Molecular Characterization of Major Cat Allergen Fel d 1 EXPRESSION OF HETERODIMER BY USE OF A BACULOVIRUS EXPRESSION SYSTEM*

Fel d 1 is amajor cat allergen inducing allergic rhini-tis and asthma in sensitized individuals. It has a more complex structure when compared with other allergens and therefore expression of recombinant Fe l d 1 has been considered a challenge. The present study shows for the first time that a Baculovirus expression system is able to produce an intact rFel d 1 molecule that is glycosylated and structurally equivalent to the natural cat allergen, nFel d 1. Enzymatic digestion of rFe l d 1 and further analysis by use of matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) resulted in a complete coverage of the amino acid sequence of rFel d 1. In addition, the three disulfide bridges at the positions (cid:1) 70- (cid:2) 7, (cid:1) 44- (cid:2) 48, and (cid:1) 3- (cid:2) 73 were verified. The N -glycan structure of rFel d 1 was investigated by a combination of MALDI-TOF MS and monosaccharide analysis by high performance anion exchange chromatography with pulsed amperomet-ric detection (HPAEC-PAC). The N -glycosylation analyses of rFel 70% acetonitrile, 0.1% trifluoroacetic acid. To analyze the molecular mass of the (cid:1) - and (cid:2) -chain, 20 pmol of rFel d 1 was reduced by 45 m M dithio- threitol (Sigma) at 56 °C for 30 min and alkylated by 100 m M iodoacet-amide (Sigma) for 30 min at room temperature in the dark. Desalting and the MALDI-TOF MS analyses of the sample were then performed as described for the intact protein. Trypsin (NovoNordisk A/S, Bags- vaerd, Denmark), acyl-CoA-binding protein (Sigma), and a Se-quazyme TM peptide mass standard kit (Applied Biosystems) were used as external calibrants. Theoretical masses were calculated by GPMAW (Lighthouse Data, Odense, Denmark). Proteolytic Digestions— 500 pmol of rFel d 1 was reduced and alkylated as described above. The sample(s) were dissolved into 50 m M high performance anion exchange chromatography with pulsed amper- ometric detection; Tricine, N -[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]-glycine; Pipes, 1,4-piperazinediethanesulfonic acid; PBMC, peripheral blood mononuclear

Domestic cat (Felis domesticus) is one of the most common household pets, and cat dander is a significant source of indoor allergens. Approximately 20 -30% of individuals suffering from allergic asthma respond in vitro and/or in vivo to cat allergens (1). Sensitization to cat allergens leads into formation of allergenspecific IgE antibodies. Allergic reaction arises when these antibodies are able to facilitate activation of effector cells within the immune system. Activation of effector cells in return leads into release of histamine, eicosanoid synthesis, and cytokine gene expression (2). The subsequent immunological response per se is an outcome of cell mediators rather than allergen(s).
Specific allergen immunotherapy has been shown to be ef-fective in modulating allergic responses and resulting in downregulation of allergen-specific T-cell responses (3). Conventional specific allergen immunotherapy is performed by the use of standardized natural allergen extracts (4). Cat allergen extract contains a number of allergenic molecules, such as Fel d 1 (5)(6)(7), cat albumin (Fel d 2) (8), and cystatin (Fel d 3) (9). Fel d 1, however, has been considered to be the major allergen (1). Immunodominant characteristics have made it a prominent candidate in efforts to develop novel vaccines for treatment of cat allergy (10 -14). A foremost ongoing approach is the development of hypoallergenic vaccines, which are based on recombinant technology (3). Natural Fel d 1 is a noncovalently linked ϳ38-kDa dimer that is composed of two ϳ18-kDa subunits. Each subunit comprises ␣and ␤-chains that are encoded by two separate genes (15,16). Natural Fel d 1 isolated from cat dander is composed of a mixture of variants comprising both full-length and truncated versions of ␤-chain (17)(18)(19). The folding of the polypeptide chains results in an anti-parallel orientation of the ␣and ␤-chain(s) held together with three disulfide bridges (20). Correct orientation of the ␣and ␤-chain(s) is considered to be critical because IgE-related antigenic determinants are conformation-dependent (17)(18)(19)(20)(21)(22)(23). Further analyses of the natural molecule have verified that the single N-glycosylation site in the ␤-chain is carrying heterogeneous triantennary complex type of structures. This heterogeneity is caused by terminal sialic acids, fucose, and ␤-galactose residues (20).
