Transcriptome analysis reveals regulatory framework for salt and osmotic tolerance in a succulent xerophyte

Background Zygophyllum xanthoxylum is a succulent xerophyte with remarkable tolerance to diverse abiotic stresses. Previous studies have revealed important physiological mechanisms and identified functional genes associated with stress tolerance. However, knowledge of the regulatory genes conferring stress tolerance in this species is poorly understood. Results Here, we present a comprehensive analysis of regulatory genes based on the transcriptome of Z. xanthoxylum roots exposed to osmotic stress and salt treatments. Significant changes were observed in transcripts related to known and obscure stress-related hormone signaling pathways, in particular abscisic acid and auxin. Significant changes were also found among key classes of early response regulatory genes encoding protein kinases, transcription factors, and ubiquitin-mediated proteolysis machinery. Network analysis shows a highly integrated matrix formed by these conserved and novel gene products associated with osmotic stress and salt in Z. xanthoxylum. Among them, two previously uncharacterized NAC (NAM/ATAF/CUC) transcription factor genes, ZxNAC083 (Unigene16368_All) and ZxNAC035 (CL6534.Contig1_All), conferred tolerance to salt and drought stress when constitutively overexpressed in Arabidopsis plants. Conclusions This study provides a unique framework for understanding osmotic stress and salt adaptation in Z. xanthoxylum including novel gene targets for engineering stress tolerance in susceptible crop species. Electronic supplementary material The online version of this article (10.1186/s12870-019-1686-1) contains supplementary material, which is available to authorized users.


Background
Drought and salinity are two major environmental stressors that impact crop yields worldwide [1]. Salinity threatens approximately one-fifth of cultivated lands globally. Meanwhile, up to one third of land surfaces are exposed to drought [2,3]. Many experts agree that using molecular genetics to breed crops with higher yields and improved tolerance to abiotic stresses is an effective strategy in safeguarding food supplies. Yet, this task remains one of the greatest challenges faced by modern agriculture [4]. Deciphering key genes and regulatory mechanisms in salt and drought tolerance and adaptation is a key step in engineering stress-tolerant crop plants [5,6].
Mechanisms of abiotic stress tolerance have mainly been studied in model plants. High throughput sequencing and functional genomics tools in model plants have yielded numerous abiotic stress tolerance genes grouped into two major classes: functional genes or regulatory genes [7]. Functional genes include important enzymes and metabolic proteins including detoxification enzymes, water channels, ion transporters, heat shock proteins and late embryogenesis abundant proteins, which directly function to protect cells from stress. Regulatory genes include important signaling and regulatory proteins that modulate protein activity during stress exposure including hormone and stress signaling pathway components, transcription factors, and ubiquitinmediated proteolysis machinery. A large number of studies show that scope for enhancing stress tolerance in plants by altering single functional genes is limited, due to the complexity of stress responses [8,9]. Thus, alteration of key regulatory genes is desirable because this can mimic or enhance stress signals by regulating a large spectrum of downstream stress-responsive genes in conferring tolerance [10,11]. Knowledge of regulatory genes is therefore important from an genetic engineering perspective in plants [11,12].
Despite vast knowledge derived from Arabidopsis thaliana (Arabidopsis) and Oryza sativa (rice) model plants, a low capacity for stress tolerance limits their usefulness as discovery tools. By contrast, xerophyte and halophyte species, widely distributed in arid and saline regions, have evolved multiple protective mechanisms that allow them to grow successfully under hostile conditions [6,13,14]. A detailed understanding of salt and drought protective mechanisms in naturally tolerant species, and the identification of key regulatory genes, is a promising new strategy for breeding salt and drought tolerant crops [13,15].
Zygophyllum xanthoxylum is a succulent xerophyte with a highly developed root system and strong stress tolerance. The natural range of Z. xanthoxylum includes arid and semiarid lands in northwestern China and Mongolia [16]. This shrub is widely planted in China for protecting fragile desert ecosystems and improving vegetation coverage [17]. Previous investigations in Z. xanthoxylum have focused on growth properties, nutritive characteristics, and transpiration resulting in the characterization of several drought and salt stress response functional genes [17][18][19][20][21][22]. Previously, we generated transcriptome datasets of roots and leaves of Z. xanthoxylum to identify differentially expressed genes (DEGs) under osmotic stress and salt treatments [20]. Attention was focused on several important classes of functional genes traditionally associated with drought and salt stress responses, including ion transporters, reactive oxygen species (ROS) scavenging systems, and photosynthesis [20].
