Mitochondrial Proteomics and Transcriptional Analysis of Cytoplasmic Male Sterility in Sugar Beet using iTRAQ and qRT–PCR

Sugar beet (Beta vulgaris L.) is an important raw material for the sugar industry, and its output is second only to sugar cane. Cytoplasmic male sterility (CMS) is a phenomenon of pollen abortion that has important implications in sugar beet hybrid breeding. Male plant sterility is usually considered to be associated with mitochondrial dysfunction. Although mitochondrial genes associated with male sterility have been well explored, the different mitochondrial proteomics of CMS in sugar beet are still poorly understood. In this study, differentially expressed mitochondrial proteomic analysis was performed on the ower buds of the male sterile line (DY5-CMS), its maintainer line (DY5-O) and a fertility restorer line (CL6), using an isobaric tag for relative and absolute quantitation (iTRAQ) technology. A total of 2260 proteins were identied by mass spectrometry, of which 538 were differentially expressed proteins. Most of them were involved in protein metabolism, carbohydrate and energy metabolism, and binding. More specically, some cysteine and methionine metabolism proteins (A0A0J8BGE0, A0A0J8CZM6, A0A0J8D7W0 and A0A0J8BCR7) may play important roles during the formation of CMS. This study provided an in–depth understanding of the CMS molecular mechanism at the protein level in sugar beet. biosynthesis of secondary metabolites (14.79%); biosynthesis of amino acids (8.28%); ribosome (7.69%); carbon metabolism (6.51%); cysteine and methionine metabolism (4.14%); arginine biosynthesis (3.55%); valine, leucine and isoleucine degradation 2-oxocarboxylic acid metabolism (3.55%); amino sugar and nucleotide sugar metabolism alanine, aspartate and glutamate metabolism (2.96%); phenylalanine, tyrosine and tryptophan biosynthesis arginine and proline metabolism glycine, serine and threonine metabolism and avonoid biosynthesis DY5-CMS/CL6 comparison, the DEPs in ribosome biosynthesis of amino carbon metabolism 2-oxocarboxylic acid metabolism (9.09%); cysteine and methionine metabolism carbon xation in photosynthetic organisms arginine biosynthesis and alanine, aspartate and glutamate metabolism Last, for the CL6/DY5-O comparison, biosynthesis amino carbon aminoacyl-tRNA biosynthesis glycolysis/gluconeogenesis avonoid biosynthesis valine, leucine and isoleucine nitrogen and arginine biosynthesis synthetase (GSH-S); Cysteine synthase (Cs); Glyceraldehyde-3-phosphate dehydrogenase (GAPDH); Histone 2A (H2A); Isobaric tag for relative and absolute quantitation technology (iTRAQ); Quantitative real-time polymerase chain reaction (qRT–PCR); Serine hydroxymethyltransferase (SHMT); Superoxide dismutase (SOD); Glutamate-cysteine ligase (GCL); Glutathione (GSH); Programmed cell death (PCD)


