IgE-Binding Epitopes of Pis v 1, Pis v 2 and Pis v 3, the Pistachio ( Pistacia vera ) Seed Allergens

: Sequential IgE-binding epitopes were identiﬁed on the molecular surface of the Pis v 1 (2S albumin), Pis v 2 (11S globulin / legumin) and Pis v 3 (7S globulin / vicilin)—major allergens from pistachio ( Pistacia vera ) seeds—using the Spot technique. They essentially consist of hydrophilic and electropositively charged residues well exposed on the surface of the allergens. Most of the epitopic regions identiﬁed on Pis v 1 and Pis v 3 do not coincide with the putative N -glycosylation sites and thus are not considered as glycotopes. Surface analysis of these epitopic regions indicates a high degree of conformational similarity with the previously identiﬁed epitopic regions of the corresponding allergens Ana o 1 (vicilin), Ana o 2 (legumin) and Ana o 3 (2S albumin) from the cashew ( Anacardium occidentale ) nut. These results o ﬀ er a molecular basis for the IgE-binding cross-reactivity often observed between pistachio and cashew nut. They support the recommendation for prescribing pistachio avoidance in cashew allergic patients. Other conformational similarities were identiﬁed with the corresponding allergens Ses i 1 (2S albumin), Ses i 3 (vicilin) and Ses i 6 (legumin) from sesame ( Sesamum indicum ), and Jug r 1 (2S albumin), Jug r 2 (vicilin) and Jug r 4 (legumin) from walnut ( Juglans regia ). Conversely, conformation of most of the epitopic regions of the pistachio allergens often di ﬀ ers from that of epitopes occurring on the molecular surface of the corresponding Ara h 1 (vicilin), Ara h 2 (2S albumin) and Ara h 3 (legumin) allergens from peanut ( Arachis hypogaea ).


Introduction
The IgE-mediated anaphylaxis to tree nuts, such as almond (Prunus dulcis), Brazil nut (Bertholletia excelsa), cashew nut (Anacardium occidentale), hazelnut (Corylus avellana), pecan nut (Carya illinoinensis), pistachio nut (Pistacia vera), walnut (Juglans regia), has now become a public health concern, responsible for a major proportion of often severe anaphylactic shocks in both children and adults [1][2][3]. Although belonging to different botanical families, tree nuts share with peanut (Arachis hypogaea) and other edible seeds from lentil (Lens culinaris), pea (Pisum sativum), kidney bean (Phaseolus vulgaris), soybean (Glycine max), sesame (Sesamum indicum) and buckwheat (Fagopyrum esculentum), three groups of 2S albumins, 7S globulins or vicilins, and 11S globulins or legumins that have been recognized for a long time as being the most frequent allergen sources in seeds [4]. However, except for peanut and Table 1. List of sera from subjects allergic to cashew nut and pistachio.

Subjects
Sex/Age (Years) ** Pistachio Specific IgE (KU/L) *** Cashew Specific IgE (KU/L) Sera from patients allergic to cashew were used as probes to identify the IgE-binding cross-reacting epitopic stretches from Pis v 1, Pis v 2, Pis v 3 and Ana o 3, immobilized on cellulose sheets. Sera from patients allergic to sesame (Table 2), were used as probes to identify the IgE-binding epitopic stretches from Ses i 3 and Ses i 6, immobilized on cellulose sheets.

Analytical Methods
Protein extract was prepared from frozen seeds by two grinding steps of 40 s each in a Fast Prep-24 homogenizer (MP Biomedicals, Illkirch, France), in 20 mM Tris-HCl buffered saline (pH 7.4). The resulting slurry was centrifuged at 15,000× g for 10 min at 4 • C. The supernatant fraction was carefully collected while avoiding the floating lipid layer, filtered through a 0.2 µm membrane, and stored at −20 • C until used.
The protein content of the pistachio extract was estimated using the bicinchonic acid kit reagent (Pierce) [24] with bovine serum albumin as a standard. The protein extract was checked by SDS-PAGE in 15% polyacrylamide gels using Tris-glycine as trailing ion [25] and staining with Coomassie blue. Coomassie blue stained bands were digested with trypsin in the gel and mass mapped by MALDI-TOF analysis as previously described [26]. The software Protein Prospector was used for the identification of the protein using the NCBI non-redundant database.

