De Novo Assembly and Analysis of Tartary Buckwheat (Fagopyrum tataricum Garetn.) Transcriptome Discloses Key Regulators Involved in Salt-Stress Response

Soil salinization has been a tremendous obstacle for agriculture production. The regulatory networks underlying salinity adaption in model plants have been extensively explored. However, limited understanding of the salt response mechanisms has hindered the planting and production in Fagopyrum tataricum, an economic and health-beneficial plant mainly distributing in southwest China. In this study, we performed physiological analysis and found that salt stress of 200 mM NaCl solution significantly affected the relative water content (RWC), electrolyte leakage (EL), malondialdehyde (MDA) content, peroxidase (POD) and superoxide dismutase (SOD) activities in tartary buckwheat seedlings. Further, we conducted transcriptome comparison between control and salt treatment to identify potential regulatory components involved in F. tataricum salt responses. A total of 53.15 million clean reads from control and salt-treated libraries were produced via an Illumina sequencing approach. Then we de novo assembled these reads into a transcriptome dataset containing 57,921 unigenes with N50 length of 1400 bp and total length of 44.5 Mb. A total of 36,688 unigenes could find matches in public databases. GO, KEGG and KOG classification suggested the enrichment of these unigenes in 56 sub-categories, 25 KOG, and 273 pathways, respectively. Comparison of the transcriptome expression patterns between control and salt treatment unveiled 455 differentially expressed genes (DEGs). Further, we found the genes encoding for protein kinases, phosphatases, heat shock proteins (HSPs), ATP-binding cassette (ABC) transporters, glutathione S-transferases (GSTs), abiotic-related transcription factors and circadian clock might be relevant to the salinity adaption of this species. Thus, this study offers an insight into salt tolerance mechanisms, and will serve as useful genetic information for tolerant elite breeding programs in future.


Introduction
Soil salinization has been an afflicting factor for plant agriculture [1]. More than 800 million hectares of land are suffering from salinization throughout the world [2]. Breeding and application of salt-tolerant elites for each food crop is becoming extremely urgent to feed the global population in future decades. Prior to achieving this goal there should be a good knowledge of the fundamental salinity treatment by supplying with 200 mM sodium chloride solution. Three independent biological replicates with eight seedling plants per replicate for each sample were collected at 0 h and 24 h after salinity treatment. The 0 h (control) and 24 h (salt treatment) samples were used for physiological analysis and total RNA extraction.

Physiological Analysis
The relative water content (RWC) was measured according to the method of Li et al. [29], and the electrolyte leakage (EL) was measured according to the protocol described by Ishitani [30]. The malondialdehyde (MDA) content, peroxidase (POD) and superoxide dismutase (SOD) activities were tested by the methods used in previous study [31].

RNA Extraction and cDNA Library Preparation
Total RNA of three control and three salt treated biological replicates was extracted using TaKaRa MiniBEST Universal RNA Extraction Kit (TaKaRa, Shiga, Japan) according to the manufacturer's protocols. The integrity and quantity of total RNA samples were assessed by using 1% agarose gel electrophoresis, Qubit RNA HS Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) and 2100 Bioanalyzer RNA 6000 Nano Assay Kit (Agilent Technologies, Singapore). mRNA was isolated from 5 ug total RNA (RNA Integrity Number (RIN) > 7.0, pooled in equal quantity from the total RNA of three replicates) of each sample by magnetic oligo(dT) beads, fragmented into short sequences, and then subjected to complementary DNA (cDNA) library construction by using Illumina TrueSeq RNA library method according to TrueSeq RNA Sample Preparation guide (Illumina Technologies, San Diego, CA, USA).

Illumina Sequencing and Sequences Assembly
The cDNA libraries of control and salt treatment samples were sequenced using Illumina Hiseq 2500 platform (Illumina Technologies) to generate 150 bp paired-end reads. The unqualified reads, including (1) adaptors contaminated reads; (2) unknown nucleotides more than 5% reads and (3) reads of which more than 50% of bases showed a q-value less than 5, were removed from the raw reads to obtain clean reads. Trinity program [32] was used to de novo assemble high-quality clean reads into transcriptome. The longest sequences of redundant and chimeric transcripts were defined as unigenes, and then were subjected to downstream functional annotation and coding sequence (CDS) prediction.

