Giant mitogenomes in Rhynchospora are a result of nuclear gene and retrotransposon insertions in intergenic spaces

Abstract

Background and Aims

The mitochondrial genomes of plants are large, with the majority ranging between 500 and 800 kb. However, the mitochondrial genomes of Cyperaceae (sedges) species were found to be much larger, exceeding 1 Mb in size. Here we aimed to investigate the gigantism of the mitochondrial genomes of three Rhynchospora (beak sedges) species and one related species of the sister family Juncaceae, the common rush (Juncus effusus).

Methods

Long PacBio HiFi reads were sequenced and assembled using hifiasm software. The mitochondrial genomes were annotated using Geneious and Mitofy software. Transposable elements were annotated using DANTE and RepeatModeler pipelines, and gene prediction in intergenic regions was conducted using Augustus. The predicted genes were annotated using BLAST and gene ontology terms.

Key Results

The mitogenome of R. breviuscula was 2 222 920 bp, that of R. pubera was 2 064 773 bp, that of R. tenuis was 1 678 054 bp and that of J. effusus was 553 985 bp. The results revealed giant intergenic spaces in all three Rhynchospora species, containing predicted nuclear genes and LTR retrotransposons. BLASTn revealed a high migration of DNA from the nucleus to the mitogenome.

Conclusions

Our findings show that the Rhynchospora mitogenome is the largest among the monocotyledons. These mitogenomes feature giant intergenic spaces, incorporation of chloroplast DNA and numerous rearrangements. Gigantism of the intergenic spaces is associated with the movement of nuclear DNA segments, suggesting a mechanism of DNA transfer from the nuclear genome to the mitochondrial genome.

INTRODUCTION

Plant mitogenomes exhibit significant variation in genome size and structure, with numerous rearrangements and a notable increase in the size of intergenic regions (Gualberto et al., 2014). The study of plant mitochondrial genomes (mitogenomes) has increased with the development of third-generation sequencing, allowing the generation of long DNA sequences and thus obtaining complete genomes. However, the number of sequenced mitogenomes remains small. These genomes mostly exhibit a circular chromosome structure; however, what intrigues researchers about plant mitogenomes is the considerable variation in structure, intergenic sequences and horizontal transfer of foreign sequences. Recent studies have also shown the prediction of transposable elements (TEs) (Souza et al., 2024), as well as the transfer of segments from chloroplasts to plant mitogenomes (Martins et al., 2019; Lee et al., 2023). The transfer of DNA segments from mitochondria to the nucleus has been observed in plants (Zhang et al., 2020, 2023a, b); however, the insertion patterns in plants vary from species to species. When the mitochondrial genome is large, a proportion of large segments are observed in the nuclear genome (Ko and Kim, 2016).

The largest sequenced mitogenome is from Silene conica, with a size exceeding 11 Mb and containing 99 % non-coding regions (Sloan et al., 2012). Recently, the nuclear genome of S. conica was sequenced (Fields et al., 2023), revealing regions with significant similarity between the nuclear genome and the cytoplasmic genome. In the mitogenome of Malus domestica (396 947 bp), it was estimated that 71.5 % of the sequences are similar to nuclear DNA (Goremykin et al., 2012). Recently, the mitogenomes of Syagrus coronata (Melo et al., 2024) and Hancornia speciosa (Souza et al., 2024) revealed that the intergenic spaces contain pseudogenes and TEs, suggesting the incorporation of nuclear DNA.

The Cyperaceae family is characterized by having holocentric chromosomes (Hofstatter et al., 2022). It is the second-largest family of monocotyledons, is monophyletic, is a sister to the Juncaceae family, and includes species of economic importance (Larridon et al., 2021). Among the Cyperaceae species, three species of the genus Rhynchospora have complete nuclear genomes: R. breviuscula (415 Mb), R. pubera (1.61 Gb) and R. tenuis (394 Mb) (Hofstatter et al., 2022). In Cyperaceae, the mitochondrial genome of Cyperus esculentus, which is characterized by being large, >1 Mb in size, contains many repeats and sequences derived from chloroplasts (Niu et al., 2022), and the mitogenome of Carex breviculmis is >1.4 Mb in length (Xu et al., 2023).

