Introduction

Plant cytoplasmic male sterility (CMS) has a primary characteristic: it prevents the production of viable pollen. This issue is visible under the microscope due to abnormal microspore development. Cotton CMS line failures occur at different times from the spore-making cell period to the tetrad period. However, the molecular mechanisms of CMS are complex and involve the rearrangement of the mitochondrial genome and the expression of chimeric genes. Reactive oxygen species (ROS) are important signaling molecules involved in many biochemical processes during normal plant growth and development. ROS play a very important role in plant growth and development, stress and response processes, seed germination, programmed cell death (PCD) and other physiological processes1. The sources of ROS in plants include chloroplast bodies, mitochondria, glyoxylate cyclers, cytoplasm, nucleus, peroxisomes, and membrane systems2. Chloroplasts are the main source of ROS during photosynthesis in plants3,4,5, whereas mitochondria are the main source of ROS in processes other than photosynthesis6. The toxic effects of ROS are mainly manifested in oxidative damage to macromolecules and even cell death. Excessive ROS accumulation can seriously affect the stability of mitochondrial DNA (mtDNA). Previous studies have shown that the stability of mtDNA is associated with many nuclear genes, such as MSH1, RECA3, OSB1 and other genes7,8,9, which regulates the stable expression of mtDNA. If these genes are abnormal, mtDNA will be uncontrolled, making rearrangement occur continuously, and CMS will be produced in this process10. Shao et al.11 studied an Arabidopsis msh1 mutant and found that when the expression of this gene was changed, the level of ROS was reversed, the stability of mtDNA was decreased, and the recombination rate was increased. This is because mtDNA is directly exposed to ROS generated by the oxidative phosphorylation process, which is an important target molecule for free radical attack. High levels of ROS can lead to mtDNA breakage, destroying its structure and function12. Additionally, ROS can induce point mutations, deletions and insertions in mtDNA13,14. Changes in ROS levels can lead to male sterility by preventing PCD from occurring over time in the downy mildew layer15,16. It has been shown that the accumulation of ROS in CMS plants is associated with the phenomenon of pollen abortion17,18,19. Therefore, exploring the interaction between CMS and ROS is important for understanding the molecular mechanism of CMS and developing new crop breeding strategies. Active oxygen elimination system mainly consists of two important systems: (1) enzyme preservation system composed of various antioxidant enzymes including catalase (CAT), superoxide dismutase (SOD), peroxide dismutase (POD) and ascorbate peroxidase (APX); (2) Non-enzymatic systems consisting of multivitamins such as ascorbic acid and small molecular compounds such as glutathione20. The abnormal function of reactive oxygen species elimination system in plants can lead to excessive accumulation of reactive oxygen species and eventually lead to the outbreak of reactive oxygen species. This phenomenon has been verified in male sterile lines of many plants. As CMS is a type of male sterility that is jointly controlled by cytoplasmic sterile genes and corresponding nuclear genes. When the sterile gene S exists in the cytoplasm and the recessive sterile genes rr are present in the nucleus (genotype S(rr)), the plant exhibits male sterility. For example, in rice, the sterile gene WA352 of CMS-WA type is located in the mitochondrial genome, and it interacts with the fertility restoration genes Rf3 and Rf4 in the nucleus to determine the fertility of rice. In CMS, the proteins encoded by mitochondrial genes (such as WA352) interact with the mitochondrially encoded nuclear proteins (such as COX11), inhibiting their normal functions in the peroxide metabolism process and leading to the accumulation of reactive oxygen species (ROS). The accumulation of ROS affects the permeability of mitochondria, promotes the release of cytochrome c into the cytoplasm, and then leads to advancement of PCD in the felt layer, resulting in abnormal PCD of the tapetum and ultimately pollen abortion. According to the nucleo-cytoplasmic complementation control hypothesis (Genetics textbook for agricultural universities, 2018 edition21), the three-line hybrid system requires synergistic interactions between cytoplasmic and nuclear factors: Cytoplasmic male sterility (CMS) arises when the mitochondrial genome carries the S-type sterility-inducing cytoplasm interacting with homozygous recessive nuclear r alleles (genotype S(rr)). Rf genes, as dominant nuclear restorers, encode mitochondrial-targeted proteins (e.g., pentatricopeptide repeat (PPR) family members) that suppress S-cytoplasm-induced sterility. Mechanistically, these proteins rescue mitochondrial dysfunction caused by structural variations (rearrangements, chimeric ORFs) in S-type mtDNA through RNA editing or transcriptional regulation22,23. This explains why restorer lines (R-line, genotype N/S(RR)) restore fertility via Rf gene products, whereas maintainer lines (B-line, N(rr)) lacking functional Rf alleles cannot. For example, in rice, the fertility restoration gene Rf4 mediates the degradation of WA352 mRNA, reduces the production of WA352 protein, and further reduces the accumulation of ROS to restore the fertility of CMS-WA. Previous studies have also shown that: The proteins encoded by CMS genes generated by mitochondrial genome rearrangement can interfere with normal mitochondrial functions and ROS-mediated metabolic processes, resulting in abnormal pollen development. While the fertility restoration genes in the nucleus can regulate these processes to restore the fertility of plants24,25,26.

