Physiological and Proteomic Adaptation of the Alpine Grass Stipa purpurea to a Drought Gradient

Stipa purpurea, an endemic forage species on the Tibetan Plateau, is highly resistant to cold and drought, but the mechanisms underlying its responses to drought stress remain elusive. An understanding of such mechanisms may be useful for developing cultivars that are adaptable to water deficit. In this study, we analyzed the physiological and proteomic responses of S. purpurea under increasing drought stress. Seedlings of S. purpurea were subjected to a drought gradient in a controlled experiment, and proteins showing changes in abundance under these conditions were identified by two-dimensional electrophoresis followed by mass spectrometry analysis. A western blotting analysis was conducted to confirm the increased abundance of a heat-shock protein, NCED2, and a dehydrin in S. purpurea seedlings under drought conditions. We detected carbonylated proteins to identify oxidation-sensitive proteins in S. purpurea seedlings, and found that ribulose-1, 5-bisphosphate carboxylase oxygenase (RuBisCO) was one of the oxidation-sensitive proteins under drought. Together, these results indicated drought stress might inhibit photosynthesis in S. purpurea by oxidizing RuBisCO, but the plants were able to maintain photosynthetic efficiency by a compensatory upregulation of unoxidized RuBisCO and other photosynthesis-related proteins. Further analyses confirmed that increased abundance of antioxidant enzymes could balance the redox status of the plants to mitigate drought-induced oxidative damage.


Introduction
Global warming during the last 100 years has caused substantial changes in temporal and spatial precipitation patterns in many parts of the world [1]. The Tibetan Plateau, sometimes called "the roof of the world", is an important region for studying plants' responses to such Stipa purpurea, a member of the Poaceae family, is a constructive perennial grass species that grows on the alpine steppe and in meadow environments of the Qinghai-Xizang Plateau, the Pamirs Plateau, and high mountains in central Asia. It is a forage species that feeds grassland animals, and it plays important roles in conserving and preserving soil and water, as a windbreak, and in stabilizing sandy soils because of its extensive distribution [16]. On the Qinghai-Xizang Plateau, it is the dominant species in alpine arid (northwest) and semi-arid (midlands) meadows, and it helps to stabilize diverse landscapes because of its outstanding tolerance to harsh environments, including cold, drought, and gale conditions. However, little is known about the adaptive mechanisms that confer its resistance to the harsh environment on the Tibetan Plateau. Because drought is one the main stress factors limiting plant growth on the Tibetan Plateau, we examined the proteomic profile of S. purpurea to identify the key intrinsic response to drought stress under laboratory conditions. We also detected carbonylated (oxidized) proteins, and quantified the activities of selected proteins that were differentially regulated in plants under drought stress. These results help us to better understand the adaptation of S. purpurea to drought conditions on the Tibetan Plateau.

Ethics statement
This field study was permitted by the meadow managers at the Institute of Tibetan Plateau Research, Chinese Academy of Sciences (CAS) and the Grassland Station of Tibet Province. The work was conducted with the approval of, and in cooperation with, the Grassland Station of Tibet Province and local herders. We confirm that the field studies did not involve endangered or protected species. There are no conflicts of ethics or interest to declare.

Sample collection and drought treatments
Seeds of S. purpurea were collected from the Tibetan Plateau, China (35°14 0 37@ N, 98°51 0 19@ E) for these experiments. The seeds were brought to the laboratory, cleaned and surface-sterilized in a solution of 2% sodium hypochlorite for 15 min, rinsed five times in sterilized water, and then placed in plastic trays lined with wet paper towels. The seeds were left to germinate for 36 h in the dark at 23°C. The seedlings were grown individually in soil pots under controlled conditions (25°C day/23°C night cycles; 200 mmol photons m −2 s −1 light intensity; relative humidity 75%-80%). When the third leaf was fully expanded, a continuous drought treatment was applied by withholding watering. Samples of S. purpurea were harvested at 0 d (as control), 1 d, 3 d, 7 d, and 14 d of the drought treatment, immediately frozen in liquid nitrogen, and kept at −80°C until protein and RNA isolation was performed. Three replicates per treatment were collected and analyzed. Soil samples were collected at each sampling time. The soil fresh weight (FW) was measured, then dry weight (DW) was determined after drying soil samples at 105°C for 48 h. Relative soil water content (RWC) was calculated using the formula RWC = [(FW−DW)/DW] × 100%.

