Background

H. ammodendron is a small C4 perennial tree belonging to the Amaranthaceae family, formerly known as Chenopodiaceae [1]. The genus Haloxylon encompasses approximately 11 species, distributed from Iran to Mongolia and Northwestern China [2]. Referred to as the King of psammophytic woody plants, H. ammodendron demonstrates remarkable adaptability to severe environmental conditions, including desert conditions and sandstorms [3]. Widely found in the deserts of northwest China, it plays a crucial role in environmental protection through wind control, carbon sequestration, and sand fixation [4]. However, due to increasing overexploitation and the dramatic climate change occurring in central Asia, Haloxylon plants face a threat [5].

The declining numbers of these plants and the subsequent ecological impacts have sparked significant interest in genetic research on Haloxylon resources. Long et al. [6] provided the first transcriptome resource of H. ammodendron to investigate genes that confer drought stress resistance to the plant. For a thorough understanding of the resilience of H. ammodendron to drought stress, Gao et al. [7] conducted a complete transcriptome analysis under osmotic conditions, paired with evaluations of physiological factors such as proline, betaine, soluble sugar levels, and peroxidase activity. Using cpDNA genes (trnS-trnG and trnV) and ITS (Internal transcribed spacer) (ITS1-ITS4) sequences, Chen at al [8]. assessed 420 individuals from 36 populations of Haloxylon for their genetic diversity and historical demography. Suo et al. [9] utilized SSR marker-derived DNA markers to distinguish H. ammodendron and Haloxylon persicum species. Later, the cpDNA of H. ammodendron and H. persicum were assembled using short-read sequencing data. Recently, a high-quality chromosome-level assembly of H. ammodendron was released [10], paving the way for an in-depth investigation of genes of interest concerning its high tolerance to drought and salinity. Despite the availability of these genomic resources, the mitochondrial genome (mitogenome or mtDNA) of H. ammodendron is lacking, hampering mtDNA-based structural and functional evolutionary studies.

The mitogenome, often referred to as the cell’s powerhouse, plays a crucial role in cellular energy production [11]. Housed in the mitochondria, this distinct genome is responsible for producing essential proteins for oxidative phosphorylation, leading to the creation of adenosine triphosphate (ATP)– the primary energy currency of the cell [12, 13]. High variations in mtDNA size coupled with structural complexity, are the major factors slowing the pace of fully and accurately assembled mitogenome projects. For example, mitogenome sizes vary over 40-fold in the single genus Silene, from 253 kbp in Silene latifolia to more than 11 Mbp in Silene conica [14]. Variation in repetitive sequences, the incorporation of foreign sequences, and gain or loss of non-coding regions are among the key factors contributing to genome size variability [15, 16].

Furthermore, while a single circular mitogenome is common in angiosperms, various configurations may occur, including linear, fragmented, loop-like, branched, and multi-chromosomal structures [11]. The dynamic nature of mtDNA presents challenges in fully assembling the genomes of certain species compared to cpDNA assembly.

However, advancements in long-read sequencing technologies have made the assembly of plant mitochondrial genomes more achievable, facilitating comparative studies of plant mitogenome structure and evolutionary implications [17, 18]. In the present study, short and long-read sequencing data were used to generate the first mtDNA of H. ammodendron. The characterization of repeats content, sequence collinearity, DNA migration and RNA editing sites were explored.

