Effects of drought and salt-stresses on gene expression in Caragana korshinskii seedlings revealed by RNA-seq

Background Drought and soil salinity are major abiotic stresses. The mechanisms of stress tolerance have been studied extensively in model plants. Caragana korshinskii is characterized by high drought and salt tolerance in northwestern China; unique patterns of gene expression allow it to tolerate the stress imposed by dehydration and semi-desert saline soil. There have, however, been no reports on the differences between C. korshinskii and model plants in the mechanisms underlying their drought and salt tolerance and regulation of gene expression. Results Three sequencing libraries from drought and salt-treated whole-seedling- plants and the control were sequenced to investigate changes in the C. korshinskii transcriptome in response to drought and salt stresses. Of the 129,451 contigs, 70,662 (54.12 %) were annotated with gene descriptions, gene ontology (GO) terms, and metabolic pathways, with a cut-off E-value of 10−5. These annotations included 56 GO terms, 148 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, and 25 Clusters of Orthologous Groups (COG). On comparison of the transcriptomes of the control, drought- and salt-treated plants, 1630 and 1521 contigs showed significant differences in transcript abundance under drought and salt stresses. Compared to the differentially expressed genes (DEGs) in drought- or salt-treated Arabidopsis in the database, 542 DEGs in drought-treated C. korshinskii and 529 DEGs in salt-treated samples were presumably unique to C. korshinskii. The transcription profiles revealed that genes related to transcription factors, protein kinases, and antioxidant enzymes are relevant to the tolerance of drought and salt stress in this species. The expression patterns of 38 randomly selected DEGs were confirmed by quantitative real-time PCR and were essentially consistent with the changes in transcript abundance identified by RNA-seq. Conclusions The present study identified potential genes involved in drought and salt tolerance in C. korshinskii, as well as many DEGs uniquely expressed in drought- or salt-treated C. korshinskii samples compared to Arabidopsis. To our knowledge, this study is the first exploration of the C. korshinskii transcriptome under drought and salt conditions, and these results will facilitate the discovery of specific stress-resistance-related genes in C. korshinskii, possibly leading to the development of novel plant cultivars through genetic engineering. Electronic supplementary material The online version of this article (doi:10.1186/s12864-016-2562-0) contains supplementary material, which is available to authorized users.


Background
Environmental stresses or abiotic constraints such as salinity, drought, heavy metals and low temperature are major obstacles to plant survival and development in many places worldwide. Plant responses and adaptations to abiotic stresses are complex [1]. Many cellular and physiological processes such as signal perception and transduction [2,3], regulation of transcription, membrane trafficking, energy metabolism [4], and protein phosphorylation and dephosphorylation are involved in plants' toleration of excessive salt and drought stresses. Several studies on drought and salt stresses using microarrays, transcriptome sequencing, and microRNA sequencing [5][6][7] have identified stress-inducible genes involved in water transport (aquaporins), ion transport (plasma membrane (PM) H + -ATPase, Na + /H + exchanger or Na + /H + antiporter), cellular membrane integrity (proline, glycine betaine, mannitol), scavenging of free oxygen radicals (superoxide dismutase, catalase and peroxidase), and protecting macromolecules (late embryogenesis abundant proteins and chaperones). Other proteins, such as regulatory proteins (transcription factors, protein kinases, protein phosphatases, and calmodulin-binding proteins), were found to be involved in signal transduction [8,9]. ABA-dependent and ABA-independent signaling pathways were induced by plants in response to stress [10]. The ABA-dependent transcription factors include MYC/ MYB and ABA-responsive element binding/ABA-binding factor (AREB/ABF). The ABA-independent transcription factors include dehydration-responsive element-binding proteins (DREB), C-repeat/drought-responsive element (CRT/DRE), and CRT/DRE-binding factor (CBF) [11,12]. Other transcription factors responding to abiotic stress conditions are zinc-finger proteins [13], basic-domain leucine-zipper (bZIP) [14], WRKY [15], and NACs [16].
The development of next-generation sequencing technology (NGS) has led to major progress in understanding plant responses to drought and salt stress [17][18][19]. Some recent efforts include tissue-selective signaling and hormone crosstalk in response to salt and osmotic stresses and, recently, the dynamics of transcriptional networks that are functional at the seedling stage in response to water stress in Gossypium arboreum [20]. The transcriptional characteristics of genes related to the reactive oxygen species (ROS)-scavenging system are important to the salt tolerance of Reaumuria trigyna [21]. Further, genome-wide gene-expression profiling was used to examine the molecular mechanism and physiological response of Gossypium herbaceum to drought [22]. Thumma et al. [23] used RNA sequencing (RNA-seq) to study the effect of water stress on gene expression in Eucalyptus camaldulensis seedlings derived from three natural populations. RNA-seq technology and its applications have dramatically accelerated plant-genomics research, including high-throughput sequencing of non-model-plant transcriptomes and large-scale genome-wide expression analysis.
Caragana korshinskii is a leguminous shrub with highly developed root systems and strong stress tolerance [24,25] that is commonly found in arid and semiarid lands in northwestern China and Mongolia [26,27]. Therefore, C. korshinskii is widely planted in China to prevent desertification and improve vegetation coverage. The effect of drought or salt stress on plant gene expression has been intensely studied in numerous species including Arabidopsis [14], rice [28], maize [15], wheat (Triticum aestivum L.) [29], and Sorghum bicolor [30]. In C. korshinskii, some investigations have focused on growth properties, nutritive characteristics, evapotranspiration, and genetic diversity [31][32][33], and several drought-and salt-stress-response genes have already been characterized [34][35][36], but there have been no studies on the stresssignaling pathways. This lack of molecular studies is a significant obstacle to understanding the molecular mechanisms underlying drought or salt adaptation in C. korshinskii.
In this study, we generated transcriptome datasets to explore the salt-and drought-tolerance mechanisms of C. korshinskii using the Illumina HiSeq™ 2000 platform. The main objectives of this study are to identify genes showing transcriptional differences and to identify their putative functions, as well as to identify genes putatively unique to C. korshinskii compared to the model plant Arabidopsis and describe the patterns in their transcript abundance under salt-and drought-treated conditions. The assembled, annotated transcriptome data and differential expression profiles will facilitate further genetic and genomics studies on the molecular mechanisms of salt and drought tolerance in C. korshinskii and most likely in other leguminous plants.