Physiological analyses
Chlorophyll fluorescence was measured using Pulse Amplitude Modulation (PAM) Chlorophyll Fluorometer. To measure the maximum quantum yield of photosystem II (PS II), plants were dark-adapted for 30 min. The Fv/Fm value was recorded during a saturating photon pulse (4000 μmol m −2 s −1 ) using a whole plant. Leaf stomatal conductance was measured using a LICOR 6400 portable gas analyzer with a light-emitting diode light source during the drought stress treatment. The RWC of S. purpurea seedlings was measured according to Yamasaki et al. with slight modifications [17]. To estimate FW, each sample was weighed immediately after collection. To obtain turgid mass (TM), each sample was floated in distilled water in a closed petri dish, and then weighed when its moisture content was saturated. The samples were then desiccated using silica in a sealed box and finally oven-dried for 24 h at 80°C to a constant DW.
The malondialdehyde (MDA) content in leaf tissues was measured as described by Duan [18]. Fresh leaves (approximately 0.5 g) were homogenized in 10 mL 10% (w/v) trichloroacetic acid (TCA) and then centrifuged at 12 000 × g for 10 min. Then, 2 mL 0.6% (w/v) thiobarbituric acid in 10% TCA was added to 2 mL of the supernatant. The mixture was heated in boiling water for 30 min and then quickly cooled in an ice bath. After centrifugation at 10 000 × g for 10 min, the absorbance of the supernatant was measured at 450, 532, and 600 nm.

Protein extraction
Each plant sample (1 g) was ground in liquid nitrogen using a mortar and pestle, and then total soluble proteins were extracted at 4°C in 5 mL 50 mM Tris-HCl buffer (pH 7.5) containing 20 mM KCl, 13 mM dithiothreitol (DTT), 2% (v/v) NP-40, 150 mM PMSF, and 1% (w/v) PVPP. The homogenate was centrifuged at 12 000 × g for 15 min at 4°C, then five volumes of acetone containing 10% TCA and 1% DTT was added to the supernatant. The samples were kept at −20°C for 4 h and then centrifuged at 25 000 × g for 30 min at 4°C. The resulting pellet was washed with acetone containing 1% (w/v) DTT, incubated at −20°C for 1 h, re-centrifuged, and then resuspended in acetone. After another centrifugation step, the pellet was vacuumdried and then dissolved in urea buffer [8 M urea, 20 The solution was vigorously mixed with a vortex mixer for 1 h at room temperature, centrifuged at 20°C for 20 min at 25 000 × g, and then the supernatant was collected. The final supernatant was used as the total soluble protein extract. The protein concentration in extracts was measured using the Bradford method [19] with bovine serum albumin as the calibration standard. Samples were used immediately or stored at −80°C until analysis.

Two-dimensional electrophoresis and image analysis
For 2-DE, first-dimension electrophoresis was performed using immobilized pH gradient (IPG) strips (pH 4-7, nonlinear, 17-cm). The IPG strips were rehydrated overnight in 340 μL rehydration buffer [8 M urea; 4% (w/v) CHAPS; 20 mM DTT; 2% (w/v) IPG buffer, pH 3-10 or 4-7; bromophenol blue]. Each strip was loaded with 900 μg protein. The proteins were then subjected to isoelectric focusing at 20°C under the following conditions: active rehydration at 50 V for 14 h, a linear increase to 200 V over 1 h, 500 V for 1 h, a linear increase to 1000 V over 1 h, 8000 V for 2 h, 8000 V to 60 000 VÁh, and held at 500 V for 1 h. After isoelectric focusing, the IPG strips were equilibrated for 20 min in equilibration buffer [6 M urea, 20% (w/v) glycerol, 2% (w/v) sodium dodecyl sulfate (SDS), and 50 mM Tris-HCl, pH 8.8, containing 1% (w/v) DTT], and then alkylated with 2.5% (w/v) iodoacetamide in equilibration buffer for 20 min. The strips were transferred to 12% SDS-PAGE gels with no stacking gel. The proteins were then electrophoretically separated at 15°C and 3 W/gel for 1 h and then 15 W/gel. The 2-DE micropreparative gels (triplicates per experimental treatment) were stained with Coomassie Brilliant Blue R-250, then scanned using a GS-800 calibrated densitometer and analyzed with PDQuest software. The optimized parameters were as follows: partial threshold, 4; saliency, 2.0; minimum area, 50. To verify the autodetected results, all spots were manually quantified by determining the ratio of the volume of each detected spot to the total volume of all spots on the gels.