Research on CMS at the molecular level began in 1976, when Levings and Pring’s comparison of mtDNA zymography profiles of maize T-type cytoplasm and normal cytoplasm revealed significant differences27. The study of plant organelles by high-throughput sequencing dates back to the first acquisition of the genome of tobacco chloroplasts in 1986 and the mitochondrial genome of ground money in 199228. Studies have hypothesized that the main cause of CMS in plants may be mitochondrial structural or sequence alterations29. Therefore, sequence studies and analysis of plant mitochondrial genomes are necessary to understand the nucleoplasmic interactions and the mechanism of CMS. Arabidopsis thaliana was the first higher plant for which mtDNA was sequenced30. To date, the mitochondrial genomes of more than five hundred plants have been sequenced31, including maize32, rice33, tobacco34, rapeseed35, sugar beet36, kenaf37, carrot38, and others. With the accumulation of increasing mitochondrial sequencing data, studies in recent years have produced a growing body of evidence suggesting that CMS may be related to the rearrangement of the mitochondrial genome. Mitochondrial gene rearrangements often result in the formation of chimeras with neighboring sequences, which are cotranscribed with unmutated sequences in the form of chimeric genes, leading to CMS.

The rearrangement pattern of cotton mtDNA during the evolutionary process has been a topic of interest for scientists. Previous studies have shown that the inclusion of many repetitive sequences in mitochondria is closely related to CMS. It is hypothesized that this is due to the formation of open reading frames (ORFs) by repeated sequence-mediated gene rearrangements. ORFs generated by mitochondrial genome reorganization are usually located upstream or downstream of mitochondrial ATPase complex subunit genes, such as atp6, atp8, and atp9, and coexpressed with these genes. On the one hand, coexpression products interfere with normal mitochondrial function, such as toxic proteins competing with mitochondrial genes for substrates. On the other hand, cotranscription with neighboring mitochondrial protein-coding genes can affect RNA editing, leading to posttranslational protein dysfunction and microspore abortion39. There are also numerous cytochrome oxidase genes in the electron transport chain that undergo rearrangements to form sterility-associated genes, such as the rearrangement of the genes atp6 and orf2540, the orf256 gene upstream of cox1 in wheat41, and the atpA and atp6 loci in sugar beets42. Alterations in the DNA sequences of these genes may affect the activity of cytochrome oxidase, which in turn affects ATP synthesis and leads to the development of septoria. Notably, such genomic rearrangements may represent an evolutionary response to nuclear selection pressure in heteroplasmic environments, reflecting nuclear-mediated remodeling of mitochondrial genomes through epigenetic regulation (e.g., DNA methylation) or retrograde signaling. Emerging evidence highlights the bidirectional nature of nucleus-organelle interactions: Nuclear-encoded PPR proteins (e.g., rice Rf1/Rf2) target mitochondrial transcripts to precisely regulate CMS-associated chimeric genes (e.g., transcriptional repression of orf79), constituting the core molecular pathway for Rf-mediated fertility restoration23. Hu Jiazhi’s group from Peking University found that during gene editing, whether targeting nuclear DNA or mitochondrial DNA, mitochondrial DNA fragments can be transferred into the nuclear genome. This phenomenon was observed in various cell types, and the use of base editors (such as BE3, BE4max, and ABEmax) could significantly reduce this fusion. The study also pointed out that mitochondrial editing tools (such as mitoTALEN and DdCBE) can also cause mitochondrial DNA to fuse with nuclear DNA when editing mitochondrial DNA. To mitigate this risk, co-expression of exonucleases TREX1 or TREX2 was found to significantly reduce the transfer of mitochondrial DNA into the nuclear genome43.

Mitochondrial functional states feedback regulate nuclear fertility-related genes via ROS signals, metabolic intermediates (ATP/NAD + levels), or epigenetic modifications, ensuring dynamic equilibrium in fertility restoration44,45.

In this study, we compared and analyzed the effects of cytoplasmic-nuclear interactions between homologous and heterologous cytoplasmic-nuclear CMS lines on microspore abortion, reactive oxygen metabolism and mtDNA rearrangement using maintainer lines as controls. We also systematically investigated the different types of cytoplasmic effects and the relationships of cytoplasmic-nuclear interactions from multiple perspectives, which provide not only theoretical guidance for the creation of optimal sterile lines and cytoplasmic-nuclear interactions but also theoretical and technological support for the creation of CMS lines for the cotton industry as a whole.

Results

Morphological and cytological observations

Flowers are the organs of sexual reproduction in plants and play an important role in the reproduction of offspring; therefore, the study of plant fertility cannot be separated from the study of flowers. Comparison of the floral organ phenotypes of homologous and heterologous cytoplasmic-nuclear CMS lines and their maintainer lines revealed that although there were no significant differences in the morphology of the floral organs and leaves between the CMS lines, the styles of the heterologous CMS lines were relatively protruding, and the filaments and anthers were more shortened and shriveled (Fig. 1).

Fig. 1
figure 1

Morphological structure of floral organs. (a, b) Heterologous CMS lines (homonucleus but alloplasms) and their maintainers; (c) Heterologous CMS lines (homoplasm but heteronuclei) and their maintainers. Notes: Morphological structures were compared between CMS lines with different cytoplasmic-nuclear combinations; C2-113 A/C4-113 A: homonucleus but alloplasms lines (paternal parent 07-113 A); C2-113 A: homoplasm but heteronuclei lines relative to 276 A (maternal origin: 276B); no significant phenotypic divergence was observed in floral organs under these conditions.

Full size image

Comparative analysis of paraffin sections from homologous and heterologous cytoplasmic-nuclear CMS lines

Four critical periods of anther development were observed in the pollen sac cell structures of one homoplasm but heteronuclei cotton CMS line, two homonucleus but alloplasms cotton CMS lines, and maintainer lines (Figs. 2 and 3); the four periods were the pollen mother cell (PMC) stage, tetrad (Td) stage, early unicellular (early Uni) stage, and mature pollen grain (MP) stage.

