Proteomic profiling of proteins associated with the rejuvenation of Sequoia sempervirens (D. Don) Endl

Background Restoration of rooting competence is important for rejuvenation in Sequoia sempervirens (D. Don) Endl and is achieved by repeatedly grafting Sequoia shoots after 16 and 30 years of cultivation in vitro. Results Mass spectrometry-based proteomic analysis revealed three proteins that differentially accumulated in different rejuvenation stages, including oxygen-evolving enhancer protein 2 (OEE2), glycine-rich RNA-binding protein (RNP), and a thaumatin-like protein. OEE2 was found to be phosphorylated and a phosphopeptide (YEDNFDGNSNVSVMVpTPpTDK) was identified. Specifically, the protein levels of OEE2 increased as a result of grafting and displayed a higher abundance in plants during the juvenile and rejuvenated stages. Additionally, SsOEE2 displayed the highest expression levels in Sequoia shoots during the juvenile stage and less expression during the adult stage. The expression levels also steadily increased during grafting. Conclusion Our results indicate a positive correlation between the gene and protein expression patterns of SsOEE2 and the rejuvenation process, suggesting that this gene is involved in the rejuvenation of Sequoia sempervirens.


Background
Plant maturation involves sequential developmental stages or phases that can be categorized as embryo, juvenile, transitional, and adult. The development of reproductively mature adult plants usually begins with a strictly vegetative juvenile phase. Maturation or a phase change is completed within weeks among annuals but can proceed for several years among perennials. In trees, the process is frequently accompanied by ancillary morphological and physiological traits, most commonly a loss of competence for adventitious rooting and a loss of overall vigor. Because a plant's developmental phase is determined in its shoot apical meristems, reversing the phase of the meristems should result in the emergence of rejuvenated shoots [1]. Indeed, new growths with reversed phases have been obtained by applying gibberellin [2] and cytokinin [3,4], continuously subculturing shoots, especially in cytokinin-containing media [5][6][7], and repeatedly grafting shoot apices from mature trees onto juvenile rootstocks in vivo [8,9] and in vitro [10][11][12][13][14][15].
Sequoia sempervirens (D. Don) Endl, a coastal redwood, can be rejuvenated through repeatedly grafting its adult-phase shoot tips onto juvenile rootstocks in vitro [16], which can result without phytohormonal supplements. Adventitious rooting is notably depressed and essentially absent in tissues of adult trees. The stability of the rejuvenated state by restoring rooting competence and other juvenile characteristics can be retained in vitro even after 30 years. The reversion has been associated with distinctive leaf proteins, including clearly apparent changes in iso-esterase and iso-peroxidase [17], different tyrosine phosphorylation patterns, and higher total protein phosphorylation in juvenile shoots [18]. Juvenile and rejuvenated shoots also released more ethylene [19], had higher total nitrogen content, and were more active in photosynthesis and respiration [20]. This diversity of differences in physiology between juvenile and adult Sequoia shoots reflect the complexity of the developmental phase change.
In 1991, a 16 kDa protein was found to be only produced in juvenile or rejuvenated meristems [21]. In 1996, a 28 kDa protein was found to be expressed in greater amounts in juvenile shoot tips than mature shoot tips [22]. Moreover, Gil et al. (2003) identified a cDNA clone termed Quercus robur crown preferentially expressed (QRCPE) that was differentially expressed in juvenile-like and mature shoots in Quercus robur. QRCPE appeared to be a cell-wall protein [23]. However, the proteomic differences in adult, juvenile, and rejuvenated tree species are not well known. Using a proteomics approach, candidate regulatory components of the rejuvenation of tree species may be identified. In the present study, Sequoia was utilized as our rejuvenation system, and gel-based proteomic analysis was performed. Three proteins with differential abundance in adult, juvenile, and rejuvenated Sequoia were identified by mass spectrometry (MS), including oxygen-evolving enhancer protein 2 (OEE2), glycine-rich RNA-binding protein (RNP) and a thaumatin-like protein (TLP). Results from the gene expression and protein accumulation pattern indicated a positive correlation between OEE2 and the rejuvenation stages, suggesting that OEE2 may be involved in the rejuvenation of Sequoia.