Protein identification and database searching
Protein spots were manually excised, washed three times with Milli-Q water, and then digested in-gel as described by Wang et al., with minor modifications [20]. Gel slices were reduced with 30% (v/v) acetonitrile in 50 mM ammonium bicarbonate before DTT/iodoacetamide alkylation. The proteins in the gel were digested with trypsin at 37°C for 16 h. The peptides were extracted three times with 0.1% (v/v) trifluoroacetic acid and 50% (v/v) acetonitrile, then MS analysis was conducted using a MALDI-TOF/TOF mass spectrometer 4800-plus Proteomics Analyzer. Primary and secondary mass spectra were transferred to Excel files and submitted to Mascot (http://www.matrixscience.com) for protein identification, applying the following parameters: no molecular weight restriction, one missed trypsin cleavage allowed, iodoacetamidetreated cysteine, oxidation of methionine, a peptide tolerance of 100 ppm, and an MS/MS tolerance of 0.25 Da. Protein identifications were validated manually with at least three peptides matched. Keratin contamination was removed, and the MOWSE threshold was set over 60 (p < 0.05). In the Mascot probability analysis, only significant hits were accepted for the identification of a protein sample.

Analysis of antioxidant enzymes
The activities of APX and SOD were determined using methods described elsewhere [21,22]. Briefly, to measure SOD activity, each 3 mL reaction mixture consisted of 50 mM sodium phosphate buffer (pH 7.8), 13 mM methionine, 75 μM nitroblue tetrazolium, 16.7 μM riboflavin, and an appropriate volume of enzyme extract. The reaction was initiated by light illumination. One unit of SOD activity was defined as the amount of enzyme that causes a 50% decrease in the inhibition nitroblue tetrazolium reduction under these conditions. The activity of APX was determined by following the decrease in absorbance at 290 nm. The reaction mixture contained 50 mM sodium phosphate buffer (pH 7.0), 1 mM ascorbate, 2.5 mM H 2 O 2 , and an appropriate volume of enzyme extract.

In-gel enzyme activity staining
Equal amounts of protein from each sample were loaded onto discontinuous PAGE gels, separated under nondenaturing conditions (4°C, 4 h, 35 mA), and then further separated for native-PAGE activity staining using 10% (w/v) acrylamide slab gels as described by Davis [23]. The activities of SOD and CAT were then determined by in-gel staining as described by Greneche [24].

Detection of hydrogen peroxide
The formation of H 2 O 2 was visualized as fluorescence of 2 0 , 7 0 -dichlorodihydrofluorescein diacetate (H 2 DCF-DA) infiltrated into the root regions as described by Behl [25]. Briefly, S. purpurea roots were washed twice to remove traces of the treatment agents, and then incubated with H 2 DCF-DA (15 μM) for 30 min. A laser-scanning confocal microscope was used to observe fluorescence (excitation at 488 nm, emission at 515 nm).

