Fresh blades of Q. litseoides were collected from Wutong Mountain, Shenzhen (113°17′E, 22°23′N; Alt. 943.7 m a.s.l), immediately frozen in liquid nitrogen, and stored at -80°C until DNA extraction. Total genomic DNA was extracted using the improved CTAB method (Porebski et al., 1997). The size and integrity of the extracted DNA were evaluated by electrophoresis on a 0.75% agarose gel, which enables assessment of DNA fragmentation and degradation. DNA purity and concentration were assessed using a NanoDrop One spectrophotometer (Thermo Fisher Scientific, USA) and a Qubit 3.0 fluorometer (Life Technologies, Carlsbad, CA, USA), respectively.

Prior to conducting Single Molecule Real Time (SMRT) sequencing, high-quality genomic DNA underwent stringent quality control. The genomic DNA was fragmented into large fragments, and then damage repair, adapter ligation, and fragment selection were performed to construct a PCR-free SMRT bell library. After size selection and quantification, the SMRT bell library was sequenced on the PacBio Revio platform. Following quality control and filtering using the SMRT Link v12.0 software, the sequencing yielded 20.16 Gb of high-throughput third-generation data.

The third-generation HiFi data was assembled using PMAT software (https://github.com/bichangwei/PMAT.git) (Bi et al., 2024), yielding a draft mitogenome assembly in Graphical Fragment Assembly (GFA) format. The GFA file was visualized using Bandage software (Wick et al., 2015) to examine the overall genome structure and detect complex connections mediated by repeat sequences. The final mitogenome was obtained by manually identifying and resolving circular molecules through tracing valid paths in the graph. To assess the accuracy and completeness of the mitogenome assembly, we realigned the PacBio HiFi long reads to the assembled mitogenome and calculated the coverage depth for each of the three molecular forms. The coverage plots were subsequently visualized using R. The assembly was annotated using the online annotation tool PMGA (http://www.1kmpg.cn/pmga/) (Li et al., 2025b). The annotation errors were carefully checked and manually corrected using Geneious R9.0.2 software (Kearse et al., 2012). Finally, the mitogenome map was drawn using the Plant Mitochondrial Genomes Map (PMGmap) (http://www.1kmpg.cn/pmgmap) (Zhang et al., 2024).

Simple sequence repeats (SSRs) were identified utilizing the MISA web service (https://webblast.ipk-gatersleben.de/misa/) (Beier et al., 2017), employing the parameters “1-10 2-5 3-4 4-3 5-3 6-3”. Subsequently, tandem repeats were detected using the TRF tool (https://tandem.bu.edu/trf/trf.html) (Benson, 1999) with the parameters “2 7 7 80 10 50 500”. Interspersed repeats were discerned through the REPuter web server (https://bibiserv.cebitec.uni-bielefeld.de/reputer) (Kurtz et al., 2001), specifying the parameters “-c -f -p -r -l 30 -h 3 -best 50”. Repetitive sequences were visualized using Origin software.

The coding sequences (CDS) of the mitogenome were extracted using Geneious R9.0.2 software (Kearse et al., 2012). The codon preferences for protein-coding genes (PCGs) were analyzed with CodonW software (http://codonw.sourceforge.net/), and relative synonymous codon usage (RSCU) values were calculated. An RSCU value greater than 1 indicates a codon is used more frequently than expected under equal usage assumptions. Stacked bar charts were generated using the ggplot2 package in R.

RNA editing, a post-transcriptional modification, is ubiquitously observed in eukaryotes, including plants. It entails alterations such as the addition, deletion, or conversion of bases within the coding region of transcripts (Jobson and Qiu, 2008; Wang et al., 2024b). To predict RNA editing sites in Q. litseoides, we employed the Deepred-Mt software (Edera et al., 2021). This tool utilizes a convolutional neural network (CNN) model and shows higher accuracy than previous methods (Edera et al., 2018). We chose results with probability values greater than 0.9.

Previous research suggests the possibility of genetic material transfer between cellular compartments or organelles (Huang et al., 2003; Wang et al., 2024a). This study investigated gene transfer between organelles in Q. litseoides. We retrieved its chloroplast genome (NCBI: ON598394) and performed BLASTn analysis (v2.9.0) with the following parameters: an E-value of ≤ 1e−5, match rate of ≥ 70%, and alignment length of ≥ 40 bp. The results of the BLAST pairwise sequence alignment were subsequently visualized using the circlize package in R.