Here, we further analyzed the transcriptome and digital gene expression profiling data of Z. xanthoxylum roots under osmotic stress and salt treatments to identify potential upstream regulators of these functional genes ( Fig. 1). Our transcriptome analysis focused on signaling pathways important for stress tolerance, Fig. 1 An overview of relationships between the regulatory gene groups studied in this work and their downstream functional genes. The 6 h and 24 h DEG transcriptomes of Z. xanthoxylum roots under osmotic stress and salt treatment were analyzed. Regulatory gene categories, blue boxes. Functional gene categories, gray boxes transcription factors important for signal output, and ubiquitin proteasome system enzymes important for protein turnover (Fig. 1). To further analyze this transcriptome, DEGs in Z. xanthoxylum roots were matched to putative orthologs in Arabidopsis, which allowed us to perform in silico functional inference, including gene network analysis for protein function, protein subcellular localization, and gene co-expression. Finally, candidate genes were selected and functionally characterized. Among these, Z. xanthoxylum unigenes orthologous to Arabidopsis NAC transcription factor genes NAC083 and NAC035 conferred drought and salt tolerance when constitutively overexpressed in Arabidopsis plants, confirming relevance of the dataset for crop engineering.

Results and discussion
Transcriptome profile of roots during osmotic stress and salt treatments To learn more about how Z. xanthoxylum adapts to drought and saline environments, we further analyzed the transcriptome data of seedling roots exposed to osmotic stress and salt treatments for 6 h and 24 h [20]. Roots were studied because they are the first organs to be exposed to osmotic and salt stress conditions [23]. 6063 and 6258 DEGs were identified in Z. xanthoxylum roots exposed to osmotic stress or salt treatments for 6 h, respectively (Additional file 1: Figure S1a). In osmotic-stressed seedling roots, 4000 DEGs were up-regulated and 2063 DEGs were down-regulated. In salt-treated seedling roots, 4140 DEGs were up-regulated and 2118 DEGs were down-regulated (Additional file 1: Figure S1a). Venn-diagram analysis shows that a large number of 6 h DEGs overlap in expression, including 2780 up-regulated genes and 1402 down-regulated genes that are present in both osmotic stress and salt treatments, suggesting that signaling pathways controlling these responses in Z. xanthoxylum are interacting (Additional file 1: Figure S1a). By comparison, only 2708 and 1307 DEGs were identified in roots exposed to osmotic stress or salt treatments for 24 h, respectively (Additional file 1: Figure S1b). In osmotic-stressed seedling roots, 1723 DEGs were up-regulated and 985 DEGs were down-regulated. In salt-treated seedling roots, 657 DEGs were up-regulated and 750 DEGs were downregulated (Additional file 1: Figure S1b). Similarly, Venn-diagram analysis revealed that 329 up-regulated genes and 309 down-regulated genes overlapped in osmotic stress-and salt-treated seedling roots. Overall, the 6 h dataset was more complex than treatment for 24 h, reflecting that regulatory genes are generally early responders to environmental signals [7,20]. We therefore selected the 6 h dataset as the focus of further analysis.