Detection of carbonylated proteins
Carbonylated proteins were analyzed according to Job [26] with slight modifications, as follows. The protein extraction buffer [50 mM Tris-HCl, 20 mM KCl, 13 mM DTT, 2% (v/v) NP-40, 150 mM PMSF and 1% (w/v) PVPP, pH 7.5] contained SDS at final concentration of 0.8% (w/v). Following dialysis, four volumes 10 mM DNPH/2 M HCl was added to the supernatant. The samples were vibrated for 30 min at room temperature, five volumes of 20/80 ice-cold TCA-acetone containing 1 mM DTT was added, and then the mixture was centrifuged at 25 000 × g for 30 min at 4°C. The resulting pellet was washed three times with 1 mL 1:1 (v/v) ethanol: ethyl acetate, and then incubated at −20°C for 1 h. After a further centrifugation step, the pellet was vacuum-dried and then dissolved in urea buffer [8 M urea, 20 mM DTT, 4% (w/ v) CHAPS, and 2% (w/v) ampholyte pH [4][5][6][7]. The S. purpurea protein samples were separated by 1D-or 2D-PAGE using 12% (w/v) polyacrylamide slab gels and protein identification and database searching were conducted as described above. Carbonylated proteins were detected by western blotting using the anti-DNP immunoassay, as described below.
Measurement of ABA, glycine betaine, and soluble sugars contents, and RuBisCO activity Abscisic acid (ABA) was quantified as described previously [27]. Samples were extracted overnight in 1 ml ABA extraction buffer (methanol, containing 100 mg/L butylated hydroxyl toluene, 500 mg/L citric acid monohydrate) according to the instructions of the Phytodetek competitive ELISA kit. Samples were then centrifuged 25 000 × g for 30 min at 4°C. The supernatant was dried and re-suspended in 100 μL 100% (w/v) methanol, and then 900 μL 1 × TRIS-TBS buffer was added. Then, ABA was quantified using the Phytodetek kit, according to the manufacturer's instructions.
The glycine betaine concentration in the samples was determined according to Costa et al. [39]. Briefly, 0.5 g dry plant material was homogenized in 2 mL distilled water and mechanically shaken for 4 h at 25°C. The sample was then centrifuged, and 250 μL of supernatant was mixed with an equal volume of 2 mol/L H 2 SO 4 and incubated in ice-cold water for 1 h. Cold potassium iodine-iodine reagent (200 μL) was added, and the mixture was gently mixed by vortexing. The samples were incubated at 4°C for 16 h, and then centrifuged at 10 000 × g for 15 min at 4°C. The supernatant was carefully aspirated with a 1-mL micropipette, and the pellet was dissolved in 3 mL 1,2-dichloroethane. After 2-3 h, the absorbance was measured at 365 nm with an ultraviolet-visible spectrophotometer. Glycine betaine reference standards were prepared in 2 mol/L sulfuric acid according to Costa [28].
The concentrations of soluble sugars in the samples were determined following Kochert [29]. The activity of RuBisCO was determined by enzymatically coupling ribulose-1,5bisphosphate carboxylation to nicotinamide-adenine dinucleotide oxidation, which was monitored at 340 nm as described by Xu [30].

Western blotting analysis
For western blotting analyses, proteins were electrotransferred to polyvinylidene difluoride membranes using a Trans-Blot SD instrument. After transfer, the membranes were blocked with 5% (w/v) powdered milk solution for 1 h at room temperature (23-25°C) and then incubated with the primary antibody, which was diluted to 1:3000 (anti-DNP), 1:1000 (anti-ABA), 1:3000 (anti-dehydrin), 1:3000 (anti-HSP) and 1:3000 (anti-Actin), at room temperature for 1 h. Then, the membranes were incubated with horseradish peroxidase-conjugated secondary antibody for 1 h at room temperature. Antibodies were obtained from Agrisera, and chemiluminescence signals were detected using an ECL kit.

Physiological responses of S. purpurea seedlings to drought stress
To analyze the physiological responses of S. purpurea to drought, we determined the water content of leaves under a drought gradient. The leaves of S. purpurea remained greenish and vigorous even when the RWC decreased to approximately 5% ( Fig. 1A and B). Stomatal conductance determines the speed at which water evaporates through the stomata of plants, and is directly related to the size of the stomatal aperture. Leaf stomatal conductance did not change significantly after 1 d of drought treatment, but decreased rapidly from 3 d to 14 d of drought treatment (Fig. 1B).
Next, we compared the maximum photochemical efficiency of PS II in the dark-adapted state (Fv/Fm), which reflects the photosynthetic efficiency of the plant. The PS II maximum efficiency, as reflected by Fv/Fm values, was lower in drought-treated plants than in controls, but did not decrease significantly in drought-treated plants from 3 d to 14 d of drought treatment ( Fig. 1B) (p<0.05). The concentration of MDA is a widely used marker of the degree of oxidative damage to plasma membrane lipids. Drought caused a significant change in MDA concentrations in S. purpurea leaves. The leaf MDA concentration increased to only twice its minimum value by the end of the drought treatment. These results suggest that the drought treatment enhanced the potential of S. purpurea to tolerate further drought.