Immunoblotting
IgE-containing sera from allergic patients were used in Western blotting as probes for the pistachio allergens. Following 1D SDS-PAGE, proteins were transferred onto a Protran nitrocellulose 0.2 µm membrane (Sigma-Aldrich, L'Isle d'Abeau Chesnes, France) at 20 V during 45 mn using a 48 mM Tris/39 mM glycine/20% (v/v) methanol mixture. After an overnight incubation in 10 mM PBS (pH 7.4) containing 0.2% (v/v) tween and 5% (v/v) skimmed milk, the membrane was soaked in the patient IgE-containing sera diluted 1:10 in the same buffer and incubated for 2 h in a moist chamber. After three washings of 10 min each with the same buffer, the membrane was soaked in rabbit HRP-labelled anti-human IgE diluted 1/5000 in the buffer and incubated for 1 h under gentle stirring. Following three washings of 10 min each with buffer, the immunolabelled spots were detected using the ECL Plus detection (ThermoScientific, Illkirch, France) after 3 min exposure in cassette. All the handling was carried out at room temperature.

Peptide Synthesis and IgE-Binding Epitope Mapping
Overlapping 15-mer peptides, frameshifted by three residues, covering the entire amino acid sequences of Pis v 1, Pis v 2 and Pis v 3 were prepared by using the SPOT technique [27]. The protocol previously described in detail [28] was followed with the exception of the utilization of the Multipep automatic Spot synthesizer (Intavis). Briefly, peptides were assembled using the Fmoc chemistry on a cellulose membrane harboring an amino polyethylene glycol moiety. The C-terminal residue of each peptide was coupled to the activated membrane. After Fmoc deprotection, the following amino acids were sequentially added. At the end of the synthesis, side chain protecting groups were removed by a trifluoracetic acid treatment while the linkage of the peptides to the membrane was maintained.
The membrane was soaked overnight into 20 mL of Tris-buffered saline (TBS) containing 2 mL blocking buffer (ThermoFisher, Montigny-le-Bretonneux, France) and 1 g sucrose (pH 7.0), and then washed three times with TBS containing 0.1% (v/v) tween (TBSTw). A 1:10 (v/v) diluted pool of patient Table 3. Geometric and thermodynamic qualities of the three-dimensional models built by homology modeling with YASARA.

Number of Models Built by Homology Modelling
Residues out of the Allowed Areas in the Ramachandran Plot
The pistachio allergens Pis v 1, Pis v 2 and Pis v 5, and Pis v 3, expressed in mature pistachio seeds, accumulate gradually in protein bodies during the seed ripening process, as seed storage proteins ( Figure 1A). Major allergens of protein bodies are easily detected in immunofluorescence experiments, using sera from patients allergic to cashew as primary antibody and an anti-human secondary antibody coupled to Alexa 633, respectively ( Figure 1B). Allergens allergens from pistachio seeds have been identified and character (2S albumin) [10] and Pis v 2 (11S globulin/legumin), and three add ulins, legumin) [acces. B7SLJ1], Pis v 3 (7S globulin, vicilin) [11] an ed in the WHO/IUIS Allergen Nomenclature database (http://ww uary 2021) [53]. In addition, another vicilin-like, two additional 11 en identified from the genome assembly available for pistachio (GC ow whether the identified potential allergens are actually expres o allergens Pis v 1, Pis v 2 and Pis v 5, and Pis v 3, expressed in m e gradually in protein bodies during the seed ripening process, 1A). Major allergens of protein bodies are easily detected in immu ng sera from patients allergic to cashew as primary antibody and dy coupled to Alexa 633, respectively ( Figure 1B).  Upon SDS-PAGE, the major allergens from pistachio are well separated and readily interact with the IgE-containing patient sera in Western blot experiments ( Figure 2). However, depending on the patient sera, different allergens are predominantly recognized by the IgE antibodies. Accordingly, several different patient sera should be used to accurately identified the IgE-binding epitopic stretches of Pis v 1, Pis v 2 and Pis v 3 mapped on the corresponding activated membranes.

