Transcriptomic analysis of Pak Choi under acute ozone exposure revealed regulatory mechanism against ozone stress

Ground-level ozone (O3) is one of the major air pollutants, which cause oxidative injury to plants. The physiological and biochemical mechanisms underlying the responses of plants to O3 stress have been well investigated. However, there are limited reports about the molecular basis of plant responses to O3. In this study, a comparative transcriptomic analysis of Pak Choi (Brassica campestris ssp. chinensis) exposed to different O3 concentrations was conducted for the first time. Seedlings of Pak Choi with five leaves were exposed to non-filtered air (NF, 31 ppb) or elevated O3 (E-O3, 252 ppb) for 2 days (8 h per day, from 9:00–17:00). Compared with plants in the NF, a total of 675 differentially expressed genes (DEGs) were identified in plants under E-O3, including 219 DEGs with decreased expressions and 456 DEGs with increased expressions. Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses revealed that O3 stress invoked multiple cellular defense pathways to mitigate the impaired cellular integrity and metabolism, including ‘glutathione metabolism’, ‘phenylpropanoid biosynthesis’, ‘sulfur metabolism’, ‘glucosinolate biosynthesis’, ‘cutin, suberine and wax biosynthesis’ and others. Transcription factors potentially involved in this cellular regulation were also found, such as AP2-ERF, WRKY, JAZ, MYB etc. Based on the RNA-Seq data and previous studies, a working model was proposed integrating O3 caused reactive oxygen burst, oxidation-reduction regulation, jasmonic acid and downstream functional genes for the regulation of cellular homeostasis after acute O3 stress. The present results provide a valuable insight into the molecular responses of Pak Choi to acute O3 stress and the specific DEGs revealed in this study could be used for further functional identification of key allelic genes determining the O3 sensitivity of Pak Choi.


Background
Tropospheric ozone (O 3 ) is generated by reactions between reactive nitrogen oxides (NO x ) and volatile organic compounds (VOCs) in the presence of sunlight, and is one of the key atmospheric pollutants due to anthropogenic activities [1]. Ozone impairs the health of human beings, and also causes serious threats to crops, forests and other ecosystems [2,3].
At the cellular level, we know now a great deal about the mechanisms by which O 3 causes oxidative damages to plants [4,5]. Ozone enters the leaves through the stomata and dissolves in apoplastic solutes, where it then subsequently triggers the production of reactive oxygen species [ROS, such as superoxide anion radical (O 2 .-) and hydrogen peroxide (H 2 O 2 )] in the plant cell wall in a highly regulated manner, and ultimately leads to activation of defense and other metabolic signaling pathways among which ROS scavenging is the most remarkable cellular response to high O 3 concentration [6][7][8].
Analyses on transcriptomic profiling in response to O 3 stress have been reported in a number of model plants, crops and trees, such as Arabidopsis [9][10][11], Medicago truncatula [12], pepper [13], soybean [14], and aspen tree [15], by means of microarray or next-generation sequencing technologies. Many kinds of leafy vegetables are sensitive to O 3 pollution [16]. However, knowledge about transcriptomic response to O 3 stress in leafy vegetables is very limited [17].
Pak Choi (Brassica campestris L. ssp. chinensis L. Makino) is one of the most consumed vegetables in East Asia with high healthy and commercial values [18]. In our previous study, we found that Pak Choi is sensitive to acute O 3 fumigation as reflected by decreased chlorophyll content, increased anthocyanin content, damaged cell membrane integrity, enhanced antioxidative enzyme activities, depressed photosynthetic rate and stomatal conductance, inhibited maximal quantum yield and effective quantum yield of PSII photochemistry [19]. However, the underlying molecular mechanisms for such physiological alterations are not clear. To our best knowledge, there is no previous report on transcriptomic changes of Pak Choi upon O 3 exposure. The goal of this study was to identify and characterize key genes encoding the protein and metabolic pathways involved in O 3 responses in Pak Choi by using RNA-Seq as the methodological approach with Brassica rapa subsp. pekinensis (Lour.) Hanelt as the reference genome.

