IntroductionTop

Drought has become a global problem that severely limits plant growth and crop productivity and threatens world food security (Li et al., 2014; Chmielewska et al., 2016; Wu et al., 2016). This phenomenon has exacerbated global climate change (Abedi & Pakniyat, 2010; Taniguchi, 2016; Duran-Encalada et al., 2017). According to the 4th Intergovernmental Panel on Climate Change (IPCC) report (Alley et al., 2007), Earth is constantly warming, and the drought is gradually becoming a complicated issue; it has been reported that drought can result in 80% yield loss of the common bean (Zadraznik et al., 2017). Food security is the foundation of human survival and social stability (Wang et al., 2013), and China is a country with a huge population and heavily dependent on agriculture. Previous study has demonstrated that precipitation has been on the decline since the 1980s in North China (Ma et al., 2005), and this finding has prompted us to begin to actively respond to the impact of drought.

A clear understanding of the molecular mechanisms of plant response to drought is essential for plant biolo­gy. Knowledge about these mechanisms is also crucial for the continued development of rational breeding and transgenic strategies to improve stress tolerance in crops (Cui et al., 2005). Therefore, overcoming the influence of drought on agricultural production has been the focus of research. Drought resistance is complex, and many proteins related to various metabolic processes are involved in the process (Ge et al., 2012). So far, changes in physicochemical and molecular mechanisms come down to plants under drought stress have been reported. Proteins participating in chlorophyll biosynthesis (Wu et al., 2016), nitrogen metabolism (Chmielewska et al., 2016), defense and stress (Zadraznik et al., 2017), storage starch biosynthesis (Ganeshan et al., 2010), transcription and translation (Cao et al., 2017) play a crucial role in drought response. In addition, previous studies have found that grain protein content, glutenin content and the ratio of glutelin to gliadin increased greatly under stress conditions (Lan et al., 2004; Lu & Lu, 2013). Therefore, research on proteomic changes involved in drought stress plays an important role in understanding the mechanism of drought in plants.

Foxtail millet (Setaria italica (L.) Beau), an ancient crop that originated from China in 6000 BC, has gradually become an important food and fodder grain crop in arid and semi-arid regions in Asia, especially in China and India (Bettinger et al., 2010; Wang et al., 2016), because of its abundant proteins, dietary fibres, phenolics, flavonoids and minerals in its grain (Liu et al., 2017) and its high tolerance to natural abiotic stress (Yadav et al., 2016). Foxtail millet can be used as a model crop for drought resistance research because of its excellent performance in drought tolerance. It is characterized by a small genome, low amount of repetitive DNA, inbreeding nature and short life cycle (Lata et al., 2013; Yi et al., 2015). Therefore, foxtail millet plays a crucial role in plant research. Foxtail millet proteins are known to be involved in several NaCl stress-related progresses, such as signal transduction, photosynthesis, cell wall biogenesis, stress related and several metabolisms, including energy, lipid, nitrogen, carbohydrate and nucleotide metabolisms (Veera­na­gamallaiah et al., 2008). Previous study also found that SiLEA14, a novel atypical late embryogenesis abun­dant (LEA) protein, plays important roles in resistance to abiotic stresses in foxtail millet (Wang et al., 2014). To the best of our knowledge, the response of foxtail millet grains to drought stress at a proteomic level has yet to be reported. Therefore, investigating the respon­se mechanisms of foxtail millet grain proteins under drought stress offers great theoretical and practical importance.

Material and methodsTop

Plant materials and experimental design

‘Huangjinmiao’, a foxtail millet landrace widely cul­­tivated in Northwestern China and characterized by high drought resistance and excellent quality, was used as the experimental material. The seeds used in the experiment were provided by the Chifeng Academy of Agriculture and Animal Husbandry, Chifeng, Inner Mongolia Autonomous Region, China. Experiments were conducted in Dunhuang, Gansu Province, China (92°13′-95°30′E, 39°40′-41°40′N, altitude 1139 m) in 2014. The area is characterized by a typical warm temperate arid climate with the following conditions: annual average precipitation of approximately 45.4 mm, annual evaporation of 2486 mm, dryness of 19.6%, average annual temperature of 10.2 °C, frost-free pe­riod of 142 days, and annual sunshine duration of 3246.7 h. In the experimental site, 0–40 cm soil sho­wed a loess-like loam texture with 8.60 g kg-1 organic carbon, 20.83 mg kg-1 available nitrogen, 40.40 mg kg-1 available phosphorus and 112.79 mg kg-1 availa­ble potassium. The pH value of the soil was 7.90.