The increase in plant mitogenome size is still a mystery, and the mechanisms involved in the transfer of foreign DNA into mitogenomes are not known. Therefore, in the present study the mitogenomes and chloroplast genomes of the species R. breviuscula, R. pubera and R. tenuis and the outgroup Juncus effusus were assembled to analyse the vertical transfer of DNA segments from the nuclear and chloroplast genomes to the mitogenomes. The results obtained in the present study provide insights into plant evolution, as the mitogenomes of closely related individuals share certain characteristics, suggesting that mitogenome evolution plays an important role in the speciation process.

MATERIALS AND METHODS

Mitogenome assembly and annotation

The PacBio HiFi reads were developed by Hofstatter et al. (2022) and obtained from the National Center for Biotechnology Information (NCBI) (see codes in Supplementary Data Table S1). Hifi reads were subjected to assembly using hifiasm (Cheng et al., 2021), available at https://github.com/chhylp123/hifiasm, with the following command: hifiasm -o output.asm -t 40 reads.fq.gz. After assembly, the contigs of the mitogenomes were identified using core genes from Cyperus esculentus mitochondria, and the contigs were manually merged using Geneious software, by mapping the reads at their ends and joining contigs with continuous read mapping. The chloroplast genome was identified through mapping using C. esculentus as a reference. Additionally, assemblies were performed using the Flye (Kolmogorov et al., 2020) and PMAT (Bi et al., 2024) pipelines to aid in the merging of contigs obtained with hifiasm (Supplementary Data Fig. S1), allowing the reconstruction of a chromosome that better represents the complete mitogenome. The mitogenomes were mapped with the HiFi reads using BWA (bwa bwasw $genome.fasta reads.fastq > coverage.sam), and coverage was obtained using BamToCov (Birolo and Telatin, 2022) to verify the chromosome’s continuity (Supplementary Data Fig. S2). Gene annotation was performed through homology using the Geneious annotation tool (https://www.geneious.com), with C. esculentus as a reference and a minimum identity cut-off of 70 % between the genomes, and it was validated with the annotations obtained using Motify (Alverson et al., 2010). All annotations were manually reviewed and corrected to include start and stop codons. A graphical representation was generated using Organellar Genome Draw (Lohse et al., 2013).

Mitogenome analysis

Repetitive regions within the mitogenomes were identified using the Repeat Finder tool in Geneious software, REPuter (Kurtz, 2001), BLASTn, Asgart (Delehelle et al., 2018) and Biser (Numanagić et al., 2018). The comparison between mitogenomes was assessed using the AliTV software (Ankenbrand et al., 2017), where the genomes are aligned using lastz and then visualized using a graph approach. Microsatellite analysis was conducted using GMATA software (Wang and Wang, 2016) by searching for di-, tri-, tetra-, penta-, hexa- and heptanucleotide motifs, with a minimum of five repetitions. The repeat sequences were analysed using REPuter software (https://bibiserv.cebitec.uni-bielefeld.de/reputer/) with the following parameters: the repeat sequence was at least 20 bp in length, and the repeat identity was >90 % (Kurtz, 2001). The tandem repeat elements were investigated using Tandem Repeats Finder (TRF) v4.07 (Benson, 1999) with the parameter ‘2 5 7 80 10 50 2000578010502000 -h -d’. The results were analysed using circlize (Gu et al., 2014) or Circos software (Krzywinski et al., 2009).

Chloroplast phylogenetic analysis

The chloroplast genome sequences were aligned using MAFFT v7.017 (Katoh and Standley, 2013), implemented as the Multiple align tool in Geneious R9. The evolutionary history was inferred using the maximum likelihood method, and branch support was assessed with 1000 bootstrap replicates conducted in IQ-TREE2 (Minh et al., 2020), using Juncus effusus as an outgroup.

Insertion of DNA into the mitogenome

Genetic identity analysis between the mitogenomes and the chloroplast and nuclear genomes was performed using BLASTn. The nuclear genomes of the species were obtained from the NCBI in FASTA format and used for similarity analysis via BLASTn, with the commands ‘makeblastdb -dbtype ‘nucl’ -in mitogenome.fa -out mitogenome’ and ‘blastn -db mitogenome -query $genome -outfmt 6’. The results were represented in a circular plot using Circos software, and statistics were obtained using R software.

