Molecular allergen profiling in horses by microarray reveals Fag e 2 from buckwheat as a frequent sensitizer

Abstract Background Companion animals are also affected by IgE‐mediated allergies, but the eliciting molecules are largely unknown. We aimed at refining an allergen microarray to explore sensitization in horses and compare it to the human IgE reactivity profiles. Methods Custom‐designed allergen microarray was produced on the basis of the ImmunoCAP ISAC technology containing 131 allergens. Sera from 51 horses derived from Europe or Japan were tested for specific IgE reactivity. The included horse patients were diagnosed for eczema due to insect bite hypersensitivity, chronic coughing, recurrent airway obstruction and urticaria or were clinically asymptomatic. Results Horses showed individual IgE‐binding patterns irrespective of their health status, indicating sensitization. In contrast to European and Japanese human sensitization patterns, frequently recognized allergens were Aln g 1 from alder and Cyn d 1 from Bermuda grass, likely due to specific respiratory exposure around paddocks and near the ground. The most prevalent allergen for 72.5% of the tested horses (37/51) was the 2S‐albumin Fag e 2 from buckwheat, which recently gained importance not only in human but also in horse diet. Conclusion In line with the One Health concept, covering human health, animal health and environmental health, allergen microarrays provide novel information on the allergen sensitization patterns of the companion animals around us, which may form a basis for allergen‐specific preventive and therapeutic concepts.

chronic coughing, recurrent airway obstruction and urticaria or were clinically asymptomatic.
Results: Horses showed individual IgE-binding patterns irrespective of their health status, indicating sensitization. In contrast to European and Japanese human sensitization patterns, frequently recognized allergens were Aln g 1 from alder and Cyn d 1 from Bermuda grass, likely due to specific respiratory exposure around paddocks and near the ground. The most prevalent allergen for 72.5% of the tested horses (37/51) was the 2S-albumin Fag e 2 from buckwheat, which recently gained importance not only in human but also in horse diet.
Conclusion: In line with the One Health concept, covering human health, animal health and environmental health, allergen microarrays provide novel information on the allergen sensitization patterns of the companion animals around us, which may form a basis for allergen-specific preventive and therapeutic concepts.

| INTRODUCTION
Allergies also affect domestic animals, such as dogs, cats and horses. 1 The same allergen sources as in human allergy may be relevant for animals, such as pollen 2 or food allergens. 3 The importance of comparing human and animal allergic diseases and causative allergens has recently been acknowledged by the establishment of an interest group for Comparative and Veterinary Allergology in the European Academy of Allergy and Clinical Immunology (EAACI). In fact, comparative allergology ideally fits into the "One Health Concept," which relies on three main pillars: human health, animal health and environmental health. 4 In horses, allergen hypersensitivity 5 leads to cutaneous symptoms, that is eczema 6 or urticaria, 7 and/or to respiratory symptoms in form of chronic coughing or recurrent airway obstruction (RAO). 8,9 In equine RAO, the exposure to allergens from hay and straw dust and induced symptoms are associated with increased plasma histamine levels 10 ; however, at least in case of the fungus Aspergillus, in vitro IgE diagnosis and intradermal IgE levels do not support IgE-mediated mechanisms. 11 In general, involvement of IgEmediated mechanisms in RAO is still controversial.
The phenomenon of insect bite hypersensitivity (IBH) which especially occurs in Icelandic ponies and leads to "summer eczema" with severe pruritus, alopecia and crusting 12 is the best investigated equine atopic disease thus far. 13 IBH can also be associated with bronchial hyper-reactivity 14 and hence mirrors the human atopic syndrome characterized by cutaneous and respiratory symptoms. 15 Interestingly, environmental allergen sources such as house dust mites (HDM), moulds 16 and pollen 2,17 likewise elicit allergic symptoms in horses.
In addition, several food allergen sources have been described being potentially relevant for horses, including oats, wheat or corn. 18 A major preventive and therapeutic strategy is to avoid feeding potential allergens. Hence, identification of such allergens is absolutely crucial. 19 Diagnosis of allergic conditions and desensitization therapy in domestic animals is today performed with allergen extracts. 17,20 Intradermal challenge with extracts is a well-established test procedure in equine allergology. 21 However, some studies have yielded conflicting results, that is either better diagnostic accuracy than conventional serological tests 22 or weaker reproducibility. 23 In this context, sensitive assessment of horse serum for the presence of allergen-specific IgE may represent a valuable diagnostic option, all the more as blood can be collected on-site without sedation. 1,[24][25][26][27] In recent years, innovative methods for reliable allergy diagnosis have been developed. In contrast to crude allergen extracts, molecular allergy diagnosis is based on single natural or recombinant allergen molecules. Molecular allergy diagnosis has entered clinical practice in humans, 28 allowing either approaches "from clinics to molecules" or "from molecules to clinics". 29 It is debated whether it might replace skin testing screenings in the future, 30 or not. 31 In particular, the immuno-solid-phase allergen chip (ISAC) has revolutionized human allergy diagnosis. The first ISAC microarray has enabled serum IgE testing on 94 allergens 32,33 ; the next-generation microarray ISAC112 extended to 112 molecules is applied in today 0 s daily allergy diagnosis. 28  reasons, it is unfortunate that component-resolved diagnosis has not yet reached veterinary allergy diagnosis. This is one reason why today there is still a knowledge gap regarding allergen components relevant in the animals. 35 To this aim, we expanded an allergen chip for IgE serology in animals adding 19 more molecules of potential relevance in veterinary allergy diagnosis (Table S1). We considered proteins from the following allergen sources potentially important: from other animals (albumins from rat, guinea pig and rabbit; lipocalin from mouse; alpha-Gal; from human (uteroglobin, IgG, profilin and Hom s 2 36 ), from the commensal yeast Malassezia associated with a number of different skin disorders such as atopic eczema, 37,38 from midges, 39,40 relevant via skin [41][42][43] and from plant food. 44 In a pilot study, using sera from clinically well-characterized allergic horses with various symptoms and horses without clinical allergy (