The study involved a completely randomized block design with three replications and two treatments, control (Ctrl) and drought stress (DS) treatment. Impe­rvious belts of at least 1 m wide were set between treatments. Foxtail millet ‘Huangjinmiao’ was sown on April 18, 2014, and harvested on August 28, 2014. A sufficient amount of water was poured before the test to ensure that the seed germinated normally. For the control group, foxtail millet was irrigated once at each of the main growth stages of jointing stage, heading stage and filling stage. For drought stress treatment, the crops were not irrigated during the entire growing stages, and the source of water was completely dependent on soil moisture and precipitation. The temperature and precipitation amount during the foxtail millet growth period of 2014 were shown in Fig. 1. In the experiment, the fertilization was performed before sowing by using 354 kg ha−1 (received 165 kg N ha−1) urea, 321.8 kg ha−1 (received 195 kg P2O5 ha−1) calcium superphosphate and 194.4 kg ha−1 (received 105 kg K2O ha−1) potassium sulfate. During the growth period, no additional fertilizers were applied. During the maturation period of foxtail millet, the yield and yield-related indexes such as panicle weight per plant, grain weight per plant, stem and leaf weight per plant, the number of effective panicle and 1000-grain weight were investigated with three replications. The harvested mature grains were used for further experiments.

Figure 1. Monthly precipitation and average temperatures during the foxtail millet growth period in 2014. The bar chart represents the precipitation of each month, and the line chart illustrates the average temperature.

Extraction and identification of protein compo­nents in foxtail millet grains

The protein components in the mature seeds of foxtail millet were extracted in accordance with the method of Agboola et al. (2005) with some modifications. Defatted grain powders (approx. 3 g) were added to a 50 mL centrifuge tube that was filled with water up to the 50 mL mark. The centrifuge tubes were placed on a table concentrator, shaken at room temperature for 3 h and centrifuged at 4500 rpm and at 4˚C for 40 min. Approximately 30 mL of the supernatant was considered as the test solution, then we added, following this order, 5% NaCI solution, 70% ethanol and NaOH solution, 1% sodium dodecyl sulfate (SDS) solution and 1%SDS+2% 2-mercapto-ethanol (2-ME) solution into the precipitate. Albumin, globulin, gliadin, glutelin, 1% SDS soluble protein and 1% SDS+2% 2-ME soluble protein were extracted by repeating the above procedures. The contents of protein components were determined by the Kjeldahl method (Kjeltec 8400 type Kjeldahl apparatus, FOSS Company, Denmark). The coefficient 6.25 was used for crude protein calculation (Yu et al., 2007). Each experiment was performed in triplicate.

Extraction and quantification of total proteins

The total proteins in the mature seeds of foxtail millet were extracted in accordance with the methods of Wang et al. (2007) and Cui (2012). Approximately 1 g foxtail millet grains was weighted and then rapidly ground into powder after liquid nitrogen was added. The powder was transferred into a 10 mL centrifuge tube and we added five volumes of precooled trichloroacetic acid (TCA) extract containing 10% TCA-acetone, 0.07%-0.1% DL-dithiothreitol (DTT) and 0.5% beta-mercaptoethanol. The solution was mixed thoroughly by shaking, and then placed in a refrigerator at -20 ˚C overnight. The solution was centrifuged at 13,000 rpm and at 4 ˚C for 30 min and the supernatant was discar­ded. The precipitate was washed three times with pre-cooled acetone containing 0.07%-0.1% DTT and dried for 45 min under vacuum. Then we added to the dried powder 5 mL of protein extract solution (containing 0.5 mol·L-1 pH 7.5 tris-HCl, 30% sucrose, 50 mmol

L-1 EDTA-Na2, 2% SDS, 2% β-mercaptoethanol and 100 mmol L-1 KCl) and 5 mL of pH 7.8-8.0 tris-satura­ted phenol. The solution was bathed for 5 min after it was blended in warm water (50 ˚C), and centrifuged for 20 min; 1 mL of phenolic layer was pipetted into a 10 mL centrifuge tube, then we added 5 volumes of ammonium acetate solution containing 0.1% β-mercaptoethanol; and tubes were placed in a refrigerator -80 ˚C overnight. The solution was centrifuged at 12,000 g for 10 min at 4 ˚C and the supernatant was discarded. The precipitate was washed three times with pre-cooled methanol containing 0.1% β-mercaptoethanol, and stored in a refrigerator at -80˚C.

