Proteomic analysis of differential anther development from sterile/fertile lines in Capsicum annuum L.

Plant materials and growth conditions

Male fertile and sterile pepper lines were grown in a greenhouse according to standard agronomic practices. During flowering, flower buds were collected at different development stages (meiosis, microspore tetrad, binucleate, and microspore stages) and cytological observations were conducted using an Olympus CX31 microscope. In addition, three biological replicates of fertile and sterile anthers were collected and frozen rapidly in liquid nitrogen, before storing at –80 °C for proteomic analysis.

Cytological observations

Flower buds were collected and fixed in formalin:acetic acid:alcohol (ethanol:acetic acid:formaldehyde:double-distilled H₂O = 10:1:2:7) for 24 h at different development stages to make cytological observations of semithin sections. The materials were then washed three times with 200 μL of 0.2 M phosphate-buffered saline and dehydrated with an ethanol concentration gradient. Finally, the samples were embedded in resin for 24 h and sectioned to obtain transverse slices using a microtome. The transverse slices were stained with 1.5–2% toluidine blue (Liu et al., 2014; Lou et al., 2007), before photographing using an Olympus CX31 microscope. In addition, pollen grains were collected from anthers using tweezers to determine their viability. The pollen grains were stained with aceto-carmine and photographed using an optical microscope.

Detection of enzyme activities and plant hormone contents

Samples of flower buds at different stages were collected and ground in liquid nitrogen to determine the oxidordeuctase enzyme activities. The ground powdered samples were extracted and incubated in extraction buffer. The homogenate was centrifuged at 8,000g and 4 °C for 15 min, and the supernatant solution was used to determine the enzyme activities with extraction kits according to the manufacturer’s instructions.

To determine the plant hormone contents, the soluble supernatant was extracted from 0.1 g of flower bud samples collected at different stages after homogenizing at 4 °C overnight in 1 mL of chilled buffer (methanol:double-distilled H₂O:acetic acid = 80:20:1). The residue was extracted in chilled buffer after centrifugation at 8,000g for 10 min. The residue was centrifuged and heated up to 40 °C under a stream of nitrogen. The sample was extracted and decolorized three times in 0.5 mL of petroleum ether at 60–90 °C. Saturated citric acid aqueous solution was then added to the sample. Finally, the sample was spun to mix well after adding 0.5 mL of methanol and used for spectrophotometric assays.

Protein extraction, TMT labeling, HPLC (High Performance Liquid Chromatography) fractionation, and LC-MS/MS analysis

All samples for protein extraction were ground in liquid nitrogen and sonicated three times using an ultrasonic processor in cool lysis buffer (8 M urea and 1% protease inhibitor). Each solution was centrifuged at 12,000g and 4 °C for 10 min. The protein concentration in the supernatant was assayed using a BCA kit according to the manufacturer’s instructions. To digest the proteins, 5 mM dithiothreitol was added to the supernatant solution, which was heated for 30 min at 56 °C and then treated with 10 mM iodoacetamide to alkylate for 15 min at 56 °C in darkness. The protein solution was diluted with 100 μL of 50 mM triethylammonium bicarbonate. Finally, trypsin was added to the protein sample solution at a trypsin:protein mass ratio of 1:50 and the first protein digestion was conducted overnight at 37 °C. For the second protein digestion, the trypsin:protein mass ratio was 1:100.

For the TMT/iTRAQ assay, 0.5 M triethylammonium bicarbonate was added to the peptide solution and reconstituted using the TMT/iTRAQ kit. One unit of reagent was thawed in acetonitrile for 2 h at 25 °C. All peptides were subjected to LC-MS/MS analysis using nano-flow liquid chromatography tandem mass spectrometry (LC-MS) with an EASY-nLC 1200 UPLC system connected to a Q Exactive™ Plus quadrupole orbitrap mass spectrometer.

Protein identification and bioinformatics analysis

For protein identification, raw data were searched against Capsicum_annuum_4072 database (35,548 entries) concatenated with reverse decoy database by Proteome Discoverer search engine (V2.4.1.15). Two missing cleavages were allowed using Trypsin cleavage enzyme. The mass tolerance for precursor ions was 10 ppm and fragment ions was 0.02 Da. Carbamidomethyl on Cys, TMT-6plex (N-terminus) and TMT-6plex (K) were specified as fixed modifications. In addition, oxidation on Met and acetylation (N-terminus) were as variable modifications. The quantitative method was set as TMT-6plex. FDR was adjusted to 1% and minimum score for modified peptides was set >40. Minimum peptide length was set at 6. The raw data and results were submitted on ProteomeXchange (accession: PXD031331).

