Comparative transcriptome and metabolome profiling reveal molecular mechanisms underlying OsDRAP1-mediated salt tolerance in rice

Integration of transcriptomics and metabolomics data can provide detailed information for better understanding the molecular mechanisms underlying salt tolerance in rice. In the present study, we report a comprehensive analysis of the transcriptome and metabolome of rice overexpressing the OsDRAP1 gene, which encodes an ERF transcription factor and was previously identified to be conferring drought tolerance. Phenotypic analysis showed that OsDRAP1 overexpression (OE) improved salt tolerance by increasing the survival rate under salt stress. OsDRAP1 affected the physiological indices such as superoxide dismutase (SOD), catalase (CAT) and malondialdehyde (MDA) to enhance redox homeostasis and membrane stability in response to salt stress. Higher basal expression of OsDRAP1 resulted in differential expression of genes that potentially function in intrinsic salt tolerance. A core set of genes with distinct functions in transcriptional regulation, organelle gene expression and ion transport were substantially up-regulated in the OE line in response to salt stress, implying their important role in OsDRAP1-mediated salt tolerance. Correspondingly, metabolome profiling detected a number of differentially metabolites in the OE line relative to the wild type under salt stress. These metabolites, including amino acids (proline, valine), organic acids (glyceric acid, phosphoenolpyruvic acid and ascorbic acid) and many secondary metabolites, accumulated to higher levels in the OE line, demonstrating their role in salt tolerance. Integration of transcriptome and metabolome analysis highlights the crucial role of amino acids and carbohydrate metabolism pathways in OsDRAP1-mediated salt tolerance.


Results
Phenotype differences between OsDRAP1 OE transgenic and WT lines under salt stress conditions. The OsDRAP1 gene (LOC_Os08g31580) is 1651 bp in length with a 497 bp 5′UTR, a 311 bp 3′UTR, and a 843 bp coding region composed of one exon. The OsDRAP1 gene encodes a polypeptide of 280 amino acids with an AP2 domain (103-166 aa) and the transactivation activity is located primarily in the region of 207-280 aa 24 .
To investigate the molecular function of OsDRAP1 in response to salt stress, rice lines overexpressing OsDRAP1 under UBI promoter were generated as described in our previous study 24 . Sucrose phosphate synthase gene (OsSPS, LOC_Os11g12810) as the reference gene and hygromycin resistance gene (HPT) in plasmid pCUbi1390 was used as a marker gene for quantifying transgene copy numbers analysis by qRT-PCR (quantitative Real Time Polymerase Chain Reaction) 25 . The correlation coefficients R 2 are 0.991 and 0.979 respectively, indicating that the data is reliable ( Supplementary Fig. 1a). The copy number of transgenic lines (T0 generation) was analyzed by using the equation: N × 10 (−0.303Ct1+12.397) = 10 (−0.368Ct2+12.806) (Supplementary Table S1). Three homozygous OsDRAP1-overexpressing lines (T3 generation) with one copy number were obtained with the criterion that all the selected 12 plants from each OE lines were positive by PCR (Polymerase Chain Reaction) validation ( Supplementary Fig. 1b). The qRT-PCR results showed that the expression of OsDRAP1 was upregulated in three OE lines to various extents compared with wild type (WT) under the normal growth conditions ( Supplementary Fig. 1c). Three OE lines (OE-6, OE-7, OE-9) were evaluated for phenotype performance under salt stress imposed by exposing three-leaf-stage seedlings of the OE lines and WT to 120 mM NaCl. After 7 days, three transgenic lines all exhibited enhanced salt tolerance relative to WT with significantly higher seedling survival rates (Fig. 1). Besides, the survival rates of OE lines were also significantly higher than WT under 150 mM NaCl treatment ( Supplementary Fig. 2). These results indicated that overexpression of OsDRAP1 improved salt tolerance in rice.

Physiological indices of transgenic and WT lines under normal and salt stress conditions.