RNA-sequencing of Pak Choi and DEGs between different O 3 treatments
The second fully expanded leaves from the top of Pak Choi plants under NF and E-O 3 were used for RNA-Seq. The sequencing results were deposited in the NCBI SRA database (Accession number: SRP100739). In total, 172,380,454 and 164,874,922 raw reads were generated from the three replicated NF libraries and the three replicated E-O 3 libraries, respectively (Additional file 1: Table S1). To ensure the quality of the libraries, adaptor reads, ambiguous reads and low-quality reads were removed (Additional file 2: Figure S1). Finally, a total of 165,289,336 and 158,294,170 clean reads were obtained for NF and E-O 3 , respectively (Additional file 1: Table S1), among which 67.63% reads in NF and 69.89% reads in E-O 3 were mapped in the Brassica database (BRAD) (Additional file 1: Table S1). The principle component analysis (PCA) of the RNA-seq data showed that reads of the three NF libraries clustered together while those of E-O 3 clustered together (Additional file 3: Figure S2), further supporting the validity of the experimental design and RNA-seq data. Comparing Pak Choi plants under E-O 3 exposure with those in the NF, a total of 675 DEGs were identified, including 219 DEGs with decreased expressions and 456 DEGs with increased expressions (Additional file 4).

Verification of RNA-Seq data by qRT-PCR
Transcriptional levels of 12 selected DEGs were examined by qRT-PCR in order to validate the reliability of the RNA-Seq data. Among the selected DEGs, three encoded proteins with oxidoreductase activities, two involved in glutathione metabolism, one involved in cell wall formation, and the rest encoded stress-related transcription factors ( Table 1). The result showed that qRT-PCR data were in similar trend to those of the RNA-Seq, proving the reliability of RNA-Seq results (Table 1).

Functional classification of DEGs
GO enrichment analysis were conducted in order to classify the possible functions of DEGs. From the DEGs with increased expressions in plants under E-O 3 exposure, the top enriched terms were 'response to stimulus' , 'oxidation-reduction process' , 'response to stress' , 'multi-organism process' , 'regulation of biological quality' , 'pathogenesis' , and 'defense response' in the category of biological process; in the molecular function category, the 'oxidoreductase activity' , 'sequence specific DNA binding transcription factor activity' , 'nucleic acid binding transcription factor activity' , 'hydrolase activity, acting on glycosyl bonds' and 'hydrolase activity, hydrolyzing O-glycosyl compounds' were the mostly highly enriched; 'extracellular region' and 'apoplast' were highly enriched in the category of cellular component (Fig. 1). For the DEGs with decreased expressions in plants upon E-O 3 treatment, 'organic substance metabolic process' , 'primary metabolic process' , 'phosphorus metabolic process' , 'phosphate-containing compound metabolic process' and 'cellular protein metabolic process' were the highly enriched ones in the category of biological process; and in the category of molecular function, the top enriched terms were 'catalytic activity' and 'anion binding' (Fig. 2).
KEGG pathway enrichment analysis was performed to reveal the enriched pathways. As shown in Fig. 3  . Four DEGs were involved in 'carbon fixation in photosynthetic organisms' , five were in 'glycolysis/gluconeogenesis' (Fig. 8), three were in 'tryptophan metabolism' , four were in 'pyruvate metabolism' , and two were in 'ether lipid metabolism'. Among them, some DEGs were involved in more than one metabolism pathway such as Bra017856, Bra026068, Bra005526, Bra002822, and Bra009352.
Transcription factors (TFs) play key roles in the regulation of gene expression. Candidate TF genes potentially involved in plant's responses to O 3 stress were further analyzed ( Table 3). The results showed that most of the DEGs encoding transcription factors that positively associated with stress-tolerance (e.g. AP2-EREBP, WRKY, Trihelix, MYB, C2H2, TAZ, LOB, SNF2 and PHD) had increased expressions in plants  (Table 3).