Results

Genome assembly and annotation

In this study, we utilized Illumina short-reads and PacBio HiFi long-reads to assemble the mtDNA of H. ammodendron. Importantly, the data was not generated within this study; instead, we sourced the Illumina short-reads and PacBio long-reads from a previously conducted study on the nuclear genome of the species. Specifically, we accessed 49.4 Gbp of clean short reads (Table S1) from the NCBI SRA database under accession number SRR17127859. The assembly was facilitated using the GetOrganelle software, version 1.7.5 [19]. In parallel, the PacBio dataset comprised 22.6 Gbp of long-reads, also retrieved from NCBI SRA under the accession SRR17129371. When visualizing the short-read assembly graph using Bandage version 0.8.1 [20], an intricate and circular multi-branch mtDNA structure was assembled encompassing nodes or contigs (Fig. 1). These nodes were delineated by overlapping regions depicted by black lines on the assembly graph. Overlapping regions were resolved by aligning associated branch nodes with the PacBio HiFi dataset. Due to the presence of repeats, recombinant sequences could occur, leading to different configurations. Therefore, four potential genomic paths were inferred. Paths 1–1 and 1–2 were considered major configurations, while paths 2 − 1 and 2–2 represented minor configurations (Fig. 1, Table S2, Table S3). The major configuration was obtained by extending 2000 bp on both ends based on the repetitive sequence. As for the minor configuration, the 2000 bp extension was not supported by the long-reads data (Fig. 1, Table S2, Table S3). Consequently, two fully circular sequences representing the mtDNA of H. ammodendron, were obtained (Fig. 1). The two circularized contigs, namely Chromosome 1 and Chromosome 2, exhibited similar GC content (44%) spanning 121,403 bp and 88,746 bp respectively (Table 1).

Fig. 1
figure 1

Diagram of mitogenome recombination. The major configuration of Haloxylon ammodendron mitogenome is two independent circular mtDNA, with three pairs of repeats that can mediate gene recombination: HRS1, HRS2 and HRS3. According to the supported long-reads, the frequencies of mitogenome recombination mediated by HRS1, HRS2 and HRS3 were 44.2%, 4.92%, and 0.21%, respectively

Full size image
Table 1 Basic statistics of H. ammodendron mitochondrial genome
Full size table

The annotation of the H. ammodendron mtDNA is summarized in Table 2. A total of 31 protein-coding genes, 18 tRNA and three rRNA were predicted. Among the protein-coding genes, 24 core genes typically found in angiosperms [21], and seven variably present genes were identified. The core genes include five ATP synthase genes (atp1, atp4, atp6, atp8, and atp9), nine NADH dehydrogenase genes (nad1, nad2, nad3, nad4, nad4L, nad5, nad6, nad7, and nad9), four cytochrome c biogenesis genes (ccmB, ccmC, ccmFC, and ccmFN), three cytochrome c oxidase genes (cox1, cox2, and cox3), one membrane transport protein gene (mttB), one mature enzyme gene (matR), and one ubiquinol-cytochrome c reductase gene (cob). The non-core genes include one ribosomal large subunit gene (rpl5), five ribosomal small subunit genes (rps3, rps4, rps7, rps12, rps13), and one succinate dehydrogenase gene (sdh4). Regarding tRNA, three were in duplicate copy (trnC-GCA, trnG-GCC, trnP-UGG), while one was in triplicate copy (trnfM-CAU). Among the three rRNA, rrn26 was present in duplicate copy.

Table 2 Gene content information in H. ammodendron mitochondrial genome
Full size table

Gene transfer

Putative gene transfer between organelles was unveiled through alignment of the newly assembled cpDNA (Fig. 2A) with the mtDNA of H. ammodendron. Homologous sequence fragments between the organelles are illustrated in Fig. 2B. According to the sequence similarity analysis, 17 DNA fragments were identified as homologous between the mtDNA and cpDNA, with a total length of 5233 bp, accounting for 2.49% of the mtDNA (Table S4). Among them, the longest fragment measured 1158 bp in length. By annotating these homologous sequences, eight complete genes were found on the 17 homologous fragments, including one protein-coding gene (petG) and seven tRNA genes (trnD-GUC, trnH-GUG, trnM-CAU, trnN-GUU, trnP-UGG, trnS-GCU, trnW-CCA). Additionally, we investigated the mtDNA trace in the nuclear genome (Table S5). The longest alignment, with a length of 15 Kbp, was identified on chromosome 1 (Table S5).

