1. IntroductionNot every protein is able to become under natural conditions an allergen. As such, allergens are clustered in protein families with the major allergens from mammalian systems always belonging to the “lipocalin” superfamily [1] and major allergens from plant origins usually belonging to either the pathogenesis-related proteins PR10 family or to the prolamin superfamily which comprises major allergens from the seed storage protein families being 2S albumins as well as the non-specific lipid transfer proteins nsLTPs [2,3]. Major allergens from plant origin also belong to the Gibberellin-regulated proteins GRPs [4], cupin protein superfamily (legumins-7S and vicilins-11S protein) and Ole e 1 families [5]. In 2012, our group made the serendipity discovery that the major birch pollen allergen Bet v 1 shares a lipocalin-like structure with the endogenous human lipocalin 2 LCN2 [6], which linked one of the largest major plant allergen family with a similar large major mammalian allergen family and with human LCN2. Indeed, LCN2 is decreased in allergic subjects [7] and as innate defense protein is usually present at mucosal surfaces and lymphoid sites. It is involved in numerous iron-dependent and immunoregulatory processes, where it can bind iron, but only when the iron is complexed by catechol-based siderophores, since lipocalins usually have no measurable affinity for iron alone [8]. LCN2′s ability to bind to iron makes it an antioxidant as it is able to prevent oxidative stress and prevent ferroptosis [9]. The iron-containing form of LCN2 (holoLCN2) is also present and released by anti-inflammatory macrophages [10] and contributes to tissue healing and recovery upon injury. In contrast, inflammatory macrophages release iron-free apo-LCN2 in response to invading bacteria [10].In plants, immunity can be triggered via different pathways. One is the recognition via the recognition of pathogen-associated molecular patterns that leads to the induction of antimicrobial proteins, secondary metabolites, reactive oxygen species ROS formation and cell wall reinforcements. The second is the so-called effector-triggered immunity ETI, which can be initiated in plants by the perception of pathogen activity and usually leads to local program cell death named “hypersensitive response” (as a hallmark for systemic acquired resistance [11]). This ETI is triggered by salicylic acid accumulation to limit the pathogen spread [12]. Interestingly, iron depletion in plants is sufficient to prime plant immunity [13] leading to salicylic acid accumulation [13,14,15] and increasing coumarin and flavonoid synthesis [16,17,18,19] as well as the transcription of pathogenesis-related PR genes [15,20,21,22]. There seems to be a direct link between plant iron homeostasis and PR gene expression as iron treatment downregulates in plants the expression of PR10 proteins [23].Interestingly, similar to mammalian lipocalins binding to iron via catechol-based siderophores, in 2014 the natural flavonoid ligand of Bet v 1 has been identified as quercetin-3-O-sophoroside [24], which likewise contains a catechol-moiety [25] with a very high affinity to iron [26]. As such, flavonoids seem to be the plant counterpart to bacterial-derived siderophores, with both considered secondary metabolites in the respected plants and bacteria/fungi and having both strong anti-oxidative and anti-inflammatory features, which are linked with their ability to complex iron.

In this respect, we hypothesized that similar to the mammalian lipocalin family, also in plants the natural function of the PR10 protein Bet v 1 is associated with plant iron homeostasis and immunity. We hypothesized that the allergenic or tolerogenic feature of Bet v 1 is linked to its iron-scavenging ability that it can exert cross-species, thereby enabling the modulation of the immune response in human immune cells with the iron-laden versus the iron-free form affecting the allergenicity of Bet v 1.

