Disruption of Allergenic Activity of the Major Grass Pollen Allergen Phl p 2 by Reassembly as a Mosaic Protein¹ Free

IgE-mediated allergy affects more than 25% of the population and hence is one of the major immunologically mediated diseases (1). In sensitized allergic individuals allergen contact induces crosslinking of mast cell- and basophil-bound IgE Abs and thus the typical immediate inflammatory reactions caused by the release of inflammatory mediators, proinflammatory cytokines, and proteases (2). Additionally, allergen contact activates allergen-specific T cells to proliferate and secrete proinflammatory cytokines underlying the typical Th2 phenotype of allergies (3, 4, 5). Besides various antiinflammatory treatment strategies, allergen-specific immunotherapy (SIT)⁵ is the only allergen-specific and disease-modifying treatment for IgE-mediated allergies (6, 7). SIT is based on the administration of increasing doses of the disease-eliciting allergens by injection or mucosal application. Mechanisms underlying SIT include the induction of allergen-specific IgG Abs, which inhibit IgE recognition of allergens and thus allergen-induced allergic inflammation, and boosting of the allergen-specific IgE production by allergens, immunomodulatory effects that reverse the Th2 phenotype and possibly immunoregulation (8).

SIT has been shown to be clinically effective in numerous clinical trials (9). It can prevent the progression of allergic disease from mild (e.g., rhinoconjunctivitis) to severe manifestations (e.g., asthma) and has a long-lasting effect (10, 11). However, the broad application of SIT has been limited by side effects caused by the administration of allergenic materials and the poor quality of natural allergen extracts (12). Due to the advances made in the field of allergen characterization through the application of recombinant DNA technology, DNAs coding for the most relevant allergens have become available and the corresponding allergens can now be produced as recombinant proteins equaling their natural counterparts (13). Recombinant allergens have opened the door for new forms of allergy diagnosis that allow the precise determination and monitoring of the molecular reactivity profiles of allergic patients. Based on the knowledge of the primary allergen structures, a variety of technologies are currently applied to develop new allergen-specific vaccination and tolerance induction protocols (14, 15). These technologies include allergen-derived T cell epitope-containing peptides for immunomodulation, recombinant allergens, allergens coupled or adsorbed to immunomodulatory components, allergen-derived peptides for the induction of blocking Abs, and the engineering of genetically modified recombinant allergen derivatives, which are selected to exhibit less IgE-mediated allergenic activities, to preserve allergen-derived T cell epitopes for immunomodulation and the ability to induce allergen-specific protective IgG responses (15, 16). The advantage of recombinant hypoallergenic allergen derivatives is that they seem suitable both for T cell modulation and vaccination approaches (16). Several strategies used for the reduction of allergenic activity include fragmentation, introduction of mutations and molecular inversion, and random gene shuffling (14, 15, 16, 17, 18, 19, 20). However, the approaches of fragmentation, deletion, or mutation may have disadvantages. It may be technically difficult to prepare several fragments for one allergen, and small protein fragments may induce low levels of protective IgG Abs. Mutants or deletion variants of allergens may lack T cell epitopes or important structures necessary to induce protective IgG responses. Randomly shuffled allergens can be relatively easily produced, but it may be quite time-consuming to select from a variety of mutants those with the most advantageous properties (21, 22). To overcome these disadvantages, we have developed an alternative strategy for the conversion of an allergen into a hypoallergenic vaccine. This strategy is exemplified for Phl p 2, a major grass pollen allergen, which is recognized by more than 200 million grass pollen-allergic patients, and according to prevalence data, allergenic activity and abundance in grass pollen represent essential components in a clinically effective vaccine against grass pollen allergy (23, 24, 25, 26, 27). The conversion of the biologically active Phl p 2 allergen into a hypoallergenic vaccine was achieved in a two-step process. First, non-IgE-reactive, Phl p 2-derived peptides were identified by peptide mapping. In a second step, non-IgE-reactive peptides were recombined in the form of a mosaic molecule by PCR-based mending of cDNAs coding for the peptides and subsequent expression of the mosaic molecule.

Sera were obtained from patients suffering from grass pollen allergy according to case history and skin testing. Serum IgE Abs specific for recombinant Phl p 2 (rPhl p 2) were determined by CAP FEIA (Phadia) and dot blot analysis as previously described (25).