Results
The restored rooting competence of repeatedly grafted Sequoia shoots is retained after 30 years of culture in vitro Sequoia shoot cultures initiated in 1976 and 1994 had retained their hallmark characteristics. Those in the adult phase were rootless, had enervated shoots, and grew very slowly compared with juvenile phase cultures ( Figure 1). These developmental stage-specific samples included adult cultured shoots established in 1976 (Adult76) and 1994 (Adult94) and juvenile and rejuvenated (Rejuvenated) shoots. However, rooting was substantially restored among adult shoots that had been grafted four times onto juvenile rootstocks in 1976 and after three times in 1994.

Two-dimensional gel electrophoresis of proteins
Mass spectrometry (MS)-based proteomic analysis has been widely used to identify differentially expressed proteins at different stages of plant development [24].  used two-dimensional gel electrophoresis and discovered pattern differences among proteins extracted from adult, juvenile, and rejuvenated shoot cultures, in which the rejuvenated cultures were obtained by repeatedly grafting adult onto juvenile stock [16]. However, the proteins were not identified. Thus, the present study identified some of the differentially expressed proteins.
Total proteins were isolated from A cultured shoots established in 1976 (Adult76) and 1994 (Adult94) and juvenile and rejuvenated cultured shoots (Rejuvenated76 and Rejuvenated94). The protein extracts were subjected to two-dimensional gel electrophoresis with at least three biological replicates analyzed for each type of tissue. The protein gel images are shown in Figure 2. At least three distinctive protein spots with differential intensity were detected. The relative intensity of each protein spot was quantified using SameSpots software (Progenesis, USA).
Identification of Sequoia proteins exhibiting up-or downregulated protein abundance during rejuvenation Three protein spots that showed differential accumulation (labeled 1, 2, and 3) in each gel were excised and subjected to in-gel digestion followed by MS ( Figure 2). Proteins were identified by a database search using inhouse Mascot software (Table 1). Protein identification was validated using Scaffold software to provide confidence interval percentages (Additional file 1). Oxygenevolving enhancer protein2 (OEE2), glycine-rich RNA binding protein (RNP) and thaumatin-like protein (TLP) were consistently identified in the three biological replicates. Specifically, two identified proteins, RNP and OEE2, showed increased protein abundance during the rejuvenation stage (Figure 3), whereas TLP showed decreased protein abundance during the rejuvenation stage ( Figure 3). In addition to spots 1, 2, and 3, three other proteins from other spots were also identified. The proteins included glyoxysomal malate dehydrogenase, class I chitinase and a LRR-repeat protein (Additional file 2). However, these proteins were not consistently identified in all five protein gel samples. Our detailed studies were confined to OEE2, RNP and TLP. The deduced amino acid sequences of OEE2, RNP and TLP were shown in Figure 4A,F, Figure 5 and Figure 6A, respectively. These sequences were entered into the database, and an additional database search was performed with in-house Mascot software. More peptides were identified, and the sequence coverage was greatly increased ( Table 1 Figure 6). OEE2 was identified in spot 1, with a predicted molecular mass of 27.7 kD. However, the apparent molecular mass of OEE2 on the protein gel was approximately 21 kD, which was much smaller than expected. Additionally, the N-terminus peptide fragments of OEE2 were not discovered in this study (Table 1). Altogether, the results suggest that the identified OEE2 is a chloroplast form without the transit peptide. The transit peptide of OEE2 could be cleaved by a protease before translocation into the chloroplast. Without including the Sequoia OEE2 protein sequence in the database, only two OEE2 peptides were identified (Table 1). A Mascot protein score of 2017 was obtained when including Sequoia OEE2 in the database. Thirteen OEE2 peptides, including a phosphopeptide, were identified. The protein coverage was 52%. The phosphopeptide of OEE2 was doubly phosphorylated (YEDNFDGNSNVSVMVpTPpTDK), and the MS/ MS spectrum of the phosphopeptide is shown in Figure  7. This phosphopeptide was validated by Scaffold, with a confidence interval > 95% (Additional file 1).
RNP was identified in spot 3. Without including the RNP protein sequence in the database, only one peptide was identified ( Table 1). The Mascot protein score was 67, and the protein coverage was 11%. The Mascot protein score was 1155 when including the Sequoia RNP protein sequence in the database. Fourteen peptides were identified, and the protein coverage was 61%. TLP was identified in spot 2. The Mascot protein score was 833, and three peptides were identified.
Protein levels of OEE2, RNP, and TLP in repeatedly grafted Sequoia shoots All of the above three proteins were consistently identified in the three biological replicates. OEE2 protein abundance was found to be up-regulated in Rejuvenated76 and Reju-venated94. The protein level in the adult specimen (Adult76 and Adult94) was lower than the juvenile and rejuvenated samples (Figure 3). Adult76 and Adult94 were significantly different from Juvenile, Rejuvenated76 and Rejuvenated94 (p < 0.05; one-way analysis of variance [ANOVA] followed by Duncan's multiple range test for statistical analysis of means) (Additional file 3 &4). The protein abundance of OEE2 through the steps of repeated grafting showed a positive correlation between rejuvenation and protein abundance in Sequoia shoot cultures initiated in both 1976 and 1994 ( Figure 3).
By contrast, the protein abundance of RNP was found to be up-regulated in Rejuvenated94 but not Rejuve-nated76. The protein level of RNP in the adult plantlet (Adult76 and Adult94) was lower than the Juvenile and Rejuvenated94 samples ( Figure 3). However, Adult76 and Adult94 were not significantly different from Juvenile, Rejuvenated76 and Rejuvenated94 (p < 0.05; oneway ANOVA followed by Duncan's multiple range test for statistical analysis of means) (Additional file 3, 4). Therefore, the protein abundance of RNP through the steps of repeated grafting showed no correlation between rejuvenation and protein abundance in Sequoia shoot cultures initiated in 1976 or 1994 ( Figure 3). TLP abundance was found to be down-regulated in Rejuvenated76 and Rejuvenated94. The protein level in the adult specimen (Adult76 and Adult94) was higher than the Juvenile and Rejuvenated samples ( Figure 3). Adult76 and Adult94 were significantly different from Juvenile, Rejuvenated76 and Rejuvenated94 (p < 0.05; one-way ANOVA followed by Duncan's multiple range test for statistical analysis of means) (Additional file 3, 4). The protein abundance of TLP through the steps of repeated grafting showed a reverse correlation between rejuvenation and protein abundance in Sequoia shoot cultures initiated in both 1976 and 1994 ( Figure 3).