We added to the treated samples a hydrated solution consisting of 7 M urea, 2 M thiourea, 4% 3-[(3-cholamidopropyl) dimethylammonio] propane­sulfonate (CHAPS), 0.2% immobilized pH gradient (IPG) buffer and 65 mM DTT and allowed to crack in a water bath at 30 ˚C for 2 h. The samples were then centrifuged for 20 min, and the supernatant was pipet­ted into a cuvette to determine the protein concentra­tion. Protein quantification was conducted via Bradford’s method. A standard curve of bovine serum albumin was made, and absorbance was determined at 595 nm by using a DU800 spectrophotometer (Beckman Com­pany, USA).

Two-dimensional electrophoresis (2-DE), image analysis and matrix-assisted laser desorption/ionization-tandem time-of-flight (MALDI-TOF/TOF) analysis

For the first dimension of electrophoresis, 0.001% bromophenol blue was added to 300 μL of 800 μg of hydrolyzed protein samples in a hydrated plate. The 17 cm IPG preformed strips were allowed to equilibrate to room temperature for 10 min and placed in a hydra­tion plate surface down to avoid bubble forma­tion. Subsequently, 5 mL of vegetable oil was ad­ded. Isoelectric focusing (IEF) was performed by applying a voltage of 50 V for 14 h, 250 V for 30 min, 1000 V for 16 h and 8500 V for 5 h. A voltage of 10,000 V was maintained until a total of 80 kVh was reached. A voltage of 500 V was then maintained for 24 h. The second dimension of SDS-PAGE was performed immediately after IEF, and the strips were removed, balanced for 15 min in mother liquor twice, and placed above 11.5% gel. The strips were sealed with low-capacity and low-sodium agarose at 10 mA voltage for 40 min and set to 30 mA towards end.

After 2-DE was completed, the gel was stained with Coomassie brilliant blue (CBB) G-250 (Candiano et al., 2004) and de-stained with double-distilled water. Each gel was then scanned using a UMAX PowerLook 2100XL image scanner (UMAX Systems GmbH, Willich, Germany) at a resolution of 300 dpi. The trans­mission mode was selected, and the picture was set to TIFF format. Image analysis was performed using PDQuest 8.0.1 (Bio-Rad Laboratories, Hercules, CA, USA), including automatic protein spot detec­tion and matching, background elimination, protein spot manual editing and matching and protein spot abundance uniformity. Differential protein spots were subjected to qualitative and quantitative analysis. The number of differentially abundant protein spots (DAPs) should satisfy the t-test of 0.05 level and the multiple of change was more than 1.5 times including induced and missing protein spots simultaneously. The DAPs were manually excised from the gels and analyzed by Hoogen Biotech Co., Ltd., Shanghai, China.

Mass spectrometry data was analyzed by GPS 3.6 (Applied Biosystems) and Mascot 2.1 (Matrix Science). The detailed parameter settings were as follows: Database, NCBI; taxonomy, Viridiplantae (900091); type of search, peptide mass fingerprint (MS/MS Ion Search); enzyme, trypsin; fixed modifications, carbamidomethyl (C); mass values, monoisotopic; protein mass, unrestricted; peptide mass tolerance, ± 100 ppm; fragment mass tolerance, ± 0.4 Da; peptide char­ge state, 1+; maximum missed cleavages, 1; pro­tein score confidence interval (CI%) ≥ 95.

Statistical analysis

Statistical analysis was performed in accordance with the method of Wu et al. (2015) with some modifications. The figures used in this article were drawn using Microsoft Office Excel 2016 coupled with OriginPro 2016 (OriginLab Corporation, Nor­thampton, MA, USA). Comparisons were statistically evaluated ANOVA in SPSS 23.0 (SPSS Science, Chicago, IL, USA). Significance was set at p<0.05 or p<0.01.