2. Materials and Methods 2.1. Structural and Docking AnalysisAtom coordinates of Bet v 1 were taken from the high-resolution (1.24 Å) crystal structure of its complex with naringenin (protein data bank (PDB) entry 4A87) [27]. The geometries of quercetin were obtained upon energy minimization with the MM2 force field of initial structures drawn using the ChemBioDraw/ChemBio3D Ultra 12.0 package. Docking input files for protein and ligands were prepared with reduce v3.23 [28], the ADFR software suite (https://ccsb.scripps.edu/adfr; accessed on 12 August 2021) and AutoDockTools [29]. Docking calculations were performed with AutoDock Vina [30,31]. The docking solution with the lowest affinity energy Eaff was selected. Estimates of dissociation equilibrium constants Kd were then calculated for the protein-ligand complexes by assuming Eaff~ΔG with Kd = exp(−ΔG/RT) at T = 298.15. Protein structural visualizations were prepared with UCSF Chimera [32,33]. 2.2. Recombinant Bet v 1Bet v 1 was produced as described [34]. In short, a codon-optimized synthetic gene of Bet v 1.0101 was obtained from Eurofins MWG Operon (Ebersberg, Germany) and cloned into the expression vector pET-28a(+) (Merck Millipore, Darmstadt, Germany). The proteins were expressed in E. coli BL21[DE3] in LB medium at 37 °C after induction with 1 mM isopropyl--D-thiogalactopyranoside. Bet v 1 was purified by a combination of hydrophobic interaction and ion exchange chromatography. SDS-PAGE, MALDI-TOF MS (Bruker Ultraflex II, Bruker Daltonics, Billerica, MA, USA), and circular dichroism (CD) spectroscopy was used to verify protein purity and identity, mass, and secondary structure. Measurement of endotoxin content was done by Hyglos EndozymeII Kit (Bernried am Starnberger See, GER) and total protein content by BCA assay according to the manufacturer’s instructions (Pierce BCA Protein Assay Kit, Thermo Scientific, Rockford, IL, USA). 2.3. Generation of Apo- and Holo Bet v 1

Bet v 1 (1.1–1.89 mg/mL) was dialyzed three times against 10 µM deferoxamine mesylate salt (Sigma D9533, St. Louis, MO, USA) following dialyzation against deionized water to generate apoBet v 1. HoloBet v 1 was generated by adding pre-formed iron-quercetin complexes (FeQ2). 3 mM FeQ2 complexes were generated by dissolving quercetin (Sigma 1592409) in 1 M NaOH and adding acidic iron (Iron-standard AAS, Sigma 16596) at a ratio 2:1, before adding them to apoBet v 1 to a final concentration of, e.g., 10 µM Bet v 1, 20 µM quercetin and 10 µM iron or dilutions thereof.

2.4. Spectral AnalysisFor spectral analysis, a physiological saline solution (0.89% NaCl) was used as a buffer to minimize iron contamination from the air. The pH was kept constant at pH 7 for Figure 1c,d and was >7.3 for Figure 1e–g. Optical density was measured using (1) 200 µM quercetin (=100 µM Q2) with increasing concentrations of ferric iron (40–200 µM), (2) 100 µM FeQ2 with increasing concentrations of Bet v 1 (2–20 µM) or (3) 100 µM FeQ2 with increasing concentrations of gel-filtrated apoBet v 1 or holoBet v 1 (0.5–8 µM). All measurements were repeated at least three times with similar results. 2.5. Anti-Oxidative AssayAnti-oxidative status was measured as described before [35,36]. Shortly, a 7 mM ABTS radical stock solution was prepared by diluting 19 mg of 2,2′-Azinobis-(3-ethylbenzothiazoline-6-sulfonic acid) (Sigma A1888) (ABTS+.) and 20 mg ammonium persulfate (Sigma A3678) in 5 mL distilled water at room temperature. The stock solution was further diluted 1:4 in distilled water to obtain an ABTS working solution. As standard either Trolox (Sigma 238813, Austria, St. Louis, MO, USA), quercetin Q2 or FeQ2 was used. Samples of apoBet v 1 (0.5 mg/mL) and/or preincubated with FeQ2, Q2 or ferroxamine (FO) were subjected to gel filtration (PD MiniTrap G-25 Columns, GE Healthcare 28-9180-07) to remove unbound ligands. 50 µL of diluted samples or diluted standards (Trolox, Q2, FeQ2) were incubated with 50 µL of ABTS working solution for 2 min in a 96 well plate before measurement of absorbance at 740 nm using an Infinite M200Pro microplate reader (Tecan, Grödig, Austria). Data are shown as antioxidative capacity using Trolox-equivalent normalized to apoBet v 1 in Figure 1f,g using FeQ2 and Q2 as standard to calculate the molar ratio of quercetin to Bet v 1. 2.6. Bet v1-Specific SeraSerum samples derived from two clinical trials (NCT0381598, NCT03816800), which were approved by the ethics committee (#1972/2017 and #1370/2018) of the Medical University of Vienna. Patients’ characteristics have been described previously in detail [37,38]. Briefly, sera of subjects with allergic rhinitis, but were otherwise healthy, were used with Bet v 1-specificity of patients to ascertain by testing for specific IgE by ALEX®-tests (Macroarray Diagnostics, Wien, Austria) and ALEX®-test specific IgE-level against Bet v1 being at least 1 kUA/L. The concentration of individual sera and in the pools for Bet v 1-specific IgE ranged from 10 to 60 kUA/L. 2.7. Bet v1-Specific IgE ELISA