The regions of the nuclear and chloroplast genomes that share genetic identity with the mitogenome the three Rhynchospora species were analysed to assess overlap. For this analysis, gene annotations of the nuclear genome, developed by Hofstatter et al. (2022), were obtained from NCBI in GFF format. Chloroplast genome annotations were generated in the present study. For repetitive DNA, the genomes were annotated using the DANTE tool available from the RepeatExplorer2 Galaxy portal (repeatexplorer-elixir.cerit-sc.cz/galaxy). The intersections were identified using the Intervene pipeline (Khan and Mathelier, 2017).

Analysis of intergenic spaces in the mitogenome

Intergenic spaces were analysed through ab initio gene prediction using Augustus software (Stanke et al., 2008) implemented on the Galaxy platform (http://usegalaxy.eu). Functional annotation of the predicted genes was performed using homology (BLASTx) and ontology genes through InterProScan (https://www.ebi.ac.uk/interpro/search/sequence/) (Blum et al., 2021). BLASTx and ontology analyses were performed using Blast2GO (Conesa et al., 2005), using the non-redundant protein sequence. The annotation of TEs was carried out using Domain-based Annotation of Transposable Elements (DANTE) (https://repeatexplorer-elixir.cerit-sc.cz/galaxy/) and RepeatModeler2 (Flynn et al., 2020). The results were analysed using circlize (Gu et al., 2014), Circos (Krzywinski et al., 2009) and Intervene (Khan and Mathelier, 2017) software.

The distribution of protein-coding regions, predicted genes, TE annotation, and BLAST results were depicted in a graph generated with the R circlize package (Gu et al., 2014). The intersections among protein-coding regions, predicted genes and TE annotations were analysed using Intervene software (Khan and Mathelier, 2017). The reads from RNAseq were obtained from NCBI (see codes in Supplementary Data Table S1) and were mapped to the mitogenome to identify the expressed regions. The mapped regions were compared with other features using Intervene software. To analyse the hypothesis of extrachromosomal circular DNA (eccDNA) involvement in the expansion of intergenic spaces in mitogenomes, an eccDNA prediction was performed on the nuclear genome of R. breviuscula using the CReSIL pipeline (Wanchai et al., 2022). The eccDNA was then annotated with DANTE.

RESULTS

Mitogenome assembly and annotation

The de novo assembly of the mitogenomes resulted in several contigs. However, by manually merging the contigs into a single contig, it was possible to obtain a ‘master chromosome’. This approach of generating a single chromosome facilitates the presentation of results and the understanding of genome rearrangements. However, the mitogenome exhibits other configurations, such as small circular chromosomes and even small linear chromosomes (Supplementary Data Fig. S1). The construction of a master chromosome representing the mitochondrial genome was verified using the alignment of PacBio reads, showing continuous coverage, with low-coverage peaks for R. breviuscula and R. pubera due to repetitive DNA throughout the mitogenome (Supplementary Data Fig. S2).

The mitogenomes were 2 222 920 bp for R. breviuscula, 2 064 773 bp for R. pubera and 1 678 054 for R. tenuis. For the outgroup J. effusus, the mitogenome was 553 985 bp (Fig. 1 and Supplementary Data Fig. S3). Mitogenome annotation revealed a GC content ranging from 40.79 % to 42.42 % in Cyperaceae and 44.58 % in J. effusus. Annotation of gene content revealed variations in 38–40 protein-coding genes, 3–9 ribosomal genes and 23–32 transfer RNA genes (Table 1). For the chloroplast genomes, sequences were obtained and annotated for R. breviuscula, R. pubera and R. tenuis, which presented a typical quadripartite structure ranging in size from 178 889 bp in R. tenuis to 183 299 bp in R. pubera (Supplementary Data Fig. S4).

Fig. 1.

Characteristics of the Rhynchospora and Juncus effusus mitogenomes. The first circle contains annotated genes (blue), the second circle contains tandem repetitive sequences (dark blue), the third circle contains microsatellite sequences (green), the fourth circle contains AT/GC content (black and red), and at the centre are links of repetitive DNA obtained by Repeat Finder.

Table 1.

Characteristics of mitogenomes of Cyperaceae and Juncaceae.

