Distinct epitope structures of defensin‐like proteins linked to proline‐rich regions give rise to differences in their allergenic activity

Abstract Background Art v 1, Amb a 4, and Par h 1 are allergenic defensin‐polyproline–linked proteins present in mugwort, ragweed, and feverfew pollen, respectively. We aimed to investigate the physicochemical and immunological features underlying the different allergenic capacities of those allergens. Methods Recombinant defensin‐polyproline–linked proteins were expressed in E. coli and physicochemically characterized in detail regarding identity, secondary structure, and aggregation status. Allergenic activity was assessed by mediator releases assay, serum IgE reactivity, and IgE inhibition ELISA using sera of patients from Austria, Canada, and Korea. Endolysosomal protein degradation and T‐cell cross‐reactivity were studied in vitro. Results Despite variations in the proline‐rich region, similar secondary structure elements were observed in the defensin‐like domains. Seventy‐four percent and 52% of the Austrian and Canadian patients reacted to all three allergens, while Korean patients were almost exclusively sensitized to Art v 1. This was reflected by IgE inhibition assays demonstrating high cross‐reactivity for Austrian, medium for Canadian, and low for Korean sera. In a subgroup of patients, IgE reactivity toward structurally altered Amb a 4 and Par h 1 was not changed suggesting involvement of linear epitopes. Immunologically relevant endolysosomal stability of the defensin‐like domain was limited to Art v 1 and no T‐cell cross‐reactivity with Art v 125‐36 was observed. Conclusions Despite structural similarity, different IgE‐binding profiles and proteolytic processing impacted the allergenic capacity of defensin‐polyproline–linked molecules. Based on the fact that Amb a 4 demonstrated distinct IgE‐binding epitopes, we suggest inclusion in molecule‐based allergy diagnosis.


| INTRODUCTION
Allergy to weed pollen has been extensively documented worldwide, being the third most important cause of pollen allergies, after grasses and trees. Relevant allergenic weeds of the Asteraceae family are mugwort (Artemisia vulgaris), ragweed (Ambrosia artemisiifolia), Santa Maria feverfew (Parthenium hysterophorus), and sunflower (Helianthus annuus). 1 Mugwort and ragweed are preferentially growing in the temperate climate zone of the Northern Hemisphere and Australia, while feverfew grows in the subtropical areas of America and Asia. In Europe, exposure to mugwort pollen is more abundant in central and northern regions, ragweed in south and eastern areas, while feverfew is not endemic (www.pollenwarndienst.at). In northern America, ragweed represents the most prevalent Asteraceae pollen source, while in Korea exposure to mugwort and ragweed is observed. 2,3 Atypical distribution and intensified flowering seasons of weeds are anticipated which potentially lead to more allergic reactions as a consequence of climatic changes. [4][5][6] Extract-based allergy diagnosis of weed pollen is frequently challenging due to (i) polysensitization of patients, (ii) overlapping flowering periods, and (iii) similar allergen profiles. 7,8 Although molecule-based approaches can be exploited for diagnosis, solely Amb a 1, Art v 1, and Art v 3 are currently available for routine allergy diagnosis of Asteraceae pollen allergy. 2 Several studies showed that sensitization profiles to distinct allergens from the same weed are highly diverse in different geographic regions. 4,7,9,10 Even molecules from the same protein family like lipid transfer proteins (LTP) or pectate lyases can show different sensitization potencies, and in this context, allergenic defensin-like proteins linked to a polyproline-rich region are interesting molecules. Those defensinpolyproline-linked proteins are exclusively found in pollen of the Asteraceae family, while defensin-like proteins alone are prevalent in higher plants and were recently also described as allergens in legumes. 1,11,12 Art v 1 is a major allergen with a sensitization rate of 95% among mugwort pollen-allergic patients. 13 Amb a 4, the homolog in ragweed, is considered a minor allergen with 20%-40% sensitization rate among ragweed-allergic patients. 14 Recently, the complete sequence of Par h 1 from feverfew was identified and a sensitization rate of 60% and 40% among Austrian and Indian pollen-allergic patients was reported. 15 A common feature of this allergen family is O-linked glycosylation of hydroxyproline residues. 13,14,16,17 Despite the fact that low levels of IgE antibodies against glycan moieties of Art v 1 were detected, they did not convey mediator release and thus demonstrated low clinical significance. 18 If O-linked glycans contribute to unspecific cross-reactivity in diagnosis remains to be determined. While structural and immunological features of Art v 1 have been extensively investigated, 13,16,19,20 only limited information is available for Amb a 4 and Par h 1. 14,15 In the present study, we investigated structural and immunological features of allergenic defensin-polyproline-linked molecules from mugwort, ragweed, and feverfew pollen. Therefore, recombinant allergens were produced to evaluate intrinsic physicochemical characteristics. IgE sensitization profiles and cross-reactivity pattern were investigated in patients from three geographically distinct regions.
Endolysosomal degradation assays were performed to monitor proteolytic susceptibility, and T-cell assays were conducted to analyze cross-reactivity at the T-cell level.

