Mitochondrial genome assembly, comparative analysis of two caprifoliaceae species, and insights into adaptive evolution

Abstract

Caprifoliaceae is a cosmopolitan plant family encompassing a large number of species, with significant economic and ecological values. Species divergence and adaptation in Caprifoliaceae are not well understood due to extensive hybridization and rapid radiation. Here, we assembled and compared the complete mitochondrial genomes of Kolkwitzia amabilis and Triplostegia glandulifera, which are distributed in distinct habitats with unique life forms, to elucidate the structural variation of mitogenomes and their implications for species divergence and environmental adaptation. The mitochondrial genome of K. amabilis consists of two circular molecules with sizes of 540,375 bp and 197,940 bp, respectively, while T. glandulifera has a single circular genome of 642,933 bp. The two genomes exhibit a similar preference for A/U bases in codon usage, mainly driven by natural selection, but present significant differences in mode and the number of repeat types. Among 25 shared genes in six Caprifoliaceae species, atp1 had the highest nucleotide diversity; meanwhile, atp4, ccmB, and mttB were under positive selection. Notably, rps7 in T. glandulifera had potentially undergone species-specific positive selection. A phylogenetic tree of Caprifoliaceae species based on 19 conserved protein-coding genes (PCGs) showed that K. amabilis and T. glandulifera formed a sister group, providing molecular evidence at the mitogenome level for the systematic classification of plants in this family. Furthermore, analyses of ancestral genome reconstruction indicated that both mitogenomes exhibited significant gene order rearrangements and gene deletions, and their duplication patterns of nad1 and nad2 genes differed from those of other Caprifoliaceae species, suggesting that dynamic structural changes were an important feature of mitogenome evolution in Caprifoliaceae. In conclusion, the firstly reported mitogenomes of two distinctive species provided important molecular data for phylogenetic, genomic evolution, potential divergence and adaption in Caprifoliaceae.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-026-08519-5.

Introduction

The Caprifoliaceae family is a highly diverse and ecologically important group, comprising 41 genera and approximately 960 species distributed globally [1]. Species in Caprifoliaceae present multiple life forms, including small trees, shrubs, vines and herbs, and span diversified habitats from evergreen broadleaf forests to high-altitude screes of the Himalayas and circumpolar regions [2, 3]. Caprifoliaceae is globally concerned due to its important economic values. It includes well-known horticultural species (e.g., Viburnum, Lonicera, Weigela) [3–5], traditional medicinal plants (e.g., certain Lonicera and Sambucus species) [6, 7], and fruit-bearing taxa used in winemaking, integrating ornamental, medicinal, and economic significance. Despite its ecological and practical importance, species differentiation and adaptation to divergent environments in Caprifoliaceae are not well understood for widespread hybridization, introgression, incomplete lineage sorting, and morphological convergence [8, 9]. Exploration of evolutionary trajectories among species in unique habitats is highly valuable to elucidate the mechanisms of diversification and adaptation of Caprifoliaceae.

Kolkwitzia amabilis, a deciduous shrub endemic to North China, is the only number of the Tertiary relic genus Kolkwitzia in Caprifoliaceae and listed as a third-class nationally protected plant in China. Its name derives from its distinctive fruits, which are densely covered in prickly hairs and shaped like hedgehogs [10]. These hedgehog-like fruits enhance ornamental value while playing a pivotal role in seed dispersal strategies, rendering K. amabilis a valuable material for ecological research [11]. K. amabilis is a xerophytic to mesophytic species with relatively strong cold tolerance. Its central distribution area lies in semi-humid and semi-arid regions, characterized by dry, cold winters and springs, along with hot, rainy summers and autumns [12]. Triplostegia glandulifera, a perennial herb with slender rhizomes, is primarily distributed in southwestern China, thriving in alpine meadows, forest edges, or valley areas at elevations ranging from 1,500 to 4,000 m [13]. Its unique adaptation to high-altitude habitats makes it an ideal model for studying mechanisms of plant responses to extreme environments, such as low temperatures, intense ultraviolet radiation, and large diurnal temperature fluctuations. Notably, the phylogenetic status of the genus Triplostegia has remained controversial for over two centuries [14]. Recently, molecular evidence supported that Triplostegia should be removed from Dipsacaceae and classified within Caprifoliaceae [15]. The two species present distinct differences in their morphological traits and habitats, implicating for deeply species divergence and environmental adaptation associated with different evolutionary histories. But their genetic basis and the mechanism underlying divergence and adaptation are largely unknown. Understanding the genetic differentiation of the two species will provide critical clues for deciphering species divergence patterns and adaptation to the heterogeneous habitats of Caprifoliaceae plants.

Mitochondria are key organelles in plant cells. They not only act as the main site of cellular respiration, providing core energy for cell life activities, but also play irreplaceable roles in important physiological processes of angiosperms, including the regulation of growth and development, the initiation of programmed cell death, and the mediation of male sterility [16]. Generally, gene compositions of the mitochondrial genome (mitogenome) in plants are highly conserved [17], which provides stable and reliable molecular anchors for comparing evolutionary relationships across species, serving as an important foundation for phylogenetic research [18]. In contrast to the chloroplast genome, plant mitogenomes exhibit significant interspecific differences in structural morphology and molecular size [19, 20]. Recombination mediated by repeat sequences is considered as the main factor of mitogenome size variation and the complex structures formation. This process has been verified in various plants such as carrot (Daucus carota), cucumber (Cucumis sativus) and rice (Oryza sativa) [21].

