The TSM high-generation selfing line B186 was sown in the Yangtze Normal University squash trial site in the fall of 2021 after germination. Per the normal growth requirements of TSM, field management was performed during growth. The leaves of B186 were collected at the 4-leaf stage for BjuAOP2 cloning, and the expanded tumorous stems of B186 were collected at 4, 7, 10, 13, 16, 19, and 22 weeks after sowing to analyze the expression pattern of BjuAOP2 at different developmental stages. In addition, roots, stems, leaves, tumorous stems, flowers, and siliques of B186 were collected at the shooting stagegib for analysis of the gene’s tissue expression pattern. Each sample had three biological replicates that were collected and placed in liquid nitrogen and kept at -80°C for subsequent use.

The protein sequence of the Arabidopsis thaliana AOP2 gene was retrieved online from TAIR (http://www.arabidopsis.org/). Subsequently, a BlastP search was performed in the Brassica genome database (http://brassicadb.cn/#/BLAST/) to search for amino acid sequences and nucleic acid sequences of TSM AOP2 gene family members. In addition, HMMsearch (Johnson et al., 2010) was used to identify the 2OG-FeII_Oxy (PF03171) and DIOX-N(PF14226) structural domains for all possible AOP2 genes. Finally, conserved structural domain confirmation was performed through NCBI-CDD (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) in NCBI to exclude sequences without typical structural domains.

Total RNA of TSM leaves was isolated with Tiangen RNA Extraction Kit TRNzol Universal Reagent (Tiangen Biotech, Beijing, China) and with a subsequent reversal to single-stranded cDNA using TransScript One-Step gDNA Removal and cDNA Synthesis SuperMix kit (TransGen Biotech, Beijing, China). The designing of the primers ( Supplementary Table S1 ) and amplification of the full-length CDS sequences using Primer premier 6.0 software was the next step, using the BjuAOP2 gene sequence obtained from the stem mustard database as a template. The PCR reaction system and reaction procedure were performed according to the instructions of TransStart FastPfu DNA Polymerase from TransGen, and the PCR product recovery and purification were performed according to the instructions of the TransGen Gum Recovery Kit (EG101). In addition, the PCR-recovered products were homologously cloned with the vector pTF101-GFP using the homologous recombination method of the pEASY-Basic Seamless Cloning and Assembly Kit (TransGen Biotech, Beijing, China) from TransGen. Subsequently introducing the recombinant vector into E. coli receptor DH5α, plasmids were extracted from 12 positive clones for validation and sequencing. The protein sequences encoding the BjuAOP2 gene obtained by sequencing were subjected to multiple-sequence alignment with the AOP2 protein sequences of Arabidopsis thaliana using DNAStar 7.1 software. The CDS length and amino acid numbers of the BjuAOP2 gene were obtained from the clone sequencing results. Data such as chromosome position was obtained from the reference genome. Isoelectric points and molecular weights were obtained from the pI/Mw calculation tool on the website ExPASy (http://www.expasy.org/tools/). The prediction of subcellular localization was obtained using online tool BUSCA (http://busca.biocomp.unibo.it/).

Genomic data of 19 sequenced cruciferous species were obtained from The Brassicaceae Genome Resource (http://www.tbgr.org.cn) (Liu et al., 2022). The protein sequence of the Arabidopsis thaliana AOP gene was accessed at the TAIR website as a reference sequence and the AOP family gene sequence was obtained using the Quick Find Best Homology tool in TBtools software (Chen et al., 2020). Clustal W was employed to carry out the multiple-sequence alignment of AOP family proteins from sequenced species in the Brassicaceae family in MAGE 11 software. Subsequently, utilizing the neighbor-joining (NJ) method phylogenetic trees were drawn, with the Bootstrap method set to 1000 (Tamura et al., 2021). The online tool, GSDS (http://gsds.cbi.pku.edu.cn/index.php) was employed for mapping the intron-exon structure pattern of the AOP gene. The analysis of the conserved motif of the gene was performed using the MEME (http://meme-suite.org/) online tool. The upstream promoter sequences (1.5-kb) of each gene of the identified AOP family were accessed at the Brassica genome website (http://brassicadb.org) with the promotor-bound cis-acting elements were examined by the software PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html). Schematic diagrams of all the above analysis results were illustrated utilizing the software Domain Illustrator software (http://dog.biocuckoo.org/) (Ren et al., 2009).