Illumina sequencing and de novo assembly
In total, 120.26 million raw reads were generated from control samples, 39.78 million raw reads were generated from drought-treated samples, and 39.90 million raw reads were generated from salt-treated samples. We obtained a total of 6.66 gigabase pairs (Gbp) with an average GC content of 47.66 %. We obtained approximately 199.94 million total reads, of which approximately 182.93 million passed the Illumina quality filtering, yielding a quality rate of over 91.49 %. This result indicated that the read number and quality were high enough for further analysis (the number of clean reads for each sample is shown in Additional file 1). The 182.93 million high-quality reads were assembled into 129,451 contigs with an N50 length of 1332 bp and an N90 length of 283 bp (Additional file 1, Table 1 After Blast2GO analysis, the contigs were classified into 56 terms from three ontologies involved in cellular components, biological processes, and molecular function ( Fig. 1). In each of the three main GO classifications (biological process, cellular component, and molecular function), "metabolic process", "cell" or "cell part", and "catalytic activity" were dominant among the returned terms. Based on a comparison to the KEGG database, 21,104 (39.57 %) of the 70,491 annotated contigs had significant matches and were thus assigned to 148 KEGG pathways. The most-represented pathways were "metabolic pathways" (4425 contigs), "biosynthesis of secondary metabolites" (1967 contigs), "ribosome" (660 contigs), "purine metabolism" (425 contigs), and "plant hormone signal transduction" (423 contigs). A total of 94,820 contigs showed a COG classification. Among the 25 COG categories, the cluster for "General function prediction only" (30,370,32.03 %) was the largest, followed by "Posttranslational modification, protein turnover, chaperones" (7851, 8.28 %), "Signal transduction mechanisms" (7739, 8.16 %), "Carbohydrate transport and metabolism" (4411, 4.65 %), and "Transcription" (4163, 4.39 %). The categories "Cell motility" (44, 0.000464 %), and "Nuclear structure" (244, 0.26 %) had the fewest corresponding genes (Fig. 2, Additional file 6).
Analysis of differential expression of assembled C. korshinskii transcripts under drought and salt treatments Compared with the control, 1630 and 1521 contigs were classified as differentially expressed genes (DEGs) in drought-and salt-treated C. korshinskii, respectively (Additional file 7). In the drought-treated seedlings, 731 contigs showed increased transcript abundance and 899 contigs showed decreased transcript abundance. In the salt-treated seedlings, 729 transcripts were upregulated and 792 transcripts were downregulated (Figs. 3 and 4).
Venn-diagram analysis revealed the specificity and overlap among differentially expressed genes (Fig. 4). Among 1630 and 1521 differentially expressed genes, numerous genes overlapped in expression in response to the salt and drought treatments. For example, in droughtand salt-treated plants, 148 upregulated transcripts overlapped and 368 downregulated transcripts overlapped. In addition, 1005 of 1521 contigs appeared only in salt-treated samples, while 1111 of 1630 contigs were expressed only in drought-treated samples.
GO annotation and statistical analyses demonstrated that 1844 tested DEGs (some contigs have more than one GO annotation) were classified into three GO ontologies and 60 terms in drought-treated plants. Of these DEGs, 843 were associated with biological processes, 419 with cellular components, and 612 with molecular function. Among these GO categories, "oxidoreductase activity" (GO:0016491,168 DEGs), "oxidation-reduction process" (GO:0055114,146 DEGs), and "response to stress" (GO:0006950, 149 DEGs) were significantly enriched among DEGs compared to the whole transcriptome background. In the molecularfunction category, 2640 tested DEGs (some contigs have more than one GO annotation) were enriched in 63 terms in salt-treated samples. The overwhelming majority of the genes with differential expression between the control and salt-treated conditions were also related to "oxidoreductase activity" (GO:0016491,186 DEGs), "oxidation-reduction process" (GO:0055114,169 DEGs), and "response to stress" (GO:0006950,131 DEGs). Approximately 41.7 % of 60 GO terms in drought-treated C. korshinskii were also found in salt-treated C. korshinskii, such as oxidoreductase activity, oxidation-reduction process,  A total of 731 DEGs that were significantly upregulated in drought-treated C. korshinskii were categorized into 46 GO terms based on sequence similarity. Accordingly, 899 significantly downregulated DEGs were categorized into 61 GO terms (Fig. 5a, Additional file 9). In addition, 6 GO terms were found in two samples, including extracellular region, monooxygenase activity, serine-type endopeptidase inhibitor activity, iron ion binding, electron carrier activity and heme binding. Interestingly, 25 DEGs were classified into the category "iron ion binding" and 21 DEGs into the category "electron carrier activity" out of 731 upregulated DEGs in drought-treated C. korshinskii. Another 22 DEGs were classified into the category "iron ion binding" and 27 DEGs into the "electron carrier activity" among 899 downregulated DEGs in drought-treated C. korshinskii (Additional file 9). A total of 729 DEGs that were upregulated in salt-treated C. korshinskii were assigned to GO classes with 19 functional terms. Accordingly, 792 DEGs that were significantly downregulated were categorized into 72 GO terms (Fig. 5b, Additional file 9). Generally, 6 GO terms, such as structural constituent of ribosome, extracellular region, ribosome, translation, heme binding, and oxidation-reduction process, were found in both samples. Furthermore, 56 DEGs were classified into the category "oxidation-reduction process" in 729 upregulated in salt-treated C. korshinskii, and 95 DEGs were classified into the category "oxidation-reduction process", out of 792 DEGs in salt-treated C. korshinskii, which were are valuable for studying salt-treated C. korshinskii (Additional file 9).