Fig. 2
figure 2

H. ammodendron plastome overview and sequence similarity. (A) Depiction of the H. ammodendron plastome, providing a visual representation of its organization and key features. (B) Illustration highlighting sequence similarity between the mitochondrial and chloroplast organelles. Yellow arcs denote regions of the mitochondrial genome, while the purple arc signifies the chloroplast genome. The interconnecting green lines emphasize homologous segments shared between the two genomes

Full size image

Repeat sequences analysis

Microsatellites were screened for each set of chromosomes in the H. ammondendron mtDNA (Fig. 3A). For chromosome 1, a total of 42 simple sequence repeats (SSRs) were identified, comprising eight monomeric SSRs, eight dimeric SSRs, four trimeric SSRs, 17 tetrameric SSRs, four tetrameric SSRs, and one hexameric SSRs (Fig. 3A). Among the SSRs, the repeat motifs (A)10, (GAAA)3 and (CT)5 were found to be prevalent (Table S6). Seventeen tandem repeat sequences were identified, exhibiting a matching identity greater than 70% and lengths ranging from 14 to 31 bp (Fig. 3B, Table S7). Dispersed repeat sequences in chromosome 1 were also examined, detecting 467 repeat sequence pairs with a length equal to or greater than 30 bp. Among these, 234 pairs were palindromic repeats, 233 were forward repeats, and one was found to be a reverse repeat type. The longest observed palindromic repeat sequence spanned 123 bp, while the longest forward repeat sequence was 175 bp in length (Table S8).

Fig. 3
figure 3

Repeat patterns in the H. ammodendron mitochondrial genome. (A) Chart showing the count of Simple Sequence Repeats (SSRs) distributed in chromosome 1 and chromosome 2. (B) Visualization of the number of tandem and dispersed repeats found in both chromosomes, offering insight into the genome’s complexity and evolutionary history

Full size image

In chromosome 2, 24 SSRs were detected (Fig. 3A). Among these SSRs, nine were monomeric, two were dimeric, three were trimeric, nine were tetrameric, and one was the pentameric type. No hexameric SSRs were found in this chromosome. The most prominent SSR motif was thymine (T)10, among others (Table S7). A total of 10 tandem repeat sequences with a matching identity greater than 78% and lengths ranging from 18 to 42 bp were identified (Fig. 3B, Table S8). Dispersed repeat sequences in chromosome 2 were also examined, detecting 148 pairs of repeat sequences with a length equal to or greater than 30 bp. Among these, 69 pairs were palindromic repeats, and 79 were forward repeats. No reverse or complement repeats were detected in both chromosomes. The longest observed palindromic repeat sequence spanned 56 bp, while the longest forward repeat sequence was 114 bp in length (Table S8).

RNA editing

RNA editing, a widespread phenomenon in land plants, entails modifications within the coding region of the transcript involving the addition, loss, or conversion of nucleotides [22]. The current study identified 337 RNA editing sites, all of which were cytidine to uridine (C to U) conversion types. The predicted RNA editing sites in various genes are shown in Fig. 4. The ccmB and ccmFN genes exhibited the highest number (25) of RNA editing sites. Conversely, no RNA editing sites were observed in the sdh4 gene. The amino acid transition exhibiting the highest abundance was from serine to leucine, comprising 19.58% (66 sites) of the observed transitions (Table S9).

Fig. 4
figure 4

RNA editing sites in H. ammodendron mitochondrial genes. This figure presents a comprehensive count of RNA editing sites discovered within 31 mitochondrial protein-coding genes of H. ammodendron, shedding light on post-transcriptional modifications and their potential implications

Full size image

Codon usage

The codon usage pattern of the H. ammodendron mtDNA was presented in Fig. 5. Codons exhibiting a relatively synonymous codon usage (RSCU) value greater than one were regarded as being favored by the corresponding amino acids. Apart from the start codon AUG and the tryptophan codon UGG, both of which had an RSCU value of 1, there were widespread codon usage preferences in the H. ammodendron mitochondrial protein-coding genes (Fig. 5). For example, alanine (Ala) showed a strong preference for the codon GCU, with the highest RSCU value of 1.64. In addition, leucine (Leu) preferred the codon UUA, with an RSCU value of 1.61 (Table S10).