Protein subcellular localization of BjuAOP2 was performed by transient expression in tobacco leaf epidermal cells. Agrobacterium tumefaciens GV3101 containing the ProCAMV35S::GFP::BjuAOP2 vector plasmid was activated and enlarge-cultivated. Subsequently, the organisms were collected, resuspended in a resuspension solution (OD600 almost 0.5), left for 2-3 h, and introduced into the lower epidermis of 3-4 weeks old tobacco leaves with a syringe. Observations were made after 3 days of transformation. The leaves transformed with empty pTF101-GFP were used as control. In addition, laser confocal microscopy (Nikon, Tokyo, Japan) was utilized for the observation of GFP fluorescence with excitation light at 488 nm and emission light at 510 nm. Furthermore, chloroplast autofluorescence showed excitation light at 640 nm and emission light at 675 nm.

The design of the primers and amplification of full-length CDS sequences were executed through Primer premier 6.0 software, using the BjuAOP2 gene sequence obtained from the stem mustard database as a template ( Supplementary Table S2 ). The pET-32a expression vector (Novagen. Madison, WI, USA) was constructed using the full-length CDS sequence of the gene through homologous recombination with the pEASY-Basic Seamless Cloning and Assembly Kit (TransGen Biotech, Beijing, China). Subsequently, plasmids were extracted from 12 positive clones for validation and sequencing. The correctly sequenced recombinant plasmids for the prokaryotic expression of the gene were transformed into Escherichia coli strain Transetta (DE3). The recombinant pET vector contains the T7 promoter and capable of expressing a fusion protein containing thioredoxin. The E. coli DE3 strain containing prokaryotic expression of the recombinant plasmid was incubated in LB medium at 37°C while also being shaken until OD600 was approximately = 0.6. Then, a final concentration of 0.5 mM was obtained by adding IPTG, and the expression of recombinant protein was induced by shaking the bacteria overnight at 16°C. The bacteria were isolated after induction by centrifuging the E. coli broth for five minutes at 6000 rpm/min. These bacteria were then suspended in PBS and placed on an ultrasonic cell crusher for 3 min at 100MHz (5-second on-10-second off cycle). The obtained cell-disrupted solution was centrifuged (12000 rpm/min and 20 min) for supernatant collection. Subsequently, the supernatant was purified by fusion protein according to the instruction procedures of Ni IDA Beads 6FF Kit (Changzhou Smart-Lifesciences Biotechnology, Changzhou, China). Finally, the purified fusion protease was detected by SDS-PAGE electrophoresis.

Kliebenstein et al. (2001a) validated the Arabidopsis thaliana AOP2 protein by in vitro experiments. In this procedure, the same method was used to validate the function of BjuAOP2 protein in Arabidopsis thaliana, with a slight modification of the enzyme activity assay. The procedure was as follows: 400 μL of purified BjuAOP2 protein solution, 10 mM ascorbate, 15 mM α-ketoglutarate, 200 mM sucrose, 200 μM FeSO4, and 100 μL GIB were added to five 10 mL centrifuge tubes (total reaction volume: 4 mL). A control 10 mL centrifuge tube was added with 400 μL ddH2O instead of purified BjuAOP2 protein solution, and the other components remained unchanged. Subsequently, the six centrifuge tubes were placed on a shaker at 28°C (110 rpm) for 4 hours. The reaction solution was purified by desulfurization and subjected to high-performance liquid chromatography (HPLC) to analyze the GSL fraction

