Introduction

Cotton stands out as both a valuable fiber crop and an oil-producing crop on a global scale. For many years, China has been the major cotton producer and consumer in the world (Li 2017). The male sterility (MS) system is widely used to facilitate the utilization of heterosis in major crops (Budar and Pelletier 2001), and also has important practical significance in hybrid cotton seed production (Kubo et al. 2011). Most of the current commercial cotton hybrids have been produced by manual emasculation and pollination in China (Yang 2018) and India (Singh et al. 2002; Santhy et al. 2008; http://www.cicr.org.in/). However, due to increasing labor costs and higher requirements for seed purity, manual emasculation has become increasingly inadequate to meet the needs of hybrid production. MS has gradually become the research hotspot in the cotton heterosis utilization field and has been used for cotton hybrid production (Xing et al. 2017; Yang et al. 2018; Wu et al. 2017). In recent years, with the successful construction of cotton sterile, maintainer and restorer lines, the “three-line” cross breeding system has been successfully used in cotton research and breeding (Zhang et al. 2012; Cui et al. 2014; Horn et al. 2014; Wu et al. 2011, 2014, 2017; Kong et al. 2017). However, with just a single source of MS lines and a limited number of restorer lines with high combining ability, the application of the “three-line” cross breeding system is greatly limited (Wang et al. 2010a, b). Cotton heterosis involves polyploid heterosis and hybrid heterosis (hybrid vigor) in hybrid breeding (Xing et al. 2017). The complex allotetraploid characteristic of the cotton genome makes genetic, genomic and functional analysis extremely challenging, and it is difficult to dig up functional genes related to fertility and to create novel male sterile lines.

Germin-like proteins (GLPs) were firstly identified in germinated wheat and worked as multifunctional enzymes, such as superoxide dismutase (SOD), oxalate oxidase (OXO) and ADP glucose pyrophosphatase or phosphodiesterase (AGPPase) (Wang et al. 2013; Pei et al. 2019). Phylogenetic analysis classifies germins and GLPs into five sub-families including true germin subfamily (wheat and barley germins), subfamily 1, 2, 3 and gymnosperm germin protein. Among these, GLP subfamily 1, 2 and gymnosperm germin protein subfamily all had SOD activity (Li et al. 2010).

Reactive oxygen species (ROS) produced by the NADPH oxidase, such as superoxide anion radical (O2−) and hydrogen peroxide (H2O2), are essential in several types of signal transduction in plants (Yamauchi et al. 2017; Mittler et al. 2011). ROS signaling plays a key role in the acclimation of plants to low-oxygen conditions (Bailey-Serres and Voesenek 2008; Sauter, 2013; Steffens et al. 2013; Pucciariello and Perata 2017), such as lysigenous aerenchyma formation in maize roots (Rajhi et al. 2011) and wheat roots (Yamauchi et al. 2014), and also in rice shoots (Colmer and Pedersen 2008; Parlanti et al. 2011; Steffens et al. 2011; Yamauchi et al. 2017). Rboh (Respiratory burst oxidase homolog), a plant homolog of mammalian NADPH oxidase, converting O2–O2− (Torres and Dangl 2005), constitutes a multi-gene family (Suzuki et al. 2011). The Arabidopsis and rice genomes have ten and nine Rboh genes, respectively (Torres and Dangl 2005; Wong et al. 2007), and each homolog has a specific role in a broad range of biological processes (Marino et al. 2012). The decrease of a pollen-specific NADPH oxidase in tobacco (NtNOX) resulted in a lower NOX activity, reduced tip-localized ROS production and inhibited pollen tube growth (Potocký et al. 2007). Deficiency of Rboh genes has been implicated in reduction of total ROS levels in NbRboh-silenced plants, which caused the reduction of positive signal (H2O2) for HR by a scavenging effect of NO on O2− and resulted in an imbalance of those molecules (Yoshioka et al. 2003).