FOR PEER REVIEW
. SDS-PAGE of a protein extract from pistachio nut (line 1) and Western blots of ra that preferentially react with Pis v 1 (line 3) and Pis v 2 (lane 4). Molecular weight ted on lane 2. Protein fractions corresponding to the Pis v 1, Pis v 2 and Pis v 3 allerge by MALDI-TOF after trypsic digestion in gel of the protein bands (line 1). * and ** IgE-reactive protein fractions. d Closely Related 2S Albumins gy modelling of the Pis v 1 allergen from appropriate templates yielded cture built up of two α-helix-containing polypeptide chains linked toge disulfide bridges ( Figure 3C,F). As predicted from the GlyP .glycosciences.de/modeling/glyprot/php/main.php, accessed on 9 January 2 utative N-glycan site at 13NLS should be actually glycosylated. in IgE-binding epitopic stretches were identified along the amino acid seq periments using sera that predominantly recognized Pis v 1 in Western blo ). These 3 stretches correspond to four distinct IgE-binding regions exp rface ( Figure 3D,G). Most of these epitopes contain charged residues and c sitively and electronegatively charged regions occurring at the molecular s ure 3E,H). elled 2S albumin allergen from cashew nut, Ana o 3, exhibit a very similar

Pis v 1 and Closely Related 2S Albumins
Homology modelling of the Pis v 1 allergen from appropriate templates yielded a typical 2S albumin structure built up of two α-helix-containing polypeptide chains linked together by two conserved disulfide bridges ( Figure 3C,F). As predicted from the GlyProt server (http://www.glycosciences.de/ modeling/glyprot/php/main.php, accessed on 9 January 2021) [54], the N-terminal putative N-glycan site at 13NLS should be actually glycosylated.
Four main IgE-binding epitopic stretches were identified along the amino acid sequence of Pis v 1 in Spot experiments using sera that predominantly recognized Pis v 1 in Western blots as a probe ( Figure 3A,B). These 3 stretches correspond to four distinct IgE-binding regions exposed on the molecular surface ( Figure 3D,G). Most of these epitopes contain charged residues and coincide with the electropositively and electronegatively charged regions occurring at the molecular surface of the allergen ( Figure 3E,H).
The modelled 2S albumin allergen from cashew nut, Ana o 3, exhibit a very similar fold ( Figure 4C,F), and four main IgE-binding epitopic stretches were similarly identified along the amino acid sequence of Ana o 3 ( Figure 4D,G), which also correspond to electropositively and electronegatively charged regions occurring on the molecular surface of the allergen ( Figure 4E,H). Very similar pictures were previously obtained with other 2S albumin allergen Jug r 1 from walnut (Juglans regia) [55] and Ses i 1 from sesame (Sesamum indicum) [56].     Other less phylogenetically closely related 2S albumin allergens, such as Ses i 1 and Ara h 2, share much less similarity with Pis v 1 and Ana o 3.      Other less phylogenetically closely related 2S albumin allergens, such as Ses i 1 and Ara h 2, share much less similarity with Pis v 1 and Ana o 3.

