Genome stabilization in the neotetraploid wasabi (Eutrema japonicum): subgenome dominance and extensive chromosomal rearrangements

Abstract

Wasabi (Eutrema japonicum), a commercially important crop in the Brassicaceae family, is prized for its unique pungent flavor, yet its genome evolution and polyploid origin remain poorly understood. This study presents a detailed evolutionary analysis of the wasabi genome. The diploid progenitors of E. japonicum diverged approximately 2 MYA, while the allotetraploid hybridization event that formed E. japonicum occurred more recently (~ 0.268 MYA). Following this whole-genome duplication, its genome stabilized through extensive chromosomal rearrangements and subgenome dominance. We also characterized transposable elements, including long terminal repeat retrotransposons (LTR-RTs), and analyzed gene expression patterns between subgenomes to further understand genome evolution and functional differentiation in E. japonicum. Notably, wasabi's distinct flavor is linked to a unique glucosinolate gene profile. This profile is characterized by a significant expansion of the epithiospecifier-modifying protein (ESM1) gene family, which promotes pungent isothiocyanate formation, and a contraction of the competing epithiospecifier protein (ESP) gene. The resulting high ESM1:ESP gene copy ratio is a key contributor to wasabi's characteristic properties. This study offers crucial insights into wasabi's genomic evolution, establishing a foundation for future research into the genetic basis of its unique flavor and other valuable traits.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-026-08635-2.

Introduction

Eutrema japonicum, also known as wasabi or Japanese horseradish, is a perennial plant belonging to the Brassicaceae family [1]. It is mainly cultivated in East Asian countries, including South Korea, China, and Japan, and known for its unique pungent taste. With wasabi’s expanding use as a seasoning in diverse culinary applications, the demand for this crop has experienced a significant increase recently [2]. However, wasabi cultivation presents significant challenges, requiring highly specific conditions. These include a consistently cool temperature range of 8–18 °C, a constant supply of fresh, running water, shade from direct sunlight, and a lengthy maturation period of at least two years [3]. Given the stringent cultivation requirements of wasabi, developing robust varieties that are resilient to environmental stress and that produce high yields is crucial. A thorough understanding of genomic information, including gene function and genetic diversity, is valuable for the development of such improved cultivars.

Polyploid crops have been suggested to exhibit higher yields, improved product quality, and increased tolerance to biotic and abiotic stresses over diploids [4, 5]. These phenomena, particularly common in the Brassicaceae family, highlight valuable agricultural characteristics of this family [6]. Allopolyploids, hybrids formed from different species, face significant genomic challenges during their evolution, sometimes called "genome shock." Hybridization is followed by widespread genetic alterations, and genomes that survive these alterations appear to achieve stabilization that maintains integrity [7]. The subgenome dominance hypothesis addresses one dimension of this stabilization, suggesting that one subgenome tends to become dominant, helping to mitigate epigenetic and genetic conflicts [8]. However, the exact mechanisms and long-term consequences of subgenome dominance in polyploid evolution are still not fully understood.

The pungent taste of wasabi originates from isothiocyanates, not produced as direct secondary metabolites in plant tissues but formed through the hydrolysis of glucosinolate precursors. When plant tissues are damaged, myrosinase and glucosinolates are released from their respective cellular compartments. Myrosinase then initiates the breakdown of glucosinolates into various degradation products [9]. Glucosinolates are largely responsible for the characteristic pungent taste of most plants in the Brassicaceae family, but the genetic variations underlying differences in pungency and flavor among different Brassicaceae plants remain unclear. Therefore, research on glucosinolate genes in wasabi, which exhibits particularly strong pungency even within the Brassicaceae family, can provide a fundamental basis for future studies investigating the relationship between glucosinolates and pungency levels in Brassicaceae plants.

In this study, we analyzed the evolution of subgenomes and glucosinolate biosynthetic genes, utilizing the wasabi genome (794.6 Mbp) that we previously assembled [10]. First, we investigated the formation of the allotetraploid wasabi genome. We reconstructed the ancestral karyotype for the genus Eutrema and investigated how the two subgenomes hybridized, as well as the functional roles of the two subgenomes in the nascent allotetraploid. Second, we studied the unique evolutionary and distributional patterns of glucosinolate genes in wasabi compared to other members of the Brassicaceae family. This study provides new insight into the interaction between subgenomes and the evolution of the glucosinolate biosynthetic pathway.

Materials & methods

Phylogenetic analysis of the brassicaceae family

To investigate the evolution of the wasabi genome, we selected 12 representative plant genomes. The protein sequences of Eutrema salsugineum (NCBI accession number: GCF_000478725.1), E. japonicum (GCA_041074935.1), E. yunnanense (GCA_002933935.1), E. heterophyllum (GCA_002933915.1), Brassica napus (GCF_020379485.1), B. rapa (GCF_000309985.2), B. oleracea (GCF_000695525.1), Raphanus sativus (GCF_000801105.2), Arabidoposis thaliana (GCF_000001735.4), Sorghum bicolor (GCF_000003195.3), and Armoracia rusticana (FigShare accession number: FigShare.21780176.v2) were obtained (accessed July 17, 2024). However, as the protein sequences of E. yunnanense, E. heterophyllum, and E. salsugineum were not publicly available, gene prediction was performed in-house using BRAKER3 (v3.0.8) based on the assembled genomes. For homology-based prediction, protein sequences from the OrthoDB Viridiplantae database (accessed August 30, 2024) were collected and used to guide GeneMark-ETP (v1.0) [11]. For the overall evolutionary comparative analysis, OrthoVenn3 was utilized [12]. The phylogenetic tree was constructed using OrthoFinder (v2.5.4), with sequence alignment performed by MUSCLE (v5.2). Trimal (v1.5.0) was used to trim and refine the conserved sequences, and the phylogenetic tree was inferred using the maximum likelihood method with FastTree (v2.1). The JTT + CAT model was employed for phylogenetic analysis [13, 14]. Gene family contraction and expansion were analyzed using CAFE5 (v1.1). To estimate divergence times, we used two calibration points based on the TIMETREE database (accessed September 9, 2024; http://www.timetree.org/) [15]. First, the divergence time between Sorghum bicolor and Arabidopsis thaliana was set to 160 million years ago (MYA). Additionally, we incorporated a secondary calibration point representing the divergence between B. napus and B. oleracea, which was estimated at 16.1 MYA.