The anti-parallel orientation of the ␣and ␤-chains together with conformation-dependent allergenicity has caused major challenges in efforts to express Fel d 1 as a recombinant (r) allergen. Both the ␣and ␤-chain(s) have been expressed separately in Escherichia coli (18,19,24,25). Individual chains, however, show markedly reduced immunogenicity (17-19, 24, 25). Recent studies have focused on expression of Fel d 1 as a fusion protein in a Baculovirus expression system (12) and as a His-tagged homodimer in E. coli (14). The recently solved three-dimensional structure of a head-to-tail construct expressed in E. coli suggests that Fel d 1 belongs to the secretoglobin family of proteins (26). Prokaryotic expression systems, however, are lacking machinery that are needed for construction of specific post-translational modifications (27). Posttranslational modifications such as glycosylation patterns are considered to be important when assessing the structure and immunological properties of recombinant glycoproteins (28).
Here we report the results of the structural characterization of rFel d 1 that was expressed in Trichoplusia ni insect cells. In the present study mass spectrometry (MS) 1 and Edman degra-* This work was part of the Therapeutic recombinant allergens from structural allergology (TRAFSA) project, funded by European Union 5th Framework Programme Grant QLK3-CT-1999-00620. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ Supported by a long term postdoctoral fellowship from the Federation of European Biochemical Societies.
ʈ To whom correspondence should be addressed. dation was combined to analyze the amino acid sequence and disulfide bridge formation in rFel d 1. The N-glycan structure of rFel d 1 was investigated by combination of MALDI-TOF MS and monosaccharide analysis by HPAEC-PAD. The immunobiological characterization and comparison with nFel d 1 was performed in vitro using IgE inhibition, histamine release, and lymphocyte proliferation assays.

Construction of the Synthetic Genes Encoding Fel d 1 Polypeptide
Chains-Synthetic genes encoding Fel d 1 ␣and ␤-chains (␣: M74952/␤: M77341) were assembled by PCR using overlapping oligonucleotides (Table I). In short, primers A1ϩA2 and A3ϩA4 for the ␣-chain and B1ϩB2 and B3ϩB4 for the ␤-chain were annealed and amplified in six cycles. Extension primers ␣-chain (ARF/ARR) and ␤-chain (BRF/BRR) with specific restriction sites were used to rescue the full-length products. The fragments were then cloned into pCR4-TOPO vector (Invitrogen), and their sequences were confirmed (ABI PRISM® 377 DNA sequencer; Applied Biosystems, Framingham, MA).
Construction of Recombinant Baculovirus Vector-The construction of the recombinant Baculovirus vector and the expression of the rFel d 1 was performed by CeNeS Pharmaceuticals plc (Cambridge, UK). The cDNA encoding the mature ␣and ␤-chains were subcloned into Baculovirus shuttle vector pFastBacDUAL (Invitrogen). The mellitin signal sequence was engineered to the N-terminal of both genes by PCR (Fig.  1). The recombinant dual vector was then transformed into MAX efficiency DH10Bac TM competent cells (Invitrogen) containing the baculovirus genome. Within the cell a transposition takes place between a mini-attTn7 target site and the mini-Tn7 element on the vector when recombinant virus is generated. The resulting recombinant bacmid was confirmed according to the suppliersЈ manual (Invitrogen).
Expression of the rFel d 1-Spodoptera frugiperda (Sf21) (Invitrogen) cells were used to generate the primary virus titer for the expression of rFel d 1. The Sf21 cells were grown in 47.5% ExCell 401(JHR Biosciences), 47.5% TC100 (Invitrogen) and 5% heat-inactivated fetal bovine serum (Invitrogen) as suspension cultures in shaker flask(s) (24). Sf21 cells were transfected with recombinant bacmid DNA in the presence of Lipofectin (Invitrogen) according to the supplier's manual (Invitrogen). The culture medium was collected 7 days post-transfection. The virus titer was assessed by plaque assay (29). For large virus stock production SF21 cells were infected with the recombinant virus at a multiplicity of infection of 0.5 at a cell density of 1 ϫ 10 6 cells/ml in a spinner bottle. The virus was harvested 7 days post-infection and titered again by plaque assay. High Five TM cells (Invitrogen) were used for protein production and cultured according to the manufacturer's instruction. For protein production the cells were infected at a multiplicity of infection of 10. The cell supernatant was harvested 96 h post-infection.