Among hormone signaling pathways, abscisic acid (ABA) signaling components were significantly enriched followed by auxin, ethylene, and cytokinin signaling pathway terms. ABA plays a well-established role in plant stress response signaling including salinity, drought, osmotic, and cold stresses [24]. During stress responses, cellular ABA levels increase via ABA biosynthesis. PYL/PYR/RCAR (PYRABACTIN RESISTANCE1/ PYR1-LIKE/REGULATORY COMPONENTS OF ABA RECEPTORS) receptors bind to ABA and interact with PP2C (protein phosphatase 2C) negative regulators thereby releasing SnRKs (sucrose non-fermenting like kinases). Activated SnRKs phosphorylate downstream proteins, including bZIP (basic leucine zipper) transcription factors, thereby activating stress tolerance genes [24,25]. DEGs encoding all of these core positive components of ABA signaling were upregulated in Z. xanthoxylum roots under osmotic stress and salt treatments ( Fig. 2b and c, Additional file 2: Table S1). PP2C components showed the most change, with 8 up-regulated and 3 down-regulated genes under osmotic stress conditions and 8 up-regulated and 1 down-regulated gene under salt conditions ( Fig. 2b and c, Additional file 2: Table S1). These data indicate that ABA signaling is an important factor in Z. xanthoxylum stress tolerance, similar to other plants. ABA signaling directly activates stress tolerance genes as well as initiates secondary mechanisms for stress tolerance [26]. Auxin is an important regulator of plant growth and development but its role in abiotic stress responses is poorly studied [27]. Interestingly, numerous auxin signaling components were differentially expressed in osmotic-stressed and salt-treated Z. xanthoxylum roots. Under osmotic stress, 28 auxin signaling pathway genes were upregulated, including 2 TIR1/AFB (Transport Inhibitor Response1/Auxin Signaling F-box) receptor genes, 12 AUX/IAA (Auxin/Indole-3-Acetic Acid) repressor genes, 3 ARF (Auxin Response Factor) transcription factor genes, 3 GH3 (Gretchen Hagen 3) and 8 SAUR (Small Auxin-Up RNA) auxin early-response genes ( Fig. 2d and Additional file 2: Table S2). Under salt stress, 24 auxin signaling pathway genes were upregulated including 2 TIR/ AFB genes, 8 AUX/IAA genes, 4 ARF genes, 4 GH3 genes, and 6 SAUR genes ( Fig. 2e and Additional file 2: Table S2). This enrichment suggests that auxin signaling is an important element in Z. xanthoxylum stress tolerance. Unlike Arabidopsis, Z. xanthoxylum can effectively maintain growth and regulate its root architecture under osmotic stress and salinity [18][19][20]. In turn, auxin plays a wellcharacterized role in stimulating root growth [27]. Collectively, steep changes in ABA and auxin signaling genes at 6 h under osmotic-stress and salt treatments suggests an important interconnecting role for these hormone pathways in the stress response. Osmotic stress-and salt-responsive regulatory genes in Z. xanthoxylum roots Perception of stress signals leads to signal transduction and the activation of protective physiological and metabolic responses [12,28]. To further analyze signal transduction genes in osmotic or salt-treated Z. xanthoxylum DEG libraries, we used Gene Ontology (GO) enrichment analysis. Over-represented categories were found to include protein kinases (253 and 204 kinase genes in osmotic-stressed and salt-treated DEGs libraries, respectively) ( Fig. 3a and b), transcription factors (126 and 143 transcription factor genes in osmotic-stressed and salt-treated DEGs libraries, respectively) ( Fig. 3c and d) and ubiquitin proteasome system (UPS) genes (80 and 85 UPS genes in osmotic-stressed and salt-treated DEGs libraries, respectively) ( Fig. 3e and f).

Protein kinases
Protein kinases play key roles in the perception of stress-related signals. Major classes involved in signal perception are leucine-rich repeat receptor-like kinases (LRR-RLKs), cell-wall associated kinases, lectin-domain-containing receptor kinases, and receptor-like cytoplasmic kinases [29,30]. A large number of these genes are rapidly induced in osmotic or salt stress treated Arabidopsis plants [30,31]. Several have confirmed roles in drought and salt tolerance, including FERRONIA [32] and RECEPTOR-LIKE PROTEIN KINASE 1 [33]. Numerous DEGs encoding RLKs are upregulated in our dataset ( Fig. 3a and b and Additional file 3: Figure S2a). For example, the Z. xanthoxylum FERRONIA-like Uni-gene61098_All is up-regulated 4.6-and 7.5-fold during osmotic stress and salt treatment, respectively (Additional file 2: Table S3). Many RLK genes not previously linked to stress are also differentially upregulated in Z. xanthoxylum roots including LRR-RLK-like Uni-gene9800_All, Unigene176_All, and Unigene12756_All with highly significant blast hits to Hevea brasiliensis BAK1 (BRASSINOSTEROID INSENSITIVE 1-ASSOCIATED RE-CEPTOR KINASE 1) [34], Ricinus communis ERECTA [35], and Prunus avium BAM3 (BARELY ANY MERISTEM 3) [36], respectively (Additional file 2: Table S3). The Arabidopsis counterparts of these genes are well studied but not in the context of abiotic stress responses [34,[36][37][38][39]. Future work to evaluate the role of these genes in stress-related functions using Arabidopsis model plants is a valuable future direction.