Coding Sequence Prediction, Functional Annotation and Classification of Unigenes
For CDS prediction, unigenes were compared with GenBank NR (NCBI non-redundant protein) [33] and Swiss-Prot databases [34] by BLAST [35] (E-value cut-off = 1 × 10 −5 ), and the best alignment was used to determine the 5 -3 orientation and the putative coding sequence region. ESTScan 3.0.3 [36] was used to predict the putative open reading frame (ORF) region of the unigenes without alignment in GenBank NR and Swiss-Prot databases.

Identification of Transcription Factor Families
To identify the potential transcription factor encoding genes in the transcriptome data, the assembled unigenes sequences were analyzed using PlantTFcat online tool (http://plantgrn. noble.org/PlantTFcat) as previously described [40].

Expression Analysis
The clean reads were mapped onto the assembled unigenes transcriptome using Bowtie [44], and the read count number for each gene was measured by RNA-Seq by Expectation-Maximization (RSEM) [45]. The estimated expression levels of all genes were calculated and normalized to fragments per kilobase of transcript per killion fragments mapped (FPKM) [46]. To identify the differentially expressed genes (DEGs), read count for each gene was normalized with the Trimmed mean of M-values (TMM) method [47], and then DEGseq [48] was used to screen the genes whose q-value were less than 0.005 and |log2FoldChange| > 1 between control and salinity treatment samples. Further, the identified DEGs were subjected to GO enrichment analysis by GOseq based on the Wallenius non-central hypergeometric distribution with the corrected p-value less than 0.05 [49]. Moreover, KEGG pathway enrichment analysis of the DEGs was also implemented using KEGG Orthology Based Annotation System (KOBAS) [50].

Identification of Simple Sequence Repeats (SSRs)
MISA (MIcroSAtellite) V1.0 [51] with default parameters was used to analyze distribution of various types of SSRs in all F. tataricum unigenes. The minimum number of repeat units for mono-nucleotide and di-nucleotide repeats were ten and six, respectively, while that for tri-nucleotide, tetra-nucleotide, penta-nucleotide and hexa-nucleotide repeats was five.

Reverse Transcription and Real-Time PCR
Two micrograms of total RNA for each sample were subjected to DNA digestion and reverse transcription to synthesize first-strand cDNA using EasyScript One-Step gDNA Removal and cDNA Synthesis SuperMix Kit (Transgen Biotech, Beijing, China). The cDNA reaction product was diluted 10-fold and 2 uL of the dilution was used to carry out real-time PCR using TransStart Green qPCR SuperMix (Transgen Biotech) on a real-time PCR system (LineGene 9600, BIOER, Hangzhou, China). The real-time PCR program consists of 94 • C for 5 min, 39 cycles of denaturing at 95 • C for 10 s, annealing and extension at 59 • C for 20 s. An actin gene was used as the internal control [52]. The relative expression level of detected genes was quantified by the 2 −∆∆Ct algorithm [37] using Microsoft Excel 2015 (Microsoft, Redmond, WA, USA). For each gene, we performed at least three biological replicates to determine the relative expression level. The related primers used for real-time PCR were list in Table S8.

Statistical Analysis
The data of various physiological indices in Figure 1 were subjected to Student's t-test analysis using Microsoft Excel 2015.

Physiological Analysis
In order to evaluate the effects of salt on physiological activities in F. tataricum seedlings, we measured a series of key physiological indices including relative water content (RWC), electrolyte leakage (EL), malondialdehyde (MDA) content, superoxide dismutase (SOD) and peroxidase (POD) activities. Two-week-old Chuan Qiao No.1 seedlings were treated with 200 mM NaCl solution, and then the aerial parts of seedling plants were collected at 0 h (control) and 24 h (salt treatment) after salt treatment. The collected seedling samples were used for physiological analysis. Values are mean ± SD (n = 3). Single asterisk (*) and double asterisks (**) stand for p < 0.05 and p < 0.01, respectively, determined by student's t-test.
As shown in Figure 1A, the RWC in salt stress plants was decreased (p-value < 0.05) than that in control plants. The EL analysis demonstrated that the EL was significantly increased (p-value < 0.01) from 4.7% to 7.2% after salt treatment ( Figure 1B). The MDA content was sharply increased (p-value < 0.01) by 25.9% after salt treatment than control ( Figure 1C). Meanwhile, we found both SOD and POD activities were significantly increased (p-value < 0.01) under salt stress ( Figure 1D,E). These results suggest that salt stress of 200 mM NaCl solution is able to significantly affect the physiological activities in F. tataricum seedlings.