Purification of Natural and rFel d 1-Natural Fel d 1 was isolated by use of monoclonal immunoaffinity chromatography from dried cat allergen extract (ALK-Abelló) as described (20). Natural Fel d 1 was then subjected to HR 5/5 Mono Q (Amersham Biosciences) column and eluted with a linear gradient of 0 -100% 20 mM Tris-Cl, 0.5 M NaCl, pH 7.5, in 20 min. Eluted fractions containing nFel d 1 were pooled and subjected to size exclusion chromatography (Amersham Biosciences) in a 10 mM NH 4 (HCO) 2 buffer and freeze dried. For the immunological assays, nFel d 1 was dissolved into sterile Dulbecco's phosphate-buffered saline (Invitrogen). Endotoxin levels in the purified Fel d 1 preparations were Ͻ15 EU/mg as determined by Limulus amoebocyte lysate assay (Bio Whittaker, Walkersville, MD).
Recombinant Fel d 1 was isolated from the High Five TM cell culture supernatant. The cells were gently centrifuged, and the culture supernatant was collected and sterile filtrated. The purification of rFel d 1 followed the protocol described for the nFel d 1.
SDS-PAGE Analysis of Affinity Purified rFel d 1-16% Tris-Tricine SDS-PAGE (Invitrogen) was used to analyze collected fractions (10 l/10 ml) from monoclonal affinity chromatography. The electrophoresis was performed in reducing conditions according to the manufacturer's instructions and stained with silver (Invitrogen).
Protein Measurements-A Lambda 800 UV-visible spectrophotometer (PerkinElmer Life Sciences) was used to measure protein concentration at 280 nm using the absorption coefficient A 280 (1 mg⅐ml Ϫ1 cm Ϫ1 ) ϭ 0.356 for both rFel d 1 and nFel d 1.
The fractionation of the peptide mixtures were performed in a 4.6 ϫ 250-mm Jupiter C18 (Phenomenex) column. The peptides were eluted with 0.05% trifluoroacetic acid, 80% acetonitrile in a gradient of 5-45% in 30 column volume or 45-80% in 5 column volume. The separated peptides were dried in a vacuum Speedvac, redissolved into water, and stored at Ϫ20°C.
Mass Spectrometric Analyses-Mass spectrometric analysis was performed on a Voyager-DE TM STR Biospectrometry TM (Applied Biosystems, Foster City, CA) MALDI-TOF MS instrument by use of acceleration voltage of 25 kV and a nitrogen laser at 337 nm. The spectra were acquired in the positive ion mode and calibrated externally. MS/MS of selected peptides was performed on a Micromass® Q-Tof Ultima MALDI mass spectrometer (Waters Corporation, Manchester, UK). 20 pmol of intact rFel d 1 was desalted by custom-made micro columns packed with POROS R1/50 (Applied Biosystems, Framington, MA) reversed-phase resin (30). The samples were eluted directly on MALDI targets with 1.0 l of matrix solution containing 10 g/l ␣-cyano-4-hydroxycinnamic acid or 2,5-dihydroxybenzoic acid in 70% acetonitrile, 0.1% trifluoroacetic acid. To analyze the molecular mass of the ␣and ␤-chain, 20 pmol of rFel d 1 was reduced by 45 mM dithiothreitol (Sigma) at 56°C for 30 min and alkylated by 100 mM iodoacetamide (Sigma) for 30 min at room temperature in the dark. Desalting and the MALDI-TOF MS analyses of the sample were then performed as described for the intact protein. Trypsin (NovoNordisk A/S, Bagsvaerd, Denmark), acyl-CoA-binding protein (Sigma), and a Sequazyme TM peptide mass standard kit (Applied Biosystems) were used as external calibrants. Theoretical masses were calculated by GPMAW (Lighthouse Data, Odense, Denmark).
Acid Hydrolysis and Monosaccharide Analysis-1 mg of rFel d 1 was hydrolyzed in 2 M trifluoroacetic acid at 100°C for 4 h, subsequently lyophilized in a Speedvac, and resuspended in water. The monosaccharides were analyzed by HPAEC-PAD using a Dionex (Sunnyvale, CA) P-500 chromatographic system consisting of Carbo PacPA-10 preanalytical and analytical columns, an AS50 auto sampler, a GP40 gradient pump, and an ED40 electrochemical detector. The flow rate was 1 ml/min, and the injection volume was 10 l. Separation was achieved isocratically with water as eluent. A pump with 600 mM NaOH as eluent was used to improve the sensitivity of detection. L-fucose (Merck), D-xylose (Fluka), D-mannose (Merck), D-glucose (Merck), Dgalactose (BDH Chemicals Ltd., Poole, UK), GlcNAc (BDH Chemicals Ltd.), N-acetylgalactosamine (BDH Chemicals Ltd.), and D-arabinose (Sigma) were treated as described above and used as standards.