Functional genes related to protein kinase
We identified 131 differentially expressed transcripts predicted to encode protein kinases (Fig. 7, Additional file 13), of which 80 and 20 contigs encoding receptorlike protein kinase were differentially expressed in drought-treated and salt-treated libraries, respectively. Among these receptor-like protein kinases, 24 and seven contigs predicted to encode leucine-rich repeat (LRR) receptor-like protein kinase were differentially expressed in response to drought stress and salt stress. Among these contigs, Cko_contig_70567, Cko_contig_105960 and Cko_contig_30368 were upregulated more than 32-fold under drought stress.
We further identified 289 contigs responsible for mitogen-activated protein kinases, of which two contigs (Cko_contig_220 and Cko_contig_10124) were upregulated dramatically by drought or salt stress (Additional file 14). In addition, eight G-type lectin Sreceptor-like serine/threonine-protein kinases were identified. Among these kinases, three (Cko_contig_99573, Cko_contig_39198, and Cko_contig_36276) were upregulated more than 32-fold under drought stress (Additional file 13). We also found that 12 cysteine-rich repeat Transcription factors responding to drought and salt treatments in C. korshinskii We performed global transcription-factor classification for differentially expressed transcripts and identified 53 transcripts belonging to 12 transcription-factor families ( Fig. 8, Additional file 15). Several transcription-factor family genes, e.g., zinc-finger protein and NAC, were dramatically differentially expressed in both drought and salt treatments. In addition, 12 WRKY members were strongly upregulated under drought treatment. Among RING-finger proteins or MYB-family members, more were downregulated under drought treatment. Among DREs, the up-and downregulated members were evenly balanced.

Genes of unknown function and genes unique to C. korshinskii
In this study, many contigs had no homologous matches among database sequences and thus have no known To identify genes that might be expressed uniquely in C. korshinskii under salt and/or drought stresses, we performed a new set of BLAST searches to compare these contigs to genes expressed in Arabidopsis under salt and/or drought stresses [37].
A total of 295 DEGs that were downregulated in drought-treated C. korshinskii were also determined to be downregulated in drought-treated Arabidopsis (Fig. 9, Additional file 16). Some of these genes encode nitrate reductase, glutathione S-transferase, etc.; however, some genes (29 DEGs) that were strongly downregulated in drought-treated C. korshinskii did not significantly change in expression in drought-treated Arabidopsis (Additional file 16), such as peroxidase and thiol protease. Many genes (249 DEGs) that were significantly downregulated in drought-treated C. korshinskii were upregulated in droughttreated Arabidopsis (Fig. 9, Additional file 16), including genes encoding subtilisin-like protease, ubiquitin carboxylterminal hydrolase, etc.
A total of 188 DEGs were upregulated in salt-treated Arabidopsis and C. korshinskii, such as spore-specific catalase and aldehyde dehydrogenase; however, many genes (17 DEGs) were strongly upregulated in salttreated C. korshinskii but remained nearly constant in expression in salt-treated Arabidopsis (Fig. 9, Additional file 16), such as plasma membrane H + -ATPase and 60S ribosomal protein. Some genes (210 DEGs) that were significantly upregulated in salt-treated C. korshinskii were slightly downregulated in salt-treated Arabidopsis (Additional file 16), including genes encoding ADP, ATP-carrier protein, and mannitol dehydrogenase.
A total of 241 DEGs were downregulated in salt-treated C. korshinskii and Arabidopsis (Fig. 9, Additional file 16), including genes encoding LRR receptor-like serine/threonine-protein kinase and beta-glucosidase; however, some genes (33 DEGs) were strongly downregulated in salt-   Among 368 DEGs that were downregulated in both drought-and salt-treated C. korshinskii, 88 were upregulated in both drought-and salt-treated Arabidopsis and 117 were downregulated in both drought-and salttreated Arabidopsis. Of the 1521 DEGs downregulated in drought-treated C. korshinskii, 95 were upregulated and 121 downregulated in drought-treated Arabidopsis, while, 93 of these DEGs were upregulated and 120 downregulated in salt-treated Arabidopsis. Twelve DEGs that were downregulated in both drought-and salttreated C. korshinskii remained relatively constant in both drought-and salt-treated Arabidopsis, except that Cko_contig_45026 remained relatively constant in drought-treated Arabidopsis but not in salt-treated Arabidopsis (Additional file 17).
In drought-treated C. korshinskii, however, 542 DEGs (33.25 %) did not match any Arabidopsis sequences in this study and thus were presumably unique to C. korshinskii. The expression levels of 17 genes changed significantly in drought-treated C. korshinskii; for example, Cko_con-tig_61538 was predicted to encode E3 ubiquitin-protein ligase and was upregulated 790.9-fold (Additional file 18). In addition, genes for proteins such as phosphoinositide phospholipase C (Cko_contig_42646), trypsin-protein inhibitor (Cko_contig_23762, Cko_contig_27564), and bZIP transcription factor (Cko_contig_23833) were expressed uniquely in drought-treated C. korshinskii.
In addition, 529 DEGs (34.80 %) in salt-treated C. korshinskii had no homologous sequence in the Arabidopsis datasets and were considered unique to C. korshinskii. The expression levels of 27 genes changed dramatically in salt-treated C. korshinskii, such as Cko_contig_58120, predicted to encode polyphenol oxidase, which was upregulated 313.3-fold, and Cko_contig_92720, predicted to encode serine/threonine kinase, which was upregulated 2678.4-fold (Additional file 18). Other genes, such as thiol protease (Cko_contig_105443), UDP-glucosyltransferase (Cko_contig_3136), and polyphenol oxidase (Cko_contig_58120), were found to be uniquely expressed in salt-treated C. korshinskii (Additional file 18). Although we cannot yet be certain which genes are responsible for C. korshinskii's stress resistance, we are convinced that the majority of predicted unigenes unique to C. korshinskii represent a valuable basis on which to explore gene diversity in C. korshinskii and support drought-or salt-stress resistance in C. korshinskii.

