Assembly and comparative analysis of the initial complete mitochondrial genome of Verbascum thapsus (Scrophulariaceae): a medicinally and ecologically significant species

Abstract

Background

Verbascum thapsus belongs to Scrophulariaceae (a family with ~ 62 genera and ~ 1830 species) and has significant medicinal and ecological value. However, its mitochondrial genome (mtDNA) has not been reported previously. This study aimed to sequence, assemble, and annotate the complete mtDNA of V. thapsus, to clarify its structural characteristics, evolutionary dynamics, and phylogenetic relationships.

Results

The assembled V. thapsus mtDNA is a circular molecule of 478,020 bp, with a GC content of 45.02%. It contains 36 protein-coding genes (PCGs), 21 tRNA genes, and 3 rRNA genes. Notable features include abundant repetitive sequences (simple sequence repeats and dispersed repeats), and 468 RNA editing sites (mainly C-to-U conversions). Comparative analysis revealed homologous sequences between mitochondrial and chloroplast genomes, accounting for 3.02% of the mtDNA length. Phylogenetic analysis (based on 14 conserved mitochondrial PCGs) placed V. thapsus within the order Lamiales, supporting its taxonomic position.

Conclusions

This study presents the first complete mtDNA of V. thapsus, which sheds light on mitochondrial genome evolution in Scrophulariaceae. The identified genomic rearrangements, codon usage biases, and RNA editing events form a foundation for future phylogenetic and functional research on V. thapsus and related species.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-026-08964-2.

Introduction

The Scrophulariaceae family, commonly known as “figwort”, mainly comprises herbs and shrubs, with a small number of trees [1]. This family has approximately 62 genera and about 1830 recognized species [2]. The phylogenetic relationships within the Scrophulariaceae remain an unresolved issue in angiosperm systematics [3]. To date, only the chloroplast genomes of three genera within the family have been investigated: Buddleja [4], Scrophularia, [5] and Verbascum [6, 7], with one genomic data available for Buddleja alternifolia [8], while no mitochondrial genomic data have been reported. The genus Verbascum, commonly referred to as “Mullein”, is herbaceous and consists of around 360 species that are widely distributed in the temperate regions of Europe, Africa, and Asia [9–12]. The primary hotspot for the diversity of this genus is Turkey, with 235 existing species [13] and new species to science also being identified and described (Ghahremaninejad et al. 2014). Morphologically, the genus has simple and alternate leaves, and the basal leaves often form a rosette. The flowers are arranged in terminal spikes, racemes, or panicles, and the corolla is usually yellow, rarely purple [1, 14]. Extracts, decoctions, and infusions of Mullein have a long—standing history of use in traditional medicine. The leaves are utilized as a diuretic, purgative, expectorant, and sedative, and the flowers possess mucolytic and expectorant properties [15, 16].

The delimitation and attribution of the genus Verbascum is challenging [17] and have long been the focus of debate among researchers. In the study of the family Scrophulariaceae, the systematics of the genus Verbascum has received relatively limited attention, making it one of the least understood genera in terms of infrageneric categorization. In 1933, Murbeck employed seed morphology to classify this genus into two groups: the group with longitudinally furrowed seeds (Aulacospermae Murb.) and the group with obliquely furrowed and concavely notched seeds (Bothrospermae Murb.) [18, 19]. In 1978, Huber—Morath proposed an alternative classification system for the genus Verbascum in his treatment of the Turkish taxa of this genus. He re—grouped 243 Verbascum species (including 129 hybrids) into a total of 13 groups from A to M and concluded that all the Turkish mullein species belong to Bothrospermae Murb. [20]. Another analysis from the former Soviet Union once again separated the genera Verbascum and Celsia and documented 51 species [19]. A regional classification published in Iran in 1981 combined the genera Verbascum and Celsia, however, it did not offer any infrageneric classification for the 49 species and 4 additional hybrids [20]. In contrast, the European taxonomic literature records nearly 100 species of Verbascum, including those named after Celsia [21].

In recent times, an increasing number of molecular phylogenetic methods have been utilized to comprehend the phylogenetic relationships within the genus Verbascum. In 2014, 41 taxa from two distinct genera were analyzed using the nuclear internal transcribed spacer region (ITS) and three non—coding chloroplast DNA regions (trnY—T, psbA—trnH, and trns—G) [22]. This research corroborates the monophyletic branching of the genus Verbascum and the establishment of a sister relationship with the genus Scrophularia, which is consistent with prior morphological findings regarding the family Scrophulariaceae [18]. Nevertheless, this study failed to offer supportive evidence for the infrageneric categorization of Verbascum. Another systematic analysis of the genus distributed in the Iranian region conducted by Sotoodeh 2015 also exhibits a high degree of similarity to the results obtained by Ghahremaninejad et al. 2014. More recently, researchers have also investigated and clarified the chloroplast genomes of 28 organisms in the purple Mullein (V. phoeniceum L.) and lyre leaf mullein (V. chinense L.), as well as within the Lamiales, and the taxonomic position of the genus Verbascum. The study by Bi et al. 2020b demonstrated that the genus Verbascum forms a monophyletic branch that is sister to the genus Scrophularia. In contrast, another study by He et al. 2020 classified Verbascum as sister to Scrophularia, and Buddleja as a sister clade of both Verbascum and Scrophularia, which contradicts previous studies indicating that the genus Verbascum is monophyletic. Consequently, the systematic position of Verbascum within the family Scrophulariaceae remains controversial and unresolved, and additional molecular data is required to address the existing problem.

Following over four decades of sequencing and analysis, it has become increasingly evident that the mitochondrial genome (mitogenome) represents an evolutionarily dynamic entity, displaying remarkable diversity in size, structure, and content across eukaryotes [23–25]. When examining the extremes of mitochondrial genome diversity, the mitochondrial genomes of plants stand out prominently. Plant mitochondrial genomes encompass nearly the entire spectrum of size variation observed in eukaryotic mitochondrial genomes. For instance, they range from the 66 kb genome of the hemiparasitic plant mistletoe (Viscum scurruloideum) [26] to the > 11 Mb mitochondrial genome of Silene conica, which holds the record for size [27]. In 2011, approximately 60 plant mitochondrial genomes had been sequenced. Half of these were from angiosperms (primarily crops), while the remainder were from other land plants and algae that have evolved at rates distinct from those of angiosperms [28]. By this time, angiosperm mitochondrial genomes were already well—known for their generally large and highly variable sizes [29], their proclivity for genome rearrangements [30], and their highly variable gene content [31]. This early genomic revolution, marked by the accumulation of a diverse array of mitochondrial genes in land plants and green algae, has facilitated the determination of the extent to which these angiosperm mitochondrial traits have spread to other land plants. It has also provided a more in—depth context for discerning the trajectory of mitochondrial gene evolution across the plant kingdom [32]. Over the past 12 years, with the advancement in sequencing technology, a greater number of plant mitochondrial genomes have been sequenced, either fully or partially. The majority of these newly sequenced mitochondrial genomes are from angiosperms. In addition to several economically significant crops, an increasing quantity of wild plant resources have also been sequenced. More than half of the 64 angiosperm orders within the current APG IV (Angiosperm Phylogeny Group 4) system [33] have at least one sequenced representative in GenBank.