RNA-Seq and De Novo Assembly of the F. tataricum Transcriptome
To elucidate the underlying salt stress responsive mechanisms in tartary buckwheat, we carried out RNA-seq analysis. Total RNA of control and salt treated samples were extracted, followed by cDNA library construction and paired-end Illumina deep sequencing. After removing the lowquality reads and adaptor sequences from reads, in total we obtain 23.81 million and 29.34 million clean reads from control and salt treatment libraries, which yielded 7.14 billion and 8.80 billion nucleotides, respectively (Table 1). With respect to GC content, control library reached 48.76% and that of salt treatment library accounted for 46.70% (Table 1). The percentage of Q20 sequences (with an error probability of 0.01) in control library reached 97.85%, while that in salt treatment library was 98.03% (Table 1). Using Trinity platform [32], we de novo assembled the two libraries clean reads into 75,287 transcripts, from which 57,921 unigenes were generated (Table 1). All the unigenes length ranged from 201 bp to 18,866 bp. The mean length, median length and N50 length of all assembled unigenes were 768 bp, 387 bp, and 1400 bp, respectively (Table 1).
An overview of the unigenes size distribution was showed in Figure 2. Among these unigenes, 23,845 (41.17%) were longer than 500 bp, and that longer than 1000 bp reached 13,930 (24.05%). The assembled unigenes dataset was used as reference transcriptome for further analysis. The reference transcriptome data generated in this study could be found at the Gene Expression Omnibus (GEO), Values are mean ± SD (n = 3). Single asterisk (*) and double asterisks (**) stand for p < 0.05 and p < 0.01, respectively, determined by student's t-test. Figure 1A, the RWC in salt stress plants was decreased (p-value < 0.05) than that in control plants. The EL analysis demonstrated that the EL was significantly increased (p-value < 0.01) from 4.7% to 7.2% after salt treatment ( Figure 1B). The MDA content was sharply increased (p-value < 0.01) by 25.9% after salt treatment than control ( Figure 1C). Meanwhile, we found both SOD and POD activities were significantly increased (p-value < 0.01) under salt stress ( Figure 1D,E). These results suggest that salt stress of 200 mM NaCl solution is able to significantly affect the physiological activities in F. tataricum seedlings.

RNA-Seq and De Novo Assembly of the F. tataricum Transcriptome
To elucidate the underlying salt stress responsive mechanisms in tartary buckwheat, we carried out RNA-seq analysis. Total RNA of control and salt treated samples were extracted, followed by cDNA library construction and paired-end Illumina deep sequencing. After removing the low-quality reads and adaptor sequences from reads, in total we obtain 23.81 million and 29.34 million clean reads from control and salt treatment libraries, which yielded 7.14 billion and 8.80 billion nucleotides, respectively (Table 1). With respect to GC content, control library reached 48.76% and that of salt treatment library accounted for 46.70% ( Table 1). The percentage of Q20 sequences (with an error probability of 0.01) in control library reached 97.85%, while that in salt treatment library was 98.03% (Table 1). Using Trinity platform [32], we de novo assembled the two libraries clean reads into 75,287 transcripts, from which 57,921 unigenes were generated ( Table 1). All the unigenes length ranged from 201 bp to 18,866 bp.
The mean length, median length and N50 length of all assembled unigenes were 768 bp, 387 bp, and 1400 bp, respectively (Table 1).
An overview of the unigenes size distribution was showed in Figure 2. Among these unigenes, 23,845 (41.17%) were longer than 500 bp, and that longer than 1000 bp reached 13,930 (24.05%). The assembled unigenes dataset was used as reference transcriptome for further analysis. The reference transcriptome data generated in this study could be found at the Gene Expression Omnibus (GEO), National Centre for Biotechnology Information (NCBI) (https://www.ncbi.nlm.nih.gov/geo/) under the accession number GSE104167. CQ_0h and CQ_Na24h represent control and salt-treated samples respectively.