Protein profiling of S. purpurea seedlings across drought gradients
To evaluate the effects of drought on the proteomic profiles of S. purpurea seedlings, total proteins were extracted from leaves of seedlings at various RWCs and then separated by 2-DE ( Fig. 2A). We detected 58 differentially expressed proteins (at least 1.5-fold difference between control and drought-stressed plants, p<0.05). These proteins were detected and analyzed using PDQuest 7.1 and Genesis 1.7 software (Fig. 2B) [31], and by MALDI-TOF/TOF MS analysis. The identified proteins were assigned Gene Ontology (GO) annotations by comparing their peptide sequences with those in NCBInr databases using the Mascot search engine and Blas-t2GO software [32]. The proteins and their putative molecular functions are shown in Table 1. The 58 proteins were classified into nine functional categories: photosynthesis (26%), material and energy metabolite (17%), response to stress (16%), ion binding (14%), antioxidant activity (7%), signal transducer activity (7%), transport (5%), cell structure (2%), and predicted protein (7%) (represented by proteins detected in 17, 5, 2, 3, 8 and 6 spots, respectively) (Fig. 3A). The abundance of these proteins changed to different extents during the drought treatment (Fig. 3B). The number of down-regulated proteins decreased during the drought treatment, while the number of upregulated proteins increased, especially those that were upregulated by more than 1.5-fold (Fig. 3B). Thirty proteins were identified as upregulated in the following comparisons: 1 d/0 d, 3 d/0 d, 7 d/0 d, and 14 d/0 d (Fig. 3C). These results indicated that the processes involving these proteins were important for drought adaptation in S. purpurea seedlings.

Effect of drought stress on antioxidant enzyme activities
The results of the 2-DE and MS analyses indicated that antioxidant enzymes accumulated in seedlings of S. purpurea under drought stress. This finding was expected, because drought stress is known to induce increases in the activities of antioxidant enzymes such as SOD, APX, and CAT, which increase the capacity to eliminate ROS, in various plant species [33]. The highest concentrations of SOD and APX were observed on days 14 and 3 of the drought treatment, respectively. The abundance of these proteins increased by 2.28-fold (SOD; Fig. 2A, spot 80) and 1.42-fold (APX; Fig. 2A, spot 57), compared with their respective levels on day 0. There was also an increase in the abundance of 2-cys peroxiredoxin BAS1 ( Fig. 2A, spots 49 and 50; 1.83-fold on day 7, 4.28-fold on day 14), suggesting that this protein also contributed to the antioxidant system in S. purpurea. To confirm these results, we measured the activities of antioxidant enzymes in the drought-stressed seedlings. There were significant increases in Images were acquired at indicated times during the drought treatment (top); enlarged images (middle); photosynthetic capabilities recorded by Fv/Fm imaging using a PAM chlorophyll fluorometer (bottom). Pseudocolor code depicted at the bottom of the images ranges from 0 (red) to 1.0 (purple). (B) Effects of drought stress on photosynthetic Fv/Fm ratio, stomatal conductance, and MDA content of S. purpurea seedlings. Relative water content (RWC) of soil and S. purpurea seedlings were measured as described in "Materials and Methods". Data are mean values ± SE obtained from three independent experiments (n = 15). Different letters within a column indicate significant differences (p<0.05; one-way analysis of variance and Tukey's test).
doi:10.1371/journal.pone.0117475.g001 SOD and APX activities as the RWC decreased during the drought treatment ( Fig. 4A and C). Similar trends were observed in the native gel activity assays of drought-stressed S. purpurea seedlings ( Fig. 4B and D).

Drought stress induced the accumulation of ROS
To investigate the interaction between antioxidant proteins and ROS in drought-treated S. purpurea seedlings, we quantified H 2 O 2 production in the root tips. Compared with the control, the drought-stressed plants showed a slow increase in ROS production under prolonged drought stress, as revealed by the increase in H2DCF-DA fluorescence as the drought treatment continued ( Fig. 4E and F). The level of H 2 O 2 peaked when the RWC decreased to approximately 5%. As shown in Fig. 4F, weak fluorescence (green) was emitted from the driest sample