Structural Analysis of the N-Glycan-Removal of N-glycan was assessed with recombinant N-glycosidase F from Flavobacterium meningosepticum (Roche Applied Science) and N-glycosidase A from almonds (Roche Applied Science) at 37°C for 18 h. The resulting products were then analyzed by MALDI-TOF MS (Applied Biosystems) (30). The sequential digestions with glycosidases were performed following the procedure of Kroll-Kristensen et al. (20).
IgE Inhibition Analyses-The IgE inhibition experiments were performed on an ADVIA Centaur Immunoassay system (Bayer Diagnostics, Denmark) (31). Pooled human serum IgE from cat-allergic individuals (n ϭ 4) was coupled to the solid phase (ADVIA Centaur Universal Reagent Pack, no. 123736), after which the surface was washed with 0.01 M sodium phosphate, 0.1% (w/v) human serum albumin, pH 7.4 (ADIVIA Centaur reagent, no. 1011132). Mixtures of a constant amount of biotinylated cat allergen extract and diluted allergen preparations were added. The amount of biotinylated Fel d allergens bound to the solid phase absorbed IgE was determined as the number of relative light units after the addition of ADVIA Centaur Lite reagent (Universal reagent pack). All of the inhibition experiments were performed as triplicates, and the data sets (log 10 (concentration), mean (DoB)) were fitted to a four-parameter logistic function using GraphPad Prism version 4.01 (GraphPad Software, San Diego, CA).
Histamine Release Assay-Histamine release assay was performed using freshly drawn blood from cat-allergic individuals (n ϭ 4) and controls (n ϭ 4). The tested individuals gave their informed consent to donate blood for research purposes. The antigens, rFeld1 and nFeld1, were diluted into Pipes buffer (Invitrogen). Antigens were then mixed with the blood samples and incubated for 30 min at 37°C. Analyses were performed in a final concentration ranging from 1.5 pg/ml to 500 ng/ml of antigen(s). The samples were then centrifuged, and the supernatants were analyzed by enzyme-linked immunosorbent assay (kit IM 2015; Immunotech). The release of histamine was measured at 405 nm by an EL 340 Biokinetics reader (Bio-Tek Instruments, Winooski, VT).
Lymphocyte Proliferation-T-cell lines specific to Fel d 1 were established from peripheral blood mononuclear cells (PBMCs) of cat-allergic patients (n ϭ 5) as described previously for grass allergen-specific T-cell lines (32). In short, freshly isolated PBMCs (2 ϫ 10 6 /ml) were stimulated in 1-ml bulk cultures with natural or rFel d 1 (2 g/ml) for 14 days with the addition of recombinant interleukin-2 from day 5. After 14 days T-cells were restimulated with irradiated autologous PBMCs, Fel d 1 (2 g/ml), and 0.05 g/ml phytohemagglutinin-P (Difco, Detroit, MI), and recombinant interleukin 2 was added at days 3, 4, and 5.
T-cell Stimulation Assay-On day 10 after restimulation, T-cells (2 ϫ 104/well) were cultured with autologous PBMCs (105/well, irradiated 2500 Rad) in 200 ml of RPMI 1640 medium supplemented with 5% v/v AB serum (Cambrex Bio Sciences), 100 units/ml penicillin, and 0.1 mg/ml streptomycin (Sigma). The culturing was performed with or without antigen in 96-well round-bottomed microtiter plates (Nunc). The cells were cultured for 72 h in a humidified atmosphere at 37°C and 5% CO 2 , followed by a 16-h pulse with 0.5 C [ 3 H]thymidine/well, and thymidine incorporation was determined by scintillation counting. The results are expressed as the mean cpm values of four replicate cultures.