| Testing for specific equine IgE by ISAC131
Serum samples of all horses were tested on ISAC131 multiplex array ( Figure S1A, B) for IgE binding to 131 allergens: 112 corresponding to commercial ISAC112, plus 19 additional allergens (Table S1). The additional allergens were selected due to potential impact for veterinary patients 35 and comprised molecular allergens from other animals (albumins, lipocalins and alpha-Gal), human antigens, 36 Figure 2C) and food allergens ( Figure 2D). Only results accounting for over 10% of the cohort will be discussed in depth in the light of clinical symptoms. In the allergen group A (Figure 2A) Figure S1, additional allergens on ISAC131 in Table S1 [Colour figure can be viewed at wileyonlinelibrary.com] reacted specifically to fungal Alt a 1, to worm Ani s 1, cockroach Bla Overall, the most abundant IgE sensitizations were seen to Fag e 2 > Cyn d 1 > Aln g 1, but we were not able to find correlations between sensitization patterns in the different clinical cohorts (Table 1).
All clinical groups show a comparable prevalence of IgE reactivity to buckwheat allergen Fag e 2, between 62% of sera from eczema, up to 100% in RAO patients ( Table 2). Highest IgE-binding intensities above 10 ISU were observed in 1 eczema, 2 coughing and 1 urticaria patients (Figure 3). The specificity of equine IgE binding to buckwheat and its 2S-albumin Fag e 2 was confirmed by IgE inhibition experiments in immunoblotting (Data S1 and Figure S2).