Pollens from different grass and corn species (Lolium perenne, ryegrass; Triticum aestivum, common wheat; Hordeum, barley; Zea mays, maize; Phleum pretense, timothy grass; and Secale cereale, rye) were purchased from Allergon. Aqueous pollen extracts were prepared by homogenizing 2 g of pollen in 50 ml of distilled water containing 5 mM PMSF with an Ultraturax (IKA) and extracting the homogenate for 3 h at 4°C under continuous shaking (28). Homogenates were centrifuged at 20,000 × g for 30 min at 4°C to remove insoluble particles. Supernatants were lyophilized and checked for quantity and quality of proteins by SDS-PAGE and Coomassie blue staining. Total protein amounts were determined by a Micro BCA protein assay (Pierce). Purified recombinant Phl p 2, Phl p 1, and Phl p 5 were obtained from Biomay. Recombinant Sec c 3, the Phl p 3 homologous allergen from rye, was expressed in Escherichia coli and purified via nickel affinity chromatography (S. Laffer and R. Valenta, unpublished).

Three peptides, comprising aa 1–33 (VPKVTFTVEKGSNEKHLAVLVKYEGDTMAEVELC), aa 34–65 (REHGSDEWVAMTKGEGGVWTFDSEEPLQGPFNC), and aa 66–96 (CFRFLTEKGMKNVFDDVVPEKYTIGATYAPEE) of the Phl p 2 (Table I) molecule, were synthesized using Fmoc (9-fluorenylmethoxycarbonyl) strategy with HBTU (2-(1H-benzotriazol-1-yl)-1,1,3,3, tetramethyluronium hexafluorophosphate) activation (0.1 mmol small-scale cycles) on an Applied Biosystems peptide synthesizer (model 433A). Preloaded PEG-PS (polyethylene glycol polystyrene) resins (0.15–0.2 mmol/g loading) (PerSeptive Biosystems) were used as solid phase to build up the peptides. Chemicals were purchased from Applied Biosystems. Coupling of amino acids was confirmed by conductivity monitoring in a feedback control system. One cysteine residue was added to each peptide at the N or C terminus to facilitate coupling of the peptides to carriers (Table I). Peptides were cleaved from the resins with a mixture of 250 μl distilled water, 250 μl triisopropylsilan (Fluka), and 9.5 ml TFA for 2 h and precipitated in tert-butylmethylether (Fluka). The identification of the peptides was checked by mass spectrometry. Peptides were purified to >90% purity by preparative HPLC (piCHEM) (29).

The cDNA coding for a hexahistidine-tagged rPhl p 2 allergen was obtained by PCR, using the 5′ primer (GGA TTT CCA TAT GGT CCC GAA GGT GAC GTT CAC G) and the 3′ primer (CGC GAA TTC TCA GTG GTG GTG GTG GTG GTG CTC TTC TGG CGC GTA GGT GGC) and the Phl p 2 cDNA as a template (24).

Based on the Phl p 2-derived peptides, three fragments of the Phl p 2 cDNA were generated and rearranged by PCR. Fragment 1–3 were fused in the order 1-3-2 using the primers P2/1–P2/6 (see Fig. 1 D) for overlap extension, as described previously (30).

The resulting cDNAs were ligated into the expression vector pET17b (Novagen), which had been cut with NdeI/EcoRI. The DNA sequences of the two constructs were confirmed by sequence analysis. The recombinant proteins were expressed in E. coli BL21 (DE3) (Novagen) by adding isopropyl-β-thiogalactopyranoside (final concentration of 0.5 mM) to a liquid culture (Luria-Bertani medium containing 100 mg/L ampicillin) at an OD₆₀₀ of 0.4 and incubation for 4 h at 37°C. E. coli cells from a 500 ml culture were harvested by centrifugation and prepared for purificaton under native (rPhl p 2) or denaturing conditions (rPhl p 2 mosaic) (Qiagen). Protein samples were analyzed for purity by SDS-PAGE and protein concentrations were determined with a Micro BCA kit (Pierce).

Laser desorption mass spectra of rPhl p 2 and rPhl p 2 mosaics were acquired in a linear mode with a MALDI-TOF instrument (Kratos Analytical) operating at 20 kV acceleration voltage and equipped with a nitrogen UV laser (337 nanometers (nm), pulse duration 3 nanoseconds) (piCHEM). The m/z values were calibrated externally. Samples were dissolved in 10% acetonitrile (0.1% TFA). Alpha-cyano-4-hydroxycinnamic acid was used as a matrix dissolved in 60% acetonitrile (0.1% TFA). For sample preparation, a 1:1 mixture of protein solution and matrix solution was deposited onto the target and air-dried.