Plant mitochondrial genomes are being increasingly utilized in phylogenetic studies as they carry relevant genetic information. With the improvement of sequencing and assembly technologies, the application of mitochondrial information for plant phylogenetic analysis has received growing attention [34–36]. As an independent genetic system, external to the cell nucleus, the mitochondrial genome is characterized by rapid evolution and a high copy number. It has been extensively employed in studies related to species identification, intraspecific matrilineal evolution, interspecific phylogeny, phylogenetic evolution, and genetic diversity [37]. However, the number of complete plant mitochondrial genomes is significantly lower than that of chloroplast genomes and animal mitochondrial genomes. This is due to the incorporation of extraneous DNA, the increase in repetitive fragments, and the frequent addition and subtraction of fragments within the genome. The limited number of plant mitochondrial genomes makes it challenging to study the major patterns of variation in multiple mitochondrial genomes at both inter—and intraspecific levels. Some scholars have hypothesized that the major pattern of mitochondrial genome variation should be genome loss events subsequent to large—scale ancestral expansions [38], yet they were unable to draw conclusive results from the limited mitochondrial genome data available. Specifically, for the taxon Verbascum, and even within the family Scrophulariaceae, there have been no relevant mitochondrial genome studies to date. Therefore, mitochondrial genome studies in Scrophulariaceae and Verbascum are urgently required.

Materials and methods

Sample collection, total DNA extraction and sequencing

The materials utilized in this study were obtained from the public database of the National Center for Biotechnology Information (NCBI) and the newly collected specimen of Verbascum thapsus (HGW—2029) from Yunnan, China (Table 1). The voucher specimen was deposited in the Herbarium of the Wuhan Botanical Garden (HIB). The mitochondrial genome data of additional species involved were retrieved from the NCBI Public Data.

Table 1.

Summary of mitochondrial genome sequences of plant species

Familiy	Genus	GeneBank
Scrophulariaceae	Verbascum thapsus	PX_570624
Orobanchaceae	Pedicularis chinensis	NC_072955.1
	Pedicularis kansuensis	NC_072932.1
	Castilleja paramensis	NC_031806.1
	Aeginetia indica	NC_069194.1
	Rehmannia glutinosa	OM397952.1
	Christisonia kwangtungensis	OM219025_7.1
Lamiaceae	Platostoma chinense	NC_069573.1
	Salvia miltiorrhiza	NC_023209.1
	Rotheca serrata	NC_049064.1
	Pogostemon heyneanus	MK728874.1
	Scutellaria tsinyunensis	MW553042.1
	Scutellaria barbata	NC_065025.1
	Scutellaria franchetiana	NC_065026.1
	Ajuga reptans	NC_023103.1
	Ajuga ciliata	MT075725_6.1
	Vitex trifolia	NC_065806.1
Plantaginaceae	Aragoa cleefii	OK514182.1
	Aragoa abietina	OK514181.1
	Plantago ovata	NC_072975.1
Gesneriaceae	Boea hygrometrica	NC_016741.1
	Haberlea rhodopensis	MH757117.1
Lentibulariaceae	Utricularia reniformis	NC_034982.1
	Genlisea tuberosa	OK274069.1
Oleaceae	Hesperelaea palmeri	NC_031323.1
	Osmanthus fragrans	NC_060346.1
	Jasminum sambac	NC_069589.1
	Olea europaea	MW262896.1
	Ligustrum quihoui	MN723864.1
	Chionanthus rupicola	MG372115.1
	Forsythia suspensa	OQ534870.1
Solanaceae	Nicotiana tabacum	NC_006581.1
	Solanum melongena	NC_050334.1

Leaves were obtained either through silica—gel drying after collection from natural habitats in the field or from live plants cultivated in the botanical garden. Total DNA was extracted employing a modified CTAB method [39], and the quality of DNA extraction was evaluated via NanoDrop spectroscopy and 1% agarose gel electrophoresis. The Truseq Nano DNA HT Sample Preparation Kit (Illumina USA) was utilized for library construction, in accordance with the manufacturer's guidelines. Sequencing was carried out using double—ended PE—150 bp. The DNA samples were subjected to sonication and fragmentation to a size of 350 bp, processed, and further amplified by PCR for Illumina sequencing. The PCR products were purified using the AMPure XP purification kit, and the size distribution of the libraries was analyzed using the Agilent 2100 Bioanalyzer and quantified through real—time PCR. Finally, sequencing based on the Illumina HiSeq 2500 (Illumina, USA) platform was conducted at Wuhan Benagen Technology Company. The identical DNA sample was also employed for single—molecule real—time sequencing of Nanopore—based ONT nanopore electrical signals, which generated a total of 10.72 Gb of sequence reads, with an average read length of 89.85 kb for sequencing.

Assembly and annotation of mtDNA

In the present study, the sequencing results from Illumina (short—reads) and Nanopore (long—reads) were combined to assemble the mitochondrial genome of Verbascum thapsus. Initially, the long—reads were processed. Assembly was carried out in Flye [40] software with default parameters, and graphical assembly results in GFA format were acquired. For all the contigs assembled in fasta format, libraries were constructed using makeblastdb. Subsequently, the BLASTn [41] program was executed, with conserved plant mitochondrial genes from Arabidopsis thaliana (NC_037304) [42], Liriodendron tulipifera (NC_021152.1) [43], and Scutellaria tsinyunensis (MW553042.1) [44] serving as query sequences. Long—contig fragments containing mitochondrial genomes were identified using the parameters “-evalue 1e—5—outfmt 6—max_hsps 10—word_size 7—task blastn—short”. The GFA files were visualized via Bandage [45] software. Eventually, the long—reads data were aligned to graphical mtDNA fragments using BWA software [46], and mitochondrial long—reads were extracted for further analysis of repetitive sequence regions in the graphical plant mtDNA. Following that, the short—reads were processed. The long—contig fragments obtained in the previous steps were utilized to align the short—reads to the mitochondrial long—contigs using the BWA software, and the mitochondrial short—contigs were exported. Finally, the ultimate mitochondrial genome sequence was assembled based on the obtained mitochondrial short—contigs and long—contigs through a mixing approach using Unicycler [47].