Due to its favorable characteristics, the mitogenome has gradually become an important tool for deciphering the phylogenetic relationships and evolutionary histories of plant taxa. Many works indicate that the variability of plant mitogenomes is closely related to the genomic ploidy, phylogenetic position, status of life history and even the niches of species. For example, among 13 species of the genus Gastrodia (Orchidaceae), mitogenomes with extreme differences in chromosome numbers varies by nearly 4 times in size, providing clues for the specialization mechanism of mitochondria in heterotrophic plants [22]. In the genus Gossypium, mitogenomes from 19 representative species (including diploid and polyploid taxa) showed that the mutation rate in coding regions was low while structural variations in non-coding regions were significant, offering key molecular evidence for the origin of allopolyploidy [23]. Additionally, the holoparasitic plant Rhopalocnemis phalloides has 21 small circular chromosomes of only 130 kb in length with extremely high sequence heterogeneity in individuals [24], reflecting the special evolutionary path of mitogenomes driven by parasitic habits.

In this study, the complete mitogenomes of K. amabilis and T. glandulifera were first assembled and characterized by using the high-throughput sequencing technology. We aim to investigate their characteristics, structural variation, and intergenomic gene transfer. Then we further analyzed the phylogenomic relationships and reconstructed ancestral mitogenomes among Caprifoliaceae plants. Our study not only brings novel insights into mitogenome evolution and potential adaptation of both plants, but also offers molecular data for phylogenetics of Caprifoliaceae and conservation genomic works of rare species in Caprifoliaceae (e.g., K. amabilis).

Materials and methods

Plant sample collection, mitogenome sequencing, and data retrieval

The materials of K. amabilis was collected from a cultivated individual grown on the Northwest University Taibai campus, located in Xi’an, Shaanxi, China (108.92°E, 34.25°N). Its original cultivated germplasm was provided by Shaanxi Baoji Seedling Breeding Center (107.87°E, 34.21°N) and then transplanted to the campus. The center’s standardized K. amabilis germplasm ensures the authenticity and stability of the materials. Fresh and healthy leaves were harvested and then stored at -80℃ for subsequent genomic DNA extraction. Total genomic DNA was extracted using a modified CTAB technique [25], followed by the determination of its quality and concentration using an ultraviolet spectrophotometer (MAPADA LIV-3200) and agarose gel electrophoresis (0.8%), respectively. Then, the qualified library was sequenced on the Illumina NovaSeq 6000 and PacBio SMRT(CCS) platform by Beijing BioMarker Technologies Company (Beijing, China). The PacBio and Illumina sequences of T. glandulifera (CRA012478) were downloaded from the Natinal Genomics Data Center (NGDC) (https://ngdc.cncb.ac.cn/?lang=zh).

Mitochondrial genome assembly and annotation

To obtain the mitogenome, PacBio HiFi reads were de novo assembled using smartdenovo (https://github.com/ruanjue/smartdenovo). The generated contigs were singled out as the seed sequence with the most homology to the honeysuckle (Lonicera japonica) mitogenome by BlastN [26]. The seed sequence was corrected using Racon (https://github.com/lbcb-sci/racon) and minimap2 [27], respectively. Then, the software pilon (https://github.com/broadinstitute/pilon) was employed to polish the raw genome with second-generation sequencing data. The successfully assembled mitogenome was reviewed and evaluated independently for assembly quality using BWA [28] and samtools [29]. The program Geseq [30] was applied to annotate the mitogenome using the following species as references: Lonicera japonica (MZ504724.1), Triosteum pinnatifidum (NC_064333.1) and Heptacodium miconioides (PP857826.1). The threshold for protein search identity was set to 55%, and those of rRNA and tRNA were 85%. Any annotation errors in mtDNA were manually corrected using Apollo software (https://github.com/GMOD/Apollo). The circular map of the mitogenome was plotted by the online software OGDDRAW (https://chlorobox.mpimp-golm.mpg.de/OGDraw.html).

Repeat sequence identification

To identify the simple sequence repeats (SSRs), the online tool MISA (https://webblast.ipk-gatersleben.de/misa/) was applied with a minimum repeat number of 10, 5, 4, 3, 3, and 3 for mono-, di-, tri-, tetra-, penta-, and hexa-repeat units, respectively. The maximum distance between two adjacent SSRs (for compound SSRs) was set to 100 bp. Interspersed repeat sequences (forward, reverse, complement, and palindromic repeats) were detected using the online tool REPUTER (https://bibiserv.cebitec.uni-bielefeld.de/reputer) with a minimum repeat size of over 30 bp and a Hamming distance of 3. Tandem repeat sequences were detected using the online tool TRF (https://tandem.bu.edu/trf/trf.html) with the following specific parameters: the minimum period size (repeat unit length) was over 30 bp, the minimum alignment score was 50, the maximum period size was 500 bp, the match weight was 2, the mismatch penalty was 7, and the indel penalty was 7.

