Hazelnut allergome overview and Cor a gRNAs identification

Abstract

Background

Corylus species (hazelnuts) are a valuable source of nutrients and are widely consumed worldwide. Nevertheless, Corylus avellana (Cor a) contains 13 allergens (Cor a 1, Cor a 2, Cor a 6, Cor a 8, Cor a 9, Cor a 10, Cor a 11, Cor a 12, Cor a 13, Cor a 14, Cor a 15, Cor a 16, and Cor a TLP) that have been deposited into the official database (WHO/IUIS) for allergen nomenclature. The recent availability of several Corylus genomes provided opportunities to explore allergome variability, and thus to develop hypoallergenic varieties using modern biotech approaches. Certainly, the identification of CRISPR-Cas9 guide RNA (gRNA) is a pivotal step in achieving this goal. User-friendly web tools include limited reference genomes to design CRISPR-Cas9 gRNAs, while bioinformatic software for custom analysis require advanced command-line skills.

Results

This work explored the evolutionary trajectories of allergenic Cor a homologs in C. avellana, C. americana, C. heterophylla, and C. mandshurica genome assemblies. 52 Cor a orthologs were found in the analyzed species, and a recent tandem duplication of Cor a 1 was found in C. americana. Three new gene models were predicted in C. avellana and C. mandshurica for Cor a 16 and Cor a 10. Additionally, we identified 56 Cor a isoallergens, of which ten Cor a isoforms. Furthermore, phylogenetic analysis sheds light on the evolutionary dynamics of three hazelnut allergens revealing the evolutionary complexity of Cor a 1, Cor a 2, and Cor a TLP within the Corylus genus. A list of multiple gRNAs designed for the CRISPR-Cas9 system was provided for the singular and multiple silencing of Cor a homologs in each Corylus genome.

Conclusions

This study enhances our knowledge on the evolutionary path of Cor a allergens among Corylus species and provides highly accurate on-target guides targeting hazelnut allergome.

Background

Hazelnuts, scientifically known as Corylus spp., belong to the family Betulaceae and are a source of valuable nutrients, which makes them one of the most nutritive among nuts [1]. Because of its nutritional properties, the worldwide economic relevance of hazelnuts reaches 2.3 billion (USD) of exports per year, with European countries considered the main importing countries [2]. Lamentably, hazelnuts also represent the main culprit of nut allergy in Europe, which could be justified by their high consumption rate [3]. The National Institute of Allergy and Infectious Diseases (NIAID) describes food allergy (FA) as an IgE-mediated response of the immune system to specific food proteins [4]. This process triggers the release of potent inflammatory mediators from mast cells and basophils in sensitive subjects, causing itching, hives, swelling, and, in severe cases, anaphylactic shock [5]. The allergens found in the Corylus genus have been extensively researched in C. avellana, leading to the identification and characterization of several hazelnut proteins [6]. So far, hazelnut allergens found in C. avellana, are Cor a 1 (PR-10), Cor a 2 (Profilin), Cor a 6 (Isoflavone reductase), Cor a 8 (Non-specific lipid transfer protein, nsLTP), Cor a 9 (11 S globulin), Cor a 10 (Heat shock protein, HSP), Cor a 11 and Cor a 16 (7 S globulins), Cor a 12, Cor a 13 and Cor a 15 (Oleosins), Cor a 14 (2 S albumin), which have been included in the Allergen nomenclature (WHO/IUIS) database (https://allergen.org/), and Cor a TLP (thaumatin like-protein) (not included). Cor a 1 is related to mild local hypersensitivity reactions and is heat labile [7], while the risk of systemic reactions increases with seed storage proteins (SSPs) Cor a 9, Cor a 14, and Cor a 11, and the lipid transfer protein (LTP) Cor a 8 [8,9,10,11]. Otherwise, Cor a 2, Cor a 6, and Cor a 10 are most abundant in pollen, and it is thought that sensitization to these proteins primarily occurs at the mucosal membranes of the respiratory system, leading to localized symptoms [12]. Cor a 12 and Cor a 13, classified as oleosins, present an unclear clinical significance, while Cor a 15 has been classified as a new allergen in pediatric patients [13]. More recently, Cor a 16 was classified as a new hazelnut allergen [14]. Moreover, Cor a TLP described by Palacín et al. [15], showed cross-reactivity with sera from fruit-allergic patients, which present high sequence similarity with allergens Mal d 2 (apple) and Pru p 2 (peach) [12]. Modern biotechnological approaches such as CRISPR-based genome editing have revolutionized the field of plant breeding, offering precise and efficient tools for targeted modifications. The most popular CRISPR system utilizes the Cas9 nucleases and requires the design of highly efficient guide RNAs (gRNAs) to successfully generate mutated plants [16]. However, so far, CRISPR-Cas9 in hazelnut is impeded by the lack of efficient regeneration protocols and user-friendly online tools integrating hazelnut genomes for gRNA design [17]. Indeed, currently available tools for custom gRNA design require bioinformatics knowledge and command-line skills. Therefore, the design of efficient gRNAs could provide a further step toward the obtainment of hypoallergenic hazelnut crops [18]. Furthermore, the sequencing of several Corylus genomes in recent years [12, 19,20,21,22,23,24] allows the exploration of the variability of allergenic gene families and the identification of efficient on-target gRNAs. The task to specifically predict the gRNAs of Cor a genes directly within Corylus genomes of studied taxa has not yet been addressed. This work encompasses the comprehensive identification of hazelnut allergens in four different Corylus species, aiming to investigate the evolution events that occurred along the Corylus genus speciation. Moreover, we leverage computational analysis for the generation of efficient gRNAs for all 13 Cor a genes and relative orthologs and paralogs for singular and simultaneous targeting of allergenic isoforms.