Fig. 5
figure 5

Codon usage preferences in H. ammodendron‘s mitochondrial genome. Graphical representation illustrating the Relative Synonymous Codon Usage (RSCU) values. This showcases the preferences of certain codons encoding the same amino acid, hinting at evolutionary pressures and translational optimization

Full size image

Phylogenetic analysis and collinearity analysis

The phylogenetic tree based on mtDNA (Fig. 6) was constructed using a maximum likelihood approach, incorporating a diverse set of 30 species representing various orders (Table S11). The resulting tree topology aligns with the most recent classification the Angiosperm Phylogeny Group (APG) provided. H. ammodendron and the Amaranthaceae species Spinacia oleracea, Chenopodium quinoa, and Suaeda glauca were placed within the Caryophyllales order. Interestingly, H. ammodendron exhibited a close phylogenetic relationship with S. glauca, another member of the Amaranthaceae family, widely distributed in the arid areas of northwest China [23]. These results were robustly supported by Bayesian phylogenetic tree inference (Figure S1). To identify both conserved and unique mitochondrial regions among Caryophyllales species, we performed pairwise alignments using Spinacia oleracea (NC_035618.1), S. glauca (NC_060419.1), Alternanthera philoxeroides (MN166292.1), Beta macrocarpa (NC_015994.1), Silene latifolia (NC_014487.1), Mirabilis jalapa (NC_056991.1), and Fallopia aubertii (MW664926.1) mtDNA (Fig. 7). The alignment results were provided in Fig. 7 and Table S12. Although collinear blocks were noticeable within Caryoplyllales species, their length is relatively short, indicating a non-conservative mitogenome structure. It is worth pointing out that the homologous block arrangement is inconsistent between H. ammodendron and its congeners, implying an evolutionary mtDNA rearrangement occurrence. Additionally, some blank regions were observed in H. ammodendron, representing species-specific sequences lacking homology with other species.

Fig. 6
figure 6

Phylogenetic placement of H. ammodendron within Caryophyllales. A Maximum Likelihood phylogenetic tree that delineates the evolutionary relationship of H. ammodendron with other species in the Caryophyllales order. Node labels display support values, with maximum-likelihood bootstrap values offering a measure of the tree’s reliability

Full size image
Fig. 7
figure 7

Comparative collinearity in caryophyllales mitogenomes. A landscape view showcasing the genome collinearity between H. ammodendron and its Caryophyllales counterparts. The diagram helps in understanding genomic rearrangements, evolutionary events, and conservation levels among these species

Full size image

Discussion

In land plants, mtDNA has undergone extensive genomic reorganization and gene arrangements, resulting in rapid structural evolution. Studying plant mtDNA offers a path into their structural organization, genetic diversity, and evolutionary dynamics, which have implications for understanding their functional role in biological processes such as plant growth, energy production, and respiration.

In the present study, we took advantage of both short- and long-read sequencing data to assemble the first mtDNA of H. ammodendron. The assembly mitogenome length was 210,149 bp, which is about 50% shorter than S. glauca (NC_060419.1, assembly size: 474,330 bp), and P. aculeate (NC_067638.1, assembly size: 515,187 bp) but in similar range with M. jalapa (NC_056991.1, assembly size: 267,334 bp), all belonging to the same order: Caryophyllales. The observed genome size variation could be imputable to several factors, including repeat elements density, intragenic regions gain or loss, and the introgression of foreign DNA [24,25,26].

Repeats elements such as microsatellites, tandem and dispersed repeats are prone to recombination leading to isomeric sequence sets in the mitogenome [14, 27, 28]. Therefore, the repeats dynamism could impact DNA maintenance and evolution of the plant mtDNA [16]. Unusually sized repeats within plant mtDNA exert significant influence, marked by their excessive size relative to coding capacity, a low mutation rate in genes, and a substantial rearrangement rate [16]. Besides, ubiquitous short tandem repeats play pivotal roles in mtDNA evolution, contributing to tandem duplications and structural variations [29]. It is worth to mention that homologous recombination, facilitated by rich repeated sequences in plant mtDNA, emerges as a major evolutionary driver [30]. Additionally, repeat-mediated recombination stands out as a highly active and substantial contributor to structural rearrangements in mtDNA [31].