Characteristic	C. esculentus	R. tenuis	R. pubera	R. breviuscula	J. effusus
Genome size (bp)	1 002 696	1 678 054	2 064 773	2 222 920	553 985
GC content (%)	40.79	42.16	41.69	42.42	44.58
Genes (bp)	48 988	43 751	32 723	40 254	44 152
CDS (n)	–	40	40	37	38
rRNA (n)	–	6	3	9	6
tRNA (n)	–	23	25	32	24
Plastid-derived	10 290	21 936	52 957	69 747	6933
Nuclear-derived	–	271 011	1 365 644	1 860 865	—
REPuter	28 962	116 692	98 912	182 686	97 368
Repeat Finder	57 413	210 527	149 827	318 943	86 344
Blast (Intragenome)	157 922	854 030	576 388	1 422 936	208 750
SSR (>2, <7)	1124	858	2388	935	241
Tandem Repeat Finder (>8, <2000) (bp)	6774	12 176	12 925	15 246	2959

CDS, coding sequences; SSR, simple sequence repeats.

The distribution of dispersed repetitive DNA was obtained using Repeat Finder (red links in Fig. 1) and BLASTn, and the results showed that the J. effusus mitogenome had fewer regions of dispersed repetitive DNA, while the Cyperaceae mitogenomes presented a significant amount of repetitive DNA dispersed throughout the genome. The self BLASTn analysis in R. breviuscula revealed a total of 1 422 936 bp of repetitive sequences in R. breviuscula, 1 365 644 bp in R. pubera and 854 030 bp in R. tenuis (Table 1). For tandem repetitive DNA, the analyses revealed a few sequences distributed throughout the genome (Fig. 1), with 15 246 bp in R. breviuscula and 2959 bp in J. effusus (Table 1). Microsatellite sequences contribute very little to genome expansion and were uniformly distributed throughout the genome. Rhynchospora pubera exhibited the highest amount of microsatellite sequences (2388 bp), while J. effusus presented 241 bp. For segmental duplications obtained using Asgart, Biser and REPuter, fewer repetitive segments were revealed, ranging from 28 962 bp in C. esculentus to 182 686 bp in R. breviuscula using REPuter (Table 1). These duplicated segments are distributed in both small and large fragments (Supplementary Data Fig. S5).

Incorporation of repetitive DNA

The integration of chloroplast DNA into the mitogenomes was detected in all four genomes, ranging from 6933 bp in J. effusus to 69 747 bp in R. breviuscula (Table 1 and Supplementary Data Fig. S6). Notably, the chloroplast genome contributes to the expansion of mitogenome size, with R. breviuscula exhibiting the highest chloroplast DNA insertions, including both gene and intergenic regions (Supplementary Data Fig. S6). However, the primary factor driving mitogenome expansion was the incorporation of nuclear DNA, as BLASTn analysis revealed a substantial number of nuclear-derived sequences integrated into the mitogenome, with 1 860 865 bp in R. breviuscula, 1 365 644 bp in R. pubera and 271 011 bp in R. tenuis (Table 1), distributed throughout the genome (Fig. 2 and Supplementary Data Fig. S7). We found that R. breviuscula and R. pubera have numerous insertions of nuclear DNA, which are distributed across all five nuclear chromosomes of each species. Within each species, these insertions were spread across the chromosomes (Supplementary Data Fig. S7). In R. tenuis, which has only two nuclear chromosomes, these insertions were less frequent (Table 1 and Fig. 3). The regions integrated into the mitogenomes overlapped with genes and TEs, as shown in the Venn diagram (Fig. 2). Interestingly, the nuclear genome regions that share similarity with the mitochondrial genome overlap with TEs and annotated genes in the nuclear genome of Rhynchospora species, as shown in the Venn diagram (Fig. 2).

Fig. 2.

The first row shows BLASTn analysis between the mitogenomes and the respective nuclear and chloroplast genomes of the species, with curved ribbons connecting pairs of syntenic blocks. The second row shows a Venn diagram depicting the intersections of nuclear gene annotations, nuclear TEs and BLASTn between the mitogenomes and the respective nuclear genome.

Fig. 3.