Quantitative real-time-PCR validation of differentially expressed transcripts from RNA-seq
To confirm the accuracy and reproducibility of the Illumina RNA-Seq results, a small number of contigs were chosen for quantitative real-time (qRT)-PCR detection. The expression of those contigs in root, stem, leaf samples of drought-or salt-treated C. korshinskii were analyzed ( Fig. 10a and b). Detailed information on the primers is provided in Additional file 19. The qPCR results for 38 selected contigs showed general agreement with their transcript-abundance changes as determined by RNA-seq, suggesting the reliability of the transcriptomic profiling data. For example, under drought stress, Cko_contig_59435, which shows strong homology to cationic peroxidase 1, was upregulated 10.29-fold in C. korshinskii: 22.85-fold in the stem and 5.06-fold in the leaf (Fig. 10b). Cko_contig_12764, a homolog of L-type lectin-domain-containing receptor kinase, was upregulated 19.23-fold in C. korshinskii under drought stress: upregulated 13.68-fold in the root, 7.05-fold in the stem and 24.01-fold in the leaf (Fig. 10a).
Cko_contig_95695, a homolog of glutathione Stransferase, was upregulated 22.93-fold under drought stress and 8.08-fold under salt stress: 31.31-fold in the drought-treated root, 4.04-fold in the salt-treated root, 13.05-fold in the drought-treated stem, 10.03-fold in the salt-treated stem, 7.08-fold in the drought-treated leaf and 1.14-fold in the salt-treated leaf (Fig. 10b). Nonetheless, moderate discrepancies between the expression levels and RPKM values were observed for three contigs: Cko_contig_12764, Cko_contig_59435, and Cko_con-tig_27564 ( Fig. 10a and b).

Discussion
In this study, we generated de novo transcriptomes of C. korshinskii under drought and salt treatment and compared the expression profiles between control and drought and salt conditions, as well as between C. korshinskii and the model plant Arabidopsis. From our analysis, we found that many genes/contigs related to protein kinases, certain transcription factors, the ROSscavenging system and ion transport proteins, gibberellin biosynthesis, carbon metabolism and osmotic regulation, and brassinosteroid-biosynthesis pathways are involved in drought-and salt-stress responses in C. korshinskii, with some of these genes being expressed significantly between C. korshinskii and Arabidopsis, as discussed in detail below. These results provide a molecular basis for understanding the specific drought-and salt-tolerance mechanisms of C. korshinskii.

Construction of an informative transcriptome dataset for C. korshinskii
Transcriptome sequencing of C. korshinskii is fundamental to the identification of functional genes and the elucidation of the molecular mechanisms involved in drought or salt adaptation and tolerance in C. korshinskii. In a recent study in C. korshinskii [38], a total of 66.35 million raw sequencing reads were generated, 64.03 million clean reads were obtained, and 86,265 contigs were assembled. The percentage of annotation was 48.09 %. In our study, a total of 199.94 million raw reads were generated from the samples, and 182.93 million clean reads were assembled into 129,451 contigs (Table 1). Notably, the percentage of annotation was 54.45 %. The numbers of total raw reads, clean reads, assembled contigs and the percentage of annotation were all higher than those obtained previously [38], likely because we used several samples under different growth conditions for sequencing analysis, whereas only a sample under normal growth conditions was used by Long et al. [38]. Our de novo assembly of RNA-seq improves the genome annotation of C. korshinskii.
Based on comparisons with different plant species, the top four species with BLAST hits to annotated unigenes were Glycine max, Medicago truncatula, Vitis vinifera, and Lotus japonicus, species whose genome annotations are comprehensive and largely accepted. The low rate of annotation compared with other plants such as Safflower Flowers (58 %) [39], Aegilops variabilis (66.46 %) [40], and Ma Bamboo (78.9 %) [41] might be due to technological limitations, such as sequencing depth, splice junctions and read length [42], which are common in de novo assembly and RNA-seq analysis. We can speculate that the unannotated contigs (45.55 %, 58,960 of 129,451 contigs) represent a specific genetic resource for C. korshinskii, which is worthy of further investigation. Unannotated sequences were on average much shorter than the annotated contigs (334 bp vs 975 bp); however, a considerable percentage of contigs (20.2 %, or 14,209 of 70,491) that were longer than 500 bp and that had RPKM values [43] over three had no known homologs in known plant species.