To determine how OsDRAP1 affects the physiological indices in response to salt stress, we comparatively analyzed the contents of malondialdehyde (MDA), soluble sugar, glutathione (GSH), and ascorbic acid (AsA), the activities of superoxide dismutase (SOD) and catalase (CAT) in OE-7 and WT plants under normal and salt stress conditions. The MDA content in the OE-7 line was significantly lower than that of WT after 12 h of salt stress, even though there was no remarkable difference between the OE and WT lines after 24 h of salt stress (Fig. 2a), indicating that the membrane system may be more stable in OE-7 than that in WT at early stages of salt stress. The soluble sugar content was significantly higher in OE-7 than in WT after 24 h of salt stress treatment (Fig. 2b). The activities of SOD and CAT, the levels of GSH and AsA were also significantly higher in OE-7 relative to WT under salt stress (Fig. 2c,d,e,f), indicating that the reactive oxygen species (ROS) scavenging ability was remarkably improved in the OE-7 line in response to salt. All these results suggested that OsDRAP1 overexpression improved salt tolerance in transgenic rice plants by maintaining membrane integrity and enhancing the ROSscavenging system. Genes with more than two-fold change in expression in the OE line relative to WT with False discovery rate (FDR) < 0.01 were defined as the differentially expressed genes (DEGs); a total of 293 and 163 genes were found to be significantly up-and down-regulated, respectively (Supplementary Table 2). Gene Ontology (GO) enrichment analysis revealed that the up-regulated genes were enriched in the protein modification process (GO:0006464), programmed cell death (GO:0012501) and response to stress (GO:0006950) terms (Supplementary Table 3). GO enrichment analysis showed that the down-regulated genes (Supplementary Table 4) were related to photosynthesis (GO:0015979) and generation of precursor metabolites and energy (GO:0006091), indicating that photosynthesis and energy metabolism are repressed in the OE transgenic plants.
To investigate common sequence motifs in the genes up-regulated in the OE-7 line relative to WT under normal condition, we identified the cis-regulatory elements in the 2-kb regions upstream of the 293 up-regulated genes. There were several highly abundant elements such as the ATCTA, CBFHV (RYC GAC ), DRECRTCOREAT, and GCCCORE motifs (Supplementary Table 5). A total of 85.0% the genes contain 1-12 copies of the ATCTA motif, which has been identified as the ethylene responsive factor (ERF) binding element 26 , and around 75.4% of genes harbor 1-12 copies of the CBFHV motif, which is a dehydration-responsive element (DRE) 27 . More than 60% of the genes have at least one copy of the GCCCORE or DRECRTCOREAT motif; the GCCCORE element has an important role in regulating jasmonate-responsive gene expression 28 , and DRECRTCOREAT is a cis-element known to be recognized by AP2/ERF proteins 29 . All the enriched cis-elements could be involved in OsDRAP1-mediated regulation of downstream gene expression in response to salt stress.  Table 7).
Among the up-regulated genes, eleven genes encoding ERF transcription factors including OsERF2, OsERF3, OsERF61 and OsERF62, and other TFs such as three bZIPs (OsbZIP09, OsbZIP23 and OsbZIP62), three MYBs (OsMYB2, OsMYB30 and OsPHR3) and four NACs (OMTN1, OsNAC6, OsNAC10, OsNAC31) were highly induced in the OE line (Supplementary Table 6), suggesting that these TFs are involved in the OsDRAP1mediated regulation of transcription in response to salt stress.
Two sodium exchanger genes, OsCAX1b and OsNHX2, a set of 87 genes encoding pentatricopeptide repeat (PPR) domain proteins and nine genes encoding proton-dependent oligopeptide transporter (POT) proteins were found to be significantly up-regulated in the OE line. These two classes (PRR and POT) of genes are reported to be  www.nature.com/scientificreports/ involved in a wide range of biological functions [30][31][32] . Six chitinase genes and three genes encoding mitochondrial transcription termination factors (mTERFs) were also remarkably induced in the OE line; chitinases and mTERFs are extensively involved in biotic and abiotic stress response 33,34 . All these results suggest that over-expressing the OsDRAP1 gene affected many TFs and genes related to stress response under salt stress conditions. 21 genes were randomly selected to verify the accuracy of the transcriptome data with qRT-PCR. The primers for qRT-PCR were listed in Supplementary Table 8, and the date were presented in Supplementary Table 9. There was a high correlation (R 2 = 0.96) between expression levels of 21 genes determined from qRT-PCR and transcriptome data, showing that the transcriptome data were accurate and reliable ( Supplementary Fig. 3).