GSLs were extracted and assayed per the prior procedure (He et al., 2002). First, 1 mL of the enzyme-activated reaction solution was transferred to a DEAE Sephadex A-25 column for overnight desulfurization at room temperature using lipase sulfate (Sigma, E.C. 3.1.6.) and eluted twice with 750 µL of deionized water. The filtration of the eluent was carried out through a 0.22 μm Hydrophilic polyethersulfone (PES) Syringe Filter (Shanghai Anpel) for HPLC analysis. The elute, the external standard 3-GIB, and SIN were analyzed by HPLC.

Total RNA was obtained from different tissues and developmental stages of tuberous mustard through TRNzol Universal Total RNA Extraction Reagent from Tiangen Biotech (Beijing, China). cDNA was synthesized based on the CDS sequences obtained by homologous recombinant cloning and sequencing using Primer premier 6.0. The afore-mentioned software designed real-time fluorescence quantification primers ( Supplementary Table S3 ). The reaction system was as follows: 2×SuperReal PreMix Plus 10 μL; upstream and downstream primers, 0.5 μL each; cDNA template, 2 μL; ddH2O, 7 μL. In addition, a three-step amplification method was used: 95°C pre-denaturation for 5 min; 95°C for 30 s, 58°C for 30 s, and 72°C for 30 s (40 cycles). Dissociation curve: 65-95°C (0.5°C increase per cycle) for 5 s (1 cycle). The entire fluorescent quantitative PCR reaction was performed on a Roche Light Cycler 480 instrument. The proportionate expression of the five BjAOP2 was calculated using the 2^-ΔΔCT calculation method with actin as an internal reference gene (Livak and Schmittgen, 2001). Technical and biological replicates, three each were utilized in each experiment.

One-way analysis of variance (ANOVA), significance analysis, and Duncan’s multiple-comparison were performed on the experimental data using SPSS analysis software, with a significance threshold of p-value < 0.05.

BLAST alignment of the Arabidopsis thaliana AOP2 gene protein sequences with the TSM database was performed. Subsequently, the 2OG-FeII_Oxy (PF14226) structural domain was identified for all possible AOP2 genes. In total, five members of the gene family distributed in different chromosomes, named BjuAOP2.1-BjuAOP2.5, were identified from the TSM genome database ( Table 1 ). The CDS sequences of the genes mentioned above were obtained by PCR using the cDNA obtained by reverse transcription as a template. BjuAOP2 genes have CDS sequences in the length range of 1284-1428 bp, with encoding amino acid numbers between 427-475 aa, molecular weight ranging from 46.8 to 55.2 kDa, and isoelectric point variation ranging from 4.77 to 5.66. The optimal prediction of subcellular localization is nucleus.

Multiple-sequence alignment with the Arabidopsis thaliana AOP2 gene ( Figure 1 ) depicted that the BjuAOP2 gene family’s five members contain the key structural domains, DIOX-N and 2 OG-Fell-Oxy, and the sequences in this range are highly conserved. However, the protein sequence similarity of the middle part of these five members is low, leading to the divergence of gene functions among the members.

An AOP gene evolutionary tree was constructed for TSM and 18 other cruciferous crops that have been sequenced to further understand the evolutionary origins of the five members of the BjuAOP2 gene family in TSM ( Figure 2 ; Supplementary Table S4 ). The AOP gene family is divided into two major categories: (1) AOP1s consisting of 33 AOP1 amino acid sequences; (2) AOP2s and AOP3s consisting of 25 AOP2 amino acids and 3 AOP3 amino acid sequences. In addition, small branches (N ≤ 3) formed by the genes are all derived from different Brassicaceae species, and the small branches further form large branches in the form of clusters containing two and three genes. Members of the TSM AOP gene family are more similar to those in B. carinata and B. rapa than those in Arabidopsis thaliana. The TSM BjuAOP2 gene has the closest affinity to B. rapa, and B. nigra, which is consistent with the evolutionary origin doctrine in U’s Triangle Brassica Species (Cheng et al., 2016).

