Mechanisms of Soybean Roots' Tolerances to Salinity Revealed by Proteomic and Phosphoproteomic Comparisons Between Two Cultivars*

Understanding molecular mechanisms underlying plant salinity tolerance provides valuable knowledgebase for effective crop improvement through genetic engineering. Current proteomic technologies, which support reliable and high-throughput analyses, have been broadly used for exploring sophisticated molecular networks in plants. In the current study, we compared phosphoproteomic and proteomic changes in roots of different soybean seedlings of a salt-tolerant cultivar (Wenfeng07) and a salt-sensitive cultivar (Union85140) induced by salt stress. The root samples of Wenfeng07 and Union85140 at three-trifoliate stage were collected at 0 h, 0.5 h, 1 h, 4 h, 12 h, 24 h, and 48 h after been treated with 150 mm NaCl. LC-MS/MS based phosphoproteomic analysis of these samples identified a total of 2692 phosphoproteins and 5509 phosphorylation sites. Of these, 2344 phosphoproteins containing 3744 phosphorylation sites were quantitatively analyzed. Our results showed that 1163 phosphorylation sites were differentially phosphorylated in the two compared cultivars. Among them, 10 MYB/MYB transcription factor like proteins were identified with fluctuating phosphorylation modifications at different time points, indicating that their crucial roles in regulating flavonol accumulation might be mediated by phosphorylated modifications. In addition, the protein expression profiles of these two cultivars were compared using LC MS/MS based shotgun proteomic analysis, and expression pattern of all the 89 differentially expressed proteins were independently confirmed by qRT-PCR. Interestingly, the enzymes involved in chalcone metabolic pathway exhibited positive correlations with salt tolerance. We confirmed the functional relevance of chalcone synthase, chalcone isomerase, and cytochrome P450 monooxygenase genes using soybean composites and Arabidopsis thaliana mutants, and found that their salt tolerance were positively regulated by chalcone synthase, but was negatively regulated by chalcone isomerase and cytochrome P450 monooxygenase. A novel salt tolerance pathway involving chalcone metabolism, mostly mediated by phosphorylated MYB transcription factors, was proposed based on our findings. (The mass spectrometry raw data are available via ProteomeXchange with identifier PXD002856).

maintaining tolerable salt levels inside the plant cells through exportation or compartmentalization are generally accepted as two major strategies used by plants to survive salinity stress (10). Plants have evolved a series of adaptive mechanisms to sense and respond to salinity cues and these include active involvements of multiple phosphorylation cascades, such as salt overly sensitive (SOS) pathway, phosphatidic acid (PA)-mediated activation of calcium-dependent protein kinase (CDPK), abscisic acid (ABA)-regulated activation of mitogen-activated protein kinase (MAPK) cascades (11)(12)(13)(14). Phosphorylation of specific signaling components are known to be initiated at critical time points after plants been subjected to the salt stresses (15) and they coordinate specific metabolic processes, cell-wall porosity and lateral root initiation to help plants adapt to salt stresses (10,13,16).
Recently, major high throughput strategies including transcriptomic, proteomic, and metabolomic approaches, have been used to dissect the responses of soybean root to salinity stress (17)(18)(19)(20)(21). However, most of these studies were focused on relatively late responses to salinity (e.g. over 48 h after Na ϩ treatment), earlier signal events minutes after the treatments were apparently ignored. Signaling events through protein phosphorylation are well known to play critical roles mediating appropriate physiological responses in determining the salttolerant capability of different soybean species. Many techniques have recently been developed for the specific enrichment of phosphopeptides; these includes immobilized metal affinity chromatography (22), strong cation-exchange chromatography (23,24), and TiO 2 affinity chromatography (25). The TiO 2 affinity chromatography has been generally accepted as one of the most effective approaches in enrichment of phosphopeptides (26).
Glycine max cultivar Union85140 and Glycine soja cultivar Wenfeng07 are salinity sensitive-and tolerant-cultivar, respectively; their drastic difference in salt tolerance enable us to explore the critical proteins contributing the salt tolerance in cultivated soybeans (27,28). In the present research, we compared the proteomes and phosphoproteomes of these two soybean species at different time points after salinity treatment. Technologies including TiO 2 affinity chromatography, 2-DE MS/MS, and LC-MS/MS were used to generate the row proteome and phosphoproteome data; large-scale bioinformatic analyses including gene ontology (GO) enrichment and phosphorylation motif enrichment were conducted to identified interested targets; functional characterization of selected target genes using gain-of-function composites in soybean and loss-of-function mutant of their homologs in Arabidopsis were conducted to confirm their role in regulating plant tolerance to salt stresses. Our results reveal that normal chalcone metabolism plays a potential role in regulating plant responses to salt stresses in soybean and provide new insights into the mechanism contributing to the difference in salt tolerance of these two soybean cultivars.

EXPERIMENTAL PROCEDURES
Plant Materials and Stress Treatments-Seeds of Glycine max cultivar Union85140 (a salt sensitive species) and Glycine soja cultivar Wenfeng07 (a salt tolerant species) were kindly provided by Prof. Lijuan Qiu from the Chinese Academy of Agricultural Sciences. The seeds were surface sterilized with 5% NaClO for 5 min and rinsed three times with sterile distilled H 2 O. Seeds were germinated in wet filter paper at room temperature (about 22-25°C) with 40 -60% humidity. The seedlings were transferred to 1/4 fold Hoagland's solution. Seedlings at three-trifoliate stage were treated with 150 mM NaCl for 0 h, 0.5 h, 1 h, 4 h, 12 h, 24 h, and 48 h before the root samples were collected for analyses. All the samples were stored at Ϫ80°C until use.