ResultsTop

Changes in the indexes related to the yield of foxtail millet under drought stress

After the foxtail millet matured, the changes in the yield and yield-related indexes under Ctrl and DS treatments were investigated (Table 1). The results indicated that yield, panicle weight per plant, grain weight per plant, and 1000-grain weight greatly decreased (p<0.01) under DS and the number of effective panicle significantly decreased (p<0.05). The stem and leaf weight per plant also decreased but this decrease was not significant under DS.

Changes in protein components in foxtail millet grains under drought stress

The contents of four protein components (albu­min, globulin, prolamin and glutelin) under Ctrl and DS conditions are shown in Fig. 2. The results indicate that the albumin content greatly decreased (p<0.01) under DS, whereas the gliadin content considerably increased (p<0.01). The contents of glutelin and globulin also in­creased but this increase was not significant under DS.

Figure 2. Content of protein components in foxtail millet grains under control (Ctrl) and drought stress (DS) treatment. Error bars represent standard deviation (n = 3). The ** above the histogram indicate statistical significance at 0.01 (p<0.01) according to Duncan’s test.

Identification of differentially abundant proteins under drought stress

The total protein of the matured seeds was extrac­ted and subjected to 2-DE analysis with three indepen­dent biological replicates to investigate the proteomic responses of foxtail millet grains under DS (Fig. S1 [suppl]). More than 1000 protein spots were detected using PD-Quest 7.0, and a total of 104 protein spots whose abundance was significantly altered (p<0.05) by more than at least 1.5-fold (Fig. 3) were screened. Amongst these protein spots, 57 were up-regulated and 47 were down-regulated under the DS treatment. All of the DAPs were analyzed by MALDI-TOF/TOF and 104 DAPs were successfully identified (Table S1 [suppl]). A total of 73 proteins (duplicate proteins were not included) with known functions were finally obtained by searching the database. Amongst the 104 identified DAPs, 40 were classified into 12 kinds of proteins (Table 2), namely, 1,4-alpha-glucan-branching enzyme 2 (spots 13 and 71), glucose-1-phosphate ade­ny­lyltransferase small subunit (spots 17 and 31), heat shock 70 kDa protein (spots 21 and 84), fructose-bisphosphate aldolase cytoplasmic isozyme-like (spots 79, 83 and 85), protein disulfide-isomerase (spots 5, 11, 19 and 81), globulin-1 S allele-like (spots 6, 9, 48, 56, 59, 73, 74, 75, 82, 97, 99, 101 and 103), globulin- 1 S allele, partial (spots 42, 44 and 93), glucose-1-phosphate adenylyltransferase large subunit 1 (spots 86 and 87), putative aconitate hydratase (spots 90 and 91), trypsin inhibitor (spots 10 and 95), glucose-1-phosphate adenylyltransferase (spots 62, 63 and 65), adenosine triphosphate (ATP) synthase F1 beta subunit (spots 32 and 35).

Figure 3. Representative two-dimensional gel electrophoresis graphs of foxtail millet grain proteins. Protein spots labelled with numbers on the map are identified through MALDI-TOF/TOF MS. A, Drought stress (DS); B, control (Ctrl).

Functional classification and subcellular loca­lization prediction of differentially abundant proteins under drought stress

Based on molecular function and metabolic path­way, these 104 DAPs were classified into 13 catego­ries, including storage proteins (17.31% of the total 104 DAPs); protein folding (15.38% of the total 104 DAPs); starch and sucrose metabolism (14.42% of the total 104 DAPs); glycolysis/gluconeogenesis (10.58% of the total 104 DAPs); biosynthesis of amino acids, detoxification and defense, protein degradation and TCA cycle (6.73% each of the total 104 DAPs); protein synthesis (5.77% of the total 104 DAPs); energy metabolism (4.81% of the total 104 DAPs); transporter (2.88% of the total 104 DAPs); pentose phosphate pathway (0.96% of the total 104 DAPs) and signal transduction (0.96% of the total 104 DAPs). Approximately 50% of these identified proteins were involved in the first four functional groups. The regulation of differentially abundant proteins with dif­ferent functional classification was diverse under DS (Fig. 4; Table S2 [suppl]). Most of the DAPs, that is, 50 and 20, were respectively located in cytoplasmic and vacuole based on subcellular localiza­tion prediction (Fig. 5).