Pis v 3 and Closely Related 7S Globulins (Vicilins)
The modelled vicilin allergen Pis v 3 corresponds to a homotrimer built from the tail to tail non covalent association of three identical single-chain protomers made of a core of two cupin motifs, extended at both ends by two side arms made up of α-helices ( Figure 8C,F). Each protomer contains a putative N-glycosylation site at 243NIT, which is predicted to be actually glycosylated according the GlyProt server (http://www.glycosciences.de/modeling/glyprot/php/main.php, accessed on 9 January 2021) [54]. This type of structural organization currently occurs in many other 7S globulin/vicilin allergens from other tree nuts and legume allergens [57]. Up to nine main IgE-binding stretches identified along the amino acid sequence of the Pis v 3 protomer using the Spot method ( Figure 8A,B), correspond to more or less exposed IgE-binding epitopic areas arrayed on both faces of the Pis v 3 homotrimer ( Figure 4C,F). In fact, no information is available on the exposition of epitopes #1 and #2 at the surface of Pis v 3 since both epitopes occur in the N-terminal region of the polypeptide chain which is lacking in the three-dimensional model built for Pis v 3 by homology modelling. Two other epitopic region corresponding to epitopes #4 (colored magenta) and #8 (colored sienna), are almost completely buried and very little exposed on the surface of the allergen ( Figure 4D,G). Like for other pistachio allergens, the well exposed epitopic regions corresponding to epitopes #3, #5, #6, #7 and #9, respectively, mostly coincide with both electronegatively (colored red) and electropositively (colored blue) charged regions, and their coalescence creates more extended epitopic regions on the molecular surface of the allergen. Other single surface-restricted epitopes (colored medium blue) are usually well exposed and mostly coalescent with the more extended IgE-binding patches. In addition, the exposed region of epitope #7 (colored sienna) could correspond to a CCD since it contains the putative 367NIT N-glycosylation site occurring on the amino acid sequence of Pis v 2.
Up to twelve IgE-binding stretches were similarly identified along the amino acid sequence of the corresponding protomer Ses i 3 from sesame seeds, using the Spot method ( Figure 9A,B) and, similarly, some epitopic regions corresponding to epitopes #4 (colored magenta) and #6 (colored purple), are poorly exposed at the surface of Ses i 3 whereas other exposed epitopes #2, #3, #5, #7, #8, #9, #10, #11 and #12 ( Figure 9C,D,F,G), readily coincide with electronegatively (colored red) and electropositively (colored blue) charged regions arrayed on the molecular surface of Ses i 3 ( Figure 9E,H). According to the common and superposable fold of the 7S globulin/vicilin allergens together with their epitopic similarities, the multiple alignment of these seed allergens exhibits a high degree of both identity and similarities, especially for Pis v 3 and Ana o 1, which belong to the same family of Anacardiaceae, and to a lesser extent, for Cor a 11 and Jug r 6 (Juglandaceae) and Ses i 3 (Pedaliaceae) ( Figure 10). Accordingly, Pis v 3 and Ana o 1, and Cor a 11, Jug r 6 and Ses i 3, are distributed in two phylogenetically closely related clusters in the unrooted phylogenetic tree built for the 7S globulin/vicilin allergens ( Figure 11).  120  130  140  150  160  170  180  190  200  210  220  230  Pis v 3 KQLCRFRCQEKYKKER  580  590  600  610  620  630  640  650  660  670  680  Pis v 3 VASGNQNLEILCFEVNAEGNIRYTLAGK    Interestingly, the first epitopic stretch identified at the N-terminal end of Pis v 3 amino acid sequence (epitope #1), exhibits some amino acid identities with linear IgE-binding epitopes previously identified at the N-terminal end of Ana o 1, Ara h 1 and Len c 1 (Figure 12). This is also the case of an epitopic stretch recently characterized at the N-terminus (26-83) of Ara h 1, recognized as a major epitope responsible for the IgE-binding activity of the 7S basic peanut protein fraction [58]. In fact, this IgE-binding epitope coincides with two other linear IgE-binding epitopes that had been previously identified along the Ara h 1 amino acid sequence [59,60].