| METHODS
Detailed description of all experimental procedures is provided in the Data S1.

| Patients and sera
Weed pollen-allergic patients from Austria (n = 36), Canada (n = 38), and Korea (n = 24) were selected on the basis of typical case history, that is, rhinitis and/or conjunctivitis during late summer, and allergen-specific IgE to mugwort and/or ragweed pollen determined by ImmunoCAP or skin prick test (Table S1) 13,15 Protein expression of nontagged proteins was performed in E. coli Rosetta-gamiB (DE3) pLysS as described. 15 Briefly, Art v 1 was obtained by ultrafiltration followed by cation exchange and size exclusion chromatography. Amb a 4 and Par h 1 were purified using ammonium sulfate precipitation followed by hydrophobic interaction and size exclusion chromatography.

| Physicochemical characterization
Purified allergens were analyzed by reducing SDS-PAGE and Coomassie staining. Protein concentrations were determined by amino acid analysis, and identity was verified by intact mass analyses using a Q-Exactive mass spectrometer. The aggregation behavior of the proteins was analyzed by dynamic light scattering (DLS) and highperformance liquid chromatography (HPLC). Secondary structure elements were determined by Fourier transform infrared (FTIR) spectroscopy using a Tensor 27 spectrophotometer. Circular dichroism spectra of native and heat-treated proteins were recorded with a JASCO J-815 spectropolarimeter.

| IgE reactivity, cross-inhibition, and mediator release assay
IgE reactivity of weed pollen extracts and purified allergens was evaluated by ELISA using sera from weed pollen-allergic patients. The biological activity of the allergens was verified in a mediator release assay. 21 Detection of IgE reactivity to purified allergens was performed using a colorimetric ELISA. IgE reactivity to reduced/alkylated (R/A) allergens was assessed using a chemiluminescence ELISA. IgE cross-reactivity between purified allergens was studied by cross-inhibition ELISA. Sera were pre-incubated overnight with the inhibitor molecules or buffer.

| Proteolytic stability
The proteolytic stability was studied with the endolysosomal degradation assays. 22 Recombinant allergens were incubated with the microsomal fraction of a murine dendritic cell (DC) line or purified recombinant cathepsin S. Samples were analyzed at different time points using gel electrophoresis and mass spectrometry.

| T-cell reactivity
Monocyte-derived dendritic cells were differentiated as described 23,24 and incubated with the allergens or Art v 1-peptide.
Art v 1 25-36 -specific TCR tg Jurkat T-cells 25,26 were added and cocultured and IL-2 promoter-driven luciferase activity was determined. 25 Art v 1 25-36 -specific T-cell lines and clones established from mugwort-allergic donors were tested for proliferation in response to the allergens or Art v 1-peptide by 3 H-thymidine uptake. 20

