Protocorm-like bodies (PLBs) were generated using a previously described in vitro regeneration protocol (Zeng et al., 2023), and cultivated in Murashige and Skoog (MS) medium (Murashige and Skoog, 1962) containing 0.5 mg·L^-1 6-benzyladenine (6-BA; Aladdin, Shanghai, China), 0.2 mg·L^-1 1-naphthaleneacetic acid (NAA; Aladdin), 20% banana puree (prepared fresh), 2% (w/v) sucrose (Aladdin) and 0.7% agar to induce proliferation (Zeng et al., 2023). Cultures were placed under controlled environmental conditions (25°C; 12-h photoperiod; 50 µmol m^-2 s^-1 cold fluorescent white light). Following the acclimatization of in vitro-regenerated plants (Zeng et al., 2023), the leaves of 36-month-old plants with uniform growth were sprayed with 0.1 mM methyl jasmonate (MeJA) (98% purity, Sigma-Aldrich, St. Louis, MO, USA), which was dissolved in trace ethanol (99.5% purity, Merck, Darmstadt, Germany). Leaves were sprayed every 30 d for a total of 90 d to investigate the effect of MeJA on WSP content and transcript levels of D. officinale genes. The treated stems were collected at three time points (30, 60, and 90 d), with three replications per time point. Plants sprayed with distilled water supplemented with trace ethanol served as the control (CK). Each treatment group included three replications. Samples were stored immediately at -80°C. Frozen samples of plants that had been treated with 0.1 mM MeJA for 30 d were sent to Biomarker Technologies (Beijing, China) to determine the transcript levels of D. officinale genes.

RNA-seq libraries were carried out as reported (Si et al., 2022), and the cDNA products accord with quality assessment were sequenced through the BMKCloud platform (www.biocloud.net). High-quality clean reads were cherry-picked by removing adapters and low-quality sequences, compared with the chromosome-level D. officinale genome, and normalized to fragments per kilobase per million fragment-mapped reads (FPKM). Differentially expressed genes (DEGs) were identified using DESeq2 version 1.38.3 based on a false discovery rate (FDR) < 0.05, and |log₂FoldChange| > 1. Gene Ontology (GO), Eukaryotic Orthologous Groups (KOG), and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed using the BMKCloud server with the P-value less than 0.05.

The Hidden Markov Model (HMM) of Dof proteins’ structural domain (PF02701) was downloaded from password-encrypted Interpro (https://www.ebi.ac.uk/interpro/). The whole genome sequences of D. nobile (accession no. 94219), D. huoshanense (accession no. 154293) and D. officinale (accession no. 142615) which were downloaded from IMP (https://www.bic.ac.cn/IMP) were analyzed by HMM Search of TBtools v2.118 (Chen et al., 2023), using an E-value > 0.05 as the filtering criterion to remove redundancy. After removing incomplete sequences in the conserved structural domains, using Conserved Domains Database (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi), 22 DoDof proteins, 29 DhDof proteins and 29 DnDof proteins ( Supplementary Tables S1 - S3 ) were finally identified. In addition, the 80 Dof proteins were analyzed at the ExPASy online website (https://www.expasy.org/) to determine the size, molecular weight (MW), isoelectric point (pI), instability index, aliphatic index, and grand average of hydropathicity (GRAVY) for each protein. Finally, the subcellular localization of the Dof proteins were predicted in WoLF PSORT (https://wolfpsort.hgc.jp/).

To explore the evolutionary relationships between the 80 Dof proteins, they were compared with 36 A. thaliana Dof proteins and 30 Oryza sativa Dof proteins that were downloaded from PlantTFDB ( Supplementary Table S4 ). They were subjected to multiple sequence alignment using ClustalX v2.1 (http://www.clustal.org/). A phylogenetic tree of these 80 proteins was constructed with MEGA v11.0 (https://megasoftware.net/) using the neighbor-joining method (Saitou and Nei, 1987) with a bootstrap value of 1000 and was visualized using Chiplot (https://www.chiplot.online/).

Collinearity analysis of the Dof proteins was performed and visualized in TBtools using the One Step MCScanX Super Fast modulation. The genome files and annotation files of A. thaliana and O. sativa, as representative dicot and monocot model plant, respectively, were downloaded from CNCD-NGDC (https://download.cncb.ac.cn/gwh/Genome/Plants/). Collinearity analyses were performed in TBtools with the Dual Synteny Plot.

