Structure, immunogenicity, and IgE cross-reactivity among walnut and peanut vicilin buried peptides

Abstract

Vicilin Buried Peptides (VBPs) from edible plants are derived from the N-terminal leader sequences (LS) of seed storage proteins. VBPs are defined by a common α-hairpin fold mediated by conserved CxxxCx_(10–14)CxxxC motifs. Here, peanut and walnut VBPs were characterized as potential mediators of both peanut/walnut allergenicity and cross-reactivity despite their low (~17%) sequence identity. The structures of one peanut (AH1.1) and 3 walnut (JR2.1, JR2.2, JR2.3) VBPs were solved using solution-NMR, revealing similar α-hairpin structures stabilized by disulfide bonds with high levels of surface similarity. Peptide microarrays identified several peptide sequences primarily on AH1.1 and JR2.1 which were recognized by peanut, walnut, and dual-allergic patient IgE, establishing these peanut and walnut VBPs as potential mediators of allergenicity and cross-reactivity. JR2.2 and JR2.3 displayed extreme resilience against endosomal digestion, potentially hindering epitope generation and contributing to their reduced allergic potential.

Graphical Abstract

Introduction

Peanut allergy is among the most commonly reported food allergies, with a prevalence of 1–9% among Western European cultures.¹ Peanut allergy is responsible for over 50% of food-induced anaphylaxis cases and is rarely outgrown, making it a key health concern.^2,3 Furthermore, up to 86% of peanut-allergic patients show some level of sensitization to tree-nuts, particularly walnut, with ~30% displaying clinically-relevant levels of IgE cross-reactivity.^4,5 While allergens from peanuts and tree-nuts can be grouped into similar protein families, their overall sequence identity falls well below the 70% threshold previously considered to support cross-reactivity, raising questions regarding the basis for this phenomenon.⁶

Vicilins are a family of ubiquitous seed storage proteins which are translated as preproproteins consisting of three conserved regions: a short, hydrophobic signal peptide that is removed upon location to a storage vacuole; an N-terminal, cysteine-rich leader sequence (LS) which is removed by asparaginyl endopeptidase (AEP)⁷; and the mature vicilin domain.^8–10 Recent studies have identified a new family of peptides, termed Vicilin-Buried Peptides (VBP) within this LS region. These VBPs are defined by a common cysteine motif (CxxxCx_(10–14)CxxxC), and are widely dispersed across most angiosperm families including trees, grasses, nightshades and legumes^9,10 where they have been implicated in a variety of biological functions including ribosome inactivation, protease inhibition and antimicrobial defense.^11–16 Despite this diversity, the structure of all VBPs that have been solved to date reveal a common α-hairpin fold/motif mediated by disulfide bonds between the highly conserved CxxxC motifs, potentially providing a common structural scaffold that can mediate peanut/tree-nut cross-reactivity.^17,18

The walnut vicilin Jug r 2 leader sequence (JRLS) contains three VBP motifs (JR2.1, JR2.2, JR2.3), while the peanut 7S globulin Ara h 1 contains just one (AH1.1) (Figure 1). These VBPs have been identified in both peanut and walnut extracts,^19,20 and are shown to bind IgE from allergic patients. Curiously, sensitization to the peanut VBP among peanut-allergic patients did not correlate with that of its parent vicilin domain, suggesting that VBP’s could potentially represent an independent family of allergens with distinct physiochemical determinants of allergenicity.^18,20,21 However, information on the cross-reactivity and biophysical properties of these allergen-derived VBPs remains sparse relative to the more studied vicilin/globulin domains. In this work, we present full 3D structures of all four peanut and walnut VBPs, revealing the characteristic α-hairpin motif with a high degree of surface similarity both among themselves, and with other previously characterized VBPs from non-allergic sources. Peptide microarrays confirm the presence of numerous peanut and walnut epitopes on the structured regions of AH1.1 and JR2.1 with IgE prevalence of up to 100% among allergic patients, establishing the VBP α-hairpin as a potent mediator of both peanut and walnut allergy. Despite their shared motif, the IgE reactivity of JR2.2 and JR2.3 was significantly diminished. This correlated with subtle structural differences in and amino acid physiochemical properties between the various VBP sequences. In addition, JR2.2 and JR2.3 showed a markedly higher resistance to simulated endosomal degradation, providing potential insight into the biophysical determinants of VBP sensitization. These studies establish VBPs as a potential family of pan-allergens, whose highly conserved α-hairpin structure provides a potent scaffold for cross-reactivity across distantly related species.