Analysis of intergenic regions in Rhynchospora mitogenomes. In the first row, the first circle contains annotated genes, the second circle contains genes predicted using ab initio methods, the third circle represents annotations using DANTE, the fourth circle contains regions mapped with RNAseq reads, the fifth circle represents annotations with LTR/Copia, the sixth circle annotations with LTR/Gypsy, and the seventh circle shows BLAST for predicted genes. The second row shows intersection between genes predicted using Augustus, annotations for protein-coding sequences (CDS), and mapping of reads obtained by RNAseq.

Intergenic region analysis

The mitogenome expansion primarily occurred in intergenic regions, measuring 409 951 bp in the J. effusus mitogenome. In Rhynchospora, these regions were even larger, ranging from 1 634 303 bp in R. tenuis to 2 182 666 bp in R. breviuscula (Table 2).

Table 2.

Genomic characteristics of intergenic regions.

Characteristic	R. breviuscula	R. pubera	R. tenuis	J. effusus
GC content	42.35	41.65	42.09	44.55
Minimum length (bp)	5	17	48	27
Maximum length (bp)	162 859	128 275	80 961	31 838
Number	87	76	81	79
Length (bp)	2 182 666	2 032 050	1 634 303	409 951
Ab initio prediction	817	664	551	259
Ab initio length (bp)	537 294	425 730	339 432	151 755
BLASTx	341	274	205	88
DANTE annotation	295	291	234	50
Orthology annotation	158	107	80	36

To determine the origin of these sequences, we performed gene prediction using Augustus, revealing the presence of genes, with 259 genes predicted in J. effusus, 551 genes in R. tenuis, 664 genes in R. pubera and 817 genes in R. breviuscula, distributed throughout the mitogenome, without forming blocks (Fig. 3 and Supplementary Data Fig. S8). These predicted genes were analysed by BLASTx using non-redundant protein sequences, and the results showed that many genes had homology with protein sequences (Table 2 and Fig. 3). In J. effusus, out of 259 predicted genes, 88 showed homology with proteins, while in R. breviuscula 341 out of 817 predicted genes showed homology with proteins (Table 2). Gene Ontology annotation revealed that the predicted genes were distributed across many functions, including biological processes, molecular functions and cellular components (Supplementary Data Fig. S9). The RNAseq analysis showed that many predicted genes are expressed (see the green circle in Fig. 3); however, it is not possible to affirm that they are expressed from the mitogenome. Most likely, the transcripts originate from the nuclear genome, indicating that these are active genes. In R. breviuscula 131 predicted genes were expressed, in R. pubera 56 were expressed, and in R. tenuis transcripts of 195 genes were detected (Fig. 3).

The intergenic regions were annotated with DANTE to analyse TEs, and the results showed that in R. breviuscula the most abundant LTR-Copia was Ale, with 70 annotations, and the most abundant LTR-Gypsy was chromovirus Reina, with 56 annotations. This pattern was also observed in R. pubera, where the most abundant LTR-Copia was Ale (65 regions) and the LTR-Gypsy chromovirus Reina had 54 annotations. In R. tenuis, the most abundant was also LTR-Copia Ale with 39 regions, followed by LTR-Gypsy chromovirus Reina with 43 annotations (Supplementary Data Table S2). Interestingly, the predicted genes were annotated with TEs, as in R. breviuscula, where 196 predicted genes were also annotated with TEs (Supplementary Data Fig. S8), which was an expected annotation since ab initio prediction also detects TE genes.

Comparison of mitogenomes in Cyperaceae

The comparison of the mitogenomes of Cyperaceae and J. effusus revealed that the Rhynchospora mitogenomes shared the most sequences, followed by those of C. esculentus and J. effusus (see links in Fig. 4). This greater sharing is related to the phylogeny obtained with the chloroplast genome (inset in Fig. 4), where the species R. breviuscula and R. pubera are more closely related and exhibit greater sequence sharing. Additionally, there is a notable variation in the structure of Cyperaceae mitogenomes, where large colinear fragments are not observed, reflecting a lack of synteny among the mitogenomes.

Fig. 4.

Dynamic rearrangements among mitogenomes of Cyperaceae and Juncus effusus, with curved ribbons connecting pairs of syntenic blocks and width proportional to block size. (Inset) Phylogenetic tree using the chloroplast genome.