Total RNA was isolated from different organs of D. officinale (roots, leaves, flowers, and stems), MeJA-treated leaves (0.1 mM, 30 d), and control leaves using the Quick RNA Isolation Kit (0416–50 GK, Huayueyang, Beijing, China). After ensuring the purity and concentration of RNA by 1% agarose gel electrophoresis (Bio-Rad Laboratories, Hercules, CA, USA) and a NanoDrop2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), the Evo M-MLV RT Kit II (Accurate Biology, Hunan, China) was used to obtain first-strand cDNA. The quantitative real-time polymerase chain reaction (qRT-PCR) reaction, which contained 5 μL 2× iTaq™ universal SYBR^® Green (Bio-Rad Laboratories), 1 μL cDNA, 0.4 μL of each primer (10 μM), and RNase-free water supplemented to 10 μL, was run in an Applied Biosystems 7500 (Applied Biosystems, Foster City, CA, USA). ELONGATION FACTOR 1 ALPHA (EF-1α) was used as the reference gene (Yu et al., 2021b). Relative gene expression of the 22 DoDof genes was calculated by the 2^-ΔΔCt method (Livak and Schmittgen, 2001). The quantitative PCR primers, listed in Supplementary Table S5 , were generated by Integrated DNA Technologies (https://sg.idtdna.com/).

Multiple enzyme-encoding genes involved in the WSP metabolic pathway were previously identified (Yu et al., 2021b; Si et al., 2022). The correlations between these enzyme-encoding genes and the DoDof genes was assessed as Pearson’s correlation coefficients with SPSS v. 27.0 (IBM, Armonk, NY, USA), and a co-expression network was generated in Cytoscape v3.10.0 (https://cytoscape.org/).

The coding sequence of DoDof4 (without the stop codon TGA) was cloned using the PrimeSTAR Max premix (Takara, Dalian, China). To determine the subcellular localization of DoDof4, DoDof4 was inserted into the SpeI and BamHI sites of a 35S promoter-driven pHB vector (Haq et al., 2021) containing yellow fluorescent protein (YFP) with In-Fusion Cloning Kit (Takara). After verification by sequencing (Zhejiang Sunya Co., Hangzhou, China), recombinant plasmid pHB-DoDof4-YFP and the negative control pHB-YFP were transformed into Agrobacterium tumefaciens GV3101 (Weidi Biotechnology Co., Shanghai, China) through a previously published freeze-thaw protocol (Yu et al., 2021b), and used to infect the one-month-old Nicotiana benthamiana leaves. After 48 h, subcellular localization of YFP in N. benthamiana leaves, and the staining leaves with 4’,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich), were observed on a confocal microscope (Zeiss, Oberkochen, Germany) under a 488 nm excitation filter. To evaluate the transcriptional activation or repression of DoDof4, coding sequence of DoDof4 without the termination codon TGA was cloned into pGBKT7 (Takara) to construct the pGBKT7-DoDof4 as previously reported (Yu et al., 2021b). Yeast AH109 (Weidi) transformed with pGBKT7-Empty and pGBKT7-DoDof4 were grown on a synthetic dropout (SD) without tryptophan (SD/-Trp) medium and SD without adenine, tryptophan, and histidine (SD/-Ade-Trp-His) medium respectively at 29°C. X-α-Gal (Coolaber, China) was additionally dropped to displayed color reaction. pGBKT7-Lam and pGBKT7-P53 were selected as negative and positive controls, respectively.