Figure 1:

Amino acid sequence of Peanut/Walnut VBPs. A) N-terminal leader sequence (LS) from Ara h 1 and Jug r 2. Individual VBP sequences are highlighted in orange. Residues present in the NMR structures, but not the native sequence are denoted in parentheses. Conserved cysteine forming the characteristic CxxxC motif highlighted in red B) sequence identity matrix comparing the various Peanut/Walnut VBP sequences.

Materials and Methods

Constructs and Purification

A full description of the expression and purification methods is provided in the supplemental material. In brief, VBP sequences from the peanut (Arachis hypogaea) and English walnut (Juglans regia) allergens Ara h 1 and Jug r 2 were identified and cloned into the pDest expression system with an N-terminal Glutathione S-transferase tag separated from the main sequence by a tobacco etch virus (TEV) protease cleavage site. The vectors were expressed in E. coli (BL21 DE3, Millipore Sigma) and the resulting protein purified using an immobilized glutathione column. TEV protease was used to remove the GST tag, and the final product was isolated using a Superdex75 26/600 gel filtration column in phosphate-buffer saline (PBS).

Patient sera

Sera from 40 individuals (male/female ratio: 17/23) with walnut (12 patients), peanut (12 patients), and peanut/walnut allergy (16 patients) were collected in accordance with rules and regulations of the institutional review board of their respective institutions (Tulane University Biomedical IRB 09-00231, REF #: 140613; University of California at Davis IRB protocol No. 200210194-6) and in accordance with U.S. federal policy for the protection of human subjects (Table 1). Patients were over the age of 18 and have experienced recurrent severe, systemic allergic reactions to peanut, walnut or both. These patients were included in this study based on their convincing clinical history of food allergy and did not undergo food challenge due to the potential severity of reactions.

Table 1.

Patient allergies

Patient #	Age	Sex, race	Symptoms	IgE kU/L	PN Allergy	WN Allergy	Other allergy
1	54	M, W	UA, GI, nLT			✓	Sesame, other TN
2	33	F, W	OAS, nLT			✓	Pumpkin, other TN
3	22	F, W	OAS, nLT			✓	Coconut, fruits, carrots other TN
4	-	M	MO, UA, LA, nLT			✓	Other TN
5	50	M, W	MO, UA, LA, LT			✓	Coconut, other TN
6	38	M	MO, UA, LA, LT			✓	Coconut, fish, other TN
7	35	F	MO, UA, LA, LT			✓
8	53	F	MO, UA, LA, LT			✓	Tomatoes, other TN
9	44	M	MO, UA, LA, LT			✓
10	48	F	MO, UA, LA, LT			✓	Apple, pear, other TN
11	40	F	MO, UA, LA, LT			✓	Sesame, fruits, other TN
12	36	F	MO, UA, LA, LT			✓	Fruits, other TN
13	-	F	MO, nLT, OFC	59	✓		Sunflower
14	24	M	MO, nLT, OFC	>100	✓		Soy, other TN
15	25	M, W	MO, nLT, OFC	>100	✓		Cod, egg
16	35	F, W	MO, nLT, OFC		✓		Soy, other TN
17	12	F, B	MO, LT	68	✓
18	40	F, W	UA, OAS, nLT		✓
19	25	F, W	UA, LA, nLT	40	✓
20	35	M, W	MO, UA, LA, LT	36	✓
21	45	M, W	UA, OAS, nLT		✓
22	40	F, W	MO, nLT, OFC		✓		Shrimp
23	19	M	MO, LT	75.4	✓
24	29	M, W	MO, nLT		✓		Shrimp
25	63	M, W	MO, LT	>100	✓	✓	Crab, lobster, flax, soy, other TN
26	25	F, W	MO, LT	>100	✓	✓	Other TN
27	21	F, W	MO, LT	52.7	✓	✓
28	38	F, A	MO, LT		✓	✓	Poppy, sesame, coconut, other TN
29	42	M, W	OFC, nLT		✓	✓	Other TN
30	48	F, W	MO, nLT		✓	✓	Sesame, other TN
31	62	M, W	MO, nLT, both		✓	✓	Crab, lobster, other TN
32	38	M, W	MO, LT, both	46	✓	✓	Shellfish, other TN
33	36	F, W	MO, LT, both		✓	✓	Shellfish, other TN
34	25	F, W	MO, LT, both		✓	✓	Other TN
35	36	F, W	MO, LT, both		✓	✓
36	33	F	MO, LT, both		✓	✓
37	17	F	MO, LT, both		✓	✓	Shellfish, flax, soy, other TN
38	27	M	MO, LT, both		✓	✓	Poppy seeds, other TN
39	30	M	MO, LT to PN, class 2 to WN	PN: 82, W: 2.2	✓	✓
40	42	F	MO, nLT, both		✓	✓	sunflower