Fig. 2
figure 2

Microspore development of homoplasm but heteronuclei CMS lineage of 276B (a1–a4), 276 A(b1–b4), C2-113 A(c1-c4). a1, b1,c1; a2, b2,c2; a3, b3, c3 and a4,b4,c4 correspond to microspores of pollen mother cell, tetrad, early uninucleate and mature pollen grains stages, respectively. PMC: Pollen mother cell stage, Td: Tetrad microspore stage; early Uni: early uninucleate stage; MP: Mature pollen grain stage. Notes: C2-113 A, homoplasm but heteronuclei relative to 276 A (maternal origin: 276B); Red arrows indicate the presence of tetrads and the commencement of abortion in the homologous CMS line 276 A. Conversely, blue arrows denote the absence of tetrads and near-complete abortion in the heterologous CMS line C2-113 A.

Full size image
Fig. 3
figure 3

Microspore development of homonucleus but alloplasms CMS lineage of 07-113B (a1–a4), 07–113 A(b1–b4), C2-113 A(c1-c4), C4-113 A(d1-d4) a1, b1,c1,d1; a2, b2,c2,d2; a3,b3,c3,d3 and a4,b4,c4,d4 correspond to microspores of pollen mother cell, tetrad, early uninucleate and mature pollen grains stages, respectively. PMC: Pollen mother cell stage, Td: Tetrad microspore stage; early Uni: early uninucleate stage; MP: Mature pollen grain stage. Notes: C2-113 A/C4-113 A, homonucleus but alloplasms lines (paternal parent 07–113 A); arrow: At Td, 07–113 A(homologous CMS) formed tetrads, while C2-113 A/C4-113 A(heterologous CMS lines) showed near-complete abortion and C2-113 A lacked tetrads.

Full size image

Comparative analysis of paraffin sections of homoplasm but heteronuclei sterile lines

Comparative analysis of anther cell structures during cotton development in homoplasm but heteronuclei CMS lines revealed that: (1) The pollen mother cells (PMCs) were observable in the center of the pollen sac during meiosis; (2) These PMCs were relatively large and had distinct nuclei; (3) There were no significant differences in the stages of PMC development (Fig. 2, a1, b1, c1). In the Td stage, the homologous cytoplasmic-nuclear CMS line 276 A had tetrad structures in the anthers, while the heterologous cytoplasmic-nuclear CMS line C2-113 A had no tetrad structures; both microspores were aborted, with 276 A starting to abort in the Td stage and C2-113 A almost completely aborted in the Td stage (Fig. 2, a2, b2, c2).

The cells of the chorioallantoic layer of the monokaryotic maintainer line 276B were highly vesicularized and began to degrade. The callus encasing the tetrads was degraded and released microspores, while a large central vesicle began to appear that gradually pushed the nucleus toward the cell wall. The microspores of the sterile lines were completely degraded. In the two sterile lines, the chorionic layer was not degraded, and the degraded PMC structure was further vesicularized (Fig. 2, a3, b3, c3).

In the MP stage, the pollen grains of the maintainer line 276B were mature, with conspicuous punctures and complete degradation of the felted cells, whereas in both CMS lines, the pollen sacs were completely crumpled, forming a darker solid structure (Fig. 2, a4, b4, c4).

Fig. 4
figure 4

Active oxygen metabolism analysis. (a) POD activities of anthers in 07-113B, 07–113 A, and C2-113 A; (b) POD activities of anthers in 07-113B, 07–113 A and C4-113 A; (c) POD activities of anthers in 276B, 276 A, and C2-113 A; (d) CAT activities of anthers in 07-113B, 07–113 A, and C2-113 A; (e) CAT activities of anthers in 07-113B, 07–113 A and C4-113 A; (f) CAT activities of anthers in 276B, 276 A, and C2-113 A; (g) MDA contents in anthers of 07-113B, 07–113 A and C2-113 A; (h) MDA contents in anthers of 07-113B, 07–113 A and C4-113 A; (i) MDA contents in anthers of 276B, 276 A and C2-113 A. Notes: PMC, Pollen mother cell; Td, Ttrad; MP, Mature Polle. Significant differences were assessed by Student’s t-test (*P < 0.05, **P < 0.01, ***P < 0.001).

Full size image

These results indicated that abortion of the heterologous cytoplasmic-nuclear CMS lines occurred earlier than that of the homologous CMS lines.

Comparative analysis of paraffin sections of homonucleus but alloplasms sterile lines

Through the comparative analysis of anther cell structure during the development of cotton CMS lines, it was found that in the PMC stage of both sterile and maintainer lines, the PMC was located in the center of the pollen sac, the chorion cells were larger, and the differences in the PMC stage of the different materials were not significant (Fig. 3, a1, b1, and c1).

In the Td stage, 07–113 A/B anthers had tetrad structures, while C2-113 A and C4-113 A anthers had no tetrad structures; the chorionic layer was not degraded; and the microspores were aborted, which indicated that the homologous cytoplasmic-nuclear CMS lines began to abort at the Td stage, and the heterologous cytoplasmic-nuclear CMS line C2-113 A was almost completely aborted in the Td stage (Fig. 3, a2, b2, c2).

The pollen sac cavity of the monokaryotic line 07-113B increased, the tetrad microspores encapsulated in the anthers separated into monokaryotic nuclei, and the chorionic layer was completely degraded; the sterile line still exhibited a tetrad structure, and the microspores had been completely degraded (Fig. 3, a3, b3, c3).