Methods

Taxa data set

To investigate the evolution of Cor a genes along the Corylus plant species, we used the genomic data of four taxa (Corylus avellana, Corylus americana, Corylus heterophylla, and Corylus mandshurica) [12, 22,23,24]. The genome sequence, gene model annotation, and protein sequences were downloaded from three (Phytozome, GigaDB, and GenBank) public databases (Supplementary Table 1). Moreover, for comparative purposes, we also added 12 well-characterized reference Cor a genes (Cor a 1, Cor a 2, Cor a 6, Cor a 8, Cor a 9, Cor a 10, Cor a 11, Cor a 12, Cor a 13, Cor a 14, Cor a 15 and Cor a 16 (Supplementary Table 2) retrieved from the WHO/IUIS Allergen Nomenclature Database (http://allergen.org, accessed on 1 February 2024), and Cor a TLP [12].

Identification and manual curation of Cor a paralogs and orthologs

In order to retrieve predicted Cor a genes, we scanned the C. avellana, C. americana, C. heterophylla, and C. mandshurica proteome data set (Supplementary Table 1) with a well-known algorithm for inferring orthologs based on all-versus-all sequence comparison with BLAST [25], implemented in InParanoid (v4) [26] with default parameters (Supplementary Tables 3–5). We used a confidence score threshold of 1 to directly estimate gene–gene correspondences between the Cor a proteins. Cor a protein domain architecture was annotated using InterProScan [27] and Conserved Domain search [28] with default parameters. Corylus orthogroups were identified by Orthofinder v2.3.3 [29], using the default settings. To improve the gene prediction procedure for Cor a 10 (GenBank ID: AJ295617) and Cor a 16 (GenBank ID: ON060881) gene loci in C. mandshurica and in C. avellana genome assemblies, we used the AUGUSTUS-PPX (Protein Profile eXtension) module implemented in Augustus (v2.5.5) pipeline, with Arabidopsis parameters [30]. AUGUSTUS-PPX analysis used amino acidic profiles of conserved, ungapped blocks from the full-length protein alignments of Cor a 16 and Cor a 10 orthologs. The predicted genes encoded proteins with a confidence score > 0.