Codon usage bias analysis and RNA editing sites prediction

The conserved protein-coding genes (PCGs) of K. amabilis and T. glandulifera mitogenomes were extracted by PhyloSuite (http://phylosuite.jushengwu.com/) with default settings. To avoid sample bias in the calculation of codon bias, gene sequences greater than 300 bp with a start codon of ATG and a stop codon of TGA, TAA, or TAG, were selected for subsequent analysis. The program CodonW (https://codonw.sourceforge.net/culong.html#References) was operated to analyze the GC content, relative synonymous codon usage (RSCU) and the effective number of codons (ENC). The ENC plot was used to analyze base composition other than the Codon bias factors. The standard curve was generated using the equation ENC = 2 + GC3 + 29/(GC3^2 + (1 - GC3)^2) [31]. Using the GC3 value as the horizontal coordinate and the ENC value as the vertical coordinate, each gene point was scattered around the standard curve to form a scatter plot. As the ENC value fell away from the curve, it indicated that the gene was impacted by natural selection [32]. The ENC plot was generated by the R software [33]. Neutral plots were made with the calculated GC3 as horizontal coordinates and the mean number of GC1 and GC2 (GC12) as vertical coordinates to determine the major influences on codon bias [34]. A strong correlation between GC12 and GC3 nucleotide compositions, particularly when demonstrating a regression slope approaching unity (slope ≈ 1), suggested that mutational biases primarily governed codon usage patterns, while significant departures from this equilibrium relationship pointed to substantial selective constraints shaping codon preference evolution [35]. The Parity Rule 2 (PR2) analysis was performed using A3, T3, C3, and G3 values, with G3/(G3 + C3) values as the horizontal coordinates and A3/(A3 + T3) values as the vertical coordinates [36]. If there were no mutation or selection bias between the two complementary strands of DNA, the coordinates of the centers of distribution of both A3/(A3 + T3) and G3/(G3 + C3) were 0.5 [37].

For an in-depth observation of RNA editing sites, RNA editing events were predicted in this research using the online tool PREPACT3 (http://www.prepact.de/). The Arabidopsis thaliana (OY747154.1), Cucurbita pepo (NC_014050.1), and Vitis vinifera (NC_012119.1) mitochondrial proteins were selected as references. The cutoff value was set at 0.001.

Phylogenetic tree construction and comparative genomic analysis

23 complete mitogenomes sequences from three orders (Asterales, Apiales, and Dipsacales) were obtained from GenBank for phylogenetic analysis (Table S1). Together with K. amabilis and T. glandulifera, 19 shared PCGs were extracted using PhyloSuite [38], aligned using MAFFT [39] and then concatenated to generate the datasets for phylogenetic tree construction. To verify the topological stability and enhance the reliability of phylogenetic conclusions, both Maximum Likelihood (ML) and Bayesian Inference (BI) methods were employed for phylogenetic analysis. The Maximum Likelihood (ML) method was applied with 1,000 bootstrap replications in the IQ-TREE [40] software. The best-fit nucleotide substitution model was selected using ModelFinder implemented in IQ-TREE based on the Bayesian Information Criterion (BIC), and the ML tree was constructed with the selected optimal model to assess branch support. For the Bayesian Inference (BI) analysis, MrBayes v3.2.7 software [41] was used with the same optimal nucleotide substitution model as the ML method for consistency. Two independent Markov Chain Monte Carlo (MCMC) runs were performed (4 chains each), with 1,000,000 iterations and sampling every 100 generations. The first 25% of trees were discarded as burn-in, and the remaining trees were used to construct a 50% majority-rule consensus tree. Posterior probabilities (PP) were used to evaluate branch support (PP ≥ 0.95 = strong support), and ML/BI tree consistency was compared to confirm phylogenetic topology reliability.

The collinearity analysis among six Caprifoliaceae species was performed utilizing the software NGenomeSyn [42]. A ML tree was reconstructed with the same methods as described above but using 25 shared PCG genes in Caprifoliaceae species. The nucleotide diversity (π) of these 25 shared genes was generated using DnaSP with a window length of 600 bp and a step size of 200 bp. Using A. thaliana as a reference, the non-synonymous to synonymous substitution ratio (Ka/Ks) was calculated with Ka/Ks_Calculator [43]. The results were plotted using the R package ggplot2 [44].

Chloroplast-mitochondrial communication and gene rearrangement

The chloroplast genomes of K. amabilis (NC_029874.1) and T. glandulifera (NC_045051.1) were downloaded from GenBank, and their sequence were compared with the mitogenome to identify potential regions of nucleotide exchange, respectively. Subsequently, the BlastN with default settings was employed to analyze the sequence similarities between the chloroplast and mitochondrial genomes. The identified regions were visualized using the Advance Circos module in TBtools [45]. Similarly, the reference nuclear genomes of the T. glandulifera were downloaded (10.6084/m9.figshare.25018103), and the NGenomeSyn [42] software was employed to identify potential sequence exchange regions between the mitochondrial and nuclear genomes of each species.

The core genes were extracted to construct the ancestral states and their evolutionary process. The interspecific phylogenetic relationships were inferred based on the phylogenetic tree constructed with 25 PCG genes. The ancestral states of mitochondrial gene orders were reconstructed using SPP (Ancestral Genomic Reconstruction) implemented in Maximum Likelihood for Gene Order Analysis (MLGO) (http://www.geneorder.org/server.php).

Results

The mitogenome composition and features

The mitogenomes of K. amabilis and T. glandulifera were assembled with PacBio data. For K. amabilis, it comprises two circular contigs of 540,375 bp and 197,940 bp in length with an average GC content of 45.50% (Fig. 1A). The chromosomes have similar nucleotide compositions, with GC contents of 45.48% and 45.54% respectively. While in T. glandulifera (Fig. 1B), the mitogenome is 642,933 bp in size, with a GC content of 45.16%.