Pis v 2 and Closely Related 11S Globulins (Legumins)
The cupin allergen Pis v 2 consists of a legumin homotrimer resulting from the non covalent association of three protomers built up from two cupin motifs ( Figure 13C,F). In fact, each protomer consists of a large acidic N-terminal and a shorter basic C-terminal subunit covalently associated by a single disulfide bridge. Finally, two homotrimers associated face to face to build an hexameric structure corresponding to a dimer of homotrimers. Like in other legumin hexamers [57], the face to face association of both homotrimers should predominantly result from electrostatic interactions occurring between the oppositely charged faces of the homotrimers, one face being predominantly electronegative (colored blue) whereas the other is essentially electropositive (colored red) ( Figure 13E,H).
Up to ten main IgE-binding epitopic stretches were identified along the amino acid sequence of the Pis v 2 protomer together with single epitopic spots (colored grey in Figure 13B), using the Spot method ( Figure 13A,B). They correspond to IgE-binding patches well exposed at the surface of Pis v 2, except for epitopes #1 (colored red) and #3 (colored green), which are predominantly buried and much less exposed on the surface ( Figure 13C,D,F,G). As with other allergens with cupin motifs, the more exposed IgE-binding patches coincide with the localization of electropositively (colored blue) and electronegatively (colored red) regions on the the molecular surface ( Figure 13E,H). Ten distinct IgE-binding epitopic stretches were revealed along the amino acid sequence of Ses i 6, the legumin allergen from sesame, associated to some IgE-binding single-spots, using the Spot method ( Figure 14A,B). Except for epitope #2 (colored blue), other epitopes #1, #3, #4, #5, #6, #7, #8, #9 and #10, are nicely exposed on the surface of the Ses i 6 homotrimer ( Figure 14C,D,F,G) and most of them coincide with electropositively (colored blue) and electronegatively (colored red) charged regions on the surface of Ses i 6 ( Figure 14E,H). Multiple amino acid sequence alignment of 11S globulin/legumin allergens shows a high degree of both identity and homology/similarity between the members of this protein family, especially between the members of the Anacardiaceae (Ana o 2, Pis v 2, Pis v 5), Juglandaceae (Jug n 4, Jug r 4) and Betulaceae (Cor a 9) families ( Figure 15). Accordingly, both Ana o 2, Pis v 5, Jug n 4, Jug r 4 and Cor a 9, are closely clustered in the dendrogram built for the 11S globulin/legumin allergens ( Figure 16). However, Pis v 2 appears as being more distantly related to Ana o 2, compared to Pis v 5.    In agreement with the phylogenetic relationships observed between Pis v 2, Pis v 5, Ana o 2, Jug n 4, Jug r 4, and Cor a 9, the sequential IgE-binding epitopic stretches identified along their amino acid sequences exhibit pronounced similarities (Figure 17), that most probably account for the IgE-binding cross-reactivity reported between some of these legumin allergens [61]. However, less similarity occurs between Pis v 2 and other phylogenetically related legumin allergens whereas Pis v 5, the other legumin allergen from pistachio, appears as more closely related to other legumin allergens than Pis v 2. Compared to Pis v 2, Pis v 5 exhibits a very similar amino acid sequence with, however, a few amino acid changes. In spite of these changes, regions in Pis v 5 corresponding to the IgE-binding epitopes identified in Pis v 2, look like very similar ( Figure 17). Conversely, Ara h 3, which belongs to the rather distantly related Fabaceae family, exhibits a higher degree of epitopic similarity with the other legumin allergens (Figure 17).   A detailed surface analysis of epitope #1 of Pis v 1 reveals that it shares a similar topographical distribution of the amino acid residues along the α-helix, associated to partial conformational similarities, with the corresponding epitope #1 from Ana o 3 ( Figure 19A,B). Similarly, epitope #8 from Pis v 2, exhibits a similar topographical distribution of amino acids and partial conformational similarities with the corresponding epitope from Ana o 2 ( Figure 19C). Obviously, these epitopic similarities should account for the IgE-binding cross-reactivity often reported among the different allergens from pistachio and cashew nut [61]. However, these conformational similarities could only be observed between closely related epitopes sharing a high degree of identity. In other cases of A detailed surface analysis of epitope #1 of Pis v 1 reveals that it shares a similar topographical distribution of the amino acid residues along the α-helix, associated to partial conformational similarities, with the corresponding epitope #1 from Ana o 3 ( Figure 19A,B). Similarly, epitope #8 from Pis v 2, exhibits a similar topographical distribution of amino acids and partial conformational similarities with the corresponding epitope from Ana o 2 ( Figure 19C). Obviously, these epitopic similarities should account for the IgE-binding cross-reactivity often reported among the different allergens from pistachio and cashew nut [61]. However, these conformational similarities could only be observed between closely related epitopes sharing a high degree of identity. In other cases of couple of epitopes sharing a moderate level of amino acid sequence identity, no relevant conformational similarities could be identified.  In addition to the IgE-binding epitopic similarities observed among the different members of the same allergen family, other cross-reactivity has been previously reported between the CD 4 (+) T cell epitopes from cashew nut, hazelnut and pistachio [62].

Structural Similarities Observed among the Tree Nut Epitopic Regions
These IgE-binding epitopic amino acid identities, sometimes associated with conformational similarities, come in support to the well recognized cross-reactivities among the allergens from different species belonging to the same family of Anacardiaceae (pistachio, cashew), Juglandaceae (walnut), Fabaceae (peanut), Pedaliaceae (sesamum) and Betulaceae (hazelnut), previously reported from immunodiffusion and crossed immuno-electrophoresis experiments, Western blotting experiments, recognition of recombinant allergens and clinical assessments [63][64][65] (Figure 20).