Introduction
Cytoplasmic male sterility (CMS) is a maternally inherited condition in which higher plants fail to produce functional pollen but maintain female fertility [1]. CMS is a useful trait for commercial hybrid breeding because inactive pollen eliminates the need for the expensive process of stamen removal [2]. CMS has been reported in over 150 plant species and is often associated with chimeric mitochondrial open reading frames (ORFs) [3][4]. A variety of key events focus on mitochondria, including oxidative phosphorylation and programmed cell death (PCD). In many cases, transcripts originating from these altered mitochondrial ORFs are translated into proteins that have been identi ed in many crops and appear to interfere with pollen development and mitochondrial function [5]. It has been demonstrated that these mitochondrial proteins are associated with CMS [5] and co-segregated with the sterility phenotype [6]. Nuclear genes (Rf) that restore the fertility of the male-sterile cytoplasm prevent the deleterious effects of mitochondrial abnormalities, and reduce the amount of sterility-related proteins [5][6][7].
Sugar beet (Beta vulgaris L.) is an important raw material for the sugar industry and a potential source of alternative energy [8]. CMS in sugar beets was rst discovered by F.V. Owen [9], and the so-called Owen cytoplasm has subsequently been used worldwide for hybrid seed production. The complete nucleotide sequences of the mitochondrial genomes from fertile and Owen CMS beets have been determined [10][11], which allows the in-depth study of the protein products of these mitochondrial genomes. Structural differences in the mitochondrial loci of atp1, atp6, cob, cox1, cox2 and rps3 have been found in the comparison of sugar beet CMS and fertile lines [12][13][14][15]. Four transcribed open reading frames (ORFs) of Satp6, Scox2-2, Sorf324 and Sorf119, which may produce novel proteins that affect mitochondrial function in pollen-producing tissues, have also been identi ed [11,16]. It was reported that polypeptides of 6 kDa, 10 kDa, 35 kDa and 12 kDa may be associated with the expression of CMS in sugar beet [7,[16][17], and the decreased activity of complex V and the increase of other complexes with ATPase activity may affect CMS in sugar beet [18]. However, these ndings are not enough to explore the secrets of beet CMS systems. A large number of CMS-associated mitochondrial proteins have been identi ed and described in rice [2,19], wheat [20][21], maize [22][23], brassica [1,24], tomato [25] etc. However, no two CMS mutations described to date are identical [6]. Sugar beet, as a complex glycophytic member of Chenopodiaceae, is different from other plants. Little is known about the mitochondrial proteins involved in the expression of the CMS phenotype in sugar beet, and the molecular mechanism that these proteins cause remains an enigma.
Although the molecular mechanisms of CMS have been extensively investigated, there is little information on quantitative comparisons of mitochondrial proteomics in fertile and CMS beet lines, especially with reference to differentially expressed proteins (DEPs) and metabolic pathways. In this study, a comprehensive analysis of mitochondrial proteomics and transcription of the CMS phenotype in sugar beet was performed by using an isobaric tag for relative and absolute quantitation (iTRAQ) technology and quantitative real-time PCR (qRT-PCR). More mitochondrial proteins associated with CMS beets were found through mass spectrometry analysis, and the reliability of the results was further veri ed by quantitative analysis of the genes. The molecular function of these proteins, as well as the metabolic pathways involved, are described in detail. Possible interactions among these proteins are also discussed. This report thus provides the general characteristics of the CMS-associated mitochondrial proteome in beets. The candidate protein and metabolic pathway provided in this report would be conducive to revealing the molecular mechanism of CMS in beets.

Plant materials
Three lines of sugar beet (Beta vulgaris L.) were used in this study, including the male-sterile line DY5-CMS, maintainer line DY5-O and fertility-restored line CL6. The DY5 pair is composed of two lines that have the same genetic background through repeated backcrossing, but their cytoplasmic genomes are different. CL6 is the continuous system selection offspring of GW-65 in the 1960s (in the USA). Three biological replicates of all the materials were sampled during the vegetative growth stage. These replicates were mixed to create sample pools for subsequent experiments.