DISCUSSION

Mitogenomes of Cyperaceae

Cyperaceae is the second largest family in the monocot Poales order; however, only a few mitochondrial genomes are available. The mitochondrial genomes of Cyperus esculentus (Niu et al., 2022) and Carex breviculmis (Xu et al., 2023) have been recently published. These mitogenomes are large (1 002 696 bp in C. esculentus and 1 414 795 bp in C. breviculmis), exhibiting extensive intergenic spaces and numerous rearrangements, characteristics typical of Cyperaceae mitogenomes. This characteristic of giant mitogenomes in Cyperaceae was confirmed in the present study, where the mitogenomes of R. breviuscula, R. pubera and R. tenuis exceeded 1.5 Mbp, being the largest reported for monocots.

The mitogenomes obtained in this study exhibit the typical characteristics of mitochondrial genomes in plants, such as the presence of dispersed and tandem repetitive DNA, numerous rearrangements, the presence of chloroplast DNA, microsatellites and similar gene content among the three analysed mitogenomes. However, these characteristics are not sufficient to explain the gigantism of intergenic spaces, with intergenic spaces being the cause of large mitogenomes. These events of mitochondrial genome gigantism are directly related to the speciation process, as evidenced by the high sequence sharing among the three Rhynchospora mitogenomes, suggesting that the expansion of intergenic spaces occurred before the differentiation of the genus.

Gigantism of intergenic spaces

The enlargement of mitochondrial genomes remains a mystery, and some studies have shown the incorporation of chloroplast DNA (Melo et al., 2024; Souza et al., 2024), while numerous rearrangements partially explain the size of plant mitogenomes. Plant cells contain three types of genome-containing compartment – the nucleus, chloroplast and mitochondria – with communication between these three compartments. The phenomenon of DNA transfer from mitochondria to the nucleus of eukaryotic genomes has been documented, including in plants (Ko and Kim, 2016; Zhang et al., 2020, 2023a, b); however, the transfer mechanism is still unclear. A few mechanisms have been proposed: (1) degradation of abnormal mitochondria, mediated by the action of a yme gene, (2) lysis of the mitochondrial compartment or mitophagy, (3) encapsulation of mitochondrial DNA within the nucleus, (4) direct physical contact between the mitochondria and the nucleus followed by membrane fusions, and (5) mitochondrial DNA that enters the nucleus being integrated into the chromosome during the repair of double-strand breaks in a mechanism known as non-homologous end joining (NHEJ) (Hazkani-Covo et al, 2010).

In the present study we investigated these nuclear DNA insertions and observed that the gigantism of Rhynchospora mitogenomes is due to the transfer of nuclear DNA to the mitogenomes (mitochondrial integrants of nuclear DNA [MINTs]). The frequency of organelle-to-nucleus DNA transfer has been reported, although with a lower intensity than that of nuclear-to-organelle DNA transfer (Zhang et al., 2020). However, in Rhynchospora the frequency of nuclear DNA segment transfer to mitochondria is much greater. Interestingly, these fragments present in mitochondria are associated with TEs, suggesting that the transfer mechanism involves TEs. This hypothesis becomes better understood through RNAseq analysis, where many of the predicted and annotated genes in the intergenic spaces of Rhynchospora mitogenomes are expressed, suggesting that these genes are expressed in the nuclear genome. Gene Ontology annotation revealed that these predicted genes are associated with biological processes, cellular components and molecular functions, suggesting that they originated from the nuclear genome.

DNA transfer between nuclear and cytoplasmic genomes has been suggested in various plants, including the eudicotyledons Syagrus coronata (Souza et al., 2024), Hancornia speciosa (Melo et al., 2024), and Malus domestica (Goremykin et al., 2012). The largest known plant mitogenome was observed in Silene conica (Sloan et al., 2012), and with the recent availability of its nuclear genome a high abundance of mitochondrial DNA in the nuclear genome was detected, suggesting the movement of DNA from the nucleus to the mitogenome (Fields et al., 2023).