In the MP stage, the cells of the felid layer of the maintainer line 07-113B were almost completely degraded, with some cell fragments remaining, and the pollen grains were mature, with obvious punctures and pollen-dispersing anthers, whereas the pollen sacs of the two CMS lines were completely crumpled, forming a darker solid structure (Fig. 3, a4, b4, c4).

Comparative analysis of the anther cell structure during the development of homonuclear and heterologous CMS lines showed that the homologous cytoplasmic-nuclear CMS line 07–113 A was abortive from the Td stage to the early monokaryotic stage, whereas the heterologous cytoplasmic-nuclear CMS lines C2-113 A and C4-113 A were almost completely abortive at the Td stage. These results indicated that the abortion period of the heterologous cytoplasmic-nuclear CMS lines preceded the Td stage and was earlier than that of the homologous cytoplasmic-nuclear CMS lines.

Activity of enzymes involved in reactive oxygen scavenging pathways

All datasets met normality assumptions (p > 0.05, Fig. S1), supporting the use of parametric tests. For the homologous cytoplasmic-nuclear CMS line 07–113 A, POD activity was increased in the PMC stage and decreased in the Td stage and the MP stage when compared with the sterile line 07-113B. For the heterologous cytoplasmic-nuclear CMS lines C2-113 A and C4-113 A, POD activity was also higher in the PMC stage and lower in the Td stage and the MP stage compared with that of the maintainer line 07-113B (Fig. 4a, b). In contrast, the POD activity of the heterologous cytoplasmic-nuclear CMS line C4-113 A was consistently lower than that of the homologous cytoplasmic-nuclear CMS line 07–113 A at the Td stage and the MP stage (Fig. 4b), while the POD activity of C2-113 A showed a decreasing trend and was lowest at the MP stage (Fig. 4a). Similarly, comparative analysis of the homologous cytoplasmic-nuclear line 276 A and its maintainer line 276B revealed that the POD activities of 276 A relative to 276B showed a low-high-low trend, and the POD activities of C2-113 A were consistently lower than those of the maintainer line 276B during the period of pollen development (Fig. 4c), and the heterologous cytoplasmic-nuclear CMS lines (C2-113 A) had consistently lower POD activity than the corresponding homologous cytoplasmic-nuclear CMS line (276 A).

The CAT activity of the homologous cytoplasmic-nuclear CMS line 07–113 A was higher in the PMC stage and lower in the Td stage and the MP stage than that of the maintainer line 07-113B (Fig. 4d, e). For the heterologous cytoplasmic-nuclear CMS line C2-113 A, POD activity was also higher in the PMC stage and lower in the Td stage and the MP stage compared to the maintainer line 07-113B (Fig. 4d). The activity of C4-113 A was lower than that of the maintainer line 07-113B in all three periods(Fig. 4e). The CAT activity in anthers of the sterile line 276 A was lower than that of the maintainer line 276B at the PMC stage; the CAT activity increased at the Td stage, and the activity of 276 A was significantly higher than that of 276B at the MP stage, which may be due to the high level of ROS scavenger enzyme in 276 A, resulting in a low level of ROS and affecting pollen development (Fig. 4f). The CAT activity of C2-113 A was consistently lower than that of 276B during pollen development, reflecting the weaker ability of anthers in C2-113 A to eliminate peroxides than that of the fertile variety 276B as microspore development occurs; the CAT activity of the heterologous cytoplasmic-nuclear CMS lines was consistently lower than that of the homologous cytoplasmic-nuclear CMS lines (Fig. 4f). The lower enzyme activity of the heterologous CMS lines predicted a lower ability to eliminate ROS. It is worth noting that Panel f is very different from panel d and panel e, indicating that the content of CAT is significantly different in homonucleus but alloplasms and homoplasm but heteronuclei CMS lines. As for why there is such a big difference, it is worthy of follow-up attention.

The MDA content reflects the degree of lipid peroxidation in organisms and may indirectly reflect the degree of cellular damage. Compared with CMS line 07-113B, the MDA content of CMS line 07–113 A was higher during the PMC stage, but the MDA content of CMS line 07-113B was higher during the Td stage and lower during the maturation stage. At the onset of microspore abortion, the heterologous cytoplasmic-nuclear CMS lines (C2-113 A and C4-113 A) showed a low-low-high trend of MDA content compared with the maintainer line 07-113B, in which the MDA content always increased (Fig. 4g, h). The MDA content of the heterologous cytoplasmic-nuclear CMS lines was higher than that of the homologous cytoplasmic-nuclear CMS lines at the time of pollen grain maturation (Fig. 4g–i). The MDA content of 276 A was higher than that of its maintainer line 276B in all three periods and showed a positive trend. C2-113 A also followed this pattern compared with 276B, and the MDA content was higher in heterologous CMS lines than in homologous CMS lines (Fig. 4i). Although our results on POD activity are consistent with previous reports in CMS systems46, the CAT/MDA observed in heterologous CMS strains is very different from that observed in rice and soybean47,48. This discrepancy may stem from species-specific nuclear-cytoplasmic interactions or differences in enzyme assays, and further cross-species comparisons are necessary to clarify these mechanisms.

Overview of cotton MtDNA sequencing data

The libraries were constructed against mtDNA and used a combination of second- and third-generation sequencing approaches, data filtering and other processing to obtain the data information statistics of each sample (Table 1). The valid data and quality distribution obtained indicate that the sequencing results are reliable.