Fig. 1.

Circular map of the K. amabilis (A) and T. glandulifera (B) mitogenomes. Genes with different functions are labeled in distinct colors. The gray region in the inner circle represents the GC content

The two mitogenomes were successfully annotated, presenting a significant difference in gene contents. The mitogenome of K. amabilis contains 42 PCGs, 25 tRNA genes, and 2 rRNA genes, while that of T. glandulifera comprises 37 PCGs, 13 tRNA genes, and 1 rRNA gene (Table 1). In contrast to the PCGs in K. amabilis, T. glandulifera exhibits fewer duplications of ATP synthase genes (atp6, atp9) and the cytochrome c oxidase gene (cox2). Its mitogenome lacks the ubiquinol cytochrome c reductase gene (ccmFC), the ubichinol cytochrome c reductase gene (cob) as well as NADH dehydrogenase genes (nad2, nad5, nad7). Furthermore, the mitogenomes of T. glandulifera and K. amabilis have no common shared ribosomal RNA (rRNA) genes. Significant disparities were observed in the transfer RNA (tRNA) genes of the two species, where T. glandulifera possessed eleven fewer tRNA genes in comparison to K. amabilis.

Table 1.

Gene composition in the mitogenome of K. amabilis and T. glandulifera

Gene Function	K.amabilis	T.glandulifera
ATP synthase	atp1 atp4 atp6(2) atp8 atp9 atp9(2)	atp1 atp4 atp6 atp8 atp9
Cytohrome c biogenesis	ccmB ccmC ccmFC* ccmFN	ccmB ccmC ccmFN
Ubichinol cytochrome c reductase	cob
Cytochrome c oxidase	cox1 cox2** cox3	cox1 cox2*(2) cox3
Maturases	matR	matR
Transport membrance protein	mttB	mttB
NADH dehydrogenase	nad1* nad2** nad3 nad4*** nad4L nad5** nad6 nad7* nad9	nad1*** nad3 nad4*** nad4L nad6 nad9
Ribosomal proteins (LSU)	rpl10 rpl16 rpl2* rpl5	rpl10 rpl16 rpl5
Ribosomal proteins (SSU)	rps1 rps10* rps11 rps12 rps13 rps14 rps19 rps3* rps4 rps7	rps1(2) rps10* rps12 rps13 rps14 rps19(2) rps3* rps4 rps7
Succinate dehydrogenase	sdh3(3)	sdh3 sdh4
Ribosomal RNAs	rrn18 rrn5	rrn26
Transfer RNAs	trnC-GCA(2) trnE-UUC trnG-GCC trnH-GUG trnI-CAU(2) trnK-UUU trnN-GUU(2) trnS-CGA trnW-CCA trnY-GUA(2) trnD-GUC trnF-GAA trnI-GAU* trnK-UUU trnM-CAU trnP-UGG trnS-GCU trnS-UGA trnV-GAC trnfM-CAU	trnE-UUC trnF-GAA trnG-GCC trnH-GUG trnI-CAU(2) trnK-UUU trnM-CAU trnM-CAU* trnP-UGG trnS-GCU trnS-UGA trnfM-CAU
Other

*:intron number;Gene(2):Number of copies of multi-copy genes

Analysis of repeat sequences

A total of 199 SSRs were identified across the two chromosomes of K. amabilis. The distribution of these SSRs was uneven, with 140 located on chromosome 1 and 59 on chromosome 2 (Fig. 2A). Among the detected SSRs, tetramer repeats were the most abundant, accounting for 33.17% of the total, followed by mononucleotide repeats (28.64%) and dinucleotide repeats (21.11%). In contrast, trimer (20) and pentamer repeats (11) were less frequent. Generally, the remaining five types of SSRs were more abundant on chromosome 1 than on chromosome 2, except for hexanucleotide repeats. Meanwhile, the mitogenome of T. glandulifera exhibited 228 SSRs, among which mononucleotide repeats were the most abundant (83), followed by tetramer repeats (71). The number of dinucleotide repeats (32) and hexanucleotide repeats (2) in T. glandulifera was lower than that in K. amabilis.

Fig. 2.

Analysis of mitochondrial genome repeat fragments in K. amabilis and T. glandulifera. A: Distribution of simple sequence repeats (SSRs). The p1-p6 denote monomeric, dimeric, trimeric, tetrameric, pentameric, and hexameric SSRs, respectively. B: Statistics of long repeats. Counts are presented for complementary repeats (CR), forward repeats (FR), palindromic repeats (PR), reverse repeats (RR), and tandem repeats (TR), respectively

The repeat structures within each genome were scanned, including complementary, forward, palindromic, reverse, and tandem repeats (Fig. 2B). In K. amabilis, a total of 137 interspersed repeats were identified, where forward repeats were the most abundant (54), followed by palindromic repeats (46) and tandem repeats (37). Chromosome 1 possessed the largest number (80) of interspersed repeats, while chromosome 2 contained the minimum number (7) of tandem repeats. Among 96 repeat units in the mitogenome of T. glandulifera, tandem repeats were the most frequent (48), while forward and palindromic repeats occurred 44 and 4 times, respectively. Similarly, the complementary and reverse repeats were not detected in both the K. ambilis and T. glandulifera mitogenomes.