Extraction and quanti cation of mitochondrial proteins
Mitochondria were isolated from 20 g of sugar beet roots using the Plant Mitochondrial Extraction Kit (BestBio, Shanghai, China). The principle of the kit is that mechanical shear force is used to destroy the cell wall and cell membrane and release organelles, and then, based on the difference in the sedimentation coe cient of organelles, mitochondria are separated using differential centrifugation combined with density gradient centrifugation. According to previous studies, this is an improved optimization method [26][27][28][29][30]. Ensuring the quality of mitochondria isolated from test material is a key step in proteomics. Janus Green B is a vital stain used for the identi cation of mitochondria, which binds to cytochrome c oxidase to give a blue-green color [20,21,31]. The isolated mitochondria were stained with 0.5% Janus Green B (Solarbio, Beijing, China) for 1-3 min, and their activity was observed under an optical microscope (COIC, Chongqing, China). Scanning electron microscopy (SEM; Hitachi SU8010, Tokyo, Japan) of frozen samples was used to examine mitochondrial integrity. Mitochondria extracted from 20 g of beet roots were dissolved in 5 mL of SDT buffer (4% SDS, 100 mM DTT and 150 mM Tris-HCl, pH 8.0). The suspension was incubated in boiling water for 30 min followed by ultrasonication (80 W, 10 s ultrasonication at a time, every 15 s, for 10 times) [32]. The supernatant obtained after centrifugation (25°C, 14000 × g, 45 min) contained mitochondrial proteins, and the protein concentration was determined following the manufacturer's protocol (BCA Protein Assay Reagent, Promega, USA).

iTRAQ labeling and strong cation exchange fractionation
Brie y, 100 µg mitochondrial proteins from each sample was digested by trypsin (37°C, 16-18 h), and the Thermo Scienti c EASY columns (2 cm × 100 µm × 5 µm-C18 and 75 µm × 100 mm × 3 µm-C18) for analytical separation at a ow rate of 300 nL min −1 for 60 min. The gradients used were as follows: 0-50 min, B from 0-40%; 50-57 min B from 40-100%; 57-60 min, B maintained at 100%. Data were acquired using a Thermo Q-Exactive mass spectrometer (Thermo Fisher Scienti c) and the experimental procedure was conducted according to a previous method [33]. Mass spectrometry (MS) data were chosen for higher-energy collisional dissociation (HCD) fragmentation from the most abundant precursor ions detected in the survey scan (300-1800 m/z). Determination of the target value was based on predictive automatic gain control (pAGC). Dynamic exclusion duration was 40 s. Survey scans were acquired at a resolution of 70,000 at m/z 200, and a resolution of 17,500 at m/z 200 was set for HCD spectra. The normalized collision energy was 30 eV and the under ll ratio was de ned as 0.1%. Enzyme: Trypsin, max missed cleavage: 2, Fixed modi cation: Carbamidomethyl (C), iTRAQ4plex (K), iTRAQ4plex (N-term), Variable modi cation: Oxidation (M), Peptide mass tolerance: 20 ppm, MS/MS tolerance: 0.1 Da. Protein identi cation was calculated by only unique peptides, and the normalization method and protein change ratio type (up-or downregulated) were both set as medians. The results were ltered based on a false discovery rate (FDR) of no more than 1% to guarantee the result's con dence [34] and a Mascot probability of 95%. When the abundance level of a protein showed a difference corresponding to a 1.3-fold or 0.77-fold change from DY5-CMS/DY5-O, DY5-CMS/CL6, and CL6/DY5-O comparisons, and had a statistically signi cant level (P value<0.05) by signi cance A analysis, the protein was then considered to be a differentially expressed protein (DEP).

Bioinformatics
Functional analysis and pathway analysis of proteins were performed using GO (http://www.geneontology.org/) and KEGG (http://www.kegg.jp/kegg/kaas/), respectively. Proteinprotein interaction networks were constructed using the publicly available program STRING (http://stringdb.org/). STRING, a database of known and predicted protein-protein interactions, quantitatively integrates the interaction data (from four sources: the genomic context, high-throughput experiments, coexpression and previous knowledge) for a large number of organisms and transfers information between these organisms under applicable conditions [33].
For statistical analysis, Fisher's exact test with the hypergeometric algorithm was used to calculate P values, and the cutoff of P values lower than 0.05 was set to select signi cantly enriched terms in categories. FDR attained by the Benjamini-Hochberg method was used to adjust P values.