Two µg/mL of apoBet v 1, holoBet v1-FeQ2, Bet v 1-Q2 or Bet v 1-FO (ferroxamine) diluted in 0.89% NaCl were coated (100 µL per well) overnight at 4 °C, blocked for 2 h at room temperature with 1% BSA in 0.89% NaCl containing 0.05% Tween 20 (200 µL per well), before incubated with 100 µL of diluted serum from birch-pollen allergic and non-allergic subjects (1:10 diluted in 0.89% NaCl/0.05% Tween-20) overnight at 4 °C. Bound IgE was detected with horseradish–peroxidase-conjugated goat anti-human IgE antibody (Invitrogen A18793, Waltham, MA, USA) diluted at 1:4000 in 0.89% NaCl containing 0.05% Tween-20 and using tetramethylbenzidine (eBioscience, San Diego, CA, USA) as a substrate. Color development was stopped with 1.8 M sulfuric acid. The optical density was measured at 405 nm by using an Infinite M200Pro microplate reader (Tecan, Grödig, Austria). Between the steps, rigorous washing was performed with 0.89% NaCl containing 0.05% Tween-20. Data are shown as a percentage normalized to apoBet v 1 IgE binding levels or as OD at 450 nm.

2.8. Bet v 1-Specific IgE Inhibition ELISA

Inhibition experiments were carried out similarly as described for the Bet v 1-specific IgE ELISA. Briefly, plates were coated with 2 µg/mL of apoBet v 1 or holoBet v 1-FeQ2 (100 µL per well) overnight at 4 °C. In the meantime, pooled serum from birch-pollen allergic subjects (diluted 1:10 in 0.89% NaCl) was preincubated with increasing doses of apoBet v 1 (0–1000 ng/mL) or holoBet v 1-FeQ2 (0–1000 ng/mL) also overnight at 4 °C. After washing and blocking, preincubated serum samples were applied (100 µL per well) overnight at 4 °C. Detection was performed by using horseradish–peroxidase-conjugated goat anti-human IgE antibody (Invitrogen A18793) diluted at 1:4000 in 0.89% NaCl containing 0.05% Tween-20, with use of tetramethylbenzidine (eBioscience) as a substrate and 1.8 M sulfuric acid to stop color development. The optical density was measured at 450 nm with the Infinite M200Pro microplate reader (Tecan, Grödig, Austria). Data are shown as OD at 450 nm or as kU/L IgE binding to plate-bound apoBet v 1.

2.9. SDS-PAGE and Western Blot

ApoBet v 1 and holoBet v 1-FeQ2 were mixed with 4x non-reducing sample buffer and 12 µg/slot was applied on two sodium dodecyl sulfate polyacrylamide (SDS–PAGE) 4–20% gradient gels (Biorad 4568095, Tokyo, Japan), run at 90 V for about 1 h and subsequently one gel was stained with Roti-Blue quick solution (Roth 4829.2, Roth, Germany) and the other gel underwent immunoblotting. Briefly, the gel was blotted on a methanol-activated PVDF-membrane (Immobilon IPVH00010, 0.45 µm), blocked with 1% BSA in tris buffered saline containing 0.05% Tween-20 (TBS-T) for 18 h, before incubating with 1:10 diluted serum pool of allergic donors (n = 17) for 18 h at 4 °C. Bound IgE was detected with horseradish peroxidase labeled anti-human IgE antibody (Invitrogen A18793) and using ECL substrate (clarity TM Western ECL Substrate, BioRad 170-5061) for detection. Between each step, rigorous washing was performed with TBS-T. Luminescence was imaged with the ChemiDoc™Touch Imaging System (BioRad).