Phylogenetic analysis of OEE2, TLP and RNP
The full-length cDNAs of the SsOEE2, SsTLP, and SsRNP genes were cloned from Sequoia shoots in vitro and further analyzed. The cDNA length of SsOEE2 was 981 bps and encoded a predicted 27.7 kDa protein. The isoelectric point of SsOEE2 was 9.18. SsOEE2 contained a signal peptide for chloroplast in the N-terminal. Phylogenetic analysis of OEE2 by the neighbor-joining method revealed that the Sequoia sempervirens protein had 49-71% homology in the amino acid sequence with the protein of other species ( Figure 4B). The peptide sequence of OEE2, among gymnosperms, was conserved and displayed diversity from those of algae and other angiosperms. The cDNA length of SsTLP was 896 bps and encoded a predicted 24.7 kDa protein. The isoelectric point of TLP was 8.62. It contained a secretory signal peptide in the N-terminal. The phylogenetic analysis of TLP, based on calculations by the neighbor-joining method, revealed that Sequoia sempervirens had 43-71% homology in the amino acid sequence with the other species ( Figure 5B). The peptide sequence of TLP was conserved among gymnosperms and angiosperms.
The cDNA length of SsRNP was 1123 bps and encoded a predicted 18.5 kDa protein. The isoelectric point of RNP was 8.05 and contained a nuclear  localization signal in the C-terminal. The phylogenetic analysis of RNP, based on calculations by the neighborjoining method, revealed that Sequoia sempervirens shared 49-79% homology in the amino acid sequence with the other species ( Figure 6B). The peptide sequence of RNP, among gymnosperms, was conserved and displayed diversity from those of angiosperms.
Gene expression patterns of up-or down-regulated genes in repeatedly grafted Sequoia shoots To further explore the role of up-or down-regulated genes in phase changes, the expression patterns of SsOEE2, SsRNP, and SsTLP in the two adult shoot cultures, established in 1976 and 1994, respectively, were monitored following repeated grafting ( Figure 8). Both SsOEE2 and SsRNP displayed higher expression levels in the juvenile phase and a lower expression level in the adult phase ( Figure 8). Furthermore, their expression levels also steadily increased with the number of repeated grafts. Specifically, a higher expression level of SsOEE2 was present in the Sequoia shoots during the juvenile and rejuvenated stages, which is consistent with the protein accumulation pattern ( Figure 3). However, the expression level of SsRNP displayed a similar pattern at different stages in Sequoia shoots initiated in 1976 but increased slightly with repeated grafting initiated in 1994. By contrast, the expression level of SsTLP in adult shoots was initially high but progressively decreased with repeated grafting. The expression pattern of SsTLP was opposite to SsOEE2 and decreased with repeated