To annotate the protein—coding genes of the Verbascum mtDNA, the mtDNA of Arabidopsis thaliana [42], Liriodendron tulipifera [43], and Scutellaria tsinyunensis [44] were chosen as reference genomes. The primary reasons for selecting these three datasets as reference genomes are as follows: (1) Arabidopsis thaliana has been comprehensively studied, and the boundaries of each gene are precisely annotated; (2) The plant mitochondrial genome comprises 24 core genes and 17 variable genes. The variable genes exhibit significant variation among different species and also experience different degrees of loss. Liriodendron tulipifera possesses all 17 variable genes, making it more accurate for annotating variable genes; (3) Scutellaria tsinyunensis belongs to the Lamiaceae family and is closely related to Verbascum thapsus of the Scrophulariaceae family, as they both belong to the same Lamiales order. The genes of mtDNA were then annotated using the Geseq software [48]. The tRNAs of mtDNA were annotated using tRNAscan – SE [49] software, and the rRNAs of mtDNA were annotated using BLASTn [41] software. For each mitochondrial genome, potential errors and inaccuracies in the annotations were manually corrected using Geneious [50] software.

Codon usage analysis of mtDNA

Codon preference was investigated and analyzed in mitochondrial DNA (mtDNA) analysis using the Relative Synonymous Codon Usage (RSCU) metric. The protein—coding sequences of the genome were acquired using the software Phylosuite [51]. The codon preference of the protein—coding genes (PCGs) of mtDNA was then analyzed, and the RSCU values computed using the software Mega6 [52]. The resulting RSCU patterns were visualized using an interactive RSCU plot generated with the RSCU-Plot Shiny app (https://pcg-lab.shinyapps.io/RSCU-Plot/) [53–56].

Analysis of repeated sequences of mtDNA

Forward, reverse, complementary and palindromic sequences were detected using the online software REPuter [57], with specific parameters set to a minimum size of 20 bp and a maximum size of 200 bp and an E-value of less than 0.01. Tandem repeats were identified using the Tandem Repeat Finder [58]. The corresponding parameters were set as follows: maximum period size = 500, minimum alignment score to report repeat = 80, indels = 7, mismatch = 7, maximum TR array size = 2, alignment parameters of match = 2. Finally, the identification of simple repeat sequences was accomplished on MISA Perl script [59], and the thresholds for the number of repeats between single and hexanucleotides were determined to be 12, 6, 4, 3, 3, 3, 3. To further characterize SSRs in distinct genomic compartments, we separately analyzed coding regions (CDS) and intergenic spacer (IGS) regions using cpstools (https://github.com/awanshri/cpstools) [60]. Specifically, SSR identification was performed with the command cpstools SSRs -i < input_file > -k 10,5,4,3,3,3, where the “-k” parameter defines the minimum repeat units for mono‑, di‑, tri‑, tetra‑, penta‑, and hexanucleotide SSRs, respectively. This allowed us to compare SSR abundance and characteristics between CDS and IGS regions. Lastly, The results were visualized using the Circos package [61].

Structure elucidation of mtDNA

The raw long-reads data were assembled using the Flye [40] software with default parameters and the graphical mtDNA files in GFA format generated by Flye were obtained. To resolve duplicate regions in the obtained graphical genome, long-contigs from mitochondria were compared to duplicate sequences using the BWA [46] software to determine the pathways supported by more long-contigs. The same method was applied for the rest of the contigs that contained multiple connectors, which in turn led to the derivation of the most likely genomic composition of mitochondrial genes.

Migration of mitochondrial and chloroplast genome sequences

The mtDNA of Verbascum thapsus was employed as the database through makeblastdb library construction, and the chloroplast genome of V. thapsus (HGW—2029) [62] assembled in the aforementioned process was utilized as the query sequence. Homologous fragments were analyzed using the BLASTn [41] software with the following parameters: -evalue 1e—10, -word_size 7, -outfmt 6. Ultimately, fragments with homology exceeding 80% were selected from the results, and then the results were visualized and presented via the Circos package [61].

Phylogenetic tree construction based on mtDNA

The DNA sequences of 14 conserved mitochondrial protein-coding genes (PCGs) were employed to construct a phylogenetic tree for 33 species within eight angiosperm families. The mitochondrial genome sequences of specific plant species are presented in Table 1. The common PCGs included atp1, atp4, ccmC, ccmFC, ccmFN, cob, cox2, cox3, matR, nad1, nad2, nad5, nad6, and rps13. Two mitochondrial genomes of the Solanaceae family were designated as outgroups. The acquisition of protein-coding genes was carried out in Phylosuite [51, 63], and the gene matrix obtained after acquisition was aligned in MAFFT [64] with default parameters. To minimize the influence of poorly aligned loci on phylogenetic inference, TrimAI [65] was utilized for trimming subsequent to sequence alignment. The entire process was visually inspected and manually adjusted using MEGA [52] and Geneious [50] software. Subsequently, the gene matrix was further concatenated in Phylosuite to obtain the data matrix utilized for phylogenetic tree construction. Based on the AIC (Akaike Information Criterion) model, the most appropriate model for the data matrix was determined in ModelFinder [66]. The maximum likelihood (ML) tree for this matrix was constructed using IQ-TREE [67], and the bootstrap support (BS) of each branch was obtained based on 10,000 repetitions of ultrafast bootstrap analysis [68, 69]. Finally, the ITOL [70] software was employed to visualize the results of the phylogenetic analysis.

Analysis of mitochondrial RNA editing events

The type sequences of the various protein—coding genes (PCGs) numbered in the mtDNA of the species were employed as input files for the prediction of C—to—U RNA editing sites of mitochondrial PCGs using Deepred – mt [71]. This tool is based on the Convolutional Neural Networks (CNN) model for prediction and shows high accuracy in comparison to previous prediction tools. All results with probability values greater than 0.9 were retained.