2.10. Human Mast Cell Degranulation AssayHuman peripheral blood mononuclear cell-derived mast cells were generated as previously described by Folkerts et al. [39]. Briefly, peripheral blood mononuclear cells were obtained from buffy coats of healthy blood donors and CD34+ precursor cells were isolated using the EasySep Human CD34 Positive Selection Kit (STEMCELL Technologies). CD34+ cells were maintained for 4 weeks under serum-free conditions using StemSpan medium (STEMCELL Technologies) supplemented with recombinant human IL-6 (50 ng/mL; Peprotech), human IL-3 (10 ng/mL; Peprotech) and human Stem Cell Factor (100 ng/mL Peprotech, Rocky Hill, NJ, USA). After 4 weeks, the cells were cultured in Iscove’s modified Dulbecco’s medium/0.5% bovine serum albumin with human IL-6 (50 ng/mL, Peprotech, Rocky Hill, NJ, USA) and 3% supernatant of Chinese hamster ovary transfectants secreting murine stem cell factor (a gift from Dr. P. Dubreuil, Marseille, France). The mature MCs were identified by flow cytometry based on positive staining for CD117 (eBioscience) and FcεRIa (eBioscience) using BD FACS Canto II (approximately 90%). Degranulation assay was performed by incubation of human mast cells with serum pools of birch pollen allergic or non-allergic subjects followed by incubation with 5 nM of apoBet v 1, holoBet v1, Q2, FeQ2 or buffer, respectively. Degranulation was assessed by measurement of the activity of released ß-hexominidase in the supernatant and unreleased enzyme in the respective cell lysate. The presented results were calculated as percentage released versus total ß-hexominidase activity, with a release from unstimulated controls being 0.003%, from positive controls with anti-human IgE being 33.4% and with ionomycin being 84.6%. 2.11. AZ-AhR Reporter AssayAZ-AhR assay was done as previously described [40]. Briefly, AZ-AhR cells were plated on 96-well plates at a density of 2 × 104 cells/well for 18 h. Subsequently, cells were stimulated for 18 h in triplicates with 45 µM of iron-quercetin complexes alone or in addition with 2 µM or 10 µM Bet v 1. The positive control cells were treated with 20 nM indirubin. Cells were washed once with 0.89% NaCl, before the lysis buffer of the luciferase assay kit (Promega E4530, Tokyo, Japan) was added. After a single freeze–thaw cycle, 20 μL/well of lysates were transferred into a black 96-well flat-bottom plate (Thermo Scientific) and bioluminescent reaction were started with addition of 100 μL/well of luciferase assay reagent (Promega). Chemiluminescence was measured (10 s/well) using a spectrophotometer (Tecan InfiniteM200 PRO). Data from six independent experiments are shown as normalized to medium levels. 2.12. Flow Cytometric Analysis of Human PBMCsPBMCs were isolated by Ficoll-Paque (GE Healthcare) and washed with 0.9% NaCl before incubation with 5 µM iron-quercetin (FeQ2) alone and/or in combination with 5 µM apoBet v1 in media containing neither phenol red nor FCS for 18 h as previously described [41]. Subsequently, cells were stained with combinations of calcein-AM (Thermo-Fisher, Waltham, MA, USA), CD3-APC-Cy7 (eBioscience, clone SK7), CD14-APC-Cy7 (Biolegend, clone M5EZ), HLADR-PE (Biolegend, San Diego, Calif, clone L243PC), and CD86-PE-CY7 (Biolegend, clone IT2.2) for flow cytometric analysis. Calcein-AM violet was used as a living marker and to determine the labile iron load in living cells. Doublets were excluded before gating on CD3+ cells in the lymphocyte population or on CD14+ cells in the monocytic population. In monocytic cells this was followed by consecutive gating for HLADR+, CD86+ and calcein+ and geometric mean fluorescence intensity (MFI) was calculated for each fluorochrome. Acquisition and analyses were performed on a FACS Canto II machine (BD Bioscience, San Jose, CA, USA) using the FACSDiva Software 6.0 (BD Biosciences). 2.13. Statistical Analyses

Anti-oxidative assay and mast cell degranulation were compared with one-way ANOVA following Holm-Sidák’s multiple comparisons test, with a single pooled variance. Data from IgE ELISA were compared with RM one-way ANOVA with Geissner Greenhouse correction and Holm-Sidák’s multiple comparisons test and inhibition ELISA with two-way ANOVA using Holm-Sidák’s multiple comparisons test, with a single pooled variance. AhR activation and flow cytometry of PBMCs was analyzed by mixed-effect analysis or with RM one-way ANOVA including Geisser-Greenhouse correction followed by Tukey’s multiple comparisons test. All tests were considered significant when p < 0.05. Statistical analysis was performed with Graphpad prism 9.4.1 (San Diego, CA 92108, USA).