Protein kinases are also important for signal transmission. Major classes include CDPKs (calcium-dependent protein kinases) and MAPKs (mitogen-activated protein kinases) [40]. In Arabidopsis, several CDPKs and MAPKs are involved in abiotic stress signaling pathways [41][42][43]. Arabidopsis plants overexpressing CPK6 have enhanced tolerance to salt and drought stresses [42]. Similarly, transgenic barley (Hordeum vulgare) plants overexpressing HvMPK1 showed greater tolerance to salt and other abiotic stresses [44]. Our DEG libraries include 31 and 28 predicted calcium response protein kinase genes, and 28 and 24 predicted MAPK genes in salt-and osmotic stress-treated Z. xanthoxylum roots, respectively ( Fig. 3a and b, Additional file 3: Figure S2b and Additional file 2: Table S3). These data are consistent with known roles for Arabidopsis CDPKs and MAPKs in abiotic stress signaling.

Transcription factors
Transcription factors are important outputs of signaling pathways, directly responsible for the activation or repression of stress-responsive genes [1,12,45]. Major classes include NAC (NAM/ATAF/CUC), AP2/ERF (APETALA2 and ethylene-responsive element binding proteins), bHLH (basic helix-loop-helix), MYB (myeloblastosis), WRKY (WRKY-domain) and bZIP transcription factors whose members confer drought and salt stress tolerance in various plants [1]. For example, many rice WRKY genes exhibit obviously different functions under drought or salt stress [46] and soybean WRKY54 over-expressed in transgenic Arabidopsis confers tolerance to drought and salt stress [47]. bZIP24 was identified by screening salt-inducible transcripts in Arabidopsis and a closely-related halophyte species with functional analysis showing that repression increased salt tolerance and the transcript abundance of numerous stress response genes [48]. Multiple bHLH transcription factors also have stress-related roles such as MYC2 (Myelocytomatosis protein 2) involved in jasmonic acid signaling and AIB (ABA-inducible bHLH-type transcription factor) whose overexpression enhances the drought tolerance of transgenic plants [49,50]. Consistent with these studies, 126 and 144 transcription factors DEGs were identified in Z. xanthoxylum under osmotic stress and salt treatment, respectively. NAC, AP2/ERF, bHLH, MYB, WRKY, and bZIP families were the most abundant ( Fig. 3c and d). Many of these DEGs matched to putative orthologs in other plants with a characterized role in stress biology. For example, CL8432.Contig3_All, CL6477.-Contig2_All, Unigene2817_All, Unigene2222_All, Uni-gene2285_All were up-regulated in osmotic-stressed and salt-treated Z. xanthoxylum (Additional file 3: Figure S2c and Additional file 2: Table S4). These genes encode predicted orthologs of Arabidopsis ERF96 [51], bHLH106 [52], MYB3 [53], WRKY22 [54], and bZIP53 [55], which are previously identified as salt or drought responsive genes with important roles in abiotic stress signaling pathways.
Numerous other transcription factor genes with less-defined roles in abiotic stress were up-or down-regulated in osmotic-stressed and salt-treated Z. xanthoxylum roots ( Fig. 3c and d and Additional file 2: Table S4). Examples include Unigene57160_All, Unigene16368_All, CL6534.Con-tig1_All and Unigene9789_All, which are orthologous to Arabidopsis MYB40, NAC083, NAC035, and WRKY69, respectively (Additional file 2: Table S4). Several transcription factor families not previously associated with abiotic stress mechanisms were also identified. Examples include GATA, GRAS (GAI/RGA/SCR) and ARF transcription factors. In rice, abiotic stress signaling by ABA and salt treatment causes the induction and differential splicing of 28 GATA transcription factor genes revealing tight regulation of transcript abundance and splice variants and possible diverse roles in abiotic stress signaling [56]. Within these families, the expression level of some genes changed significantly. For example, CL11183.Contig1_All predicted to encode a GRAS transcription factor and was up-regulated more than 10-fold in response to osmotic stress and salt treatment in Z. xanthoxylum (Additional file 2: Table S4). Some of these previously uncharacterized transcription factors might represent novel regulators of osmotic stress or salt tolerance.