Acronyms: PN: peanut; WN: Walnut; TN: Tree-nut; MO: Multi-organ involvement; LT: Life Threatening; nLT: non-life threatening; OAS: Oral allergy syndrome; OFC+: Oral food challenge positive; UA: upper airway; LA: lower airway

In sex/race column: M=male; F=female, W=white; B=black; A=Asian. In some cases the race was not specified. Final M/F: 17/23

Peptide microarrays

The entire amino acid sequences of Ara h 1 and Jug r 2 were printed onto microarray slides (JPT Peptide Technologies GmbH, Berlin, Germany) as sequential overlapping 15 amino acid spots, offset by 5 amino acids. Slides were placed into a HS400 Pro (Tecan, San Jose, CA), blocked in filtered SuperBlock TBS (Thermo Fisher Scientific, Waltham, MA) for 30 minutes at room temperature under agitation and then washed with Tris-buffered saline containing 0.5% Tween-20 (TBST) (Bio-Rad, Hercules, California). After centrifugation, 200 μL of each patient’s undiluted sera was injected into individual chambers containing microarray slides and incubated at 4°C for 16 hours with agitation. Microarray slides were then washed and injected with 170 μL of mouse α-human IgE (Life Technologies, Grand Island, NY) diluted into filtered Superblock at a dilution of 1:5000 for 30 minutes at room temperature. The slides were washed and dried before scanning on a GenePix-4000B (Molecular Devices, San Jose, CA). IgE binding was measured by Cy3 green fluorescence at 532 nm. The data was analyzed by GenePix Pro 7.2 software.

Statistics

Statistical analyses were performed using R (version 3.6.3). Modified z values were calculated from the microarray median signal-to-noise ratios (SNRs). IgE reactive peptides were defined as being recognized by 50% or more or patients within an allergy group.

Structural and biochemical characterization of VBPs

Detailed NMR and biophysical characterization methods can be found in supplemental material. Briefly, triple-resonance and NOESY spectra were collected on 0.1–1 mM protein samples in PBS using either a 600 or 800 MHz Agilent DD2 console equipped with cryogenically cooled probe. Backbone and side-chain assignments and T1/T2 relaxation times for the oxidized samples were obtained using standard triple resonance techniques employing either the standard VARIAN Biopack or modified BEST-TROSY pulse sequences as described previously^22,23 supplemented by additional 3D or 4D experiments to resolve peak overlap in the reduced samples.^24,25 3D structures were calculated using the PONDEROSA web server.^26,27 Surface similarity was assessed using the SPADE (Surface comparison-based Prediction of Allergen Discontinuous Epitopes) algorithm accessed via the provided web server.^28,29 Amino acid physicochemical property similarities assessed using the Property Distance (PD) tool accessed via the Structural Database of Allergenic Proteins.³⁰

Simulated gastric and endosomal digestion was carried out as described previously using 1.15 μg/mL trypsin (Sigma) an 0.25 U Cathepsin S (Human, recombinant from E-coli – Millipore) respectively in the absence or presence of 2 mM DTT.³¹ Circular dichroism spectra were collected using a Jasco J-815 CD spectropolarimeter (Jasco, Easton MD) and analyzed using the BESTSEL web server.^32,33

Results

Structural characterization of peanut and walnut VBPs reveal a conserved α-helical scaffold