Analysis of Differentially Expressed Genes
To investigate the DEGs between control and salt treatment libraries, we mapped the clean reads from each sample onto the F. tataricum reference transcriptome using Bowtie. Approximately 81.3% and 84.7% of the total high-quality reads from control and salt treatment libraries were mapped to the reference transcriptome, respectively. Using DEGseq, with the cutoff of q-value < 0.005 and |log2FoldChange|>1, we obtained 455 DEGs ( Figure S1). Among these DEGs, 404 unigenes had higher expression levels under salt treated conditions, whereas only 51 unigenes had higher expression levels under control conditions.
To understand the biological function of the DEGs, we performed GO term enrichment using Blast2GO. Out of 455 DEGs, 363 DEGs were highly abundant in 25 GO terms (Corrected p-Value < 0.05). GO enrichment analysis revealed that these DEGs were overrepresented in ´oxidationreduction process´, ´endopeptidase inhibitor activity´, ´endopeptidase regulator activity´, ´peptidase inhibitor activity´, ´peptidase regulator activity´, ´cell morphogenesis´, and ´cellular component morphogenesis´ ( Figure 7A and Table S5). Most of the up-regulated DEGs in salt treatment were

Analysis of Differentially Expressed Genes
To investigate the DEGs between control and salt treatment libraries, we mapped the clean reads from each sample onto the F. tataricum reference transcriptome using Bowtie. Approximately 81.3% and 84.7% of the total high-quality reads from control and salt treatment libraries were mapped to the reference transcriptome, respectively. Using DEGseq, with the cutoff of q-value < 0.005 and |log2FoldChange|>1, we obtained 455 DEGs ( Figure S1). Among these DEGs, 404 unigenes had higher expression levels under salt treated conditions, whereas only 51 unigenes had higher expression levels under control conditions.
In addition, in order to further predict which biochemical pathways these DEGs may be involved in, we carried out KEGG enrichment analysis. The results revealed that the DEGs were highly represented in pathway terms 'Antigen processing and presentation' (ko04612), 'Protein processing in endoplasmic reticulum' (ko04141), 'Estrogen signaling pathway' (ko04915), 'Toxoplasmosis' (ko05145), and 'Plant-pathogen interaction' (ko04626) ( Figure 7B and Table S6). enriched in ´cell morphogenesis´, ´cellular component morphogenesis´, ´oxidation-reduction process´, ´endopeptidase inhibitor activity´, ´endopeptidase regulator activity´, ´peptidase inhibitor activity´, ´peptidase regulator activity´, ´cell wall macromolecule metabolic process´, and ´oxidoreductase activity´ (Table S5). Meanwhile, the down-regulated DEGs under salt treated conditions were overrepresented in ´ATP metabolic process´, ´purine nucleoside triphosphate metabolic process´, ´ribonucleoside triphosphate metabolic process´, ´purine ribonucleoside triphosphate metabolic process´, ´nucleoside triphosphate metabolic process´ and ´monovalent inorganic cation transport´ (Table S5). These results indicated the biological functions described above might be activated under salt stress in F. tataricum.  In addition, in order to further predict which biochemical pathways these DEGs may be involved in, we carried out KEGG enrichment analysis. The results revealed that the DEGs were highly represented in pathway terms ´Antigen processing and presentation´ (ko04612), ´Protein processing in endoplasmic reticulum´ (ko04141), ´Estrogen signaling pathway´ (ko04915), ´Toxoplasmosis´ (ko05145), and ´Plant-pathogen interaction´ (ko04626) ( Figure 7B and Table S6).