Collinearity Analysis

Utilizing the BLAST [41] program, the BLASTN results of pairwise comparisons of individual mitochondrial genomes were acquired, and homologous sequences with a length exceeding 500 bp were retained as conserved collinear blocks for constructing the Multiple Synteny Plot. Based on sequence similarity, the Multiple Synteny Plot of V. thapsus and its relatives was generated using the source program of the MCscanX [72] software.

Results

Characteristics of the mtDNA of V. thapsus

The main structure of the mtDNA of V. thapsus (PX_570624) is a single cyclic molecule (Fig. 1). After excluding the repetitive regions using Nanopore data, the predominantly present one cyclic contig with a total length of 478,020 bp and a GC content of 45.02% was obtained. Annotation of the mtDNA of V. thapsus showed that a total of 36 unique PCGs (including a total of 24 mitochondrial core genes and 12 non-core genes), 21 unique tRNA genes (trnF-GAA presents three copies), and 3 rRNA genes were annotated. The core genes included 5 ATP synthase genes (atp1, atp4, atp6, atp8, and atp9); 9 NADH dehydrogenase genes (nad1, nad2, nad3, nad4, nad4L, nad5, nad6, nad7, and nad9); 4 cytochrome C biogenesis genes (ccmB, ccmC, ccmFC, and ccmFN); 3 cytochrome C oxidase genes (cox1, cox2, and cox3); 1 protein transport subunit gene (mttB); 1 maturase gene (matR), and 1 Cytochrome b gene (cob). Non-core genes included 4 ribosomal protein large subunit genes (rpl2, rpl5, rpl10, rpl16); 6 ribosomal protein small subunit genes (rps3, rps4, rps10, rps12, rps13, rps14); and 2 succinate dehydrogenase genes (sdh3, sdh4) (Table 2).

Fig. 1.

The circular map of the Verbascum thapsus mitochondrial genome. Note: Different colored boxes represent different types or uses of genes. Genes displayed outside the circle are transcribed counterclockwise, while genes displayed inside the circle are transcribed clockwise

Table 2.

Mitochondrial genome coding genes of Verbascum thapsus

Group of genes	Name of genes
ATP synthase	atp1, atp4, atp6, atp8, atp9
NADH dehydrogenase	nad1, nad2, nad3, nad4, nad4L, nad5, nad6, nad7, nad9
Cytochrome b	cob
Cytochrome c biogenesis	ccmB, ccmC, ccmFC, ccmFN
Cytochrome c oxidase	cox1, cox2, cox3
Maturases	matR
Protein transport subunit	mttB
Ribosomal protein large subunit	rpl2, rpl5, rpl10, rpl16
Ribosomal protein small subunit	rps3, rps4, rps10, rps12, rps13, rps14
Succinate dehydrogenase	sdh3, sdh4
Ribosome RNA	rrn5, rrn18, rrn26
Transfer RNA	trnC-GCA, trnD-GUC, trnE-UUC, trnF-GAA (× 3), trnfM-CAU, trnG-GCC, trnH-GUG, trnI-CAU, trnI-GAU, trnK-UUU, trnL-CAA, trnM-CAU, trnN-GUU, trnP-UGG, trnQ-UUG, trnS-CGA, trnS-GCU, trnS-UGA, trnV-GAC, trnW-CCA, trnY-GUA

The number in the brackets represents the number of copies of the gene, e.g. (× 3) means there are three copies

Codon usage analysis of mtDNA

Codon usage analysis was conducted on 36 unique PCGs of V. thapsus mitochondria, and the usage of codons by individual amino acids is shown in Fig. 2. Codons with relative synonymous codon usage (RSCU) exceeding 1 indicate that, it is used more frequently in amino acid encoding. As shown in Table 3, in addition to the start codon AUG and tryptophan (UGG), both of which have RSCU values of 1, there was a general codon usage preference in mitochondrial PCGs. For example, alanine (Ala) has a high usage preference for GCU and has the highest RSCU value among mitochondrial PCGs at 1.58, followed by arginine (Arg) with a usage preference for AGA and proline (Pro) with a usage preference for CCU, which both have an RSCU value of 1.53. It is worth noting that the maximum RSCU value for phenylalanine (Phe) is less than 1.2 implying weak codon usage preference.

Fig. 2.

Relative Synonymous Codon Usage analysis of the mitochondrial genome of Verbascum thapsus

Table 3.

Relative synonymous codon usage of individual amino acid pairs in the mitochondrial genome

Amino	Codon 1	Codon 2	Codon 3	Codon 4	Codon 5	Codon 6
	RSCU	RSCU	RSCU	RSCU	RSCU	RSCU
Ala	GCU	GCA	GCC	GCG
	1.58	1.01	0.89	0.52
Arg	AGA	CGA	CGU	AGG	CGG	CGC
	1.53	1.36	1.18	0.73	0.65	0.55
Asn	AAU	AAC
	1.37	0.63
Asp	GAU	GAC
	1.34	0.66
Cys	UGU	UGC
	1.21	0.79
End	UAA	UGA	UAG
	1.32	0.88	0.79
Gln	CAA	CAG
	1.49	0.51
Glu	GAA	GAG
	1.4	0.6
Gly	GGA	GGU	GGG	GGC
	1.45	1.26	0.73	0.57
His	CAU	CAC
	1.49	0.51
Ile	AUU	AUC	AUA
	1.35	0.84	0.81
Leu	UUA	UUG	CUU	CUA	CUC	CUG
	1.44	1.22	1.22	0.92	0.7	0.51
Lys	AAA	AAG
	1.2	0.8
Met	AUG
	1
Phe	UUU	UUC
	1.16	0.84
Pro	CCU	CCA	CCC	CCG
	1.53	1.1	0.73	0.64
Ser	UCU	UCA	AGU	UCC	UCG	AGC
	1.43	1.13	1.04	0.98	0.81	0.61
Thr	ACU	ACC	ACA	ACG
	1.39	1.03	1	0.58
Trp	UGG
	1
Tyr	UAU	UAC
	1.51	0.49
Val	GUU	GUA	GUC	GUG
	1.24	1.19	0.79	0.79