The DoDof4 coding sequence without the termination codon TGA was inserted into pCAMBIA1301 vector (CAMBIA, Canberra, Australia) to overexpress the DoDof4 gene. DoDof4 gene into pCAMBIA2300 (CAMBIA) using sgDoDof4-F and sgDoDof4-R primers, which were designed on the CRISPRdirect website (https://crispr.dbcls.jp/) to construct pCAMBIA2300-DoDof4-CRISPR/Cas9 (DoDof4CRISPR). The verified plasmids, pCAMBIA1301 (overexpression-empty), pCAMBIA2300-CRISPR/Cas9 (knockout-empty), pCAMBIA1301-DoDof4 (DoDof4OE), and DoDof4CRISPR were transferred into 25 mL of A. tumefaciens GV3101 following the method described in 2.8. A single colony was selected and sub-cultured in yeast mannitol medium containing 0.1 mg·mL^-1 rifampicin and 0.1 mg·mL^-1 kanamycin (Macklin, Shanghai, China) for 2 d until OD₆₀₀ was 0.8, then centrifuged at 5000 rpm·min^-1. MS, which was used as infiltration solution (1 mL) that contained 1 mol·L^-1 acetosyringone (Sigma-Aldrich), 1.95 g·mL^-1 2-morpholinoethanesulfonic acid (MES; Sigma-Aldrich), and 0.1% Tween-20 (Sigma-Aldrich), was injected to a depth of 0.5 cm into the center of each D. officinale PLB. PLBs were subsequently soaked in the infiltration solution for 4 d (25°C, 24h darkness) to generate pCAMBIA1301, DoDof4OE, pCAMBIA2300-CRISPR/Cas9 and DoDof4CRISPR lines, respectively. pCAMBIA1301 and DoDof4OE transgenic lines were selected on PLB proliferation medium containing 30 mg·L^-1 hygromycin B (Yeasen). pCAMBIA2300-CRISPR/Cas9 and DoDof4CRISPR transgenic lines were selected on PLB proliferation medium containing 25 mg·L^-1 kanamycin. After 30 d of transient transfection, PLBs were collected and stored immediately at -80°C. Using the technique outlined in 2.7., total RNA was extracted from transgenic PLBs. DoDof4 expression was examined by qRT-PCR (see 2.6.). The primers used for qRT-PCR are listed in Supplementary Table S6 .

WSPs in D. officinale samples (100 mg of leaves or PLBs) were ultrasonically extracted using an established method (Yu et al., 2018). WSP content was determined on an UV-3600 multi-directional spectrophotometer (Shimadzu, Kyoto, Japan) at 488 nm using the phenol-concentrated sulfuric acid method (Ogura et al., 2023) and was expressed as mg·g^-1. To investigate the composition of WSPs, rapid hydrolysis and pre-column derivatization were executed using 0.5 mol·L^-1 1-phenyl-3-methyl-5-pyrazolopyrimidone (PMP; Sigma-Aldrich), as described by Yu et al. (2019). Thereafter, 5 μL of filtered supernatant was injected into an Agilent 1260 (Santa Clara, CA, USA) equipped with an Ultimate XB-C18 column (Welch Materials, Shanghai, China) to detect monosaccharide content. Glucose (Sigma-Aldrich) and mannose (Sigma-Aldrich) were used as internal standards, and their content was expressed as mg·g^-1.

All experiments in this paper were performed as three or more independent replicates. The mean ± standard deviation (SD) of the data were calculated and graphs were created in GraphPad Prism v. 8.0.2 (GraphPad, Boston, MA, USA). Statistical analyses were performed in SPSS Statistics v. 27.0. In all graphs, *** represents a very highly significant difference (P < 0.001), ** represents a highly significant difference (P < 0.01), and * represents a significant difference (P < 0.05). RNA-Seq analysis (including transcriptome sequencing, library preparation, transcriptome sequencing, and transcriptome analysis) was conducted at BMKCloud (https://www.biocloud.net/).

Various hormones, as well as biotic and abiotic stresses affect plant health and growth, and ultimately influence the regulation of a number of related genes. MeJA is a signaling molecule that stimulates stress responses and induces the synthesis of a wide range of metabolites. We took D. officinale as a model system to explore the effect of MeJA on Dendrobium species. WSP content increased in MeJA-treated plants, when compared to the control, even more so when treatment period was extended from 30 to 90 d. These findings confirmed that MeJA promotes the accumulation of WSPs in D. officinale ( Figure 1 ).