Validation of Differentially Expressed Genes by Real-Time PCR
To confirm the reliability of expression profile data obtained by RNA-seq analysis, we conducted real-time PCR analysis. Fourteen unigenes randomly selected from the DEGs were further qualified the expression levels in control and salt treated samples. The fold changes of these genes obtained by real-time PCR were consistent with that obtained by RNA-seq analysis (according to Pearson correlation analysis, R 2 = 0.9372) (Figure 8), thus confirming the accuracy and strong reliability of the expression analysis results.

Validation of Differentially Expressed Genes by Real-Time PCR
To confirm the reliability of expression profile data obtained by RNA-seq analysis, we conducted real-time PCR analysis. Fourteen unigenes randomly selected from the DEGs were further qualified the expression levels in control and salt treated samples. The fold changes of these genes obtained by real-time PCR were consistent with that obtained by RNA-seq analysis (according to Pearson correlation analysis, R 2 = 0.9372) (Figure 8), thus confirming the accuracy and strong reliability of the expression analysis results.

Identification of Simple Sequence Repeats
Molecular markers, such as Random Amplified Polymorphic DNA (RAPD), Restriction Fragment Length Polymorphism (RFLP), Single Nucleotide Polymorphism (SNP) and Simple Sequence Repeats (SSRs), have been extensively used as powerful tools for genetic study and markerassisted selection (MAS) in many food crops [54]. In this study, we performed SSRs screening using MISA search tool [51].

Identification of Simple Sequence Repeats
Molecular markers, such as Random Amplified Polymorphic DNA (RAPD), Restriction Fragment Length Polymorphism (RFLP), Single Nucleotide Polymorphism (SNP) and Simple Sequence Repeats (SSRs), have been extensively used as powerful tools for genetic study and marker-assisted selection (MAS) in many food crops [54]. In this study, we performed SSRs screening using MISA search tool [51].

Discussion
Soil salinization is a severe threat to the yield of many crops. Recently, transcriptome studies on a large number of species, such as rice, Arabidopsis, smooth cordgrass (Spartina alterniflora) [14], wheat (Triticum aestivum) [15] and Caragana korshinskii [55], have been intensively carried out to decode the salt stress response. These studies have largely improved our understanding of salt-responsive mechanisms of plants. However, more investigation is needed especially for the uncharacterized species. In this study, we characterized the salt-responsive transcriptome of F. tataricum. To our knowledge, before this study, there are seldom reports focused on globally deciphering the salt response via RNA-seq in F. tataricum.
Chuan Qiao No. 1 is one of the most widely grown tartary buckwheat cultivar in Sichuan Province. Considering the importance of seedling stage for tartary buckwheat development and yield, we used two-week-old seedlings as the experimental materials for physiological activities analysis and RNA-seq analysis. By measuring the RWC, we found salt stress could cause water loss in seedling plants at 24 h after salt treatment. EL and MDA content are usually used to evaluate the integrity of plant cell membranes under abiotic stress [56]. Significantly increased EL and MDA content indicated 200 mM NaCl solution has largely affected the robust status of cell membrane in tartary buckwheat seedlings. When exposed to abiotic stress, plant antioxidant enzymes such as POD and SOD are usually activated to protect cells and subcellular systems from damage caused by reactive oxygen species (ROS) [57]. Consistently, we tested and found increased POD and SOD activities in seedling plants after salt treatment for 24 h.
Due to unavailable genome information, we de novo assembled a transcriptome of F. tataricum.  [59] in F. tataricum. Thus, our high-throughput RNA-seq yielded considerably more genes. We speculated the gaps could result from the differences between salt stress and aluminum stress processes, or from the sequencing approaches.
Among all transcriptome sequences, 36,688 (63.34%) unigenes were annotated in public databases. Through GO classification, 20,834 (35.96%) unigenes were categorized into 56 subcategories, out of which ´metabolic process´, ´binding´, ´cellular process´ and ´catalytic activity´ were

