyr 186 and Asp 199 are not only involved in interactions with the antibody, but they are also involved in stabilization of the surfaces of the allergens. Differences in effects of particular mutations on IgE binding may be explained by the fact that, unlike Arg 157 , Tyr 186 and Asp 199 are involved in many interactions through their side chains, and their mutation impact a larger area of the surface of the allergens.
In our studies, we were able to identify cross-reactive determinants that could be tested for IgE recognition. This approach was successful in that we identified specific residues that were molecular determinants for IgE responses in a number of miteallergic patients. The success of such mAb-directed approaches will vary for different allergens depending on the specificity of mAb and IgE responses. In addition, these approaches will lead to the production of hypoallergens with low IgE antibody binding capacity. These mutants will need to be tested for their suitability for immunotherapy, aiming to reduce IgE-related side effects that may occur when increasing doses of natural allergen are administered to patients. The significant reduction of IgE binding that was observed by performing single site mutations on the allergen surfaces shows that recombinant mutants of Der p 1 and Der f 1 may find application in vaccines for the treatment of dust mite allergy.