To explore the conserved mitochondrial regions among closely related species, including Castanea henryi, C. mollissima, Castanopsis carlesii, Fagus sylvatica, Lithocarpus litseifolius, Quercus acutissima, Q. cerris, Q. ilex, Q. petraea, Q. robur, and Q. variabilis, we performed pairwise alignments utilizing BLASTn (Chen et al., 2015). We retained only matches that exceeded 500 bp in order to evaluate the synteny patterns of the mitogenomes under investigation. Subsequently, we visualized the synteny of multiple genomes using NgenomeSyn v1.41 (He et al., 2023).

To investigate the evolutionary placement of Q. litseoides within Quercus, we performed phylogenetic analyses based on both mitochondrial and chloroplast genomes. Eight Quercus species with available complete mitogenomes were included: Quercus acutissima, Q. cerris, Q. chenii, Q. ilex, Q. litseoides, Q. petraea, Q. robur, and Q. variabilis. Fagus sylvatica was selected as an outgroup to root the phylogenetic trees. Accession numbers and genome sources for the nine species were provided in Supplementary Table S1 .

For mitochondrial phylogeny reconstruction, 29 shared mitochondrial PCGs were identified using the PhyloSuite v1.2.3 software and extracted for analysis (Xiang et al., 2023; Zhang et al., 2020). For chloroplast phylogeny reconstruction, the complete chloroplast genomes were used. The multiple sequence alignments were executed via the MAFFT v7.524 software with default parameters (Katoh and Standley, 2013). Phylogenetic trees were reconstructed using the Maximum Likelihood (ML) method implemented in the IQ-TREE v2.1.3 software, with automatic model selection (-m TEST) and 1,000 ultrafast bootstrap replicates (-bb 1000) to assess nodal support (Nguyen et al., 2015). The phylogenetic analysis results were ultimately visualized through FigTree v1.4.4 software.

The mitogenome of Q. litseoides displayed a multipartite architecture, comprising three discrete molecules. The assembly revealed two circular and one linear mitochondrial DNA molecules, totaling 516,686 bp with 45.67% GC content ( Figure 1 ). Specifically, the circular molecules 1 and 2, along with the linear molecule 3, have lengths of 337,926 bp, 111,038 bp, and 67,722 bp respectively, with corresponding GC contents of 45.91%, 45.06%, and 45.44% ( Figure 2 ). Coverage depth analysis revealed that all three molecular structures exhibited continuous and uniform coverage curves, with no low-depth or ambiguous regions detected at junctions or repeat regions ( Supplementary Figure S1 ).

In the mitogenome of Q. litseoides, a total of 38 PCGs are annotated, comprising 25 mitochondrial core genes and 13 non-core genes, along with 23 transfer RNA genes (tRNAs) (trnL-GAG and trnP-UGG being multi-copies) and three ribosomal RNA genes (rRNAs). The core genes comprised 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 ubiquinone cytochrome c reductase genes (ccmB, ccmC, ccmFC, and ccmFN), three cytochrome c oxidase genes (cox1, cox2, and cox3), one transmembrane protein gene (mttB), one maturase gene (matR), and one cytochrome c biogenesis gene (cob). Notably, atp9 was duplicated ( Table 1 ). The exon-intron organization suggested that nad1, nad2, and nad7 each contained five exons ( Supplementary Figure S2 ). While nad7 showed cis-splicing, other genes exhibited trans-splicing ( Supplementary Figure S3 ), indicating distinct splicing mechanisms in Q. litseoides.

In the mitogenome of Q. litseoides, we identified 112, 34, and 17 SSRs in molecule 1, molecule 2, and molecule 3, respectively ( Figure 3A ; Supplementary Table S2 ). Tetrameric repeats constituted the predominant repeat type in this mitogenome, with 67 identified instances, accounting for 41.10% of all SSRs. In addition, 41 monomeric repeats (25.15%) and 34 dimeric repeats (20.86%) were identified, although they were less abundant than tetrameric repeats. Notably, pentameric repeats occurred exclusively in circular molecule 1, while hexameric repeats were absent throughout the mitogenome. The composition analysis revealed that most SSRs consisted of adenine (A) and thymine (T) base pairs. Within the mitogenome of Q. litseoides, we detected 15 tandem repeats ( Figure 3B ; Supplementary Table S3 ). Among interspersed repeat sequences, only palindromic and forward repeats were identified, with no reverse or complement repeats detected. Palindromic repeats were found in molecule 1 and 3 (17 and 3 pairs, respectively), while forward repeats were restricted to molecule 2 (11 pairs) ( Figure 3B ). All interspersed repeats exceeded 30 bp in length ( Supplementary Table S4 ).