Discussion
Soil salinization is a severe threat to the yield of many crops. Recently, transcriptome studies on a large number of species, such as rice, Arabidopsis, smooth cordgrass (Spartina alterniflora) [14], wheat (Triticum aestivum) [15] and Caragana korshinskii [55], have been intensively carried out to decode the salt stress response. These studies have largely improved our understanding of salt-responsive mechanisms of plants. However, more investigation is needed especially for the uncharacterized species. In this study, we characterized the salt-responsive transcriptome of F. tataricum. To our knowledge, before this study, there are seldom reports focused on globally deciphering the salt response via RNA-seq in F. tataricum.
Chuan Qiao No. 1 is one of the most widely grown tartary buckwheat cultivar in Sichuan Province. Considering the importance of seedling stage for tartary buckwheat development and yield, we used two-week-old seedlings as the experimental materials for physiological activities analysis and RNA-seq analysis. By measuring the RWC, we found salt stress could cause water loss in seedling plants at 24 h after salt treatment. EL and MDA content are usually used to evaluate the integrity of plant cell membranes under abiotic stress [56]. Significantly increased EL and MDA content indicated 200 mM NaCl solution has largely affected the robust status of cell membrane in tartary buckwheat seedlings. When exposed to abiotic stress, plant antioxidant enzymes such as POD and SOD are usually activated to protect cells and subcellular systems from damage caused by reactive oxygen species (ROS) [57]. Consistently, we tested and found increased POD and SOD activities in seedling plants after salt treatment for 24 h.
Due to unavailable genome information, we de novo assembled a transcriptome of F. tataricum.  [59] in F. tataricum. Thus, our high-throughput RNA-seq yielded considerably more genes. We speculated the gaps could result from the differences between salt stress and aluminum stress processes, or from the sequencing approaches.
Among all transcriptome sequences, 36,688 (63.34%) unigenes were annotated in public databases. Through GO classification, 20,834 (35.96%) unigenes were categorized into 56 sub-categories, out of which 'metabolic process', 'binding', 'cellular process' and 'catalytic activity' were strongly represented. Meanwhile, KEGG and KOG classification were used as alternative approaches to map the unigenes to biological pathways. The results showed that metabolic pathways of 'Posttranslational modification, protein turnover, chaperones', 'Signal transduction mechanisms', 'Intracellular trafficking, secretion, and vesicular transport' in KOG and 'Signal transduction', 'Carbohydrate metabolism' and 'Translation' in KEGG were overrepresented. These results can supply with important cues for dissecting the molecular mechanisms of salt responses in F. tataricum.
Through comparison of gene expression patterns between control and salt treatment, we obtained 455 DEGs. This is relatively a smaller number compared with the DEGs numbers obtained in Arabidopsis [60,61], wheat [15], and rice [62]. We speculated that this gap could result from the fact that our RNA-seq data mainly reflected the early-stage gene expression profiles under salt stress (24 h after salt treatment). Anyway, among these DEGs, the genes related to protein kinases, phosphatases, HSPs, ABC transporters, GSTs, abiotic-related transcription factors and circadian clock were dominant. Protein kinases and phosphatases act as key components in plant stress signaling [2]. Previous studies indicated that the mitogen-activated protein kinases (MAPK) pathways were tightly related to salt stress response [18,[63][64][65]. Among our DEGs, c33984_g1 and c34545_g1, two homologs of MAPK kinase, were up-regulated after salinity treatment (Table 3), suggesting MAPK signaling pathways are also involved in buckwheat salt signaling. Meanwhile, four serine-threonine protein kinase encoding genes, namely c1037_g1, c25455_g1, c33021_g2, c25455_g1 and c22689_g1 showed significantly higher expression levels in salt treatment (Table 3), which was consistent with the observation in Spartina alterniflora salt stress transcriptome data [14]. In Arabidopsis thaliana, calcium signaling is elicited by salt stress, and calcium-dependent protein kinases (CDPKs) and CBL-interacting protein kinases (CIPKs) mediate the calcium signal and downstream responses [1,4,66]. It is worth noting that c38839_g1, c31052_g1, c60566_g1 (three putative CDPKs) and c49233_g1 (a putative CIPK) were up-regulated under salt stress (Table 3). Previous study showed that overexpression of a L-type lectin-domain-containing receptor kinase (AtLPK1) enhanced the salt-stress tolerance in Arabidopsis [67]. Coincidentally, our analysis showed that the expression of c10866_g1 (a homolog of AtLPK1) was increased by about 9-fold in salt treatment than in control (Table 3). Besides, one putative protein phosphatase 2C (PP2C) gene (c38748_g1) and one tyrosine-protein phosphatase gene (c60922_g1) were significantly induced by salt stress (Table 3), suggesting protein phosphatases may play important roles in F. tataricum under stress.
