Background

Eleutherococcus senticosus (Rupr. et Maxim) Maxim, also known as Siberian ginseng, is a valuable medicinal plant [1]. The extracts of its roots and bark have various physiological effects, such as modulation of the immune response [2]. Among the various bioactive constituents of E. senticosus, triterpenoid saponins, flavonoids, coumarins, and lignans play important roles in regulating these physiological activities [3]. The farnesyl pyrophosphate synthase [4] (FPS), squalene synthase (SS), squalene epoxidase [5, 6] (SE), and mevalonate pyrophosphate decarboxylase [7] (MDD) of E. senticosus are key enzymes that catalyze the biosynthesis of triterpenoid saponins. In previous studies, it was found that DNA methylation in the promoter regions of FPS, SS, SE, and MDD [7, 8] genes in E. senticosus inhibited the synthesis of the mevalonate pathway, thus reducing the accumulation of saponins.

Drought stress is usually considered a negative factor in plants and an important cause of yield loss. However, studies have shown that plants exposed to drought stress accumulate higher concentrations of secondary metabolites [9]. For example, drought stress dramatically decreases spica biomass production but increases rosmarinic acid (RA), ursolic acid (UA), and oleanolic acid (OA) contents in Prunella vulgaris L. [10]. In Salvia miltiorrhiza Bunge, water stress significantly increased the salvianolic acid B yield and decreased that of tanshinone IIA [11]. However, there are no reports on drought stress in the Araliaceae family. Drought stress is a type of adversity stress, in which plants respond to the external environment through various mechanisms. DNA methylation is one of the earliest and most studied mechanisms that regulate genome function and induce plant resistance and abiotic stress adaptation [12]. This provided a new idea for our research. Because the secondary metabolism of medicinal plants is affected by drought stress, they are inevitably subjected to epigenetic modifications. However, it is unclear how plants affect metabolite accumulation through epigenetic modifications under drought conditions.

DNA methylation is an epigenetic modification that plays an important role in regulating gene expression. Methyl-binding domain (MBD) protein is a class of methyl CpG-binding domain protein that selectively bind to fully methylated CpG dinucleotides and participate not only in the recruitment of nucleosome remodeling factors and histone deacetylases, but also in the formation of transcriptional repressor complexes that regulate gene expression through epigenetic modifications [13]. In humans, abnormalities in the MBD genes are strongly associated with the development of cancer and psychiatric disorders [14]. For example, overexpression of MBD1 in the human body is usually associated with carcinogenesis, leading to the silencing of tumor suppressor genes through the formation of the abovementioned complexes [15]. In plants, MBD genes also play important functions: in Arabidopsis thaliana (L.) Heynh. AtMBD4 mediates phosphates uptake [16]. Similarly, AtMBD6 is involved in RNA-mediated gene silencing [17], whereas AtMBD9 is involved in regulating flowering time and shoot branching in A. thaliana, with mutant plants exhibiting an abnormal phenotype of early flowering and branching [18]. These studies on MBD have demonstrated their importance. Therefore, a holistic analysis of EsMBD in E. senticosus could provide a new research perspective for exploring the mechanism of DNA methylation of secondary metabolism-related genes.

Results

Identification and analysis of the physicochemical properties of the EsMBD gene family

Given the Hidden Markov Model of the MBD gene family (PF01429), 20 EsMBD genes were screened out from the whole-genome sequencing data of E. senticosus, and sequentially named EsMBD1-EsMBD20 according to their individual positions on the chromosomes (Table 1). Based on the Hidden Markov Model of the MBD gene family, 10 AeMBD genes were screened from the whole genome sequencing data of Aralia elata (Miq.). Seem and consecutively named AeMBD1-AeMBD10 according to their separate positions on the chromosomes. The results showed that the quantity of amino acids in the EsMBD gene family of proteins was in the range between 166 and 2,015 aa, and the molecular weight ranged from 18,865.60 to 226,144.33 Da, whereas the isoelectric points were below 7.00 except for EsMBD7 (9.27), EsMBD11 (9.39), EsMBD14 (8.68), EsMBD15 (7.08), and EsMBD17 (8.27), indicating that most EsMBD proteins were acidic. Moreover, all EsMBD proteins exhibited hydrophobicity of less than 0 and an instability index greater than 40, indicating that all EsMBD proteins were unstable and hydrophilic.