RESULTS
Purification of rFel d 1-Expression of rFel d 1 was demonstrated from the insect cell culture supernatant by SDS-PAGE and Western blotting (data not shown). Monoclonal affinity chromatography was used to capture rFel d 1 from culture supernatant. SDS-PAGE analysis showed that rFel d 1 was bound to the monoclonal antibody column and eluted in a single step. Both ␣and the ␤-chain were detected when stained with silver (Fig. 2). Affinity purified native rFel d 1 was then analyzed by MALDI-TOF MS, which revealed a major peak in the m/z ranging from 18,000 to 19,000. In addition, a peak at m/z 8000 was seen, indicating overexpression and co-purification of free ␣-chain (data not shown). Anion exchange and size exclusion chromatography were used to remove the free ␣-chain and other impurities. For MS analyses, rFel d 1 was subjected to reversed-phase chromatography. The total yield of purified dry weight rFel d 1 was from 1 to 3 mg/liter of culture supernatant. The purity of reduced and nonreduced natural and rFel d 1 was verified by MALDI-TOF MS as described below.
Analysis of Native and Reduced rFel d 1-Analysis of rFel d 1 by MALDI-TOF MS revealed several peaks in the m/z range from 18353.48 to 19377.20 (Fig. 3a). A closer inspection of the mass spectrum revealed spacing between peaks that is diagnostic of glycosylation. In addition, a minor peak (Fig. 3a, asterisk) at m/z ϳ17,700 referring to nonglycosylated rFel d 1 was detected. Following reduction and S-carboxyamidomethylation a peak appeared at 8043.60, and several minor peaks appeared at in the mass range from 10700.40 to 11385.68 (Fig.  3b). The observed signal at 8043.60 was in agreement with the theoretical value m/z 8043.25 for the ␣-chain. This result indicates to a correct cleavage at the N terminus and overall homogeneous expression of the ␣-chain.
The peaks in the m/z range 10,700.40 to 11,385.68 showed similar spacing as the native molecule, indicating glycosylation of the ␤-chain (Fig. 3b). As described below, subsequent diges- Primary Sequence Analysis and Orientation the rFel d 1 ␣and ␤-Chains-Reduced and carboxymethylated rFel d 1 was digested with endoproteinase Asp-N, and the peptides were separated by reversed-phase chromatography. The peptides were identified by N-terminal sequencing, MS-MS, and/or MALDI-TOF MS. These peptides are referred to as F-1 to F-20 in a Table II. Enzymatic digestion with Asp-N gave full sequence coverage for the ␤-chain. The valine in the ␣-chain at position 10 was verified from the enzymatic cleavage with combination of trypsin and chymotrypsin (data not shown).
Partial oxidation of the methionine residues was detected in peptides F-7, F-12, and F-13 when analyzed by MALDI-TOF MS, and partial deamidation was found in peptides F-16/F-17. In addition, recombinant Fel d 1 was found to be expressed as a mixture of glycosylated (F-7) and nonglycosylated (F-9) ␤-chain (Table II). Analyses of the N-glycan will be described below. Modification referring to ␥-carboxylation was found from the N terminus of the ␣-chain (F-2). In N-terminal sequencing, the peptide was found to be blocked, and it was subsequently identified by MS-MS (data not shown). This finding suggests that a subpopulation of rFel d 1 would have modified N termini.
To determine the positions of the disulfide bond linkages, native rFel d 1 was digested with endoproteinase Asp-N and a combination of trypsin and chymotrypsin. Separation of the peptides was performed by reversed-phase chromatography, and all of the resulting peptides were analyzed by MALDI-TOF MS. The combined digestion with trypsin and chymotrypsin revealed disulfide bond linkages between residue(s) ␣44-␤48 and ␣3-␤73. The third disulfide bond between residue(s) ␣70-␤7 was demonstrated by digestion with endoproteinase Asp-N. The observed m/z of this peptide was 2338.25 Da. The disulfide bond formation between all three peptides was verified by reduction with dithiothreitol (Table III). No unpredicted bonds between the cysteine residues and/or free cysteine residues were found.
Analyses of the N-Glycan Structure-Deglycosylation of the rFel d 1 glycopeptide F-7 (Table II) was achieved using N-glycosidase A (Fig. 4). Analysis by MALDI-TOF MS revealed that the expected nonglycosylated peptide was released along with several glycans of different m/z values (Fig. 4a). Deglycosylation of the same glycopeptide (F-7) using N-glycosidase F was not detected. N-Glycosidase A is found active on glycans carrying a ␣1.3-linked fucose residue attached to the innermost GlcNAc residue (Fig. 4b). These results suggest that the glycan core is carrying ␣1.3-fucosylation, which is in correlation with the previous studies of T. ni-expressed recombinant proteins (33).