The codon usage bias was analyzed in 28 screened PCGs of the Q. litseoides mitogenome. Supplementary Table S5 summarized the distribution of codon usage for each amino acid. The analysis revealed significant codon preference, with a total of 29 codons showing RSCU values > 1 ( Figure 4 ; Supplementary Table S5 ). Among these, GCU (Ala), UAU (Tyr), and CAU (His) were the three most frequently used codons in Q. litseoides. We observed that most preferred codons ended with uracil (U), which may reflect mitochondrial tRNA abundance or mutational bias.

Using the deep representation learning method (Deepred-Mt), we identified 494 C-to-U RNA editing sites across 36 PCGs in the Q. litseoides mitogenome ( Supplementary Table S6 ). The distribution analysis showed that nad4 contained the most editing sites (n=42), followed by ccmB (n=35), while atp1 had only two editing sites ( Figure 5A ). Among these sites, 464 (93.93%) caused non-synonymous changes, while 30 (6.07%) remained synonymous. These non-synonymous editing events resulted in 12 distinct amino acid substitutions. The most frequent substitutions were serine (Ser) to leucine (Leu) (n=112) and proline (Pro) to leucine (Leu) (n=108) ( Figure 5B ), suggesting potential functional implications for mitogenome protein stability.

In the mitogenome of Q. litseoides, we identified 15 homologous fragments shared with the chloroplast genome (excluding sequences aligned with chloroplast inverted repeat regions) ( Figure 6 ; Supplementary Table S7 ). These chloroplast-derived sequences transferred to the mitogenome were designated as MTPTs. These identified homologous fragments ranged from 64 to 1,448 bp in length, totaling 6,183 bp (approximately 1.20% of the mitogenome). Annotation of these sequences identified two PCGs (atpF and petG), ten tRNA genes (trnV-GAC, trnV-GAC, trnA-UGC, trnD-GUC, trnM-CAU, trnI-CAU, trnW-CCA, trnP-UGG, trnH-GUG, and trnN-GUU), and one rRNA gene (rrn16S). These results demonstrated extensive gene transfer between organelles in Q. litseoides. The gene content analysis suggested that most transferred sequences were tRNA genes, indicating potential functional retention of these chloroplast-derived sequences in mitogenomes.

Homologous block analysis was conducted for 12 species of Fagaceae family. The multi-synteny plot clearly indicated several homologous collinear regions between Q. litseoides and its closely related species ( Figure 7 ). For instance, the length of the longest matching fragment between Q. litseoides and Q. acutissima was 27,460 bp ( Supplementary Table S8 ). However, these homologous fragments were relatively short, indicating limited mitogenome structural conservation among these species. Furthermore, the observed gaps suggested unique genomic features in Q. litseoides, as most sequences lacked homology with other species. These findings demonstrated extensive gene rearrangements in the Q. litseoides mitogenome compared to its relatives.

To explore the evolutionary relationships within Quercus, we reconstructed two phylogenetic trees using both mitochondrial and chloroplast genomes ( Figure 8 ). The two phylogenetic trees exhibited similar topological structures, both with high bootstrap support values. Except for the differing placement of Q. acutissima, Q. chenii, and Q. variabilis, the remaining species showed consistent phylogenetic positions in both trees. Quercus litseoides was closely related to Q. cerris in both phylogenetic trees. Additionally, Q. robur and Q. petraea formed a strongly supported sister group (with 100% bootstrap support) in both trees, corroborating their classification within section Quercus. Overall, the phylogenetic analyses revealed both concordant and discordant evolutionary signals between mitochondrial and chloroplast genomes.