Discussion
Plant maturation involves a sequence of developmental stages or phases that can be categorized as embryo, juvenile, transitional and adult. Aging in perennials is very complex, and no consensus has been achieved in the general concepts related to this topic [25]. A juvenile phase occurs in all woody plants, which lasts up to 30-40 years in certain forest trees, during which flowering does not occur. The physiology and molecular biology aspects are then considered when listing biochemical markers of maturation in woody plants. These markers occur as part of primary and secondary metabolism (e.g., mineral and carbon nutrition, growth regulators, polyamines, phenolic compounds, and peroxidase activity) and gene expression (e.g., nucleic acids, transcription and proteic synthesis) [25][26][27][28].
The transition leading to adulthood is accompanied by changes in morphological characteristics, such as a loss of adventitious rooting, modification of leaf morphology, an enervated growth rate, and a variation in phytohormone levels [29][30][31]. The process of rejuvenation of the adult shoot apex or inflorescence reversion was identical in planta and well characterized by research that showed some of the key genes involved in the reversion of determinate floral meristem to indeterminate shoot meristem [32,33]. However, most species are less prone to maturational stage reversion because signals from the leaf are less ephemeral even when its initiation is dependent on environmental cues. Several plants undergo rejuvenation when exposed to optimal stimuli, such as gibberellins [34], cytokinin [35], ABA [36], and grafting [16]. Grafting is known to cause phase reversion in woody perennials (e.g., citrus [37], cupressus [38], eucalyptus [38], hedera [39], passiflora [40], pseudotsuga [41], and avocado [14]. The factors triggering maturation and phase transition and how this process is regulated in terms of changes in gene expression, cellular signaling, and metabolism are well known. Unfortunately, most of the established knowledge on phase transition is based on annual Arabidopsis or rice. Much less genetic evidence has been obtained regarding rejuvenation in perennials. We focused on Sequoia proteins and genes that displayed an identical pattern at different developmental stages, and advances in proteomics has helped unravel some of the particular features of rejuvenation in Sequoia.