Gel filtration was performed at room temperature on an ÄKTA Purifier system (Pharmacia Biotech) with a prepacked Superdex 200 10/30 column (Pharmacia Biotech, 24 ml bed volume). The elution buffer was 10 mM Tris-HCl (pH 7.0) and 150 mM NaCl with an isocratic flow rate of 0.5 ml/min. The molecular mass markers were thyroglobin (M_r = 670 kDa), bovine gamma-globulin (158 kDa), chicken OVA (44 kDa), bovine myoglobin (17 kDa), and vitamin B12 (1.35 kDa) as included in the gel filtration standard (Bio-Rad). The gel filtration analyses of 350 μl of Phl p 2 (0.26 mg/ml) and 350 μl of Phl p 2 mosaic (0.23 mg/ml) were performed under identical conditions.

CD measurements were performed on a Jasco J-715 spectropolarimeter (Japan Spectroscopic) using 0.02-cm pathlength cylindrical cells. Far-UV CD spectra of rPhl p 2, rPhl p 2-mosaic, P1, P2, and P3 were recorded from 180 to 250 nm (185 nm rPhl p 2, 197 nm rPhl p 2-mosaic). Each spectrum was recorded as an average of three scans, with 0.2-nm resolution at a scan speed of 50 nm/min. Concentrations used were 0.26 mg/ml for rPhl p 2, 0.56 mg/ml for rPhl p 2 mosaic, and 1 mg/ml for the synthetic peptides P1, P2, and P3. Results were expressed as the molar mean residue ellipticity (θ) at a given wavelength.

The secondary structure content was calculated using the secondary structure estimation program CDSSTR (31, 32).

For immunoblotting, the protein extracts were boiled for 10 min in SDS-sample buffer and supernatants were recovered after centrifugation at 10,000 × g for 10 min at room temperature. Approximately 100 μg/cm gel of total protein extracts was separated by preparative 12% SDS-PAGE and blotted onto nitrocellulose (Schleicher & Schuell). Nitrocellulose strips were exposed to rabbit sera diluted 1/1000, and bound IgG was detected by incubating the strips with 1/2000 diluted ¹²⁵I-labeled donkey anti-rabbit Ig Abs (Amersham Biosciences) for 3 h at room temperature. After repeated washes, strips were dried and exposed to Kodak X-OMAT S films at −70°C using intensifying screens.

Two-microliter aliquots of the rPhl p 2 wild-type, rPhl p 2 mosaic, and Phl p 2-derived peptides, as well as an unrelated control peptide generated from the birch pollen allergen Bet v 1 (preimmune serum) (31aa: DGGSILKISNKYHTKGDHEVKAEQVKASKEC) (concentration = 1 mg/ml) were dotted onto nitrocellulose strips. Strips were exposed to patients’ sera (dilution 1/10), and bound IgE Abs were detected with ¹²⁵I-labeled anti-human IgE Abs (Phadia) diluted 1/10 (25).

Basophils were enriched from one grass pollen allergic patient by dextran sedimentation. Granulocyte isolation and histamine release were performed as described (33). In brief, isolated cells were washed, resuspended in histamine release buffer, and exposed to serial dilutions of recombinant allergens (rPhl p 2, rPhl p 2 mosaic) or anti-IgE mAb E-124-2-8 (1 μg/ml) in 96-well microtiter plates (TPP) for 30 min at 37°C. After incubation, cells were centrifuged. Cell-free supernatants were recovered and analyzed for histamine content by using a radioimmunoassay (Immunotech). Histamine release was expressed as a percentage of total histamine measured in cell lysates.

Heparinized peripheral blood was obtained from two additional grass pollen allergic patients after informed consent was given. Blood samples (100 μl) were incubated with serial dilutions of recombinant allergens (10⁻³ to 10 μg/ml), anti-IgE Ab (1 μg/ml), or PBS for 15 min (37°C). Cells were analyzed by two-color flow cytometry on a FACScan (BD Biosciences). Allergen-induced up-regulation of CD203c was calculated from mean fluorescence intensities (MFIs) obtained with stimulated (MFI_stim) and unstimulated (MFI_control) cells, and expressed as stimulation index (MFI_stim:MFI_control) (34).