The members of GLP superfamily are ubiquitously expressed in various plant species and play important roles in plant development and plant responses to biotic and abiotic stresses (Lu et al. 2010). In Arachis hypogaea, Citrus limon and prunus salicina, GLPs were found to be highly expressed in flowers and fruits, and were involved in flower and fruit development (Wang et al. 2013; Bruno et al. 2014; Sharkawy et al. 2010). GLP genes also have important roles in abiotic stress response (Gucciardo et al. 2007; Yu et al. 2017; Fu et al. 2018) as well as in fungal infection resistance (Zimmermann et al. 2006; Rietz et al. 2012; Banerjee et al. 2017). A number of GLP genes respond to stresses by regulating their SOD, OXO and AGPase activity (Wang et al. 2013).

In our previous work, a genic male sterile mutant (ms1) was found in the Coker 312 mutation lines (Gossypium hirsutum L.), which has the same phenotype and genetic background except for male sterility (Fu et al. 2015). cDNA-AFLP and RNA sequencing were performed to screen the differentially expressed genes (DEGs) related to male sterility between C312 and ms1. Further study confirmed that GhACNAT gene expressed in the reproductive organs influenced cotton male fertility by regulating lipid metabolism and jasmonic acid biogenesis (Fu et al. 2015). To get a more comprehensive understanding of the male sterile mechanism in ms1, other significantly down-regulated genes in the developing anthers of ms1 were cloned and further studied in this paper. GhGLP4 (Gh_A13G0945), belonging to the germin-like protein subfamily 2 (GLP4), was predicted to be involved in the anther development and fertility. GhGLP4-silenced plants were constructed by RNA interference, and the reproductive organs of GhGLP4 interfered lines were found abnormally developed. The SOD activity and H2O2 content were analyzed in these lines to further investigate the mechanism leading to pollen abortion. The results from this study provide a foundation for further studies of the molecular mechanisms of fertility and the creation of novel male sterile lines in cotton.

Methods and methods

Plants materials

The seeds of G. hirsutum L. cv. Coker 312 (C312) were preserved in our laboratory. The genic male sterile mutant (ms1) in C312 background was bred in our laboratory from the C312 mutation lines (Fu et al. 2015). Both C312 and ms1 were planted in the experimental field of Zhejiang Sci-Tech University (Xiasha Campus, Hangzhou, Zhejiang). Vegetative organs (including roots, stems and leaves), reproductive organs at the day of flowering (including petals, bracts, stamens, stigmas and ovules) and anthers at different developmental stages of the flower buds (Fig. 1a) from both lines were collected for total RNA extraction. All samples were snap-frozen in liquid nitrogen immediately and stored in −80℃ ultra-low-temperature refrigerator. Three biological replicates were performed for each interval and each biologic replicate consisted of five buds for anther collection from different individual plants.

Fig. 1
figure 1

The expression level of GhGLP4 gene in developing anthers and other tissues. a The reproductive organs at the early, middle and late anther stages in C312 and ms1 (G. hirsutum L.). Early buds: flower buds length were 2–5 mm; middle buds: flower buds length were 15–18 mm; late buds: the day before and the day of flowering with 35–40 mm in length. The length was measured from the bract axils to the top of the bud; left: C312; right: ms1; bar = 5 mm. b Expression pattern of GhGLP4 gene in C312. The expression level of GhGLP4 gene in leaf was set as 1, and the relative expression in other tissues was compared with that in leaves. c The expression level of GhGLP4 gene in anthers at different development stages in C312 and ms1. The relative expression was compared with that in EB of C312. EB early buds; MB middle buds; LB late buds (P < 0.05, P < 0.01 and P < 0.001, Student’s t tests)

Full size image

Gene cloning and construction of VIGS vectors

The GhGLP4 gene was screened in the cotton genomes (http://www.cottongen.org) for gene accesses and sequences. The conserved domain of GhGLP4 protein was predicted by CD-Search software (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi). GenDoc (https://github.com/karlnicholas/GeneDoc) was used for multiple sequence alignment with the amino acid sequences of different GLP genes. The phylogenetic tree was constructed in MEGAX with adjacent algorithms, using bootstrap values performed on 1000 replicates.

