The study protocol was approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University, Nanjing, China. Written informed consent for the use of blood samples was obtained from all participants prior to study entry according to the declaration of Helsinki.

A total of 16 patients with allergic rhinitis with positive SPT results (allergens were supplied by ALK-Abelló, Inc., Hørsholm, Denmark) and with positive serum IgE test results to American CR extract [by using ImmunoCAP assay (Pharmacia Diagnostics AB, Uppsala, Sweden)], and 6 healthy controls (HC) were recruited in this study. Serum (4 ml) from peripheral venous blood was collected from each patient and the healthy controls for western blot analysis.

Total RNA was isolated from adult female CRs reared at our institute using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). Total RNA was quantified by measuring the absorbance ratios at 260/280 nm. cDNA was prepared by reverse transcriptase using a commercial RNA-PCR kit according to the manufacturer's instructions (Takara Biotech Co., Ltd., Dalian, China). For each reaction, 1 μg of total RNA was reverse transcribed using oligo-d(T). cDNA encoding Per a 9 was amplified by PCR using primers based on the no-coding sequence of the Per a 9 gene (AY563004.1; forward, 5′-TACA GCAAGTGGGACAGCAG-3′ and reverse, 5′-ATATGGGCAT CAAAGATATA-3′). The PCR conditions were 95°C/5 min (1 cycle), 95°C/1 min, 55°C/1 min and 72°C/1 min (30 cycles), and 72°C/5 min (1 cycle). The purified PCR product was cloned into the pMD18-T vector (Takara Biotech Co., Ltd.), before being transformed into Escherichia coli (E. coli) strain DH5α. The inserts were sequenced on an ABI Prism 377 DNA sequencer (Applied Biosystems, Foster City, CA, USA). DNA sequence data were translated to amino acid sequence using the Show Translation tool in the SMS software package (http://www.bioinformatics.org/SMS/).

The Per a 9 gene was subcloned into the pET15b vector (Novagen, Madison, WI, USA) using the NdeI and BamHI sites and verified by DNA sequencing. The recombinant pET15b-Per a 9 plasmid was transformed into the ArcticExpress™ (DE3) RP host strain. A colony of the selected transformed ArcticExpress™ (DE3) RP E. coli on an overnight LB-ampicillin agar plate was inoculated into 5 ml of LB-ampicillin broth, and incubated at 15, 25 or 37°C, respectively overnight. One milliliter of the culture was inoculated into 50 ml of fresh LB-ampicillin broth and incubated at 37°C with shaking at 250 rpm until the optical density (OD) at A_600nm reached 0.6. Subsequently, IPTG was added to a final concentration of 1 mM and the culture was incubated for a further 4 h. The bacterial cells were harvested by centrifugation at 4,000 × g at 4°C for 20 min, and were lysed in lysis buffer by sonication at 20 kHz, 2 min pulse-on, 3 min pulse-off. Cell debris was removed by centrifugation at 12,000 × g at 4°C for 20 min. The supernatant was loaded on a Nickel column (Genscript, Nanjing, China), washed with running buffer containing 50 mM Tris-HCl, 300 mM NaCl and 5% glycerol (pH 8.0), and eluted with elution buffer containing 50 mM Tris-HCl, 300 mM NaCl, 50 and 250 mM imidazole and 5% glycerol (pH 8.0). The eluted fractions washed with 250 mM imidazole were obtained and identified as Per a 9.

A 96-well plate was coated with purified recombinant Per a 9 at 10 μg/ml in carbonate-bicarbonate buffer (0.05 M, pH 9.6) overnight at 4°C, 100 μl/well. Human serum samples [1:20 dilution in phosphate-buffered saline (PBS)-Tween-20 with 2% BSA] were then added to the plates for 2 h at room temperature. Following IgE binding, the plates were incubated with horseradish peroxidase-labeled goat anti-human IgE (1:2,500 dilution) (KPL, Inc., Gaithersburg, MD, USA), and the color was developed with tetramethylbenzidine peroxidase substrate. The plates were read on a microplate reader (Eon; BioTek, Winooski, Vermont, USA) at an absorbance of 405 nm. The cut-off of the enzyme-linked immunosorbent assay (ELISA) was calculated as the mean of the negative controls plus 2 standard deviations (SDs).