Skin prick tests were performed on the forearms of two more grass pollen allergic patients with equimolar concentrations of proteins after informed consent was given. Twenty-microliter aliquots of rPhl p 2 and rPhl p 2 mosaic (1, 2, 4, 8, and 16 μg/ml), diluted in sterile water as well as commercially available prick solutions (timothy grass pollen extract, histamine) (Allergopharma), were applied and pricked with sterile lancets (Allergopharma). Reactions were recorded after 20 min by photography and by transferring the ballpoint pen-surrounded wheal area with a transparent scotch tape to paper (26). Wheal sizes (mm²) were calculated using the formula ((D1 + D2)/2)² (26).

Rabbits were immunized with uncoupled and keyhole limpet hemocyanin-coupled proteins. Recombinant Phl p 2 mosaic was coupled to keyhole limpet hemocyanin using an Imject immunogen EDC conjugation kit (Pierce), whereas peptides were coupled via their cysteine residues using an Imject maleimide activated immunogen conjugation kit (Pierce).

Rabbits were immunized with the immunogens (200 μg/injection) using CFA (first immunization) and IFA (booster injections after 4 and 7 wk) (Charles River Breeding Laboratories). Rabbits were bled 8 wk after the first immunization.

Sera containing rPhl p 2- and rPhl p 2 mosaic-specific IgG Abs were stored at −20°C until use.

Rabbit Abs specific for Phl p 2 as well as for Phl p 2 mosaic were tested for reactivity with rPhl p 2, rPhl p 1, and rSec c 3, the Phl p 3-related allergen from rye, by ELISA. ELISA plates (Greiner Labortechnik) were coated with rPhl p 2 (5 μg/ml in PBS) at 4°C overnight. Plates were washed twice with PBS/0.05% (v/v) Tween 20 and blocked with PBS/1% (w/v) BSA/0.05% (v/v) Tween 20 (BSA) at 37°C for 2.5 h. Sera were diluted as follows: 1/100, 1/500, 1/1000, 1/5000, 1/10,000, 1/50,000, and 1:100,000. Plate-bound allergens were incubated with 100 μl/well of the diluted sera at 4°C overnight. Bound Abs were detected with an HRP-labeled donkey anti-rabbit Ig antiserum (Amersham Biosciences), diluted 1:2000. The Ab levels correspond to the ODs measured (means of duplicates with a SD of <10%).

Six-week-old female BALB/c mice were purchased from Charles River Breeding Laboratories and kept at the animal care unit of the Department of Pathophysiology, Medical University of Vienna, according to the local guidelines. Groups of five mice were immunized subcutaneously two times in 3-wk intervals with either 10 μg rPhl p 2 or 10 μg rPhl p 2 mosaic, or with a mixture of 10 μg of each of the Phl p 2-derived peptides (peptides 1–3) (Table I), adsorbed to Al(OH)₃ (Alu-Gel-S, Serva Electrophoresis). Sera were obtained via bleeding from the tail vein on days 0 (preimmune serum), 21, and 42 and stored at −20°C until use. Measurements of IgG1 Abs specific for rPhl p 2 were done as described (35).

The ability of rPhl p 2 mosaic-induced rabbit IgG Abs to inhibit the binding of allergic patients’ IgE to rPhl p 2 was investigated by an ELISA competition assay (29). ELISA plates (Nunc Maxisorp) were coated with rPhl p 2 (1 μg/ml) and preincubated either with a 1/100 dilution of the rabbit anti-rPhl p 2 antiserum, rabbit anti-rPhl p 2 mosaic antiserum, or, for control purposes, the corresponding preimmune sera. After washing, plates were incubated with 1/4 diluted sera from 14 Phl p 2-sensitized grass pollen allergic patients, and bound IgE Abs were detected with HRP-labeled goat anti-human IgE (KPL) diluted 1/2500. The percentage of inhibition of IgE binding achieved by preincubation with the anti-rPhl p 2 or anti-rPhl p 2 mosaic antisera was calculated as follows: percentage of inhibition of IgE binding = (100 − OD_I/OD_P) × 100, where OD_I and OD_P represent the extinctions after preincubation with the rabbits’ immune sera (I) and the corresponding preimmune sera (P), respectively (36).