To further determine the composition of the glycan moiety of rFel d 1, the sample was hydrolyzed, and the released monosaccharides were analyzed by HPAEC-PAD. Fig. 5 shows the HPAEC-PAD chromatogram of monosaccharides released from rFel d 1. The results indicate that the N-glycan is composed of mannose, fucose, GlcNAc, and N-acetylgalactosamine residues. The m/z values and monosaccharide composition of the released glycans was used to predict their structures using GlycoMod (Expasy). The proposed structures are presented in Table IV. Supplementary characterization of the N-glycosidic glycans was performed by MALDI-TOF MS in combination with sequential glycosidase treatment. Removal of the terminal galactose residue was demonstrated by the sequential digestion; however, no evidence of sialic acids was detected (data not shown).
Competition for Allergen-specific IgE Antibodies-The presence of conformational specific IgE epitopes on recombinant Fel d 1 was addressed in an IgE inhibition assay using a serum pool derived from four cat-allergic patients. Serum IgE was captured by anti-IgE immobilized on paramagnetic beads. After washing, the binding of biotinylated cat extract to captured IgE was inhibited by the addition of dilution series of recombinant Fel d 1, natural Fel d 1, and extract, respectively. The inhibition curves determined for cat allergen extract, nFel d 1 and rFel d 1 are shown in Fig. 6. Natural Fel d 1 and rFel d 1  Histamine Release Assay-To investigate the biological effect of recombinant and natural Fel d 1 to stimulate basophil, degranulation was tested in histamine release assay using human basophiles derived from individual cat-allergic patients. The release of histamine was found consistent for both recombinant and nFel d 1 in all the four cat-allergic individuals tested, whereas the controls suffering from pollen allergies showed no release. Even though the molar amount of released histamine varied from patient to patient, the histamine response within the patients remained equal between the recombinant and the natural antigen (Fig. 7).
T-cell Proliferation-Allergen-specific T-cell cultures were obtained in most cases when PBMC established from cat-allergic patients were stimulated with natural and recombinant Fel d 1. However, initial stimulation with nFel d 1 resulted in specific T-cell lines in five of five patients, whereas lines were obtained from three of five patients with rFel d 1. When the individual T-cell lines were stimulated with either natural or recombinant Fel d 1, comparable responses were generally obtained even though the response differed between natural and recombinant Fel d 1 in some T-cell lines (Fig. 8). In addi-  tion, all lines examined showed a clear Th2 cytokine profile with a high interleukin 5/interferon-␥ ratio (CBA assay; BD Biosciences) (data not shown).

DISCUSSION
The ability to produce recombinant allergens with intact immunochemical properties and correct amino acid sequence and structure is of major importance for their potential use as diagnostic agents and as active ingredients in vaccines. Here we demonstrate the expression of recombinant Fel d 1 as a heterodimer that is structurally and immunochemically equivalent to the naturally occurring cat allergen Fel d 1.
The successful expression of rFel d 1 heterodimer was obtained by cloning the genes encoding Fel d 1 ␣and ␤-chains one by one into a dual gene plasmid (Fig. 1). The expressed ␣and ␤-chains were detected to from the cell culture medium as a mixture of free ␣-chain and intact ␣/␤-heterodimer. As shown in several previous studies (17)(18)(19)(20)(21)(22)(23), monoclonal immunoaffinity chromatography was found to be particularly important in purification of both natural and recombinant Fel d 1. In addition, the co-purification of free ␣-chain verified that the monoclonal antibody used in the present study was Fel d 1 ␣-chainspecific. The following purification steps using anion exchange and size exclusion chromatography verified that rFel d 1 shares similar physical and chemical properties with nFel d 1.