To examine the transcriptomic changes in plants following MeJA treatment, PE150 sequencing was performed on six D. officinale samples (MeJA-treated and distilled water-treated with trace ethanol, each in three replicates) using the Illumina NovaSeq6000 sequencing platform. After sequencing quality control, a total of 37.35 Gb of clean data was obtained, including 5.79 Gb of clean data in each sample, and the Q30 base percentage was at least 92.28%. The number of clean reads in each sample ranged from 19,347,994 to 22,825,395. Compared to the high-quality D. officinale genome, the alignment efficiency of clean reads was between 87.66 and 89.77%. Based on the transcriptomic data, a total of 5945 differentially expressed genes (DEGs) were screened, including 2981 up-regulated DEGs and 2954 down-regulated DEGs ( Figure 2A ), accounting for about half of all DEGs. The functions of these DEGs were analyzed by clusters of eukaryotic orthologous groups (KOG) annotation, revealing a total of 24 categories that were annotated based on the orthologous and evolutionary relationships between eukaryotes ( Figure 2B ). Based on KOG, only ‘general function prediction’ was most annotated functional classification, followed by ‘post translation modification, protein turnover, chaperones’, and ‘signal transduction mechanisms’, while ‘nuclear structure and cytoskeleton’ had fewest annotations. Intuitively, ‘energy production and conversion’, ‘carbohydrate transport and metabolism’, and ‘secondary metabolites biosynthesis, transport and catabolism’ were closely related to WSP biosynthesis. Volcano plots were used to show differential changes in the expression of DEGs, indicating large and significant differences in up-regulated DEGs ( Figure 2C ). KEGG enrichment analysis showed a total of 5945 DEGs that were annotated in KEGG pathways, and 20 pathways had highest P-values ( Figure 2D ). Among them, up-regulated DEGs were mainly clustered into the ‘plants hormone signal transduction pathway’, followed by ‘carbon metabolism’. For carbon metabolism, ‘glycolysis/gluconeogenesis’, ‘pentose phosphate pathway’, ‘fructose and mannose metabolism’, and ‘carbon fixation in photosynthetic organisms’ were associated with WSP biosynthesis. For the down-regulated DEGs, the only pathway was ‘photosynthesis-antenna protein’. In the biological pathways from GO enrichment analysis, the largest group was biological processes (BP), followed by molecular functions (MF) ( Figure 2E ). The number of up-regulated and down-regulated DEGs differed for each function. In general, the DEGs were concentrated in ‘cellular and metabolic processes’, ‘cellular anatomical entity’, and ‘catalytic activity and binding’. In addition, 16 TFs were predicted among the DEGs ( Figure 3A ), and seven TFs (bHLH, AP2/ERF, C2H2, bZIP, NAM/ATAF1/2/CUC2 (NAC), Golden2, ARR-B, Psr1 (GARP), B3) from over 20 DEGs accounted for 7/20 of all TFs. The number of predicted DEG members in MYB, WRKY, C3H, Dof, Trihelix, Nuclear Factor Y (NF-Y), and Fatty Acyl-CoA Reductase 1 (FAR1) TF families ranged from 10 to 20, and the number of members in TCP and Whirly TF families was less than 10 ( Figure 3B ). The WRKY, bZIP, AP2/ERF, MYB, and bHLH TF families were previously identified in D. officinale (He et al., 2017; Zeng et al., 2021; Li et al., 2022), whereas the families of TFs that the D. officinale Dofs belong to remain unclear. Furthermore, expression of the enzyme-encoding genes involved in WSP biosynthesis was positively correlated with Dof genes (Pearson’s R² > 0.8) ( Figure 3C ). Therefore, the Dof genes were studied in more detail to appreciate how they regulate the accumulation of WSPs.

After screening the D. nobile, D. huoshanense and D. officinale genome following the characteristic Dof structural domain (PF02701), and removing redundant and incomplete sequences, 80 Dof genes were identified. Among these Dof proteins, D. nobile and D. huoshanense have an identical number (n = 29) of Dof genes, and the number of DoDof genes were lowest (n= 22). Based on a distribution of individual Dof genes on the corresponding chromosomes, they were sequentially named DhDof1-DhDof29, DnDof1-DnDof29, and DoDof1-DoDof22.

To better understand the roles and mechanisms of these Dof proteins, we systematically predicted and analyzed their physicochemical properties ( Supplementary Table S7 ). Our results revealed that three Dendrobium species exhibited similar levels in terms of average size ( Supplementary Figure S1A ), MW ( Supplementary Figure S1B ), pI ( Supplementary Figure S1C ), instability index ( Supplementary Figure S1D ), aliphatic index ( Supplementary Figure S1E ), and GRAVY ( Supplementary Figure S1F ) for their Dof proteins. The size in the Dof proteins ranged from 109 (DnDof7) to 626 (DhDof1), and their molecular weights ranged between 12.05 (DnDof7) and 71.12 kDa (DhDof1), and their pIs ranged from 5.17 (DnDof14) to 9.79 (DoDof3). The instability index of all Dof proteins exceeded 40, which indicates that they are highly stable. Combined with aliphatic index and GRAVY, all Dof proteins were classified as hydrophilic proteins.