Metabolite profiling of the OE and WT lines in response to salt stress.
To globally profile the metabolite contents in leaves of the OE-7 line and WT, we performed LC/MS metabolome analysis and detected a total of 3894 biochemicals, of which 372 were known metabolites, in the two lines under normal and salt stress conditions. The metabolites covered several categories such as amino acids, carbohydrates, nucleotides, peptides, hormones and secondary metabolites (Supplementary Table 10). To reduce the dimensionality of the data and visualize the relationship among samples, we performed PCA, and the first principal component (PC1) explained 25.5% of the total variation, while the second principal component (PC3) explained 8.7% of the variation across the data set (Fig. 3). A plot of the PC1 and PC3 scores showed a clear separation in PC1 between the different treatments. The OE-7 samples and WT samples were separated by PC3 under salt stress. Compared with that under normal condition, the differences of metabolites between OE and WT samples were further increased under salt stress. This indicates changes in the metabolite profiles caused by the salt treatment as well as by the differences between the OE-7 and WT in response to salt stress.
We further performed a comparative analysis to identify the differential metabolites between the OE and WT lines as described in the Materials and Methods. Under normal growth conditions, there were 38 and 3 www.nature.com/scientificreports/ metabolites detected to be up-and down-regulated, respectively, in the OE line relative to WT (Supplementary  Table 11). The three down-regulated metabolites were 5-amino-6-(5′-phosphoribosylamino) uracil, akuammine and paucine, and the 38 up-regulated metabolites consisted of 22 unmapped metabolites and 16 secondary metabolites including eseramine, glycerylphosphorylethanolamine, pyridaben, satratoxin H, streptidine 6-phosphate, calactin and aminoadipic acid. These results suggested that overexpression of OsDRAP1 could elicit significant metabolite changes, especially the accumulation of secondary metabolites, under normal growth condition. Under salt stress condition, we detected 426 differential metabolites (267 up-regulated, 159 down-regulated) in the OE line compared with WT. Of these 267 up-regulated metabolites, only 86 were known metabolites, including amino acids (proline, valine), organic acids (glyceric acid, dehydroascorbic acid, azelaic acid, phosphoenolpyruvic acid), and many secondary metabolites (Supplementary Table 12). Amino acids such as proline and valine, which are involved in osmotic stress tolerance, are very important for plant abiotic stress tolerance 35,36 . The organic acids dehydroascorbic acid, azelaic acid, glyceric acid, and aminoadipic acid were previously reported to be highly involved in plants responding to environmental stress 37-40 . Correlation between the transcriptome and metabolome data. To explore the correlation between gene expression and metabolite accumulation, we assessed the integrated metabolome and transcriptome profiling based on a global analysis of the covariance structure of the data sets. We detected several correlations between the levels of metabolites and genes using Pearson correlation coefficient analysis as described in the Materials and Methods. The significantly enriched pathways (those associated with significantly accumulated metabolites) in the OE line are shown in Table 1.
There were 17 pathways highly enriched in the OE line compared with WT under salt stress conditions (Table 1), including several amino acid metabolism pathways such as glycine, serine and threonine metabolism (8 genes), cysteine and methionine metabolism (13 genes), arginine and proline metabolism (4 genes), and tyrosine metabolism (7 genes). A few carbohydrate metabolism pathways including the pentose phosphate pathway (6 genes), glyoxylate and dicarboxylate metabolism (9 genes) and taurine and hypotaurine metabolism (5 genes) were highly enriched; and biosynthesis of amino acids was also evidently enriched (17 genes) in OE line (Supplementary Table 13, Supplementary Fig. 4).