Repeat sequence analysis

Repetitive sequence analysis of V. thapsus mtDNA identified a total of 103 SSRs, with monomeric and dimeric forms accounting for 47.57% of the total SSRs. 15 thymine (T) monomeric repeats accounted for 60.00% of the 25 monomeric SSRs. No hexameric SSRs were detected (Fig. 3A, Table S1). The SSR sets identified by MISA and cpstools were highly consistent: MISA detected 103 unique SSR loci, while cpstools generated 104 records. The extra record corresponds to an (ACTAG)₃ repeat located at positions 155,026–155,040, which spans the boundary between the intergenic spacer (IGS) region trnK-UUU-matR and the coding region of the matR gene. Consequently, cpstools assigned this repeat to both the IGS and the gene region, whereas MISA reported it only once. Excluding this duplicate, the remaining 103 SSR positions were identical between the two methods, with perfect agreement in motif type, repeat length, and genomic coordinates, confirming the reliability of both approaches. The advantage of cpstools lies in its ability to separately classify SSRs into IGS and coding regions. This analysis revealed that 100 of the 104 records (96.15%) were located in IGS regions, and only 4 (3.85%) were located in gene regions — the latter including 3 fully genic SSRs (in rps3_2, cob, and rrn18) and the boundary‑associated (ACTAG)₃ repeat (Table S4). Tandem repeat sequences, also known as satellite DNA, are core repeat units of 7–200 bp that repeat multiple times in tandem. They are widespread in eukaryotic genomes and prokaryotes. A total of 5 tandem repeat sequences were found in the V. thapsus mtDNA with a matching degree of > 85% and a length of 12–21 bp. The dispersed repeats in the V. thapsus mtDNA were inspected, and a total of 220 pairs of repeats ≥ 30 bp in length were observed, including 117 pairs of palindromic repeats, 102 pairs of forward repeats, 1 pair of complementary repeats, and no pair of reverse repeats was detected. The longest palindromic repeat was 317 bp, and the longest forward repeat was 7,569 bp (Fig. 3B, Table S2, Table S3).

Fig. 3.

Repeat Sequence Analysis of the Verbascum thapsus mitochondrial genome. Note: A Horizontal coordinates indicate the type of SSRs, and vertical coordinates indicate the number of repetitive fragments. B Horizontal coordinates indicate the type of repetitive sequence, and vertical coordinates indicate the number of repetitive segments

RNA editing sites prediction

RNA editing events were identified and characterized on 36 unique PCGs from V. thapsus mitochondria. The setting criterion was cutoff value = 0.9. Under this criterion, a total of 468 potential RNA editing sites were identified on the 36 mitochondrial PCGs, all of which were base C to U edits. The ccmB gene, with 39 RNA editing sites identified, had the highest number of edits among all mitochondrial genes. This was followed by the nad4 gene with 35 RNA editing events (Fig. 4).

Fig. 4.

Number of predicted RNA editing sites for each coding gene in the Verbascum thapsus mitochondrial genome

Structure analysis and structural variation of mtDNA

The mtDNA of V. thapsus contain three nodes; Fig. 5A. Each node represents a sequence (contig) obtained from the assembly, and if the black line connects the two nodes, it means that there is an overlapping region between the two sequences. All these sequences constitute a complex multi-branched closed genome structure representing the complete mitochondrial genome sequence of the species. For several key nodes where branching exists, the relevant sequences at the branching nodes were derived and mapped to long-reads. When two sequences connected along a black line, appear head-to-tail on the same long-reads, it means that the long-reads support the interconnection of these two sequences. For those at branching nodes where there are multiple different connections, those supported by more long-reads are prioritized. Red nodes represent potentially repetitive sequences, which will appear multiple times in the genome. Based on the long-reads data, one loop ‘master circle’ sequence obtained after solving the branching node caused by a repeat sequence of 7,567 bp (node 3) is shown in Fig. 5B, which represents its master circle structure, and the specific solution path is shown in Table 4. In addition, node 3, as a repeat sequence, may mediate a potential alternative structure, as shown in Fig. 5C, which may split into two smaller rings mediated by node 3, representing a possible secondary conformation.

Fig. 5.

A Sketch of Verbascum thapsus mitochondrial genome assembled based on the long-reads data; B The primary ring structure of 1 ring obtained after solving the branching node caused by a repetitive sequence (node 3) with a length of 7,567 bp based on the long-reads data; C The primary ring structure, mediated by node 3, may split into two smaller rings, representing a possible secondary conformation

Table 4.

Path selection for individual nodes (repetitive zones) based on Nanopore data

Contig	Type	Path
1	circular	1–3-2–3

To further corroborate this multi-conformational architecture, we compared the sequencing depth of the three distinct genomic compartments: node 1 (large unique region, 360,015 bp), node 2 (small unique region, 102,871 bp), and node 3 (the 7,567 bp repeat). In a conventional single-master-circle genome organization (e.g., that of chloroplast genomes), any sequence that is present in two copies per genome would be expected to exhibit approximately twice the sequencing depth of a single-copy region. Node 1 and node 2 exhibited average depths of 82.4 × and 77.0 ×, respectively, representing the baseline coverage for single-copy sequences. Notably, node 3 showed a depth of 142 ×, which is approximately 1.8-fold higher than that of the unique regions. This elevated depth is close to the theoretical twofold increase expected for a sequence present in two copies per genome (assuming equal sequencing efficiency and accounting for minor technical variations). Therefore, the depth data independently confirm that the 7,567 bp repeat is duplicated in the V. thapsus mtDNA. Such a duplicated repeat can mediate homologous recombination, giving rise to two alternative sub-circles (Fig. 5C) in addition to the master circle (Fig. 5B). Given that the observed depth ratio (1.8 ×) is slightly less than the expected 2 ×, this suggests a mixed population of genome conformations. Let p be the proportion of molecules existing as the master circle (where the repeat is duplicated) and 1-p be the proportion existing as the two smaller sub-circles (where the repeat is single-copy per sub-circle). The expected average copy number of the repeat is 2p + 1 × (1-p) = 1 + p. Setting 1 + p = 1.8 gives p = 0.8. Hence, approximately 80% of the mitochondrial DNA molecules are in the master-circle conformation, while about 20% are present as the two sub-circles. This quantitative estimate further supports the conclusion that the master circle predominates, but a fraction of molecules may exist as two smaller sub-circles, where the repeat would be present only once per sub-circle and thus contribute less to the overall depth. This depth-based evidence strongly supports the complex multi-branched closed genome structure deduced from the assembly (Fig. 5A) and is fully consistent with the long-read supported connectivity at node 3.