A total of 18 proteins related to storage proteins were identified. Amongst them, 7 were up-regulated and 11 were down-regulated. Plant carbon metabolism generally included glycolysis, TCA cycle and carbo­hydrate metabolism. A total of 11 DAPs were involved in the glycolysis pathway. Of these proteins, 5 were up-regulated and 6 were down-regulated. At the same time, there were 7 proteins associated with the TCA cycle, 6 were up-regulated and 15 were involved in starch synthesis and metabolism, with 9 of them were down-regulated under DS. Further analysis showed that there were 6 proteins involved in protein synthesis, 16 proteins were involved in protein folding, and 7 proteins were associated with protein degradation, which were collectively referred to as grain proteins associated with protein metabolism. Amongst the 16 proteins involved in protein folding under drought tre­atment, 9 were up-regulated and 7 were down-regu­lated. Four proteins were up-regulated among the 7 proteins involved in protein degradation. Furthermore, 7 proteins were involved in amino acid-related meta­bolism, with 5 of them were down-regulated and 2 were up-regulated. There were 7 proteins related to detoxification and defense, with 5 of them were up-regulated. Five proteins related to energy metabolism were identified and all of them were up-regulated under DS. In addition, there were 3 proteins, 1 protein, and 1 protein related to transporter, pentose phosphate pathway, signal transduction, respectively, and they were all up-regulated under DS (Table S1 [suppl]).

Figure 4. Clustering analysis of differentially abundant proteins under drought stress (DS) and control (Ctrl). The relative abundance value of each differentially abundant protein is represented by a bar with a specific colour. The up-regulated and down-regulated proteins are expressed as red and green, respectively. As the abundance value increases, the colour of the bar covers green to red, and when the abundance value is zero, the colour of the bar is black as shown in the bar at the bottom right.

Figure 5. Subcellular localization of the differentially abundant proteins in foxtail millet grains under drought stress (DS). Black and grey bars represent the up-regulated and down-regulated proteins, respectively.

DiscussionTop

Changes in yield and protein components under drought stress

Drought stress has a great impact on crop yield and quality (Candogan et al., 2013). Similar to previous findings (Samarah et al., 2009; Kilic & Yağbasanlar, 2010; Li et al., 2011), this study indicates that yield and yield-related indexes such as panicle weight per plant, grain weight per plant, 1000-grain weight, and the number of effective panicle, are greatly affected by DS. The protein is one of the important qualities of foxtail millet. Cereal properties and quality are significantly affected by grain protein content and its composition (Pierre et al., 2007; Sunilkumar & Tareke, 2016>, and DS drastically affects crops at the transcriptional level of genes encoding storage proteins (Begcy & Walia, 2015). In this study, the protein components of foxtail millet grains were measured under different irrigation treatments and found that different protein compo­nents behaved differently. Under the Ctrl condition, the following protein components of foxtail millet grains at high to low contents were obtained: albumin (1.56%), gliadin (0.97%), globulin (0.55%) and glutelin (0.54%), the result was consistent with a previous study (Liu et al., 2014). Nevertheless, the highest content under DS was observed in gliadin (1.61%), followed by albumin (1.29%), glutelin (0.66%) and globulin (0.58%). The content of gliadin under DS was significantly higher than that under Ctrl condition, which was similar to the content in wheat (Shen et al., 2006) and rice (Lu, 2012). The ratio of glutelin to gliadin was 0.56 under DS treatment whereas the ratio of glutelin to gliadin was 0.41 under the Ctrl treatment, which was inconsistent with that in wheat (Lan et al., 2004) but was consistent with that in winter wheat (Dai et al., 2006). Previous studies had shown that changes in flour quality of wheat were causally related to changes in protein composition under DS, and grain yield was significantly negatively correlated with protein content mainly because of high rates of grain N accumulation and low rates of carbohydrate accumulation under DS (Saint Pierre et al., 2008; Candogan et al., 2013). In this study, the protein components except albumin were negatively correlated with grain yield under DS. Investigating the proteomic changes in the grains would help comprehensively elucidate the effects of DS on foxtail millet.