Mitochondria are essential organelles that function as cellular powerhouses and possess a highly complex genomic organization characterized by considerable sequence polymorphism and structural variation (Kozik et al., 2019). Although plant mitogenomes are typically depicted as circular molecules, recent studies have revealed that their actual structure in vivo is far more complex and dynamic, comprising a mixture of multimolecular forms, with linear DNA, circular and branched molecules (Gualberto et al., 2014; Kozik et al., 2019; Sloan et al., 2018). According to the most recent data accessible in the NCBI Genome Database (accessed: February 20, 2025), complete mitogenomes have been reported for 11 Fagaceae species, including Quercus acutissima (MZ636519.1) (Liu et al., 2022), Q. variabilis (MN199236.1, unverified) (Bi et al., 2019), Q. cerris (OY770018.1, unannotated), Q. robur (OW028777.1, unannotated), Q. ilex (OZ205167.1, unannotated), Q. petraea (OZ066324.1, unannotated), Fagus sylvatica (NC050960.1) (Mader et al., 2020), Lithocarpus litseifolius (NC065018.1) (Qiu et al., 2025), Castanea henryi (PP856681.1) (Tu et al., 2024), C. mollissima (OP895669.1) (Guo et al., 2023), and Castanopsis carlesii (PP853255.1) (Tu et al., 2024). Structural diversity within the Fagaceae mitogenomes is also well documented, such as the bi-circular structure in C. mollissima (Guo et al., 2023), the tripartite circular-linear organization in Q. acutissima (Liu et al., 2022) and the conventional single circle in F. sylvatica (Mader et al., 2020).

In this study, we reported the mitogenome of Q. litseoides from section Cyclobalanopsis, revealing a unique structure composed of two circular molecules and one linear molecule. This multipartite structure partially resembled that of Q. acutissima but differed in molecule number and composition, indicating lineage-specific structural diversification within Quercus. The mitogenome of Q. litseoides spanned 516,686 bp, which was intermediate in size between that of Q. acutissima (448,694 bp) and L. litseifolius (573,177 bp) (Liu et al., 2022; Qiu et al., 2025). Notably, it was the second longest among the currently known Quercus mitogenomes, following Q. ilex. The GC content serves as a crucial element in the evaluation of species (Liu et al., 2022). The GC content of this mitogenome was 45.67%, closely aligning with values observed in other Fagaceae species (such as Q. acutissima 45.72% and Q. variabilis 45.76%). This similarity suggests base composition conservation across this family, which may contribute to the maintenance of mitochondrial functional stability.

Large repetitive sequences (>1 kb), commonly present in angiosperm mitogenomes as 2–3 copies, mediate homologous recombination to form characteristic multipartite structures (Cole et al., 2018; Odahara et al., 2021). Such recombination promotes genomic heteromorphism and structural diversity (Liu et al., 2022; Wynn and Christensen, 2019). In the Q. litseoides mitogenome, simple sequence repeat (SSR) analysis revealed that tetrameric repeats were the predominant type (41.10%), consistent with findings in other Fagaceae mitogenomes (Tu et al., 2024). However, the longest interspersed repeat (282 bp) was significantly shorter than those observed in Q. acutissima (10,578 bp) and Fagus sylvatica (918 bp), suggesting potentially constrained recombination activity (Liu et al., 2022; Mader et al., 2020). This low recombination trait may reflect an adaptive evolutionary strategy that helps maintain genome integrity and structural stability.

In parallel, codon usage analysis offered additional insights into the evolutionary constraints acting on the Q. litseoides mitogenome. Variation in codon usage frequency among eukaryotes is attributed to long-term evolutionary selection pressures (Tu et al., 2024). The RSCU value, indicating the ratio of observed codon usage frequency to the expected frequency under no bias, is crucial for evaluating species-specific codon preferences. Analyzing codon preferences is essential for elucidating the evolutionary dynamics of species (Sharp and Li, 1986). In the Q. litseoides mitogenome, a total of 29 codons exhibited RSCU values greater than 1, consistent with patterns observed in other Fagaceae species, suggesting a conserved codon usage bias within this family. This conservation likely contributes to translational efficiency and reflects selective constraints acting on mitochondrial gene expression.