To identify Phl p 2-derived peptides lacking allergenic activity, three peptides, each comprising about one-third of the Phl p 2 protein, were chemically synthesized (Table I). The peptides had a length of 34 (peptide 1), 33 (peptide 2), and 32 (peptide 3) amino acids with molecular masses of ∼3.7 kDa, and together they covered the complete Phl p 2 amino acid sequence. The peptides were designed to cause a complete disruption of the structure of the Phl p 2 molecule, which consists of nine antiparallel β-strands (37). Peptide 1 consists of strand A, B, and the N-terminal part of C; peptide 2 consists of the C-terminal part of strands C, C′, D, and E; and peptide 3 consists of of strands F, G, and H. Fig. 1 A shows the three-dimensional structure of the protein in a ribbon representation with the peptides P1–P3 highlighted in different colors.

The IgE reactivity of the Phl p 2-derived peptides was compared with that of the complete rPhl p 2 protein by dot blot analysis (data not shown). Nitrocellulose-dotted rPhl p 2, r Phl p 2-derived peptides (P1–P3), an immunologically unrelated major grass pollen allergen, rPhl p 5 (38), and, for control purposes, HSA as well as an unrelated control peptide (P) were exposed to sera from 34 grass pollen allergic patients and to serum from a nonallergic individual (N). Thirty-two sera from grass pollen allergic patients showed IgE reactivity to nitrocellulose-dotted rPhl p 2, but no serum reacted with any of the three Phl p 2-derived peptides. Serum from the nonallergic individual displayed no IgE reactivity to any of the peptides or proteins.

The recombinant Phl p 2 mosaic protein was obtained by the rearrangement of the three Phl p 2-derived peptides in the altered order P1-P3-P2. This mosaic protein was created under the assumption that the de novo assembly of three nonallergenic Phl p 2-derived peptides would deliver a mosaic protein with disrupted three-dimensional structure and consequently reduced allergenic activity. Fig. 1,B shows the order of the three peptides in the natural Phl p 2 allergen and in the Phl p 2 mosaic protein. We produced recombinant Phl p 2 and recombinant Phl p 2 mosaic protein, with both containing a C-terminal hexahistidine tail (Fig. 1,B), to allow the purification of both proteins by nickel affinity chromatography (Qiagen). Recombinant Phl p 2 mosaic was constructed by PCR-based mending of cDNAs, coding for the peptides P1–P3, using the primers displayed in Fig. 1,D and the Phl p 2-encoding cDNA (24) as template as described (30). The resulting cDNA consisted of 309 bp coding for a protein with a calculated molecular mass of 11,769 Da, which is almost identical to the His-tagged recombinant Phl p 2 allergen (11,784 Da) (Fig. 1 C).

Fig. 2,A shows the purified recombinant proteins (rPhl p 2, rPhl p 2 mosaic). Although the two proteins did not show an identical migration behavior in the SDS-PAGE, mass spectrometry gave almost identical molecular masses for the two proteins (rPhl p 2, 11,775 Da; rPhl p 2 mosaic, 11,770 Da), which are in agreement with the molecular masses deduced for the cDNA sequences including the methionines at their N termini (Fig. 2, B and C).

The CD spectrum of the His-tagged rPhl p 2 protein is characterized by a broad minimum at 207 nm and two maxima, a pronounced positive band at 192 nm and a smaller one at 225 nm. This is consistent with earlier CD measurements of rPhl p 2 and other proteins featuring a β-sandwich fold (37). The positive band at 225 nm is most likely contributed by the aromatic residues (two tryptophanes and three tyrosines) of rPhl p 2 (39) (Fig. 2 D).

Secondary structure calculation using the CDSSTR program and different basis sets yielded a structure consisting of mainly β-sheet and β-turn contributions for rPhl p 2 (45% β-sheet, 27% turn, 24% coil, 99% overall fold). The CD spectrum of the rPhl p 2 mosaic shows only one pronounced negative band with a minimum at 202 nm and a slanted shoulder between 210 and 240 nm, hinting at a significantly reduced β-sheet and increased random-coil conformation (Fig. 2,D). For the rPhl p 2-mosaic protein, the CDSSTR program could not be used because of the restricted wavelength range of its CD spectrum. Each of the three synthetic Phl p 2-derived peptides exhibited a negative band around 200 nm, indicating a mainly random-coil conformation (Fig. 2 E).