Predicted Resistance of IgE-Binding Epitopic Regions to Proteolysis
The allergenicity of the vicilin allergen Ara h 1 from peanut (Arachis hypogaea), has been attributed in part to the particular arrangement of the monomers in the typical homotrimeric structure of vicilin/7S globulin proteins [66]. The quaternary association of monomers in the homotrimer has been proved to mainly depends on hydrophobic interactions responsible for the swapping of monomers by their distal ends, allowing the major IgE-binding epitopes located along these buried extremities to escape the proteolytic degradation by digestive enzymes. However, in the case of Pis v 3 (vicilin) and Pis v 2 (legumin), an additional mechanism could explain the resistance to digestive proteolysis. Looking at the distribution of the putative cleavage sites for pepsin predicted to occur at the molecular surface of Pis v 3 and Pis v 2, using the web server PeptideCutter of Expasy (https://web.expasy.org/peptide_cutter/, accessed on 15 January 2021), suggests that some of the IgE-binding epitopes previously identified does not contain cleavage sites and could remain in attacked by the protease, e.g., epitopes #3, #4, #6, #9 from Pis v 3, and epitopes #1, #2, #3, #5, #6, #7 from Pis v 2 ( Figure 21).
However, due to the occurrence of numerous putative cleavage sites for trypsin predicted in the IgE-binding epitopes of both Pis v 3 and Pis v 2, most of these epitopes should be degraded later, during the intestinal digestion in the presence of trypsin, except for epitopes #1, #2 and #8 from Pis v 2, which do not contain any K or R residues ( Figure 13).

Discussion
Using the Spot technique with a panel of IgE-containing sera from patients allergic to pistachio and cashew nut, sequential IgE-binding epitopes were identified on the molecular surface of the modelled Pis v 1 (2S albumin), Pis v 2 (legumin) and Pis v 3 (vicilin) allergens, respectively. These epitopic amino acid stretches essentially consist of hydrophilic (N,Q,S,T) and charged residues (D,E,K,R). In this respect, these residues account for ~60% of the epitopic stretches identified on Pis v 3. Accordingly, most of these epitopic regions coincide with the electropositively and electronegatively charged areas occurring on the surface of the allergens. Moreover, and whatever the size of the allergens, a few IgE-binding epitopes locate in the same area and their coalescence should create more extended epitopic surfaces susceptible to correspond to discontinuous epitopes. However, this clustering tendency is more or less pronounced depending on the allergens (see Figure  3D,G for Pis v 1, Figure 8D,G for Pis v 3, and Figure 13D,G for Pis v 2). A similar epitopic coalescence was observed for the counterpart allergens of Pis v 1 ( Figure 4D,G for Ana o 3), Pis v 3 ( Figure 9D,G for Ses I 3) and Pis v 2 ( Figure 14D,G for Ses I 6). Finally, some IgE-binding epitopes well exposed on the surface of the cupin allergen protomers, e.g., epitope #8 of Pis v 3 (which contains the putative N- , magenta (epitope #4), yellow (epitope #5), purple (epitope #6), dark green (epitope #7), brown (epitope #8) and dark blue (epitope #9), respectively. The IgE-binding epitopic regions identified on the surface of Pis v 2 are numbered and colored red (epitope #1), pale blue (epitope #2), pale green (epitope #3), magenta (epitope #4), yellow (epitope #5), purple (epitope #6), dark green (epitope #7), brown (epitope #8), dark blue (epitope #9) and orange (epitope #10), respectively.
However, due to the occurrence of numerous putative cleavage sites for trypsin predicted in the IgE-binding epitopes of both Pis v 3 and Pis v 2, most of these epitopes should be degraded later, during the intestinal digestion in the presence of trypsin, except for epitopes #1, #2 and #8 from Pis v 2, which do not contain any K or R residues ( Figure 13).