Estimation of divergence time using Ks

To detect whole-genome duplications and to estimate divergence times, we employed the WGDI (v0.74) using synonymous substitution rate (Ks) values [16]. The genomes of E. japonicum and E. yunnanense were analyzed for this purpose. Ks-based age distributions for all paralogous genes were reconstructed from the genomes, simulating the evolution of coding sequences to recalculate synonymous distances. To estimate divergence times, the following formula [17] was used:

Long terminal repeat retrotransposons (LTR-RTs) insertion time estimation

We employed LTRharvest (v2.9.0) and TEsorter (v1.4.6) to identify and to classify LTR-RTs, respectively. Subgenome-specific elements were identified via k-mer enrichment analysis. Insertion times were estimated by calculating the nucleotide divergence (K) between LTR pairs using the Jukes-Cantor model [18] and applying the formula:

Estimation of the timing of hybridization

To estimate the timing of hybridization, we analyzed transposable elements (TEs) from both subgenomes and evaluated their divergence rates. The divergence of TEs in each subgenome was measured using PercDivs (percentage of substitutions in the aligned region compared to the consensus), which was calculated using RepeatMasker (version not available) [19]. In the case of the allotetraploid E. japonicum, the high degree of overlap in TE divergence between the two subgenomes indicates a consistent evolutionary rate across both. Regions where there was less overlap in TE divergence suggest distinct phases of genome divergence prior to their merger in the allotetraploid [20].

Reconstruction of a chromosomal-level genome assembly of E. yunnanense based on subgenome A of E. japonicum

To upgrade the previously published draft genome of E. yunnanense [21] to a chromosomal-level assembly, we employed RagTag (v2.1.0), using E. japonicum Subgenome A as the reference. RagTag scaffolds a draft genome by aligning the sequences from E. yunnanense to the reference genome [22]. Through this scaffolding process, the E. yunnanense draft genome was reorganized and refined, producing a much more accurate and structured genome assembly, comparable to a reference-level genome.

Karyotype reconstruction of Eutrema

Utilizing protein sequences from E. salsugineum, E. yunnanense, and the A and B subgenomes of E. japonicum, the ancestral Eutrema karyotype was inferred using the WGDI (v0.74) with the -ak parameter. Subsequently, the reconstructed ancestral sequences were mapped back onto each individual Eutrema genome using the -km and -k parameters in WGDI [16]. To investigate chromosomal rearrangements, we focused on E. japonicum and performed a detailed analysis to identify instances of nested chromosome fusion, end-end joining, and reciprocally translocated chromosome arms [16, 23].

Subgenome dominance

To investigate subgenome dominance in E. japonicum, we performed ortholog identification, gene expression analysis, and functional enrichment analysis. Orthologous gene pairs between the A and B subgenomes of E. japonicum were identified using OrthoFinder (v2.5.4) [24]. We focused on 13,525 one-to-one orthologous pairs, excluding multi orthologous pairs. Gene expression levels were quantified by mapping RNAseq reads to the E. japonicum genome using HISAT2 (v2.2.1) [25] and assembling transcripts with StringTie (v2.2.3) [26]. Gene expression was measured in transcripts per million (TPM) to account for sequencing depth and gene length. Homoelog expression bias was defined as a fold change ≥ 2 in transcripts per million between Homoelog genes. Gene Ontology (GO) enrichment analysis of genes exhibiting homoelog expression bias was conducted using the STRING database (v11.5, accessed November 17, 2024) to identify overrepresented functional categories [27].

Comparative genomic analysis of glucosinolate-coding genes

To investigate the genes involved in glucosinolate biosynthesis and breakdown, we initially retrieved 172 protein sequences from UniProt (accessed November 20, 2024). These sequences were curated from reviewed publications and validated in A. thaliana. Subsequently, the 34 core genes were selected from the initial set of 172 glucosinolate-related genes based on their roles in the methionine-derived aliphatic glucosinolate biosynthetic pathway and subsequent hydrolysis reactions that generate sinigrin and its breakdown products, including isothiocyanates and nitriles, which are responsible for the characteristic pungency of wasabi. We then conducted a synteny analysis by comparing these glucosinolate-related protein sequences to those from the following species: E. japonicum, E. yunnanense, E. heterophyllum, E. salsugineum, A. thaliana, A. rusticana, R. sativus, B. rapa, B. oleracea, and B. napus. Using DIAMOND (v2.1.10) [28], we performed comparisons with the following criteria: an e-value threshold of < 1 × 10^–6, identity ≥ 70%, and coverage ≥ 80%. Phylogenetic analyses of the glucosinolate-related genes were performed by aligning the protein sequences using MAFFT (v7) [29], constructing phylogenetic trees with FastTree (v2.1) [30], and visualizing the trees through the iTOL (v7) [31]. The physical map of glucosinolate-encoding genes was constructed using MG₂C (v2.1) based on their positions on the chromosomes [32].

Glucosinolate gene expression profiling across the Brassicaceae family

To investigate the expression patterns of the epithiospecifier modifying protein (ESM1) and epithiospecifier protein (ESP) genes across the Brassicaceae family, we analyzed RNAseq data from leaf tissues of ten representative species. The datasets included E. japonicum (NCBI accession number: SRX9404179), E. yunnanense (SRX3406420), E. heterophyllum (SRX3406409), E. salsugineum (SRX4795955), A. thaliana (ERX13395024), A. rusticana (SRX940417), R. sativus (SRX22367879), B. rapa (SRX27192209), B. oleracea (SRX27369770), and B. napus (ERX14095866). Sequencing reads were aligned to the respective reference genomes using HISAT2 (v2.2.1) [25], and transcript abundance was quantified using StringTie (v2.2.3) [26], yielding TPM values. Genes with no detectable alignment were excluded from downstream analysis. To obtain a more detailed view of gene expression in E. japonicum, we also analyzed tissue-specific RNAseq data from leaf (SRX9404181), stem (SRX9404180), and root (SRX9404179). Genes involved in glucosinolate biosynthesis were selected based on prior annotations, and expression patterns were visualized as a heatmap.

Results

Multiple lines of genomic evidence support a recent allotetraploid origin of E. japonicum

This study utilizes the chromosome-level reference genome for E. japonicum, which we previously assembled [10]. The allotetraploid genome of E. japonicum (2n = 4 × = 28) is composed of two highly colinear subgenomes (subgenome A and subgenome B), each with seven chromosomes. This high degree of colinearity is demonstrated by a whole-genome dot plot (Fig. 1a), the alignment of conserved syntenic blocks between homoeologous chromosome pairs (Fig. 1b), and a clear one-to-one macro-synteny map between the two subgenomes (Fig. 1c). First, we constructed a phylogenetic tree from ten species within the Brassicaceae family and S. bicolor as an outgroup (Fig. 1d). Our analysis indicated that the ancestors of E. japonicum diploid genome progenitors for subgenomes A and B diverged approximately 2.009 MYA (95% CI: 1.801–2.217 MYA). Subsequently, the progenitor of E. japonicum subgenome A and E. yunnanense diverged approximately 1.256 MYA (95% CI: 1.055–1.457 MYA).