When Phl p 2 mosaic and Phl p 2 wild type were applied in a concentration of 0.23 mg/ml (Phl p 2 mosaic) or 0.26 mg/ml (Phl p 2) in a physiological buffer to gel filtration, the Phl p 2 mosaic eluted at 15.72 ml, which corresponds to a molecular mass of 34 kDa, whereas the Phl p 2 wild type eluted at 16.96 ml, corresponding to a molecular mass of 19 kDa (Fig. 2 F). Accordingly, Phl p 2 mosaic occurred as a trimer and Phl p 2 formed a dimer, whereas no monomeric forms were detected for either of the two proteins.

The IgE binding capacity of purified Phl p 2 mosaic was compared with that of the Phl p 2 wild-type protein by dot blot experiments as described for the peptides using sera from 12 timothy grass pollen allergic patients (Fig. 3). All sera contained IgE Abs against rPhl p 2, but no serum exhibited detectable IgE reactivity to the rPhl p 2 mosaic or to the negative control, HSA.

To study the allergenic activity of rPhl p 2 mosaic on effector cell activation, a human basophil histamine release assay, analysis of CD203c expression on human basophils, and skin prick test experiments were performed in grass pollen allergic patients. Fig. 4 A shows the release of histamine from isolated human basophils in response to rPhl p 2 and the rPhl p 2 mosaic. The rPhl p 2 mosaic (maximal histamine release between 1 and 10 μg/ml) exhibited a >1000-fold reduced allergenic activity compared with the rPhl p 2 wild-type allergen, which induced a maximal histamine release already at a concentration of 10⁻³ μg/ml.

CD203c has previously been described as an activation marker on human basophils, which is up-regulated upon allergen-induced crosslinking of receptor-bound IgE (34). Exposure of basophils isolated from blood samples of two additional grass pollen allergic patients sensitized to Phl p 2 resulted in a significant up-regulation (p < 0.05) of CD203c at a concentration of 0.01 μg/ml, whereas the rPhl p 2 mosaic protein failed to up-regulate CD203c up to a concentration of 10 μg/ml (Fig. 4, B and C).

The reduction of allergenic activity of the rPhl p 2 mosaic as compared with rPhl p 2 wild type was confirmed in vivo by skin prick testing in two grass pollen allergic patients comparing the mosaic protein with the rPhl p 2 wild-type protein (data not shown). Both patients showed a dose-dependent positive skin reaction to the rPhl p 2 allergen at all concentrations tested (1–16 μg/ml) and to timothy grass pollen extract. In contrast, none of the patients had a positive skin reaction to the Phl p 2 mosaic at a concentration of 1, 2, and 4 μg/ml. The wheal surface area (mm²) induced with the rPhl p 2 mosaic at 8 μg/ml (only in patient A) and at 16 μg/ml (in both patients) was markedly reduced compared with the rPhl p 2, and did not even reach the size of the wheal surface area induced with 1 μg/ml rPhl p 2 (patient A: rPhl p 2, 100 mm², mosaic of 16 mm²; patient B: rPhl p 2, 81 mm², mosaic of 9 mm²).

To test the ability of the Phl p 2 mosaic protein to induce Phl p 2-specific IgG Abs, rabbits were immunized with rPhl p 2 and rPhl p 2 mosaic. Rabbit antisera were analyzed in terms of specificity and immunogenicity by dot blot and ELISA. Fig. 5 A shows a comparison of Phl p 2-specific IgG levels in rPhl p 2- and rPhl p 2 mosaic-immunized rabbit antisera. Both antisera reacted with rPhl p 2 up to a serum dilution of 1/10,000. The Phl p 2 mosaic induced even slightly higher levels of Phl p 2-specific Abs than did rPhl p 2.

The specificity of rabbit IgG Abs to rPhl p 2 was also studied by dot blot experiments (Fig. 5,B). Both the rabbit anti-rPhl p 2 mosaic antiserum and the anti-rPhl p 2 antiserum reacted with the Phl p 2 mosaic as well as with the rPhl p 2 allergen; the anti-Phl p 2 mosaic antiserum was even stronger than that obtained by immunization with the rPhl p 2 allergen. No reactivity was observed with the rabbit preimmune sera as well as with the buffer control (Fig. 5 B).

When we tested the anti-Phl p 2 mosaic and anti-Phl p 2 Abs for crossreactivity, we found that the rabbit anti-Phl p 2 Abs crossreacted with rPhl p 1 and rSec c 3, which can be explained by the sequence homology of Phl p 2 and Sec c 3 and with the C terminus of Phl p 1. The anti-P2M Abs crossreacted with rSec c 3, but only weakly with rPhl p 1 (data not shown).