The escape of DNA from the mitochondria to the nucleus in the form of a plasmid has been documented in Saccharomyces cerevisiae (Thorsness and Fox, 1990). A similar mechanism could exist in the reverse direction, from the nucleus to the mitochondria. The fact is that the evolution of the mitochondrial genome is still poorly understood. It is known that the loss or transfer of genes from mitochondria to the nucleus via endosymbiont gene transfer has been observed and is a relatively rare event. There are obstacles to the functional transfer of genes between mitochondria and the nucleus, making it extremely rare in the direction from the nucleus to the mitochondria (Butenko et al., 2024).

Currently, there is no information providing insights into the mechanism of DNA segment exchange from the nucleus to the mitochondria. However, in Rhynchospora species this hypothetical mechanism appears to be very intense. Insights from our work suggest that TEs play an active role in the transfer mechanism, particularly the LTR-Copia Ale and LTR-Gypsy Reina elements, as they are abundant in the three Rhynchospora species. One hypothesis is that eccDNA forms in the nucleus and then, following a direct physical association between the mitochondria and the nucleus, along with membrane fusion, the eccDNA is inserted into the mitochondria and can be integrated into the mitochondrial chromosome.

In plants, eccDNA has been reported as an important form of DNA mobility in the cell nucleus, remaining an enigmatic genomic component. The formation of eccDNA involves direct repeats, inverted repeats and mobile elements, and includes the breakage–fusion–bridge cycle (Zhang et al., 2023a, b). In rice, eccDNAs are distributed unevenly along the chromosomes, and the number is proportional to the chromosome size (Zhuang et al., 2024). In Amaranthus palmeri, eccDNA is involved in the transmission of herbicide resistance and is observed near the chromosomes by fluorescence in situ hybridization (Koo et al., 2018). We performed an estimation of eccDNAs in R. breviuscula and detected a large number of eccDNAs (224) across the species’ five chromosomes. These eccDNAs were annotated, and TEs were found (Supplementary Data Table S2), suggesting a potential pathway for understanding the evolution of plant mitogenomes.

We conclude that the Rhynchospora mitogenomes are large, the largest in monocots, with gene contents similar to those of other plant mitogenomes, featuring the incorporation of chloroplast DNA and characterized by numerous rearrangements. The gigantism of the genome is associated with the enlargement of intergenic spaces, with these intergenic spaces containing segments of nuclear DNA, suggesting a mechanism of DNA transfer from the nuclear genome to the mitochondrial genome.

FUNDING

This work was supported by the Max Planck Society (to A.M.) and the National Council for Scientific and Technological Development (CNPq).

SUPPLEMENTARY DATA

Supplementary data are available at Annals of Botany online and consist of the following. Figure S1. Brached conformation of the mitogenomes obtained using PMAT, hifiasm and Flye. Figure S2. The validation of the mitogenomes assembles integrity relied on the coverage analysis of mapping reads from pacbi hifi. The first circle illustrates the coverage of the reads and the center of the circle, a histogram of the coverage. Figure S3. Graphic representation of the mitogenomes of Cyperaceae and Juncus effusus. Figure S4. Graphic representation of the plastomes of Cyperaceae. Figure S5. The first row is the representation of chloroplast DNA incorporation, the second to fourth rows represente the genome duplications using Asgart, Biser and REPUter. Figure S6. Representation of chloroplast DNA incorporation and intersection between plastid annotation and Blast of the plastid-derived sequences. Figure S7. Representation of gene annotation, LTR and Blastn (to mitogenome) in the nuclear genmes of Rhynchospora. Figure S8. Annotation of gene and intergenic regions. The first circle corresponds to annotated genes (names are on the outer), the second circle contains genes predicted using ab initio methods, the third circle represents annotations using DANTE, and the fourth circle represents AT/GC content (black and red). In the center of the circle, there is a Venn diagram depicting the intersections of gene annotations, DANTE annotations, and genes predicted by ab initio methods. Figure S9. Gene Ontology annotations for the predicted genes in the intergenic spacers of the mitochondrial genomes of Cyperaceae and Juncus effusus. Table S1. NCBI codes used in the study. Table S2. Distribution of the domain bassed annotation of transposable elements in R. breviuscula.

AUTHOR CONTRIBUTIONS

C.A. designed the research, performed the experiments and wrote the first draft. A.M. provided resources and contributed to writing the manuscript.

DATA AVAILABILITY

The data generated in this study have been deposited in the public database of the National Center of Biotechnology Information, and codes are available in Supplementary Data Table S1.