For differentially expressed proteins analysis, firstly, the expression ratio of each protein between two samples with multiple replicates were obtained and calculate the average value. Then, the significance p value of differentially expressed proteins was calculated by the two-tailed T-test method. DAPs with p value <0.05 and expression ratio >1.3 were up-regulated, while DAPs with p value <0.05 and expression ratio <1/1.3 were down-regulated proteins.

All DAPs were functionally classified into biological processes, molecular functions, and cellular components based on Gene Ontology (GO) annotations (http://www.geneontology.org/), and enriched subcategories and metabolic pathways using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (http://www.genome.jp/kegg/pathway.html). In addition, all DAPs were searched against the STRING database (V10.1) to detect protein–protein interactions (PPIs) based on confidence score >0.7. Arabidopsis thaliana was used as the reference species for establishing PPI networks. Data visualization in venn diagram were used by Tbtools software (Chen et al., 2020). The PCA diagram, volcano plot and heat map were completed by online software (https://www.ptmbiolabs.com/). The bar charts in this study were made by origin software (V.7.0).

Targeted protein quantification by PRM

Protein extraction and trypsin digestion were performed in the TMT experiment for targeted protein quantification. After LC-MS/MS analysis, the data obtained for each sample were processed using the PRM acquisition method to quantify the abundances of targeted proteins. Each protein with more than two unique peptides was quantified, but only one peptide segment was identified for some proteins due to sensitivity and other reasons. Three biological replicates of each sample were performed and 15–20 proteins were randomly selected and quantified. PRM analysis was conducted by Jingjie PTM BioLabs (Hangzhou, China) for this study.

Differential anther development in male sterile and fertile pepper lines

We screened one putative male sterile pepper line that produced few pollen grains compared with the fertile pepper line. To confirm the pollen abortion characteristics in the male sterile pepper line, we observed the morphological differences in anther development during different stages using semi-thin sections. No obvious differences were found between the fertile and sterile male lines in the sporogenous cell stage and the pollen mother cell stage (Figs. 1A, 1B, 1H, and 1I). The nucleus volume was larger and the microspore mother cells entered meiosis in the male sterile line (Fig. 1C). Observations of tapetum cells indicated that they disintegrated in the male sterile line (Fig. 1J). Furthermore, the stained microspores were dispersed in the chamber in the microspore stage in the fertile pepper line (Figs. 1D–1F). By contrast, the microspores were crushed and disintegrated, and they adhered to the tapetum to form a dense belt in the male sterile line (Figs. 1K–1M). In the last stage of pollen development, normal grains were clearly observed during pollen maturation in the fertile pepper line (Fig. 1G), whereas a thin layer of tapetum cells formed a dense belt without mature pollen in the sterile line (Fig. 1N). Thus, morphological observations indicated that premature tapetum death may have been the cause of male sterility.

To further assess the viability of the pollen grains, anthers were collected and the pollen grains were stained with aceto-carmine. Interestingly, large amounts of the pollen grains were strongly stained in the fertile pepper line (Fig. 1O), whereas none of the pollen grains stained in the male sterile line (Fig. 1P). These results suggest that male sterility may have been due to the production of few pollen grains and their infertility.

Antioxidant enzyme activities in different anther development stages in male sterile and fertile pepper lines

We investigated the differences in the activities of antioxidant enzymes during anther development in the male sterile (BY) and fertile (KY) pepper lines. In particular, the superoxide dismutase (SOD), CAT, POD, and malondialdehyde enzyme activity levels were compared in three anther development stages (microspore tetrad, meiosis, and uninucleate microspore stages). As shown in Figs. 1Q–1T, the POD enzyme activity levels (9,499–11,000 U/g) in the meiosis and microspore tetrad stages were significantly lower in the male sterile line compared with those in the fertile pepper line (11,000–13,800 U/g). However, the POD enzyme activity levels did not differ significantly in the male sterile and fertile pepper lines in the uninucleate microspore stage. The SOD enzyme activity levels and malondialdehyde contents did not differ significantly in the three stages in the male sterile and fertile pepper lines. In addition, the CAT enzyme activity was significantly lower in the fertile line in the three anther development stages compared with the male sterile pepper line, but especially in the uninucleate microspore stage.