Protein kinases are important for drought and salt tolerance in C. korshinskii
Protein kinases phosphorylate and activate many target proteins in major signaling events in response to osmotic stress in higher plants [44]. Microarray analysis revealed that Glycine soja G-type lectin S-receptor-like serine/threonine-protein kinase GsSRK significantly increased in expression under alkaline (NaHCO 3 ) treatment, identifying it as an alkaline-responsive gene [45]. In this work, Cko_contig_103451, associated with Gtype lectin S-receptor-like serine/threonine-protein kinase, was upregulated 7.6-fold in drought-treated C. korshinskii whole plants and 4.6-fold in drought-treated Arabidopsis datasets. We also found that three other homologous genes (Cko_contig_99573, Cko_contig_39198, and Cko_contig_36276) were upregulated more than 32-fold in drought-treated C. korshinskii but changed little in Arabidopsis datasets (Additional file 20). These results are also consistent with the results of studies on the expression pattern of GsSRK in Glycine soja under NaCl (upregulated 4-fold in leaves and 3.5-fold in roots) and PEG treatments (upregulated 3.5-fold in leaves and 3.3-fold in roots) [46].
Arabidopsis thaliana L-type lectin-domain-containing receptor kinase AtLPK1 is strongly induced by stress hormones and salt treatments, and overexpression of AtLPK1 in Arabidopsis enhances salt-stress tolerance [39]. In this study, Cko_contig_12764 had a highly significant BLAST hit to the L-type lectin-domaincontaining receptor kinase and showed increased transcript abundance in salt-treated Arabidopsis datasets (upregulated 3.5-fold), but was downregulated 8-fold in salt-treated C. korshinskii. Another homologous gene (Cko_contig_49124) was upregulated 7-fold in drought-(See figure on previous page.) Fig. 10 Changes in transcript levels of 38 selected genes, as detected by real-time RT-PCR. The black bars represent the relative intensity of qRT-PCR from three independent biological replicates (left Y-axis), and the red bars represent the expression level (RPKM) of the transcript (right Y-axis) (a and b). a includes 20 qRT-PCR results, and b includes another 18 qRT-PCR results. The qRT-PCR primers for each contig are listed in Additional file 19. PM, PD and PS represent that the tested materials were the whole seedlings including leaves, stems, and roots in normal growth condition, under drought treatment, and salt treatment, respectively. RM, RD and RS represents that the tested materials were the roots of the plant in normal growth condition, under drought treatment, and salt treatment, respectively. SM, SD and SS represent that the tested materials were the stems of the plant in normal growth condition, under drought treatment, and salt treatment, respectively. LM, LD and LS represent that the tested materials were the leaves of the plant in normal growth condition, under drought treatment, and salt treatment, respectively treated C. korshinskii but upregulated slightly in Arabidopsis (Additional file 20). Kempa et al. [47] found that a plastid-localized glycogen synthase kinase 3 GSK-3 was induced at high concentrations of NaCl, KCl and sorbitol and could modulate salt-stress tolerance and regulate carbohydrate metabolism in Medicago sativa cv. Europa. We identified 58 contigs related to glycogen synthase kinase and only one differentially expressed transcript (Cko_contig_85930), which was uniquely expressed and upregulated 7.7-fold under drought stress in C. korshinskii, with no homologous gene detected in the examined Arabidopsis datasets.
In C. korshinskii, a large number of protein kinases may play an essential role in sensing external drought and salt signals and regulating gene expression at the cellular level in response, as seen in Arabidopsis (Additional file 20); however, the genes encoding protein kinases in response to stresses (e.g., salt or drought) may differ between C. korshinskii and other plants.