A 433 bp fragment of GhGLP4 was amplified by PCR and the VIGS vectors of pCLCrV were constructed with the method described previously (Gu et al. 2014; Fu et al. 2015; Gao et al. 2019). All primers were listed in supplemental files (Additional file 3: Table S1). The pCLCrVA-GhGLP4 + pCLCrVB was used for target gene silencing; the pCLCrVA-empty + pCLCrVB and pCLCrV-GhPDS + pCLCrVB were used as the negative control and positive control, respectively (Fu et al. 2015). The vectors with or without target fragments were transformed into Agrobacterium tumefaciens GV3101 (Fu et al. 2015; Gao et al. 2019).

Growth conditions and agroinfiltration

C312 seedlings were grown in a growth chamber and greenhouse at 28 °C with a 14 h light and 10 h dark cycle. 2-week-old healthy seedlings were infected with different Agrobacteria carrying pCLCrVA or one of its derivatives and pCLCrVB (Gu et al. 2014; Fu et al. 2015; Gao et al. 2019). The pCLCrVA-GhGLP4 transgenic line was labeled as GhGLP4i with consecutive numbers, and the positive control pCLCrV-GhPDS was labeled as GhPDSi, the pCLCrVA-empty was labeled as negative control, and the wild-type C312 was labeled as WT (Fig. S1). The growth chambers and the greenhouse for the infected plants were kept at 22–23 °C, 14/10 h light/dark cycle. About 30 days after injection, the leaves of agro-infected lines were collected for DNA extraction to detect the positive lines by PCR with vector primers. Total DNA was, respectively, extracted from the leaves of WT, GhGLP4i, GhPDSi and control plants using a modified CTAB (cetyltriethylammnonium bromide) DNA extraction method (Fu et al. 2015; Huang et al. 2009). PCR analysis was then conducted using specific primers for A-F/A-R, A-F/V-GhGLP4-R and A-R/V-GhGLP4-F. (The primers were listed in Table S1).

RNA extraction and cDNA synthesis

Total RNA was extracted with RNA extraction kit according to the manufacturer’s instructions (RNAprep Pure TIANGEN BIOTECH, China). The extracted RNA sample was measured using a BioDrop DUO (BioDrop Ltd, UK). About 200 ng RNA of each sample was subjected to double-stranded cDNA synthesis using ReverTra Ace® qPCR RT Master Mix with gDNA Remover (TOYOBO Co., Ltd, Japan) according to the manufacturer’s protocol. The synthesized cDNA was diluted 20 times and used for quantitative real-time PCR (qRT-PCR) analysis.

Gene expression analysis by quantitative real-time PCR

qRT-PCR was performed using NovoStart® SYBR qPCR SuperMix Plus (Novoprotein Scientific Inc., Shanghai, China) with ABI QS3 fluorescence quantitative PCR instrument (ABI, USA). Transcript levels of GhRboh genes (Gh_D07G0928, Gh_D12G1932, Gh_A05G2211, Gh_D05G2471 and Gh_A07G0856) were checked to evaluate the SOD activity in C312 and GhGLP4i lines. Primers for qRT-PCR were designed using Primer Express 6.0 (Premier Biosoft, Palo Alto, CA) (Table S1). The cotton Ubiquitin7 gene (GhUBQ7, Accession number: DQ116441) was used as an internal control for the assays. The expression level of target gene in cotton was obtained and standardized to the constitutive GhUBQ7 gene expression level. The results were obtained from three biological repeats and two technical duplications, and the statistical significance of the differences was determined using the Student’s t test in Graphpad Prism 8 (version 8.0.2). Differences between treatments were considered significant when *P < 0.05, **P < 0.01 and ***P < 0.001 in a two-tailed analysis.

Pollen viability determination

Pollen grains from the opening flowers of C312 and GhGLP4i lines were stained with 1% (w/v) 2, 3, 5-triphenylterazolium chloride (TTC) (Solarbio, Beijing, China) for pollen viability examination. TTC powder (0.1 g) was dissolved in 0.1 M phosphate buffer (pH = 7.0, without NaCl) to prepare 1% TTC solution. Flowers were collected on the day of blooming, and the pollen grains were immediately immersed in TTC solution, and placed in a 34 °C water bath in the dark for 30 min. Images of pollen grains were captured under a bright filed using Olympus IX73 fluorescent inverted microscope (Olympus, Japan). The viable pollen appeared red; the apoptotic and abortive pollen appeared colorless.