Fig. 1.

Genomic architecture and evolutionary history of E. japonicum. a Dotplot of E. japonicum (ea: E. japonicum subgenome A; eb: E. japonicum subgenome B). The colors on the x and y-axes represent regions derived from the ancestral Eutrema karyotype. b Conserved synteny blocks on each chromosome of E. japonicum. The colors in blocks indicate the corresponding syntenic regions. c Macro-synteny relationships between subgenome A (top) and subgenome B (bottom) of E. japonicum. d A phylogenetic tree of 11 plant species, including ten species from the Brassicaceae family, using S. bicolor as the outgroup. Letters at each node indicate the estimated divergence time in million years ago (MYA). A timescale is displayed at the top of the figure. e Distribution of estimated insertion times for long terminal repeat retrotransposposons (LTR-RTs) specific to each subgenome. SubA: Subgenome A, SubB: Subgenome B. f Ks-based estimation of WGD and divergence times in E. japonicum, using the closely related diploid species E. yunnanense. Wasabi: E. japonicum, Yunn: E. yunnanense. g Estimation of the hybridization time of E. japonicum using TE divergence rates, calibrated by the divergence time of E. japonicum subgenomes A and B (2.009 MYA) using single-copy orthologs

Analysis of the insertion times of subgenome-specific long terminal repeat retrotransposons (LTR-RT) revealed distinct peaks for each subgenome (Fig. 1e). LTR-RT insertions often proliferate following significant genomic changes, such as divergence or hybridization. The insertion time of LTR-RTs in subgenome A displayed peaks at approximately 0.094 and 1.504 MYA, while the insertion time of LTR-RTs in subgenome B showed a single peak at 2.089 MYA (95% CI: 0.197–6.594). In allotetraploids, the oldest peak of LTR-RT insertion times is generally considered to represent the divergence time between the two diploid progenitors [33]. Therefore, the peak observed in the insertion time of LTR-RTs in subgenome B likely reflects the speciation event that separated the diploid subgenome progenitors from their ancestral species, while the older peak in subgenome A (1.504 MYA, 95% CI: 0.000–4.901) likely represents a composite signal resulting from both the speciation between the progenitors of subgenomes A and B and the subsequent speciation between the progenitor of subgenome A and E. yunnanense. The most recent peak in subgenome A (0.094 MYA, 95% CI: 0.000–4.901) may reflect the recent hybridization event that formed the allotetraploid.

To complement results from analysis of LTR-RT insertion times, we conducted Ks-based estimation of divergence times (Fig. 1f). We first calculated the Ks between paralogous gene pairs within subgenome A and subgenome B, separately, to estimate the timing of whole-genome duplication. The estimated whole-genome duplication dates were approximately 39.46 MYA (95% CI: 33.85–45.07) and 39.92 MYA (95% CI: 34.85–44.99), respectively. These dates are consistent with the timing of the most recent At-α whole-genome duplication event in the Brassicaceae family [6, 34, 35]. Further Ks analysis of orthologous gene pairs among subgenomes A and B, subgenome A and E. yunnanense, and subgenome B and E. yunnanense, indicates recent divergence events. The estimated divergence time between subgenome A and B is 1.78 MYA (95% CI: 0.92–2.62). The divergence time between subgenome B and E. yunnanense is estimated at 1.72 MYA (95% CI: 0.25–3.19), while the divergence between subgenome A and E. yunnanense is estimated at 1.11 MYA (95% CI: 0.06–2.25). The absence of a corresponding peak in subgenome B may reflect asymmetric TE dynamics between the two subgenomes, as transposable element bursts following polyploidization can occur in a subgenome-specific manner depending on differences in TE composition and regulation [33].

Finally, we estimated the timing of hybridization (allotetraploidization). Transposable elements accumulate mutations over time, providing a molecular clock from which to infer the divergence and hybridization between two subgenomes based on the density of TE mutation rates (Fig. 1g). We set the divergence time between the subgenomes at 2.009 MYA based on phylogenetic analysis. Using this divergence time, we estimated that hybridization occurred approximately 0.268 MYA, indicating that the progenitors of subgenomes A and B hybridized after diverging from the common ancestor of E. yunnanense and the progenitor of subgenome A.

Extensive chromosomal rearrangements in the allotetraploid genome of E. japonicum relative to its diploid relatives

To further investigate the evolutionary history of the wasabi genome, we inferred ancestral chromosome arrangements using synteny and comparative genomic data. First, we collected protein sequences from E. salsugineum (2n = 2x = 14), E. yunnanense (2n = 2x = 14), and subgenomes A and B of E. japonicum (2n = 4x = 28). Based on phylogenetic analysis, we scaffolded the E. yunnanense draft genome using E. japonicum subgenome A to reconstruct a pseudo-chromosome level sequence, as a chromosomal-level genome is necessary to study karyotype evolution. Subsequently, we constructed an ancestral karyotype of the genus Eutrema with seven chromosomes. Based on this reconstructed ancestral karyotype, we identified recombination events on each Eutrema genus from ancestral karyotype. We found that E. japonicum had undergone more extensive chromosomal rearrangements relative to the ancestral karyotype than had the two diploid species, E. salsugineum and E. yunnanense (Fig. 2a). Recombination from the ancestral karyotype that occurred in E. japonicum prior to tetraploidization was rare and appears to have occurred mostly for genome stabilization during the tetraploidization process (Table 1) [36, 37].

Fig. 2.

Karyotype recombination events in E. japonicum. a Karyotype reconstruction of the genus Eutrema (with seven chromosomes) and recombination events across three species within the genus (E. salsugineum, E. yunnanense, E. japonicum), inferred from protein sequences. Numbers at the nodes represent divergence times (MYA: million years ago). b Representative example of reciprocal translocation of chromosome arms (RTA) during chromosome evolution in E. japonicum. c Representative examples of nested chromosome fusion (NCF) and end-to-end joining (EEJ) contributing to chromosome restructuring in E. japonicum. RTA, reciprocal translocation of chromosome arms; NCF, nested chromosome fusion; EEJ, end-to-end joining

Table 1.