Table 1 Physicochemical properties of EsMBD gene family proteins
Full size table

Chromosomal localization and collinearity analyses of EsMBD genes

The chromosomal localization analysis of EsMBD genes (Fig. 1A) showed that they were unevenly distributed in 12 of the 24 E. senticosus chromosomes. Chromosome No. 3 contained four EsMBD genes, chromosome 17 contained three EsMBD genes; chromosomes No.1, 8, and 12 each contained two EsMBD genes; and chromosome No.4, 9, 10, 11, 13, 14, and 18 each contained one EsMBD gene. The distribution of EsMBD and tandem repeats on different chromosomes of E. senticosus are shown in Fig. 1A, in which only one pair, EsMBD18 and EsMBD19, were tandem repeat genes.

Fig. 1
figure 1

Chromosomal localization, evolutionary analysis, and collinearity analysis of EsMBD genes in E. senticosus. A: Chromosomal localization of EsMBD genes in E. senticosus. Red markers indicate tandem repeats; B: Collinearity analysis of EsMBD genes of E. senticosus; C: Phylogenetic analysis of EsMBD genes of E. senticosus; and D: Collinearity analysis of interspecies MBD between E. senticosus and A. elata. Red lines indicate the MBD genes

Full size image

The results of the collinearity analysis of EsMBD genes in E. senticosus (Fig. 1B) showed 22,568 genes in the genome, accounting for 62.05% of all genes, and collinearity was observed among the 17 EsMBD genes in the entire family. As shown in Fig. 1B, in the E. senticosus genome, these genes were of collinearity: EsMBD1/EsMBD6, EsMBD1/EsMBD14, EsMBD1/EsMBD15, EsMBD2/EsMBD5, EsMBD2/EsMBD8, EsMBD2/EsMBD16, EsMBD3/EsMBD9, EsMBD3/EsMBD13, EsMBD5/EsMBD8, EsMBD5/EsMBD16, EsMBD6/EsMBD14, EsMBD6/EsMBD15, EsMBD7/EsMBD11, EsMBD8/EsMBD16, EsMBD9/EsMBD13, EsMBD12/EsMBD18, and EsMBD14/EsMBD15, which presented many fragmental repeats among the EsMBD gene family.

The results of the collinearity analysis between the EsMBD genes of E. senticosus and the AeMBD genes of A. elata (Fig. 1D) showed significant collinearity between E. senticosus and A. elata, between EsMBD and AeMBD genes, and between the same EsMBD and different AeMBDs. This indicated that MBD genes evolved with the evolution of species in Araliaceae.

Phylogenetic analysis of MBD

The 20 EsMBD protein sequences identified from the screening and 10 AeMBD sequences from A. elata were used to construct a phylogenetic evolutionary tree, in addition to the 13 AtMBD protein sequences with well-defined functions in A. thaliana. The results (Fig. 1C) showed that the evolutionary tree was divided into eight subfamilies, α, β, γ, δ, ε, ζ, η, and θ, based on the affinities between the different members of the MBD gene families of E. senticosus, A. elata, and A. thaliana. Except for the η subfamily, which contained only EsMBD4, EsMBD20  and AtMBD8, the MBD genes of E. senticosus and A. elata were both localized in the remaining subfamilies. The α subfamily contained 4 genes, EsMBD2, EsMBD5, EsMBD8, and EsMBD16; the β subfamily contained 1 EsMBD17 gene; the δ subfamily contained EsMBD7, EsMBD11, EsMBD12, and EsMBD18; and the γ subfamily contained EsMBD3, EsMBD9, and EsMBD13. The δ subfamily contained EsMBD7, EsMBD11, EsMBD12, and EsMBD18; the ε subfamily contained EsMBD10; and the ζ subfamily contained EsMBD1, EsMBD6, EsMBD14, and EsMBD15. The η subfamily contained EsMBD4 and EsMBD20, whereas the θ subfamily contained only EsMBD19.