Transcription factors involved in drought-and salt-stress response in C. korshinskii
Many WRKY, MYB, NAM, ATAF, and CUC (NAC) transcription factors were found to confer resistance to drought or salt stress in plants, and overexpression of these transcription factors enhanced the drought or salt tolerance of transgenic plants [7,[48][49][50][51][52]. Previous research has found that the expression levels of wheat (Triticum aestivum L.) TaWRKY10 were upregulated 1.5-fold under PEG treatment and 2.6-fold under salt stress [53]. In the C. korshinskii and Arabidopsis datasets, we detected eight WRKY transcription factors that showed increased expression in both drought-treated C. korshinskii and Arabidopsis (Additional file 20), especially Cko_contig_52997 (upregulated 30.8-fold in drought-treated C. korshinskii and 12.3-fold in drought-treated Arabidopsis). The characteristics of WRKY expression in C. korshinskii and Arabidopsis were consistent with findings from research on wheat under drought stress [53]. The expression of the Arabidopsis zinc-finger proteins Zat 11 and Zat 12 is stimulated by drought (Zat 12 was upregulated 10-fold in leaves) and salt stresses (Zat 11 was upregulated 95-fold in roots) [54]. In this study, we found that a zinc finger protein, Cko_contig_79632, was upregulated 255.8-fold in drought-treated C. korshinskii but downregulated slightly in drought-treated Arabidopsis, and Cko_contig_85079 showed increased transcript abundance (20.0-fold increase) in salt-treated C. korshinskii but was only upregulated 0.9-fold in salt-treated Arabidopsis. Interestingly, we found that eight zinc-finger protein contigs were downregulated in drought-or salt-treated C. korshinskii. These findings support the possibility that zinc-finger proteins could play key roles as both positive and negative regulators of plant defenses [54].
We also identified differences in the expression profiles of some other transcription factors in C. korshinskii and Arabidopsis under the same stresses (Additional file 20); for example, some NAC, MYB, and ERFs were downregulated in drought-or salt-treated C. korshinskii. This result indicated that some MYB and ERF transcription factors might function as negative regulators of drought or salt tolerance in C. korshinskii.
Functional genes related to the ROS-scavenging system and ion transport in C. korshinskii ROS are produced in plant cells under normal and stressful conditions [55] and can be removed by the antioxidant system [55,56]. In C. korshinskii, antioxidant enzymes such as superoxide dismutase (SOD), catalase (CAT), and peroxidase (POX) protect cells and subcellular systems from drought or salt damage caused by ROS (Additional file 20). For example, Cko_contig_105070 showed strong homology to CAT and was transcribed at high levels in salt-treated C. korshinskii (180.9-fold increase) and upregulated 1.13-fold in salt-treated Arabidopsis (Additional file 20). Therefore, it is most likely associated with enhanced CAT activity and responsible for converting the increased H 2 O 2 into H 2 O and O 2 [57]. Ascorbate peroxidase (APX) represents the ascorbatedependent H 2 O 2 -scavenging mechanism in plants, which reduces H 2 O 2 to H 2 O using ascorbic acid as an electron acceptor [57]. Eleven APX proteins were uniquely expressed in treated C. korshinskii (Additional file 5) but were not detected in Arabidopsis datasets. Two of these proteins (Cko_contig_33634, Cko_contig_46972) showed increased transcript abundance under drought stress. In the glutathione peroxidase (GPX) pathway, GPX can reduce H 2 O 2 to the corresponding hydroxyl compounds [21,58]. Interestingly, in our datasets, 12 GPX proteins were expressed uniquely in drought-or salt-treated C. korshinskii. In particular, Cko_contig_17138 was upregulated 2.4-fold in drought-treated C. korshinskii and upregulated 79.1 % under salt stress, no homologous contigs were found in published Arabidopsis datasets (Additional file 5).
It is possible that the survival of plants under salt stress depends on the accumulation of high levels of ions in the tissue for the maintenance of osmotic balance [59]. Cko_contig_24748, associated with a K + -transporter, was uniquely upregulated 16.5-fold under salt stress in C. korshinskii but was not found in the considered Arabidopsis datasets.
The transcript profile was consistent with the profile of K + -transporters in R. trigyna, most of which were upregulated under salt stress [21]. This result suggests that K + -transporters were enhanced and Na + /K + homeostasis reestablished in salt-treated C. korshinskii, as seen in R. trigyna under salt stress [21]. It is possible that plasma-membrane H + -ATPases (PM-H + -ATPases) are involved in the mitigation of physiological disturbances imposed by salt stress [60]. Interestingly, in this study, Cko_contig_104930, encoding plasma membrane H + -ATPases (PM-H + -ATPases), showed significant increases in expression (3.1-fold increase) in salt-treated C. korshinskii but remained relatively constant in salttreated Arabidopsis. This finding echoes the results of other studies on salt-treated plants, such as Oryza sativa Indica cultivars [60]. Transcripts encoding proteins associated with functions such as diterpenoid biosynthesis mechanosensitive ion channels (Cko_contig_66582) accumulated in both drought-and salt-treated C. korshinskii but were not detected in Arabidopsis datasets. These results indicate that C. korshinskii may have the specific ability to transduce these stress stimuli into electrical signals to enhance drought or salt tolerance. In summary, a small number of contigs uniquely found in C. korshinskii but not in Arabidopsis suggest that C. korshinskii may have a stress-tolerance mechanism distinct from that of the model plant Arabidopsis.

Contigs in response to gibberellin biosynthesis in C. korshinskii
Plant growth and development are regulated by gibberellins (GAs), which are regulated by numerous genes involved in GA biosynthesis and degradation. GA biosynthesis is tightly controlled by Gibberellin 20-beta-dioxygenase (GA20ox) and Gibberellin 3-betadioxygenase (GA3ox). Gibberellin 2-beta-dioxygenase (GA2ox) catalyzes the deactivation of bioactive GAs [61]. In Arabidopsis, the cellular concentration of bioactive GAs was reduced via an increase in expression of GA2ox7, resulting in an accumulation of DELLA, an inhibitor of plant growth [62]. Based on transcriptomesequence analysis of C. korshinskii and Arabidopsis, we identified nine drought-or salt-responsive GA2ox proteins, but these contigs were not found in Arabidopsis datasets (Additional file 5). Only five GA3ox proteins and Cko_contig_63577 were overexpressed in both-drought treated C. korshinskii and in researched Arabidopsis datasets (Additional file 20). Two GA20ox proteins were detected at higher levels in drought-or salt-treated C. korshinskii, but no homologous gene was found in the Arabidopsis datasets (Additional file 5). In addition, three contigs associated with DELLA protein were downregulated in both treated C. korshinskii and Arabidopsis (Additional file 20). Thus, we propose that GAs are upregulated under salt stress for normal growth of halophytes such as C. korshinskii, with accompanying repression of the expression of DELLA proteins. Our prediction is supported to some degree by studies of a GA-deficient mutant of Arabidopsis [62,63].