Protein Extraction-Total proteins from roots was extracted as described by Lv et al (29) with minor modifications. Briefly, about 4 g of root tissue for each sample was ground into fine powder in liquid nitrogen. The powder was thoroughly suspended in 45 ml of precooled TCA/Acetone (v:v ϭ 1: 9); the homogenate was settled for overnight and then centrifuged at 14,000 ϫ g for 15 min. The pellet was washed three times with acetone and the residual acetone was removed by vacuum. All the above experiments were carried out at 4°C. 50 mg white powder was resuspended in 800 l SDT lysis buffer (4% SDS, 100 mM Tris-HCl, 1 mM DTT, 1 mM PMSF, pH7.6, including one-fold PhosSTOP phosphatase inhibitor mixture from Roche), and boiled for 15 min in water bath, and followed by 100 s of sonication. After centrifugation at 14,000 ϫ g for 15 min at 4°C, the protein in supernatant was quantified via BCA (bicinchoninic acid) method (30).
Protein Digestion with Prior Filter Aided Sample Preparation-Approximately 1.5 mg aliquot of dissolved protein for each sample was processed by the filter aided sample preparation method to remove SDS in the samples (31). Briefly, dithiothreitol (DTT) was added to the protein solution to reach 100 mM, and then boiled for 5 min. 25 l aliquot of each sample was mixed with 200 l UA buffer (8 M Urea, 150 mM Tris-HCl pH 8.0), loaded into a Microcon filtration devices (Millipore, with a MWCO of 10 kd), and centrifuged at 14,000 ϫ g for 15 min; 200 l of fresh UA buffer was added to dilute the concentrate in the device and centrifuged again. The volume of concentrate was brought to 100 l with UA buffer supplemented with 50 mM iodoacetamide (IAA) and the sample was shaken at 600 rpm for 1 min. After 30 min incubation at room temperature, the samples were diluted with 40 l of digestion buffer (contains 5 g of trypsin). The mixture was shaken at 600 rpm for 1 min, and incubated at 37°C for 16 -18 h. After digestion, the peptide solution was passed through a Microcon filtration device (MWCO 10 kd), and the concentration of the collected peptides was estimated based on their OD at 280 nm (32).
Eight-plex iTRAQ Labeling-For every eight-plex set, a pooled sample was obtained by combing two groups of samples representing seven time points (a control and six salt treatments) from two cultivars (Union 85140 and Wenfeng07). These pooled samples serve as normalizing reference for the peptide content in samples from all the tested eight-plex sets. A 200 g digested peptides of each sample was subjected to AB Sciex iTRAQ labeling (Fig. 1). The eight-plex iTRAQ labeling was performed according to the manufacturer's instructions. A total of six eight-plex sets of iTRAQ samples were used for the three biological replicates.
Phosphopeptide Enrichment-Phosphopeptides were enriched using TiO 2 beads as described by Ostasiewicz et al. (33) with minor changes. Labeled peptide solutions were lyophilized and acidified by dissolving into DHB buffer (3% 2, 5-DiHydroxyBenzoic acid, 80% ACN and 0.1% TFA). The 25 g of TiO 2 beads (10 in diameter, Sangon Biotech) were added to 50 l peptide solution and spun down after 2 h incubation at room temperature. The pellets were packed into plastic tips (fit to 10 l pipette), washed 3 times with 20 l of wash solution 1 (20% acetic acid, 300 mM octanesulfonic acid sodium salt and 20 mg/ml DHB) then followed by three times with 20 l wash solution 2 (70% water; 30% ACN). The enriched phosphopeptides were eluted using freshly prepared ABC buffer (50 mM ammonium phosphate, pH 10.5) and lyophilized for MS analysis.
NanoRPLC-MS/MS Analysis of Phosphorylated Peptides-The lyophilized phosphopeptides were subjected to capillary liquid chromatography tandem mass spectrometry using a two dimensional EASY-nLC1000 system coupled to a Q Exactive Hybrid Quadrupole-Orbitrap Mass Spectrometer (Thermo Scientific). In nanoLC separation system, mobile phase A solution contains 2% acetonitrile (ACN) and 0.1% formic acid in water, and mobile phase B solution contains 84% ACN and 0.1% formic acid. The Thermo EASY SC200 trap column (RP-C18, 3 m, 100 mm ϫ 75 m) was pre-equilibrated with mobile phase A before peptides loading. The phosphopeptides were initially transferred to the SC001 column (150 m ϫ 20 mm, RP-C18) using 0.1% formic acid solution. The peptides were then separated via the trap column using a gradient of 0 -55% mobile phase B for 220 min with a flow rate of 250 nL/min followed by a 8 min rinse with 100% of mobile phase B. The trap column was re-equilibrated to the initial conditions for 12 min. The MS data of each sample were acquired for 300 -1800 m/z at the resolution of 70 k. The 20 most abundant ions from each MS scan were subsequently dissociated by higher energy collisional dissociation (HCD) in alternating data-dependent mode. The HCD generated MS/MS spectra were acquired with a resolution no less than 17,500.