Structure of EsMBD genes with analyses of conserved structural domains and motifs of the protein

The structure of EsMBD genes in E. senticosus is shown in Fig. 2A C. Of the EsMBD gene sequences, 70% were within 2,000 bp in length, only one was up to 6,000 bp in length, and only 20% were approximately 1,000 bp in length. Ten genes in the EsMBD gene family contained 5`UTR, 11 genes contained 3`UTR, and 8 genes did not have UTR sequences, namely EsMBD1, EsMBD6, EsMBD7, EsMBD9, EsMBD10, EsMBD13, EsMBD15, and EsMBD20. The EsMBD genes in the same branch of the evolutionary tree also presented similar structures. Ten conserved motifs of EsMBD were identified using the online tool MEME analysis (Fig. 2D) and named as motifs 1–10. Furthermore, EsMBD17 contained only 2 motifs, the fewest in the EsMBD gene family, whereas EsMBD10 contained the most 10 motifs. Over 50% of the EsMBD gene family contained motifs 1, 2, 6, 7 and 9, indicating that these five genes are relatively conserved in this family.

Fig. 2
figure 2

Gene structure, promoter analysis, and conserved motifs of EsMBD in E. senticosus. A: Motif analysis of EsMBD genes in E. senticosus; CDS: DNA sequence corresponding to the protein sequence one by one; UTR: non-coding segment at both ends of the mRNA molecule B: Promoter analysis of EsMBD in E. senticosus; C: Exon-intron structure analysis of EsMBD genes in E. senticosus; D: Conserved motifs of EsMBD genes in E. senticosus; Motif: conserved sequences with biological functions

Full size image

Analysis of cis-acting elements of the EsMBD gene promoter

Sequence analysis of the promoter region of the EsMBD gene family in E. senticosus (Fig. 2B) revealed that four phytohormone response elements were identified in addition to the conventionally transcriptional regulatory elements of TATA-box and CAAT-box, including the methyl jasmonate (MeJA), gibberellin, salicylic acid, abscisic acid. Some environment-related transcriptional regulatory elements were also included, such as light response, anaerobic induction, drought inducibility, and low temperature. The number of light-response-related cis-acting elements was the highest in the promoter regions of all EsMBD genes (Supplementary Table S1). Additionally, cis-acting elements related to the regulation of flavonoid biosynthesis were found in the promoter region of EsMBD4. Therefore, we hypothesized that EsMBD might be involved in mediating the physiological regulation of E. senticosus and the synthesis of flavonoid substances by binding to methylated cytosine under different light conditions.

Expression analysis and quantitative real-time PCR validation of EsMBD gene family

The expression of EsMBD genes under drought stress in E. senticosus is shown in Fig. 3A. The expression of different EsMBD genes varied with different water contents: EsMBD3, EsMBD12, EsMBD14, EsMBD17, and EsMBD18 showed low expression at 90% relative water content, whereas EsMBD6 and EsMBD9 were highly expressed at 30% relative water content, and EsMBD1, EsMBD8, EsMBD10, and EsMBD16 were highly expressed at 90% relative water content. The expression of EsMBD13 gradually decreased with increasing water content. The expression of EsMBD genes in the different saponin content groups is shown in Fig. 3B. EsMBD14 and EsMBD17 were significantly expressed in the low and medium saponin content groups but were not significantly expressed in the high saponin content group, whereas EsMBD1, EsMBD5, and EsMBD16 were not significantly expressed in the low and medium saponin content groups but were significantly expressed in the high saponin content group.

Fig. 3
figure 3

EsMBD gene expression in different E. senticosus transcriptomes. A: EsMBD gene expression under drought stress. B: EsMBD gene expression in different saponin content groups. L, low saponin content; M, middle saponin content; H, high saponin content. Three replicates were performed for each group of samples

Full size image

To verify the reliability of the transcriptome data, all differentially expressed genes were screened for qRT-PCR validation, except for four EsMBDs, which were not expressed in the transcript. The results showed that the trends of other genes were consistent with the transcriptome sequencing data, indicating that the transcriptome sequencing results were reliable (Fig. 4).

Fig. 4
figure 4

Quantitative Real-time PCR validation of EsMBD genes in E. senticosus. FPKM: Fragments Per Kilobase of exon model per Million mapped fragments

Full size image

Molecular docking of EsMBD proteins to DNA sequences

Homology modeling of EsMBD proteins was performed using the SWISS-MODEL website. The GMQE ranged from 0.02 to 0.21, QMEANDisCo Global values ranged from 0.41 ± 0.11 to 0.72 ± 0.11, and sequence similarity ranged between 22.22 and 73.24%. Molecular docking was performed at the HDOCK SERVER website (Fig. 5). Among the eight best predictions acquired, each EsMBD protein was successfully docked to the methylated DNA sequence, indicating that EsMBD could bind to fully methylated CpG dinucleotides and thus play a regulatory role.