Codon usage patterns analysis

A comprehensive analysis of the codon usage bias in both species was conducted (Fig. 3). A total of 8,531 codons were identified within 29 unique PCGs in K. amabilis. The most frequently used codon was AUU (isoleucine), accounting for 3.45%, while the least used was UAG (TER), at 0.47%. At the same time, 30 codons were used more frequently than expected, indicated by relative synonymous codon usage (RSCU) values > 1, and 32 codons were used less frequently than expected (RSCU < 1). In T. glandulifera, 7,068 codons were found across 26 PCGs, with AUU (isoleucine) being the most prevalent at 3.32%, and UAG (TER) the least at 0.42%. For further insights, the analysis revealed 32 codons with RSCU values > 1 and 30 with RSCU < 1. Especially, Methionine (AUG) and Tryptophan (UGG) showed no codon preference (RSCU = 1) both in K. amabilis and T. glandulifera.

Fig. 3.

Relative synonymous codon usage (RSCU) analysis in mitogenomes of K. amabilis and T. glandulifera

The GC content was calculated for the first (GC1), second (GC2), and third (GC3) positions of these genes in K. amabilis (Table S1), revealing distinct patterns: GC1 averaged 47.38% (36.25% − 57.25%), GC2 42.42% (29.41% − 53.94%), and GC3 37.01% (24.79% − 57.93%). In T. glandulifera, similar trends emerged with GC1 at 48.21%, GC2 at 41.74%, and GC3 at 37.07%. The average GC content of the selected PCGs in these two species was 42.27% for K. amabilis and 42.35% for T. glandulifera, indicating a preference for A/T bases and A/T-ending codons.

To gain a deeper understanding of the codon usage patterns in the K. amabilis and T. glandulifera mitogenomes, the effective number of codons (ENC) values were calculated for each PCG based on the GC3 values (Fig. 4). The ENC values ranged from 37.88 to 60.10 in K. amabilis and from 38.25 to 60.11 in T. glandulifera. The ENC-plot revealed that the majority of protein-coding genes in K. amabilis fell below the standard curve, with only three genes (atp4, nad9, rps7) located above the curve (Fig. 4A). In T. glandulifera, a similar pattern was observed, with most genes below the curve with four exceptions (atp4, nad9, rpl5, rps7) (Fig. 4B). Furthermore, the neutrality plot analysis demonstrated a weak correlation between GC12 and GC3 for both species, as indicated by the relatively low R² (K. amabilis: 3.4 × 10^− 3, T. glandulifera: 9.9 × 10^− 4) (Fig. 4C and D). A binary rule 2 (PR2) analysis was conducted to explore the presence of codon bias. The PR2 plot revealed that the points of T. glandulifera were more clustered than those of K. amabilis. However, both species exhibited a similar bias pattern in the base composition at the third position of codons, with most points concentrated in the first and third quadrants. These results suggested that the bias in codon usage for both species was primarily driven by natural selection pressures rather than purely random processes (Fig. 4E and F).

Fig. 4.

Codon usage bias landscape of mitochondrial protein-coding genes (PCGs) in K. amabilis and T. glandulifera. A, B: Relationship between GC3 and effective number of codons (ENC). C, D: The analysis of neutrality analysis of PCGs. E, F: The PR2-bias result of PCGs

RNA editing sites prediction

By using the PREPACT3, a total of 686 RNA editing sites were identified in K. amabilis (Fig. 5A), while T. glandulifera exhibited 537 sites (Fig. 5B). The nad4 and mttB genes showed the highest number of RNA editing events in K. amabilis and T. glandulifera. Notably, there were 12 genes (ccmB, ccmC, nad1, nad4, nad4L, rpl10, rpl16, rps4, rps7, rps12, rps13, rps14) exhibiting conserved RNA editing events in both mitogenomes.

Fig. 5.

The distribution pattern of predicted RNA editing sites in PCGs of K. amabilis (A) and T. glandulifera (B)

Comparison of the mtDNA among Caprifoliaceae species

A ML tree was constructed based on the mtDNA sequences of 19 conserved PCGs from 25 species. The 19 shared PCGs contented atp1, atp4, atp6, atp8, atp9, ccmC, cox2, cox3, matR, mttB, nad1, nad3, nad4L, nad6, nad9, rps3, rps4, rps12, and rps13. Meanwhile, a Bayesian Inference (BI) tree was also constructed using the same dataset for verification. The topological structures of the ML tree and BI tree were completely consistent, and the branch support rates of the BI tree were higher than those of the ML tree. Both phylogenetic tree clustered the 25 species into three orders, which were consistent with the APG IV taxonomic classification. In the Caprifoliaceae clade, K. amabilis formed a group with T. glandulifera sistering to others, suggesting a closer evolutionary relationship between the two species (Fig. 6).

Fig. 6.

Maximum likelihood and Bayesian inference phylogenetic tree constructed from 25 species. The red numbers under each branch represent the support rates of the Bayesian inference tree

For further comparative analysis of the mitogenomes in Caprifoliaceae, 25 shared genes were extracted from six species for subsequent analysis. The ML tree with Arabidopsis thaliana as an outgroup revealed a similar interspecific relationship as observed in the previous analysis (Fig. 7A). The BlastN program was subsequently performed to compare homologous genes and their sequence arrangement. Mitogenomes of Caprifoliaceae species had numerous homologous collinear blocks with the relatively short sizes. Furthermore, there were eight homologous fragments and six inversions identified between the K. amabilis and T. glandulifera mitogenomes (Fig. 7B).