qRT-PCR
We performed qRT-PCR to verify whether the iTRAQ results were consistent with the quantitative results of genes in the roots and leaves. Total RNA was extracted from the roots and leaves of sugar beet with a TaKaRa MiniBEST Plant RNA Extraction Kit (Takara Bio, Osaka, Japan), and subjected to rst-strand cDNA synthesis using a PrimeScript™ RT reagent kit (Takara Bio) according to the manufacturer's instructions. Ct values were obtained via the real-time uorescence detection method with SYBR® Premix EX Taq™ ІІ (Takara Bio) on an ABI 7300 Real Time PCR System (Applied Biosystems). The primers used were designed according to the cDNA sequences published on NCBI by using Primer-BLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast/; Table 1). The BvICDH gene (GenBank AF173666) was selected as a reference gene as described previously [35]. The 2 -ΔΔCT method, SPSS 19.0 software, and Origin 8.0 software were used for statistical analysis and mapping. All reactions were performed in three biological and three technical replicates.

Isolation of mitochondria from sugar beet
The isolated mitochondria were morphologically intact, and spherical-ellipsoidal in shape ( Supplementary  Fig. S1).
The four metabolic pathways in the three comparisons all included the biosynthesis of amino acids; carbon metabolism; valine, leucine and isoleucine degradation; and arginine biosynthesis. In particular, 5 metabolic pathways, including ribosome; 2-oxocarboxylic acid metabolism; alanine, aspartate and glutamate metabolism; tropane, piperidine and pyridine alkaloid biosynthesis; and isoquinoline alkaloid biosynthesis, were only found in the DY5-CMS/DY5-O and DY5-CMS/CL6 comparisons. However, avonoid biosynthesis was only found in the DY5-CMS/DY5-O and CL6/DY5-O comparisons. These pathways are involved in carbohydrate and energy metabolism, protein metabolism, the biosynthesis of secondary metabolites and nucleotide metabolism, which means that these pathways were the most important in sugar beet under CMS conditions.

Network analysis of DEPs
To better understand how sugar beet transmits CMS signaling through protein-protein interactions, the DEPs associated with CMS were analyzed by STRING. This analysis revealed a protein association network that has a very notable interaction (Fig. 3). According to the protein association network (Fig. 3), the DEPs were highly enriched in carbon metabolism, metabolism of amino acids and ribosomes.
Abbreviations of the speci c protein names in the network are given in Supplementary Table S3-3. These pathways are involved in carbohydrate and energy metabolism, protein metabolism and nucleotide metabolism. Interestingly, 5 DEPs occupied the key regulatory sites of this protein network, including serine hydroxymethyltransferase (SHMT), D-3-phosphoglycerate dehydrogenase (PGDH), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), aspartate aminotransferase (AAT) and citrate synthase (CS) (Fig. 3). Among these proteins, only AAT is involved in protein metabolism, and the rest are involved in carbohydrate and energy metabolism.

Gene expression analysis of speci c DEPs
qRT-PCR analysis was performed to investigate gene expression changes at the mRNA level, and thirteen genes were selected for this analysis (Fig. 4). The genes analyzed in the roots were different from those in the leaves based on the pattern of changes at the mRNA level. Furthermore, on the basis of the pattern of changes at the protein and mRNA levels, the genes analyzed in the roots were clustered into three groups: Group I, consistent changes at the transcript and protein levels; group II, inverse changes at the transcript and protein levels; and group III, changes only at the protein level. More speci cally, ve genes, chalcone-avanone isomerase family protein (CHI 2-A), ATP synthase subunit α (ATP-α), CS, PGDH and glutamine synthetase (GS), were clustered in Group I; seven genes, glutathione synthetase (GSH-S), GAPDH, histone 2A (H2A), AAT, SHMT, 60S ribosomal protein L36 (L36) and annexin, were clustered in group II; and only Cu/Zn-superoxide dismutase (Cu/Zn-SOD) was clustered in group III.