Homologous analysis of genome sequences

Through sequence similarity analysis, a total of 20 fragments were included as homologous fragments between the mtDNA and the cpDNA in the species of V. thapsus, with a total length of 14,415 bp, which accounted for 3.02% of the total length of the mtDNA (Fig. 6, Table 5). Among these, MTPT4 was the longest at 4,074 bp. By annotating these homologous sequences, 12 complete genes were identified on the 20 homologous fragments, including 6 PCGs (psbF, psbE, rpl16, rpl22, rps7, rps19), and 6 tRNA genes (trnD-GUC, trnH-GUG, trnN-GUU, trnM-CAU, trnI-CAU, trnS-GGA).

Fig. 6.

Chloroplast genome to mitochondrial genome migration map in Verbascum thapsus

Table 5.

The homologous DNA fragment in the V. thapsus mitochondrial genome

Number	Identity	Length	Mismatches	Gap Openings	cp start	cp end	mt start	mt end	MTPT annotation
MTPT1	100	28	0	0	41401	41428	293705	293732	Partial (psaA)
MTPT2	100	41	0	0	119725	119765	407430	407390	Partial (ndhA)
MTPT3	99.929	1414	1	0	151811	153224	305778	304365	Partial (rpl2)
MTPT4	99.509	4074	8	3	95218	99290	410990	406928	Partial (ndhB), Complete (rps7)
	99.509	4074	8	3	138472	142544	406928	410990	Partial (ndhB), Complete (rps7)
MTPT5	99.304	2298	10	1	83654	85951	303487	305778	Partial (rps3, rpl2), Complete (rpl22, rps19)
MTPT6	98.678	227	2	1	86451	86676	305778	306004	Complete (trnI-CAU)
	98.678	227	2	1	151086	151311	306004	305778	Complete (trnI-CAU)
MTPT7	97.436	78	2	0	1	78	268239	268162	Complete (trnH-GUG)
MTPT8	97.365	2315	40	10	81283	83592	301213	303511	Partial (rpl14, rps3), Complete (rpl16)
MTPT9	95.294	85	3	1	108738	108821	441494	441410	Complete (trnN-GUU)
	95.294	85	3	1	128941	129024	441410	441494	Complete (trnN-GUU)
MTPT10	95.181	830	24	3	455	1284	216318	217131	Partial (psbA)
MTPT11	94.774	842	17	4	39153	39994	477206	478020	Partial (psaB, psaA)
MTPT12	94.048	84	4	1	54854	54937	217417	217499	Partial (atpB)
MTPT13	93.438	320	2	3	54467	54786	217132	217432	Partial (atpB)
MTPT14	90.94	298	21	4	28315	28606	364337	364040	IGS (trnC-GCA, petN)
MTPT15	90.909	77	6	1	53198	53273	388422	388346	Complete (trnM-CAU)
MTPT16	90.226	133	11	1	35233	35365	341658	341788	Partial (psbC)
MTPT17	89.444	180	15	1	45724	45903	39068	38893	Complete (trnS-GGA)
MTPT18	86.957	138	11	2	30900	31037	39564	39694	Complete (trnD-GUC)
MTPT19	85.475	179	12	5	46472	46650	38882	38718	Partial (rps4)
MTPT20	82.817	774	81	23	65090	65843	274724	273983	Partial (psbL), Complete (psbF, psbE)

Collinearity analysis of mtDNA

The results of the collinearity analysis were shown in Fig. 7, where the red curved regions indicate points of inversion and the gray regions indicate areas with good homology. For better presentation of the results, collinear blocks less than 0.5 kb in length were not retained in the results. Homologous collinear blocks were detected between V. thapsus and the related species under the order Lamiales, but these collinear blocks were short in length. In addition, some blank regions were found, and these sequences were unique in V. thapsus with no homology to the rest of the species. The results suggest that the collinear blocks were arranged in an inconsistent order among the mtDNAs of these seven species, and the mtDNA of the V. thapsus underwent genome rearrangements with closely related species.

Fig. 7.

Collinearity analysis of mitochondrial genomes of Verbascum thapsus

Phylogenetic analysis

A phylogenetic tree was constructed for 33 species under eight angiosperm families based on the DNA sequences of 14 conserved mitochondrial PCGs. The results showed that the species V. thapsus belongs to the family Scrophulariaceae of Lamiales with high support. It is clustered in the phylogenetic tree with the family Lentibulariaceae, the family Orobanchaceae, and the family Lamiaceae, showing close affinities. And the species of each family clustered under that family in the evolutionary tree with good support and closer affinity with other species in the same family (Fig. 8).

Fig. 8.

Phylogenetic tree of 33 species based on 14 conserved mitochondrial protein-coding genes

Discussion

The structure of mtDNA

As more sequencing studies were completed, researchers gradually realized that plant mitochondrial genomes are diverse in conformation, and that assembled mtDNA may also be cyclic (most plant mtDNA), polycyclic (kiwifruit mtDNA) [73], linear (maize mtDNA) [74], and possibly multi-branched structures (Sitka Spruce mtDNA) [75], as demonstrated in several papers. Since mtDNA is subject to a large number of rearrangements and reorganizations, it is possible that it has more than just a single ring structure with other substructures existing [32].

Since the primary endosymbiotic event, the structure of plant mitochondrial DNAs (mtDNAs) has experienced rapid and substantial alterations [25, 42]. It is in line with this evolutionary pattern that the genomic composition of mitochondria has become intricate, rendering the study of plant mtDNAs relatively challenging [76, 77]. In the present study, we accomplished the sequencing and assembly of the Verbascum thapsus mtDNA, with a length of 478,020 base—pairs (bp) in length. Through the analysis of the V. thapsus mtDNA, we obtained a single circular 'master circle' (Fig. 5). Nevertheless, owing to the high frequency of recombination, plant mtDNA exhibits a diverse array of dynamic structures within mitochondria, such as master rings, secondary rings, and linear molecules [78, 79]. Thus, it cannot be precluded that there may exist potential alternative structures. For instance, the master circle structure mediated by node 3 might split into two smaller rings, representing a possible secondary conformation (Fig. 5). The mtDNA of the angiosperm common ancestor encompasses a total of 41 protein—coding genes (PCGs), including 24 core genes and 17 variable genes. The variable genes display significant variation among species and are subject to varying degrees of loss [80]. From the assembled mtDNA, 36 PCGs were annotated, including 24 unique mitochondrial core genes and 12 non—core genes. The guanine—cytosine (GC) content of the V. thapsus mtDNA was 45%, which exceeded that of its own chloroplast genome (38%). This phenomenon may be ascribed to the fact that compared with the mtDNA, the chloroplast DNA (cpDNA) encodes a greater number of genes within a shorter length (153,014 bp < 478,020 bp), and the termination codons (TAG, TAA, TGA) have a high adenine—thymine (AT) content. GCs can form three hydrogen bonds between them, one more than the number of hydrogen bonds between ATs. Consequently, DNA molecules with a high GC content are more stable (with higher denaturation temperatures). In general, the more primitive the organism, the higher the GC content of its DNA [81]. This can also be corroborated by the fact that the cpDNA is evolving at a faster rate than the mtDNA, which also possesses a higher degree of conservation [32]. Recent studies have further elaborated on the structural diversity of plant mtDNAs, including comparisons with species like Primulina hunanensis, which exhibits a similar GC content and complex conformations [82].