Shared fusion between two subgenomes of E. japonicum

Sub A	Start A	End A	Sub B	Start B	End B	P-value
A01	1063	4153	B01	979	3995	0.0015
A01	2208	2652	B06	2192	2750	0.0442
A02	3088	3357	B02	1	266	0.0002
A02	638	2477	B02	2502	846	0.003
A02	3093	3356	B05	2866	2477	0.0396
A03	1238	1881	B03	993	1615	0.0015
A03	392	926	B03	334	820	0.0011
A04	1176	3357	B04	2334	189	0.0017
A05	1076	1501	B02	6	283	0.047
A05	298	498	B02	846	1001	0.0266
A05	2149	2305	B04	1354	1270	0.0286
A05	1	2422	B05	3884	1515	0.0009
A05	3231	4017	B05	809	88	0.0062
A05	2458	3007	B05	1514	949	0.009
A06	1	3636	B06	3394	3	0.004
A07	1	2185	B07	3108	1065	0.0017
A07	3030	3386	B07	328	4	0.0045

Sub A Subgenome A chromosome number, Start A/End A Start and end positions of protein-based reconstructed sequence in subgenome A, Sub B Subgenome B chromosome number, Start B/End B Start and end positions of the protein-based reconstructed sequence in subgenome B

In addition to recombination from the ancestral karyotype, we identified recombination between the two subgenomes of E. japonicum after tetraploidization (Fig. 2b-c; S1). Chromosome fusion patterns like nested chromosome fusion and end-end joining characteristically occur between homoeologous chromosomes and are more readily detectable if they have arisen recently [38]. In general, nested chromosome fusion or end-end joining events would typically result in patterns where one A chromosome aligns in a split manner across two or more B chromosomes (or vice versa) [23]. The absence of these specific signatures, in conjunction with the observed comprehensive 1:1 synteny, indicates that such large-scale inter-subgenomic fusions have not taken place (Fig. 1a). This suggests that following wasabi's relatively recent polyploidization event (~ 0.27 MYA), no major structural reorganization via chromosome fusion has yet occurred. In conclusion, E. japonicum was formed through significant recombination from the ancestral karyotype, but its two subgenomes have remained structurally stable without major recombination after the polyploidization event.

Subgenome dominance contributes to functional complementarity and genome stability in E. japonicum

We conducted a comprehensive analysis to investigate subgenome dominance, identifying 13,525 homoeologous gene pairs in colinear blocks. Of these, 2,496 genes in subgenome A and 2,794 genes in subgenome B, respectively, showed highly biased expression compared to their homoelog (Fig. 3a). Analysis of GO terms associated with subgenomes A and B of E. japonicum revealed distinct functional characteristics, suggesting potential functional divergence.

Fig. 3.

Subgenome dominace of E. japonicum. a Expression bias between homoelog genes in E. japonicum. The gray histogram bars represent the distribution of homoelog expression bias for gene pairs with a Log₂ fold-change of less than one. Homoelog genes that are biased toward the subgenome A are shown in red, while pairs that are biased toward the subgenome B are displayed in blue. Gene ontology (GO) analysis of homolog biased genes. b Molecular function of subgenome A. c Molecular function of subgenome B. d Biological process of subgenome A. e Biological process of subgenome B. f Cellular component of subgenome A. g Cellular component of subgenome B. The horizontal axis represents the signal, which likely corresponds to a measure of enrichment strength (e.g., the proportion of genes in a group that are associated with a GO term). The vertical axis lists various GO terms. Each GO term is represented by a horizontal bar and a circle. Bar color indicates significance based on the false discovery rate (FDR)

Regarding molecular function, subgenome A exhibits a prevalence of GO terms associated with heterocyclic and organic cyclic compound binding, DNA binding, and cis–trans isomerase activity, suggesting its greater involvement in transcriptional regulation, DNA replication and repair, and, potentially, signal transduction pathways involving isomerization (Fig. 3b). In contrast, subgenome B is enriched for terms related to ion binding, small molecule binding, and transferase activity, implying potential specialization in ion homeostasis, signal transduction, and metabolic regulation (Fig. 3c).

In terms of biological processes, subgenome A shows enrichment for GO terms related to carboxylic and organic acid metabolism, particularly for biosynthetic processes, as well as the determination of bilateral symmetry during development (Fig. 3d). Conversely, subgenome B is predominantly associated with general metabolic processes, macromolecule modification, and RNA modification, indicating a role in diverse metabolic pathways, protein modification, and RNA maturation (Fig. 3e).

For the cellular component, both subgenomes share many common terms, including "Intracellular anatomical structure," "Intracellular organelle," "Intracellular membrane-bounded organelle," "Cellular anatomical entity," "Cytoplasm," "Chloroplast," and "Plastid," suggesting that both subgenomes contribute to the maintenance of fundamental cellular structures and functions, especially those related to organelles (Fig. 3f, g). Notably, subgenome A exclusively includes the term "Nucleus." This result indicates a potential enrichment of genes involved in nuclear functions, such as transcription regulation, DNA replication, and DNA repair, thus implying a more prominent role in nuclear processes compared to subgenome B.

Genomic identification of glucosinolate biosynthesis genes in E. japonicum

Glucosinolates are a diverse group of sulfur-containing compounds that act as natural pesticides and contribute to the characteristic pungent flavors of plants in the Brassicaceae family, such as cabbage, broccoli, and mustard. Secondary metabolites, activated upon tissue damage, are stored in plant tissues and play a crucial role in defense against herbivores and pathogens. We reconstructed the glucosinolate biosynthesis pathway in E. japonicum by referencing multiple established glucosinolate pathways (Fig. 4). The glucosinolate biosynthesis pathway can be divided into three distinct stages: (1) side-chain elongation of precursor amino acids, (2) core structure formation, and (3) secondary modification of the side chain [39–41]. We determined the copy numbers of genes involved in each step of the pathway in E. japonicum and analyzed their distribution across the genome.

Fig. 4.

Glucosinolate biosynthesis and breakdown pathways in wasabi. Pathways were categorized into three main types based on their amino acid precursors: Aliphatic, Indolic, and Aromatic glucosinolate (GSL). Green boxes represent amino acid chain elongation steps, blue boxes indicate core GSL structure formation, pink boxes denote side chain modifications. The numbers in parentheses indicate the number of gene copies in E. japonicum corresponding to each pathway. ESP: epithiospecifier protein, ESM1: epithiospecifier modifier 1, NSP: nitrile-specifier protein, TFP: thiocyanate-forming protein