Discussion
Cytoplasmic male sterility (CMS) is a complex phenomenon of the abnormal development of stamens and pollen in owering plants and involves the interaction of nuclear and cytoplasmic genes, a variety of metabolic processes and structural changes [24]. A mitochondrial proteomics study on CMS of sugar beet has previously been reported [18], but gel-based methods limited the detection of proteins. iTRAQ technology, as a nongel-based quantitative proteomics approach, has been used in research on the male sterility of some plants due to its unique advantages [36][37], but to date, there is no related report on CMS in sugar beet. To further elucidate the molecular mechanism of CMS in sugar beet, a comparative mitochondrial proteomic analysis was performed on the CMS line DY5-CMS, its maintainer line DY5-O and the restorer line CL6 using iTRAQ and qRT-PCR. By using iTRAQ technology, we identi ed a large number of DEPs that were either up-or downregulated in CMS beet relative to the other lines. The possible potential effects of some DEPs and metabolic pathways on male sterility in sugar beet are discussed below.

DEPs involved in protein metabolism
Proteins perform physiological functions and play essential roles in the growth and development of organisms by interacting with other molecules. Our understanding of CMS in beets was advanced by proteins involved in protein metabolism that included cysteine and methionine metabolism, nitrogen metabolism and ribosomes. In this study, cysteine and methionine metabolism proteins AAT (A0A0J8BGE0), AAT (A0A0J8CZM6), GCL (A0A0J8BCR7) and Cs (A0A0J8D7W0) were more accumulated in the CMS line than in the maintainer and restorer lines ( Table 2 and Fig. 5). In contrast, GSH-S (A0A0J8FCL3) was downregulated in the CMS line compared to the restorer line (Supplementary  Table S2). AAT is an important enzyme in amino acid metabolism that catalyzes the reversible transfer of an α-amino group between aspartate and glutamate [38]. A previous study of AAL toxin-induced cell death in Arabidopsis thaliana indicated that AAT was included in the most upregulated genes 24 h after AAL treatment [39]. This result suggested that AAT was a potentially important gene for PCD and played a role in the activation of cell death. In humans and other animals, AAT has long been used as an established serum marker for cardiac and liver damage and may also be hypothesized as a useful indicator of cell death [40]. PCD, as an active suicidal process, is an essential part of many developmental processes and responses to plant pathogens [41][42]. The results from a previous study also indicated that the PCD process induced by oxidative stress may be the physiological cause for the abortion of microspores in the maize CMS-C line [21]. In addition, GCL is the rate-limiting enzyme in the GSH biosynthesis pathway [43]. GSH is an antioxidant and eliminates free radicals, and strains lacking glutathione have been reported to be susceptible to oxidative stress induced by toxins of peroxide and superoxide anions and lipid hydroperoxides [44][45]. Elevated GCL in the CMS line may limit the GSH biological pathway, which cannot normally participate in cellular protection by hindering the scavenging of free radicals and enzymatic reduction reactions. This nding was also supported by the downregulation of GSH-S in the CMS line. The abundance changes of GSH-S may have broken the process of response to stimulus and even be a key factor in CMS, and explanations for this could be: (i) Insu cient GSH-S might inhibit glutathione synthesis, which may reduce the ability of free radical scavengers in CMS line and cause an improper function in plant defense and stress tolerance; (ii) The level of ROS as a toxin is elevated due to the downregulation of GSH-S, which may be involved in the PCD process. In contrast, Cs was de ned as a sensor whose activity is balanced between cysteine consumption and sulfate assimilation and controls cysteine syntheses within the cell [46]. As KEGG results of cysteine and methionine metabolism (bvg00270) show (Fig. 5), cysteine is converted from serine (via acetylserine) by the transfer of hydrogen sul de in plants. Interestingly, elevated cysteine is used as the sulfur donor for GSH synthesis [46], while GSH-S was downregulated in this study. Previous research also shows that Cs catalyzes the biosynthesis of cysteine on the basis of a regulatory network, which mediates between inorganic sulfur supply and the demand for reduced sulfur during plant growth and in response to environmental changes [47][48].
In this study, GS (A0A0J8BRK7) was downregulated in the CMS line compared to the maintainer line, and this effect was reversed upon fertility restoration (Supplementary Table S2). These ndings were validated by qRT-PCR analysis (Fig. 3). GS is one of the key enzymes of nitrogen metabolism and catalyzes the synthesis of glutamine with NH 4 + and glutamic acid as substrates [49]. GS plays a central role in maintaining the balance of carbon and nitrogen, such as preventing toxins generated by excessive NH 4 + [49]. Therefore, the synthesis of glutamine may be inhibited by the downregulation of GS in CMS beets, which disrupts the balance between carbon and nitrogen. CMS may be caused by toxins generated by excessive NH 4 + . Consistent with our results, GS was also downregulated or absent in sterile lines of pepper and wolfberry, which affected amino acid metabolism and cytoskeleton biosynthesis, thereby inducing male sterility [50][51]. Many ribosomal proteins that differentiate DY5-CMS from DY5-O and CL6 in the same way were also found in this study ( Table 2). They are major components of ribosomes and play important roles in protein synthesis.
In many species, a large number of DEPs involved in protein metabolism were identi ed, which con rmed that this pathway was associated with CMS [20, 21, 24, 25, 31, 36 and 37]. According to the results in Table 2, cysteine and methionine metabolism was preferentially upregulated under CMS conditions.
There were more products of protein metabolism in the CMS line than in the maintainer and restorer lines (