Discussion
Using the Spot technique with a panel of IgE-containing sera from patients allergic to pistachio and cashew nut, sequential IgE-binding epitopes were identified on the molecular surface of the modelled Pis v 1 (2S albumin), Pis v 2 (legumin) and Pis v 3 (vicilin) allergens, respectively. These epitopic amino acid stretches essentially consist of hydrophilic (N,Q,S,T) and charged residues (D,E,K,R). In this respect, these residues account for~60% of the epitopic stretches identified on Pis v 3. Accordingly, most of these epitopic regions coincide with the electropositively and electronegatively charged areas occurring on the surface of the allergens. Moreover, and whatever the size of the allergens, a few IgE-binding epitopes locate in the same area and their coalescence should create more extended epitopic surfaces susceptible to correspond to discontinuous epitopes. However, this clustering tendency is more or less pronounced depending on the allergens (see Figure 3D,G for Pis v 1, Figure 8D,G for Pis v 3, and Figure 13D,G for Pis v 2). A similar epitopic coalescence was observed for the counterpart allergens of Pis v 1 ( Figure 4D,G for Ana o 3), Pis v 3 ( Figure 9D,G for Ses I 3) and Pis v 2 ( Figure 14D,G for Ses I 6). Finally, some IgE-binding epitopes well exposed on the surface of the cupin allergen protomers, e.g., epitope #8 of Pis v 3 (which contains the putative N-glycosylation site at 243NIT), become partly buried upon the oligomeric association of the protomers. Accordingly, the allergenic potency of the Pis v 3 protomer should slightly differ from that of the Pis v 3 homotrimer. Obviously, these discrepancies depend on the mode of association of the protomers in the vicilin and legumin homotrimers [62]. In addition, the occurrence of a N-glycan chain on each Pis v 3 protomer at the N-glycosylation site 243NIT located on epitope #8, which is predicted to be actually glycosylated by GlyProt (http://www.glycosciences.de/modeling/glyprot/php/main.php, accessed on 15 January 2021) server, should participate in the allergenic potency of the allergen. N-glycans have been demonstrated to participate in the IgE-binding activity of N-glycosylated epitopes of e.g., the structurally related Ara h 1 vicilin allergen from peanut [63]. The N-glycan chain of the 13NLS glycosylation site of Pis v 1, which is adjacent to the IgE-binding epitope #1, could similarly participate in the allergenicity as a carbohydrate determinant (CDD), even though there are no evidence for such a role of the N-glycan chains of pistachio allergens.
The approach which has been used to identify the linear or continuous epitopes, which combines the Spot technique with molecular modelling, provides results which need to be interpreted with caution since it does not allow to identify discontinuous or conformational epitopes, depending on the length of the synthetic peptides used as IgE-binding probes. In addition, while the molecular modelling techniques have improved significantly during the past decade, they still remain less reliable than the conventional approaches by X-ray crystallography or NMR to identify and map the IgE-binding epitopes on the molecular surface of the allergen-IgE complexes. However, the coupling of Spot and molecular modelling techniques is a reasonable and expedient compromise to get an insight into the IgE-binding epitopes from allergenic proteins.
Depending on the localization of the exposed pepsin and trypsin cleavage sites on the three-dimensional model built for the Pis v 3 and Pis v 2 protomers (Figure 21), a central core structure remaining unaltered upon the digestive proteolytic attack by pepsin is predicted to occur in both allergens. Interestingly, this core structure still retains some epitopes that should remain protected from the proteolytic attack till Pis v 3 is recognized by the immuno-competent cells of the gastro-intestinal tract. However, except for a few IgE-binding epitopic stretches from Pis v 2, apparently devoid of lysine and arginine residues, most of the epitopes remaining unaltered upon the pepsin attack, should be further degraded due to their richness in electropositively charged residues.