Fig. 7.

Phylogenetic analysis (A) and collinearity analysis (B) among K. amabilis, T. glandulifera, and related species. In the collinearity analysis, species are ordered from top to bottom to match the phylogenetic tree. Pink lines represent collinear blocks, and gray lines indicate inverted regions

The nucleotide diversity (π) of 25 shared genes among Caprifoliaceae species was also calculated (Fig. S1). The π values ranged from 0 to 0.21, with the highest diversity observed in atp1. Most of these genes did not exhibit significant differences in nucleotide diversity.

To assess selection pressures during the evolutionary process of closely related species, the ratio of nonsynonymous (Ka) to synonymous (Ks) substitutions, known as the Ka/Ks, was calculated (Fig. 8). The Ka/Ks ratios varied across genes, with most values below 1, indicating purifying selection. However, atp4, ccmB, and mttB exhibited ratios above 1 in all six Caprifoliaceae species, suggesting positive selection. Notably, the Ka/Ks ratio for ccmB was exceptionally high among these PCGs in six species, ranging from 1.44 to 1.68, implying a strong adaptive evolution. Interestingly, compared to its conservation in other species, the rps7 gene displayed a high value of 1.15 in T. glandulifera, hinting its species-specific adaptation to high elevation.

Fig. 8.

The substitution rates (Ka/Ks) of PCGs in 6 Caprifoliaceae species compared with A. thaliana

Intracellular gene transfer and gene rearrangement

The chloroplast genome of K. amabilis (NC_029874.1) was downloaded from NCBI and subsequently analyzed for homologous fragments with the mitochondrial genome. Sequences with > 70% similarity were determined as homologous fragments. A total of 34 homologous fragments were shared between the chloroplast and mitochondrial genomes, with alignment lengths ranging from 40 to 3,628 bp. The number of mismatches in these fragments varied from 0 to 176. These homologous fragments totaled 19,692 bp, accounting for approximately 2.67% of the K. amabilis mitogenome. These fragments were referred to as mitochondrial plastid DNAs (MTPTs), which represent the migration of sequences from the chloroplast to the mitochondrial organelle. Of these 34 homologous fragments, MTPT 1 was the largest, measuring 3,628 bp in length. These homologous sequences were further annotated to identify functional regions, which included 12 tRNA genes (Fig. 9A). Unfortunately, no chloroplast-derived fragments were detected in the mitogenome of T. glandulifera when compared with published chloroplast genomes (NC_045051.1). Subsequently, the published nuclear genomes of T. glandulifera were utilized to analyze the homologous fragments between the mitochondrial and nuclear genomes. A total of 51 homologous fragments were identified between the nuclear and mitochondrial genomes, with a total length of 190,488 bp. Among these, the longest homologous fragment was located at 38,200,182 bp-38,217,459 bp on chromosome 3 of the nuclear genome, with a length of 17,273 bp. Further annotation revealed 11 mitochondrial genes within these homologous fragments, including 7 PCGs and 4 tRNA genes (Fig. 9B). Notably, since the nuclear genome of K. amabilis has not yet been published, homologous sequence analysis between its nuclear and mitochondrial genomes could not be conducted.

Fig. 9.

(A) The MTPTs between the mitogenome and chloroplast for K. amabilis. Gray lines represent collinear blocks, and pink lines indicate inverted regions. B The NMTPs between the mitogenome and nuclear genome for T. glandulifera. Lines of different colors represent the homologous fragments between different chromosomes and the mitochondrial genome

The gene arrangement patterns within Caprifoliaceae mitogenomes were predicted, with Arabidopsis thaliana as the outgroup (Fig. 10). The gene order of all mitogenomes had undergone dominant rearrangements except for two gene block pairs (atp8-cox3 and atp4-nad4L) when compared with the ancestral gene state (A5). Mitogenomes of K. amabilis and T. glandulifera presented significant gene arrangements different from those of other taxa of Caprifoliaceae, in which the nad1 gene and nad2 gene underwent substantial duplication.

Fig. 10.

Inferred gene rearrangement events among K. amabilis (Kam), T. glandulifera (Tgl), Heptacodium miconioides (Hmi), Triosteum pinnatifidum (Tpi), Lonicera macrantha (Lma), and Lonicera japonica (Lja) with the A. thaliana (Ath) as the outgroup. Black boxes indicate the ancestral state of each branch

Discussion

Mitogenome composition and characterization

In the plant kingdom, mitochondrial genomes are well-known for their remarkable diversity in size and structure. Mitogenomes in plants are typically larger than their animal counterparts and exhibit substantial length variability, a characteristic largely attributed to their higher proportion of repetitive sequences [21, 46]. Even closely related species can exhibit striking differences genome size. For instance, within the genus Silene, the mitogenome size spans from 253 kb in S. latifolia [47] to 6.728 Mb in S. noctiflora [48]. Nowadays, there are few reports on the mitogenomes of Caprifoliaceae [49]. To deepen the understanding of the evolutionary process of Caprifoliaceae, the mitogenomes of K. amabilis and T. glandulifera were assembled and annotated for the first time in this study. Their genomes are 738 kb and 643 kb, respectively, presenting the high variability of mtDNA sizes in Caprifoliaceae as previously reported with approximately 500 kb to 1,000 kb in length (Fig. 1). However, all mitogenomes exhibit highly conserved GC contents (about 45%) - a feature that may help preserve overall genomic functionality and the integrity of genetic information [50, 51].