Discussion
In the diet of human beings, leafy vegetables always play an important role. Ozone pollution may cause serious injury to leafy vegetables and result in great losses of productivity and quality [16]. In our previous study, E-O 3 caused visible injury and physiological alterations in Pak Choi but the molecular mechanism was not clear [19]. Previous reports showed that high O 3 concentration triggers O 3 -responsive genes [20]. For examples, Ludwików et al. [9] reported the expression patterns of O 3 stress-responsive Arabiodopsis genes with an emphasis on ROS-scavenging (e.g. catalase genes) and secondary metabolism (e.g. phenylpropanoid-related genes); Mahalingam et al. [21] classified O 3 -responsive genes into down-regulated (e.g. coding for proteins that function in chloroplast), early up-regulated (e.g. coding for membrane proteins and those involved in transcription and signaling), and late up-regulated genes (e.g. coding for membrane-associated or secretory proteins, and those involved in ROS-scavenging). In aspen trees, the genes related to defense and signaling were up-regulated while the genes in the carbohydrate metabolism were downregulated [15]. Ludwików et al. [8] summarized the possible functions of differential expressed genes under O 3 stress into several groups as follows: redox control, transcription, signal transduction, metabolism, and defense. In this study, transcriptomic profiles of Pak Choi under NF or E-O 3 were compared using RNA-Seq technology for the first time. It is noteworthy that the experiment was designed to simulate an acute O 3 exposure according to realistic O 3 concentrations in relatively clear area and in a highly O 3 -polluted area or a projected future [22,23]. Based on the results of our previous physiological study with this species [19], we postulated that 2 days were sufficient to analyze the O 3 -responsive genes in Pak Choi. The current results not only characterized the essential key genes, but also unveiled potential metabolic pathways or TFs that are potentially involved in cellular regulation or adaptation against O 3 stress. According to GO enrichment analysis, potential functions of DEGs under O 3 stress have been summarized in several aspects, such as redox control, defense, transcription, signal transduction, and metabolism [8].
In this study, we found that the DEGs between E-O 3 and NF were mainly related to defense, redox control and metabolism. In particular, for the DEGs with increased expressions in plants under E-O 3 , the most enriched ones were 'response to stimulus' , 'oxidation-reduction process' , 'response to stress' , 'pathogenesis' and 'oxidoreductase activity'. The enrichment of genes which related to 'pathogenesis' indicated the similarity between plant response to O 3 and pathogen infection, as both of these stresses triggered the production of ROS [24]. This result is consistent with the other reports in which pathogenesis-related (PR) proteins were induced by O 3 [13,[25][26][27].
Ozone enters the leaves and subsequently triggers the production of ROS that leads to activation of ROS scavenging systems [6][7][8]. In our previous study, we showed that O 3 treatment caused enhanced antioxidant enzyme activities in Pak Choi [19]. ROS are signaling molecules that often result in cell wall strengthening. In this study, cell wall-related genes were also found with increased expressions under O 3 treatment. For example, xyloglucan endotransglycosylase (XETs) played key roles in repairing the damage of cell caused by O 3 attack, promoting cell wall biogenesis and increasing cell or stomatal density. Our result showed that the expressions of XET coding genes increased in response to O 3 , which was consistent with previous reports [28][29][30][31], suggesting that cell wall modification is a common strategy for plant to adapt to high O 3 .
Previous reports showed that the expressions of some genes in 'oxidation-reduction process' or with 'oxidoreductase activity' increased in response to O 3 stress in different plant species [4,[32][33][34][35], and similar results were observed with Pak Choi in this study. Proteomic analysis in poplar also showed that two enzymes with oxidoreductase activity increased under O 3 exposure [36]. These activated oxidoreductase genes potentially function to avoid severe oxidative damage at the cellular level. Furthermore, KEGG pathway analysis also revealed significantly increased expressions of genes involved in the glutathione metabolism. Glutathione S-transferases (GSTs) have functions in preventing oxidative damage and keeping redox homeostasis [37][38][39][40]. Our results showed that the expression of DEGs encoding GSTs increased, suggesting that GSTs were involved in protecting the cell against O 3 stress in this experiment. In addition, expressions of genes (Bra025351 and Bra039047) homologous to CYP81D11, which putatively encodes cytochrome P450 monooxygenases, increased in the present study. It has been reported that expression of CYP81D11 could be induced by osmotic stresses, ABA-, SA-and JA-treatments that the gene probably played important roles in plant detoxification processes [41,42]. Furthermore, cytochrome P450 monooxygenases might be conjugated to glutathione or sugar moieties by either glutathione S-transferases or glycosyl transferases, and then the conjugates were transported to the apoplast or vacuole [43]. The higher expression of GST genes and the GO enrichment in 'extracellular region' or 'apoplast' probably revealed the detoxification processes of Pak Choi to O 3 stress.
Oxidation of membrane lipid in the presence of ROS could lead to accumulation of omega-3 trienoic fatty acids (TFAs) that was the primary precursor of jasmonic acid (JA), one important phytohormone for defense [44][45][46]. As a feedback reaction, jasmonic acid (JA) acts to limit O 3 -lesion spread [7]. In the oxidative cell death cycle, jasmonates protect tissues from ROS-induced cell death and thus counteract the effects of salicylic acid and ethylene [7]. In the present study, the expression level of one putative JAZ7 encoding gene (Bra021923), which belonged to Tify family protein, was silenced when under O 3 treatment. Yu et al. [47] reported that mutation of JAZ7 in the darkness might cause the up-regulation of the genes involved in sulphate metabolism, indole-glucosinolate biosynthesis, callose deposition, and JA-mediated signaling pathways. While the expression of gene (Bra025713) encoding JAZ1 protein, which acts to repress JA signaling, increased in this study [48,49]. Our results of KEGG enrichment showed that the expressions of DEGs related to 'sulfur metabolism' and 'glucosinolate biosynthesis' increased, suggesting the JAZ7 and JAZ1 were both involved in O 3 stress responses and might play opposite functions. This result indicated that O 3 stress potentially activated JA-responsive genes and limited the spread of leaf cell death lesions to protect the healthy tissue, which was confirmed by the observation of foliar visible injury in our previous study [19].