Contigs associated with carbon metabolism and osmotic regulation in C. korshinskii and Arabidopsis
Responses of carbon metabolism or starch metabolism to drought stress have been found in many plants, such as alfalfa (Medicago sativa) [64] and maize (Zea mays) [65]. In C. korshinskii, seven sugar-transport proteins, 13 carbohydrate-transport proteins, one granule-bound glycogen (starch) synthase protein and two sucrose-UDP glucosyltransferase proteins were found to be differentially expressed under drought or salt stress (Additional file 5). In particular, Cko_contig_43585, encoding a sugar-transport protein, was upregulated 568.5-fold in salt-treated samples. The expression data suggest that sugar or carbon metabolism, as well as various sugarrelated signaling pathways, are influenced by drought stress in C. korshinskii.
The synthesis and transport of substances involved in osmotic regulation (such as proline and other amino acids) promote salt tolerance in most plants and crops [3,66,67]. Previous studies have reported that ProT accumulation is increased under salt stress in several plants, such as barley and mangrove [68,69]. Our experimental results showed an increase in expression of the proline transporter gene (ProT, Cko_contig_90699), which was uniquely upregulated 2.6-fold and 1.7-fold in drought-and salt-treated C. korshinskii, respectively, but not in Arabidopsis. This drought-or salt-treated expression pattern in C. korshinskii was similar to the findings in barley and mangrove [68,69]. In yeast, amino-acid permeases (AAPs) complement proline-uptake deficiency and mediate the efficient transport of proline, alanine, and valine, but the expression of most members of the AAP-gene family was repressed under high-salt conditions [70]. Surprisingly, the transcription of AAP (Cko_contig_4505) was uniquely induced (3.3-fold increase) in drought-treated C. korshinskii, and no homologous gene was found in published Arabidopsis datasets. This finding suggests that C. korshinskii had the distinct ability to synthesize and transport various osmoprotectants, such as proline, osmotin, and amino acids, to maintain cell turgor under osmotic stress, which is not seen in the model plant Arabidopsis or in other eukaryotes.

Contigs associated with brassinosteroid-biosynthesis pathways in C. korshinskii
Brassinolides regulate plant growth and development and have been known to improve plant tolerance to abiotic stresses [2]. The results of the RNA-seq expression analysis showed that 30 contigs involved in brassinosteroid biosynthesis pathway were regulated in treated C. korshinskii (Additional file 5). Cko_contig_7817 (85A1/2 or 85A2), a homolog of brassinosteroid-6-oxidase family protein, was differentially regulated in treated  1.4), that are involved in starch and sucrose metabolism in drought-or salt-treated C. korshinskii, but no homologous genes were detected in Arabidopsis datasets (Additional file 5). These genes are necessary to enhance the biosynthesis of soluble sugars such as glucose, sucrose, and fructose and to maintain cell osmotic adjustment and balance (Additional files 22 and 23).

Conclusions
This work presents an original transcriptome-sequencing analysis of mixed RNA from C. korshinskii whole seedlings (leaves, stems, and roots) under drought and salt stresses. A total of 129,451 cDNA contigs from whole C. korshinskii plants were obtained, of which 54.51 % were annotated with gene descriptions, while the remaining 45.49 % were unknown. A large number of candidate functional genes were identified as being potentially involved in multiple mechanisms that might contribute to drought or salt tolerance, meriting further investigation. Several key pathways, such as metabolic pathways, biosynthesis of secondary metabolites, ribosome, purine metabolism and plant hormone signal transduction, and GO terms, such as metabolic process, cell or cell part, and catalytic activity, were identified as being involved in drought and salt tolerance in C. korshinskii. In addition, hundreds of DEGs appeared to be expressed uniquely in C. korshinskii under drought and salt stress, respectively, compared to the published Arabidopsis database. These unknown genes and uniquely expressed DEGs were presumably genes unique to C. korshinskii that are involved in drought or salt tolerance. This information of C. korshinskii genes will be very useful in future studies on molecular adaptations to abiotic stress in leguminous plants and will facilitate the genetic manipulation of important plants.

Plant materials, growth conditions, and stress treatments
Seeds of C. korshinskii collected from Turpan Desert Botanical Garden, Chinese Academy of Sciences were sown in plastic pots (12 × 12 cm) filled with a nonsterile sandy soil irrigated with 1 L sterile water. After germination, C. korshinskii seedlings were grown in a greenhouse with a 16 h light/8 h dark photoperiod at a temperature of 22°C ± 1°C and a relative humidity of 50 ± 10 % and reirrigated with 0.7 L sterile water. Two weeks later, the salt-treated plants were watered with 400 mM NaCl, the drought-treated plants were unwatered, and control plants were irrigated with 1 L sterile water. One month later, when seedlings were approximately 10-cm high, all seedlings, leaves, stems, and roots were harvested for analysis.

cDNA-library preparation and Solexa sequencing for transcriptome analysis
Total RNA was extracted from C. korshinskii whole seedlings treated with 400 mM NaCl or drought stress using a Plant RNA Extraction Kit (Autolab, China), following the manufacturer's protocol. The concentration and quality of each RNA sample were determined using a NanoDrop 2000TM micro-volume spectrophotometer (Thermo Scientific, Waltham, MA, USA) and gel electrophoresis. The poly (A) mRNA was isolated from 10 mg total RNA using magnetic oligo(dT) beads, then divided into short fragments by fragmentation buffer (Ambion, Austin, TX, USA). The first-strand cDNA was synthesized by randomhexamer primers, and then the second-strand cDNA was synthesized using DNA polymerase I (New England Biolabs), RNase H (Invitrogen), buffer, and dNTPs. Short DNA fragments were purified with a QIAquick PCR Purification Kit (Qiagen Inc., Valencia, CA, USA), then subjected to end repair and poly(A) addition and ligated to sequencing adaptors. Suitable fragments (350-450 bp) were purified by agarose-gel electrophoresis and gathered by PCR amplification. One cDNA library (Control sample) was sequenced on a PE flow cell and three cDNA libraries (Control, drought-treated and salt-treated samples) were sequenced on a SE flow cell using the Illumina HiSeq™ 2000 platform.