Identification of OEE2 protein
OEE2 is a core component of the PSII complex in the chloroplast in plants [42]. Evidence has shown that OEE2 in higher plants can only associate with the PSII core complex through OEE1 and OEE3. The oxygen-evolving capacity of PSII decreased to as low as 5-10% in the absence of OEE2 [42]. Based on the literature, the accumulation of OEE2 can be regulated by plant hormones. The abundance of OEE2 increased in the cytokinin-treated moss Physcomitrella patens [43]. The accumulation of OEE2 can also be regulated by abiotic stress. Up-regulated OEE protein levels have also been detected in E. elongatum and Norway spruce during drought stress [44,45] and in rice undergoing salt stress and metal treatment [46]. Furthermore, the accumulation of OEE2 can be regulated by development. The photosynthetic rate and expression level of OEE increased in leaves of maize which roots were colonized by Trichoderma virens [47].
A decrease in photosynthesis with increased aging has been shown to be a significant feature of several trees [27]. Higher rates of photosynthesis and chlorophyll a/b ratios were present during the juvenile and rejuvenate stages, but quantum efficiencies of PSII indicated equivalent effectiveness in the juvenile, rejuvenate, and adult stages. Recent studies have demonstrated RNA interference (RNAi) in PsbP protein, which has greater homology with SsOEE2, with similar results in tobacco [48]. Therefore, the increasing expression level of SsOEE2 is a critical event for restoring the net photosynthesis rate. Furthermore, the photosynthesis rate and expression level of OEE were increased in maize leaves when the roots were colonized by Trichoderma virens [47]. Moreover, the plant in the adult stage is much shorter in the juvenile stage (Figure 1). With PsbP protein knockout in tobacco, the transgenic plant phenotype displayed dramatic growth retardation and also decreases in the chlorophyll a/b ratio and oxygen evolution [49]. This indicated the effect of roots on sink activity and in directing carbon partitioning toward the roots and promoting their development. Therefore, this indicated that active OEE2 is related to nutrient recomposition associated with restoration of rooting and foliar morphology, which were other significant features of Sequoia rejuvenation.
The protein accumulation and gene expression level of SsOEE2 increased through successive grafts in the present study (Figure 3 Figure 8). A positive correlation was found between rejuvenation and protein and transcript abundance of the SsOEE2 gene. This suggested that SsOEE2 may be involved in Sequoia rejuvenation. Specifically, based on the molecular mass, the OEE2 identified in this study appeared to be localized in the chloroplast. Altogether, these results suggest that the chloroplast protein may be involved in Sequoia rejuvenation. Because only plants possess chloroplast, and the fact that rejuvenation found only in plants implicates chloroplast may be one of the regulatory components in plant rejuvenation. In addition, expression of a chlorophyll a/b binding gene was also found to be greater in developing juvenile than in mature foliage [50,51]. It is possible that chloroplast proteins are a key of rejuvenation process in plants. Whether photosynthesis efficiency is altered in Sequoia rejuvenation requires further studies.

Identification of RNP protein
Plant RNP is a class of genes with diverse functions, and the functions of many of these genes are still unknown. The expression of RNP genes is regulated by abiotic stress (e.g., cold, wounds, and hormones such as auxin and abscisic acid [52]. In Arabidopsis, AtGRP7 has been widely studied. AtGRP7 has been reported to promote phase transition in Arabidopsis and is recognized as a novel autonomous pathway component [53]. However, transgenic plants overexpressing AtGRP7 and wildtypes using the ATH1 GeneChip did not show significant changes in FCA or FY under a long-day photoperiod. FLC level was lowered in AtGRP7OX plants. This indicated that RNP may play a multifunctional role in different phase transitions in planta. Many types of RNA-binding proteins (RBPs) play diverse roles in regulating RNA metabolism in a variety of cellular processes [54]. While analyzing the cellular localization of two rice OsRZ2-GFP and OsRZ3-GFP fusion proteins using a confocal microscope, strong GFP signals were observed in the nuclei of roots of OsRZ2and OsRZ3-expressing Arabidopsis plants [55]. Moreover, the growth of roots in OsRZ2-overexpressing grp7 plants was higher than wildtype and grp7 plants, demonstrating that RNPs influence root growth in Arabidopsis plants [56,57]. The restoration of rooting competence is an obvious correlative characteristic of Sequoia rejuvenation (Figure 1). In Figure 6SsRNP displayed greater homology in the amino acid sequence with AtGRP7 and demonstrated a potential role RNPs in root growth.
Although the expression level of SsRNP in the steps of repeated grafting in the present study suggested no correlation between rejuvenation and protein/transcript abundance of the SsRNP gene in Sequoia shoot cultures initiated in 1976 or 1994, SsRNP still appeared to play a possible role in the restoration of rooting competence during rejuvenation. A previous study showed that OEE2 can be phosphorylated by a RNP-mediated wallassociated kinase [58]. Additionally, a phosphopeptide of OEE2 was identified in this study (Table 1). Whether the phosphorylation of OEE2 is involved in the rejuvenation of Sequoia and whether RNP and OEE2 are coordinated during rejuvenation require further studies.
Glycine-rich protein (GRP) has been shown to be abundant in the xylem cell wall and participate in wood formation [59][60][61][62][63]. GRPS were also shown to play an important role in cold adaptation in a biosphere [57,60,62,64]. Therefore, a higher expression level of GRP was present in the later wood, consisting of cells with smaller lumens and thicker cell walls during the chill season [59,60]. Glycine-rich RNA binding proteins (RNP) belong to the Class IV GRPs and are involved in stress adaptation [53,54,65,66]. The expression level of SsRNP in Sequoia shoot cultures initiated in 1976 displayed a similar expression level during the adult, juvenile, and rejuvenile stages but exhibited slightly increased levels during the juvenile and rejuvenile stages compared with the adult stage in shoots initiated in 1994 (Figure 8).