Fig. 5
figure 5

Model of EsMBD protein docking with methylated DNA sequences. MBDs are depicted in yellow, methylated DNA sequences are displayed in blue, and 5-methylcytosine is located between the third and ninth positions

Full size image

Subcellular localization of EsMBD proteins

Subcellular localization can reveal the working positions of proteins at the cellular level. The WoLF PSORT website (https://wolfpsort.hgc.jp) was used to predict the subcellular localization of EsMBD proteins. The results showed that most EsMBD proteins were predicted to be located in the nucleus, EsMBD12, EsMBD16, EsMBD17, EsMBD18, and EsMBD20 were predicted to be located in the cytoplasm, and EsMBD2 and EsMBD5 were predicted to be located in the nucleus and cytoplasm (Supplementary Table S2). To verify the prediction of subcellular localization, we selected 6 EsMBD proteins involved in molecular docking to fuse Green Fluorescent Protein (GFP) and expressed them instantaneously in Allium cepa L. Five EsMBD proteins were localized in the nucleus and cytoplasm, among which EsMBD2 and EsMBD6 were expressed in both the nucleus and cytoplasm and EsMBD12, EsMBD16, and EsMBD18 were expressed only in the cytoplasm (Fig. 6).

Fig. 6
figure 6

Subcellular localization of the five EsMBD proteins. Green fluorescence (GFP) signals were observed under a laser scanning confocal microscope to indicate the subcellular localization of EsMBD-GFP fusion proteins. The ruler in the image is 50 μm

Full size image

Discussion

In the model plant A. thaliana, mutations are not completely random throughout the genome because of epigenetic modifications and other effects, and the mutation probability of important genes is lower than that of non-important genes [19]. MBD genes are inseparable from DNA methylation, and have been identified and validated in various plants. In these plants, MBDs are involved in a several physiological processes and stress regulation [20]. MBD genes are not only important for plant growth and development but also pose great relevance to humans [21]. In this study, we conducted a comprehensive bioinformatics analysis of EsMBD genes in E. senticosus, identified EsMBD family members by sequence similarity, and analyzed their evolutionary amplification patterns, sequence characteristics, expression motifs, and homology to better model them, which not only provided a basis for further studies of MBD genes, but also laid a foundation for epigenetic modifications affecting the genome and physiological activities of the organism.

Evolution of MBD genes

Gene duplication drives the evolution of genetic systems by providing a rich source of raw material. Polyploidization is a major mechanism of environmental adaptation and species formation in organisms [22], and the major mechanism of biological genome structure and gene evolution is through a single gene, chromosome, and whole-genome duplication (WGD) [23]. The sequencing of the genome of E. senticosus showed that in addition to the γ genome-wide triploidization event common to all core members of the family Pentacanthaceae, E. senticosus also experienced two WGD events [24]; thus, it could be hypothesized that the E. senticosus MBD gene family expanded during these two WGD events. Further analysis of the collinearity between E. senticosus and chromosomal localization information revealed that the genomic collinearity level of E. senticosus was as high as 62.05%, and there was also a significant collinearity among EsMBD gene families, while the EsMBD gene families were scattered in arrangement on chromosomes, and only one tandem repeat gene existed. Therefore, it could be inferred that the main method of amplification of EsMBD gene families was gene fragment duplication. The collinearity analysis showed significant collinearity between E. senticosus and A. elata, which are both species of Araliaceae, while there was also significant collinearity between the EsMBD and AeMBD genes. Therefore, it can also be inferred that MBD genes evolved within the Araliaceae species before eventually turning into interspecies-specific MBD genes through amplification and duplication.

Phylogeny of the MBD gene family

MBD proteins are key determinants of DNA methylation status [25]. Twenty EsMBD genes and ten AeMBD genes were identified in E. senticosus and A. elata. Together with the 13 AtMBD genes in A. thaliana, a phylogenetic tree was constructed and branched according to a previously constructed phylogenetic tree between Brassica napus L. and A. thaliana [26]. Although there was only AtMBD8, EsMBD4, and EsMBD20 in one branch, the MBD gene family members of E. senticosus and A. elata were identified in all the other branches. This might be attributed to the fact that the relatives of E. senticosus being closer to each other than those of A. thaliana and A. elata. The phylogenetic tree was also verified using the inter-species collinearity map of E. senticosus and A. elata. MBD gene family members have different abilities to bind to methylated DNA [27]. Therefore, we speculated that the functional differences in MBD genes among different branches might be significant, indicating that the functional differences among MBD proteins were gradually realized during the evolutionary process. This result is consistent with the conclusions of previous studies that MBD genes have a long evolutionary history, with MBD genes existing and beginning to differentiate as early as the formative stages of monocotyledons and dicots [26].