The MS analyses of rFel d 1 were designed based on the methods that were previously optimized for nFel d 1 (20). We presumed that expressed as a glycosylated heterodimer, rFel d 1 would show structural characteristics similar to nFel d 1. This was first demonstrated when the three disulfide bridges were assessed. Only after sequential cleavage by trypsin and chymotrypsin were two of the rFel d 1 disulfide bridges (␣44-␤48 and ␣3-␤73) verified. Similar results were previously obtained with nFel d 1 (20). The third bridge (␣70-␤7) was demonstrated after cleavage by endoproteinase Asp-N (Table III). These results indicate that rFel d 1 is a stable molecule resisting enzymatic cleavage in a manner similar to nFel d 1. Demonstration of the disulfide bridges also showed that rFel d 1 ␣and ␤-chains are in anti-parallel orientation as described for the nFel d 1 (20). Correct orientation of the two-polypeptide chains has been shown to be crucial for the allergenicity of this molecule, because the B-cell epitopes of nFel d 1 have shown to be conformation-dependent (17)(18)(19)(21)(22)(23). In inhibition assays, natural and rFel d 1 showed similar immunochemical characteristics in competition for cat allergen-specific IgE antibodies. This result suggests that rFel d 1 is able to cover all of the conformational IgE epitopes presented by nFel d 1. This conclusion was further supported by basophil histamine release results. Surface-bound IgE antibodies were stimulated by recombinant and nFel d 1, showing similar dose-dependent release of histamine. Furthermore, the molar release of histamine was found to be consistent when compared with individual total IgE levels (Fig. 7).
Detailed characterization of post-translational modifications from the natural allergen source is necessary before the expression of recombinant allergens. Post-translational modifications can affect the specific activity, recovery, half-life in the circulation, and immunogenicity of proteins (28). Sequence analysis of rFel d 1 by MS showed few but significant modifications within the molecule (Table II, footnote a). Partial deamidation of the asparagine residue was found from peptide F-16 (Table  II). Previously deamidation was considered to be a purification artifact. Currently, it is viewed as a signal for protein aging or damage (34). Furthermore, in individuals suffering from celiac disease, deamidation induced by tissue transglutaminase is found to be clinically significant, resulting from immunomodulation of gliadin peptides (35). In the present study, partial deamidation in rFel d 1 may result from extended expression time (96 h). However, considerable changes for the immunogenicity in vitro were not detected either in IgE inhibition and/or histamine release assays. Deamidation of rFel d 1 may, however, have influenced the initial stimulation of some T-cell lines, modifying T-cell-specific epitopes (34,36). On the other hand, comparable results were obtained in many cases when the lines were tested with the two Fel d 1 preparations.
Few groups have explored receptor-specific interactions between allergens and cells that are involved with allergen uptake (37). ␥-Carboxylation was first discovered in proteins of the blood coagulation cascade (38), and it has recently been shown to facilitate the interaction of proteins with membrane receptors (39). Because this modification was not previously detected in nFel d 1 (20), the potential immunostimulatory role of ␥-carboxylation in addition to deamidation(s) found in the subpopulation of rFel d 1 needs to be further investigated.
Numerous allergens are glycoproteins; however, the role of glycosylation versus allergenicity is poorly understood. Furthermore, the studies of glycosylation have mainly been focused on IgE or IgG binding and/or histamine release responses (22)(23)40). In the present study, the rFel d 1 ␤-chain was expressed with and without N-glycosylation (Table II, F-7/F-9). Furthermore, the glycosylation analysis suggests that the Nglycosylation profile of the rFel d 1 is different from nFel d 1. Natural cat allergen was shown to carry terminal sialic acids as well as terminal fucose residues, whereas they were not detected when T. ni expressed the rFel d 1. In addition, rFel d 1 was shown to carry core ␣-1.3-fucosylation, which is not found present in nFel d 1 (20). Core ␣1.3-fucosylation is commonly found in plant and bee venom allergens (40). Previously studies have suggested that glycosylation is playing important structural and immunolobiological roles in nFel d 1 (20,23). In the present study different glycosylation patterns between natural and rFel d 1 did not show a significant effect for IgE-related immunological assays in vitro. However, recent findings by Cobb (41) show that some carbohydrates may facilitate important immunological responses through T-cell activation. Therefore, further studies are needed to elucidate the immunoregulatory role of specific carbohydrate structures in allergy. FIG. 8. T-cell lines generated from the peripheral blood of two cat-allergic patients through repetitive stimulation with natural or recombinant Fel d 1. Comparable levels of T-cell activation were found when Fel d 1-specific T-cell lines were initiated with natural and recombinant Fel d 1. The lines were subsequently stimulated with each allergen preparation as indicated for comparison. a, T-cell lines from 5 patients scored as follows. b, the exact response pattern of two individual patients (SI ϭ cpm in allergen stimulated/cpm in cultures grown in medium alone (background)). ϩ, SI 3; ϩϩ, SI Ͼ 10; ϩϩϩ, SI Ͼ 30.