Phosphopeptide Identification and Quantitative Analysis-The raw HCD files were analyzed by Mascot2.2 and Proteome Discoverer1.4 and searched against a peptide database derived from the Glycine max genome sequence ("uniprot_Glycine_74305_20140429.fasta" downloaded from http://www.uniprot.org/on April 29, 2014, which includes 74, 305 nonredundant predicted peptide sequences) (34). The Mascot search parameters were list in Table I. The Proteome Discoverer 1.4 was used for integrating the spectra intensity (Ͼ 200) of the eight-plex reporter ions. The quantitative value of phosphopeptides at different treatment time points was normalized using the pooled sample as a reference and converted to log 2 value of foldchange. The phosphopeptides pass the cutoff and detected in at least two replicates were used for assessment of significant change in response to NaCl stress. In this research, two statistical approaches were used for significance analysis. The "significance A" value previously described by Cox and Mann (35) was adapted to evaluate the changes between the treated (samples) and untreated (control, T0) root tissues in each biological replicate with each of which includes three technical replicates. A Student's t test was performed using the standard deviation of the pooled sample (standard) between different biological replicates for assessing the global variability of all tested samples (29). The phosphopeptides that passed both Significance A Ͻ 0.05 and p value Ͻ 0.05 were considered significantly changed (36).
Protein Shotgun Identifications by Thermo Scientific LTQ Velos-To construct a comprehensive database of salt responsive proteins in soybean, the LTQ Velos Mass Spectrometer coupled to Zorbax 300SB-C18 peptide traps (Agilent Technologies, Wilmington, DE), was used for protein identifications (37). In which, the analytical column is 0.15 mm ϫ 150 mm (RP-C18) (Column Technology Inc., Fremont, CA). Each sample was analyzed three times and the peptides/proteins identified were combined and listed in supplemental Table S1 and S2.

2-DE Gel Based MALDI-TOF/TOF Mass Spectrometer Analysis for
Protein Identification-0.2% (w/v) DTT and 0.5% IPG buffer (Lot No.: 17-6000-87, GE Healthcare Life Sciences, Piscataway, NJ) were added into the 200 g samples before IEF. Total 250 l samples containing about 200 g proteins were applied to the dry IPG strips (13 cm, pH 3-10 nonlinear, GE healthcare). The program of IEF was as followed: rehydration at 20°C for 12 h, 30 V for 8 h, 150 V for 2 h, 500 V for 0.5 h, 1000 V for 0.5 h, 4500 V for 4000 v⅐hrs, 8000 V for 66000 v⅐hrs. Focused strips were first equilibrated by incubating in equilibration buffer (pH 8.8, 2% (w:v) SDS, 6 M urea, 50 mM Tris-HCl, 30% glycerol (v:v) containing 1% DTT (w:v) for 15 min, followed by incubation in the abovementioned equilibration buffer containing 4% (w:v) iodoacetamide (IAA) for also 15 min. The second dimension separation was conducted on the 12% acrylamide SDS-PAGE. The PAGE gels were stained with Coomassie brilliant blue for over 2 h. Then, all these gels were captured by magic scanner with the same contrast and brightness. Sequentially, spots in these gel images were analyzed using ImageMasterTM 2D Platinum 5.0 software (GE Healthcare) and their relative volumes (% Vol) were represented as relative abundances. Each sample had at least two independent replicates and the differentially expressed protein spots' relative volumes were compared with Student's t test analysis (p Յ 0.05). Spots with significant changes were excised out, and destained with 100 l destaining solution combined with 25 mM ammonium bicarbonate and 50% (v: v) methanol in Milli-Q water. The gel crystals were dehydrated in 100% acetonitrile and vacuum-dried. Then, gel plugs were rehydrated with 10 g/l of trypsin in 25 mM ammonium bicarbonate on ice for 40 min and transferred into 30°C incubator for 16 -18 h digestion. Finally, 80% acetonitrile with 20% trifluoroacetic acid (v:v) was used to extracted digested peptides from the gels. MALDI-TOF/TOF mass spectrometer 4700 Proteomics Analyzer (Applied Biosystems, USA) was applied to identify mass spectrometry of digested peptide. The MS scans were acquired among the mass range from m/z 700 to 3500 Da and the mass errors were less than 50 ppm. The  Quantitative RT-PCR Analysis-RNA isolation, mRNA reverse transcription and qRT-PCR methods were performed as described by Wang et al. (38) with mini modifications. The root samples were frozen with liquid N 2 and total RNAs were extracted with TRIZOL Reagent (Invitrogen). The genomic DNA was removed with DNase I and cDNA was synthesized using the Plant RNeasy Mini kit (Qiagen) according to the manufacturer's instructions. The primers were generated with NCBI online Primer-BLAST against the G. max genome (39). The soybean actin11 gene was used as a reference for normalization. Quantitative RT-PCR used SYBR TM Premix Ex Taq™ (TaKaRa, Shiga, Japan) and the reaction was conducted on a CFX96 System (Bio-Rad). The gene specific primers are listed in supplemental Table S3.
Bioinformatic Analysis-Peptide motifs were extracted using the motif-X algorithm (40). The width of the generated motifs was set as seven amino acids and serine or threonine was selected as the central amino. Gene oncology (GO) analysis was carried as described by Lv et al. (29). The cis-elements recognized by transcription factor binding were identified using JASPAR software (41,42).