Backbone and side-chain assignments for all four VBPs were obtained using standard triple-resonance approaches. Analysis of the resulting chemical shifts using the TALOS prediction algorithm revealed two α-helical regions consistent with the expected VBP motif (Figure 2). The downfield shift (>33 ppm) of the cysteine Cβ peaks, along with the presence of NOESY cross-peaks between the Cβ protons on the opposite CxxxC repeats support the presence of an α-hairpin disulfide bonding pattern (Figures 1, S1). AH1.1 also contains an additional pair of cysteine residues flanking the main CxxxC motif, which were found to form a third disulfide bond that was verified by mass-spectrometry (Figures 1, S1). This information, along with structural restraints derived from the available NMR data was used to determine the 3D structure of all four VBPs. As shown in Figure 2, all four constructs adopt an unambiguous α-hairpin structure with low backbone RMSD values (Figure S1). Few long-range NOE interactions were detected beyond the disulfide bonded cysteines in each of the VBPs; a rather unusual observation for an ordered protein (Fig. S1). Reducing these disulfide bonds resulted in a decrease in peak dispersion in the ¹H-¹⁵N HSQC spectra of all four peanut/walnut VBPs indicative of a loss of secondary structure (Figure 2). Cα/Cβ chemical shift analysis and CD spectroscopy confirm this conclusion, suggesting that the VBP sequence itself possesses a low degree of conformational stability in the absence of its covalent disulfide linkages (Figures 2, 3). Taken together, this data suggests that AH1.1, JR2.1, JR2.2 and JR2.3 each adopt a common α-hairpin structure maintained almost exclusively by the conserved disulfide bonds. This architecture is common to the VBPs from other plant species such as tomato and buckwheat (Fig. S2),^9,10,15 potentially facilitating cross-reactivity across evolutionarily distant species.

Figure 2:

Peanut and walnut VBPs adopt a common α-hairpin structure mediated by conserved disulfide bonds. A) Secondary structure plot for the peanut/walnut VBPs under oxidizing (black) and reducing (orange) conditions, as calculated based on the backbone and side chain NMR chemical shifts using the TALOS algorithm. Average secondary structure content for all α-hairpin residues are shown in B. C) NMR structures of AH1.1, JR2.1, JR2.2, and JR2.3. Cartoon figures depict representative conformations of each structure. Inter-helix disulfides shown as sticks and highlighted in yellow. Complete NMR data tables are available in Figures S1.

Figure 3:

Loss of VBP structure under reducing conditions. A) Representative circular-dichroism spectra of peanut/walnut VBPs in the absence (black) and presence (orange) of the reducing agent tris(2-carbocyethyl)phosphine (TCEP). The local minima at 220/210 nm and 205 nm are indicative of α-helical and random-coil secondary structure respectively. The relative loss of α-helical structure upon addition of TCEP is quantified in B. Data values and error bars represent the mean and standard deviation obtained from at least three trials from two biological replicates.

The disulfide-mediated α-hairpin allows peanut and walnut VBPs to adopt a conserved motif with similar physiochemical properties despite their low sequence identity (Figure 1). Indeed, previous papers using the Structural Database of Allergenic Proteins (SDAP) Property Distance (PD) index - a tool that compares peptide sequences based on the physicochemical properties of their constituent amino acids -^30,34,35 identified several potential cross-reactive regions on evolutionarily distant allergen families including peanut/walnut VBPs, their parent 2S albumins, and 7s vicilin domains.^17,18 Drawing inspiration from these studies, the surface similarity of the peanut/walnut VBP structures was quantified using the SPADE computational tool developed by Dall’Antonia et al.^28,29 The SPADE algorithm takes into account the physical properties of each residue, similar to the PD metric employed for sequence analysis, but includes an additional layer examining the solvent accessible area and spatial positioning of each residue based on the available 3D structures to calculate a surface similarity score (ΔSIM). Comparing the structure of AH1.1 with the three walnut VBPs reveals a high degree of similarity (Figure 4), supporting the potential for VBPs to mediate peanut/walnut cross-reactivity. Repeating this analysis in the reverse direction yields similar results. In both cases the overall surface similarity (Σ[ΔSIM]) values were noticeably higher between JR2.1, JR2.2, and AH1.1 than JR2.3, potentially reflecting subtle differences in the cross-reactive potential of the latter.

Figure 4:

Quantifying structural similarity across peanut and walnut VBPs A) Structure of AH1.1 colored by surface similarity (ΔSIM) values when compared against JR2.1, JR2.2, and JR2.3 as calculated using the SPADE surface comparison algorithm.^28,29 Cyan and purple represent regions of low and high surface similarity respectively. Total ΔSIM (Σ[ΔSIM]) for all residues is indicated. B) Structure of WN VBPs colored by surface similarity (ΔSIM) values against AH1.1 as calculated using the SPADE surface comparison algorithm. Total ΔSIM (Σ[ΔSIM]) for all residues is indicated. Additional SPADE comparisons were performed on VBP homologues from tomato (VBP-8) and buckwheat (BWI-2c) (Figure S2).