The pungent taste of E. japonicum is primarily due to the hydrolysis of sinigrin into allyl isothiocyanate, a reaction catalyzed by the enzyme myrosinase upon tissue disruption. In wasabi, the biosynthesis of sinigrin, a specific type of aliphatic glucosinolate, begins with the amino acid methionine and proceeds through a series of enzymatic transformations (Fig. 4). Initially, methionine is converted into 2-oxo acid by the action of branched-chain amino acid aminotransferase 4 (BCAT4) and BCAT6. Next, methylthioalkylmalate synthase 1 (MAM1) and methylthioalkylmalate synthase 2 (MAM2) catalyze the chain elongation of 2-oxo acid, producing a chain-elongated methionine derivative with a 3-carbon structure. This intermediate is then transformed into an aldoxime by cytochrome P450 enzymes CYP79F1 and CYP79F2, followed by conversion into S-alkyl-thiohydroximate via CYP83A1. Subsequently, S-alkyl-thiohydroximate is converted into thiohydroximate acid by superroot 1 (SUR1), and then into desulfo-glucosinolate by UDP-glycosyltransferase 74B1 (UGT74B1). This desulfo-glucosinolate is further converted into methylthiopropyl (also known as glucobrassicanapin) through the action of sulfotransferases SOT16, SOT17, and SOT18. Methylthiopropyl is subsequently oxidized by flavin-monooxygenase glucosinolate-S-oxygenase 5 (FMOgs-ox5) to form methylsulfinylpropyl. Finally, methylsulfinylpropyl is converted into sinigrin by the 2-oxoglutarate-dependent dioxygenase AOP1.

When a plant experiences tissue damage, sinigrin is hydrolyzed by myrosinase (β-thioglucoside glucohydrolase, TGG) into an aglucone intermediate, which can be converted into different breakdown products depending on the enzymes present. The formation of isothiocyanates, known for their role in plant defense and potential health benefits, is facilitated by ESM1. If ESP is present, the aglucone is converted into epithionitrile, while the production of thiocyanates requires both ESP and thiocyanate forming protein (TFP). Additionally, nitriles are formed through the action of ESP1 and of nitrile specifier protein 1 (NSP1), NSP2, NSP3, and NSP5. This glucosinolate degradation pathway plays a crucial role in plant defense, influencing the bioavailability, toxicity, and characteristic pungency of plants like wasabi [42]

Distribution of glucosinolate-related genes within the Brassicaceae family

To investigate the evolution and diversification of glucosinolate biosynthesis within the Brassicaceae family, we focused on a core set of 34 genes directly involved in the pathway, selected from an initial pool of 172 genes based on established pathway information (Fig. 5). We performed a synteny analysis across ten representative Brassicaceae species: E. japonicum, E. yunnanense, E. heterophyllum, E. salsugineum, A. thaliana, A. rusticana, R. sativus, B. rapa, B. oleracea, and B. napus. We observed a significant expansion of the indole glucosinolate O-methyltransferase (IGMT) gene family across the Brassicaceae. IGMTs catalyze the methylation of indole glucosinolates, a crucial step in the biosynthesis of these defense compounds. As anticipated, the allotetraploid species B. napus generally exhibited a higher number of glucosinolate biosynthesis genes compared to the diploid species, a likely consequence of its polyploidy. In contrast, although E. japonicum is tetraploid, it exhibits an overall reduction in glucosinolate gene retention compared to B. napus. The pattern of gene retention is generally similar across the Eutrema genus; however, compared to the diploid species (E. heterophyllum, E. salsugineum, and E. yunnanense), allotetraploid E. japonicum displays approximately twice as many genes.

Fig. 5.

Comparative analysis of gene copy number for glucosinolate-related genes across species in Brassicaceae. The color of the circle transitions from light blue to orange, and the size of the circles increases, to indicate a greater number of genes. EJ: Eutrema japonicum, EH: Eutrema heterophyllum, EY: Eutrema yunnanense, ES: Eutrema salsugineum, AT: Arabidopsis thaliana, AR: Armoracia rusticana, RS: Raphanus sativus, BR: Brassica rapa, BO: Brassica oleracea, BN: Brassica napus. Each dot represents the gene's copy number, with larger, darker dots indicating higher copy numbers

The pungency of E. japonicum is attributed to the hydrolysis of sinigrin into allyl isothiocyanate (Fig. 4) [9]. This reaction is modulated by two key enzymes: ESP, which promotes nitrile production, and ESM1, which promotes isothiocyanate formation and inhibits nitrile production [43]. Our genomic analysis revealed that E. japonicum possesses seven copies of ESM1 and only one copy of ESP, representing the highest ESM1 gene copy number among the analyzed species (Fig. 5, Table 2).

Table 2.

Gene copy numbers of ESM1 and ESP in ten species of Brassicaceae family

Gene	EJ	EY	EH	ES	AT	AR	RS	BR	BO	BN
ESM1	7	2	5	6	3	3	1	3	2	6
ESP	1	1	0	0	2	0	0	4	5	9

EJ Eutrema japonicum, EY Eutrema yunnanense, EH Eutrema heterophyllum, ES Eutrema salsugineum, AT Arabidopsis thaliana, AR Armoracia rusticana, RS Raphanus sativus, BR Brassica rapa, BO Brassica oleracea, BN Brassica napus

Across the Brassicaceae family, ESM1 gene copy numbers show notable variation among lineages. Species within the Eutrema genus generally exhibit relatively higher ESM1 copy numbers compared with many other Brassicaceae members. For example, E. heterophyllum and E. salsugineum contain five and six copies, respectively, while E. yunnanense has two copies. In contrast, most other Brassicaceae species possess fewer ESM1 copies, although B. napus also shows a relatively high copy number.

A different pattern is observed for ESP genes. Members of the Eutrema genus typically contain very few ESP copies, with E. japonicum and E. yunnanense each possessing a single copy and E. heterophyllum and E. salsugineum lacking the gene entirely. In contrast, ESP genes are expanded in the Brassica lineage, where B. rapa, B. oleracea, and B. napus contain multiple copies. Taken together, these results indicate that the Eutrema genus, particularly E. japonicum, is characterized by a relatively high ESM1/ESP gene copy ratio compared with many other Brassicaceae species.

Additional phylogenetic analyses clearly demonstrated that the ESM1 (Fig. 6a) and ESP (Fig. 6b) genes are evolutionarily conserved across the Brassicaceae family, suggesting their functional importance in glucosinolate metabolism, and underwent independent evolutionary trajectories within E. japonicum (Fig. 6c). Notably, ESM1 genes form a distinct clade in the phylogenetic tree, separate from other genes involved in the glucosinolate breakdown pathway (ESP, TFM, NSP genes). This strongly supports the hypothesis that ESM1 genes experienced unique functional divergence and independent evolution within the glucosinolate degradation process. This separation in the model organism A. thaliana is well-supported: their protein interaction networks are largely independent (Fig. 6d), phylogenetic analysis places ESM1 and ESP in distinct evolutionary clades (Fig. 6e), and their predicted protein structures show fundamentally different functional domains (Fig. 6f, g). In Brassica species, which generally lack pungency, ESM1 genes show consistently low expression levels across leaves. In contrast, species known for their pungent taste—such as those in the Eutrema genus and horseradish (A. rusticana)—exhibit high expression of ESM1 genes, particularly in leaf tissues (Fig. 6h). Additionally, ESP gene expression tends to be high in Brassica species (Fig. 6i). These expression patterns, when considered alongside gene copy number, suggest that the transcriptional activity levels of ESM1 and ESP genes play critical roles in determining pungency in Brassicaceae. A high ESM1/ESP gene expression ratio appears to be a defining molecular signature of pungent species, reinforcing the hypothesis that ESM1-driven isothiocyanate formation is essential for the pungency characteristic.