Scavenging Activity of the Superoxide Anion (O 2 Ϫ ) Assay-This assay was based on the method of Zhang et al. (43) with slight modifications. Antioxidant enzymes were extracted with 10 ml of 0.05 M phosphate buffer (pH 5.5) from 0.5 g root homogenate. The extract was centrifuged at 12,000 ϫ g (4°C) for 10 min. 1 ml collected supernatant (crude enzyme extract) was added into 4 ml the reaction buffer, which was consist of 2 ml 0.05 M phosphate buffer, 1 ml 0.05 M guaiacol (substrate, overdose) and 1 ml 2% hydrogen peroxide (H 2 O 2 ). The increased absorbance at 470 nm due to the enzyme-dependent guaiacol oxidation was recorded every 30 S until the reaction time reached 4 min. The enzyme's radical scavenging activity (RSA) was defined as: RSA ϭ V ⑀ V t ϫ 1 w ϫ ⌬OD ⌬t (g/min), where w is the weight of fresh root (g), V t is the volume of crude enzyme used in the reaction mixture (ml), V e is the total volume of extracted crude enzyme (ml), ⌬t is the cost time of the reaction (min). Free Radical Scavenging Activity on ABTS ⅐ϩ -The ABTS cationic radical (ABTS ⅐ϩ ) decolorization assay was done by the method of Re et al. (44). ABTS ⅐ϩ working solution was generated by adding 2.45 mM potassium persulphate (final concentration) to 7 mM ABTS (final concentration). This working solution was incubated in dark at room temperature for 12-16 h until it gave an absorbance of 0.70 Ϯ 0.02 at 734 nm. Ten microliters of extracts were mixed with 1.0 ml of working ABTS ⅐ϩ solution and incubated at 30°C for 30 min and the absorbance of reaction mixture was measured at 734 nm. The enzyme's radical scavenging activity was expressed as: RSA ϭ ⌬OD ⌬t where ⌬OD is the reduced absorbance value, ⌬t is the reaction time (min), D f is the dilution factor, w is the weight of fresh root (g), M 0 is the original ABTS ⅐ϩ concentration. Na ϩ and K ϩ Ion Content Analyses-Na ϩ and K ϩ ion contents were detected followed the methods proposed by Qi et al. (45) using the flame atomic absorption spectrophotometer (Shimadzu AA-6300C). The content was expressed as: milligram ion per gram fresh weight (mg/g FW).
Gain-of-function Test of GmCHS, GmCHI, and GmCPM in Soybean Hairy Root System-The full-length CDSs of GmCHS (Glyma01g43880.1), GmCHI (Glyma04g40030.1), and GmCPM (Glyma07g14460.1) from Wenfeng07 was cloned into the pCAM-BIA1301 vector between NcoI and BglII sites downstream of the 35 S promoter. The original pCAMBIA1301 vector was used as a negative control. All these constructs were transformed into the salt-sensitive cultivar Union85140 via agrobacterium rhizogenes strain K599 as previously described (3). The composites were treated in 1/2 fold MS medium with 100 mM NaCl or without NaCl. The seedlings were weighted 10 days after salt treatment.
Loss-of-function Test of AtCHS, AtCHI and AtCPM in Arabidopsis thaliana-The seeds of deletion mutants chs, chs/chi, chs/cpm (Seed stock number: CS85, CS8584, CS8592) were got from ABRC (Arabidopsis Biological Resource Center) and germinated on 1/2 Murashige and Skoog (MS) medium. 5 days after germination, the seedlings were transferred to 1/2 MS medium with or without 150 mM NaCl. The photos of plants were taken 10 days after salt treatment.

RESULTS
Salt stress is well known to cause leaf chlorosis by reducing chlorophyll a, b, and total chlorophyll content (46). After NaCl treatment, we found that the relative contents of chlorophyll a, b, and carotenoids in Union85140 decreased more than that in Wenfeng07 (Fig. 2). In addition, the chlorophyll a/b ratio in Union85140 increased less than that in Wenfeng07 at each time points. These results confirmed that Wenfeng07 is significantly more tolerant to salt stress than Union85140 at the physiological level.
ROS Elimination Capacity and Na ϩ /K ϩ Content in Roots of the Two Soybean Cultivars-The antioxidant property in plant tissue is generally accepted to correlate with plant tolerance to various abiotic stresses including salinity, and it is usually represented by general radical scavenging capacities of peroxidases (POD), ascorbate peroxidase (APX), glutathione Stransferase (GST), and superoxide dismutase (SOD).
The antioxidant properties in salt-treated root tissue of the two cultivars were analyzed using H 2 O 2 guaiacol, DPPH (2, 2-diphenyl-1-picrylhydrazyl) and ABTS (2, 2Ј -azinobis (3-ethylbenzothiazoline 6-sulfonate)) radical scavenging capacity assays as previously described (47). As shown in Fig. 3A and 3B, there was no significant difference (p Ͼ 0.05) between Wenfeng07 and Union 85140 in their superoxide scavenging capacities under normal condition (T0). The scavenging capacities of the superoxide anion (SASA) in these two variants increased consistently at the early stage after salt-treatment (from 1.39 Ϯ 0.03 g Ϫ1 *min Ϫ1 in T0 to 4.87 Ϯ 0.12 g Ϫ1 *min Ϫ1 at time point T4). Starting from 4th hr (T4) of salt treatment, SASA values in these two cultivars were found to decline from their climaxes. Interestingly, SASA values in the salt-tolerant Wenfeng07 were shown to be higher than that in the saltsensitive Union 85140 at the first four sampling times after salt treatment (from T0.5 to T12), but declined quicker and to a much lower level than that in Union 85140 24 h after the treatment. Similar to SASA, ABTS •ϩ scavenging potentials in the two tested cultivars displayed peak values at T1 after a short increase, then started to decrease in the rest of the stress treatment. Consistent with their salt tolerance, the ABTS •ϩ scavenging capability of Wenfeng07 were found to be significantly higher than that of Union 85140 (p Ͻ 0.05) all the time.