Structure of MBD genes and protein-conserved motifs

UTRs are closely associated with gene expression motifs that enhance plant perception of stimuli from developmental and environmental regulatory factors [28]. In the MBD gene family of E. senticosus, UTRs were present in 12 genes, accounting for 60% of all EsMBDs. EsMBDs lacking UTR structures were in the same branch or adjacent ones, implying that the expression motifs of the MBD gene family shared some similarity between the same and adjacent branches, whereas the structures and differences of genes were more significant between different branches. Studies have shown that dehydration stress can induce the extension of the 3’UTR transcript, and that these stress-induced 3’UTR extensions have a new function: regulating the long non-coding RNA of its adjacent genes, which confirms the importance of UTR length in plant adaptation to stress conditions [29]. The UTR lengths of different MBD genes in E. senticosus were different, which also indicated that different genes may have functional differences and play different roles under various stresses.

EsMBD sequences exhibit significant conservation and inter-group specificity across adjacent branches, and conserved structural sequences are prerequisites for conserved biological functions [30]. The EsMBD motifs were similar in structure between adjacent branches but differed significantly between branches that were further apart from each other. It was speculated that this was due to the different functional differentiation of MBD genes during the process of evolution. Moreover, the 20 EsMBD protein sequences obtained from the screening also presented conserved motifs in most EsMBDs, which is consistent with the results of earlier studies comparing conserved motifs of MBD in B. napus [26]. These conserved motifs of different MBD genes in E. senticosus also showed different similarities.

Promoter analysis of the MBD gene family

A promoter is a small region of a DNA sequence that responds to various transcription factors and triggers specific gene expression [31]. Cellular metabolism in organisms is largely regulated at the transcriptional level, and promoters are responsible for basic regulation of transcription initiation [32]. The promoters of the MBD genes in E. senticosus were found to contain a large number of cis-acting elements, and the quantity of different cis-acting elements in adjacent branches showed some similarity, suggesting that there is a certain degree of functional differentiation among the MBD genes in E. senticosus, which is consistent with a previous hypothesis that there is functional differentiation of MBD genes during the evolutionary process [26]. Among the various cis-acting elements of the promoter, light response elements were the most abundant, followed by anaerobic induction elements, indicating that the EsMBD genes of E. senticosus were subjected to maximum regulation by light and anaerobic conditions (Supplementary Table S1). It has been demonstrated that different light qualities like light intensities regulated the growths of roots, stems, and leaves as well as the synthesis of secondary metabolites in E. senticosus [33, 34]. For example, red light was found to induce the synthesis of Eleutheroside E [33] and LED lighting promoted the development of fine roots in E. senticosus seedlings [34]. In addition, shade treatment enhances photosynthesis in E. senticosus by promoting the growth and accumulation of secondary metabolites [35]. The EsMBD promoter regions, which contain substantial light-responsive and cis-acting elements involved in the regulation of flavonoid synthesis, were also identified. We speculated that the EsMBD gene of E. senticosus could participate in regulating the physiological activities of E. senticosus under certain light stress conditions, thereby affecting the primary and secondary metabolism of E. senticosus.

EsMBD gene expression analysis

Drought stress alters plant growth and developmental processes, causing morphological changes in plant height and leaf size. Plants also respond to drought stress by reducing water loss and enhancing water absorption capacity [36]. It has been suggested that drought induced increases in DNA methylation may be one of the main mechanisms by which plants respond to drought stress [37]. In maize, the MBD genes of Zea mays L. showed a similar expression motif under drought stress [38]. Interestingly, EsMBD genes in E. senticosus also exhibited similar expression motifs under drought stress, with some EsMBD gene expression showing similar up- or down-regulation trends at different relative water contents, and most EsMBD genes were up-regulated under relatively dry conditions. In a previous study by Cui et al. [7], the saponin content of E. senticosus decreased with increasing DNA methylation ratio. Meanwhile, EsMBD14 and EsMBD17 were significantly expressed in the low and medium saponin content groups but were not significantly expressed in the high saponin content group. This indicates that EsMBD14 and EsMBD17 were not significantly expressed in the low methylation ratio group. This suggests that EsMBD14 and EsMBD17 are involved in regulating saponin synthesis under drought stress through DNA methylation.