Characterization of carbonylated proteins in S. purpurea under drought stress
To further characterize the oxidative stress induced by drought stress in S. purpurea, we used a proteomic immunochemical approach to identify carbonylated (oxidized) proteins. Proteins were extracted from leaves of plants in the control (0 d) and from those subjected to 14 d of drought, and then separated by 2-DE. The carbonylated proteins were cut from 2D gels, digested with trypsin, and then identified by MS as described above. The identified protein spots are shown in enlarged images of 2D gels ( Fig. 5A and B) and further details are provided in Table 2. The carbonylated proteins showing increased abundance under drought stress included RuBisCO activase (spots C3, C4, C7, C8 and 32), ATP synthase beta subunit (spots C1 and C2), phosphoglucan, water dikinase (spot C5), oxygen-evolving enhancer protein 1 (spot C6), 2-phosphoglycerate dehydratase 1 (spot C9), chloroplastic aldolase (spot 33), and fructose 1, 6-bisphosphate aldolase precursor (spot 35). In another study, the abundance of oxidized proteins increased after exposure to different biotic and abiotic stresses [26]. Our results showed that proteins in the chloroplasts, plastids, mitochondria, and cytosol were oxidized under drought stress in S. purpurea. Dynamic changes in levels of osmolytes, ABA, and RuBisCO in S. purpurea under drought stress Osmolytic adjustment is a vital mechanism in the tolerance of plants to dry environments [34], and glycine betaine and soluble sugars are important metabolites in the drought stress response.
The proteomic analyses showed that several osmolyte proteins responsible for the synthesis of glycine betaine and soluble sugars were upregulated under drought stress ( Fig. 2A, spots 1, 2,  and 65). The concentrations of glycine betaine and soluble sugars were significantly higher (p<0.05) in drought-stressed seedlings than in control seedlings after 14 d of drought treatment ( Fig. 6A and B). The ABA content in S. purpurea seedlings gradually increased under prolonged drought stress, reaching a maximum at day 14 (Fig. 6C). There were no significant differences in RuBisCO activity between drought-stressed plants and controls (Fig. 6D).

Western blot analyses
The proteomic analysis revealed that levels of ABA signaling-related proteins increased in response to drought stress. The plant hormone ABA, which is synthesized by NCED2, modulates tolerance to drought stress [8]. Our proteomic profile showed that NECD2 was upregulated under drought stress. This was validated by the western blot analysis. The increased abundance of several defense proteins [dehydrin COR410-like protein (DHN) and heat shock protein 70 (HSP 70)] was also confirmed by western blot analyses.

Changes in growth and photosynthesis-related proteins under drought stress
Photosynthesis is an important metabolic process that plays a central role in regulating plant responses to drought stress. It is mainly regulated via photochemical reactions related to energy production, gas exchange, and CO 2 fixation and assimilation. Some of the proteins involved in photochemical pathways, such as ATP synthase family proteins ( Fig. 2A, spots 47, 46, 47, and 52), showed increased abundance in S. purpurea under drought stress, suggesting that they were involved in providing adequate and affordable energy. Increased ATPase activity may also function in homeostasis, to maintain the cytoplasmic pH near neutral. The 1.28-to 4.07-fold increases in the abundance of ATP synthase family proteins under drought stress was closely correlated with increased abundance of RuBisCO activase ( Fig. 2A spots 14, 16, and 70), which is involved in regenerating carbamylated active sites in RuBisCO to maintain its activity [35]. In another study, the RuBisCO transcript level in pea increased two-fold when the soil RWC decreased by 50% [36].  Table 2. doi:10.1371/journal.pone.0117475.g005 Phosphoribulokinase (PRK) is involved in regenerating RuBisCO during the carbon assimilation process [37]. In the present study, the increased abundance of PRK ( Fig. 2A, spot 17) (1.62-to 4.17-fold) during days 1-7 of drought stress suggested that RuBisCO regeneration  was promoted under drought stress. A set of proteins, including PRK, that showed 1.3-to 5.0-fold increases under water stress were identified in maize [38]. The accumulation of RuBisCO activase in S. purpurea under drought stress suggested that the photosynthetic CO 2 fixation capability was enhanced via increased RuBisCO activity. RuBisCO for carboxylation and glyceraldehyde 3-phosphate dehydrogenase (GADPH) for carbon reduction were identified as key enzymes for maintaining higher photosynthetic capacity in drought-tolerant compared with drought-sensitive Kentucky bluegrass under drought stress [30]. In the present study, GADPH A ( Fig. 2A, spot 73) and GADPH B ( Fig. 2A, spot 23) were differentially regulated in the leaves of S. purpurea under drought conditions. Together, these results indicated that a higher photorespiration rate contributed to drought tolerance in S. purpurea.
Our proteomic analysis showed that drought stress resulted in oxidative damage to proteins (Fig. 5B) in S. purpurea seedlings. This was revealed by the increased abundance of carbonylated proteins (i.e., oxidized proteins) in drought-stressed plants, as detected in the DNP immunoassay. One of the oxidation-sensitive proteins in S. purpurea seedlings was RuBisCO (Fig. 5, spots C3, C8, C4, C7 and 32). Photosynthetic efficiency can be enhanced by increased carboxylase activity and/or by decreased oxygenase activity of RuBisCO [39]. Carbonylation of RuBisCO subunits resulted in decreased photosynthesis in pea under cadmium stress [40]. In the present study, the main carbonylated proteins were RuBisCO (Fig. 5, spots C3, C8, C4, C7, and 32). However, some RuBisCO proteins were not oxidized (Fig. 2, spots 14, 16, and 70) and were expressed at high levels under drought stress. There were no significant differences in RuBisCO activity in S. purpurea seedlings between drought-stressed and control plants. This result suggested that drought stress might inhibit photosynthesis via oxidation of RuBisCO, and that a compensatory upregulation of unoxidized RuBisCO proteins occurred during the response of S. purpurea seedlings to drought.