Post-translational modification of proteins under drought stress

Proteomic approaches have emerged as efficient tools to study plant responses to stress (Veeranagamallaiah et al., 2008). These approaches have identified not only changes in protein abundance, but also proteins that change through post-translational modifications (Mann & Jensen, 2003). In this study, some proteins were shown in different positions in 2-DE but were identified as the same protein, indicating the possibility of post-translational protein modification or degradation, such as proline hydroxylation, glycosylation, phospho­ry­lation and proteolytic cleavage in foxtail millet grains under DS. This result was similar to previous findings (Holmes‐Davis et al., 2005; Gao & Wu, 2009; Simon, 2010). Globulin, a storage protein, was the only protein component that detected as DAPs through 2-DE. Of the 18 storage proteins, 16 were globulins. Of these globulins, 7 were up-regulated and 9 were down-regu­lated under DS. Therefore, the overall change in globulin was not evident, and this observation was consistent with the previous determination of protein components. Post-translational modifications could regulate protein activity, localization and protein–protein interactions in numerous cellular processes, thus leading to the elaborate regulation of plant responses to various external stimuli (Hashiguchi & Komatsu, 2016). Proteins which were present in multiple spots could be attributed to translation from alternatively spliced mRNAs (Ishikawa et al., 1997). Veeranagamallaiah et al. (2008) showed that post-translational modifications can change the molecular weights and/or charge of proteins. Studies on post-translational modifications under stress conditions have been conducted in various crops, such as rice (Chastain & Gu, 2006), wheat (Zhang et al., 2014), maize (Zörb et al., 2010), sugar beet (Bing et al., 2016) and canola (Koh et al., 2015). To the best of our knowledge, fox­tail millet under DS has not been investigated. There­fore, the post-translational modification mechanisms of foxtail millet should be intensively studied.

Proteins related to protein metabolism under drought stress

Protein metabolism generally included protein folding, protein synthesis and protein degradation. In this study, of the 29 DAPs involved in protein metabolism, 18 were up-regulated and 11 were down-regulated, indicating that protein metabolism was considerably enhanced to some extent under DS. Of the DAPs related to protein folding, 4 were heat shock related proteins (Spots 21, 37, 67 and 84). In general, heat-shock proteins were known for their overexpression in response to chemical and physiological stresses (Wu, 1995). Heat-shock proteins played an important role in normal cellular homeostasis and stress response (Kregel, 2002). They participated in protein folding, assembly, translocation and degradation in many normal cellular processes, stabilize proteins and membranes, and could assist in protein refolding under stressful conditions (Wang et al., 2004). Genomic investigations have shown that heat-shock proteins are highly expressed in DS-exposed tissues (Cho & Hong, 2004). Furthermore, 5 of the 7 protein spots involved in protein degradation were proteasomes (Spots 15 and 26) that were up-regulated under DS. The degradation of many cytoso­lic and nuclear proteins and signalling pathways are affected by the proteasome pathway (Callis & Vierstra, 2000; Criqui et al., 2002) and implicated in plant defense reactions (Suty et al., 2003). Of the 6 DAPs related to protein synthesis, 5 were up-regulated under DS, indicating that drought greatly affected protein synthesis. These results demonstrate that DS greatly influenced protein metabolism.

Proteins related to energy metabolism under drought stress

In clustering heat map (Fig. 4), all of the DAPs related to energy metabolism were up-regulated. The result was consistent with previous studies showing that the abundance of ATP synthesis related proteins was greatly enhanced under stressful conditions (Parker et al., 2006; Guo et al., 2012). According to the study of Kottapalli et al. (2009), water deficit stress in barley may be alleviated by inducing ATP synthases. ATP synthase F1 beta subunit coordinated ATP production with the demand for ATP-fuelled calcium pump activity and regulates cytosolic calcium concentrations (Hubbard & Mchugh, 1996).

In conclusion, this study aimed to characterize the response of foxtail millet under DS at a proteomic level. The results showed that DS greatly influenced foxtail millet in terms of its various properties, such as yield, yield-related indexes and protein components. A total of 104 DAPs were successfully identified through 2-DE coupled with MALDI-TOF/TOF analysis. Amongst them, 57 were up-regulated and 47 were down-regulated under DS. Notably, the identified proteins were involved in various biological processes, such as storage proteins, protein folding and starch and sucrose metabolism. Post-translational protein modifications might also occur. This study revealed changes in the protein components and proteome levels of foxtail millet grains under DS. However, a comprehensive analysis should be performed to further understand the variability in the response of foxtail millet to DS.