4. DiscussionThe present study supports our notion that the natural function of Bet v 1 is strongly linked with its ability to bind to complexes of iron, a redox-active element essential for most life [43]. The strong inherent affinity of particularly ferric iron to flavonoids/polyphenols is highlighted by the fact that flavonoid-iron complexes can self-assembly into huge polyphenol-metal networks [47,48], which is used in the industry to coat microorganisms, but also by the fact that solubilized iron-flavonoid complexes serve as an iron source for commensal and opportunistic microbial pathogens [42]. Importantly, in plants intracellular iron depletion is sufficient to lead to the expression of pathogenesis-related proteins [15,20,21,22], further emphasizing that the natural function of Bet v 1 is linked with plant defense and nutritional immunity (particularly of iron), similarly as the mammalian counterpart LCN2. Indeed, using the cow milk lipocalin protein beta-lactoglobulin, we could show in previous in vivo [40,49] and in clinical studies [38,50,51] that nutritional support to immune cells was essential to induce immune resilience, amelioration of symptoms and tolerance in allergic rhinitis patients.

Here, we show similarly, that the binding affinity to quercetin-iron by Bet v 1 is exceptionally high with calculated affinity being in the lower nM-range, which is about 1000× stronger than to quercetin alone. Moreover, we show that binding to iron-quercetin complexes was superior to quercetin alone in hampering epitope-recognition by specific IgE antibodies from birch pollen-allergic individuals. Importantly, this had also implications for the effector phase as crosslinking by holoBet v 1 was not as effective as with apoBet v 1 without any ligands, in which epitope masking did not hinder crosslinking. It also indicates that during allergic sensitization Bet v 1 was presented to the human immune system without any ligand, as otherwise the generation of IgE against these epitopes would have been impeded.

In our study, the impact of the iron-quercetin ligand on rendering Bet v 1 less allergenic was not restricted on IgE recognition. Indeed, here we give evidence that the carrier function of Bet v 1 is crucial to provide an anti-inflammatory and anti-oxidative stimulus to human immune cells. Both the transport of quercetin as well as iron was facilitated by Bet v 1 with quercetin able to activate the arylhydrocarbon receptor, but also to increase the labile iron pool in human monocytic cells, in which a large labile iron pool is a major characteristic of anti-inflammatory macrophages [43].Activation of the cytoplasmic promiscuous AhR, which interacts with a plethora of exogenous ligands including quercetin [52], is described to mediate anti-inflammatory stimuli [53] that promote regulatory T [54,55], but not Th2 [56] cell differentiation and, keep antigen-presenting cells such as dendritic cells or macrophages in an immature state [57].Iron seems to have a similar impact- when not present in a free form to generate reactive oxygen species- with an increase of the labile iron content in macrophages promoting an anti-inflammatory phenotype and iron depletion promoting an inflammatory phenotype [58,59]. Indeed, iron deficiency is associated in vitro [60,61,62] and in human clinical studies with Th2 inflammation [63,64,65,66,67,68,69]. Iron is also central to the human immune defense, which is highlighted by the fact that the macrophages are the central hub for iron handling, recycling and distribution [70]. Importantly, iron distribution and recycling stops when the macrophage changes due to immune activation to an inflammatory phenotype. As such, both too much and too low intracellular iron levels have been reported to cause oxidative damage in mitochondria and trigger the expression of inflammatory-related genes [71]. This also explains that in humans every chronic immune activation will lead over time to anemia of chronic inflammation [43]. Despite the phylogenetic distance between plants and human, the innate immune mechanisms are evoked by the same triggers, with not only humans being affected by a hypersensitive response towards allergens, but also plant mounting a hypersensitive response that leads to the expression of pathogenesis-related proteins in response to iron sequestration. Although this immune priming is a very desirable response to infections, it also turns otherwise harmless proteins into allergens.Our current understanding indicates that the major birch pollen allergen Bet v 1—as a prime example of pathogenesis-related proteins [72]—has the ability to sequester iron, similarly to the mammalian lipocalin family is able to bind iron, and that exactly this feature makes them potent allergens. When they do not carry ligands, they are able to deprive their surrounding of iron complexes, thereby actively leading to immune activation and inflammation [70]. In contrast, when they do carry iron, they indeed are able to augment the labile iron pool in macrophages and thereby promote an immature, tolerogenic and anti-inflammatory phenotype in macrophages [58].