Multiple sequence alignment and phylogenetic analysis

Sequence similarities were determined by performing a ClustalW (v2) [31] multiple alignments with default settings, using the full-length protein sequences of annotated Cor a proteins as input. Phylogenetic analysis was performed by using newly identified Cor a homologs. The dataset was completed with the following proteins: Cor a 1, Cor a 2, Cor a 6, Cor a 8, Cor a 9, Cor a 10, Cor a 11, Cor a 12, Cor a 13, Cor a 14, Cor a 15, Cor a 16 and Cor a TPL (Supplementary Table 2). Evolutionary relationships between Cor a proteins were inferred using the maximum likelihood method based on the Whelan and Goldman model [32], using the MEGA6 software (http://www.megasoftware.net, accessed on 1 February 2024) [33]. The model with the lowest BIC (Bayesian Information Criterion) score was considered to describe the substitution pattern. The bootstrap consensus tree, obtained from 100 replicates, was taken to represent the Cor a family phylogenetic history [34].

Estimation of average evolutionary divergence

The evolutionary divergence (π) was estimated as number of base substitutions per site from averaging over all sequence pairs. Standard error was obtained by a bootstrap procedure (100 replicates). Analyses were conducted using the Maximum Composite Likelihood model [35]. All positions containing gaps and missing data were eliminated. Selective pressures acting on the Cor a TLP genes were investigated by determining the nonsynonymous to synonymous nucleotide substitution (dN/dS) indicated as ω. Tests were conducted to estimate the evolution of each codon: positive (dN > dS); neutral (dN = dS); and negative (dN < dS). Analyses were conducted using the SLAC algorithm [36]). The Tajima’s D test is to distinguish between genes evolving randomly (“neutrally”) and those evolving under a non-random process, including directional selection or balancing selection and, gene family expansion or contraction using the MEGA6 software [37]. All the Cor a TLP -coding DNA sequences were aligned using ClustalW [31].

Single- and multi-targeting guide RNAs (gRNAs) design

Single- and multi-targeting guide RNAs (gRNAs) for the CRISPR-Cas9 system were designed using CRISPR-Local, a high-throughput tool [38]. Reference gRNAs database (RD)-build model was used to build the whole-genome gRNA database from reference genomes of Corylus spp. and corresponding annotation files. The RD-build program was executed with -U 15 -D 3 settings (Supplementary Tables 6–9). The screening of all possible on-target gRNAs was realized with the Rule Set 2 algorithm [39]. The seqMap program was used to predict the effects of each off-target site with the highest cutting frequency determination (CFD) score for each gRNA [40]. All target and off-target data determined across the entire genome were exported into RD. Database (DB)-search model was used to obtain sorted results from all annotated Cor a genes. Paralogs (PL)-search model was used to output gRNAs targeting Cor a gene homologs (Supplementary Table 10).

Validation of prediction results

All gRNAs predicted in C. avellana were used to perform molecular validation. DNA was extracted from leaf tissue of genotype Corylus avellana cv Mortarella using DNeasy Plant mini kits (Qiagen, Valencia, CA, USA). PCR was executed with 25 ng of genomic or complementary DNA, 10 pmol primers, 1 U of Taq DNA polymerase Kit (Invitrogen, Carlsbad, CA, USA), 10 pmol dNTPs, and 2 mM MgCl2 in 25 µl reaction volumes. Amplification was performed using the following cycling conditions: 1 min at 94 °C, followed by 30 cycles of 1 min at 94 °C, 1 min 30 s at 60 °C and 2 min at 72 °C, with a final extension for 7 min at 72 °C. Amplicons were separated by electrophoresis on agarose gel (1.5%), and photographed by a GelDoc apparatus. Primers were designed with Primer3 (http://frodo.wi.mit.edu), with a length between 18 and 22 bp. The length of the amplified fragments and the Tm of the specific primers were reported in Supplementary Table 11. In particular, 13 out of 52 guides were used as forward primers in combination with two different reverse primers for on and off-targets in C. avellana, respectively (Supplementary Table 11).

Results

Allergenic genes annotation in four Corylus spp.