Discussion
In a previous study, we showed that overexpression of the gene OsDRAP1 in transgenic rice plants enhances drought tolerance without retarding growth 24 . OsDRAP1 is a transcription factor in the AP2/ERF family, members of which are involved in development and abiotic/biotic stress signal transduction in higher plants 41 . In the present study, we combined transcriptome and metabolome analysis to dissect the physiological, gene expression and metabolites changes of OsDRAP1-mediated salt stress tolerance. Our result showed that overexpression of OsDRAP1 could improve rice salt tolerance. In addition, a number of genes were either up-or down-regulated, and a few metabolites were differentially produced in a transgenic OsDRAP1 OE line relative to WT under normal   43 . ROS scavenging enzymes and antioxidants such as GSH and AsA play an important role in reducing oxidative stress 44,45 . In the present study, the activities of ROS enzymes (SOD and CAT) and the contents of GSH and AsA were significantly higher in the OsDRAP1 OE line than in WT under salt stress condition (Fig. 2), indicating that overexpression of OsDRAP1 could strongly enhance the ability of rice plants to maintain redox balance and protect against oxidative stress under salt stress. Meanwhile, the increased soluble sugar and decreased MDA contents in the OE line under salt stress demonstrated that OsDRAP1 is involved in maintaining cell membrane integrity and energy supply under salt stress; a higher concentration of soluble sugars can provide carbohydrate energy in plants under various unfavorable environmental conditions 46 and MDA is an indicator of oxidative damage in the plant cell membrane induced by stress 47,48 .
To explore the downstream genes regulated by overexpression of OsDRAP1, we first compared the whole genome gene expression profiles in the OE line and WT under normal growth condition. The expression of a set of genes with distinct functions was found to be up-regulated by overexpression of OsDRAP1 (Supplementary  Table 2). These genes include six WAK family genes (OsWAK34, OsWAK60, OsWAK38, OsWAK76, OsWAK40 and OsWAK59), which are involved in cell wall, signaling transduction and environmental stress response [49][50][51] . In particular, a few TFs functioning in transcriptional regulation in response to environmental stresses were synergistically up-regulated in the OE line. For example OsDREB1A and OsDREB1B were previously characterized as cold stress regulators 29,52 ; OsBIERF3 and OsERF105 were reported to be involved in rice development and osmotic stress tolerance 53,54 ; and several WRKY genes such as OsWRKY62, OsWRKY67 and OsWRKY69 are highly involved in basal biotic stress response [55][56][57][58] . All these genes with evidently higher expression in the OE line relative to WT could play a crucial role in OsDRAP1-mediated stress tolerance.
Further comparative analysis of the transcriptomes after 24 h of salt stress revealed more genes differentially expressed between the OE line and WT. Several functional categories of genes including transcription regulation, photosynthesis, cell wall process, peptide transport and oxidoreductase activity were evidently up-regulated in OE line under salt stress, showing diverse functional genes are involved in salt stress response.
Fourteen TF genes including ERFs, bZIPs, MYBs and NACs had significantly higher expression in the OE line compared with WT under salt stress, implying that they coordinately regulate transcription in response to salt. For example, OsERF61, OsERF63, OsERF2 and OsERF3, which are members of the same AP2/ERF subfamily, were up-regulated, and these four ERF genes were previously found to be important factors regulating growth and metabolism in plants in response to abiotic stress [59][60][61] . Members of a large family of NAC TFs such as OsNAC1, OsNAC6, OsNAC10 and ONAC131 were all up-regulated in the OE line. Overexpression of OsNAC10 in rice roots enhances drought tolerance and grain yield 62 , and ONAC131 plays and important role in rice disease resistance by regulating the expression of other defense-and signaling-related genes 63 . Higher expression of these TFs likely contributes to OsDRAP1-mediated salt stress tolerance in the OE line.  www.nature.com/scientificreports/ PPRs and mTERFs are involved in the regulation of organelle gene expression, and play crucial roles in plant growth and development by tightly coordinating nuclear gene expression 34,64 . Several PPR genes in rice were found to be involved in salt stress response 65 , and a few mTERFs in Arabidopsis were found to increase adaptability to environmental changes 66,67 . In the present study, a large set of PPR genes and three mTERF genes were up-regulated in the OE line, showing that regulation of organelle gene expression is closely associated with salt tolerance.
Eleven genes encoding POT proteins and two sodium exchanger genes (OsCAX1b and OsNHX2) were upregulated in the OE line relative to WT, suggesting their role in osmotic regulation. POT proteins function in potassium translocation and shoot potassium homeostasis, regulating nutrient balance, growth, and stress tolerance in plants 32,68 , while OsCAX1b and OsNHX2 are reported to be involved in sodium/calcium transport 69,70 . These up-regulated POT genes and sodium exchanger genes might increase the potassium content and decrease the sodium level to help maintain ion homeostasis in the OE line under stress.