DEPs involved in carbohydrate and energy metabolism
The main physiological function of carbohydrate and energy metabolism is to provide energy and a carbon source. Some proteins affected include GAPDH (A3FMH0), ATP-α (Q9XPH4), SHMT (A0A0J8C0D9), PGDH (A0A0J8BRE3) and CS (A0A0J8BI86) ( Fig. 2; Table 1). Stamen and pollen development is a high-energy-requiring process [5], and the failure of pollen development in CMS lines re ects the abnormal expression of these proteins. Several studies on plant proteomics have also suggested that carbohydrate and energy metabolism is closely related to male plant sterility [37,52].
Most previous studies showed that the downregulation of genes or proteins associated with carbohydrate and energy metabolism in sterile lines resulted in insu cient energy supply and led to abortion in plants.
Similar to the above, two proteins downregulated in the CMS line were also found in this study. In detail, the expression of GAPDH in the CMS line was downregulated compared with that in the restorer line, and its mRNA abundance was signi cantly lower in CMS than in the maintainer line (Fig. 3a). GAPDH is a key enzyme in carbohydrate metabolism and is involved in the formation of ATP in the glycolytic pathway, DNA repair, tRNA export, and membrane fusion and transport [53]. Furthermore, the association of GAPDH with cell survival may occur by providing ATP to maintain mitochondrial membrane potential via ATPase, helping to counteract the effects of energy collapse by the loss of mitochondrial function [53].
Decreased GAPDH may block the carbohydrate metabolism pathway, resulting in a shortage of energy supply, which affects anther development. In wolfberry and soybean, GAPDH downregulation in the sterile line led to insu cient energy and sterility, which was consistent with our conclusion [51,54]. Furthermore, the expression level of PGDH was lower in the CMS line than in the restorer line. PGDH is a member of the oxidoreductase homolog family and plays a role in allosteric regulation [55], which suggests that decreased PGDH would hinder allosteric regulation, resulting in a harmful effect on ATP synthesis. Thus, there may be a direct relationship between the occurrence of CMS and allosteric regulation reduction.
In contrast to previous research, three proteins upregulated in the CMS line that were associated with carbohydrate and energy metabolism were identi ed in this study. The expression of ATP-α was higher in the CMS line than in the maintainer line according to our iTRAQ and qRT-PCR results (Supplementary Table S1 and Fig. 3). An investigation of the mechanism underlying CMS in kenaf showed that ATP-α and ATP-β were differentially expressed in the CMS line [31]. ATP synthase is a key enzyme in the process of oxidative phosphorylation of mitochondria [24] and plays a critical role in energy metabolism. ATP-α and ATP-β work together, but the overexpression of ATP-α affects the normal functioning of ATP synthase, suggesting that microspore abortion in the CMS line may be related to abnormal expression of ATP-α.
SHMT catalyzes the conversion of glycine to serine and participates in the photorespiration process [56][57]. A previous study in temperature-sensitive wheat reported a decrease in SHMT in the sterile line, which may inhibit photorespiration [58]. However, this nding differed from that of our iTRAQ results, which revealed the different expression patterns of SHMT in different species, but all affected the photorespiration process (Fig. 3). This phenomenon is similar to previous studies, in which CMS-related genes in many crops showed rare similarity to each other, and all led to a similar CMS phenomenon [59]. This implies that common pathways in all kinds of CMS types must exist and be more important than each CMS gene. The TCA cycle is the most important metabolic pathway for energy production under aerobic conditions [60]. CS distributed in mitochondria is a critical enzyme in this cycle. CS catalyzes the reaction of oxaloacetic acid (OAA) and acetyl-CoA to form citric acid and CoA [61]. Surprisingly, both iTRAQ and qRT-PCR (Fig. 3a) revealed that the CS expression level was higher in the CMS line than in the maintainer and restorer lines, which would normally convert more acetyl-CoA into energy. However, CS may not have a regulatory role in those tissues where its activity is in considerable excess of the maximal rates of acetyl-CoA production or isocitrate oxidation [62], which still needs to be further veri ed.