Homology analysis

Gene transfer within cells transpires among different organelles, such as chloroplasts, mitochondria, and nuclei [83, 84]. Specifically, gene transfer from chloroplast DNA (cpDNA) to mitochondrial DNA (mtDNA) is rather prevalent during the long—term evolution of plants [83, 85]. The mechanisms of gene transfer between cpDNAs and mtDNAs encompass intracellular gene transfer and horizontal gene transfer [86, 87]. Currently, the event of mtDNA transfer to cpDNA has only been discovered in cashews (Anacardium occidentale) [88]. In Verbascum thapsus, a total of 20 fragments exhibit homology between the mtDNA and cpDNA, with an overall length of 14,415 base pairs (bp), accounting for 3.02% of the entire length of the mtDNA. Among these fragments, MTPT4 is the longest, measuring 4,074 bp. These large mitochondrial—plastid transfer regions (MTPTs) are considered to have a broad influence on eukaryotic evolution and contribute to genetic diversity [89]. The results of the collinearity analysis also demonstrated that although homologous collinear blocks were detected in V. thapsus and its closely related species within the order Lamiales, these collinear blocks were shorter in length, and some exclusive non—homologous regions were identified. This suggests that there is no homology among these species, and genome rearrangement has occurred between the V. thapsus mtDNA and its closely related species. One potential reason for this is that intracellular gene transfer results in a high degree of mtDNA rearrangement [90]. Moreover, as the cpDNA fragments transferred into mitochondria show a high degree of consistency with the original cpDNA sequences, this gene transfer may lead to mtDNA disassembly [91]. Additionally, research has shown that the origin of transfer RNAs (tRNAs) in plant mitochondria can be divided into two sources. One is inherited from the mitochondrial ancestor, and the other is acquired from chloroplasts through horizontal gene transfer (HGT) [92]. In the context of the present study, 6 tRNA genes (trnD—GUC, trnH—GUG, trnN—GUU, trnM—CAU, trnI—CAU, trnS—GGA) were transferred from the cpDNA to the mtDNA, accounting for nearly one—third of the tRNAs in V. thapsus.

The phenomenon of trnW—CCA gene transfer has been extensively reported in other angiosperms, and it appears to be homologous to the cpDNA. However, similar to Amborella trichopoda, V. thapsus possesses a native trnW—CCA gene, which is uncommon in other angiosperms [93]. Furthermore, a bacterial—derived tRNA (trnC—GCA) was identified in the native tRNA gene pool of V. thapsus. Although this event has been reported in some previous studies [93, 94], it is infrequent in the mtDNA of plants [95].

Analysis of codon usage and repeat sequences

The eukaryotic genome comprises 64 codons, which encode 20 distinct amino acids and three stop codons. With the exception of methionine and tryptophan, multiple codons encode all other amino acids. This phenomenon leads to the situation where amino acids exhibit differential preferences for various codons, and such preference is considered to be the outcome of a relative equilibrium that gradually emerges within the cell through long—term evolutionary selection. Undoubtedly, there are substantial variations in genomic codon usage frequencies among different species and organisms. The codon preference was ascertained through the calculation of the relative synonymous codon usage (RSCU). The findings indicated that there was a high utilization rate of A or T at the third codon position encoding the protein genome in the mitochondrial DNA (mtDNA) of Verbascum thapsus, a phenomenon commonly observed in plant mtDNAs [90, 96]. Except for phenylalanine (Phe), which had a maximum RSCU value of less than 1.2 and lacks a strong codon usage preference, the codon usage patterns in the V. thapsus mtDNA exhibit a high degree of preference.

DNA sequences in biological cells contain many repetitive sequences, which can be categorized into two main groups, namely, tandem repeats and dispersed repeats. The difference between tandem and dispersed repeats is whether the repeats are distributed adjacent to each other. Being adjacent to each other is a tandem repeat, while not being adjacent to each other is a dispersed repeat. Among the tandem repeats, the simple sequence repeat (SSR) is a special type of tandem repeat, generally not exceeding 6 bp. SSRs are often used in molecular marker development because of their dominant inheritance and other characteristics. Repetitive sequences are ubiquitously present in mtDNAs, and the vast majority of variations in the size of plant mtDNAs can be attributed to differences in the size of repetitive sequences [97]. A total of 103 simple sequence repeats (SSRs) were identified in V. thapsus, with monomeric and dimeric forms constituting 47.57% of the total SSRs. Fifteen thymine (T) monomeric repeat sequences accounted for 60.00% of the 25 monomeric SSRs, and hexameric SSRs were not detected in this mtDNA. SSRs serve as significant molecular markers for biometrics, evolutionary analysis, and genetic diversity studies [98].The mtDNAs of nearly all angiosperms possess large (> 1 kb) non—tandem repeat sequences with active recombination [99]. The longest non—tandem repeat in the V. thapsus mtDNA is 7,569 base pairs (bp), which might be one of the factors contributing to the multiple conformations of mitochondria. Secondly, the mtDNA of V. thapsus contains a substantial number of dispersed repeats, among which forward repeats and palindromic repeats are prevalent, with a ratio between them approaching 1:1, which is also comparable to that of other angiosperms [36].