To explore the evolutionary relationships within the Dof gene family of Dendrobium, a phylogenetic tree of AtDof, OsDof, DhDof, DnDof and DoDof genes was shown in Figure 4 . Dof genes were divided into four groups according to their phylogenetic relationships ( Figure 4 ). Group A contained the largest number of Dof genes (n = 48), groups B/C/D contained 34, 32, and 32 Dof genes. Further analysis revealed that the Dof genes from Dendrobium species clustered closely together, rather than randomly distributed with those from A. thaliana and O. sativa. Multiple sequence alignment showed that all Dof genes consist of 50 highly conserved amino acids that form a CX₂CX₂₁CX₂C sequence at the N-terminus, namely the Dof structural domain ( Supplementary Figure S2 ). This suggests that Dof genes are relatively conserved during the evolution of Dendrobium species.

Supplementary Figure S3 showed the location of the Dof genes in D. huoshanense ( Supplementary Figure S3A ), D. nobile ( Supplementary Figure S3B ) and D. officinale ( Supplementary Figure S3C ) chromosomes. The 29 DhDof genes were unevenly mapped on the 15 chromosomes while no DhDof genes localized on Chr3/10/12/15. Chr8 possessed the highest number (n = 5) of DoDof genes (DoDof12-DoDof 16). The 29 DhDof genes exhibited random distribution across 15 chromosomes, mirroring the distribution pattern observed in D. huoshanense. Among all chromosomes, Chr3 exhibited the highest density (n = 5) of DhDof genes (DhDof2-DhDof6) while Chr2/7/13/16 lacked DhDof genes. The 22 DoDof genes were unevenly distributed across the 13 chromosomes. Chr19 had the highest number (n = 5) of DoDof genes (DoDof18-DoDof 22) while five chromosomes (Chr7/8/10/15/16) had no DoDof genes. An interspecies (i.e., between three Dendrobium species) analysis revealed that D. officinale and D. nobile displayed the highest level of collinearity (n = 32), and D. officinale and D. huoshanense had 21 pairs ( Supplementary Figure S4A ). This indicates that D. officinale and D. nobile have a closer relationship than D. officinale and D. huoshanense. In addition, there were 25 collinear genes when compared with A. thaliana, and even fewer (n = 8) between D. officinale and O. sativa ( Supplementary Figure S4B ).

The expression of genes in plants is closely related to their functions, so the expression of four tissues in D. officinale (flower, leaf, stem, and root) were analyzed ( Supplementary Table S8 ). The expression pattern revealed that 22 DoDof genes exhibited a distinct organ-specific expression. These were divided further into 5 groups I-V ( Figure 5A ). In group I, 6 DoDof genes exhibited high expression in many tissues. For example, DoDof12 exhibited high expression in leaf and stem, DoDof14 had high expression in root and leaf, DoDof17 showed high expression in flower and root, and DoDof4/6/8 had high expression in flower, root, and stem. In group II, eight genes (DoDof3, DoDof5, DoDof13, DoDof15, DoDof16, DoDof18, DoDof20, and DoDof22), and in group III, DoDof2, DoDof7, and DoDof19 displayed high expression in root and stem, and relatively uniform expression in flower, root, and stem. DoDof10 and DoDof11, which formed part of group IV, displayed low expression in all tissues. In group V, the remaining DoDof genes were expressed mainly in flower and had similarly low expression in other tissues. Overall, most genes were highly expressed in flower, but many genes were also expressed in root and stem. Highly expressed DoDof genes were also identified in each tissue ( Figure 5B ). Compared with the expression patterns of the 22 DoDof genes, DoDof17 displayed the highest expression in flower and root, and DoDof14 displayed the highest expression in leaf. Expression in stem was dominated by DoDof2 and DoDof4, then by DoDof6, and finally DoDof8 and DoDof12. A Venn diagram ( Figure 5C ) revealed that no genes were highly expressed in all tissues, while three genes (DoDof4/6/8) were highly expressed in flower, stem, and root.

In order to assess the relationships between DoDof genes and MeJA, the relative expression levels of DoDof genes were examined by qRT-PCR after 30 d of MeJA treatment to validate the corresponding transcriptomic data. The change in patterns of DoDof genes was consistent with the trend of the transcriptome data, which indicates that the transcriptomic data was reliable ( Figure 6 ). After MeJA induction, the expression of 14 DoDof genes (DoDof1/2/3/4/5/6/8/11/16/17/18/20/21/22) was up-regulated. The expression of DoDof9 did not change before and after induction, and the remaining seven genes showed different degrees of down-regulated expression after induction. DoDof4/6 were highly up-regulated, 2.44- and 3.79-fold, respectively, after MeJA treatment. Very few genes of same subfamilies showed same expression trend. For example, relative expression of DoDof1/11 was lower before and after induction, and relative expression of DoDof7/14/19 decreased after induction. These findings suggest that MeJA has a significant effect on expression of DoDof genes and may thus play a role in the regulation of plant WSP biosynthesis.