SOD activity and H2O2 content measurement

Measurement of SOD activity was performed following the adapted NBT method (Li, 2000). The SOD reaction mixture included 0.05 mol/L phosphate buffer (PBS, pH = 7.8), 14.5 mM methionine solution, 30 μM methylenediamine tetraacetic acid disodium salt solution, 60 μM riboflavin solution and 2.25 mM nitroblue tetrazolium chloride. The anther samples from C312 and GhGLP4i lines, respectively, were mixed and grinded sufficiently and quickly with 2 mL PBS on the pre-cooled mortar, and then centrifuged at 12,000 rpm for 20 min at 4 °C. The supernatant was kept as the enzyme solution. 3 mL SOD reaction mixture and 10 μL enzyme solution were left to react for 20 min under 4000 lx light.

The H2O2 content was measured following the adapted H2O2 Quantitative Assay Kit (Water-Compatible) (Sangon Biotech, Shanghai, China). The anther samples from C312 and GhGLP4i lines, respectively, were mixed and grinded sufficiently and quickly with liquid nitrogen; then, the powders (approx. 0.1 g) were soaked into a 1.5-mL pre-cooled acetone, and shook in a low-temperature shaker for 20 min. The extraction buffers were centrifuged at 12,000 rpm for 15 min at 4 °C.

The SOD enzyme mixed solution and H2O2 supernatant were assayed spectrophotometrically to calculate the absorbance units (A560) per gram fresh weight. A spectrophotometer (UV-2600, Shimadzu, Japan) was used for the absorbance measurements at 560 nm. The results were determined based on SOD activity calculation formula (Li, 2000), and the H2O2 content was determined based on the H2O2 standard curve line. All data presented from at least three independent experiments with three replicates each, and the values were expressed as the mean ± standard deviation (SD). The statistical significance of the differences was determined using the Student’s t test. The figures and the linear regression curve with Pearson correlation coefficient were made by Graphpad Prism 8 (version 8.0.2). Differences between treatments were considered significant when *P < 0.05, **P < 0.01, and ***P < 0.001 in a two-tailed analysis.

Results

Expression analysis of GhGLP4 in C312 and ms1

GhGLP4 gene was significantly differentially expressed in the tested tissues and organs in C312 and its expression level in the reproductive organs such as petals, stigmas and stamens was significantly higher than that in the vegetative organs including roots, stems and leaves. Among the tested reproductive organs, GhGLP4 gene was especially extra-predominantly expressed in stamens, but expressed at very low level in the ovule (Fig. 1b). The expression level of GhGLP4 gene increased significantly in the middle and late stages compared to the early stage of anther development in C312. In ms1, the expression level of GhGLP4 gene had no considerable change during the anther development, whereas, in C312, the expression level of GhGLP4 gene sharply declined in the middle and late anthers in ms1 (Fig. 1c).

Identification and phylogenetic analysis of GhGLP4 gene

GhGLP4 (Germin-like protein) has a conserved cupin OxOx domain from the 22nd to 213th amino acids and has oxalate oxidase activity, belonging to the Cupin superfamily (Fig. 2a). The homologs of GhGLP4 in different species have the same structural characteristics in the GLP family, containing a conserved β-barrel protein with a metal ion binging ability. The full-length cDNA with 654 bp of GhGLP4 was cloned from C312. The GhGLP4 was highly conserved in Gossypium (Fig. 2b), and one GhGLP4 gene was found to be locate in the AT sub-genome of G. hirsutum and two homologous genes of GhGLP4 in the AT and DT sub-genome of G. barbadense (Fig. 2c, d).