Identification of thaumatin-like protein
Thaumatin-like proteins (TLP) are a class of superfamily proteins participating in several biological processing [67]. The expression level of SsTLP was higher in the adult stage and decreased with successive grafts. The expression level showed an opposite pattern with SsOEE2. TLP is well known to belong to pathogenesisrelated (PR) protein family 5 (PR5) [68]. Most TLPs have strong antifungal activity [69][70][71], but some TLPs have little or no antifungal activity [72,73]. SsTLP displayed high homology with two TLPs-Cryj 3.1 and Cryj 3.2 isolated from Cryptomeria japonica [74]. Several TLPs isolated from Cryptomeria japonica have different expression patterns in plant tissue, with a higher expression level of Cryj 3.1 in the reproductive organ. Furthermore, gSN-TLP isolated from elderberry trees (Sambucus nigra L.) also displayed higher expression levels in older leaves compared with younger leaves [75]. These studies indicate that TLPs could be an adult trait and worthy of future investigations of their function in rejuvenation.

Conclusions
In the present study, three proteins differentially accumulated during different rejuvenation stages in Sequoia sempervirens. These included OEE2, TLP and RNP. The protein abundance and transcript abundance of a phosphoprotein SsOEE2 were consistently lower in Sequoia shoots during the adult stage but increased with repeated grafting. This correlation between rejuvenation and protein and transcript abundance of the genes suggested that SsOEE2 is associated with rejuvenation, and may be involved in Sequoia sempervirens rejuvenation.

Tissue culture
Continuously cultured Sequioa sempervirens (D. Don) Endl. shoots from freshly germinated seedlings served as the juvenile rootstocks, and those from mature trees provided the adult shoot meristems. The continuous cultures were initiated from terminals of juvenile and adult shoots and from scion-growth remnants following regrafting of shoot tips. One centimeter long terminals were subcultured at 6 week intervals in nutrient media containing Murashige and Skoog salts [76], 3% sucrose, 0.2% gelrite, 555 μM myo-inositol, 3 μM thiamine HCl, 2.4 μM pyridoxine HCl, 4.1 μM nicotinic acid, and 26.6 μM glycine. Grafting was performed by inserting the obliquely cut base of a 1.5 cm long shoot terminal into a longitudinal incision made in a rooted 1 cm tall juvenile stem segment. Regrafting was performed at 8 week intervals. Shoot cultures and grafted plants were maintained at 27°C with 16 h daily exposure to 22.5 μmol m -2 s -1 cool-white fluorescent light.