α-hairpin scaffold supports cross-reactive IgE epitopes despite low sequence identity

Overlapping 15-mers representing the complete sequence of all four VBPs were printed onto peptide microarrays, and assessed for IgE binding using sera from peanut (PN), walnut (WN) and dual-sensitized (PW) patients (Table 1). Three immunoactive peptides were identified in AH1.1 (A4, A12 and A13), which show considerable overlap with previously identified immunodominant epitopes of Ara h 1 (Figure 5),^36–38 while seven peptides were identified in the walnut VBPs (JR3, JR5, JR6, JR7, JR8, JR9, JR18) (Table 2). Two trends were clear from the prevalence of IgE binding to the peptide series. First, both PN and WN mono-allergic patients recognized peptides derived from the sequences of both peanut and walnut despite the low sequence identity, suggesting a high degree of cross-reactivity. Second, sera from PN and PW patients displayed higher % IgE binding than the WN patients to all the microarray peptides – including those from the walnut VBPs, further supporting the presence of cross reactivity between the peanut and walnut VBPs.

Figure 5:

Potential IgE-reactive and cross reactive regions on peanut and walnut VBPs. A) Sequence of peanut/walnut VBPs used in the peptide microarray analysis. Residues depicted in the solution-NMR structures are shown in capital letters. Residues from the expression system used to generate the NMR samples but are not present in the peptide microarrays are denoted with brackets. Residues highlighted in blue represent alpha-helices identified in the available NMR structures. Epitopes identified in previous works by Maleki et al.¹⁷ and Burks et al.³⁶ are indicated in grey rectangles, whereas cross-reactive peptides identified in this present work are indicated by colored rectangles. B) IgE-reactive peptides mapped onto the structure of peanut/walnut VBPs. Peptides color-coded as in (A).

Table 2:

IgE-reactive PN/WN VBP peptides.

Peptide #	Amino acid sequence	% Patients IgE binding			Average z value
		WN	PN	PW	WN	PN	PW
A4	VLASVSATHAKSSPY	25	58	38	2.9	3.5	2.2
A10	CQQEPDDLKQKACES	8	42	19	1	5.1	6.4
A11	DDLKQKACESRCTKL	0	0	0	0.7	0.4	0.5
A12	KACESRCTKLEYDPR	42	50	50	3.1	4.2	3.2
A13	RCTKLEYDPRCVYDP	50	92	81	5.3	8.1	7.2
JR3	PRDPREQYRQCQEYC	92	100	94	28.5	24.5	27.4
JR4	EQYRQCQEYCRRQGQ	0	33	19	0.6	2.9	2.7
JR5	CQEYCRRQGQGQRQQ	17	58	50	2.4	4.8	4
JR6	RRQGQGQRQQQQCQI	50	75	75	6.5	17.7	16.3
JR7	GQRQQQQCQIRCEER	75	100	75	10.9	17.2	16.3
JR8	QQCQIRCEERLEEDQ	67	75	81	7.4	9.5	11.2
JR9	RCEERLEEDQRSQEE	50	100	69	6.2	10.3	9.5
JR18	QRRGQEQTLCRRRCE	50	83	69	4.8	8.8	6.8

Acronyms: WN=walnut; PN=peanut; PW=peanut and walnut

Mapping the individual immunoactive peptides onto the VBP structures suggests this cross-reactivity is mediated by the shared α-hairpin motif (Figure 5). Given the overlapping nature of the microarray peptides, A12 and A13, along with a neighboring peptide which barely missed the IgE binding prevalence threshold (A10) likely encompasses a single conformational epitope. Notable, these peptides incorporates significant regions of the ordered α-hairpin structure and is recognized by all three serum groups (PN, WN, PW), demonstrating the cross-reactive potential of the conserved VBP motif (Figure 5). In contrast, the IgE-reactive peptide A4 encompasses the disordered N-terminal region of AH1.1 (Figure 5), potentially indicating a second epitope that is independent of the α-hairpin motif. With regards to the walnut VBPs, the majority of IgE reactive sequences are localized to JR2.1, with only one peptide (J18) on JR2.2, and none on JR2.3 (Figure 5). Of the IgE-reactive walnut peptides, all were localized to the structured α-hairpin region and were recognized by all three patient groups, confirming the VBP motif as a potent mediator of cross-reactive peanut/walnut IgE binding. Taken together, the data suggest that the conserved α-hairpin structure of JR2.1 and AH1.1 mediates both allergic sensitization and cross-reactive IgE binding across peanut and walnut-allergic patients.