Thirteen Corylus avellana (Cor a) proteins were identified and reported as allergens triggering specific immunoglobulin E (IgE) reactivity (Supplementary Table 2). In order to identify their coding genes in the published genome assemblies of Corylus spp., we used the BLAST-based reciprocal best hits (RBHs) method. A total of 52 Cor a genes were discovered in four analyzed proteomes (Supplementary Tables 3, 4). Four Corylus species presented all the 13 Cor a orthologs located on nine chromosomes (1, 2, 3, 4, 5, 6, 8, 9, and 11) suggesting that these allergens were preserved along the speciation continuum (Fig. 1). Moreover, the average amino acidic identity among Corylus orthologs was more than 90% (Supplementary Table 3). A lineage-specific micro-synteny profile for all allergens was observed in C. avellana, C. americana and C. heterophylla. Differently, Cor a 8, Cor a 10, Cor a 13, Cor a 15, Cor a 16 and Cor a TLP in C. mandshurica were located on different chromosomes, as highlighted in Fig. 1. Two Cor a 1 ortholog genes (CamerRush.01G109100.1 and CamerRush.01G109000.1), with confidence score = 1, were identified in Corylus americana genome. Despite this, CamerRush.01G109100.1 showed a higher global identity (97.5%) to Cor a 1. Cor a 6 orthologs showed the highest identity (99.1%) among the four hazelnut species. While the wide diversity range was observed for Cor a 1 (90.2%) and Cor a 16 (90.4%) orthologs (Supplementary Table 3). Unexpectedly, RBH analysis identified no Cor a 16 locus among predicted C. avellana gene models [12]. Therefore, we executed an augustus-PPX prediction on C. avellana genome, and a new locus named Cav04g12115 with an amino acidic identity ~ 87% to Cor a 16 gene (ID: ON060881) was predicted (Fig. 2A). Similarly, we have reannotated Cor a 16 (ID: ON060881) and Cor a 10 (ID: AJ295617) loci in C. heterophylla and C. avellana genomes, respectively. Initially, very low identity values (< 70%) were detected between homology gene pairs Cor a 16: ON060881-EVM0010550 and Cor a 10: AJ295617-Cav03g20430 (Fig. 2B and C). Our predicted Cor a 16 (EVM0010550_2) in C. mandshurica was located on the same region of EVM0010550 gene model (Fig. 1B), while Cor a 10 (Cav03g20435) locus resulted longer than ~ 1.2Kb compared to Cav03g20430 gene model (Fig. 2C). Hence, coding DNA sequences of re-annotated exon-intron structures showed higher amino acidic identities (> 90%) to Cor a 10 and Cor a 16 compared to protein sequences released by consortia (Fig. 1B).

Fig. 1

Circos plot summarizes the genomic positions of hazelnut allergenic loci and their orthology relationships. Insert legend provides information for the data rings. (a) The 11 chromosomes of C. avellana, C. americana, C. heterophylla, and C. mandshurica genome assemblies. The length of each circle segment is proportionated to the size of chromosomes. (b) The gene IDs of allergens annotated in analyzed Corylus spp. (c) The orthology relationships between putative allergens, while red lines indicated no synthenic Cor a genes

Fig. 2

Agustus-PPX Cor a 16 and Cor a 10 gene prediction in two hazelnut genome assemblies. (a) Genomic region including Cor a 16 locus (Cav04g12115) on Corylus avellana chromosome 4. (b) Re-annotation of exon-intron structures erroneously predicted for Cor a 16 (EVM0010550) on Corylus mandshurica linkage group (LG) 4. (c) Re-annotation of exon-intron structures erroneously predicted for Cor a 10 (Cav03g20435) on Corylus avellana chromosome 3 (pchr = pseudo-chromosome; LG = linkage-group)