In addition, the Na ϩ content and Na ϩ /K ϩ ratio were compared in the two cultivars. Our results showed that, changes in Na ϩ content and the Na ϩ /K ϩ ratio exhibited similar dynamic patterns at different time points in these two cultivars ( Fig. 3C and 3D). The salt tolerant wenfeng07 accumulated higher level of Na ϩ (7.0671 Ϯ 0.5495 mg/g FW) than the salt sensitive Union85140 (1.5189 Ϯ 0.0026 mg/g FW) under control condition. The root Na ϩ /K ϩ ratio in wenfeng07 (0.1829 Ϯ 0.0143-fold) was also significantly higher than that in Union85140 (0.0350 Ϯ 0.0001-fold). After treatment, two peak values of the root Na ϩ content were observed at time points T4 and T24 (Fig. 3C).
Protein Expression Profiles Revealed by LC-MS/MS-To obtain a comprehensive observation on soybean responses to salinity and to search for clues to the mechanistic differences resulting drastic difference in their tolerance, LC-MS/MS was used to analyze root samples of the two compared species of the soybean subjected to salt stress as described in the previous section. Results of three biological replicates are included in supplemental Tables S1 and S2, and major discoveries are summarized in Table II. A total of 46410 peptides out of 14702 proteins were identified from Wenfeng07 and 46710 peptides out of 14585 proteins from Union85140 (Table II). Of these, 4464 and 4409 nonredundant proteins were found from Wenfeng07 and Union85140 respectively.
In total, there were 89 differential expressed proteins been identified by LC-MSMS between these two cultivars (Table III). In detail, 25 and 20 proteins were specifically detected in Wenfeng07 and Union85140 roots, respectively (Table III). Among the 25 proteins specifically presented in Wenfeng07, many of them including MYB transcription factors (TFs), ethylene-responsive transcription factor 6, chalcone synthase, cytochrome P450 monooxygenase CYP51G1, glutamate receptor and a PDR-like ABC-transporter were previously reported to be related with stress responses (23,25,29). Although among the 20 proteins specifically detected in Union85140, the auxin pathway related proteins (such as auxin response factor, auxin-induced protein AUX22 and PIN6a), drought stress responsive protein (KS-type dehydrin SLTI629) and many kinases (such as serine/threonine protein kinase and stress-induced receptor-like kinase 2), might contribute to its general response to salinity stress. Additionally, different homologs of a protein family presented differential expressions in the two varieties. For example, for eukaryotic translation initiation factor 3, the subunit F was expressed with higher abundance in Wenfeng07 roots, whereas the M subunit was expressed with higher abundance in Union85140 roots. Similar dynamics were found in hypersensitive induced reaction protein and nodulin proteins. In addition, the ascorbate peroxidase 2, GST 8, pathogenesis-related protein and two superoxide dismuta-ses (I1LKZ3 and I1LR93) showed opposite trends in these two varieties-they were down-regulated in Wenfeng07, but up-regulated in Union85140.
The results showed that the chalcone flavonone isomerase/ chalcone isomerase was also up-regulated in both Wen-feng07 and Union85140, supporting the LC-MS/MS observations. Interestingly, the ascorbate peroxidase 2 (APX2) protein showed two different isoelectric points (Spots 1757 and 1810 in Fig. 5) in Wenfeng07 roots. In addition, Spot 1757 was up-regulated whereas Spot 1810 was down-regulated. Similarly, the stress-induced protein SAM22 and copper amino oxidase proteins also had two different isoelectric points and different mass weights. These two proteins showed opposite down/up change trends after salt treatment in Wenfeng07. In Union85140, the peroxisomal voltage-dependent anion-selective channel protein, fumarylacetoacetase-like, isocitrate dehydrogenase (NADP) (EC 1.1.1.42) and glyceraldehyde-3phosphate dehydrogenase protein showed three or more spots in the 2-DE gels. Altogether, 14 nonredundant proteins were identified from two or three DEP spots with different isoelectric points and/or molecular weights in the two soybean varieties (Table VI). This implied that the isoforms of the abovementioned proteins might play significant roles in the two varieties.
Phosphopeptide Identification and Quantitative Analysis-The intensity of each phosphopeptide was normalized to the mean of intensities of all phosphopeptides within each biological replicate. Subsequently, the log 2 intensity value changes (salt stress time point Tx/T0) in each condition were calculated for each phosphopeptide (supplemental Table S4). The Student's t test (p values) was performed using the standard deviation of the pooled sample (standard) between different biological replicates for assessing the global variability of all tested samples (supplemental Table S4).