DoPMM (KF195558), DoGMP (KF195560, KP203853), DoGMT (KY851106, MF278268), DoUGP (KF711982), DoUGE (KX54540), and DoCSLA (KM980199, KM980200, KP003920, KM980201, KM980202, KF195561, KP205040, KP205041), which were highly expressed in flowers and stems ( Figure 7A ), were identified. According to the 2020 edition of the Chinese Pharmacopoeia (Chinese Pharmacopoeia Commission, 2020), the stems of D. officinale are usually used as medicine, while the flowers, roots, and leaves are non-medicinal parts. Analysis of the expression of DoDof genes in different tissues revealed that DoDof2/4/6/8/12 had highest expression in stem, so they were selected for correlation analysis with these 16 enzyme-encoding genes by SPSS Statistics version 27.0 ( Figure 7B ). DoDof4 had a notable co-expression relationship with 8 enzyme-encoding genes (KM980199, KM980202, KF195561, KF195558, KX545406, KF711982, KP203853, and MF27968). DoDof2 was correlated with 6 enzyme-encoding genes (KP203853, KX545406, KF711982, MF27968, KF195561, and KF195558), DoDof8 was highly and positively correlated with KF711982, no enzyme-encoding gene had a high positive correlation with DoDof6 or DoDof12.

DoDof4 showed 99.65% similarity with the original sequence. DoDof4 open reading frame (ORF) consisted of 876 bp encoding 292 amino acids ( Figure 8A ) and the predicted molecular weight was 27.84 kDa. The protein sequence contains one Dof domain at the N-terminus indicating that cloned DoDof4 belongs to the Dof family. Subcellular localization analyses revealed that pHB-DoDof4-YFP was localized in the nucleus ( Figure 8C ). Furthermore, pGBKT7-DoDof4 plasmid and positive control grew well on SD/-Ade-Trp-His medium, which degraded the X-α-Gal and changed the color of yeast colony from white to blue. pGBKT7-Empty and negative control did not grow on the SD/-Ade-Trp-His medium ( Figure 8B ), suggesting that DoDof4 was a transcriptional activator.

Plasmids (pCAMBIA1301, pCAMBIA2300-CRISPR/Cas9, DoDof4OE, and DoDof4CRISPR) were then separately transformed into the D. officinale PLBs ( Figure 9A ). After 4 d of transient transformation, DoDof4OE transgenic lines were greener than other lines ( Figure 9B ). After 30d, RNA was extracted from the transgenic lines, and their cDNA was reverse transcribed. The qRT-PCR results showed that the relative expression of the DoDof4OE line was higher than that of the pCAMBIA1301 line ( Figure 9C ). Sequencing results revealed a 3 bp deletion in DoDof4CRISPR ( Supplementary Figure S5 ). These results validated successful genetic transformation.

WSP content in the DoDof4OE transgenic line was 1.5-fold higher than that in the pCAMBIA1301 transgenic lines ( Figure 9D ). HPLC detection revealed that the WSPs in D. officinale were predominantly composed of glucose and mannose. The contents of glucose and mannose in DoDof4OE showed a sharp increase relative to pCAMBIA1301. Furthermore, the mannose/glucose ratio in DoDof4OE was 1.87-fold higher than that in pCAMBIA1301 and the mannose/glucose ratio in DoDof4CRISPR was the lowest. These results imply that DoDof4 may significantly stimulate the biosynthesis of WSPs in D. officinale, possibly by regulating mannose transportation ( Supplementary Figure S6 ). Therefore, the relative expression of KM980199, KM980202, KF195561, KF195558, KX545406, KF711982, KP203853, and MF278268 were characterized in pCAMBIA1301 and DoDof4OE transgenic PLBs to explore how DoDof4 regulates the biosynthesis of WSPs. The relative expression of KM980199 and KP203853 increased in the DoDof4OE line ( Figure 9E ), indicating that DoDof4 may regulate WSP biosynthesis by converting mannose-1P to GDP-mannose.