Plant materials
Sequoia shoots at different developmental stages or phases were cultured as previously described [16]. The adult shoot stocks were established by culturing shoot tips excised in 1976 and 1994. Cultures of juvenile shoot cuttings maintained as shoot stocks were established using fresh in vitro germinated seeding. Rejuvenated shoots were obtained by grafting the shoot tips five times from the two adult Sequoia shoots by grafting the adult shoot tips onto juvenile rootstocks. After grafting, shoots were severed and maintained in stocks by subculturing 2-3 cm terminal sections every 2 months in MSbased medium [76].

Protein extraction
This method was modified from a previous study [77]. Two grams of the plant sample were ground with a mortar and pestle in liquid nitrogen. 0.15-0.2 g of the sample was transferred to 10 Eppendorff centrifuge tubes. Each tube was washed with a 10-fold volume of cold 10% trichloroacetic acid/acetone containing, successively, 0.07% 2-mercaptoethanol, cold 80% methanol containing 0.1 M ammonium acetate and 0.07% 2mercaptoethanol, and cold 80% acetone containing 0.07% 2-mercaptoethanol and cold 100% acetone. After each wash, all tubes were centrifuged at 16,000 × g at 4°C for 5 min, and the supernatant was discarded. When all washes were finished, the pellets were air dried at room temperature for at least 10 min, and 0.8 mL 0.1 g -1 of a starting sample of 1:1 phenol/ sodium dodecyl sulfate (SDS) buffer was added, mixed thoroughly, incubated for 10 min, and centrifuged at 16,000 × g at 4°C for 10 min. The phenol phase was pipetted into a 50 ml falcon tube. At least 3-5 volumes of cold methanol containing 0.1 M ammonium acetate and 0.07% 2-mercaptoethanol were added. The tube was stored at -20°C overnight. The falcon tube was centrifuged at 3,000 × g at 4°C for 10 min the next day. The precipitates were washed twice with cold methanol containing 0.07% 2-mercaptoethanol and twice with 80% cold acetone containing 0.07% 2mercaptoethanol. Following each washing step, the falcon tube was centrifuged at 3,000 × g at 4°C for 10 min. Finally, the pellet was resuspended with 16 mL cold 100% acetone and aliquoted into eight 2.0 mL Eppendorff centrifuge tubes. These tubes were then centrifuged at 16,000 × g at 4°C for 5 min, and the supernatant was discarded. The pellet was air dried at room temperature for at least 10 min and dissolved in 200 μL rehydration buffer (6 M urea, 2 M thiourea, 0.5% triton X-100, and a trace of bromophenol blue), and the tubes were stored at -80°C overnight.
Isoelectric focusing and two-dimensional gel electrophoresis A 1.5 mL Eppendorff centrifuge tube was prepared. An appropriate volume of protein sample containing 800 μg of protein was mixed with 3.8 mg dithiothreitol (DTT) and 0.5 μL ampholyte in the tube. Rehydration buffer was added to achieve a final volume of 500 μL. The protein sample was loaded onto the focusing tray and covered with a 17 cm, pH 3-10, immobilized pH gradient (IPG) strip (linear). Active rehydration was performed on a Bio-Rad Protean IEF cell under 50 voltage for 14 h. Isoelectric focusing was performed on the same machine using the following program: 250 V, linear ramp for 20 min; 10,000 V, linear ramp for 3 h; and 10,000 V for a total of 60,000 Vh. Strips were stored at -80°C overnight. The strips were thawed, and equilibration began with SDS equilibration buffer (50 mM Tris-HCl, pH 8.8, 6 M urea, 30% glycerol, 2% SDS, and a trace of bromophenol blue) the following day. The strips were kept in equilibration buffer 1 (SDS equilibration buffer containing 1% DTT) and shaken for 15-20 min. The strips were moved into equilibration buffer 2 (SDS equilibration buffer containing 2.5% iodoacetamide) for 15-20 min. The strips were then layered on 12% acrylamide gel and embedded in the place of agarose. Electrophoresis was performed on a Bio-Rad Protean II xi Cell (step 1: 32 mA for 30 min; step 2: 90 mA for 270 min) and stopped when the indicated blue line reached the bottom of the gel. Gels were then washed with distilled and deionized water once for 5 min and fixed with fixative buffer (50% methanol and 7% acetic acid) overnight. Gels were washed three times with ddH 2 O (10 min each), and then the stained gels were stained with Sypro Ruby for 4-5 h. Stained gel images were scanned using a GE Typhoon 9400. Gel images were compared using SameSpots (Progenesis, UK) software.