Phosphorylation Motif Analysis for Quantitative Phosphopeptides-To extract overrepresented patterns from the 1164 quantitative phosphorylated peptides with differential  changes between the two cultivars, the software MEME Suite and motif-X were used to analyze the motifs generated at different time points after salinity treatment from the two soybean cultivars. The intensities of phosphopeptides from Wenfeng07 (I pW ) were compared with those from Union85140 (I pU ) and the ratio values (I pW :I pU ) with significant (p value Ͻ 0.05) differences were divided into two groups. When the intensity value I pW Ͼ I pU , its corresponding phosphopeptide was categorized into the Up group, whereas the phosphopeptide with I pW Ͻ I pU was categorized into the Down group. The Up group represented the peptides with higher phosphorylation level in the salt-tolerant cultivar and lower phosphorylation level in the salt-sensitive cultivar. There were ten phosphorylation motifs enriched from the Up group (Fig.  6A) and 14 motifs enriched from the Down group (Fig. 6B). In addition, Ser and Thr were observed as the central phosphorylated amino acid residue in both groups, with much higher frequency for Ser. In both the Up and Down groups, the amino acid closely neighboring the phosphorylated Ser/Thr was mainly Pro or Asp (Fig. 6 plemental Tables S4 and S5). For example, ten GmMYB family proteins were quantitatively analyzed on one or more phosphorylated peptides. Interestingly, the phosphorylated peptide TVPSAsG in GmMYB I1KQI5 was detected in both cultivars, and another phosphorylated peptide FsPNLNQN-PNPNLGK could only be detected in Union85140 (supplemental Tables S4 and S5), indicating that phosphorylation of the same protein could be modified at different sites in the two cultivars and might generate various activations. In addition, the phosphorylated peptide QKIDDsDESPNPK in GmMYB K7MQI8 in both cultivars was only detected at late time points (T12-T48) (supplemental Table S4). Similar results were observed in GmMYBs (K7LAB8 and I1JE71), GmbZIPs (Q00M78, I1JDF7, K7MV95, and C6T6L1), GmWRKY (I1MT25) and GmERF (I1KN17). This suggested a temporary regulation of this modification in response to salt stress.
To reveal the potential interaction network between TFs and differentially expressed proteins, the TF-specific binding motifs of the promoters from enzymes involved in chalcone metabolism are summarized in Fig. 7. Motif structures of these promoters were retrieved from the JASPAR database (50). All the promoters of genes encoding chalcone synthase (GmCHS), chalcone isomerase (GmCHI), and cytochrome P450 monooxygenase (GmCMP) were predicted to contain the conserved motifs recognized by MYBs, indicating that the promoters of these 13 enzymes should be regulated by this TF family. In addition, promoters of the two GmCMP and one GmCHI also included motifs recognized by bZIP. Additionally, GmERF had potential binding motifs in promoters of some GmCHS, GmCMP and GmCHI genes. Because their activities might be regulated by phosphorylation modifications, these     TFs should play significant roles in the bridge between stress signal and the transcription of salt responsive genes.
Rapid Function Tests of the Genes Involved in Chalcone Metabolism-In this research, the enzymes involved in chalcone metabolism were proposed to have potential correlations with soybean salt tolerance at both proteomic and transcriptional levels. To further validate that the chalcone synthase (CHS), chalcone isomerase (CHI) and cytochrome P450 monooxygenase (CPM) were determinants of plant salttolerance, gain-of-function and loss-of-function analyses were tested in soybean composites (Fig. 8) and A. thaliana mutants (Fig. 9), respectively, at seedling stage.
When subjected to NaCl treatments, the Union85140/ gmchs-ox composites showed higher tolerance than Union85140/pCAMBIA1301 (negative control) (Fig. 8), indicating that chalcone was a positive regulating factor in salt tolerance. Both of the gain-of-function Union85140/gmchi-ox and Union85140/gmcpm-ox composites (Fig. 8) showed a slight lower tolerance than the negative control. Similar results were observed in Arabidopsis, the single deletion mutant (chs) showed significantly lower tolerance than wild type ( Fig. 9A and 9B), indicating that chalcone was a positive regulating factor in salt tolerance. Both of the loss-of-function double mutants chs/cpm ( Fig. 9C and 9D) and chs/chi ( Fig. 9E and 9F) also showed lower tolerance than wild type. However, these double mutants (chs/cpm and chs/chi) showed higher tolerance than the single deletion mutant (chs), suggesting that chalcone isomerase and cytochrome P450 monooxygenase were two negative regulating factors in salt tolerance. To summarize, chalcone synthase dominated the response to salt stress in chalcone metabolism.

DISCUSSION
Compared with the salt-sensitive Union85140, the salt-tolerant Wenfeng07 showed no significant advantage in exportation or compartmentalization of salts, but much higher ca-pacity for ROS elimination within 48 h of NaCl treatment. Plants have evolved very complex mechanisms for ROS elimination at the transcription, translation and post-translational modification levels (12,13,15,29,51). The present study involved a comparative analysis of salt stress responses between a salt-tolerant and a salt-sensitive soybean variety using proteomic and phosphoproteomic approaches. Among them, 89 representative differentially expressed proteins were checked with their changes at transcriptional level using quantitative RT-PCR. Our results confirmed the view that expression differences at proteomic level are involved in functional proteins, whereas differences at phosphoproteomic level are mainly related to regulatory proteins (29). Interestingly, a series of proteins related to ROS scavenging and protein folding/degradation-such as GST, APX, SOD, heat shock protein 90 -2, and Hsp70-Hsp90 organizing protein 1-were involved in salt responses of both salt-tolerant and salt-sensitive varieties, which were almost in accordance with previous studies (17,52,53). However, tolerance discriminations were possibly dominated by: (1) synthesis of flavonoid/ isoflavonoid involved in the salicylic acid defense pathway by chalcone metabolism (54,55) in Wenfeng07, compared with initiation of lateral roots by auxin response factor, auxininduced protein AUX22 and PIN6a (10) in Union85140; (2) up-regulation of ERF and MYB TFs for activating MAPK and SOS pathways to eliminate ROS and excessive salts (12,13) in Wenfeng07; and (3) regulating innate immunity via cytochrome P450 monooxygenase, chalcone isomerase, and sterol 24-C methyltransferase (56,57) 16 were also identified as phosphoproteins, including the stress-induced protein SAM22, histone H2A OS, eukaryotic translation initiation factor 3 subunit C, elongation factor 1-delta-like, fructose-bisphosphate aldolase, cytosolic glutamine synthetase GSbeta1 for signal transduction, chromosome remodeling, gene translation, energy and small molecular metabolism, etc.
