Background

As an essential component of key macromolecules, nitrogen (N) is quantitatively the most important mineral nutrient for plants [1]. In soils, N is often the most important factor limiting plant growth, and plants frequently encounter N deficiency in their natural habitats. In the past several decades, the increasing use of N fertilizers in crop production has played a major role in increasing yields [2]. However, the main problem is the fact that crop plants only use less than half of the applied N [3], with the remaining N causing severe environmental pollution. There is thus an impending need to realize high productivity while decreasing the rate of N application. This necessitates a comprehensive understanding of molecular mechanisms underlying morphological and physiological adaptation to low N (LN) stress in crops.

Under LN stress, the sessile plants have evolved many adaptive responses. Plants improves the efficiency of N uptake by modifying root architecture, enlarging root system and enhancing the expression of high-affinity transport systems for nitrate and ammonium [4,5,6,7,8,9]. Meanwhile, N utilization efficiency could be also improved in the plants [10]. In addition, the remobilization of N from source organs might be stimulated when plants are subjected to N limitation [11], resulting in the enhancement of N re-assimilation, and maintenance of N economy in plants [12, 13]. Furthermore, the expression levels of a number of genes associated with N metabolism in plants were altered to ensure the survival or complete their life cycle [14, 15].

Tibetan annual wild barley is regarded as one of the progenitors of cultivated barley [16, 17], possessing wider genetic diversity and generally better adaption to N deficiency in comparison with cultivated barley [18]. Some wild barley genotypes with high LN tolerance were identified, providing the elite genetic materials for improving LN tolerance of barley as well as other cereal crops [19]. Two wild barley genotypes differing dramatically in LN tolerance were used for transcriptome analysis at early stage of LN stress (6 h and 48 h) in our previous study [20]. However, plants respond to nutrient deficiency by inducing or repressing different sets of genes at special time [Transcription factors and hormone signaling

Expression of plant-specific Dof1 TFs has been proved to improve plant growth under LN condition [39, 40]. Several R2R3-type MYB TFs are involved in plant stress responses [41], and over-expression of OsMYB48–1 enhanced drought and salinity tolerance [

Methods

Plant materials and LN treatment

The experiment using two Tibetan wild barley accessions XZ149 and XZ56 (LN- tolerant and sensitive genotypes, respectively) was carried out in black plastic pots (5 L) in a greenhouse with natural light. The wild barley accessions were collected from Tibetan area in last century and kindly presented by professor Sun of Huazhong Agricultural University, China. The nutrition solution was used according to Quan et al. [20]. The solution was renewed every five days, continuously aerated with pumps. Treatments were conducted on three-leaf-stage seedlings with two N levels (0.2 mM N as LN treatment, 2 mM N as control).

For biomass determination, the plants were harvested at 12 d after LN treatment and separated into shoots and roots. Dry weight was recorded after the samples were dried at 105 °C for 30 min and to constant weight at 80 °C. Meanwhile fresh plant tissues were taken for use in determining nitrate reductase (NR) activity, glutamine synthetase (GS) activity and soluble protein content with three biological replications, and the content of metabolites with four biological replications. The roots of the two barley accessions under both N treatments were sampled with three biological replicates at 1 d, 2 d, 4 d, 8 d and 12 d after treatments for the time course analysis of gene HvNRT2.1 expression. For RNA-Seq analysis, the samples were taken at 12 d after treatments. Roots of four seedlings for each treatment were pooled as one biological replication. Totally eight samples [2 genotypes × 2 treatments × 2 biological replications] were taken for analysis.

Physiological measurement

N concentration in plant tissue was determined using Foss Kjeltec 8400. Soluble protein content was measured as described by Andrews et al. [52]. NR and GS activity were determined according to Kaiser et al. [53] and Masclaux-Daubresse et al. [54], respectively. The metabolites were extracted and analyzed using 7890A/5975C GC–MS system (Agilent, USA) and AMDIS 32 software according to Quan et al. [22].

cDNA library construction and sequencing for RNA-seq

The total RNA was extracted using miRNeasy mini kit (QIAGEN, Germany) following the manufacturer’s specification. RNA degradation, integrity, abundances and purity were assessed for meeting the requirements [20]. cDNA libraries were constructed using the Illumina TruSeq™ RNA Sample Preparation Kit (Illumina, San Diego, CA, USA) according to the manufacturer’s instructions. Briefly, mRNA was obtained from the total RNA using magnetic beads with poly-T oligonucleotide. Then the random fragmentation of the purified mRNA was reversely transcribed into cDNA. After ligated with the adapters on both ends, DNA fragments were selectively amplified and enriched. Subsequently, the purified PCR products were quantified using Agilent Bioanalyzer 2100 system. After cluster generation, the final cDNA library was sequenced on an Illumina NextSeq 500 platform.

Raw reads with 75 bp single-end were initially processed to remove adapters sequences, empty sequences and low-quality bases, and then the Q20, Q30, GC contents, and sequence duplication level of the clean data were analyzed. Then the clean reads were mapped against the barley reference genomes using TopHat (http://tophat.cbcb.umd.edu/), and finally the map** results were analyzed to identify splice junctions.

Identification of the DEGs and validation of RNA-Seq by real time PCR

Gene expression levels were calculated by the FPKM values (fragments per kilobase of exon per million fragments mapped reads) [55]. Fold-changes were defined as normalized read count abundance for the LN-stressed samples divided by that of the control samples. To identify differentially expression genes (DEGs), the difference in expression between control and LN treatment was analyzed using the DESeq R package (1.10.1) [56]. An FDR (false discovery rate) of 0.05 was used for determining significant DEGs [57].

To validate the reliability of the RNA-Seq results, the expression of candidate genes was determined by real time PCR assay using the RNA for RNA-Seq. The first strand cDNA was synthesized using PrimerScript™ RT reagent Kit with gDNA Eraser (Takara, Japan). The gene-specific primers, designed by primer-blast (http:/www.ncbi.nlm.nih. gov/tools/primer-blast/), were presented in Additional file 11: Table S1. All real time PCR analyses were performed on a CFX96 system (Bio-Rad USA) with two biological replicates and three technical replicates. HvGAPDH was used as an internal control. The relative expression was calculated by the comparative CT method and expressed as the fold change referred to the expression in the control plants [58].

Statistical analysis

Gene Ontology (GO) annotation and KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis for the DEGs were conducted using the Blast2GO program [59] according to Zeng et al. [60]. GO terms were tested by applying tools for GO enrichment (http://systemsbiology.cau.edu.cn/agriGOv2/) at p-values ≤0.05 [61]. Venn diagram was made on jvenn (http://jvenn.toulouse.inra.fr/app/example.html) [62]. Heatmaps and hierarchical clustering were generated with genesis 1.8.1. Significant differences of gene expression between treatments were tested using a DPS statistical software, and the difference at P < 0.05 was considered as significant.