Table 1 Sequencing data statistics of MtDNA.
Full size table

The mtDNA was assembled with the assembly software. The mt genomes of C2-113 A (GenBank accession: OR906298), C4-113 A (GenBank accession: OR906297), 276 A (GenBank accession: OR906300) and 276B (GenBank accession: OR906299), 07–113 A (GenBank accession are not yet available), 07-113B (GenBank accession are not yet available) were each assembled into a single circular molecule, and the mt genome sizes of the CMS lines were smaller than the genome size of their maintainer line (Table 2). The mitochondrial genome circle diagrams of each cotton material were plotted with the Organellar GenomeDRAW software (Figs. 5, S2). By BLASTn comparison of the assembled sequences with known plant mitochondrial sequences in the NCBI public DNA database, a total of 40 genes were annotated in 07–113 A and 07-113B, including 36 core protein-coding genes common to all plant mitochondrial genomes. A total of 28 genes were directly related to the electron transport chain (ETC) and oxidative phosphorylation, and four genes were involved in cytochrome c maturation: ccmB, ccmC, ccmFC, and ccmFN, while the remaining four genes were involved in ribosome assembly, RNA maturation, or methyltransfer, and were rps, rpl, matR, and mttB, respectively. except for sdh3, sdh4, rps, and rpl, most protein-coding genes are highly conserved in the mitochondrial genome.

Table 2 Genetic composition of mitochondria of cotton material.
Full size table
Fig. 5
figure 5

Mitochondrial genome circle map of 07–113 A and 07-113B. Notes: The outer circle is the position coordinate of genomic components such as genes and ncRNA, with corresponding gene names; the inner circle is the content of genomic GC.

Full size image

By analyzing the noncoding RNAs (ncRNAs) of the six materials, it was found that the number of transfer RNAs (tRNAs) was higher in all the maintainer lines than in the sterile lines (Table 3), and there were differences in the number of ribosomal RNAs (rRNAs) and tRNAs between the homologous CMS lines and heterologous CMS lines without any pattern.

Table 3 Composition of NcRNAs in plastid-nucleus homologous and plastid-nucleus heterologous CMS lines.
Full size table

SV detection and analysis

SVs often affect important agronomic traits49. The relationship between Structural Variation (SV) and Cytoplasmic Male Sterility (CMS) is an important research topic in genetics and plant breeding. SVs are large sequence and positional changes in the genome, such as duplications, insertions, inversions, heterozygotes, and deletions, which are the result of a combination of endogenous and exogenous factors. And may affect gene expression and function, which in turn affects CMS performance. SVs have a greater impact on the genome than single-nucleotide polymorphisms (SNPs), and they can often cause birth defects, cancer, etc., in the organisms. Comparison of the SVs in the genomes of the four sterile lines showed that the genomes of the heterologous cytoplasmic-nuclear CMS lines are more complex than those of the homologous CMS lines (Table 4).

Table 4 SV types and mutations in homonucleus but alloplasms and homoplasm but heteronuclei CMS lines.
Full size table

Analysis of mitochondrial genome collinearity

By analyzing the collinearity between two genomes, we can identify insertions and deletions between the genome of the target species and the reference genome. We can also analyze the structural changes (e.g., chromosome rearrangement) of the genomes in the process of evolution. Using the maintainer line 07-113B as the reference, 07–113 A and C2-113 A were compared, and Fig. 6 shows that the collinearity between 07 and 113 A and 07-113B was better, the collinearity between C2-113 A and 07-113B was poorer, and more translocations occurred in C2-113 A (Fig. 6a).

Fig. 6
figure 6

Comparison of Parallel Collinearity Plots in Homologous and Heterologous CMS Lines of Cotton. (a) Comparison of Parallel Collinearity of homonucleus but alloplasms CMS lines in cotton; (b) Comparison of Parallel Collinearity of homoplasm but heteronuclei CMS lines in cotton. Notes: The upper axis indicates the measured genome, and the lower axis indicates the reference sequence genome. The yellow box in the upper and lower axes indicates the forward strand of the genome, and the blue box indicates the reverse strand of the genome. The colors of the graphs linked between the upper and lower axes indicate the type of comparison: Collinear: collinearity comparison; Translocation: translocation comparison; Inversion: inversion comparison; Tran + Inver: translocation and inversion comparison. The fuchsia module represents the result of collinearity analysis. The size of the purple-red part is positively correlated with the strength of the collinearity, i.e., the larger the purple-red part, the stronger the collinearity, indicating that homozygous has higher mitochondrial genome stability than heterozygous CMS lines.

Full size image

As shown in Fig. 6b, the two sterile lines were compared with 276B for collinearity, and the collinearity of 276 A with 276B was better than that of C2-113 A with 276B.

Discussion

Periods of microspore abortion in cotton CMS lines and their cytological characteristics

Due to the different sterility mechanisms of cotton CMS lines, there are great differences in the time of abortion and abortion characteristics among different sterile lines. In the land cotton CMS line J4A, the abortion period occurs from the microsporoblast stage to the Td stage (during meiosis), and abortion is characterized by the absence of degradation of the chorionic and mesophyll cells of the anthers throughout microspore development and the absence of tetrad structure formation50. Failure of Jin A cotton CMS lines begins during the proliferation of sporulating cells, which are unable to undergo normal mitosis and often contain multiple micronuclei, followed by cytoplasmic vesiculation of the microsporangial mother cell51. Comparative observation of cotton A and B resistance found that septicity began to occur at the same time as microspore mother cells, and microspore mother cells obviously degraded and gradually disintegrated52. In CMS-D8, septicity started at the microsporoblast stage, and the microspores were completely degraded at the Td stage with no tetrad formation. During the septic stage, the pomatum cells are abnormally vesicularized and extruded into the drug compartment53. The island cotton CMS line H276A was aborted at the Td stage, characterized by vesicularization of the nuclei of the tetrad cells, and the pomace layer was intact and did not undergo degradation during anther development31,51. The abortion period of the CMS “Anti-A” sterile lines in cotton occurs at the Td stage, and no tetrad structure is formed in the sterile lines54.