RNA editing is a pivotal post-transcriptional regulatory mechanism in plant organelles (Edera et al., 2018), primarily involving cytosine-to-uracil (C-to-U) conversions that can alter amino acid sequences and expand protein diversity (Ma et al., 2022). In the Q. litseoides mitogenome, we detected a total of 494 RNA editing sites in 36 PCGs, all of which were C-to-U conversions. The most common amino acid transitions caused by these sites were serine-to-leucine (Ser-Leu) and proline-to-leucine (Pro-Leu), which was consistent with previous reports (Tu et al., 2024). Remarkably, RNA editing showed strong gene-specific variation, ranging from the highly edited nad4 to the entirely unedited rpl2. These editing events are believed to contribute to translational diversity and may play a role in functional adaptation by modulating protein structure and function at the post-transcriptional level.

In addition to RNA editing, gene transfer between organelles also shapes the functional plasticity of plant mitogenomes. During the evolution of higher plants, mitogenomes have frequently incorporated fragments from plastid DNA, known as mitochondrial plastid DNA sequences (MTPTs) (Gui et al., 2016). Gene transfer between organelles is of evolutionary significance and has been widely observed as more mitogenomes and chloroplast genomes become available (Sloan and Wu, 2014). There is considerable variation in the length of the transferred fragments across various higher plant species (Gong et al., 2024). In this study, we identified 15 homologous fragments between the Q. litseoides organelles, representing 1.20% of the mitogenome and including 13 complete genes. Notably, the transfer of ten tRNA genes substantiates the frequent and functional exchange of genetic material between organelles in angiosperms (Bi et al., 2016). These transferred fragments not only provide raw materials for mitogenome remodeling but may also enhance functional diversification of the mitogenome.

Collectively, the observed RNA editing patterns and gene transfers between organelles highlight the functional flexibility of the Q. litseoides mitogenome. These features may reflect evolutionary adaptations that help maintain mitochondrial functionality under selective pressures, and they also offer valuable insights into the genetic mechanisms potentially related to the species’ endangered status.

Synteny analysis, which evaluates the sequence similarity of homologous sequences, is considered a powerful approach for investigating evolutionary relationships among species (Xu et al., 2023). In plant mitogenomes, the extent of synteny between two species can serve as an indicator of evolutionary distance and offer insights into their phylogenetic relationships. Moreover, rearrangements in plant mitogenomes play pivotal roles in enhancing genetic diversity, driving adaptive evolution, and regulating developmental and reproductive processes (Huang et al., 2025a). In this study, Q. litseoides exhibited extensive rearrangements compared to closely related species, reflecting a highly non-conserved and dynamic structural evolution. This phenomenon may be related to the structural plasticity of these mitogenomes. Despite these rearrangements, Q. litseoides retained relatively large syntenic blocks with Q. acutissima, Q. cerris, and Q. ilex, demonstrating conserved genomic regions and a shared evolutionary ancestry with these species. These results collectively underscore both the structural flexibility and the phylogenetic proximity of Q. litseoides within Quercus.

In this study, phylogenetic trees were reconstructed based on both mitochondrial and chloroplast genomes, and the results revealed highly consistent topological structures, highlighting the robustness of evolutionary relationships within Quercus. Notably, Q. litseoides consistently clustered with species from section Cerris in both trees, further supporting its phylogenetic placement and aligning with its traditional taxonomic classification within subgenus Cerris (Denk et al., 2017). Similarly, Q. robur and Q. petraea, both belonging to section Quercus, formed a strongly supported sister group (100% bootstrap support) in both trees, indicating their close evolutionary relationships and shared origin, consistent with previous morphological and molecular evidence (Denk et al., 2017; Hipp et al., 2020). Despite the overall consistency, slight discrepancies were observed in the phylogenetic positions of several species between the two organellar trees, particularly in Q. acutissima, Q. chenii, and Q. variabilis. These differences may reflect distinct evolutionary pressures acting on the mitochondrial and chloroplast genomes. In addition, factors such as introgression, historical hybridization, and incomplete lineage sorting are also considered major contributors to the observed phylogenetic incongruence between organelle genomes (Denk et al., 2017; Manos et al., 1999).

Taken together, these findings underscore the importance of integrating multiple organellar genomes to uncover subtle phylogenetic signals and lineage relationships within Querus. However, future studies incorporating nuclear genomic data will be essential to achieve a more comprehensive and robust understanding of Quercus evolution. Moreover, expanding the taxon sampling to include a broader range of Quercus species, especially those from underrepresented sections, will further refine phylogenetic resolution and enhance our understanding of diversification patterns within this ecologically and economically important genus.