De novo transcriptome assembly
To facilitate assembly, the following criteria were used to filter low-quality reads: 1. Filter reads with adapter contamination. 2. Filter reads in which unknown nucleotides >5 %. 3. Filter reads in which more than 20 % of bases showed a Q-value less than 20.
High-quality clean reads obtained after the above filtering and correction steps were used for de novo assembly using the Trinity method [71] with an optimized k-mer length of 25 by default and with all other parameters set to default values. The assembled sequences were further incorporated into distinct contigs by using the CD-HIT [72] and CAP3 [73] programs. The parameters in CAP3 were "-p 98 -o 50". CD-HIT was also used to cluster transcripts on the basis of sequence similarity (-c sequence identity threshold >0.98) and retain only the longest transcript from each cluster; redundant contigs were excluded.

Sequence annotation
Multiple complementary approaches were adopted for alignment and analysis. All contig sequences were entered as search terms in the NCBI non-redundant nucleotide sequence (nr) database, the Kyoto Encyclopedia of Genes and Genomes (KEGG) database, the Cluster of Orthologous Groups of proteins (COG) database, and the Swiss-Prot protein database using a cut-off E-value of 10 −5 . Gene Ontology (GO) annotations of the contigs were determined using Blast2go (https://www.blast2go.com/) [74] according to the molecular function, biological process, and cellular component ontologies (http://www.geneontology.org/) with an E-value threshold of 10 −5 ; the GO level of the GO distribution figure was three. The best alignment results were used to determine the 5′-to-3′ orientation of the contigs. Annotated contigs with no GO hits were investigated in terms of protein domains using InterProScan databases (http://www.ebi.ac.uk/interpro/). The "GetORF" program from the EMBOSS software package was used to predict the open reading frames (ORFs) of each contig [75]. Enzyme mapping of the annotated contigs was performed by direct GO-to-enzyme mapping and used to query the KEGG to define the KEGG orthologs (KOs). The KEGG mapping tool was used to plot these KOs into a complete metabolic atlas [76].

Identification of differential expression transcripts (DEGs) and functional analysis
The expression of each contig was calculated by the RPKM method [43]. The method of Audic and Claverie [77] was applied to determine the P-value, which corresponds to the differential transcript expression of the contigs. The false discovery rate (FDR) is used to determine the P-value threshold in multiple tests. FDR ≤0.05 and the absolute value of log 2 Ratio ≥1 were used as the threshold to judge the significance of differential gene expression. The differentially expressed genes were used for GO-and KEGG-enrichment analyses. GO-enrichment analysis was performed using the Blast2go software, and the P-values were calculated using the Benjamini-Hochberg (BH) correction [78]. We selected a corrected P-value <0.05 as a threshold to determine which DEGs were significantly enriched in GO terms. In KEGG-enrichment analysis, the cellular metabolism, biochemical pathways, and potential biological behaviors of the differentially expressed genes were examined (www.genome.jp/kegg/). A corrected P-value <0.05 was used as the threshold to determine significant enrichment of the gene sets.
We performed a new set of BLAST searches to compare C. korshinskii contigs to genes expressed in Arabidopsis under salt and/or drought stresses, in order to identify genes that might be expressed uniquely in C. korshinskii under salt and/or drought stresses. The RNA-seq data of Arabidopsis can be found in ArrayExpress and the accession number was E-GEOD-48235.

Real-time quantitative reverse-transcription polymerase chain reaction (qRT-PCR) for validation and expressionpattern analysis
Total RNA was extracted from whole seedlings and organs (leaves, stems, and roots) of C. korshinskii treated with 400 mM NaCl or drought stress using a Plant RNA Extraction Kit (Autolab, China) according to the manufacturer's protocol. Reverse-transcription reactions were performed using Superscript II reverse transcriptase (Invitrogen, Grand Island, NY, USA) according to the manufacturer's instructions and 1 μg of total RNA and oligo(dT) primers.
Primers for qRT-PCR were designed using the Primer Premier 5.0 software (Premier Biosoft Int., Palo Alto, CA, USA). The sequences of the specific primer sets are listed in Additional file 19. The qRT-PCR analysis was performed using an ABI Prism 7500 sequence detector (Applied Biosystems, Foster City, CA, USA) and the SYBR® Premix Ex TaqTM Kit (TaKaRa, Tokyo, Japan) to detect transcript abundance. The qRT-PCR reactions were performed in 20-μl volumes containing 2 μl firststrand cDNA, 200 nM of each primer, and 10 μl of the 1 × SYBR PCR mixture with the following cycling programs: denaturation at 95°C for 30 s, then 40 cycles of denaturation at 95°C for 3 s and annealing and extension at 60°C for 30 s. The specificity of the amplifications of all products were verified by melting curve analysis, which was conducted at 95°C for 15 s, 60°C for 60 s, and 95°C for 15 s. Three replicates were conducted in parallel, and the results were normalized to the expression level of the constitutive 18S rRNA gene. The relative quantitative method (ΔΔCt) was used to evaluate quantitative variation.