Plant hormone contents in different anther development stages in male sterile and fertile pepper lines

A previous study determined significant differences in the plant endogenous hormone contents between male sterile and fertile lines in different species (Ding et al., 2018; Yang et al., 2020). Thus, in order to determine whether plant hormones are involved in differential anther development, the gibberellin, auxin, abscisic acid (ABA), and jasmonic acid contents were quantified in the male sterile and fertile pepper lines. As shown in Figs. 1U–1X, the gibberellin and jasmonic acid contents were significantly lower in the meiosis stage in the male sterile line compared with the fertile pepper line. In the microspore tetrad stage, the ABA content was slightly higher in the fertile pepper line but the same in the uninucleate microspore stage (Fig. 1W). However, the auxin content was significantly lower in the uninucleate microspore stage in the fertile line compared with the male sterile pepper line (Fig. 1V).

DAPs during anther development

To elucidate the molecular mechanisms related to male sterility, proteomic analysis was conducted to determine the differences between the male sterile and fertile pepper lines. The total proteins were extracted from the sterile and fertile lines in the meiosis, tetrad, and binucleate stages, and subjected to proteomic analysis. The proteomic analysis workflow is shown in Figs. 2A and 2B. The good repeatability of samples indicates that the result data is highly reliable (Fig. 2C). In total, 8,516 proteins and 1,646 DAPs were identified in all of the samples (Fig. 2D and Table S1). The distributions of all DAPs and their overlapping in different developmental stages were visualized by venn diagram analysis. As shown in Fig. 3, only eight DAPs (five upregulated and three downregulated) were identified in three stages (Figs. 3A and 3B). However, the number DAPs of their overlapping in two different developmental stages were no more than 50 in venn diagram (Figs. 3A and 3B). To identify proteins related to male sterility, 362 (213 upregulated and 149 downregulated), 179 (99 upregulated and 80 downregulated), and 1,104 (453 upregulated and 651 downregulated) DAPs were screened based on comparisons of BY_JS (meiosis stage in the male sterile) vs. KY_JS (meiosis stage in the male fertile), BY_SF vs. KY_SF (microspore tetrad stage), and BY_SH vs. KY_SH (binucleate microspore stage) with fold changes of 1.3 and P ≤ 0.05 (Figs. 3C–3E). The number of DAPs was significantly higher in the binucleate microspore stage than that the meiosis and microspore tetrad stages (Fig. 3E).

GO analysis of DAPs

To clarify the functional categories for DAPs in the male sterile and fertile pepper lines, the DAPs were assigned to three categories comprising biological process, cellular component, and molecular function (Fig. 4). Cellular process, metabolic process, and response to stimulus were the most highly enriched terms in all three stages in the biological process category, but especially in the binucleate microspore stage, and the numbers of DAPs assigned to other terms were slightly lower (Fig. 4). Interestingly, more DAPs were mainly assigned to peroxisome and microbody process in the microspore tetrad stage (Fig. 4 and Fig. S1). In the cellular component category, DAPs were significantly enriched in the cell, organelle, and membrane terms (Fig. 4). Many DAPs were assigned to translation, cytosolic ribosome process, and amide biosynthesis in the meiosis stage (Fig. S1). However, less DAPS were assigned to the extracellular region term in the meiosis stage compared with the binucleate microspore and microspore tetrad stages. Catalytic activity and binding were the most highly enriched molecular function terms in all three stages (Fig. 4).

The functional subcategories and metabolic pathways were assigned for DAPs, and the results indicated that DAPs in the meiosis stage were mainly enriched in the protein processing in endoplasmic reticulum, proteomic regulatory subunits, and heat shock proteins pathways (Fig. 5, Table S2). In the microspore tetrad stage, the DAPs were mainly enriched in the monoterpenoid biosynthesis pathway. However, the monoterpenoid biosynthesis and starch and sucrose metabolism pathways were enriched in the binucleate microspore stage (Fig. 5).