Fig. 6.

Independent Evolution and Functional Divergence of ESM1 and ESP Genes. a Phylogenetic relationships of ESM1 genes in Brassicaceae. b Phylogenetic relationships of ESP genes in Brassicaceae. c Phylogenetic relationships of glucosinolate biosynthesis genes in E. japonicum with A. thaliana as the outgroup. Red lines indicate E. japonicum genes related to ESM1 genes, and blue lines indicate those related to ESP genes. d Protein interaction network of glucosinolate biosynthesis genes. e Phylogenetic relationships of genes in the glucosinolate biosynthesis pathway in A. thaliana, divided into four major clades highlighted by different colors. Protein structure of (f) ESM1, where the blue region represents the transmembrane region and the black area represents the Lipase_GDSL domain, and (g) ESP, where the black region represents the Pfam Kelch_1 domain and the light orange region represents the SMART Kelch domain. Expression profiling of (h) ESM1 and (i) ESP genes in ten Brassicaceae species. Species abbreviations are as follows: EJ: E. japonicum, EY: E. yunnanense, EH: E. heterophyllum, ES: E. salsugineum, AT: A. thaliana, AR: A. rusticana, RS: R. sativus, BR: B. rapa, BO: B. oleracea, BN: B. napus. Expression values are quantified as log₂TPM

Furthermore, we studied the genomic distribution of glucosinolate biosynthesis genes in E. japonicum to explore potential regulatory mechanisms (Fig. 7a, b). Most glucosinolate biosynthesis genes, including all seven copies of the ESM1 gene, are predominantly located in subtelomeric regions, near the telomeric sequences at both ends of chromosomes. This subtelomeric localization contrasts markedly with previous findings in B. napus, where glucosinolate biosynthesis genes were broadly distributed throughout the genome [20]. Gene expression analyses in three tissues of E. japonicum further revealed that most subtelomerically located glucosinolate biosynthesis genes exhibited high expression levels in leaves (Fig. 7c, d).

Fig. 7.

Genomic locations and evolution of glucosinolate-associated genes in E. japonicum. a Physical maps of chromosomes 1 through 7 of subgenome A (ChrA01-07) in E. japonicum. b Physical maps of chromosomes 1 through 7 of subgenome B (ChrB01-07) in E. japonicum. For both sets of chromosomes, red labels indicate ESM1 genes, and black labels indicate glucosinolate genes except for ESM1 genes. The scale on the left of each chromosome represents the chromosome's size in megabases (Mb) and shows the relative positions of the genes. c The expression profile of glucosinolate biosynthesis genes in three tissues: leaf, stem, and root. d The expression profile of glucosinolate breakdown genes in the same tissues. Expression levels are quantified as Log₂TPM values., e Schematic representation of the ESM1 gene distribution in E. japonicum and A. thaliana. f Evolutionary process of ESM1 genes in E. japonicum. Starting from an ancestral species, which contained genes that are the progenitors of present-day Euj137/Euj24808, Euj138/Euj24809, and Euj6168/Euj29732, a speciation event separated the lineage into subgenomes A and B. A tandem duplication event, affecting the ancestral gene of Euj137 and Euj24808, occurred before this speciation. The allotetraploid E. japonicum arose through tetraploidization, leading to a total of seven ESM1 gene copies. The gene Euj572 originated after speciation, within subgenome A. The phylogenetic tree on the right, constructed using E. japonicum ESM1 genes and the A. thaliana ESM1 sequence (AT) as an outgroup, supports this evolutionary history

A distinct ESM1/ESP gene ratio in E. japonicum compared with other Brassicaceae species

We studied the mechanisms driving ESM1 gene copy number in E. japonicum. First, we constructed a phylogenetic tree based on the key genes of the glucosinolate pathway in E. japonicum, using A. thaliana as an outgroup (Fig. 5c). Focusing on the ESM1 genes, we elucidated the evolutionary relationships among the seven genes identified in E. japonicum (Euj137, Euj24808, Euj572, Euj138, Euj24809, Euj6168, and Euj29732) compared to the ESM1 gene in A. thaliana. The evolution of ESM1 genes in E. japonicum involved a combination of tandem duplication, speciation, and tetraploidization (Fig. 7e, f). The diploid ancestral species of E. japonicum initially possessed two ESM1 genes (A1, A2). The A1 gene underwent tandem duplication, producing the ancestral gene (A1-1, A1-2) that gave rise to Euj137 and Euj24808 (A1-1), as well as the one that gave rise to Euj138 and Euj24809 (A1-2). Although the pairs Euj137/Euj138 and Euj24808/Euj24809 seem to have duplicated in tandem after polyploidization, phylogenetic analysis indicated that the tandem duplication actually occurred in the ancestral polyploid before speciation (Fig. 7e). Following speciation, A1-1, A1-2, and A2 were partitioned into two distinct diploid species, with subgenome A retaining Euj137, Euj138, and Euj6168, and subgenome B acquiring Euj24808, Euj24809, and Euj29732. Subsequently, the subgenome A–specific Euj572 gene emerged. This finding is supported by phylogenetic analysis: Euj572 is phylogenetically distant from the ESM1 gene in A. thaliana and closely related to subgenome A. This suggests that rather than being simply lost from subgenome B after its appearance in the ancestral species, Euj572 likely arose de novo in subgenome A after speciation. Finally, through tetraploidization, E. japonicum ultimately acquired a total of seven ESM1 genes. Among these, Euj137 and Euj24808 demonstrated particularly high expression levels in leaf tissues, suggesting their prominent role in leaf-specific glucosinolate metabolism. Euj6168, on the other hand, exhibited strong expression in both leaves and roots, implying a broader functional role across different tissues. In contrast, Euj138 and Euj24809 showed markedly low expression levels, indicating potential subfunctionalization or pseudogenization after tandem duplication. These differential expression patterns further support the hypothesis that the expansion of the ESM1 gene family in E. japonicum was accompanied by functional diversification among the gene copies.