One of the complexities of the mtDNA is the variability in molecular conformation. Several configurations have been documented, including circular, linear, sigma-like, fragmented, and multi-chromosomal structures [30]. Here we reported a multi-chromosomal mitogenome architecture that consists of two circular molecules. Similar findings were reported in other Caryophyllales species, such as Fagopyrum esculentum (10 circular chromosomes) [32], and Fallopia multiflora (two circular chromosomes) [33]. In the same plant order, a multi-chromosomal architecture with more than 50 chromosomes was also found for one of the symbolic plants, Silene conica, well known for its massive mitogenome spanning 11 Mb [14]. However, the close relatives S. glauca (NC_060419.1) [18] and Chenopodium quinoa (NC_041093.1) [17] present a single molecule in a fully circularized conformation based on long reads data-based assembly. The origin and maintenance of the multi-ring structure in plant mtDNA are areas of ongoing research [34,35,36,37,38,39]. It is believed that mechanisms such as recombination and rearrangement events, possibly mediated by repeated sequences [40,41,42], could play a role in generating and maintaining this phenomenon. The balance between the recombination and replication of mitochondrial rings was proposed as a key aspect that could influence the occurrence of mtDNA multi-chromosome structure [43, 44]. Besides, gain or loss of chromosome sets is suggested to be one of the key biological events leading to the diversity of mtDNA structures across different organisms [35], providing insights into the mechanisms driving the mitochondrial evolution.

DNA transfer among chloroplasts and mitochondria allows genetic exchange between cellular compartments and can serve as a driver for expansion or contraction of mitochondrial genomes in plants. Our study detected sequence transfer between the cpDNA and the mtDNA, indicating possible horizontal or intracellular gene transfer [45]. Sequences exchange between mtDNA and cpDNA was also noted for other Caryophyllales members including Mammillaria huitzilopochtli [46].

RNA editing is a post-transcriptional mechanism in higher plant organelles, contributing to improved protein folding [47]. In angiosperm mitochondria, extensive cytidine to uridine RNA editing occurred, playing a crucial role in transcript expression [48]. Interestingly, all protein-coding genes analyzed from H. ammomendron mtDNA showed C- to-U conversion, which may affect the start or end positions of the coding genes. It is worth noting that proteins translated from edited RNAs could play a critical function in an environment change context, resulting in plant acclimatization and survival capabilities [49]. Indeed, H. ammodendron is well known for its remarkable tolerance to saline and dry environments [50, 51].

The maximum likelihood phylogenetic inference resulted in a tree with strong support values, placing H. ammodendron in the close relative taxon S. glauca. Our tree topology is in perfect agreement with the recent results based on the Angiosperm Phylogeny Group (APG IV) classification [52], ITS [53], plastid markers [54], the recent nuclear whole genome sequence data [10], and the Bayesian phylogenetic tree indicating the reliability of the mitogenome-based tree.

Conclusions

In the present study, the mtDNA of H. ammodendron was characterized. The mtDNA exhibited a multi-chromosomal conformation with two circular molecules with a total length of 210,149 bp. It contains 31 protein-coding genes, 18 tRNA and 3 rRNA. A set of 17 sequence fragments were found homologous between the mtDNA and cpDNA, spanning 5,233 bp, accounting for 2.49% of the mitogenome. RNA editing assessment revealed that the protein-coding genes were mainly subjected to cytidine-to-uridine conversion. The phylogenetic inference showed a close relationship with S. glauca. The mtDNA resource provided by this study laid a foundation for further investigations into the comparative evolution and functional role of the H. ammodendron mtDNA.

Methods

Assembly and annotation

Before the mtDNA assembly, short-reads data were trimmed using fastp v0.23.4 [55]. To assemble the H. ammodendron mitogenome, a hybrid approach encompassing both short and long-read sequencing data was implemented. Firstly, using the Illumina short-reads data, a draft mtDNA assembly was conducted with GetOrganelle v1.7.5 [19] with the following settings:-R 15 -k 21,45,65,85,105 -F embplant_mt. The resulting genome graph sketch was then visualized using Bandage v. 0.8.1 [20], manually inspected and curated for node junctions. Using minimap2.2.26 (r1175), we mapped PacBio HiFi data onto the initial assembly to address conflictual nodes and repetitive regions [56].. In cases where multiple alternative connections exist at branching nodes, preference was given to connections supported by long reads. The obtained assembly was validated by mapping both short and long reads to ensure the connectivity and consistency of the assembly.