Molecular docking model analysis and subcellular localization of EsMBD genes

The MBD protein of E. senticosus was revealed using homology modeling on the SWISS-MODEL website. The Global Model Quality Estimate [39] (GMQE) is a quality assessment criterion that integrates the properties of the target template alignment and template structure to improve the reliability of the quality estimate. The QmeandisCo global score [40] is the average QMEANDisCo score for each residue. Based on the QMEANDisCo score estimated for substantial models, it provides an error estimate and represents the standard deviation between the QMEANDisCo global score and underlying facts. Because the reliability of the predictions depends on the model size, the provided error estimates are calculated in line with models of sizes similar to the inputs. The global model quality estimates, together with the QMEANDisCo global score demonstrated both the accuracy and reliability of the prediction models. In homology modeling, EsMBD protein models were derived from the MBD proteins of Homo sapiens, and A. thaliana showed the highest sequence similarity of 73.24%. This confirmed our previous hypothesis that MBD genes evolve and are amplified within the E. senticosus genome. Among all EsMBD protein homology models, the size and morphology varied constantly. Combined with the results of motif analysis, it can be speculated that there are different degrees of functional differentiation among EsMBDs, whereas the study in Glycine max (Linn.) Merr. [41] verified that most MBDs were differentiated during early evolution, which was also consistent. MBD proteins can specifically bind to methylated cytosines, thus causing DNA methylation silencing in methylated genes [9]. All those screened from E. senticosus were able to dock successfully with methylated DNA after homology modeling, which not only verified the basic mode of action of MBD proteins but also predicted EsMBD binding to methylated functional genes as one of the pathways of DNA methylation regulating secondary metabolism in E. senticosus. Among all models, EsMBD2 and EsMBD5 presented the best quality and highest sequence similarity. To further verify the function of EsMBD proteins, we examined the subcellular localization of 5 EsMBDs that responded strongly to drought stress. Previous studies have shown that AtMBD5 protein is located in the nucleus and interacts with AtRAN3 protein [42]. In our study, all EsMBDs were located in the nucleus and cytoplasm, consistent with the results of previous studies. EsMBD proteins are widely involved in the physiological metabolism of E. senticosus and play important roles in its stress response and secondary metabolic processes. This also showed that E. senticosus is likely to affect the synthesis of saponins and other substances through epigenetic modifications under drought stress.

Conclusion

In this study, 20 EsMBD genes were identified in E. senticosus, which were mainly obtained by fragment duplication and found to be unevenly distributed on 12 of the 24 chromosomes. The gene structure and conserved motifs of EsMBD genes exhibited different degrees of similarity, with the greatest degree between adjacent branches of the evolutionary tree and substantial light-, hypoxia-, and drought-related cis-acting elements in the EsMBD promoter. Under drought stress, EsMBD genes of E. senticosus showed similar expression motifs. Comparing the transcriptome data and Quantitative Real time PCR results of different saponin content groups of E. senticosus, it was found that different EsMBD genes had different sensitivity to drought stress, among which EsMBD14 and EsMBD17 were most sensitive to drought stress. After molecular docking model analysis and subcellular localization, it can be inferred that functional differentiation occurred during the evolutionary process of EsMBDs. Based on the above analysis and transcriptome data, the cytosine of the functional gene promoter in the E. senticosus genome can be effectively methylated under various stress conditions. EsMBD14 and EsMBD17 combine with methylated cytosine, resulting in changes in the expression of functional factors, thus affecting the secondary metabolism process of E. senticosus.