In perceiving the O 3 stress signals as well as JA signal, plants employ multifaceted signaling pathways to regulate their cellular responses. In this study, defenserelated signaling kinase genes and transcription factor genes were also revealed in response to O 3 stress. For example, the expression of a SnRK2 family serine/ threonine-protein kinase gene (Bra015981) increased [50]. The expressions of a number of defense-related AP2/ERF, Trihelix, WRKY, and MYB family genes also increased in response to O 3 stress. It has been documented that members of the AP2/ERF family confer tolerance to multiple stresses [51,52] and are key regulators of redox responsive gene networks [53]. In the present study, we also found that most of the DEGs encoding AP2/ERF family proteins had increased expression levels. Among them, Bra024953 encoding ERF5 had a higher expression level upon E-O 3 than in NF. This result is also consistent with other previous studies [54]. In tomato, expression of ERF5 was induced by abiotic stress, such as drought, wounding etc. [55]. It seems that the up regulation of ERF5 is a common response to abiotic stresses. CBF4 is a regulator for adapting to drought stress [56]. Our results showed that this TF might also be involved in the response to O 3 in Pak Choi.
Trihelix family genes play important roles in plant development, but their responses to abiotic stresses are indistinct to date [57]. Only a few Trihelix family genes were found responsive to abiotic stress such as cold or salt in Brassica species [57,58]. For the first time, we found that the expressions of Trihelix family genes, Bra005127, Bra005688, and Bra028899, increased in Pak Choi when exposed to high O 3 concentration. In particular, Bra005688 encoded GT-3a ortholog, which has a function in binding GTTAC and is light inducible [59]. GT-3b, which is encoded by Bra005127 can be induced by pathogen or salt stress [59,60]. Our result suggested that these two TFs potentially played important roles in O 3 stress regulation.
WRKY proteins with conserved WRKY motif and zinc finger-like domain function as transcriptional activators or repressors [61]. WRKY TFs have been found to be responsive to abiotic stresses such as O 3 [62][63][64]. In this study, three DEGs encoding WRKY family proteins were identified and the expressions of them increased under E-O 3 exposure. WRKY transcription factors were involved in the regulation of senescence-related processes [20,62,65]. It has been reported that 22 Pak Choi WRKY genes were differentially expressed in response to abiotic stresses, such as cold and salinity. [66]. In the present study, we found three new DEGs encoded WRKY28, WRKY48, and WRKY57, respectively, and were probably specifically related to O 3 stress response.
The induction of MYBs also mediated transcriptional reprogramming in response to E-O 3 [62,64,67]. MYBs could regulate the genes related to anthocyanin biosynthesis pathway and the JAZ-DELLA-MYBL2 module upstream of the MYB/bHLH/WD40 complex together mediated abiotic stress-caused anthocyanin accumulation in Arabidopsis [68]. Our results also showed that the expression of one gene encoding MYB family protein increased, which coincided with the accumulation of anthocyanin in O 3 -treated Pak Choi [19].
The other down-stream functional genes, besides those stated above (e.g. XETs & GST), also includes genes involved in sulfur metabolism and phenylpropanoid biosynthesis. Kimura et al. [69] reported that SnRK2.3, which enzyme's putative encoding gene had increased expression, controls the production of O-acetyl-L-serine a putative signaling compound of the sulfur starvation response, and our KEGG enrichment result showed that the expression of DEG (Bra038031) encoded the enzyme catalyzing L-serine to O-acetyl-Lserine in the 'sulfur metabolism' pathway increased. These two genes (Bra015981 and Bra038031) probably work together under the O 3 caused sulfur starvation, among which Bra038031 encoded the enzyme catalyzing L-serine to O-acetyl-L-serine and Bra010645 encoded the enzyme catalyzing adenylyl sulfate to 3′-phosphoadenylyl sulfate.
The expressions of ten DEGs in 'phenylpropanoid biosynthesis' pathway increased. Proteins encoded by Bra037007 and Bra009105 were involved in catalyzing the alcohol to lignin. There were two DEGs which played a key role in catalyzing β-D-glucosyl-2-coumarinate into coumarinate. Flavonoid is a derivative product of the phenylpropanoid biosynthesis pathway. The TRANSPARENT TESTA 12 (TT12) encodes a multidrug and toxic compound extrusion (MATE) vacuolar transporter that is required for flavonoid sequestration in developing seed coat of Arabidopsis [70]. Chai et al. [71] found that the TT12 is less organ-specific and also expressed in leaves of Brassica. Our result showed that the gene (Bra020862) with increased expression and homologous to AtTT12 in the leaves of Pak Choi was probably involved in the defense to O 3 stress.
We also found that PR protein increased with the increasing expression of genes encoding GST. Plasmodesmata callose-binding protein 2 (PDCB2) is identified as a glycosylphosphatidylinositol (GPI)-anchor protein [72]. In the present study, the expression of DEG encoding PDCB2 increased under E-O 3 , suggesting that callose deposition might be enhanced, and cell to cell communication could be inhibited in O 3stressed Pak Choi [73].
The most enriched pathway of genes with decreased expression was 'glycolysis/gluconeogenesis'. Some DEGs in this pathway were also involved in 'carbon fixation in photosynthetic organisms' and 'pyruvate metabolism' pathway. The lower biosynthesis of phosphoenolpyruvate from oxaloacetate, and D-Fructose 6-phosphate from D-Fructose 1, 6-bisphosphate were probably the main reasons for the impairment of photosynthesis and respiration under E-O 3.