Isoforms and isoallergens identification

Plant genomes frequently host gene groups that have evolved from a common ancestor following gene duplication events. To discover potential ‘isoallergens’ and ‘isoforms’ of Cor a genes, we have extensively analyzed the groups of closely related proteins identified by OrthoFinder [29]. Isoallergens were defined as proteins from the same species that show at least 67% amino acid identity, while isoforms are considered variants of the same allergen, typically with > 90% identity [41]. A total of 56 isoallergens (out of orthologs) found in four Corylus species showed average protein identities ≥ 80% to the reference Cor a allergens (Table 1). Isoallergens were found for Cor a 1, Cor a 2, Cor a 9, Cor a 10, Cor a 13 and Cor a TLP; while, Cor a 6, Cor a 8, Cor a 11, Cor a 12, Cor a 14, Cor a 15, and Cor a 16 showed no paralogs. Among these, six Cor a 9 and four Cor a 13 proteins showed sequence identities higher than 93% and could be classified as isoforms (Table 1; Supplementary Table 5). Among the hazelnut isoallergens, major differences emerged for Cor a 1 and Cor a TLP, which stood out with higher gene copy numbers in C. avellana compared to the other species. By contrast, the number of Cor a 2 isoallergens was conserved among the four species.

Table 1 Hazelnut isoallergenic genes annotated in four Corylus species

Phylogenetic relationships between nut allergenic genes

Three Cor a loci (Cor a 1, Cor a 2, and Cor a TLP) were strongly duplicated in Corylus genomes. To retrace their evolutionary trajectories, we infer three phylogenetic analyses for ortholog and paralog loci. The evolutionary relationships were analyzed using a maximum likelihood method (Fig. 3). For comparative purposes, we included two well-characterized reference genes (Cor a 1 and Cor a 2), the Mal d 2 gene from Malus domestica (ID: XM008362484; Supplementary Table 2), and three out-group genes (Hor v 1, Hor v2 and Hor v TLP) identified in Hordeum vulgare (L.) (Fig. 3). A total of 13, 12, and 33 homologs to Cor a 1, Cor a 2, and Cor a TLP proteins, respectively, were aligned. The sequences were grouped into phylogenetic clades supported by bootstrap values ≥ 50%, allowing the definition of three, three, and four clades for Cor a 1, Cor a 2 and Cor a TLP allergens, respectively (Fig. 3). Clade 1 of Cor a 1 orthogroup showed two homologs in C. avellana, C. mandshurica, and C. americana (Fig. 3A). Interestingly, no C. heterophylla genes were included in clade 1, and the most similar to Cor a 1 (EVM0015794.1) was collapsed in clade 2. Moreover, no C. mandshurica genes were clustered in clade 3 (Fig. 3A). The Panel B of the phylogenetic tree provides insights into the evolutionary relationships among Cor a 2 homologs. A total of 12 genes were included in this analysis, and in each clade, a member of each species was grouped (Fig. 3B), suggesting gene duplication events probably occurred before Corylus speciation. Finally, panel C summarizes the evolutionary analysis of homologs to Cor a TLP. As expected, Mal d was significantly separated from hazelnut phylogenetic clades (Fig. 3C). Interestingly, a total of 18 Mal d homologs were found in C. avellana, of which 17 were distributed in a short genomic region of chromosome 9 (~ 112 kb) (Fig. 2C; Supplementary Fig. 1). Similarly, C. americana showed eight homologs distributed among all the clades. Differently, C. mandshurica and C. heterophylla homologs were included in three clades.

Fig. 3

Natural diversification of the Cor a 1 (panel a), Cor a 2 (panel b), and Cor a TLP (panel c) orthogroups identified in C. avellana, C. americana, C. heterophylla and C. mandshurica genomes. The evolutionary history of 58 Corylus genes was used together with three well-characterized allergenic genes to perform a maximum likelihood analysis. Labels showing the bootstrap values higher than 50 (out of 100) are indicated above the branches. The taxa to which the protein sequences belong are indicated by colored spots