Materials and methods

Identification and physicochemical characterization of the EsMBD gene family

Whole genome sequencing data for E. senticosus [24] were downloaded from the CNGB Sequence Archive (CNSA) database (https://ftp.cngb.org/pub/CNSA/data3/CNP0001682/CNS0348054/CNA0019321/). Whole genome sequencing data of A. elatae [43], which is the same species of Araliaceae, were downloaded from the Dryad database (https://datadryad.org/stash/dataset/doi:https://doi.org/10.5061/dryad.69p8cz937). The Hidden Markov Model (PF01429) for MBD gene was downloaded from the PFAM database (http://pfam.xfam.org/). The structural domains of the MBD proteins were searched for using the HMMER program, and the E value was set to 0.001. Furthermore, the online Simple Modular Architecture Research Tool (SMART) and NCBI Batch CD-Search tool were used to confirm the conserved domains of all the candidate MBD protein sequences. We retained only sequences containing significant MBD domains in all results; therefore, the sequence of E. senticosus was eliminated. MBD gene family members were identified from the whole genomes of E. senticosus and A. elata, and named according to their different positions on the chromosome (Supplementary Table S3). The obtained EsMBD protein sequences were submitted to ProtParam (https://web.expasy.org/protparam/) and analyzed for physicochemical properties, such as amino acid content, molecular weight, isoelectric point, hydrophilicity, and instability coefficient.

Chromosomal localization and collinearity analysis of EsMBD genes

For collinearity analysis, E. senticosus genome files and E. senticosus genome files with A. elata were used for sequence alignment using the Basic Local Alignment Search Tool (BLAST), with a cut-off E value of 1.0 e− 10. The MCScanX software was used to perform collinearity analysis based on BLASTP. TBtools [44] (https://github.com/CJ-Chen/TBtools) software was used to visualize collinearity.

Phylogenetic analysis of MBD genes

The 13 AtMBD protein sequences of A. thaliana were downloaded from The Arabidopsis Information Resource (https://www.arabidopsis.org/). Twenty EsMBD proteins identified in E. senticosus, 12 AeMBD proteins from A. elata, and 13 AtMBD proteins from A. thaliana were used to construct the phylogenetic tree. We used the MBD gene family of A. thaliana as an outgroup, according to previous methods [26], and constructed a phylogenetic tree with A. elata to obtain more accurate results. ClustalW was used for multiple sequence alignment of MBD protein sequences with default parameters. The alignment results were used to construct an NJ phylogenetic tree using MEGA 11.0. Bootstrap resampling (100) was used to assess the reliability of interior branches. The bootstrap value was set at 1,000. The online tool EvolView (https://www.evolgenius.info/evolview/) was used to construct a phylogenetic tree.

Gene structure and sequence analysis of the EsMBDs

The exon-intron structure analysis was performed using TBtools software by inputting gene annotation GFF files. The online MEME Suite was used to confirm the motifs of E. senticosus EsMBD protein sequences with the following parameters: maximum number of ten motifs and optimum width of 6–50.

EsMBD gene promoter’s cis-acting element analysis

The 2,000 bp nucleotide fragment upstream of the start codon of each EsMBD gene in the E. senticosus genome was screened as the promoter sequence. The cis-acting element analysis of these EsMBD genes was performed using the online software PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/).

Expression analysis of EsMBD

The transcriptome sequencing data of drought-stress-treated E. senticosus were downloaded from the NCBI database (Accession: SRR19962743-SRR19962757). Based on the results of a previous study by Cui et al. [7], transcriptome data of E. senticosus with different saponin contents (Accession: SRX13417593-SRX13417601) were downloaded from the NCBI database. Based on the different transcriptome data, differentially expressed EsMBD genes were screened (Supplementary Table S4-S5). The expression values (FPKM values) were used to compare the transcript levels of MBD in E. senticosus. The transcripts with log2FC ≥ 1.5 or ≤ -1.5 and FDR < 0.05 were selected as transcripts that were differentially expressed. Heatmaps of EsMBD gene expression were drawn using the Heatmapper website (http://www.heatmapper.ca/), based on the expression of differentially expressed genes.

To verify the accuracy of the expression levels obtained from RNA-Seq analysis, the screened EsMBD was validated by qRT-PCR. Primers for qRT-PCR amplification of differentially expressed EsMBD genes were designed using the Prime Premier software (5.0) (Supplementary Table S6). The specificity of the qRT-PCR primers was determined by analyzing the solubility curves. The qRT-PCR was performed using Talent qPCR PreMix (SYBR Green) (TIANGEN, Beijing) on an Applied Biosystems 7900HT PCR System (THERMO FISHER SCIENTIFIC, Waltham, MA, USA) with the glyceraldehyde-3-phosphate dehydrogenase gene (GAPDH) used as an internal reference gene [45]. Three biological replicates were used for each sample. The total reaction system was 10.0 µL with 1.0 µL enzyme, 2.9 µL Nuclease-Free Water, 0.3 µL each primer, 0.5 µL cDNA, 1.0 µL ROX Refernence Dye. The reaction conditions were as follows: pre-denaturation at 95 °C for 3 min, denaturation at 95 °C for 5 s, annealing at 55 °C for 10 s, and supplemental extension at 72 °C for 15 s. The reaction was completed after 40 cycles. EsMBD gene expression was calculated using SDS 2.4 software through the 2−ΔΔ Ct method [46].