Conclusion
RNA-Seq was performed for Pak Choi plants exposed to NF and high O 3 concentration. The transcriptomic comparison together with physiological analysis (published in [19]) revealed that O 3 treatment led to ROS burst and then resulted in lipid oxidation and hormonal (esp. JA) signaling alterations. The signal transduction via phosphorylation (e.g. SnRK2 family kinases) ultimately affected the expression of defense-related transcription factors and down-stream stress-related functional genes (e.g. PR genes, PDCB2 gene, sulfur metabolism and phenylpropanoid biosynthesis-related genes). The proposed working model for acute O 3 -stress signaling network was illustrated in Fig. 9. This result suggested that ROS metabolism, JA pathway and associated downstream functional genes worked together to maintain cellular homeostats and adaptation to O 3 stress. The regulation of plant gene expression included multiple regulatory steps beyond transcriptional regulation; and the current result provided an overall, but not a complete, insight into O 3 effect on Pak Choi. The specific DEGs coding for transcription factors, kinase and functional proteins could be valuable targets for genetic manipulation to improve the O 3 stress tolerance of Pak Choi through experimental approaches in the future.

Plant material and treatments
One commonly cultivated variety of Pak Choi ('Jingguan') was selected in this study. Seeds were from Beijing Vegetable Research Center, Beijing Academy of Agriculture and Forestry Sciences. The seeds were sown in pots (10 cm in diameter), filled with a mixture of peat:vermiculite (3:1,v:v) at 25°C in a greenhouse in July 2015. When they had five leaves in total, 20 seedlings were moved into open top chambers (OTCs) for O 3 treatments.
The OTCs were located in Changping cropland area (40°19'N, 116°13′E), Beijing. Details on the OTCs are in Yuan et al. [74]. After 2 days' adaption to OTC, 20 randomly-picked plants were moved into either of two OTCs. One OTC was fumigated with elevated O 3 (E-O 3 ) and the other one with non-filtered air (NF). Ozone was generated using pure oxygen by a generator and mixed with ambient air using a fan according to Hu et al. [75]. Ozone concentrations at approximately 10 cm above the plant canopy within the OTCs were continuously Fig. 9 Proposed working model for acute O 3 -stress signaling network in Pak Choi measured using an O 3 analyzer (Model 49i, Thermo Scientific, USA). Plants were fumigated for 2 days (8 h per day, from 9:00-17:00) on 21st and 22nd August 2015, and the ozone metrics were calculated as AOT40 (the sum of the differences between hourly O 3 concentrations and 40 ppb for each hour when the concentration is above 40 ppb during daylight hours) according to CLRTAP [76]. In NF, the 16 Table S2). At the end of the exposure, three biological replicates (plants) in each treatment were randomly selected for the following RNA-Seq analysis.