The fluctuation in unique metabolites is considered essential for plant abiotic stress response; these metabolites regulate osmosis, energy balance, ROS homeostasis and signal transduction 71,72 . We first compared the metabolite contents between the OE line and WT under normal conditions and found that overexpression of OsDRAP1 resulted in the accumulation of secondary metabolites. Aminoadipic acid, which is related to lysine metabolism, is an important signaling amino acid that regulates plant growth and responses to the environment 73 ; calactin and satratoxin H are involved in program cell death, which has been widely implicated in biotic stress tolerance 74,75 ; and glycerylphosphorylethanolamine is associated with osmotic stress tolerance 76 . These accumulated metabolites could have priming effect of OsDRAP1 OE line for stress tolerance.
The results presented in this study revealed alterations in large numbers of metabolites in OE relative to WT under salt stress, consistent with the findings of a previous study 19,20 . Amino acids including proline and valine accumulated under salt stress, suggesting that these metabolic pathways are related to OsDRAP1-mediated salt tolerance. Levels of several organic acids such as glyceric acid, azelaic acid, phosphoenolpyruvic acid and dehydroascorbic acid were higher in the OE line under stress. Up-regulation of glyceric acid was previously found to contribute to ROS scavenging under stress 77 . Increased levels of azelaic acid, which is involved in lipid peroxidation, enhance biotic stress tolerance 78 . A higher AsA content contributes to the protection of thylakoid membrane lipids from oxidation in stressed plants 40 . All these results indicate that the accumulation of specific amino acids and organic acids plays an important role in salt tolerance of the OE line.
Further correlation analysis revealed that several metabolism pathways were specifically enriched in the OsDRAP1 OE line, illustrating the changes in metabolism associated with salt stress response. Especially a few amino acid metabolism pathways related to salt stress were determined. Amino acids are not only indispensable for protein synthesis but also play important cellular functions. Arginine and proline have positive effects on membrane integrity along with adaptive roles in mediating osmotic adjustment in plants under stress conditions 79 . The cysteine and methionine metabolism pathway is very important for the induction of the alternative oxidase pathway in poplar response to salt stress 80 . The valine, leucine and isoleucine degradation pathways play critical roles in the regulation of energy homeostasis and nutrition metabolism in plants under environmental stress 81 , and cyanoamino acid metabolism is positively related to biotic stress tolerance 82 . Additionally, carbohydrate metabolism homeostasis plays an important role in plant stress response, and the pentose phosphate pathway is essential for the ROS scavenging activity of the ascorbate-glutathione redox cycle in response to plant stress 23 . The high levels of glyoxylate and dicarboxylate metabolites in the OE line suggests that they play roles in energy supply and protection against environmental stress, which is consistent with the findings of a previous study 83 . All these results revealed that a complex molecular mechanism is involved in salt stress tolerance in the OE line.

Conclusion
Overexpression of OsDRAP1 improves rice salt stress tolerance by enhancing redox homeostasis and membrane stability. Combined transcriptome and metabolome analysis revealed a complex molecular mechanism underlying OsDRAP1-mediated salt stress tolerance; higher expression of OsDRAP1 caused differential expression of genes and metabolites, which could be involved in basal stress tolerance of the OE line. A number of genes with diverse functions including transcription regulation, organelle gene expression and ion transportation were expressed at higher levels in the OsDRAP1 OE line under salt stress; metabolites such as amino acids, organic acids and secondary metabolites also accumulated to high levels in the OE line under salt stress, suggesting their important roles in salt tolerance. Combined transcriptome and metabolome analysis highlights the crucial role of amino acids and carbohydrate metabolism in OsDRAP1-mediated salt tolerance.