In Arabidopsis and creeping bentgrass (Agrostis stolonifera L.), heat shock proteins (HSPs) function as protein chaperones to attenuate plant abiotic stress responses [68][69][70]. A lot of potential HSPs homologs identified by our study, such as c10992_g1, c22339_g1, c26417_g3, c29031_g1, c29386_g1, c29386_g2, c30033_g1, c31857_g1, c32499_g1, c32766_g1, c35678_g1, c38902_g1 and c54547_g1, were shown to accumulate more transcripts under salt stress (Table 3), suggesting the involvement of HSPs regulatory mechanisms. ATP-binding cassette (ABC) transporters are also correlated with salt and other abiotic stress resistance [71,72]. Overexpression of AtABCG36 displayed higher fresh weight and stronger drought and salt resistance than wild type in Arabidopsis. Here, two ABC transporter B family members (c39646_g1 and c30623_g2) and two ABC transporter C family members (c38901_g1 and c33388_g1) were significantly up-regulated (Table 3), suggesting ABC transporters might share similar function in F. tataricum as well as in Arabidopsis. Glutathione S-transferases (GSTs) play vital roles in response to the oxidative stress including drought, salt and other stress [73][74][75]. Elevated expression levels of GSTs in tobacco and soybean were proven to enhance salt stress tolerance [74,75]. In our study, five putative GST encoding genes, namely c39155_g1, c39970_g1, c4684_g2, c59876_g1 and c60552_g1, showed significantly higher expression levels under salinity stress (Table 3), indicating GSTs might also impact the stress tolerance in F. tataricum.
Brassinosteroid (BR) metabolism participates in regulating plant growth and abiotic stress resistance [56]. Genetic and biochemical evidences demonstrated that somatic embryogenesis receptor kinases (SERKs) interacted with brassinosteroid insensitive 1 (BRI1) and were indispensable for early steps in BR signaling [76]. In rice (Oryza sativa), interaction between 14-3-3 proteins and OsBZR1 was essential for maintaining robust BR responses [77]. Notably, the fact that c25693_g1 (putative SERK1) and c39300_g1 (predicted 14-3-3) were remarkably up-regulated under salt treated conditions (Table 3) indicated that BR signaling might also participate in salt stress signaling pathways in F. tataricum.
Circadian clock coordinates the plant growth and development in response to environment changes [82,83]. Many microarray and metabolite-profiling studies revealed that plant circadian clock controlled stress-related genes modulate the abiotic responses [84]. Mutation of Gigantea (GI) in Arabidopsis resulted in enhanced salt tolerance, which is due to the release of SOS2 kinase [85]. In addition, the rice osgi mutant exhibited increased osmotic stress and harbored higher abundance of OsDREB1E, OsAP37, OsAP59, OsLIP9, OsLEA3, OsRAB16A, and OsSalT [86]. Consistently, in our transcriptome data, c32034_g1 (predicted FtGIGANTEA in tartary buckwheat) was decreased by more than 2-fold under salt conditions (Table 3). This finding indicates that in FtGIGANTEA may also function in salt-stress adaption.

Supplementary Materials:
The following are available online at www.mdpi.com/2073-4425/8/10/255/s1. Figure S1: Numbers of up-and down-regulated genes between control and salt treatment in Fagopyrum. tataricum. Table S1: Gene ontology of the F. tataricum transcriptome at level2. Table S2: Statistics of euKaryotic Orthologous Groups (KOG) function classification of the F. tataricum transcriptome. Table S3: Enrichment of the F. tataricum transcriptome on Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways at Hierarchy2. Table S4: Identification of the transcription factor families in F. tataricum. Table S5: Gene ontology of the differentially expressed genes (DEGs). Table S6: Enrichment of the DEGs on KEGG pathways. Table S7: Statistics of the frequency for different simple sequence repeats (SSRs) in F. tataricum. Table S8: The primers used for real-time PCR in this study.