Immunoblots for the detection of serum specific IgE were performed using recombinant Per a 9 as previously described (19,20). Recombinant Per a 9 (5 μg) was added to a sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (gel concentration of 15%) under reducing conditions and then transferred onto nitrocellulose membranes. The nitrocellulose membranes were incubated with the sera of the patients with American CR allergies (1:5 to 1:20 in PBS-Tween-20 with 1% BSA, 10% normal goat serum) for 90 min. Following rinsing with PBS, the membranes were incubated with peroxidase-labeled anti-human IgE monoclonal antibody. The positive protein bands were visualized by incubating the membranes with tetramethylbenzidine peroxidase substrate. Sera from 2 non-atopic subjects were used as negative controls.

The expression of CD63 and CCR3 on the basophil surface is considered as an indicator of basophil activation (21,22). Briefly, peripheral blood mononucleated cells (PBMCs) from 20 ml blood donated by 4 healthy volunteers were separated by Ficoll-Paque density gradient, and treated with 10 ml LS (a solution containing 1.3 M NaCl, 0.005 M KCl and 0.01 lactic acid, pH 3.9) for 2 min at 8°C. Following neutralization with 12% Tris (pH 10.9), non-specific IgE on the basophils was stripped off and the cells were passively sensitized with the sera of patients with American CR allergies or the healthy controls (n=4, 1 in 10 dilution, 2 h at 37°C) (same patients and controls as mentioned above) as previously described (22). The cells were then challenged with Per a 9 (1.0 μg/ml) for 15 min at 37°C. A goat anti-human IgE antibody (Serotec, Kidlington, UK) was used as a positive control. CCR3-PE-labeled antibody (85-12-1939–42; eBioscience Inc., San Diego, CA, USA) and anti-human CD63-FITC antibody (HH-MHCD63014; Invitrogen) were added to the cells for 15 min at 37°C. Flow cytometric analysis of surface markers was performed at 488 nm on a FACSAria flow cytometer (Becton-Dickinson, Franklin Lakes, NJ, USA) and analyzed by FACSDiva software.

The complete amino acid sequence of the cloned Per a 9 gene was used as query to search for homologous sequences through the Swiss-Prot/TrEMBL (Uniprot; http://www.uniprot.org/) and tBLASTn in NCBI (blast.ncbi.nlm.nih.gov/Blast.cgi) (10,23–25). The phylogenetic tree was obtained by using the maximum-likelihood (ML) method on the basis of the JTT amino acid sequence distance implemented in MEGA 5.1, and the reliability was evaluated by the bootstrap method with 1,000 replications (24–32).

Physiochemical analysis, including molecular weight, theoretical pI, amino acid composition, instability index, aliphatic index and the grand average of hydropathicity (GRAVY) of Per a 9 was performed using the ProtParam tool (http://web.expasy.org/protparam/), as previously described (33). The Per a 9 characteristic pattern was examined for the original sequence and further analysis was performed to highlight the presence of functional motifs using the Prosite database (http://prosite.expasy.org/), as previously described (34).

Per a 9 secondary structural elements recognition was assessed by PSIPRED (bioinf.cs.ucl.ac.uk/psipred), which threads sequence segments through Protein Data Bank (PDB) library (http://www.pdb.org/) to identify conserved substructures (35). Furthermore, the secondary structure elements were also identified and compared with the results obtained with NetSurfP ver. 1.1 (www.cbs.dtu.dk) (36).

The Per a 9 protein sequence was searched for homology in the PDB (http://www.rcsb.org/). In addition, the homologous templates suitable for Per a 9 were selected using the PSI-BLAST server (http://blast.ncbi.nlm.nih.gov/Blast.cgi) in the NCBI and Swiss-model server (http://swissmodel.expasy.org/). The best templates were retrieved from the results of PSI-BLAST and used for homology modeling. The Der f 25 modeled protein structure was built through Alignment Mode in SWISS-MODEL (http://swissmodel.expasy.org/) using the complete amino acid sequence. An initial structural model was generated and checked for recognition of errors in 3D structure by PROCHECK (37), ERRAT (verification of protein structures: patterns of nonbonded atomic interactions) (33,38) and VERIFY_3D (a method to identify protein sequences that fold into a known three-dimensional structure. Assessment of protein models with three-dimensional profiles) programs in Structural Analysis and Verification Server (http://nihserver.mbi.ucla.edu/SAVES/) (33,38).