DEPs involved in binding
Two DEPs related to ion binding and nucleotide binding were identi ed in this study. Annexin is a Ca 2+binding protein that interacts with membrane phospholipids in a calcium-dependent manner [50]. In pepper, the downregulation of annexin in the sterile line disrupted the Ca 2+ balance and reduced fertility [50]. Calcium was established as a second messenger that plays an important role during plant growth and development [52]. In this study, the abundance of annexin (A0A0J8C5G7) in the CMS line was signi cantly lower than that in the maintainer and restorer lines (Table 2), and there was no difference between the maintainer and restorer lines. These results indicated that downregulation of annexin may be associated with the onset of male infertility. In detail, the downregulation of annexin hindered the calcium-binding process, which impaired Ca 2+ homeostasis; Ca 2+ cannot function properly as a second messenger and reduced the interaction between annexin and membrane phospholipids. Furthermore, we found that H2A (A0A0J8BEW7) content was lower in the CMS line than in the restorer line. Histone is involved in nucleotide binding and plays an essential role in gene expression control and genome management. Previous studies also reported that the histone content in sterile lines was lower than that in maintainers in wheat and pepper [63][64]. The abnormal expression of H2A may disrupt nucleotide binding and gene expression, which in turn affects anther growth and development and leads to abortion of the plant. Changes in both proteins hinder the binding process, disrupt the balance of the substance during the reaction and cause plant sterility.

Conclusion
In our research, we used iTRAQ technology to identify a total of 2260 differentially expressed proteins related to the male sterility phenotype involved in protein metabolism, carbohydrate and energy metabolism, and binding from three lines of sugar beets. More speci cally, some cysteine and methionine metabolism proteins (A0A0J8BGE0, A0A0J8CZM6, A0A0J8D7W0 and A0A0J8BCR7) may play important roles during the formation of CMS. In addition, we identi ed 13 key DEGs potentially associated with sugar beet CMS. The way that the proteins associated with carbohydrate and energy metabolism and binding affect male sterility was further elucidated. It will be interesting to study the mutants associated with CMS proteins to better understand the molecular mechanisms of CMS in the future.