In-gel digestion of proteins
Two-dimensional gel spots were excised with cut tips individually. Cut gels were soaked in 50 mM dithioerythreitol (DTE)/25 mM ammonium bicarbonate (pH 8.5) at 37°C for 1 h and soaked in 100 mM iodoacetamide/ 25 mM ammonium bicarbonate (pH 8.5) at room temperature for 1 h in the dark. Gel samples were then soaked in 50% acetonitrile/25 mM ammonium bicarbonate buffer (pH 8.5) three times (15 min each), followed by centrifugation at 10,000 × g for 1 min each. Gel samples were soaked with 100% acetonitrile for 5 min. After removing acetonitrile, gel samples were dried in a Speed Vac for approximately 10 min. One hundred nanograms of trypsin and 40 μL 25 mM ammonium bicarbonate were added to each gel sample and incubated at 37°C for at least 16 h. Fifty μL of 50% acetonitrile/5% trifluoroacetic acid (TFA) was added to each gel sample and sonicated for 10 min. The above step was repeated, combined peptide solutions was dried in a Speed Vac for 90 min. Dried peptide pellets were prepared for liquid chromatography-MS/MS (LC-MS/MS) using Mass Solutions Technology.

Mass spectrometry analyses and protein identification
Peptide samples were analyzed by LC/MS/MS using Q-TOF (Waters, UK) or LTQ Velos (Thermo, US). Peak list files were generated from RAW files. Protein identification was performed using in-house Mascot software (Matrixscience, US) [78]. The mass peak list data were searched against the National Center for Biotechnology Information database using green plant taxonomy. For the Q-TOF dataset, mass tolerance was set to 0.6 D. For the LTQ dataset, mass tolerance was set to 1 D. The missed cleavage site was set to 1. To validate the identified peptides or proteins, Scaffold software (Proteome Software, US) was used to provide confidence interval percentages for the identification of peptides and proteins.

Statistical analyses of gel spot intensity
Statistical analyses were performed using SPSS 16.0 software (SPSS, Chicago, IL, USA) to compare the protein abundance of the gel spots. Significant effects (p < 0.05) in the one-way ANOVA were followed by Duncan's multiple range test.

RNA extraction
One gram of sample was ground in the mortar treated with liquid nitrogen. The sample powder was mixed with 10 mL extraction buffer (10 g polyvinylpyrrolidone, 10 g hexadecyltrimethylammonium bromide, 4.653 g ethylenediaminetetraacetic acid [EDTA], 58.44 g sodium chloride, 0.25 g spermidine, and fresh 200 μl β-mercaptoethanol). After heating the slurry at 65°C in a water bath for 10 min, delaminating was performed by 13,000 rpm centrifugation at 25°C for 15 min. The supernatant was mixed well with chloroform and isoamylalcohol twice, and the total RNA was collected after precipitation with a 1/4-fold volume of 10 M lithium chloride at 4°C overnight.

Rapid amplification of cDNA end and cloning of gene of interest
Total RNA was dephosphorylated by alkaline phosphatase and then decapped by acid pyrophosphatase. The completed mRNA was ligated using a GeneRacer RNA oligo (Invitrogen, Carlsbad, California, USA), and cDNA was formed by reverse transcription. The cDNA was used as the template for cloning the genes of interest using the following steps: denaturation at 94°C for 2 min, followed by 20 cycles of amplification (i.e., 94°C for 30 s for denaturation, 47-66°C for 30 s for annealing, depending on the genes, 72°C for 30 s for elongation)