Three immunoinformatics tools, including the DNASTAR Protean system (39), Bioinformatics Predicted Antigenic Peptides (BPAP) system (http://imed.med.ucm.es/Tools/antigenic.pl) and the BepiPred 1.0 server (http://www.cbs.dtu.dk/services/BepiPred/) were used to predict the B cell epitopes of Per a 9, as previously described (40). The ultimate consensus epitope results were obtained by combining the results of the three tools together with a previously published method (41). If the results of all 3 methods were non-epitope, then the consensus result was 0% epitope. Similarly, if the predicted results had only one or no non-epitope, the consensus result was 67 or 100% epitope, respectively. Finally, the regions whose consensus epitope result was 67 or 100% were selected as the final potential epitope regions. In the DNAStar Protean system, 4 properties (hydrophilicity, flexibility, accessibility and antigenicity) of the amino acid sequence were selected as parameters for epitope prediction. The peptide regions with good hydrophilicity, high flexibility, surface accessibility and high antigenic index were selected as candidate epitopes for further investigation. The BPAP system and BepiPred 1.0 server only need the amino acid sequence and provide more straightforward results, which combined with the physicochemical properties of amino acids such as hydrophilicity, flexibility, accessibility, turns and exposed surface (42).

T cell epitopes are principally predicted indirectly by identifying the binding of peptide fragments to the MHC complexes. However, the binding grooves of MCH-II molecules are open at both ends allowing various lengths of peptides to bind. On the other hand, the same MHC molecule can accommodate a variety of binding sequences. These two properties make the development of accurate predictive algorithms for MHC-class II binding complex. For HLA-DR-based T cell epitope prediction, the artificial neural network-based alignment (NN-align) method NetMHCIIpan-3.0 (http://www.cbs.dtu.dk/services/NetMHCIIpan/) was applied (43). For HLA-DQ alleles, NetMHCII-2.2 (http://www.cbs.dtu.dk/services/NetMHCII/) was used (44). Although HLA-DQ provides limited binding-affinity data, it was recently reported to provide the optimal performance in predicting this locus (33). The binding significance of each peptide to the given MHC molecule is based on the estimated strength of binding exhibited by a predicted nested core peptide at a set threshold level. For HLA-DR-based T cell epitope prediction, HLA-DR 101, HLA-DR 301, HLA-DR 401 and HLA-DR 501 were used. The ultimate HLA-DR-based T cell epitope results were obtained by combining those 4 results together and if 3 of these were shown to be epitope, the consensus result was then considered epitope. This method was also used in HLA-DQ-based T cell epitope prediction. For HLA-DQ-based T cell epitope prediction, only HLA-DQA10501-DQB10201, HLA-DQA10301-DQB10302, HLA-DQA10401-DQB10402 and HLA-DQA10102-DQB10602 were used. As a result, the ultimate consensus epitope results were obtained by combining the results of the HLA-DR-based T cell epitope and HLA-DQ-based T cell epitope. B cell and T cell epitopes identified by computational tools were mapped onto linear sequence and on the 3 dimensional model of Per a 9 in order to determine their position and secondary structure elements involved.

Data are expressed as the means ± SE for the indicated number of independently performed duplicated experiments. Statistical significance between means was analyzed by one-way ANOVA or the Student's t-test utilizing the SPSS 13.0 version. A value of P<0.05 was considered to indicate a statistically significant difference.

cDNA encoding Per a 9 was amplified by PCR using primers based on the no-coding sequence of Per a 9 gene. It is a 1,074 bp gene and encodes a 356 amino acid protein (Fig. 1). The sequence homology with the published one (Accession no. AY563004) was 99% (353/356) at the protein level.

The Per a 9 gene was subcloned into the pET15b vector and transformed into the ArcticExpress™ (DE3) RP host strain. Per a 9 was expressed at 15, 25 or 37°C following induction. Since Per a 9 was expressed in a soluble form in the supernatant at 15°C (Fig. 2A), the condition of 1 mM IPTG and 15°C was selected as the final condition throughout the study. The Per a 9 protein was purified by Ni column. More than 8.69 mg recombinant Per a 9 was obtained from 1 liter of cell culture. The purity of the purified Per a 9 was identified by SDS-PAGE. It showed a single band with an apparent molecular weight of 40 kDa (Fig. 2B).

In order to determine the allergenicity of Per a 9, we examined the ability of Per a 9 to bind IgE in the sera of the patients with American CR allergies using a direct ELISA technique. The sera from all patients apart from patients 4, 6 and 12 exhibited a positive IgE reactivity to Per a 9. The results revealed that 13/16 (81.3%) sera from these patients reacted to Per a 9 (Fig. 3A). The IgE binding activity of Per a 9 in a representative group of 3 patients and 2 healthy controls were assessed by western blot analysis and the results are illustrated in Fig. 3B.

Per a 9 at 1.0 μg/ml induced approximately up to a 4.2-fold increase in the number of CD63 and CCR3 double-positive cells when incubating with passively sensitized basophils from the sera of patients with American CR allergies. Per a 9 had no effect on the basophils sensitized by the sera from HC (Fig. 4).