Subcellular localizations and domain enrichment analysis for DAPs

We also analyzed the subcellular compartment localizations of the DAPs, i.e., cytoplasm, chloroplast, nucleus, plasma membrane, extracellular region, mitochondrion, vacuolar membrane, and endoplasmic reticulum. As shown in Fig. 6, the DAPs in all three stages were localized in the cytoplasm, chloroplast, and nucleus, which accounted for 20–30% of the total, thereby demonstrating that proteins in these three compartments had important roles in the male sterile and fertile lines. Interestingly, the distributions of the DAPs mainly differed in the chloroplast and cytoplasm. More DAPs were localized in the cytoplasm (29%) in the meiosis stage compared with the microspore tetrad stage (24%) and binucleate microspore stage (26%) (Fig. 6A). However, more DAPs were localized in chloroplasts in the microspore tetrad stage (30%) and binucleate microspore stage (31%) compared with the meiosis stage (25%) (Figs. 6B and 6C). These different distributions indicate that many proteins in chloroplast compartments affected pollen development in the microspore tetrad and binucleate microspore stages.

Protein domain analysis was performed at P < 0.05 using the InterPro database. In the BY_JS vs. KY_JS comparison, the enriched domains for DAPs included alpha/beta hydrolase fold, NB-ARC, and PIC domains (Fig. 6D), whereas most DAPs contained protease inhibitor and trypsin domains in BY_SF vs. KY_SF, and all were upregulated proteins (Fig. 6E). In BY_SH vs. KY_SH, the enriched domains included trypsin and protease inhibitor, PIC, and glycosyl hydrolase domains (Fig. 6F). Interestingly, the DAPs containing trypsin and protease inhibitor domains or glycosyl hydrolase domains were upregulated proteins, whereas the DAPs containing PIC domains were downregulated (Table S2).

PPI networks and cluster analysis

To understand the relationships among DAPs, we determined the PPI networks in different stages using the STRING 11.0 database. For BY_JS vs. KY_JS, 14 upregulated proteins (A0A2G2YPH5, A0A2G3AGR1, A0A1U8FF28, and others) were assigned to a ribosome network, and six upregulated proteins and one downregulated protein were clustered in a proteomic interaction network (Fig. 7A). However, only 19 DAPs comprised one small network with 13 upregulated proteins and six downregulated proteins for BY_SF vs. KY_SF (Fig. 7B). In addition, more than 100 proteins in metabolic pathways, TCA cycle (tricarboxylic acid cycle), ribosome, and basal transcription factor established a large and complex interaction network, which was closely related to plant male sterility for BY_SH vs. KY_SH (Fig. 7C). These PPI networks indicate that anther development is regulated by a complex network.

Based on their similar expression patterns, all of the DAPs identified during anther development were divided into six clusters (Fig. 8). The expression levels of cluster 2 members in the sterile group (BY_JS, BY_SF, and BY_SH) were upregulated compared with those in the fertile group (KY_JS, KY_SF, and KY_SH), but the expression levels of cluster 1 members in the sterile group (BY_JS, BY_SF, and BY_SH) were downpregulated compared with those in the fertile group (KY_JS, KY_SF, and KY_SH) (Fig. 8). Furthermore, these DAPs were mainly related to citrate metabolic process, cold acclimation, regulatory region nucleic acid binding, response to jasmonic acid, microbody, and acid phosphatase activity pathway (Fig. 8 and Fig. S2). Furthermore, the expression levels in cluster 5 were significantly downregulated in KY_SH compared with BY_SH, whereas those of some in cluster 6 were upregulated in KY_SH, where they were involved in response to stress, stimulus, peroxisome, antioxidant activity, detoxification, and other metabolism pathways (Fig. 8 and Fig. S2).

Targeted protein quantification by PRM

To confirm the reliability of the proteome data, PRM was conducted to quantify the targeted proteins. About 15–20 proteins from the KEGG pathway data in three different stages were randomly selected and quantified. Nineteen proteins from 20 selected randomly in the meiosis stage (BY_JS vs. KY_JS) had quantitative information and 1.5-fold changes occurred in all of these proteins, where 13 were downregulated and six were upregulated (Fig. 9A). Due to the properties of the proteins and their abundances, six proteins (four downregulated and two upregulated) were randomly selected from 15 proteins identified in the microspore tetrad stage (BY_SF vs. KY_SF) (Fig. 9B). Interestingly, three downregulated proteins with lipid-binding domains comprising A0A2G2YY91, A0A2G3AC90, and A0A1U8E7C6 may function in lipid transport during anther development. In addition, 20 proteins (10 downregulated and 10 upregulated) among 22 randomly selected proteins identified as proteins related to glycometabolism, including A0A2G2ZTG5, A0A1U8F8K6, and A0A1U8HAH9, were downregulated in the binucleate microspore stage in the male fertile pepper line (Fig. 9C). The PRM results indicate that the DAPs were stable and accurately quantified.