Selection pressure acting on Cor a TLP genes

To infer the direction and magnitude of natural selection acting on Corylus TLP genes, the non-synonymous substitution (dN) and synonymous substitution (dS) ratio was used. Neutrality tests, performed on the Cor a TLP gene families and TLP phylogentic clades, yielded average Tajima’s D and ω (dN/dS) values ranging from 0.669 to − 1.226 and from 0.552 to 0.102, respectively (Table 2). Single codon analysis highlighted the presence of 6, 10, 13 and 24 negatively selected sites in the C. heterophylla, C. mandshurica, C. americana and C. avellana Cor a TLP families, respectively (Table 2). Interestingly, the negative Tajima’s D value found for the TLP family of C. avellana confirms the low frequency of polymorphisms and a gene family size expansion (Table 2). Globally, a restricted evolutionary divergence was observed into the phylogentic clades, indeed the average nucleotide diversity (π) ranged from 0.019 to 0.152 (Table 2). Finally, Tajima’s highest negative D values were found for C1 (–1.035) and C4 (–1.226) clades (Table 2), in which the highest number of paralogs were revealed. Hovewer, confirming low frequency of polymorphisms, indicating gene family size expansion.

Table 2 Estimation of Non synonymous and synonymous substitutions mean dissimilarity (ω = dN /dS ) for each Corylus TLP gene family and phylogentic clade

Identification of single and multiple gRNAs for Cor a allergens

CRISPR-Cas9 genome engineering offers a real opportunity for developing hypoallergenic cultures, and gRNA design is a pivotal step. To provide high-throughput gRNAs considering the genetic variations among Corylus genomes with different genetic backgrounds, we designed gRNAs with the highest CFD score for Corylus sp. allergenic loci [38]. For instance, the gRNAs targeting Cor a 1 allergens in C. avellana, C. heterophylla, and C. mandshurica showed high on-target scores (0.746, 0.758, 0.813, respectively), indicating a strong potential for targeting this allergen. In contrast, the gRNA sequence in C. americana had a slightly lower on-target score (0.686). For Cor a 2, all the analyzed species had similar on-target scores (0.705, 0.695, 0.678 and 0.629). However, the off-target scores for C. americana and C. heterophylla were barely higher (0.017 and 0.016). Focusing on the off-target scores, the lowest values were found for the Cor a 11 in C. avellana and C. heterophylla, Cor a 6 in C. heterophylla, and Cor a 13 in C. americana (off-target value: 0). However, we found gRNA sequences with relatively low off-target value for several genes (Supplementary Tables 6–9). Indeed, the highest off-target value was related to the gRNA identified on Cor a 1 allergen in C. avellana (0.26), while all the other gRNA sequences showed lower off-target values. Moreover, the gRNA sequences for Cor a 6, Cor a 8, Cor a 9, Cor a 10, Cor a 11, Cor a 12, Cor a 13, Cor a 14, Cor a 15, and Cor a TLP showed relatively consistent on-target scores among the four species (all above of 0.62), indicating a comparable potential for targeting these allergens.

The gRNA specificity was molecularly validated for predicted C. avellana guides (Supplementary Table 7). Revealed amplicons on target genes confirmed gRNA-matching specificity (Supplementary Fig. 2). Finally, we identified multiple CRISPR-Cas9-gRNAs for silencing species-specific isoallergens (Supplementary Table 10). This allowed the selection of seven gRNAs targeting 15 Cor a paralogs. In particular, we identified common gRNAs for the simultaneous silencing of all Cor a 13 and two Cor a 1 homologs in C. heterophylla (Supplementary Table 10). Similarly, Cor a 9 homologs identified in C. avellana and in C. americana could be simultaneously targeted using a common gRNA (Supplementary Table 10). Finally, three gRNAs were identified in C. mandshurica to simultaneously silence all Cor a 1, Cor a 9 and Cor a 13 homologs (Supplementary Table 10).