Uniprot and tBLASTn were used to search the homologous sequences of Per a 9. As a result, 47 sequences were obtained. In order to determine the relationships between Per a 9 and its homologous sequences, phylogenetic analysis was performed and the evolutionary tree inferred by the ML method is presented in Fig. 5.

The primary structure of Per a 9 contained 356 amino acids and the molecular weight was 39,735. The theoretical pI was 5.58 and the aliphatic index was 80.56. The GRAVY was -0.413, which incidated that Per a 9 exhibited a hydrophilic character. The instability index was 31.36 (<40), indicating that the Per a 9 protein was stable. Per a 9 is an arginine kinase and the results from PROSITE showed that Per a 9 had two obvious characteristic patterns of phosphagen kinase, including PS51509 (phosphagen kinase N-terminal domain profile) and PS51510 (phosphagen kinase C-terminal domain profile). Moreover, PS00112 was contained in PS51510 and consisted of CPTNLGT, which was the active center of phosphagen kinase. The ATP-binding sites were His (185) and Arg (229).

Secondary structure prediction with PSIPRED identified 11 α-helices and 8 β-sheets in Per a 9. PredictProtein predicted 10 α-helices and 9 β-sheets. Alternatively, NetSurfP v1.1 predicted 11 α-helices and 9 β-sheets. The predicated secondary structure of Per a 9 is presented in Table I.

The search for the proteins of Per a 9 with known tertiary structure in the PDB yielded Litopenaeus vannamei arginine kinase (PDB Accession no. 4BG4) showing the highest sequence identity (83% with Per a 9). As a result, the 4BG4 template was used for homology modeling. The homology model that matched the aforementioned structures is shown in Fig. 6A. As indicated by the Ramachandran plot (Fig. 6B), 89.4% of the residues in the model structure were within the most favored regions, 10% of the residues were in the additional allowed region, 0.6% of the residues were in the generously allowed regions and 0% of the residues were in the disallowed region. As indicated by the ERRAT program, the results (Fig. 6C) revealed that the overall quality factor was 100, which indicated that the structure had a high resolution. As indicated by the VERIFY_3D program, the results revealed that 99.72% of the residues had an average 3D (atomic model)-1D (amino acid sequence) score of >0.2, which also indicated that the structure was good. Based on these validations, it is shown that the homology model was adopted for this study.

Surface accessibility and fragment flexibility are important features for predicting antigenic epitopes. In addition, the existence of regions with high hydrophobicity also provides strong evidence of epitope identification. The antigenic index directly revealed the epitope forming capacity of the Per a 9 sequence. Based on these sequence properties, the final predicting regions of Per a 9 by DNASTAR were obtained as: 23–31, 38–46, 57–64, 70–72, 91–117, 143–154, 159–165, 174–183, 189–19, 222–227, 292–302, 309–32 and 337–345. The predicted results of the BPAP system were 19–20, 39–46, 58–74, 90–118, 131–136, 145–154, 157–165, 175–176, 275–278, 290–292, 295–299, 309–325 and 338–344. The predicted results of the BepiPred 1.0 server were 11–19, 21–28, 46–55, 115–127, 134–142, 152–158, 176–187, 193–203, 216–222, 238–251, 264–273, 277–293 and 296–308. Furthermore, the final potential B cell epitopes of Per a 9 were selected on the basis of the results of these 3 tools. The ultimate results of the 3 immunoinformatics tools finally predicted 11 peptides (23–28, 39–46, 58–64, 91–118, 131–136, 145–154, 159–165, 176–183, 290–299, 309–320 and 338–344) and these peptides are shown in Fig. 7.

For the HLA-DR-based T cell epitope prediction of Per a 9, the final predicting regions of HLA-DR 101, HLA-DR 301, HLA-DR 401 and HLA-DR 501 are shown in Table II and the ultimate results of HLA-DR-based T cell epitope prediction finally predicted 5 peptides (119–127, 194–202, 210–218, 239–250 and 279–290). For HLA-DQ alleles, the final results of HLA-DQA10101-DQB10501, HLA-DQA10301-DQB10302, HLA-DQA10401-DQB10402, and HLA-DQA10102-DQB10602 are also shown in Table II and the ultimate results of these 4 methods finally predicted 2 peptides, 46–54 and 65–80. As a result, Per a 9 was predicted to have 7 T cell epitope sequences, 46–54, 65–80, 119–127, 194–202, 210–218, 239–250 and 279–290 as shown in Fig. 7.