Molecular docking of EsMBD proteins to DNA sequences

Homology modeling of EsMBD proteins was performed using SWISS-MODEL (https://swissmodel.expasy.org/) with the Global Model Quality Estimate [39] (GMQE), QMEANDisCo [40] global score, and sequence similarity ratio. The methylated DNA sequence models (PDB ID: 5UZ2) were downloaded from the RCSB PDB website (https://www.rcsb.org/). EsMBD proteins and methylated DNA sequences were docked on the HDOCK SERVER website (http://hdock.phys.hust.edu.cn/). The molecular docking results were obtained using the UCSF ChimeraX website (https://www.cgl.ucsf.edu/chimerax/).

Vector construction

The cloning and restriction primers were designed using Prime premier 5.0 (Supplementary Table S7) [39]. The EsMBD sequences were cloned using cloning primers and enzyme digestion primers, using E. senticosus cDNA as a template and Taq DNA polymerase (TIANGEN, Beijing). The amplification conditions were as follows: pre-denaturation at 94 °C for 5 min, denaturation at 94 °C for 30s, annealing at 55 °C for 30s, extension at 72 °C for 5 min, and an additional extension at 72 °C for 5 min. A lethal fast cloning kit (TIANGEN, Beijing) was used to connect the cloned sequences with the pLB vectors and was introduced into competent Escherichia coli Top10 cells (TIANGEN, Beijing, China). Sequencing was performed using a SinoGenoMax (Beijing, China). The recombinant pLB plasmids were extracted using a TIANprep Mini Plasmid Kit (TIANGEN, Beijing, China). The pHG vectors and recombinant pLB plasmids were digested overnight at 37 °C with BamHI-HF, PstI-HF, and SacI-HF (NEW ENGLAND BIOLABS, Beijing, China) restriction enzymes. We used T4 DNA ligase (NEW ENGLAND BIOLABS, Beijing, China) to connect EsMBD genes with pHG vectors overnight at 16 °C to construct recombinant EsMBD-GFP expression vectors. The total reaction system was 20.0 µL with 1.0 µL T4 DNA Ligase, 1.0 µL Nuclease-Free Water, 2.0 µL T4 DNA Ligase Buffer, 11.0 µL Insert DNA (20 ng / µL), 5.0 µL pHG vector.

Subcellular localization of EsMBD proteins

We selected Allium cepa L. as the infection receptor. The recombinant expression vectors were transformed into Agrobacterium tumefaciens (Biomed, Beijing, China) and cultured on LB solid medium containing kanamycin (100 ng/ml) and rifampicin (50 ng/ml). The positive colonies were selected and cultured overnight in LB liquid medium at 28 °C. The cells were then centrifuged at 3000 rpm for 10 min to collect the cells, which were resuspended in LB liquid medium without antibiotics. The medium contained MgCl2 (10 mmol/L) and acetosyringone (100 mmol/L), and was adjusted to an OD600 of 1.0.

To prepare LB solid culture medium, the outer skin of the onion was peeled off without antibiotics. Fresh and thick scales were taken and a small square with an area of 1 cm2 on the inner skin was cut using scissors. The onion skin was torn off with tweezers, placed downward close to the mesophyll in LB solid culture medium without antibiotics, and sealed with a sealing film for dark culture at 28 °C for 24 h. After 24 h, the inner skin of A. cepa was removed and soaked in LB heavy suspension solution at room temperature for 20 min. After 20 min, tweezers were used to collect the epidermis, filter out the bacterial liquid, lay on new LB solid medium without antibiotics, and incubated for 16–24 h at 25 °C with a photoperiod of 16 h/8 h. Small pieces of onion inner skin were removed, shaken, washed with clean liquid LB medium to remove attached A. tumefaciens, placed on a slide, and observed under a laser scanning confocal microscope (Leica, Germany).