RNA extraction and library preparation for transcriptome analysis
The second leaves from the top were sampled from three plants in each treatment, immediately frozen in liquid nitrogen for RNA preparation. Three μg RNAs per sample were used to generate sequencing libraries using the NEBNext® Ultra™RNA Library Prep Kit for Illumina® (NEB, USA). In brief, mRNA was isolated with poly-T oligo-attached magnetic beads, fragmented using NEBNext First Strand Synthesis Reaction Buffer (5X), and then reverse transcribed into first and second strand cDNA using random hexamer primer by M-MuLV Reverse Transcriptase (RNase H − ) and DNA Polymerase I, respectively. The residual mRNA was removed by RNase H. The remaining overhangs were converted into blunt ends via exonuclease/polymerase activities, and adenylated at the 3′ ends with NEBNext Adaptor with a hairpin loop structure. Furthermore, the AMPure XP system (Beckman Coulter, Beverly, USA) was used to select 150~200 bp cDNA fragments. Then 3 μl USER Enzyme (NEB, USA) was used with adaptor-ligated, size-selected cDNA for 15 min at 37°C followed by 5 min at 95°C before PCR with Phusion High-Fidelity DNA polymerase, Universal PCR primers and Index (X) Primer. Finally, PCR products were purified in AMPure XP system and the library quality was assessed on the Agilent Bioanalyzer 2100 system. The obtained libraries were sequenced on an Illumina Hiseq 2500 platform to generate 125 bp paired-end reads.

Analysis of Illumina sequencing results
In order to obtain clean data (clean reads), low quality reads, reads containing adapters, and reads containing poly-N of raw data were erased. After that Q20, Q30, GC-content of the clean data were calculated. Based on high quality clean data, the following analyses were conducted.

Quantification of gene expression levels and differential expression analysis
The reference genome and gene model annotation files of Brassica were from its database website (http://brassicadb.org/brad/) [77]. Using TopHat (v2.0.12), clean data were mapped back onto the reference genome. The reads numbers mapped to each gene were calculated using the HTSeq (v0.6.1). The expected number of Fragments per Kilobase of transcript sequence per Millions base pairs sequenced (FPKM) of each gene was estimated based on reads numbers mapped to this gene and the gene length. Differentially expressed genes (DEGs) of plants under different treatments (three biological replicates per treatment) were identified by DESeq (P < 0.05, |log 2 (fold change)| > 0.8) [78]. Based on Wallenius non-central hyper-geometric distribution, Gene Ontology (GO) enrichment analysis (P < 0.05) of the DEGs was conducted by the GO seq R packages [79]. The statistical analysis (P < 0.05) of DEGs enrichment in Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways was conducted using KOBAS 2.0 [80].

Real-time quantitative RT-PCR (qRT-PCR) assay
Using qRT-PCR, the expression patterns of 12 genes (Gene ID: Bra003517, Bra013923, Bra020878, Bra028899, Bra031485, Bra038089, Bra034061, Bra035732, Bra010802, Bra012938, Bra00944, and Bra025833) were analyzed (Primers are listed in Additional file 6: Table S3). Actin was used as the reference gene according to [81]. In brief, cDNA was synthesized using ReverseTra Ace qPCR-RT Kit (Toyobo, Japan). The reverse transcription reaction system included 0.5 μL primer mix, 2 μL RNA template, 0.5 μL RT enzyme mix, 2 μL 5 × RT buffer, 5 μL ddH 2 O. In reference to the corresponding unigene sequence, gene-specific primers were designed using the online tool [GenScript Real-time PCR (TaqMan) Primer Design, https://www.genscript.com/ssl-bin/app/primer]. And the efficiency of the primer pairs were checked by serial dilutions of template cDNA as shown in Additional file 7: Figure S3. The cDNA was diluted to 100 ng μL −1 and then used for qRT-PCR test with each gene-specific primers and SYBR® Green Real time PCR Master Mix (Toyobo, Japan) on the Bio-Rad iQ5 real time system. Reactions were conducted at 96°C for 1 min, 40 cycles of 95°C for 15 s, 60°C for 15 s and 72°C for 45 s.

Data analysis
qRT-PCR was conducted for each gene expression analysis using three biological replicates per treatment. The