Fig. 2
figure 2

Phylogenetic analysis of the GhGLP4 family. a Conserved domain of GhGLP4. b GhGLP4 shared high identity across different species. c The phylogenetic tree of the GLP superfamily. d Location distribution of GLP4 on different chromosomes of G. hirsutum (Gh_A13G0945, GhGLP4) and G. barbadense (Gb_A13G012740, Gb_D13G012420, GhGLP4 homologous genes). Brassica napus (P46271.1); Prunus persica (Q9ZRA4.1); Citrus limon (JQ941709.1); Hordeum vulgare (DQ647623.1); Thevetia peruviana (C0HJY2.1); Triticum aestivum (P15290.1); Oryza sativa (Q7XZY1.1); Arabidopsis thaliana (P94014.2); Pisum sativum (Q9S8P4.2); Vitis vinifera (NP 001,267,914.1); ziziphus jujuba (XP 024,926,108.1); G. arboreum (XP 017,618,973.1); G. raimondii (XP 012,461,204.1); G. barbadense (PPE01127.1); G. hirsutum (XP 016,675,454.1); G. barbadense (KAB2048717.1)

Full size image

Stamen abnormal development in GhGLP4i lines

A total of 19 individual plants were identified as positive GhGLP4 interference lines by PCR analysis with specific primers (Fig. S1, S2). During the flower bud development, the floral organs of GhGLP4i lines appeared atypical and the stigmas towered over the petals (Fig. 3a). On the flowering day, GhGLP4i lines had smaller flowers, the stamen filaments became shorter, the number of stamens decreased significantly, pollen sacs did not crack, and no pollens were released from the closed anther (Fig. 3b). Results from the TTC staining showed that the anthers of C312 were typically dehiscent to release normal pollen grains, whereas the anthers of GhGLP4i lines were almost indehiscent with no pollen scattering; the anthers in some GhGLP4i line were squashed and could release few pollens in different sizes without vigor (Fig. 3b). The stigmas were obviously elongated, and significantly longer than the stamens (Fig. 3c). The expression of GhGLP4 in the stamens of GhGLP4i lines was significantly down-regulated to 3.4–17.5% of that in WT plants (Fig. 3d). The self-pollinated GhGLP4i lines showed flowers withering, ovule browning, boll falling, and no cotton bolls setting (Fig. 3e); following pollination of the GhGLP4i plants with C312 pollens, normal bolls were harvested (Fig. 3f), indicating that the down-regulation of GhGLP4 gene influenced the development of stamen, but did not affect the development of stigma and hybrid seed set.

Fig. 3
figure 3

Phenotypic comparison of the reproductive organs and GhGLP4 expression analysis between WT and GhGLP4i lines. a Stigma exertion in the early flower buds of GhGLP4i lines. b Phenotypes of flower, stamen, pollen sac dispersion, pollen vigor stained by TTC on the day of flowing in GhGLP4i lines and WT. c Phenotype of stigmas and stamens in GhGLP4i lines. d The expression level of GhGLP4 gene in stamens in GhGLP4i lines and WT (P < 0.05, P < 0.01 and P < 0.001, Student’s t tests). The GhGLP4 relative expression was compared with that in WT. e and f Bolls from self-pollination (e) and cross pollination with WT flowers (f) in GhGLP4i lines. The red circles indicated the browning petiole of falling boll

Full size image

SOD activity and H2O2 content changed in GhGLP4i lines

To measure the SOD activity and H2O2 content in the anthers of GhGLP4i lines and WT, a H2O2 standard curve line was built (Fig. S3). According to the standard curve line, seven individual GhGLP4i lines were measured and calculated. The results showed that the SOD activity significantly decreased (Fig. 4a), and the H2O2 content had increased sharply in GhGLP4i lines compared to WT (Fig. 4b). The SOD activity and H2O2 content were further measured in the other 12 GhGLP4i lines and 6 WT plants; the significantly negative correlation (r =  −0.6878) existed between SOD activity and H2O2 content (Fig. 4c, d). The results indicated that the ROS scavenging ability in GhGLP4i lines was impaired, and the rapid release of ROS and the high H2O2 content disrupted the ROS homeostasis which influenced the normal development of stamens in GhGLP4i lines, leading to plant reproductive organ abortion.