Hu et al.55 studied the mechanism of abortion in citrus and found that the pollen abortion of male sterile Ougan mutants occurred at the microspore Td stage. Analysis of paraffin sections revealed that the CMS lines underwent abortion at different times, and none of the sterile lines showed degradation of the chorionic layer during microspore development. Comparing the homologous and heterologous cytoplasmic-nuclear CMS lines, we found that the abortion period of homologous CMS lines was generally later than that of heterologous CMS lines. It was hypothesized that because the mitochondrial and nuclear genomes of homologous cytoplasmic-nuclear CMS lines and their maintainer lines were almost identical, this kind of sterile line was less affected by the nucleoplasmic interaction effect, which led to late abortion and made them the best combinations for the study of male sterility.

Production and clearance of ROS in anther development

In the present study, it was observed that the activities of ROS-scavenging enzymes (e.g., POD and CAT) in the sterile lines were markedly lower than those in the maintainer lines. This disparity led to a progressive attenuation of the anthers’ capacity to eliminate ROS within the sterile lines. Commencing from the PMC stage, the weakened scavenging ability induced peroxidation of the mitochondrial membrane, which was evidenced by a substantial elevation in MDA content, accompanied by the disruption of the mitochondrial membrane structure and the constriction of the plasma membrane space. Subsequently, the mitochondrial function was compromised, failing to convert energy in a normal manner, thereby resulting in the depletion of ATP and NADH and the obstruction of the ascorbic acid and glutathione cycles. Furthermore, throughout the entire process from sporogenesis to the maturation of pollen grains, the activities of ROS-scavenging enzymes in the heteroplasmic CMS lines were consistently lower than those in the homoplasmic CMS lines, indicating a relatively weaker superoxide scavenging ability in the anthers of the heteroplasmic CMS lines. The MDA level, as the terminal product of ROS-mediated metabolic processes, exhibited an inverse relationship with the enzyme activities among the three materials. Specifically, the MDA level was relatively lower in the maintainer lines, whereas in the heteroplasmic CMS lines, it was ultimately higher than that in the homoplasmic CMS lines. The excessive generation of ROS in the sterile lines, combined with the significant reduction in the activities of enzymes such as SOD, POD, and CAT, led to an augmented accumulation of ROS. Eventually, this accumulation induced cytotoxicity and damage to organelles, giving rise to abnormal cell function and ultimately resulting in microspore abortion. As cytoplasmic-nuclear hybrids, the CMS lines manifested lower anti-ROS enzyme activities and higher ROS accumulation levels, thus conferring enhanced toxic effects. These results were confirmed by phenotypic and cytological observations.

Role of the mitochondrial genome in male sterility in cotton

Due to the complex and variable structure of the mitochondrial genome in higher plants, the vast majority of the genetic information in most mitochondrial genomes reflects the differences in mitochondrial evolution, while only a small part of the genetic information is related to the target traits of our research. Therefore, when isolating, identifying, and cloning mitochondrial genes related to plant cytoplasmic male sterility, it is of great importance to maintain a relatively consistent mitochondrial gene background of the research materials. That is, the application of near-isogenic lines can effectively reduce the analysis of differences unrelated to CMS that have formed during the evolution of the mitochondrial genome. Although previous studies on cotton CMS have used materials with different cytoplasmic and nuclear genetic backgrounds, few studies on materials from homologous cytoplasmic-nuclear CMS lines and heterologous cytoplasmic-nuclear CMS lines have been published. The mitochondrion is a central site for a wide range of metabolic activities and is involved in a variety of biochemical processes56. Mitochondria are involved in ATP synthesis, which provides energy for normal physiological activities in plants and is also a major source of ROS production57. The mitochondrial genome evolves through genomic rearrangements as well as insertions, deletions and base mutations in the sequence. In this study, we found that several cotton sterile line SNPs and insertions - deletions (InDels) were distributed in the intergenic region. All InDels occurred in the intergenic region. No insertion or deletion of small fragments occurred in the CDS region, and the InDels were mostly small fragments (1–5 bp) (Figs. S3, S4). All SNP types in the mtDNA of the sterile lines were inverted, and the vast majority of them were A - C transitions. We found that SNP mutations in the mitochondrial genome of the four CMS lines occurred relatively more often in genes for ribosome-encoded proteins, such as rpl1, rpl2, rpl4, rpl5, rpl10, rpl16, and rps4 (Tables S1, S2). From the results of this study, it can be hypothesized that the content of H2O2 in the CMS strain was significantly higher than that in the maintainer strain, whereas CAT, POD and other related enzymes in the CMS strain were regulated by the related genes, which resulted in under- or overactivity and an imbalance of ROS. The heterologous cytoplasmic-nuclear CMS lines had lower anti-ROS enzyme activities and higher levels of ROS accumulation, which may be an important reason for the higher number of mutation sites and more SVs in the mitochondrial genome and poorer collinearity between the maintainer lines and the heterologous CMS lines.