Thermodynamic stability and proteolytic resistance as determinants of VBP allergenicity

Previous studies identified protein stability as an important criteria for allergenicity among some allergen families.³⁹ Similar trends could influence the sensitizing potential of VBP allergens, potentially providing a physical basis for the paucity of IgE binding peptides from JR2.2 and JR2.3.^40,41 In the context of food allergens, stability plays a key role in the ability of potential allergens to resist proteolytic digestion, facilitating exposure of the intact antigen to the immune system.⁴² To assess this hypothesis, all four VBPs were subjected to simulated gastric and duodenal digestion using pepsin (pH 2.0) and trypsin (pH 7.4) respectively. The oxidized form of all four VBP’s were extremely resilient to gastric digestion, with no detectable degradation observed under the conditions employed. Reducing the intramolecular disulfide bonds significantly enhanced proteolysis, with almost complete cleavage observed even following a 20-fold reduction in pepsin concentration (Figure 6). Curiously, both JR2.2 and JR2.3 were found to be significantly more resistant than their counterparts under these conditions. Moving onto the duodenum, some VBPs such as BWI-2c from buckwheat and C2 from pumpkin have been shown to act as potent, irreversible trypsin inhibitors with K_i values in the nm range.^7,10,15 However, no such activity was observed for any of the peanut and walnut VBPs tested here (Figure 6). Instead, all four VBPs were readily degraded by trypsin. As with gastric digestion, proteolytic cleavage was noticeably enhanced under reducing conditions, reflecting the loss of the ordered α-hairpin structure though no correlation with IgE binding was observed under either condition.

Figure 6:

Conformational stability of peanut and walnut VBPs contribute to proteolytic stability. Half-life of peanut and walnut VBPs subjected to simulated gastric (A) and duodenal (B) digestion under reducing and oxidizing conditions. C) Half-life of peanut and walnut VBPs subjected to simulated endosomal digestion (reducing conditions only). Faded bars represent conditions under which no appreciable (<10%) digestion was detected. Data values and error bars represent the mean and standard deviation obtained from at least three trials from two biological replicates.

The ability of potential allergens to resist endosomal digestion can also influence the sensitization process, providing an additional avenue through which differences in VBP stability could determine allergenicity.^43–45 To assess this hypothesis peanut and walnut VBPs were subjected to simulated endosomal degradation by Cathepsin S (pH 5.4, 2 mM DTT). Both AH1.1 and JR2.1 displayed moderate stability under these conditions, with half-lives in the 30–60 minute range. In contrast JR2.2 and JR2.3 showed negligible (<10%) degradation under the assay conditions, potentially accounting for the lower sensitization potential of the latter. It should be noted that both JR2.2 and JR2.3 retain a greater proportion of their ordered secondary structure under reducing conditions (Figure 3) as assessed using CD, suggesting that the α-hairpin motif may be more resilient in these sequences, contributing to their proteolytic resistance. NMR chemical shift analysis shows a similar trend (Figure 2) though the statistical difference between AH1.1, JR2.1 versus JR2.2, JR2.3 was borderline significant (p=0.06) when assessed using a Student’s T-test. However, it is consistent with both the CD studies and proteolytic stability results, providing a compelling mechanistic link between the global conformational stability of the VBP motif, endosomal degradation kinetics, and allergenicity.