Materials and methods
Plant materials, growth conditions and stress treatments. To evaluate salt tolerance phenotypes, Oryza sativa L. cv. Nipponbare, a japonica rice cultivar, was used to generate OsDRAP1 overexpression (OE) lines by Agrobacterium-mediated transformation 25 . The background parent Nipponbare as wild-type (WT) control plants and the OE lines were grown in Yoshida nutrient solution 84  www.nature.com/scientificreports/ Physiological analysis. One OE line (OE-7) and WT at the three-leaf stage were subjected to salt treatment by transferring to nutrient solution containing 120 mM NaCl for 24 h. The leaves from three plants were collected and analyzed for soluble sugar content, malondialdehyde (MDA) content, superoxide dismutase (SOD) activity, catalase (CAT) activity, and contents of antioxidants, namely glutathione (GSH) and ascorbic acid (AsA). MDA content and total soluble sugar concentrations were measured using the method previously described by Song et al. 85 , and the SOD and CAT activities were estimated as described by Ouyang et al. 86 . GSH 87 and AsA 88 contents were calculated as described previously. All data were analyzed using the Student's t-test with three replicates.
Transcriptome analysis. Salt stress was induced by transferring plants to nutrient solution containing 120 mM NaCl as above described. The aerial parts of seedlings at the three-leaf stage were harvested at 24 h after salt treatment and normal growth condition (three replicates for each sample). Total RNA was extracted from sampled leaves by using TRIzol Reagent (Invitrogen, USA). RNA concentration was measured using a Nan-oDrop 2000 (Thermo Scientific). RNA integrity was assessed using the RNA Nano 6000 Assay Kit of the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, USA). RNA sequencing and assembly were performed by Biomarker Technologies Corporation (Beijing). Analysis of differential expression between two samples was performed using EBseq. False discovery rate (FDR) < 0.01, |log2 (fold change)|≥ 2 was set as the threshold for significant differential expression. Gene Ontology (GO) enrichment analysis of the differentially expressed genes (DEGs) was implemented by the GOseq R package based on the Wallenius non-central hyper-geometric distribution 89 , which can adjust for gene length bias in DEGs. The expression levels of the selected DEGs were further confirmed by qRT-PCR. Negative-ion ESI mode provided better sensitivity and more observable peaks in total ion chromatograms than positive-ion mode; therefore, we used negative-ion mode as the main profiling strategy, and used positiveion mode to obtain additional information for metabolite identification as previously described 93 .

Metabolome analysis.
The fold change, the P value from Student's t test and the variable importance in the projection (VIP) value from the orthogonal projections to latent structures-discriminant analysis (OPLS-DA) model were combined to identify the differential metabolites, using a threshold of with |fold change|> 2, P value < 0.05 and VIP > 1.
Principal component analysis. Principal component analysis (PCA) of metabolites was performed using the R package XCMS (version 3.2). The package prcomp was used for the analysis and factoextra was used for ggplot2-based visualization. The abscissa of the PCA score chart represents the first principal component, namely PC1, and the ordinate represents the second principal component, namely PC3.

Metabolite-transcript correlation analysis.
Pearson correlation coefficients (PCCs) were calculated for metabolite and transcript profile data as described by Cho et al. 94 and Reem et al. 95 . The mean of all the biological replications of the OE line and WT for metabolite data and the normalized mean value of each gene for transcriptome data were evaluated. The coefficients were calculated using the log 2 (fold change) values of each metabolite and gene. PCC > 0.80 and PCC P value (PCCP) < 0.05 were used as the criteria for screening.
Quantitative RT-PCR analysis. Total RNA was extracted from rice leaves using Trizol (Invitrogen) and the Direct-zol RNA MiniPrep Kit (Zymo Research) according to the manufacturer's instructions. The FastKing gDNA Dispelling RT SuperMix (TIANGEN Biotech Co., Beijing) was used for the synthesis of first-strand cDNA. qRT-PCR using the primers listed in Supplementary Table 8 was carried out with SuperReal PreMix Plus (SYBR Green) (TIANGEN Biotech Co., Beijing) following the manufacturer's instructions. qPCR was performed with an Applied Biosystems 7500 thermocycler (Thermo Fisher Scientific, USA) under the following cycling conditions: 95 °C for 15 min, followed by 40 cycles of 95 °C for 10 s and 60 °C for 32 s. The expression of rice Actin1 (LOC_Os03g50890) was used as an internal control to normalize target gene expression. The relative expression of each gene was calculated based on the 2 −△△CT method 96  www.nature.com/scientificreports/