Perception of Salinity and Signal Transduction-The SOS system (e.g. SOS1: H9CDQ2, supplemental Table S4) acts as a central hub in preventing Na ϩ toxicity in the plant, especially for Wenfeng07. The most common role of the SOS system is to sequestrate Na ϩ ions from the plant cytosol (58). In general, the high salt stress suddenly triggers a cytosolic Ca 2ϩ signature (59), which can be perceived by the calcineurin B-like protein, SOS3 and Ser/Thr protein kinase, SOS2 (60). After perceiving the Ca 2ϩ signature, SOS3 is phosphorylated by the protein kinase SOS2. The SOS2/SOS3 complex activates the plasma membrane Na ϩ /H ϩ antiporter, SOS1. Downstream of the SOS cascade, SOS1 mediates Na ϩ efflux at the root epidermis (61). In our study, there were many SOS2 and SOS3 homologs found with multiphosphorylated sites and with different regulation levels. For example, GmSOS2 (K7KTI3) was observed with four phosphorylation sites, in which phosphor-Ser in peptide LPEsPREGSEEDNFLENLTGMPIR only occurred at early time points T0.5-T4, but the phosphor-Ser in peptide EGsEEDNFLENLTGMPIR only occurred at late time points T12-T48 (supplemental Table S4). Interestingly, Gm-SOS3 (C6T458), both in Wenfeng07 and Union85140, was detected at T0 and all treatment times except T4 (supplemental Table S4). In addition, another GmSOS3 homolog FIG. 8. Effects of salinity stress on seedlings of soybean composites. The seedlings of negative control (Union85140/ pCAMBIA1301), gmchs-ox (Union85140/ GmCHS), gmchi-ox (Union85140/GmCHI) and gmcpm-ox ((Union85140/GmCPM)) composites were treated in 1/2 MS medium with or without 100 mM NaCl for 10 days. Bar: 1 cm.
The Ca 2ϩ signature could also be perceived by calciumdependent protein kinases (CDPKs or CPKs) (62). The latter two transmit the signal into phosphorylation cascades capable of modulating gene expression and target protein activity (63). CDPKs, through their interaction with ion channels and transporters, seem to represent part of membrane-delimited plant stress responses (64). In the present study, the Gm-CDPK (D3G9M7) in Wenfeng07 showed much higher phosphorylation levels than Union85140 for time points T0.5-T48 (supplemental Table S4). This suggested that this GmCDPK might significantly contribute to the salt tolerance of Wenfeng07.
Reactive oxygen species (ROS) and hormones are key elements in intricate switches used by plants to trigger highly dynamic responses to changing environment. Although ROS may have deleterious effects in cells, they also act as signal transduction molecules involved in mediating responses to environmental stresses (65). Plant plasticity in response to the environment is linked to a complex signaling module in which ROS and antioxidants operate together with hormones, including auxin (66). The auxin resistant double-mutant tir1 afb2 showed increased tolerance to salinity as measured by chlorophyll content, germination rate and root elongation. In addition, mutant plants displayed reduced hydrogen peroxide (H 2 O 2 ) and superoxide anion (O 2 ϪϪ. ) levels, as well as enhanced antioxidant metabolism (67). Microarray analyses in- dicated that auxin responsive genes are repressed by oxidative and salt treatments in rice (68). More recently, the transcriptomic data of Blomster et al. (69) showed that various aspects of auxin homeostasis and signaling are modified by apoplastic ROS. Together, these findings suggest that the suppression of auxin signaling might be a strategy of plants to enhance their tolerance to abiotic stress, including salinity. In this study, the auxin response factor K7M7H1 was found with phosphorylated serine (in peptide sPPQPR). However, this modification was only detected at late time points T12-T48 (supplemental Table S4). Recent research found that a saltresponsive ethylene response factor1 (ERF1) regulates ROS-dependent signaling during the initial response to salt stress (13). However, the GmERF (I1KN17) was only observed with phosphorylation modification in the sensitive cultivar Union85140 (supplemental Table S4).
Other reported pathways of salt signaling include mitogenactivated protein kinase (MAPK or MPK) cascades (70). A MAPK cascade consists of a MAPK kinase (MAPKkk)-MAPK kinase (MAPKK/MKK)-MAPK module that links salt-signal receptors to downstream targets (71). For a rapid signal transduction, the GmMAPKK2 (Uniprot accession no. Q5JCL0) showed a much higher level phosphorylation modification after NaCl treatment in both Wenfeng07 and Union85140 (supplemental Table S4).
Metabolism of Small Molecules Related to Detoxification and Defense Pathways-Under salinity stress, the plant employs detoxification and defense pathways to increase their tolerance (58). Several abiotic stresses, such as salt, drought and cold can induce ROS accumulation including O 2 ϪϪ. , H 2 O 2 and hydroxyl radicals (10). Suitable concentrations of ROS are acquired as substrates in lipid, sugar and protein metabolisms. Peak values of ROS concentration usually act as signals for inducing ROS scavengers, which are mainly substrates involved in these metabolisms. In this study, copper amino oxidase and quinone oxidoreductase, which produces ROS (72,73), were up-regulated after salt treatment. Meanwhile, universal scavengers, such as APX, SOD, GST and POD, also showed up-regulation in roots of both saltsensitive and -tolerant soybean. Among these scavengers, APX has been shown to reduce H 2 O 2 to H 2 O, with the concomitant generation of monodehydroascorbate. Many reports demonstrating that APX overexpression can enhance the salt tolerance of different plants have confirmed that APX plays an important role in scavenging ROS produced by salinity stress (74 -77). Moreover, the two homologs of APX might have different efficiencies in ROS elimination, because APX2 was significantly up-regulated in the tolerant cultivar (Wenfeng07), whereas APX1 was significantly up-regulated in the sensitive cultivar (Union85140) after salt treatment. This result is consistent with findings in two rice APXs (78).