| Data processing and statistical analyses
An unsupervised cluster analysis was performed with ClustVis using IgE reactivity and cross-inhibition data. 27 Figure 1B, Fig. S1). The proline-rich region could not be modeled due to low sequence similarity and high flexibility. 19 All three allergens were purified to homogeneity with purities of >98% (Fig. S2A). In line with previous results, they showed an unusual electrophoretic behavior. [13][14][15] The protein identity and formation of four disulfide bonds were unequivocally verified by mass spectrometry ( Fig. S2B). Protein preparations were monomeric with a R h 2 nm as assessed by DLS, and in HPLC, they eluted as single peaks with comparable retention times and molecular weight (Fig. S2C,D).
FTIR analysis revealed the presence of defined secondary structure elements. Consistent with the structure of Art v 1, the proteins showed a higher content of beta sheets than alpha helices, and the PABLOS ET AL.
| 433 amount of disordered structures positively correlated with the length of the proline-rich region ( Figure 1C and Fig. S3A). In CD analyses, almost identical spectra were recorded, indicating a similar folding consistent with proline-rich proteins ( Figure 1D). 13,15,21 The defensin-polyproline-linked proteins exhibited comparable stability to thermal denaturation, and structural changes observed at 222 nm were analogous. However, denaturation spectra of Amb a 4 and Par h 1 shared higher similarity compared to Art v 1 (Fig. S3B).

| Weed pollen-allergic patients from different geographic regions demonstrate distinct sensitization profiles
To assess the immunological relevance, we investigated sera of weed pollen-allergic patients from Austria, Canada, and Korea. First, we evaluated the patients' sensitization profile to pollen extracts of mugwort, ragweed, and feverfew in ELISA (Figure 2A,B). The majority of Austrian patients were multisensitized to all three sources, with ragweed presenting strongest IgE binding followed by mugwort pollen. While 53% of the Canadian patients reacted to ragweed as well as mugwort, a strong bias toward ragweed sensitization was noted which was also reflected by significantly higher IgE levels. The Korean patients showed a dominant mugwort sensitization profile with 58% mono-sensitized patients.
The allergenic activity of all recombinant allergen preparations was verified in a mediator release assay (Fig. S4), and IgE reactivity of individual sera was monitored in ELISA ( Figure 2C,D, Table S1).
Sensitization to any of the tested molecules was detected in 19 Figure 5B). Proteolytic peptides identified by mass spectrometry supported these findings (Fig. S9). Similar degradation patterns were found within the conserved defensin-like domain although kinetics varied considerably ( Figure 6). Typically, defensin-like domains were degraded from both ends to the inner core and generated peptides were spanning the immunodominant T-cell epitope of Art v 1. 20,26 Analogous degradation kinetics and cleavage sites were observed using cathepsin S (Fig. S10).
These results prompted us to investigate whether the homologous allergens were cross-reactive at the T-cell level. Therefore, we While Art v 1 is considered the major allergen of mugwort pollen, 13 homologs from ragweed and feverfew are considered minor allergens. 14,15 It is believed that distinct IgE sensitization profiles may arise from varying exposure to weeds in different geographic regions. 7,9 We further investigated whether intrinsic molecular characteristics of defensin-polyproline-linked allergens could also account for distinct sensitization profiles.
Recombinant allergens showed typical structural features of the defensin-like protein family while amino acid sequences vastly differed in the proline-rich region. [13][14][15] Structural modeling showed that IgE epitopes of Art v 1 are partially conserved in Amb a 4 and Par h 1. The proline-rich region did not allow reliable prediction due to low sequence similarity and high structural flexibility. 19,28 In FTIR, comparable distributions of secondary structure elements were found which reflected previous NMR data obtained for Art v 1 in solution. 19 Overall, proteins showed similar spectra in CD measurements and neither and length. [28][29][30] Next, the sensitization profile to weed pollen was investigated in patients from Austria, Canada, and Korea. Patients from Austria were sensitized to ragweed and mugwort due to the con- Defensin-polyproline-linked allergens show variable IgE cross-reactivity. Art v 1, Amb a 4, and Par h 1 were immobilized onto ELISA plates and patients' sera were pre-incubated with the inhibiting molecules. Graphs show mean AE SEM, ***P ≤ .001, **P ≤ .01, *P ≤ .05 and ns, not significant patients were exclusively reactive to ragweed extract. The Canadian patients showed a prevalent sensitization to ragweed, 7,31 while 55% were also reactive to mugwort due to either IgE cosensitization or cross-reactivity. 32,33 In contrast, Korean patients presented genuine sensitization to mugwort, which corresponded to the dominant Asteraceae weed sensitization in this region. 34 IgE sensitization profiles to feverfew pollen were recorded in order to evaluate cross-reactivity in Austrian and Canadian patients typically not exposed to this weed. Growth of feverfew has been reported in Southern Korea 4 (http://www.cabi. org/isc/datasheet/45573); however, its implication in allergy has not been established.
Using recombinant allergens, a dominant IgE reactivity to Art v 1 (92%) among Korean patients was noted. Art v 1 sensitization prevalence in Canada and Austria was only 55% and 53%, respectively, suggesting that those patients were also reactive to Amb a 1. 7,31 Notably, Art v 1 and Par h 1 represent the most prevalent molecules in the pollen extract, while Amb a 4 levels are outperformed by the major allergen Amb a 1 (unpublished data). Among Austrian patients sensitized to defensin-polyproline-linked proteins, 73% reacted to all three allergens in a similar manner with slightly higher IgE reactivity to Art v 1. The sensitization prevalence of 42% to Amb a 4 was in agreement with 39% as previously described. 14