Discussion

This study demonstrates that while the vicilin LSs from peanut and walnut contain a variable number of VBP motifs, each of these adopt a characteristic α-hairpin motif mediated by disulfides between the conserved CxxxC elements. While homologous structures have been described for other VBPs^9–11,15, this is the first time such a structural motif has been described for a major food allergen. The unusual nature of these sequences - being maintained almost exclusively by disulfide bonds - allows for significant variability in the amino acids surrounding the conserved CxxxC cysteine residues while retaining a high degree of structural similarity as indicated by the high surface similarity values shown in Figure 3. This gives rise to cross-reactive peanut/walnut IgE epitopes despite the low sequence identity. Indeed, it is worth noting that the sequence identity between JR2.1 and AH1.1 is a mere 17% (Figure 1) - falling well below the threshold commonly considered for cross-reactivity. While the main vicilin/albumin domain of Ara h 1 and Jug r 2 are recognized by both peanut and tree nut-allergic patient IgE,^17,18 our work represents the first time that the cross-reactivity potential of the independent VBP domains has been specifically assessed. The structural homology provided by the VBP motif could potentially facilitate cross-reactivity across other plant species. Indeed, comparing the peanut/walnut VBP structures against those of its tomato (VBP-8) and buckwheat (BWI-2c) homologues reveals a high degree of similarity, with Σ[ΔSIM] values comparable to those observed among the peanut/walnut VBPs (Figure S2). A notable exception is JR2.3, whose comparatively low Σ[ΔSIM] values might account for the absence cross-reactive peptides observed in this study. Similarly, the high Σ[ΔSIM] score observed for JR2.2 belies its scarcity of IgE-reactive peptides relative to JR2.1. It should be noted that the cross-reactive α-hairpin structure is mediated primarily by intermediate and long-range disulfide bonds, suggesting that IgE recognition is mediated by conformational rather than linear epitopes: a conjecture supported by the localization of IgE binding to the structured α-hairpin regions of the VBP’s tested (Figure 5). The drastic loss in secondary structure observed under reducing conditions suggests that the peptide microarray may not fully represent these conformational epitopes, potentially accounting for some of the discontinuities in the peptide mapping, or the aforementioned discrepancies concerning JR2.2 and JR2.3. Nonetheless, the structural similarities observed between the various VBP α-hairpins reported in this work and elsewhere, coupled with the high (up to 100%) rate of cross-reactive PN IgE binding observed in JR2.1, suggests that the contribution of these shared α-hairpins to cross-reactivity of these sequences warrants further investigation.

The α-hairpin VBP structure also bears striking similarity with other cysteine-rich seed storage proteins, potentially allowing for cross-reactivity with other allergen families. Indeed, previous studies have employed bioinformatics approaches, including the SDAP PD search tool described previously, to identify several potential peanut/tree-nut cross-reactive sequences on 2S albumins, 7S globulins, 11S globulins, and VBPs based on their biophysical similarity to the immunodominant epitope of Ara h 2, a conjecture which was verified by subsequent microarray studies.^17,18 Additionally, antibodies raised against a 13-mer consensus (13cp) sequence derived from these multiple alignments was able to successfully recognize both peanut and walnut VBPs, along with their globulin and albumin counterparts.¹⁸ Curiously, a PD search comparing the tomato and buckwheat VBPs shown in Figure S2 against this same 13cp reveals comparable scores to the peanut/walnut VBPs, suggesting that this mode of cross-reactivity is applicable to other food allergen sources beyond the peanut and walnut sequences described in this work. Finally, it should also be noted that IgE-reactive VBPs have also been identified in in almonds (Pru du 8),⁴⁶ cashew (Ana o 1),⁴⁷ pistachio (Pis v 3),^48,49 sesame (Ses i 3),⁴⁹ and macadamia nut (Mac i 1).⁵⁰ While the structure of these IgE-reactive VBPs have yet to be solved, their sequences share similar PD scores when compared against the 13cp as their peanut and walnut homologues. Taken together, these findings establish VBPs as a potential family of pan-allergens whose disulfide-bridged α-helical structure contribute to both allergenicity and cross reactivity across a wide range of species and protein families.

Stability is frequently cited as a biophysical property that can differentiate allergens from non-allergens. A systematic investigation into its role in the sensitization process is complicated by the various definitions of protein stability (eg: resistance to pressure, heat, chemical modification etc.), and the potential contribution of other physiochemical phenomenon such as aggregation, ligand binding, and post-translational modification. Nonetheless, there are several works which demonstrate this trend on both the individual protein and whole-proteome level.^39,40,51,52 With regards to food allergens, stability is proposed to enhance resistance to gastrointestinal digestion, facilitating exposure of the intact antigen to the immune system following oral consumption.⁴² However, recent work suggests that sensitization can also occur via cutaneous or airway exposure, potentially accounting for the low gastrointestinal resistance of profilins and PR-10 allergens.^39,53,54 It should be noted the walnut and other plant VBP’s naturally adopt a structured, disulfide-bonded α-hairpin motif when extracted from their natural source material, indicating that this represents the most relevant species in the context of gastrointestinal digestion.^9,19 In this form, no correlation was observed between gastrointestinal digestion and IgE reactivity, suggesting that non-oral exposure avenues may play a role in VBP sensitization.