Discussion

Genome evolution and recent allotetraploidization of E. japonicum

This study provides a comprehensive genomic analysis of E. japonicum, a commercially important Brassicaceae plant, to elucidate the evolutionary history of its allotetraploid genome. The estimated LTR-RT insertion times, phylogenetic analysis, and Ks-based divergence time analysis all show congruent results, suggesting that the speciation of the progenitors of subgenomes A and B occurred approximately 1.78 to 2.089 MYA, while the divergence between the progenitor of subgenome A and E. yunnanense occurred 1.25 to 1.11 MYA. This concordance strongly supports the estimated timing of E. japonicum genome formation (Fig. 1d-f). Moreover, hybridization occurred almost immediately after subgenome divergence, which strongly suggests that E. japonicum is a very recently formed neopolyploid. However, in our analysis, there is a slight discrepancy between the hybridization times predicted based on LTR-RTs (Fig. 1e) and those based on TE divergence rates (Fig. 1g). Although valuable, LTR-RT insertion dating may underestimate hybridization time in allopolyploids because the activity of specific retrotransposon families fluctuates, influenced by factors such as stress and epigenetics [33]. Furthermore, this peak can be obscured by signals related to the speciation of E. yunnanense and the progenitor of subgenome A. In contrast, the TE divergence rate method offers a more robust estimate by averaging mutation accumulation across a broader range of TE families, including both LTR-RTs and DNA transposons. This larger sample size can reduce influence from stochastic fluctuations in individual TE families, providing a more stable molecular clock that reflects overall genomic divergence between subgenomes. While a constant mutation rate is a simplification, averaging divergence across multiple TE families improves stable proxy compared to relying solely on recent LTR-RT insertions [44–46]. Based on TE divergence rate analysis from previously reported studies, the merger of the tetraploid Acorus calamus genome is estimated to have occurred 1.3 MYA [47], and the hybridization event in the hexaploid species Echinochloa crus-galli at 0.31 MYA [48]. The results provide strong evidence for recent hybridization events, and their reliability is supported by their congruence with established phylogenetic relationships.

Genome stabilization and subgenome differentiation after polyploidization

The extensive chromosomal rearrangements observed in E. japonicum suggest that recombination and structural reorganization may have played important roles during genome stabilization following polyploidization (Fig. 2). Research on A. thaliana [36] and B. napus [37] has also reported similar findings. Homoeologous recombination can generate genomic rearrangements and novel variations, which can be favored by natural selection despite potential instability. These patterns may reflect homoeologous recombination and structural genome reshuffling that occurred during the early stages of allotetraploid genome stabilization. However, it should be noted that our inference is based on comparative genomic patterns rather than direct measurements of meiotic recombination rates. The fact that only a limited number of chromosomal fusion events were shared between the two subgenomes lends further support to the hypothesis that the majority of chromosomal rearrangements took place after the formation of the allotetraploid (Table 1). The observed subgenome dominance in E. japonicum is consistent with patterns widely reported in many polyploid species. Following allopolyploidization, one parental subgenome often becomes more transcriptionally or structurally dominant, a phenomenon documented in diverse plant systems [49–51]. In our study, the dominance pattern observed between the two subgenomes of E. japonicum may reflect asymmetric evolutionary trajectories following hybridization (Fig. 3). Such dominance has been proposed to mitigate genomic and epigenetic conflicts that arise after polyploid formation and may contribute to long-term genome stability. In the context of the extensive chromosomal rearrangements detected in this study, the emergence of subgenome dominance may represent an additional mechanism facilitating genome stabilization during the early stages of allotetraploid evolution.

GO analysis results are consistent with the subgenome dominance hypothesis. These results further suggest functional differentiation between the two subgenomes during the stabilization of the polyploid genome. Subgenome A is enriched for genes involved in DNA binding, organic compound binding, isomerase activity, carboxylic and organic acid metabolism, developmental processes and, notably, nuclear functions. Conversely, subgenome B appears specialized for ion and small molecule binding, transferase activity, general metabolic processes, and RNA processing. (Fig. 3b-g).

Such asymmetric functional partitioning between subgenomes has been widely reported in polyploid plants and is considered an important mechanism facilitating genome stabilization following allopolyploidization [52, 53]. Functional differentiation between duplicated gene sets can reduce redundancy and allow complementary gene functions to be retained, thereby stabilizing regulatory and metabolic networks in newly formed polyploids. In E. japonicum, the enrichment of regulatory and nuclear-related functions in subgenome A suggests that this subgenome may have assumed a greater role in transcriptional regulation and genome maintenance, whereas subgenome B may contribute more extensively to metabolic and biochemical processes.

This functional differentiation may have been particularly important during the early stages of allotetraploid genome stabilization, when duplicated regulatory networks and metabolic pathways required reorganization. In Brassicaceae species, many metabolic pathways are associated with secondary metabolites involved in chemical defense, including glucosinolate-derived compounds responsible for the characteristic pungency of wasabi. Therefore, the enrichment of metabolic and transferase-related activities in subgenome B may reflect diversification of metabolic functions following polyploidization, potentially contributing to the biochemical complexity of E. japonicum. Taken together, these results suggest that subgenome differentiation in E. japonicum involved functional specialization between regulatory and metabolic processes, which may have facilitated genome stabilization following the recent allotetraploidization event.

ESM1 expansion and ESM1/ESP imbalance may underlie wasabi pungency

The significant expansion of the ESM1 gene family in E. japonicum, alongside the low copy number of ESP genes, may be consistent with enhanced pungency, although gene family expansion in glucosinolate-related pathways can also be influenced by multiple ecological and evolutionary pressures. Since ESM1 promotes the hydrolysis of sinigrin into allyl isothiocyanate—the key compound responsible for wasabi’s characteristic spicy flavor—while ESP diverts the reaction toward non-pungent nitriles, and the gene copy number imbalance likely promotes isothiocyanate dominance in E. japonicum [43, 54, 55]. This ESM1/ESP gene copy ratio pattern is consistently observed across pungent Eutrema species [56] (Fig. 6, Table 2), implying that selective pressure may have favored ESM1 gene expansion and/or ESP gene reduction to enhance pungency. Interestingly, although B. napus also shows a high ESM1 copy number (six), it has an even greater number of ESP genes (nine), which may lead to a suppressed isothiocyanate profile and a milder taste. The high expression of the ESM1 gene in Eutrema species and its low expression in Brassica species support these findings (Fig. 6h). These findings underline the importance of the ESM1:ESP gene balance and gene expression in determining glucosinolate hydrolysis outcomes and pungency levels. Future functional studies, including metabolite profiling and the development of overexpression/knockout lines, will be essential to further elucidate the genetic mechanisms underlying pungency diversification across the Brassicaceae family. However, gene family expansion in glucosinolate-related pathways is not necessarily driven solely by selection for pungency. In Brassicaceae species, glucosinolate hydrolysis products are well known to contribute to defense against herbivores and microbial pathogens. In addition, gene retention patterns following polyploidization can influence the dosage balance of metabolic pathways and may contribute to lineage-specific expansion of certain gene families. Therefore, the expansion of ESM1 genes in E. japonicum may reflect the combined effects of ecological interactions and genome evolutionary processes rather than a single selective driver.