| DISCUSSION
Little is known about possible allergen sources for horses and especially about the responsible allergen molecules, 35 including food allergy, 3 except in summer eczema, where IgE to Culicoides allergens play an important role. 40 While intradermal tests with crude Culicoides whole body often results in positive intradermal test reaction in clinically healthy horses, the use of recombinant Culicoides allergens allows a much more specific diagnosis of summer eczema, 47 in clinically healthy, but sensitized horses. 48 We designed the ISAC131 multiplex microarray and tested the IgE binding to 131 allergens using sera from 51 horses from different breeds and different countries of origin (Table 1). Equine total serum IgE levels are approximately 3 logs higher than in humans and did previously not discriminate allergic from healthy horses. 49,50 We also found specific IgE in the group of horses without allergic symptoms, which we interpret as clinically inapparent sensitizations. 49 The higher IgE levels in serum of horses could have caused high background levels in ISAC131 which in three cases impeded evaluation. Furthermore, especially high IgG concentrations and their possible cross-binding to the allergen, 51 or competition among the multiplexed allergens for such cross-reacting IgE could also influence the signal quality, especially at lower IgE levels. 52 In general, the ISAC131 results (Figures 2 and 3) appropriately reflected the known susceptibility of horses to tree and grass pollen. 2,5 Interestingly, the major alder pollen allergen Aln g 1, but not Bet v 1 from birch pollen, was identified as a major respiratory sensitizer in 18 cases. Both pollen major allergens from the botanic species Fagales belong to the PR-10 family, a protein family with innate immune function in plants. 53 They are able to ignite Th2 immune responses in humans and animals by their ligand-binding capacity. 54 PR-10 molecules are highly cross-reactive and can sensitize human atopic individuals; in humans, this is usually dominated by IgE responses to the birch major allergen Bet v 1, 55 at least in Middle and Northern Europe. We speculate that possibly around paddocks and often associated ponds, alders may be most prominent and, therefore, represent the primary sensitizing allergen source. This theory is in conflict with a recent report that for human allergics Bet v 1 is the leading allergen also in the birch-free Mediterranean area. 56 In 2 of the 4 horses reacting via IgE to PR10 allergen Mal d 1 from apple, co-sensitization to Aln g 1 could be found. To this end, it has not been shown that horses may develop oral allergy syndrome, which in humans is a common clinical problem due to sensitization to PR-10 allergens, also sporadically reported to occur in companion animals other than horses. 2 The second most abundant sensitization was found to Bermuda grass allergen Cyn d 1 in 14 of the 51 horses investigated. As in the case of tree pollen, a great degree of IgE cross-reactivity is elicited by grass pollen, with group 1 and group 5 pollen allergens being most important in humans. Cyn d 1 belongs to group 1 grass pollen allergens, but no cross-reactivity was observed with Phl p 1 from timothy grass. This could be due to grass seeding strategies on paddocks, which is a hotly debated topic among horse owners. Natural (n) Cyn d 1, nPhl p 4, nApi g 5, nCup a 1 and MUXF3 on ISAC131, expresses cross-reactive carbohydrate determinants (CCDs) which principally could lead to nonspecific IgE binding. 57 However, only in one horse positive for Cyn d 1, simultaneous IgE reactivity could be detected to Phl p 4 and/or Phl p 1. We consider thus anti-Cyn d 1 IgE to be specific and non-CCD dependent, at least in the cohort of our pilot study.
In addition to the animal allergens on commercial ISAC112, we added animal and human antigens on ISAC131, however, without revealing any significant sensitization in the investigated horse cohorts. The most frequent sensitizer was Can f 3, serum albumin from dog, whereas in only one horse IgE to Can f 5 was found, suggesting exposure to a male dog. 58 The IgE reactivities were in most incidences paired with serological IgE cross-reactivity to Equ c 1, the equine lipocalin. We did not find IgE to cat, rabbit, rat or mouse lipocalins.  Only sporadic sensitizations to Mala s 1, Mal s 6, 10 and 12 37 were found. These allergens are characterized from M. sympodialis a lipid-dependent yeast, part of our normal skin microbial flora, but also associated with several common skin disorders such as atopic eczema/dermatitis, 38,63,64 and may be associated with food allergy. 65 Malassezia has so far been detected on the skin of healthy horses. 66 We failed to establish the IgE-microarray testing for Culicoides animal studies, 71 hypertension and diabetes, 72,73 and tumour growth. 74 Interestingly, Buckwheat is a constituent of some "horse mueslis," or horse crackers. Unfortunately it contains Fag e 2, which belongs to the 2S-abumin family, 75 pepsin-resistant food allergens 76 with the capacity to exacerbate atopic dermatitis, elicit urticaria, angioedema and in severe cases anaphylaxis in human patients. [77][78][79] Few studies describe human buckwheat allergy in Asia, 80,81 Italy, 82,83 or Britain 84 . This is the first report identifying IgE to Fag e 2 in equine sera, using ISAC131. We could confirm the specificity of anti-Fag e 2 IgE binding to buckwheat in immunoblotting inhibition experiments.
This indicates that horses may be sensitized to buckwheat, which recently has gained interest as a dietary allergen relevant for human patients 78,81 in whom buckwheat has been associated with anaphylactic reactions. 81 However, it remains to be investigated by the veterinary discipline whether Fag e 2 indeed is responsible for any of the clinical allergic symptoms in horses.
In conclusion, we report here that IgE testing using a multiplex