Chalcone Metabolism Pathway is Involved in Soybean Tolerance to Salt Stress-Up to now, the chalcone metabolism pathway has mainly been considered as a feasible strategy for enhancing plant immunity to microbes (79 -81). In plants, chalcone biosynthesis begins with the hydroxylation of cinnamic acid by cytochrome P450 monooxygenase (82). The intermediate product p-coumaric acid is then activated by 4-coumaroyl:CoA ligase, yielding p-coumaroyl-coenzyme A (CoA) (83,84). Subsequently, malonyl-CoA is added to p-coumaroyl-CoA and yields tetrahydroxychalcone by the enzyme chalcone synthase. Finally, chalcone isomerase converts the C15 compound tetrahydroxychalcone into (2S)-flavanones (85)(86)(87). These flavonoids, including a diverse family of polyphenols, have been proven with health-promoting effects especially in preventing various human pathological risks (88,89). Hence, significant amounts of research have been stimulated to elucidate the biosynthetic networks of flavonoids (90,91). However, there are very few reports on the contribution of chalcone metabolism to plant salt tolerance (92,93). Recently, a cytochrome P450 monooxygenase mutant was shown to be involved in a series of abiotic stresses including ABA and salt in Arabidopsis (94). Our proteomic and phosphoproteomic analyses showed that key enzymes, such as cytochrome P450 monooxygenase, chalcone synthase and chalcone isomerase, were correlated with salt stress especially in tolerant cultivar Wenfeng07. Their salt-responsive dynamics were also confirmed at the transcriptional level. The functions of these enzymes were preliminarily tested in soybean composites and Arabidopsis mutants. Both the gain of function and loss-of-function tests demonstrated that cytochrome P450 monooxygenase and chalcone isomerase were negatively related with salt tolerance in plant seedlings, whereas chalcone synthase was positively related.
Interestingly, 10 MYB (MYB like) transcription factors (TFs) were identified with significantly changed phosphorylation sites (supplemental Table S4 and S5). Commonly, MYB TFs play crucial roles in flavonol accumulation by regulating the expression of series genes coding for key enzymes involved in chalcone metabolism in plants (95)(96)(97). In addition, three chalcone metabolism enzymes have been found in response to salt stress. These results indicate that the network between phosphorylated MYB TFs and chalcone metabolism enzymes might play potential crucial roles in soybean's tolerance to salinity. CONCLUSION Plants have evolved a set of physiological and biochemical responses for adaptation to salinity stress. Generally, glutathione and proline as well as several secondary metabolites, such as flavonoids, play a pivotal role in tolerance/detoxification of plants (98 -100).
In the present research, quantitative proteomic and phosphoproteomic analyses were conducted with both salt-tolerant (Wenfeng07) and -sensitive (Union85140) soybean varieties under salt stress. LC-MSMS and 2-D gel based proteomic analysis of these two variants from a series of time points after salt treatment identified 179 differentially expressed nonre-dundant proteins in total. Of these, 16 proteins also showed changes at phosphorylation level. These differential protein expression characteristics were mostly involved in functional pathways which possibly dominated the capacity of the two varieties concerning salt tolerance.
The quantitative phosphoproteomic analysis identified 3744 phosphorylated sites and 1163 differentially changed sites between the two cultivars, which revealed an activated signaling cascade involved in salt response. The comparison at phosphorylation level indicated that the hub signals fitted with salt tolerance in the tolerant variety.
In summary, the proteomic and phosphoproteomic comparisons between tolerant and sensitive variants could aid understanding of the response and defense mechanisms of soybean in response to salinity stress. The transcriptional and functional analyses confirmed the correlation of significantly changed proteins with salt tolerance. Moreover, the identified significantly changed proteins and phosphorylated sites provide an array of potential salt-response markers for future work. More importantly, the chalcone metabolism pathway was shown as a likely novel candidate for further research on plant salt tolerance. Based on these findings, we hypothesized a novel soybean salinity-tolerance pathway involved in chalcone metabolism (Fig. 10). After the perception of salinity signal, the GmMYBs are phosphorylated and further activated the genes GmCHS, GmCHI and GmCPM. Then, these activated key enzymes GMCHS, GmCHI and GmCPM mediated the accumulation patterns of flavonoids. Finally, these fla-vonoids appropriately reduced the ROS or play roles in other functions for enhancing the soybean's tolerance to salinity. FIG. 10. A hypothetical model for transcription factor GmMYB in regulating genes GmCHS, GmCHI and GmCPM during soybean's response to salinity. After the perception of salinity signal, the GmMYBs are phosphorylated and further activated the genes GmCHS, GmCHI and GmCPM. Then, these key enzymes GMCHS, GmCHI and GmCPM regulated the accumulation patterns of flavonoids. Finally, these flavonoids appropriately reduced the ROS or play roles in other functions for soybean's tolerance to salinity.