In E. japonicum, important glucosinolate biosynthesis genes are strategically located in dynamic subtelomeric regions. These areas are associated with high recombination rates, frequent gene duplication, and altered gene expression patterns, and often harbor genes linked to environmental adaptation and defense [57]. This placement, particularly for ESM1 genes involved in isothiocyanate production, may facilitate rapid evolutionary responses to environmental pressures [58]. Increased recombination after tetraploidization may have facilitated the redistribution and retention of these genes in these dynamic regions (Fig. 2; 7a, b).

The subtelomeric localization of E. japonicum glucosinolate biosynthesis genes contrasts with the broader distribution in B. napus [20], highlighting a unique evolutionary trajectory that may contribute to the distinctive glucosinolate and isothiocyanate profile of wasabi. Despite potential concerns regarding suppressed gene expression in subtelomeric regions, our findings indicate robust expression levels among glucosinolate biosynthesis genes in these regions (Fig. 7c, d). The high leaf-specific expression patterns, particularly evident for Euj137, Euj24808, and Euj6168, underscore their critical roles in the synthesis and transport of glucosinolates and their derivatives, such as isothiocyanates, essential for plant defense mechanisms. The high expression levels of glucosinolate genes in leaves involved in the synthesis of short-chain aliphatic glucosinolates such as sinigrin, glucoraphanin, glucoiberin, and glucoalyssin in leaves likely reflect their predominant biosynthesis in leaf tissues, followed by subsequent transport to the roots [59].

While the study provides a strong genomic foundation and compelling hypotheses, a limitation is its reliance on computational analyses, necessitating further experimental validation to confirm the functional significance of these findings, especially regarding the role of ESM1 gene expansion and subgenome dominance in shaping wasabi's unique metabolic profile. Future research should, therefore, focus on the functional characterization of key genes, including ESM1 and ESP gene, through techniques like Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)-Cas9 gene editing and RNASeq analysis across various tissues and developmental stages. Comparative studies with related Eutrema species, especially those representing potential progenitor lineages, are also essential. In conclusion, this work provides a valuable framework for future investigations into the evolution of polyploid genomes, particularly within the Eutrema genus, and the genetic basis of specialized traits in plants, highlighting the dynamic interplay between genome evolution, adaptation, and the generation of novel chemical diversity.

Conclusion

This study provides a comprehensive genomic analysis of E. japonicum and clarifies the evolutionary processes underlying its allotetraploid genome. Multiple lines of genomic evidence, including phylogenetic reconstruction, LTR-RTs insertion dating, and Ks-based divergence analyses, consistently indicate that E. japonicum originated from a recent hybridization event between two diploid progenitors approximately 0.27 MYA. Comparative genomic analyses further reveal extensive chromosomal rearrangements relative to the ancestral karyotype, whereas the two subgenomes themselves have remained largely structurally stable following polyploidization.

Subgenome dominance analysis indicates functional differentiation between the two subgenomes, with subgenome A enriched in regulatory and nuclear-related functions and subgenome B primarily associated with metabolic and biochemical processes. This asymmetric functional partitioning likely contributed to genome stabilization during the early stages of allotetraploid evolution.

Comparative analyses of glucosinolate-related genes revealed a distinctive expansion of the ESM1 gene family and a low copy number of ESP genes in E. japonicum, resulting in a uniquely high ESM1/ESP gene ratio within the Brassicaceae family. Together with expression patterns, these findings suggest that ESM1-driven isothiocyanate formation may play an important role in the characteristic pungency of wasabi.

However, this study has several limitations. Our conclusions are primarily based on comparative genomic and transcriptomic analyses, and functional validation of candidate genes such as ESM1 and ESP was not performed. In addition, metabolite profiling of glucosinolate hydrolysis products was not included, which limits direct linkage between gene copy number variation and pungency phenotypes. Future studies integrating functional genomics, metabolite analysis, and broader sampling of potential progenitor lineages will be important to further clarify the genetic mechanisms underlying pungency diversification and genome evolution in the Eutrema genus.

Abbreviations

AOP

Alkenyl hydroxalkyl producing enzyme

BCAT

Branched-chain amino acid aminotransferase

CYP

Cytochrome P450

EEJ

End-to-end joining

ESM1

Epithiospecifier-modifying protein 1

ESP

Epithiospecifier protein

FDR

False discovery rate

FMOgs-ox

Flavin-monooxygenase glucosinolate-S-oxygenase

GO

Gene Ontology

GSL

Glucosinolate

IGMT

Indole glucosinolate O-methyltransferase

Ks

Synonymous substitution rate

LTR-RTs

Long terminal repeat retrotransposons

MAM

Methylthioalkylmalate synthase

MYA

Million years ago

NCF

Nested Chromosome Fusion

NSP

Nitrile specifier protein

RTA

Reciprocal Translocation of Chromosome Arms

SOT

Sulfotransferase

SRA

Sequence Read Archive

SUR1

Superroot 1

TE

Transposable element

TFP

Thiocyanate forming protein

TGG

β-Thioglucoside glucohydrolase (myrosinase)

TPM

Transcripts per million

UGT

UDP-glycosyltransferase

WGD

Whole-genome duplication

Author’s contributions

C.K., D.J. designed the study. D.J, S.L, and S.C. performed experiments. D.J. analyzed data. C.K., and A.H.P. supervised the study. D.J. wrote the manuscript with input from all authors. D.J., C.K., and A.H.P. revised the manuscript.

Funding

This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (RS-2024–00414335).

Data availability

Protein sequences predicted from genome assemblies of E. salsugineum (GCF_000478725.1), E. japonicum (GCA_041074935.1), E. yunnanense (GCA_002933935.1), E. heterophyllum (GCA_002933915.1), B. napus (GCF_020379485.1), B. rapa (GCF_000309985.2), B. oleracea (GCF_000695525.1), R. sativus (GCF_000801105.2), A. thaliana (GCF_000001735.4), S. bicolor (GCF_000003195.3) were obtained from the NCBI, and A. rusticana were obtained from figshare (accession: 21,780,176.v2).

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.