Previous studies have shown that Phl p 2 is a poorly immunogenic protein and induces only low levels of Phl p 2-specific IgG Ab responses by immunization of mice and upon immunotherapy in allergic patients (35, 40, 41). This observation was also confirmed in our experiments, as shown in Fig. 5 C. Recombinant Phl p 2, adsorbed to alum, failed to induce relevant IgG1 Ab response in immunized mice. A mixture of the three Phl p 2-derived peptides comprising the primary sequence of the Phl p 2 allergen failed to induce a detectable Phl p 2-specific IgG1 response. In contrast, immunization of mice with the alum-adsorbed Phl p 2 mosaic protein yielded a significantly stronger (p = 0.003) Phl p 2-specific IgG1 response than immunization with rPhl p 2. Neither immunization with the wild-type rPhl p 2 allergen, the rPhl p 2 mosaic, nor with the peptide mix induced a detectable Phl p 2-specific IgE Ab response (data not shown).

Next, we studied whether immunization with the rPhl p 2 mosaic could induce IgG Abs that crossreact with group II allergens from other grass species. Fig. 5,D shows that rabbit anti-rPhl p 2 mosaic antiserum not only recognized the rPhl p 2 allergen but also group II allergens in different grass pollen species (L. perenne, ryegrass; T. aestivum, common wheat; Hordeum, barley; Z. mays, maize, P. pratense, timothy grass; and S. cereale, rye). The specificity of the Phl p 2-specific IgG reactivity is demonstrated by the lack of reactivity of the rabbit preimmune sera (Fig. 5 D, P lanes).

The bands at ∼10 kDa may represent group II and III grass pollen allergens, whereas the bands at ∼30 kDa in L. perenne and Z. mays may result from crossreactivity of the rabbit Abs with group I allergens.

IgG Abs induced by immunization with the rPhl p 2 mosaic inhibited the binding of allergic patients’ serum IgE to the wild-type rPhl p 2 in an ELISA competition assay using sera from 14 grass pollen allergic patients (Table II). The anti-Phl p 2-mosaic Abs inhibited the binding of grass pollen allergic patients’ IgE to Phl p 2, ranging from 42.7 to 95.2% (54.9% average inhibition), albeit to a lower degree than it could be achieved by preincubation with anti-rPhl p 2 Abs (88.0% average inhibition) (Table II).

We report a novel strategy for the conversion of allergens into allergy vaccines that reduced allergenic activity and preserved of immunogenicity. We chose Phl p 2, one of the four major grass pollen allergens (i.e., Phl p 1, Phl p 2, Phl p 5, and Phl p 6) required for a therapeutic grass pollen vaccine, as an example (27, 42). The strategy is based on the definition of nonallergenic fragments in a first step. After the allergen has been split into hypoallergenic peptides, these peptides are subsequently reassembled in form of a hypoallergenic mosaic molecule with the aim to disrupt the allergen’s three-dimensional structure. The Phl p 2 mosaic molecule created in this manner lacked IgE reactivity and accordingly had a strongly reduced allergenic activity as determined by basophil activation assays (i.e., histamine release assays, CD203c expression). Skin testing of allergic patients with the wild-type Phl p 2 and the Phl p 2 mosaic molecule confirmed the hypoallergenic nature of the mosaic protein. These results indicate that immunotherapy with the mosaic protein will induce less therapy-related side effects mediated by IgE Abs than its natural counterpart.

Phl p 2 has been shown to be a hardly immunogenic protein in animal models as well as in clinical trials (35, 41). The Phl p 2 mosaic protein also seems to overcome this disadvantage. We obtained significantly higher Phl p 2-specific IgG levels upon immunization with the Phl p 2 mosaic as compared with the immunization with the Phl p 2 wild-type allergen in a mouse model using aluminum hydroxide, the most frequently used adjuvant for immunotherapy.

The fact that Phl p 2 mosaic occurred as a trimer while Phl p 2 formed only dimers may explain why the mosaic induced higher Phl p 2-specific IgG levels in comparison to the Phl p 2 wild type. We have already earlier observed that a recombinant trimer of the major birch pollen allergen Bet v 1 induced higher Bet v 1-specific IgG levels than did a dimer and monomer (43). Similar increases of the immunogenic activity of oligomerized Ags have been also described for other vaccines (44).