Fig. 4
figure 4

SOD activity, H2O2 contents and the correlation analysis between them in the anthers of WT and GhGLP4i lines. a SOD activity in the anthers of WT and GhGLP4i lines, b H2O2 contents in the anthers of WT and GhGLP4i lines, c SOD activity and H2O2 contents further measured in other WT and GhGLP4i lines, d the negative correlation between SOD activity and H2O2 in the anthers of WT and GhGLP4i lines

Full size image

To understand the mechanism of the disruption of ROS homeostasis in anthers of GhGLP4i, the expression of GhRboh genes that are responsible for the ROS generation was measured in anthers of WT and GhGLP4i lines. As a result, two of the GhRboh genes in G. hirsutum, Gh_A07G0586 and Gh_D07G0928 were dramatically down-regulated, and another two GhRboh genes Gh_D05G2471 and Gh_D05G2211 were significantly up-regulated in GhGLP4i lines (Fig. 5). This result indicated that each GhRboh gene may play a different role in maintaining ROS homeostasis and the silencing of the GhGLP4 gene may change the GhRboh gene expression and produce synergistic effects on the homeostasis of ROS, causing the SOD activity to decrease and the H2O2 content to increase in the anthers, and ultimately result in anther abortion.

Fig. 5
figure 5

Relative expression of GhRboh genes in WT and GhGLP4i lines. The relative expression of Gh_A07G0586 (a), Gh_D07G0928 (b), Gh_D05G2471 (c) and Gh_D05G2211 (d) in GhGLP4i line was compared with that in WT. The expression level of Gh_D05G2211 gene in WT was set as 1

Full size image

Discussion

GLPs have been reported to play critical roles in plant development and defense in many plant species, such as peanut (Wang et al. 2013), lemon (Bruno et al. 2014), plum (Sharkawy et al. 2010), barely (Zimmermann et al. 2006), pea (Gucciardo et al. 2007) and flax (Yu et al. 2017). In upland cotton, GhGLP1 was the only identified GLP family member, which expressed specifically in fiber cell to promote fiber expansion (Kim et al. 2004). In this study, we concluded that the GhGLP4 gene in G. hirsutum was essential for stamen development, indicating each member of GhGLP family may have its distinct expression pattern and function during cotton development. A similar phenomenon is observed in other species, e.g., in Camellia sinensi, CsGLP1 and CsGLP2 showed much stronger responses to herbivore-induced damage than other GLPs, such as CsGLP4 and CsGLP7 which were almost insensitive to phenological changes (Fu et al. 2018). In Arachis hypogaea, AhGLP4 and AhGLP8 expressed with high level in flower buds, while AhGLP6 and AhGLP7 expressed at low levels in flower buds and flowers (Wang et al. 2013).

It was proposed that the mechanism of GLPs regulating plant defense responses is associated to their enzyme activity of OXO and SOD, which can balance H2O2 content to influence plant defense (Wang et al. 2013). In plant cells, the mitochondrial electron transport chain (ETC) is a major site of ROS production and excessively accumulated ROS can cause oxidative damage to various biological molecules (Moller, 2001; Wan et al. 2007). The H2O2 produced by cell metabolism has a direct harmful effect on plants and induces damaged cell apoptosis (Wang et al. 2010a, b). Pollen is a non-photosynthetic tissue, and its formation is an extremely energy-intensive process (Ji et al. 2013). ROS homeostasis is essential for normal anther and pollen development, whereas excessive accumulation of ROS in anthers leads to cell apoptosis and male sterility (Yang et al. 2018). In CMS-D2 line of G. hirsutum, the disrupted DNA methylation broke the homeostasis of ROS production and ATP synthesis in mitochondria, triggering a burst of ROS that was transferred to the nucleus to initiate programmed cell death (PCD) prematurely and leading to microspore abortion (Zhang et al. 2019). In this study, the low expression of GhGLP4 reduced SOD activity in anthers and changed the expression levels of GhRboh genes, which disrupted the homeostasis of ROS production and resulted in high levels of H2O2 content. The disrupted ROS homeostasis in the anthers then initiated PCD and caused pollen abortion. The analysis of GhGLP4 gene may provide a foundation to analyze the functions of other members of the GhGLP family; moreover, it makes a tremendous contribution toward creating new male sterile germplasm resources.

Conclusions

GhGLP4 gene was predominantly expressed in the reproductive organ and maintained high levels of SOD activity to decrease the concentration of H2O2 so as to preserve the normal pollen development.