Discussion

Genomes of hazelnut species Corylus avellana, Corylus americana, Corylus heterophylla, and Corylus mandshurica were recently assembled and released, paving new ways for genetic research within this genus [12, 22,23,24]. These species are particularly appreciated for their nutritional values [42], and genetic studies have given particular emphasis to the identification and characterization of hazelnut Cor a proteins, which have been described as the primary allergenic proteins responsible for IgE-mediated allergy induction [18]. Recently, CRISPR-Cas9 technology revolutionized plant breeding programs by enabling the direct manipulation of specific genes, enhancing speed and precision in obtaining improved genotypes [43]. However, so far, no studies have been reported about the application of CRISPR-Cas9 in hazelnuts, most likely due to the absence of efficient regeneration protocols and available user-friendly software for gRNA design [17, 44]. This lack of knowledge and resources hampered progress in exploring hazelnut genetic modifications for developing hypoallergenic crops [18].

In light of the limited genome editing studies and challenges associated with available bioinformatics tools for gRNA design [45, 46], the investigation of evolutionary relationships among allergenic Cor a genes in four Corylus species and identification of gRNAs lists for targeting Corylus allergens individually and in combination represent a further step towards genetic manipulation and biotechnological-mediated breeding in hazelnut species [43]. The conservation of Cor a genes throughout the evolutionary divergence of the Corylus species, was initially confirmed with the identification of 52 Cor a genes across C. avellana, C. americana, C. heterophylla, and C. mandshurica genomes. Moreover, gene duplication of Cor a 1 was identified in C. americana, suggesting a recent and potentially functional duplication event [47]. Notably, we identified the Cor a 16 locus on chromosome four in C. avellana cv. ‘Tumbol’ named Cav04g12115, which was unpredicted by Lucas et al. [12]. Meanwhile, new exon-intron structures were predicted for Cor a 16 in C. mandshurica and Cor a 10 in C. avellana. These newly predicted genes and structures [48] expand our knowledge of allergen genetic positions and clarify the structural organization of allergenic genes, enhancing the understanding of hazelnuts’ allergenic profile, which is a crucial step for accurately characterizing gene functional domains for future genome editing studies [49]. Moreover, the investigation of orthologous relationships and isoallergenic gene variations across different Corylus species highlighted high amino acidic sequence similarities among identified hazelnut allergens, while sequence variations could be attributed to the divergent speciation path [46, 50]. Understanding gene relationships and sequence variations enhances insights into the genetic basis of hazelnut allergenicity and potential cross-reactivity sites. Furthermore, the presence of hazelnut isoallergens with varying copy numbers across the genomes suggested genetic diversity in the allergenic composition of different Corylus species [51]. For instance, Cor a 1 and Cor a TLP exhibited differences in the number of isoallergenic genes. Cor a TLP showed almost the double copy number in C. avellana compared to the other Corylus genomes. In accordance with Lucas et al. [12], we inferred that the Cor a TLP was encoded by Cav09g00740. These proteins are members of pathogenesis response proteins-5 (PR-5), which are known to be involved in plant hypersensitive response to infections and, due to their high sequence similarity, may all be considered as potential allergens in hazelnuts [15, 52, 53]. Indeed, almost all (14 out of 15) of the Cor a TLP homologs formed a gene cluster spanning a genomic region of ∼112 kb of chromosome 9, originating by tandem gene duplications [54].

Hence, the genetic diversity and evolutionary history of Cor a 1, Cor a 2, and Cor a TLP allergens have been further elucidated across the different Corylus species by the phylogenetic analysis, which revealed evolutionary relationships among the four hazelnut genomes. By employing a phylogenetic reconstruction, we identified complex relationships between these genes and the established isoallergen nomenclature [55]. Three maximum likelihood analyses collapsed the 58 Cor a proteins into ten phylogenetic clades [12]. Our findings indicated that C. avellana and C. mandshurica showed multiple homologs closely related to Cor a 1, while in C. heterophylla orthologs to Cor a 1 showed a more divergent evolutionary path [56]. By contrast, the analysis of Cor a 2 homologs revealed species-specific patterns, with each species having representative genes in different clades, indicating unique evolutionary trajectories [57]. We can speculate that gene duplication events of the Cor a 2 homologs preceded the speciation events among the four species. Furthermore, the investigation of Cor a TLP orthologs demonstrated variations in gene numbers and distribution among the species, with tandem duplications observed on chromosome 9 in C. avellana, indicating the functional relevance of Cor a TLP in plant defense and potentially in allergenic induction [12, 58]. Therefore, despite the distinct evolution paths, all identified orthologs conserved high sequence similarities and should be considered as potential allergens. Indeed, the understanding of Cor a sequence similarities provides insights into the evolutionary history of hazelnut allergenic genes across different Corylus species, shedding light on the genetic relationships, divergence, and conservation of these allergenic molecules. The ratio of nonsynonymous (Ka) to synonymous (Ks) nucleotide substitutions among coding DNA sequences of phylogenetic clades was below 1. This indicates that negative selection has been acting against extreme polymorphic variants and in favor of purifying selection [59]. Moreover, Tajima’s negative D values of phylogentic clades grouping highest number of Cor a TLP paralogs supported the hypothesis of recent gene duplication events [60].