The gene structure diagram from the AOP gene family ( Figure 3A ) revealed that the AOP family is conserved in the number of exons, 53 of the 61 members containing three exons. Despite the similar exon length of most genes, the intron lengths differed significantly. The AOP family was analyzed for conserved motifs using the MEME online tool ( Figure 3B ). In total, 15 conserved motifs of 8-41 aa in length were contained in the AOP gene ( Figure 3C ; Supplementary Table S5 ). In addition, the structural domains of 15 motifs were analyzed using NCBI-CDD online software. The motif 10,4,5 and 7 together form the DIOX_N domain. And the motif 2, 6, 8 and 1 together form the 2OG-FeII_Oxy domain structural of the AOP gene family.To further elucidate the functional element that influences expression of AOP genes, the promoter sequences were analyzed using the PlantCARE database to identify cis-regulatory elements in the promoter region. Nineteen types of stress- and hormone-related cis-acting regulatory elements were detected in the promoters of AOP genes ( Figure 4 ; Supplementary Table S6 ). All 61 AOP genes contained element related to light. The MYB bingding site were detected in 47. Hormone-related elements (MeJA-responsiveness, salicylic acid responsiveness, zein metabolism regulation, abscisic acid responsiveness, auxin-responsive element, gibberellin-responsive element) were detected in more than 45 members of AOP genes. Other results are shown in the Figure 4 .

The green fluorescent protein (GFP) fusion transient expression vector containing the BjuAOP2 gene was injected into tobacco leaves and observed by laser confocal microscopy ( Figure 5 ). Except for BjuAOP2.4, where a green fluorescent signal was observed only in the nucleus, the other four family members were observed to fluoresce in both the cytoplasm and nucleus. Thus, BjuAOP2.1, BjuAOP2.2, BjuAOP2.3, and BjuAOP2.5 were expressed in the cytoplasm and nucleus, and BjuAOP2.4 was expressed only in the nucleus, consistent with the prediction results of subcellular localization.

The expression patterns of five BjAOP2 genes in various tissue parts of B186 and various developmental stages of the tumorous stem were analyzed using real-time PCR ( Figure 6 ). Expression of BjuAOP2 genes was detected in roots, stems, tumorous stems, leaves, flowers and siliques. BjuAOP2 genes had significantly different expression patterns in different tissues ( Figure 6A ). BjuAOP2.2 were more highly expressed in flowers than in other tissues examined ( Figure 6A ), whereas BjuAOP2.1, BjuAOP2.3 and BjuAOP2.4 were more highly expressed in stems than in other tissues ( Figure 6A ). BjuAOP2.5 were more highly expressed in tumorous stems than in other tissues ( Figure 6A ). Analysis of the expression pattern of the BjuAOP2 gene at different stages of tumorous stem development revealed different expression patterns with the expansion of TSM ( Figure 6B ). The expression of BjuAOP2.1, BjuAOP2.2 and BjuAOP2.4 peaked at 7 weeks after sowing before falling back to basal levels. BjuAOP2.3 and BjuAOP2.5 showed low expression level throughout the whole growth period.

BjuAOP2 prokaryotic expression vector was constructed using PET-32a. The protein expression was induced by IPTG, and the BjuAOP2 fusion protein was obtained by the His-tag purification column. In vitro enzymatic activity analysis of five BjuAOP2 fusion proteins was performed using GIB as substrate. As shown in Figure 7 , all five BjuAOP2 proteins successfully catalysed the conversion of GIB to SIN. The standards GIB and SIN were detected at 3.3 min and 4.6 min, where only GIB was detected in the control group. However, both GIB and SIN were detected in the protein elutes of all five members. Therefore, all five BjuAOP2 proteins of TSM have catalytic activity.