RNA editing sites prediction

In eukaryotes, RNA editing takes place in the coding regions of transcribed RNA through base insertion, deletion, or substitution [100]. Analyses have demonstrated that RNA editing occurs ubiquitously in the mitochondria of higher plants and is one of the major processes of gene expression within the plant mitochondrial DNA (mtDNA) [101, 102]. For the majority of angiosperms, the chemical reaction of RNA editing is a deamidation reaction where site—specific cytosine (C) is converted into uracil (U) [103]. Studies have proposed that RNA editing of mitochondrial genes may regulate traits related to cytoplasmic genetics in plants [104]. A total of 468 potential RNA editing sites were identified in Verbascum thapsus. The majority of these RNA editing sites were located at the first or second position, which is consistent with the patterns observed in most plants [36, 96, 105]. The functions of novel protein—coding genes can be effectively predicted by identifying RNA editing sites. Notably, it was found that the stop codons of two genes (atp6 and rps10) and the start codons of four genes (cob, cox2, rps10, and nad4L) were generated by RNA editing events. Moreover, these new start and stop codons encoded more conserved proteins and homologous proteins, which facilitated better gene expression in the mitochondria [106]. In conclusion, the newly sequenced mitochondrial genome of Verbascum thapsus offered novel insights into the function and structure of the mitochondrial genomes of Verbascum thapsus, the Scrophulariaceae family, and the order Lamiales.

New insights into the construction of phylogenetic trees with mtDNA

Previous research has elucidated the phylogenetic relationships among Lamiales through the analysis of morphological features, chromosomal characteristics, internal transcribed spacer (ITS) sequences, amplified fragment length polymorphism (AFLP) data, and chloroplast genomic information [6, 7, 22, 62]. Plant mitochondria have undergone rapid evolutionary processes, resulting in inter—specific heterogeneity of mitochondrial genomes, large—scale genomic recombination, and gene mosaicism [25, 27]. Given that plant mitochondrial genomes also harbor relevant genetic information, they are increasingly being utilized in phylogenetic investigations. Moreover, with the advancements in sequencing and assembly technologies, the application of mitochondrial information in plant phylogenetic analyses has garnered increasing attention [34–36]. Plant mitochondrial genomes exhibit significant variations in size and structure, yet the functional genes remain relatively conserved [38, 77]. In this study, a phylogenetic tree was constructed based on the DNA sequences of 14 conserved mitochondrial protein—coding genes (PCGs) for 33 species belonging to 8 families of angiosperms, along with a phylogenetic tree based on mitochondrial genome data within the Lamiales. In the mitochondrial PCG tree, it was found that the families Scrophulariaceae, Lentibulariaceae, Orobanchaceae, and Lamiaceae, as well as the remaining families within the order Lamiales, could be clearly distinguished. It was also evident that the orders, families, and genera could be effectively delineated, which is consistent with the results of the APG IV classification system [33]. However, at a finer scale, there are still certain discrepancies between the phylogenetic tree constructed from mitochondrial PCGs and that from chloroplasts. For instance, in the families Labiatae and Oleaceae, where both mitochondrial and chloroplast data are relatively abundant, the phylogenetic positions of different genera within the family vary. One potential explanation for this phenomenon is that the genetic variation of organelles within the same cytoplasm can disrupt the linkage disequilibrium between mitochondria and chloroplasts [107, 108]. When such a situation occurs, the phylogenetic reconstructions of the genomes from the two organelles may conflict. Additionally, limited sampling, incomplete lineage sorting, and differences in evolutionary rates may also contribute to the conflicts between the phylogenetic reconstructions of the two organelle genomes [109, 110]. Therefore, a large—scale sampling is essential for a more comprehensive understanding of the evolution of the mitochondrial genomes of Lamiales. Sequencing and assembling a new complete mitochondrial genome of V. thapsus represents a significant step in comprehending the variation of mitochondrial genomes within the order Lamiales, and even within the family Scrophulariaceae.

Conclusion

This study successfully assembled and analyzed the complete mitochondrial genome of Verbascum thapsus, providing valuable insights into its structural characteristics, including a genome size of 478,020 bp with a GC content of 45%. The research highlighted key aspects such as extensive homologous sequences between mitochondrial and chloroplast genomes, codon usage biases favoring A/T-rich codons, the presence of repetitive sequences including SSRs and dispersed repeats, and numerous RNA editing events predominantly involving C-to-U conversions. Phylogenetic analysis based on mitochondrial protein-coding genes supported the taxonomic positioning within Lamiales, aligning with APG IV classifications, while revealing some discrepancies with chloroplast-based phylogenies, possibly due to organellar genome evolutionary dynamics. These findings advance our understanding of plant mitochondrial genome evolution and underscore the utility of mitochondrial data in phylogenetic studies, paving the way for future research on broader sampling and functional genomics in related taxa.

Abbreviations

cpDNA

Chloroplast genome

mtDNA

Mitochondrial genome

NCBI

National Center for Biotechnology Information

PCGs

Protein-coding genes

RSCU

Relative synonymous codon usage

SSRs

Simple sequence repeats

tRNA

Tranfer RNA

rRNA

Ribosomal RNA

Authors’ contributions

Xiang Dong designed the experiments, managed and analyzed the data, and drafted the initial manuscript; Hang Huang and Peninah Cheptoo Rono were responsible for the preparation of tables and figures, and revised the manuscript; Guangwan Hu and Changli Zeng provided experimental materials, oversaw project management, secured financial support, and finalized the manuscript. All authors read and approved the final version of the manuscript.

Funding

This work was funded by grants from the Science and Technology Major Project of Xizang (XZ202501ZY0151.4), the National Natural Science Foundation of China (32270228), Hubei Provincial Science and Technology Innovation Talent and Service Project (2022EHB031), the Key R&D Program of Hubei Jiangxia Laboratory (JXBS010), the Project for Scientists in Biological Taxonomy of the Chinese Academy of Sciences (CAS-TAX-24–39), the Sino-Africa Joint Research Center, CAS, (SAJC202401), the Hubei Provincial Natural Science Foundation of China (2026AFB672), and the Shiyan Key Laboratory of Biological Resources and Eco-environmental Protection (Hanjiang Normal University) 2025 Open Fund Project (No. 001), and and the University-level Scientific Research Project of Jianghan University (2025JCYJ16).

Data availability

The assembled and annotated genomes are accessible in NCBI database under the research accession PX_570624.

Declarations

Ethics approval and consent to participate

The authors have complied with the relevant institutional, national and international guidelines in collecting biological materials for the study. The relevant permits for this research were granted by the Chinese Academy of Sciences and The Wuhan Botanical Garden of the Chinese Academy of Sciences. No materials from animals or humans were used in this research. The voucher specimen of the collected plant materials was deposited in the Herbarium of the Wuhan Botanical Garden (voucher number: HGW—2029), and was identified by Professor Hu Guangwan, who is also a corresponding author of this paper.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.