As mentioned in the introduction, cytoplasmic-nuclear interactions affect mitochondrial function and ROS-mediated metabolic processes. On the one hand, genome rearrangements may impact the integrity of the mitochondrial electron transport chain (ETC), resulting in increased electron leakage and thus elevated ROS production. On the other hand, genome rearrangements can also influence the expression and activity of antioxidant enzymes in mitochondria, thereby affecting the cell’s ability to scavenge ROS. Inducing DNA damage and mutations: ROS are powerful oxidants within cells that can directly damage mitochondrial DNA (mtDNA), causing DNA strand breaks, base damage, etc., thus increasing the risk of mutations and rearrangements in the mitochondrial genome. For example, ROS-induced DNA damage may lead to replication errors and repair defects in mtDNA, which subsequently trigger genomic structural rearrangements. Affecting mitochondrial dynamics and genomic stability: An increase in ROS levels affects the dynamic balance of mitochondria, including the processes of mitochondrial fission and fusion. Changes in mitochondrial dynamics may lead to variations in the distribution and stability of the mitochondrial genome, further promoting genomic structural rearrangements. Additionally, ROS may also regulate the degradation and renewal of the mitochondrial genome through pathways such as mitophagy, indirectly influencing the genomic structure58,59.

Interestingly, in our study, we found that the number of tRNAs was higher in the maintained lines than in the sterile lines by analyzing ncRNAs in all six cotton materials. ncRNAs have various regulatory mechanisms in plant sterility, including direct regulation of gene expression, participation in signal transduction pathways, and interactions with other regulatory molecules. There have been many studies demonstrating the association between ncRNAs and sterility. For example, a study found that a long non-coding RNA plays an important role in regulating photoperiod-sensitive male sterility in rice, revealing the key functions of lncRNAs in plant reproductive growth60.Wang et al. through genome-wide analysis, identified lncRNAs involved in fertility transition in the photo-thermosensitive genic male sterile rice line Wuxiang S, providing new insights into the mechanism of action of lncRNAs in plant sterility61.

It has been suggested that tRNA-derived small RNAs may be involved in the regulation of male sterility in crops. These small RNAs may indirectly affect male fertility by influencing gene expression and signaling pathways62.

In recent years, tRNA-derived small RNAs (tsRNAs) have gradually become a research hotspot. These small RNAs are generated through a series of complex processing procedures of tRNA and play extensive and important regulatory roles in organisms. In the context of research on male sterility in cotton, tsRNAs are highly likely to be involved in the regulatory network related to male sterility63.

On the one hand, tsRNAs may affect the stability and translation efficiency of target mRNAs through base complementary pairing, thus regulating the expression of genes related to male reproductive development. For example, certain specific tsRNAs might target genes that play vital roles in key processes such as pollen development and pollen tube elongation. In sterile lines, due to the abnormal expression of tsRNAs, the expression of these key genes becomes imbalanced, ultimately leading to male sterility.

On the other hand, tsRNAs may also be involved in epigenetic regulatory processes such as chromatin modification and gene silencing. They may be able to recruit relevant regulatory proteins to modify the structure of chromatin, thereby influencing the accessibility and expression status of genes. In cotton sterile lines, tsRNAs may change the chromatin state of regions where genes related to male reproductive development are located through this epigenetic regulatory mechanism, preventing these genes from being expressed normally and hindering the normal development of male reproductive organs.

In conclusion, the difference in the quantity of tRNA between the maintainer and sterile lines of cotton provides important clues for us to further investigate the potential regulatory roles of tRNA-derived small RNAs in male sterility. Further exploration of the specific regulatory mechanisms of tsRNAs in cotton male sterility will contribute to a more comprehensive understanding of the molecular basis of plant male sterility and provide a theoretical basis for the utilization of cotton heterosis and variety improvement.

Conclusions

Currently, there is a single narrow source of biparental germplasm resources for hybridization, and most of the CMS materials used in previous studies are heterologous cytoplasmic-nuclear materials obtained by heterologous karyoplasmic substitution. Negative cytoplasmic effects caused by cytoplasmic-nuclear heterozygosity may interfere with progeny selection for CMS, and the overabundance of genetic information that is not related to CMS may impede the study of the CMS mechanism in cotton. Therefore, in this study, we selected homologous cytoplasmic-nuclear and heterologous cytoplasmic-nuclear CMS lines and their respective maintainer lines as experimental materials and performed a comparative analysis of the morphology, cytology, physiology and biochemistry indices and the mitochondrial genome level. This study demonstrated that the homologous system is less affected by nuclear-cytoplasmic interactions, provided a theoretical basis for the creation of CMS lines optimized for cytoplasmic-nuclear interactions and laid the foundation for the analysis of cytoplasmic-nuclear homology and heterozygosity in the CMS mechanism.

Methods

Plant materials

The materials used in this study included a homologous cytoplasmic-nuclear sterile line, 07–113 A; two heterologous cytoplasmic-nuclear sterile lines, C2-113 A and C4-113 A; and their maintainer line, 07-113B. The CMS germplasm 07–113 S (referred to as the C3 cytoplasm) of Gossypium barbadense L. was obtained by continuous self-crossing of the 07-113 line, introduced by the Institute of cotton research of CAAS in 2007, and its cytoplasmic homologous CMS line is 07–113 A. The CMS germplasm 276 S (referred to as the C2 cytoplasm) of an island cotton strain, introduced from the Cotton Institute of the Chinese Academy of Agricultural Sciences, was obtained in July 2012 by transplanting the non-full-length Hcpdil5-2a gene into kenaf by the pollen tube channel method, which was saturated and backcrossed with 276B for many consecutive years to obtain a stable sterile line, which was named cytoplasmic homologous CMS line 276 A. C2-113 A and C4-113 A are stable, sterile, cytoplasmic heterologous CMS lines obtained by multigeneration saturation backcrossing with 07-113B as the paternal plant and “C2” and “C4” cytoplasmic materials as the maternal plant. Anthers of different developmental stages and buds of different sizes were taken from cottons at full bloom. The materials were collected from the experimental field of Guangxi University, and all materials were collected under the same normal growing conditions.