A strong overall IgE-binding cross-reactivity between pistachio and cashew allergens has been often reported from immunoelectrophoretic analyses, Western blotting experiments, and specific IgE measurements [63][64][65]. Both vicilins (Pis v 3, Ana o 1) [11] and 2S albumins (Pis v 1, Ana o 3) [67] have been incriminated as the main IgE-binding cross-reacting allergens. A detailed surface analysis of the epitopes occurring on the molecular surface of these allergens, revealed a rather high degree of amino acid sequence identities, associated or not with conformational similarities depending on the degree of sequence identity occurring between their epitopes. Both sequence identities and conformational similarities establish the molecular basis of the IgE-binding cross-reactivity between the allergens from pistachio and cashew, which belong to the same botanical family of Anacardiaceae. However, the surface analysis performed on the legumin allergens revealed no conformational similarities between most of the sequential IgE-binding epitopes of Pis v 2 and Ana o 2. A conformational IgE-binding epitope (called 2B5) was identified on the molecular surface of Ana o 2 that mainly consists of 24 amino acid residues (20EPDNRVEYEAGTVEAWDPNHEQFR43) located at the N-terminus of the large (acidic) subunit, which is expressed only when associated to the small (basic) subunit [68,69]. The IgE-binding epitopes #1 and #2 of Pis v 2 were identified on the corresponding amino acid sequence stretch (30EPKRRIESEAGVTEFWDQNEEQLQ53) which displays 55% identity (underlined bold letters) with the 2B5 epitope.
Obviously, the amino acid sequence identities between the IgE-binding epitopes from allergens of different origin depend on the degree of conservation the allergens have retained during evolution. Accordingly, the occurrence of IgE-binding epitopes sharing similar physicochemical properties has been pointed out as a key factor contributing to the cross-reactivity among peanut and tree nut allergens [61]. In this respect, 2S albumin allergens (Pis v 1, Ana o 3) and vicilin allergens (Pis v 3, Ana o 1), are predicted to display a degree of conservation higher than that predicted for the legumin allergens (Pis v 2, Ana o 2). Most conserved residues essentially occur along the structurally conserved secondary features, e.g., the α-helices of 2S albumins and the β-strand cupin motifs of the vicilins. These conserved regions are predominantly built up from hydrophilic residues. According to the sequence and conformational similarities observed among 2S albumin and vicilin allergens, Pis v 1 and Pis v 3 of pistachio closely cluster with the corresponding Ana o 3 and Ana o 1 allergens of cashew in the dendrograms built up from the amino acid sequence alignments with the neighbor joining method (Figures 6 and 11). Although Pis v 2 and Ana o 2 allergens appear as being more distantly related in relation with the lower degree of sequence similarities that relate both allergens, another legumin allergen related to Pis v 2, Pis v 5, is closely clustered to Ana o 2 and other tree nut allergens in the legumin dendrogram ( Figure 16).
Due to the high degree of conservation of the 2S albumin and cupin allergen structures, the occurrence of amino acid sequence similarities observed between the pistachio allergens Pis v 1, Pis v 2 and Pis v 3 and other homologous allergens from either closely or distantly related families, has an influence on their allergenicity and cross-reactivity. Especially, the location of the amino acid sequence similarities in regions corresponding to IgE-binding epitopes, greatly determine their capacity to cross-react with the corresponding regions from other homologous allergens. Depending on the degree of sequence identity between epitopes belonging to homologous allergens, the resulting cross-reactivity will be strong, moderate or low. In this respect, the sequence similarities observed between the homologous allergens of Anacardiaceae, e.g., between Pis v 1 and Ana o 3 or between Pis v 2 and Ana o 2, account for the high level of cross-reactivity and cross-allergenicity observed between pistachio and cashew. The IgE-binding cross-reactivity the major allergens from pistachio, Pis v 1, Pis v 2 and Pis v 3, share with homologous proteins from peanut, tree nuts and proteins from, sesame, buckwheat, peppercorn or mango, has a clinical incidence and in particular, helps to discriminate between allergies and co-sensitizations [19,22,[70][71][72][73][74][75][76][77][78]. Whereas co-sensitization frequently occurs between pistachio and cashew, it is clinically relevant in only one-third of cases [72]. However, skin prick tests (SPT) performed on French children indicated that a low reaction dose to cashew in cashew-allergic children would be a predictive factor of allergy to pistachio [78]. Accordingly, oral food challenge to pistachio should be avoided in cashew-allergic children exhibiting a low reaction dose to cashew nut. Acknowledgments: This article is dedicated to the memory of Fabienne Rancé, who actively contributed to the methodology and validation of the results and provided us with the patient sera used in this work. Many thanks to Alain Jauneau (CNRS, Toulouse, France), who provided us with the micrography and histo-immunochemical pistachio slices.

Conflicts of Interest:
The authors declare no conflict of interest.