Production of ROS, carbonylation of proteins, and antioxidant capacity of S. purpurea under drought conditions
The regulation of ROS that accumulate under biotic or abiotic stress is an important factor in stress responses, because excess ROS can severely disrupt metabolic processes and damage various cellular components, including nucleic acids, lipids, and proteins. Plants have evolved ROSscavenging enzymes including SOD, GR, APX, CAT, GPX, and peroxiredoxin, which provide cells with an efficient mechanism to neutralize ROS. Drought stress has been shown to activate genes encoding antioxidant enzymes such as peroxidase, SOD, and APX in Alfalfa and various other plant species [41]. In our proteomic analysis, we detected differential expression of antioxidant proteins; 2-Cys peroxiredoxin BAS1, APX 1, and SOD ( Fig. 2A). 2-Cys peroxiredoxin BAS1 reduces hydrogen peroxide and alkyl hydroperoxides using reducing equivalents produced in the thioredoxin or glutaredoxin system. This protein was shown to be upregulated by drought-induced oxidative stress in rice chloroplasts, and it was able to decrease H 2 O 2 levels [42]. It has also been shown to function as an electron acceptor in CDSP32-driven electron-transfer, thereby protecting the photosynthetic apparatus from oxidative damage in Arabidopsis [43]. Our proteomic results show that the abundance of 2-Cys peroxiredoxin BAS1 in S. purpurea increased as the RWC decreased ( Fig. 2A, spots 49 and 50).
Our results showed that the greatest abundance of APX1 and SOD was in S. purpurea seedlings in the driest soil (approximately 5% water content) ( Fig. 2A, spots 57 and 80). In plants, SOD and APX are two of the main ROS-scavenging enzymes, and constitute a highly efficient system for removing superoxide and hydrogen peroxide. Hydrogen peroxide is converted to H 2 O by APX, which uses ascorbate as the substrate, and SOD catalyzes the dismutation of O 2 − into O 2 and H 2 O 2 [44]. To further investigate the relationship between H 2 O 2 and antioxidant enzymes, we measured the activities of APX and SOD, and quantified H 2 O 2 production in drought-stressed S. purpurea. The activities of SOD and APX increased during acclimation to drought (Fig. 4). In another study, SOD activity in the leaves of Capparis ovata was 1.7-fold and 1.8-fold higher on days 7 and 14 of drought stress, respectively, compared with their levels in the control; these increased activities prevented ROS damage under drought stress [45]. The results of the enzyme activity assays and proteomics analysis were consistent in this study. Together, our results indicated that during the adaptive response of S. purpurea seedlings, drought induced the expression of antioxidant proteins that neutralized excess ROS and minimized damage.
The degree of protein oxidation appears to be related to the amount of ROS generated and the balance between the rate of protein oxidation and the rate of protein degradation under stress conditions. We observed that S. purpurea could endure oxidative damage caused by decreases in RWC to approximately 5% (Fig. 1A). Another antioxidant defense system, the NO-mediated S-nitrosylation of critical protein thiols, can prevent protein oxidation in plants.
We identified several homologs of S-nitrosylation-sensitive proteins that have been reported in Arabidopsis [46]. The abundance of methionine synthase ( Fig. 2A, spot 11) and elongation factor Tu ( Fig. 2A, spot 21) increased to different degrees in drought-stressed S. purpurea. However, further research is required to determine whether S-nitrosylation-sensitive proteins are S-nitrosylated, and whether NO provides antioxidant protection to cells by S-nitrosylation of functional proteins in the adaptive response of S. purpurea to drought conditions.