Stability can also be defined by the ability to resist endosomal degradation by antigen-presenting cells. The data herein suggests that resilience under these conditions may be a better predictor of VBP allergic potential. This is consistent with a growing body of work indicating that the kinetics of endosomal digestion, epitope generation, and MHCII presentation can influence the nature of the downstream immune response against a wide range of allergens.^43–45 Under this “stability hypothesis”, high but transient MHCII levels facilitating asymptomatic tolerance while a more persistent presentation regime promotes allergic sensitization.^43–45 This gives rise to a bimodal relationship between stability and allergenicity. Here, proteins with intermediate stabilities tend to be asymptomatic, as the majority of their epitope fragments are generated within the late endosome when MHCII loading and presentation is most effective, i.e. ‘normal’.^55,56 In contrast, proteins with above average stabilities are more likely to generate an allergic response as peak epitope generation occurs after the window for optimal MHCII loading (Figure 7).^57–59 However, proteins which stray too far toward either extreme are unlikely to generate an immune response of any kind. This phenomenon has been observed in hypoallergenic variants of the birch pollen allergen Bet v 1⁵¹ or non-allergic Amb a 1 (ragweed) homologues.⁶⁰ The observations presented in this work indicate that AH1.1 and JR2.1 lie in the stability regime required for allergic sensitization (Figure 7) while the elevated stability of JR2.2 and JR2.3 inhibits the generation of T-cell epitopes on a reasonable timescale, limiting their allergic potential. These findings provide valuable insight into possible proteomics tools to differentiate allergic VBPs from their non-allergic counterparts and raises the possibility for novel immunotherapeutic strategies centered around destabilized VBP mutants which fall into the intermediate proteolysis regime, increasing the likelihood of a tolerogenic response (Figure 7).

Figure 7:

Proposed model of VBP allergenicity. Following uptake by antigen presenting cells, potential allergens are subjected to endosomal degradation to generate T-cell epitopes. Proteins with an intermediate stability are primarily digested in the late endosome, resulting in efficient MHCII loading and a tolerogenic (IgG/IgG₄) immune response as reviewed by Foo et al.⁵⁹ Proteins with above-average stability experience peak epitope generation in the lysosome, increasing the probability of an allergic (IgE) response due to the scarcity of MHCII. Hyper-stabilized proteins are completely resilient to endosomal digestion and yield only minimal levels of epitope fragments over the endosomal lifecycle, resulting in a diminished immune response. The results reported in this work suggest that extremely immunogenic VBPs such as AH1.1 and JR2.1 belong to the former, while hyper-stabilized homologues such as JR2.2 and JR2.3 fall into the latter category, contributing to their reduced sensitizing potential.

Together, the results suggest that plant VBP’s represent an important family of potential pan-allergens whose conserved α-hairpin fold allows them to mediate cross-reactivity across evolutionarily distant plant species and protein families. The unique bisulfide-mediated architecture of the VBP motif hinders attempts to assess immunogenicity using traditional bioinformatic approaches. Instead, stability in the context of endosomal degradation was identified as a potential predictor of allergenicity within the VBP family, though future work is required to confirm these findings in a clinical setting. Nonetheless these findings provide valuable insight into the mechanism of sensitization and potential proteomic tools to identify and assess allergic VBP’s in future works.

Abbreviations Used:

AEP

asparaginyl endopeptidase

CD

Circular Dischroism

ΔSIM

Surface similarity score

DTT

Dithiothreitol

HSQC

Heteronuclear Single Quantum Coherence

LS

Leader sequence

MHCII

Major Histocompatability Complex, Class 2

NMR

Nuclear Magnetic Resonance

NOE

Nuclear Overhauser Effect

NOESY

Nuclear Overhauser Effect Spectroscopy

PD

Property Distance

SDAP

Structural Database of Allergenic Proteins

TCEP

tris(2-carboxyethyl)phosphine

VBP

Vicilin-buried peptide.