| 437
Art v 1  6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29   patients. In all patients, IgE reactivity to Par h 1 was virtually the same as for Amb a 4, also reflected by highly correlating antibody levels. Even though Par h 1 sensitization has to be considered a cross-reactivity phenomenon in the investigated context, this might have implications for patients sensitized to Art v 1 or Amb a 4.

A G S K L C E K T S K T Y S G K C D N K K C D K K C I E W E K A Q H G A C H K R E A G K E S C F C Y F D C S K S P P G A
To study reactivity pattern in detail, we investigated whether reduction and alkylation lead to loss of IgE reactivity. In line with previous studies, the appropriate folding of the Art v 1 defensin-like domain was crucial for IgE reactivity. 13 Art v 1, and in fact, the immunodominant T-cell epitope is located in this region. 20,39 In contrast, premature degradation and thus suboptimal peptide loads resulting in poor presentation on MHC-II might be a reason for a lower immunogenicity of Amb a 4 and potentially also Par h 1. 14, 37,40 As differences in the primary structure might provide distinct cleavage sites for different proteases, we additionally investigated the proteolytic stability to cathepsin S, a major component of the endolysosomal protease cocktail. 22,41,42 Similar degradation kinetics was observed as compared with the endolysosomal extract and, in general, cathepsin S cleavage sites were conserved (Fig. S11). Potentially, other proteases of the endolysosomal fraction, varying intrinsic stability, or resistance to redox environment of proteins might account for minor differences. [43][44][45] Proteolytic peptides typically harbored the immunodominant T-cell epitope previously determined for Art v 1, 20 which prompted us to investigate T-cell cross-reactivity.
Despite high sequence similarity in this region, we observed no T-cell cross-reactivity using Art v 1-specific T-cell assays. Results are largely in agreement with Jahn-Schmid et al 26 who described that the crucial minimal Art v 1 T-cell epitope (EWEKA) was restricted to HLA-DR1. It is important to note that our experimental settings focused on the immunodominant epitope Art v 1 [25][26][27][28][29][30][31][32][33][34][35][36] , while T-cell cross-reactivity between other, minor epitopes and HLA class II restrictions other than those by HLA-DR1 was not investigated. 20,26,39 However, this finding also implicates that Art v 1 alone might not be efficient when considering molecule-based immunotherapy approaches.
In summary, we show that regardless of the structural similarity shared by the allergenic defensin-polyproline-linked proteins, they differ in their immunological properties. The higher sensitization rate of Art v 1 is consistent with the increased stability to proteolytic degradation, which ensures a steady supply of T-cell epitopes. A group of patients reacted exclusively to Amb a 4/Par h 1 epitopes which are not dependent on the cysteine-stabilized fold of the defensin-like domain. We therefore suggest including Amb a 4 in molecule-based diagnosis of weed pollen allergies, thus allowing further refinement for therapeutics in allergen immunotherapy.