The assembly was annotated with GeSeq online tool (https://chlorobox.mpimp-golm.mpg.de/geseq.html#) [57] with the following parameters: Arabidopsis thaliana (NC_037304), Nicotiana tabacum (NC_ 006581), Glycine max (NC_020455), and Suaeda glauca (NC_060419.1) were set as reference mitogenomes; BLAT search (protein search identity = 25, rRNA, tRNA, DNA search identity = 85) for coding protein genes, rRNA and tRNA prediction. Additionally, tRNAscan-SE v2.0.7 [58] was used for tRNA prediction with a cut-off score for reporting tRNAs = 15. The annotated mtDNA was visualized using Apollo v2.7.0 [59] for manual correction. The genome map was rendered using the Organellar Genome DRAW (OGDRAW) tool [60].

Repeat sequence analysis

The identification of repetitive sequences, including microsatellite sequence repeats (SSRs), tandem repeats, and dispersed repeats, was performed using MISA (https://webblast.ipk-gatersleben.de/misa/) [61], TRF (https://tandem.bu.edu/trf/trf.unix.help.html) [62], and the REPuter (https://bibiserv.cebitec.uni-bielefeld.de/reputer/) [63] web servers respectively.

RNA editing and codon usage bias analyses

The protein-coding sequences of the mitochondrial genome of H. ammodendron were extracted using the PhyloSuite toolbox [64]. Prediction of RNA editing sites was performed using the online tool PREPACT3 (http://www.prepact.de/) [65], while MEGA v7.0 software [66] was employed to calculate the relative synonymous codon usage (RSCU).

Collinearity analysis

Using the BLASTN v2.10.1 + tool, we conducted pairwise alignments to assess sequence similarity between H. ammodendron and its related species (Spinacia oleracea, Suaeda glauca, Alternanthera philoxeroides, Beta macrocarpa, Silene latifolia, Mirabilis jalapa, Fallopia aubertii) [67]. Homologous sequences exceeding 500 bp were retained as conserved collinear blocks for constructing the multiple synteny plot with the help of MCscanX implemented in TBTools v1.098746 [68].

Gene transfer analysis

To detect the putative gene transfer between cpDNA and mtDNA, we performed a de novo assembly and annotation of the cpDNA of our sample using GetOrganelle v1.7.5 [19] and GeSeq web server [57], respectively. Then, an alignment of organelles was performed using BLAST v2.10.1+ [67] with the following settings: E-value ≤ 1E-10, matching rate ≥ 70%, and matching length ≥ 40 bp. Utilizing the genome of H. ammodendron, we applied a similar approach by aligning the mtDNA onto the nuclear genome to investigate the sequence transfer between the mtDNA and the nuclear genome.

Phylogenetic analysis

To infer the mtDNA-based phylogenetic tree, a set of 15 conserved genes (atp1, atp4, atp6, atp8, ccmB, ccmC, ccmFC, ccmFN, cox2, matR, nad1, nad2, nad3, nad5, and nad6) from 30 species belonging to four orders (Caryophyllales, Ericales, Santalales, and Zygopgyllales) (Table S10) were extracted from each mtDNA and aligned using MAFFT v7.505 [69]. Poorly aligned sequence regions were trimmed with trimAl v 1.4.1 [70]. The resulting data matrix was then concatenated with PhyloSuite [64] (Supplementary file 1), and the tree was constructed using IQ-TREE v2.0.3 following the Maximum Likelihood approach with the GTR + I + G model [64]. To assess the tree topology, a Bayesian method was performed using MrBayes v3.2.6 tool [71] with the following settings: number of substitution types: 6; Model: 4by4; rates variation across sites: invgamma; number of generations: 10,000, sample frequency: 10; Burnin: 250. The tree was rendered with the Interactive Tree of Life tool (iTOL) v5.0 [72].