Considering the limited research focusing on the specific application of CRISPR-Cas9 for targeting hazelnut allergens, we identified single and multi-targeting gRNAs for Cor a allergens across different Corylus species, contributing to addressing the underexplored field of CRISPR-Cas9 targeting hazelnut allergens. By employing CRISPR-Local, a high-throughput tool for gRNA design, we generated a comprehensive list of Cor a gRNAs for all analyzed genomes. Our analysis prioritized gRNAs based on their on-target and off-target scores. The screened gRNAs showed an on-target score > 0.66, with a putative cutting efficiency ranging from 50 to 75% [36, 55]. In general, identified gRNAs showed high potential for targeting Cor a allergens across the species [38]. The importance of utilizing gRNAs with low off-target effects is widely known [61,62,63,64]. Here, most of the identified gRNAs showed low off-target values, suggesting a low risk of off-target effects for the majority of gRNAs. Like most algorithms, the cut specificity score is not always a perfect predictor of off-target effects. As a general rule, however, the higher the CFD score, the more likely that a guide matches an off-target site [36]. Our work provides gRNAs without off-target sites or with low off-target scores, as also established in other studies [55, 60, 65]. In our work, we also provide multiple gRNAs designed to target paralog genes [66]. The analysis led to the selection of different gRNAs capable of targeting fifteen Cor a paralogs that could be used to edit multiple genes in Corylus species, as also evidenced in recent studies employing CRISPR-Cas9 multiplex genome editing to target simultaneously different allergenic genes and obtaining less allergenic plants [67]. Interestingly, in Corylus mandshurica, a gRNA was designed to target three different Cor a 1 paralogs, demonstrating the feasibility of targeting multiple genes simultaneously [66, 68]. Most of the detected common sequences exhibited high on-target scores, indicating a potential high efficiency in targeting the selected genes. Despite the absence of efficient regeneration protocols, recent studies have successfully demonstrated the use of Agrobacterium-mediated transient transformations for expressing the CRISPR-Cas9 system in plants, leading to efficient genome editing. In particular, Hamada et al., [69] reported a protocol to perform genome-editing in plants, without requiring callus culture and regeneration through the biolistic delivery of CRISPR-Cas9 expression cassette in embryos. Moreover, other studies generated stable genome-edited plants using particle bombardment in embryos demonstrating an efficient method to obtain CRISPR-Cas9 edited plants in species reluctant to callus culture [70]. Furthermore, the recent application of CRISPR-Cas9 system in nut species such as walnut [71] indicates a growing trend toward overcoming challenges in nut-related species. Therefore, our findings highlight the importance of in-depth research for efficient gRNA design for the CRISPR-Cas9 system, paving the way for future research to develop hypoallergenic hazelnut varieties through targeted gene editing.

Conclusion

Our findings elucidate Cor a genes evolutionary divergence and similarities within the Corylus genus, and provide insights into the genomic basis of hazelnut allergome. This research contributed to discovering new potential allergens in Corylus, pointing out new Cor a orthologs and paralogs. Furthermore, gRNAs identified in this study could be considered ideal candidates for targeting single and multiple Cor a homologs. In conclusion, this research is a first step for developing potential hypoallergenic Corylus varieties.