Through the annotation, a total of 42 PCGs were obtained in K. amabilis mitogenome, which is similar to L. macranthoides (42). K. amabilis has two and three copies of atp6 and atp9, respectively, suggesting the possibility of gene redundancy and functional divergence. In contrast, the annotation of T. glandulifera yielded 37 PCGs, smaller than K. amabilis but similar to H. miconioides (36) (Table S1). The mitogenome of T. glandulifera lacks seven genes, namely ccmFC, cob, nad2, nad5, nad7, rpl2, and rps11 (Table 1). Blast analysis revealed that only fragmented sequences of these missing genes were detected in the nuclear genome of T. glandulifera, except the gene rps11, for which a near-complete sequence was identified, implying for potential transfer to the nuclear genome or completely lost accompanied with genomic rearrangement. It has been shown that the ccmFC gene play an important role in energy metabolism in plants [52]. It interacts with the small heat shock protein gene HSP24.7 in cotton, affects mitochondrial reactive oxygen species (ROS) production, and ultimately helps the radicle break through the seed coat to complete germination [53]. The losses of the nad genes may have affected the respiratory chain function [54]. The deletion or rearrangement of these genes may be the result of the species’ response to environmental stresses, revealing the complexity and diversity of Caprifoliaceae in the evolutionary process.

Functional characteristics of repetitive sequences, codon usage bias, and RNA editing, and their impact for adaptive evolution

Repetitive sequences in the mitogenomes of higher plants promote genetic recombination and changes genome structure and size. Several studies, such as those on Melia azedarach [55] and Morus [56], have highlighted the critical role of repetitive sequences in mitogenome recombination and their intimate associations with genes. In general, the number of SSRs showed a positive correlation pattern with genome size [57]. However, a contradictory pattern was observed in this study: more SSRs were identified in the smaller mitogenome of T. glandulifera than in K. amabilis (Fig. 2A), which may be related to the distinct types and distribution characteristics of repetitive sequences between the two species. Meanwhile, more interspersed repeats were detected in the mitogenome of K. amabilis, particularly FRs and PRs (Fig. 2B). As reported in previous studies [58, 59], palindromic and forward repeats are key drivers of homologous recombination in plant mitogenomes. The higher abundance of PRs and FRs in K. amabilis may thus promote more frequent repeat-mediated recombination events, which in turn contribute to the formation of its complex bipartite mitochondrial genome structure.

Codon usage bias is an important reference for understanding gene evolution and translation mechanisms [19, 60]. A range of factors shape the usage patterns of synonymous codons, and organisms continuously refine their gene-coding features to adapt to their survival environments, ultimately leading to variations in codon preferences across both genes and species [31, 61]. In K. amabilis and T. glandulifera, codons exhibited consistent preferences across multiple analytical metrics including RSCUs, ENC, neutrality analysis, and PR2 analysis. This congruence across diverse statistical frameworks (which respectively quantify codon bias intensity, deviation from neutrality, and asymmetry in nucleotide usage) strongly indicates that these two species may have converged on similar evolutionary strategies to fine-tune translational efficiency and orchestrate gene expression regulation.

In mitochondria of higher plant, RNA editing is a ubiquitous phenomenon and is an essential step in the production of functional proteins by mitochondria. The abnormality of RNA editing affects mitochondrial function and leads to slow growth of plants [62, 63]. Among the genes encoding proteins of K. amabilis and T. glandulifera, the ccmB, mttB, and nad4 genes possessed the most RNA editing sites (Fig. 5). These three genes have different functions; the ccmB gene encodes a cytochrome C biosynthesis-related protein [64], the mttB gene is involved in mitochondrial membrane transport proteins [65], and the nad4 gene functions as encoding a subunit in the respiratory chain [54, 66]. Among them, nad4 is also the gene with the most RNA editing sites in other plants, such as walnut (Juglans regia) [67]. The PPR protein SLOW GROWTH 4 is involved in editing of the nad4 gene, and the editing of these sites is essential for the correct expression of the nad4 gene [68]. It indicates that RNA editing has a critical role in the functional fulfillment of these genes.

Phylogenetic relationships and evolutionary dynamics of mitochondrial genes in Caprifoliaceae

Phylogenetic analysis of the 25 mitogenomes found that K. amabilis and T. glandulifera formed a sister group, and interspecific relationships were identical to those in and previous works [13, 69], which further confirmed the practicability of mitogenome as a tool for phylogenetic analysis above species level. At the same time, it is suggested that more species information of Caprifoliaceae should be collected for phylogenetic analysis in the future, so as to further clarify the overall phylogenetic relationships within Caprifoliaceae. Further analyses by Ka/Ks showed significant differences in the evolutionary rates of atp4, ccmB, and mttB genes revealing their different strategies in adapting to environmental changes. Such differences may have influenced the growth and development of Caprifoliaceae species, which in turn affect their ecological niche distribution. For T. glandulifera in high-altitude habitats, low temperature and intense UV radiation are major stresses that disrupt ribosomal stability and impair protein synthesis, threatening cellular homeostasis and plant fitness. Relevant studies on rye have shown that ScRPS7 is more highly expressed in meristems than non-meristems; under low-temperature stress, its expression decreases sharply initially but gradually recovers to pre-stress levels after 30 days, thereby sustaining seedling growth [70]. Similarly, in T. glandulifera, the rps7 gene exhibits positive selection signals that may be associated with high-altitude adaptation. Given rps7’s core functions in ribosome assembly and protein translation, it is hypothesized that positive selection-driven beneficial mutations could enhance ribosomal stability and translation efficiency under low-temperature and intense ultraviolet stress, thereby maintaining stress-responsive protein synthesis in T. glandulifera. This adaptive modification alleviates high-altitude stress damage, stabilizes cellular metabolism, and improves the species’ survival and adaptive capacity in alpine habitats. In the future, transcriptome data can be used to detect the expression pattern of the rps7 gene under high-altitude stress conditions. It can also be further combined with proteomics data and functional verification experiments (such as gene overexpression and knockout) to verify the hypothesis proposed in this study and clarify the specific molecular mechanism of rps7 in mediating the high-altitude adaptation of T. glandulifera.