Morphological and cytological observations

Several CMS lines and their maintainer lines of cotton flowers were collected under the same conditions during the full bloom period, with homologous/heterologous classification determined by cytoplasmic-nuclear compatibility. and the morphology of the floral organs was observed and photographed (Canon, Tokyo, Japan). Flower buds of different diameters were also collected and fixed in Canon’s solution to make paraffin sections, which were measured with a Vernier caliper and classified into four grades: 2.0 ≤ BL < 3 (mother cell stage), 3.0 ≤ BL < 4 (tetrad stage), 4.0 ≤ BL < 5 (early uninucleate stage), and 5.0 ≤ BL < 6 (mature pollen formation stage)64.The sepals and petal tissues were removed, placed in Carnot fixative, and left to be fixed in a refrigerator at 4 °C for 24 h. The buds were dehydrated with a graded ethanol series65. Dehydrated flower buds were embedded in paraffin and sliced into 10 μm thick sections. Serial sections of flower bud tissue were placed on slides and stained with Senna-solid green. The sectioned flower buds were observed and imaged through a DMI3000B microscope (Leica, Wetzlar, Germany).

Enzyme activity and MDA content assays

The activities of POD and CAT can be used as biomarkers of oxidative stress levels. Based on their biological importance, ease of experimental manipulation, and wide application in scientific research. We chose POD and CAT as ROS-related indicators. Three biological replicates of each sample at each stage were used for the Enzyme activity assay. The enzyme activity of CAT was determined by double-antibody sandwich assay using a plant superoxide dismutase enzyme-linked immunosorbent assay (ELISA) kit (Shanghai Enzyme-Link Bio), and the activity of POD was determined by guaiacol colorimetric assay according to the experimental method of Li Hesheng66. The MDA content in cotton anthers was determined by a double-antibody sandwich assay using the plant hydrogen peroxide ELISA kit (Shanghai Enzyme-Link Bio), which was conducted with three biological replicates and the index values were measured following the manufacturer’s instructions. Statistical analyses were performed using Student’s t-test after verifying normality (Shapiro-Wilk test). The error bars represent the standard deviation of three replicates, and asterisks indicate significant differences (*P < 0.05, **P < 0.01, ***P < 0.001; Student’s t-test)67.

Mitochondrial sequencing and assembly

Seeds of cotton collected from the experimental field of Guangxi University were threshed using a threshing machine. Cotton seeds were decolorized with concentrated sulfuric acid and placed on wet gauze in Petri dishes. The seeds were decolorized with concentrated sulfuric acid, placed on wet gauze in Petri dishes, and incubated for one day at 25–28 °C in a light-free incubator. Transfer the germinated seeds to a tap water hydroponic system and incubate in a dark incubator at 25–28 °C for 7 days to obtain etiolated seedlings. Total DNA was extracted from 0.5 g of defoliated seedlings using a modified cetyltrimethylammonium bromide (CTAB) method68, and the extracted mtDNA was sent to Shanghai Lingen Biotechnology Co.

The purity and quantity of extracted DNA were assessed using Nanodrop 2000 (Thermo Fisher Scientific, Carlsbad, CA, USA) and Qubit 2.0 (Thermo Fisher Scientific, Carlsbad, CA, USA)., After passing the quality control test, the library was constructed with a starting amount of 1 µg DNA, the DNA was fragmented into 300–500 bp fragments by ultrasonication with Covaris M220, the library was enriched and amplified by polymerase chain reaction (PCR), and the target bands were recovered with 2% agarose gel (Certified Low Range Ultra Agarose). The cotton mitochondrial genome was sequenced using a high-throughput sequencing platform (Illumina NovaSeq 6000). Then, SOAPdenovo69(version 2.04) software with the default parameters (-K 121 -F -p 8) was used to assemble the sequences, and second-generation assembled sequences were aligned to Nanopore third-generation data using BWA v0.7.17 to extract the third-generation data of the target samples. Then, the extracted third-generation data were assembled by combining them with the second-generation data. Clean reads were compared to the mitochondrial genome sequence, and bases were corrected using Pilon v1.23. Finally, the start position and orientation of the mitochondrial scaffold were determined from the reference genome to obtain the final mitochondrial genome sequence. The sequencing experiment was conducted without biological replicates.

Analysis and annotations of the Mt genomes

We used the homology matching prediction method to perform gene prediction on the sample genome. The protein sequences of the National Center for Biotechnology Information (NCBI) mitochondrial reference genome were first aligned to the sample genome using the software BLAST + 2.7.1 with the threshold parameter e-value 1e-5. Poor alignment results and redundancy were removed, and the integrity of the sample genes and the exon/intron boundaries were manually corrected to obtain the conserved gene set with a high degree of accuracy. The assembled genome sequences of the sequenced samples were combined with the predicted coding genes to obtain a circle map of the sample genome. The software used was OGDRAW (https://chlorobox.mpimp-golm.mpg.de/OGDraw.html).

Structural variant detection

MUMmer software was used to compare the target and reference genomes, and then LASTZ was used for region comparison to identify structural variants (SVs).

Collinearity analysis

Large-scale collinearity between genomes was determined using MUMmer software. Interregion comparison was performed using LASTZ to confirm the local positional alignment relationships and to find the regions of translocation (Translocation/Trans), inversion (Inversion/Inv) and translocation + inversion (Trans + Inv)70.