Additionally, we found that Phl p 2 mosaic-specific Abs crossreacted with group II and group III allergens present in several other grass species. Perhaps most importantly, we found that rabbit IgG raised by injection of the wild-type or the mosaic allergen inhibited the binding of allergic patients’ IgE to the wild-type allergen similar to that observed for other hypoallergenic derivatives that have been used already in clinical trials in allergic patients (45, 46).

However, despite the fact that Phl p 2 mosaic induced higher Phl p 2-specific IgG levels than Phl p 2 upon immunization, we found that the Phl p 2 mosaic-induced IgG inhibited allergic patients IgE reactivity to Phl p 2 to a lower degree (i.e., 54.9% average inhibition) than to Phl p 2-induced IgG Abs (i.e., 88% average inhbition). There may be several explanations for the only partial inhibition of IgE reactivity by mosaic-induced IgG. One possible explanation would be that the original IgE epitopes have been destroyed and therefore no IgG can be induced against the original IgE epitopes. The inhibition of IgE reactivity will thus be caused only by IgG recognizing parts of the IgE epitopes or by binding in their close vicinity (i.e., “steric inhibition”). Another possibility would be that the mosaic-induced IgG response has lower affinity/avidity than IgG induced by the wild-type allergen.

However, the mosaic strategy may have several advantages over other strategies for generating hypoallergenic vaccines (e.g., T cell peptides, gene shuffling, fragmentation) (8, 21, 47). For example, the use of recombinant allergen fragments or synthetic peptides requires that treatment is performed with a cocktail of several different molecules to provide the necessary epitopes for Ab induction and T cell tolerance, whereas our approach allows treatment with a single reassembled molecule, which should greatly ease the production of the vaccine. The recently described random reassembly of DNA fragments of different gene fragments (21, 22) would be a similar approach but delivers a relatively large number of novel genes that have to be expressed and characterized regarding suitability. The induction of coincidently crossovers, deletions, insertions, inversions, and point mutations of genes through this molecular breeding technique or other mutational approaches may also lead to the loss of sequences necessary for the induction of Ab or T cell responses. In contrast, the construction of a mosaic molecule by allergen fragmentation and consecutive systematic recombination represents a more rational and targeted approach. It allows the retention of T cell epitopes and through additional structural analyses the reduction or even complete abolishment of allergenic activity by destroying conformational IgE epitopes.

Our approach also overcomes the problem that small hypoallergenic allergen fragments are poor immunogens and induce low or no allergen-specific IgG when injected as isolated molecules or as mix. In fact, we demonstrated that injection of rPhl p 2 and of a Phl p 2 peptide mix induced low (Phl p 2) or no detectable (peptide mix) anti-Phl p 2-specific IgG responses, whereas the P2M induced robust anti-Phl p 2 IgG responses. We also think that the mosaic approach will be applicable to many different allergens because we have recently succeeded in preparing a hypoallergenic mosaic for the major grass pollen allergen Phl p 1 (T. Ball and R. Valenta, unpublished).

Grass pollens contain several different allergens, but according to IgE prevalence testing and analysis of allergenic activity only four (i.e., Phl p 1, Phl p 2, Phl p 5, and Phl p 6) have been identified as the most important ingredients of a therapeutic grass pollen vaccine (27, 48). For three (Phl p 1, Phl p 5, Phl p 6) of these molecules, hypoallergenic derivatives suitable for vaccination have been identified (T. Ball and R. Valenta, unpublished and Refs. 36 ,49) and the Phl p 2 mosaic will now complete the panel of hypoallergenic molecules required for the formulation of a grass pollen vaccine. It has been recently shown that the expression of low immunogenic allergens from grass pollen (i.e., Phl p 2, Phl p 6) in the context of a recombinant hybrid allergen together with Phl p 1 and Phl p 5 increases their immunogenicity (41). A hybrid molecule consisting of wild-type allergens Phl p 1, Phl p 2, Phl p 5, and Phl p 6 has been shown to be sufficient for the diagnosis of grass pollen allergy in a skin prick testing study and induced higher levels of blocking Abs upon immunization of animals than did a mix of the allergens (50). It may therefore be considered to construct also hypoallergenic hybrid Ags by assembling the Phl p 2 mosaic with other hypoallergenic grass pollen allergen derivatives to further improve vaccination strategies based on hypoallergenic molecules. Clinical trials will now have to show the efficacy and safety of treatment with these molecules.

Rudolf Valenta is consultant to Biomay, Vienna, Austria