Gene rearrangement, duplication, and gene transfer

The types of protein-encoding genes are relatively uniform among higher plants, but their number, location, and order of arrangement tend to vary from species to species and even among different varieties of the same species. In a comprehensive comparative analysis of the six species of Caprifoliaceae, we found that the mitogenomes of Caprifoliaceae had undergone extensive gene loss, rearrangement, and structural variation. In general, cytochrome C synthase genes and ribosomal protein genes are more severely lost. Genes associated with the five mitochondrial respiratory chain are relatively conserved [71]. In the present study, the complex I genes nad2, nad5, and nad7 were not identified in T. glandulifera relative to the other five Caprifoliaceae species, indicating their species-specific loss during mitogenome evolution.

In addition to gene loss, frequent rearrangements are another characteristic shared by plant mitogenomes. It has been reported that angiosperm mitogenomes diverge rapidly at the structural level and show loss of collinearity even among closely related species [47, 72]. This feature was also reflected in the covariance analysis of mtDNAs of six Caprifoliaceae species (Fig. 7). Ancestral reconstruction analysis of these mitogenomes also showed obvious structural variation and rearrangement events (Fig. 10), implying for the dynamic evolutionary process of Caprifoliaceae mitogenomes. In our study, mitogenomes of K. amabilis and T. glandulifera have unique gene order different from other genera characterized by the duplicated nad1 gene and nad2 gene, indicating genus-specific recombination events.

Gene transfer between mitochondria and chloroplasts is considered an integral part of long-term plant evolution [73, 74]. In angiosperms, tRNA genes are frequently transferred from chloroplasts to mitochondrial genomes [75]. In this study, there were 26 chloroplast genome fragments identified in the mitogenome of K. amabilis. These findings demonstrate the potential for horizontal gene transfer between organelle genomes. 13 tRNA genes, including trnH-GUG, trnK-UUU, were transferred from the chloroplast genome to the mitogenome (Fig. 9A). Such gene transfer phenomenon not only enriches the diversity of the mitochondrial genome, but also may have an important impact on metabolic regulation and adversity adaptation in Caprifoliaceae. Interestingly, no MTPT was found in the mitogenome of T. glandulifera, possibly ascribed to gene loss and structural rearrangements. Studies have shown that the structural dynamics of mitochondrial genomes, including rearrangements, make it difficult to stably preserve MTPT, and gene loss may eliminate functional constraints related to MTPT, ultimately leading to undetectable MTPT in some species [76]. Additionally, a number of mitogenomic fragments transfer to nuclear genome (NMTPs) was detected in T. glandulifera (Fig. 9B), which was considered as a key factor shaping the composition, function and evolution of both organelles and nuclear genome [77]. To comprehensively understand intergenomic transfer in Caprifoliaceae species, more high-quality mitogenome and nuclear genome data and further investigations are required to explore the procedure of genome evolution in future works.

Conclusion

The mitochondrial genomes of K. amabilis and T. glandulifera were first completely assembled. Both species exhibited similar codon preferences, and positive selection was detected in atp4, mttB, ccmB. Phylogenetic analysis clarified their affinities with other Caprifoliaceae species. Comparative analyses revealed frequent genomic rearrangements and dynamic structural evolution of Caprifoliaceae mitogenomes The chloroplast-to mitochondrion gene transfer in K. amabilis reveals the genetic interactions among the organelles and provides insights into adaptive evolution. These findings not only enrich the understanding of the diversity of plant mitogenomes but also offer important clues for the molecular basis of plant adaptation to the environments.

Authors’ contributions

Project administration, ZLL; Software, Writing-Original Draft, JTW; Methodology, MZR and ZLL; Formal analysis, MMZ and HYD; Data Curation, Visualization, SNH, SYW, JTW; Resource, Investigation, RWH, XJ and CLY; Writing-Review & Editing, Validation, JTW, JFL and ZLL; Funding acquisition, JFL and ZLL. All authors have read and agreed to the published version of the manuscript.

Funding

Declaration.

This work was supported by the National Natural Science Foundation of China (32070354).

Data availability

The Illumina, PacBio data generated in this study are not publicly available due to the inclusion of unpublished raw genomic information but can be made available to qualified researchers upon reasonable request. The mitogenomes and genomes annotation have been deposited in the figshare database (10.6084/m9.figshare.30466820).

Declarations

Ethics approval and consent to participate

We are licensed to collect the plant of K. amabilis from Northwest University, Shaanxi Province, China. The collection complies with relevant institutional, national, and international guidelines and legislation.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.