Expression patterns of stress-responsive proteins in S. purpurea
Dehydrin proteins, which are members of the LEA protein family, were named for their proposed function in plants under drought stress. Accumulation of dehydrin proteins is associated with drought responses in many plant species including Arabidopsis, rice, wheat, and poplar [10,35,42,47]. Accumulation of members of the LEA protein family, including dehydrins, is often upregulated (>2-fold) in the stress responses of plants [47]. The results of our 2-DE analysis showed that a dehydrin ( Fig. 2A, spot 63) was strongly expressed in leaves of S. purpurea when the RWC decreased from approximately 80% to 5%. Similar results were observed in the western blot analysis. Together, these results indicate that accumulation of dehydrin proteins may have enhanced drought tolerance by protecting cells against dehydration (Fig. 7). Dehydrins may play an important role in S. purpurea by increasing the concentrations of cellular solutes during acclimation to drought conditions. Heat shock protein 70 (HSP70) ( Fig. 2A, spot 4) and heat shock protein cognate 2 (Hsc70-2, spot 5) showed increased abundance in leaves of S. purpurea under drought stress. The functions of Hsc70-2 and HSP70 are similar; both prevent protein aggregation, assist non-native proteins to refold, and maintain organellar precursor proteins under stress conditions. The expression of HSP70 has been reported to increase in many plant species under stress conditions; for example, by more than 8-fold in drought-stressed Arabidopsis, and by almost 10-fold in drought/heat stressed wheat, as compared with controls [48,49]. One example of the function of HSP is to control the correct folding of RuBisCO. A more efficient RuBisCO could reduce photosynthetic water use, thereby increasing tolerance to drought stress [50]. Our results showed that RuBisCO was an oxidation-sensitive protein (Fig. 5B, spot 32, 34). Together, these results suggested that drought-induced HSP70 could help to ameliorate oxidative damage to RuBisCO by ensuring its correct folding to maintain the photosynthetic capability of S. purpurea leaves.

ABA biosynthesis-related protein changes during drought stress
In plants, ABA rapidly accumulates in response to various stresses, including drought and high salinity, and mediates diverse resistance mechanisms. Our 2-DE results showed that the key ABA catabolism enzyme NCED2 ( Fig. 2A, spot 69) was expressed most strongly in plants in the driest soil. Similar results have been reported for a number of plant species including Arabidopsis, rice, and barley [51]. The results of the western blot analysis confirmed that the abundance of NCED2 increased along the drought-stress gradient (Fig. 7). In plants, ABA plays a central role in stomatal closure during drought stress [52]. The ABA signal in guard cells results in the production of nitric oxide and H 2 O 2 , leading to stomatal closure [53]. Stomatal closure is a protective mechanism to avoid excessive water loss from leaves. Our study demonstrated that accumulation of ABA induced stomatal closure, resulting in reduced stomatal conductance (Fig. 1B). These results suggested that ABA is produced by S. purpurea seedlings under drought conditions, reducing stomatal conductance and leading to relatively stable photosynthesis in the leaves.
Expression patterns of material-and energy metabolite-related proteins in S. purpurea Osmotic adjustment is a key adaptive response to drought stress in many plants and other organisms. This process can provide resistance to drought, within species-specific limits, by maintaining turgor and cell volume [4]. The most important osmolytes in plants are glycine betaine and soluble sugars. Accumulation of these substances in plants was shown to enhance the ability of thylakoid membranes to maintain photosynthetic efficiency under drought stress [54]. We detected increases in the amounts of glycine betaine and soluble sugars in S. purpurea seedlings under drought stress. The higher glycine betaine and soluble sugars concentrations in the drought-stressed plants were accompanied by increased abundance of proteins involved in the biosynthesis of glycine betaine and soluble sugars ( Fig. 2A, spots 1, 2, and 65). Thus, glycine betaine and soluble sugars appear to play important roles in osmoregulatory drought resistance of S. purpurea. Similarly, drought stress was reported to increase the levels of glycine betaine and soluble sugars in wheat, thereby improving drought resistance and maintaining photosynthetic activity and metabolism [55].

Conclusion
The responses of the stress-resistant forage species S. purpurea to drought are complex. Drought conditions led to the accumulation of compatible solutes (glycine betaine, soluble sugars) and proteins (PRK and HSPs), which allowed the plant to maintain its photosynthetic capacity under drought stress. There were also substantial increases in the abundance of antioxidant enzymes to scavenge toxic ROS, and changes in the expression of numerous other proteins, including dehydrins. These changes enhanced stress tolerance via ABA-mediated regulation of stomatal conductance to reduce water loss (Fig. 8). All of these responses may help to alleviate the negative effects of drought by preventing damage to vital cell constituents and enhancing osmoregulation.