UPS enzymes
Ubiquitination is an important mechanism for regulating protein turnover in response to stimuli and UPS enzymes play a central role [57,58]. Accordingly, numerous Z. xanthoxylum DEGs induced under osmotic-stress and salt treatment encode UPS-related components such as ubiquitins, proteasome subunits, E1 ubiquitin-activating enzymes, E2 ubiquitin-conjugating enzymes, and E3 ubiquitin ligases (Fig. 3e and f and Additional file 3: Figure S2d). Among these, 22 F-box E3 genes were induced both in osmotic stress-and salt-treated Z. xanthoxylum (Fig. 3e and  f). Four of these genes match to Arabidopsis orthologs with known roles in stress signaling including CL7586.Con-tig2_All (Additional file 2: Table S5). This F-box gene is a predicted ortholog of Arabidopsis TUBBY LIKE PROTEIN3 important for plant responses to ABA, salt, and osmotic stress [59]. Several U-box, RING (really interesting new gene) and HECT (homology to E6-APC terminus) E3 ubiquitin ligase genes are also up-or down-regulated in osmotic stress-or salt-treated Z. xanthoxylum roots ( Fig. 3e and f, Additional file 3: Figure S2d and Additional file 2: Table S5) including CL4067.Contig1_All. This gene is a predicted ortholog of Arabidopsis PUB18 (Plant U-box 18) induced by ABA, drought and salt stress, which functions as a negative regulator of ABA-mediated drought stress responses [60]. These findings highlight potentially important roles for E3 ligases as both positive and negative stress response regulators in Z. xanthoxylum roots.
In summary, 459 and 432 protein kinase, transcription factor, and UPS genes were differentially expressed in osmotic-stressed and salt-treated Z. xanthoxylum DEG libraries, respectively. Hierarchical cluster analyses indicate that transcripts for most of these genes are increased at 6 h but not 24 h (Additional file 3: Figure S2). For example, 103 up-regulated and 41 down-regulated transcription factor genes were among the DEGs of Z. xanthoxylum roots treated for 6 h with salt (Additional file 3: Figure S2) but only 9 of these transcription factor genes were among DEGs at 24 h (Additional file 4: Figure S3). Thus, 6 h DEGs are more likely to be early response regulators in Z. xanthoxylum roots under osmotic stress and salt treatments. To test of the reproducibility of our RNA-sequencing data, we used qRT-PCR to independently monitor the transcript abundance of 20 randomly selected DEGs representing protein kinases, transcription factors, and UPS enzymes. These results were generally consistent with the RNA-sequencing data (Additional file 2: Table S6).

Protein-protein interaction networks of Z. xanthoxylum in response to osmotic stress and salt treatments
To further explore functional relationships among regulatory DEGs from Z. xanthoxylum, BLASTp homology searches against the Arabidopsis TAIR 10 genome release were used to identify putative orthologs. Among 459 osmotic stress-responsive regulatory DEGs in Z. xanthoxylum, 359 unique DEG orthologs were identified in Arabidopsis. Similarly, among the 432 salt-responsive regulatory DEGs in Z. xanthoxylum, 339 unique DEG orthologs were identified in Arabidopsis. STRING analysis was performed to present known gene co-expression, genetic interactions, protein interactions, and protein subcellular localization. Cytoscape was used to calculate and display the most over-represented functional categories, demonstrating a central gene cluster enriched in osmotic stress and salt responsive genes ( Fig. 4a and b; Additional file 5: Figure S4 and Additional file 6: Figure S5). In the osmotic stress-responsive network, 220 DEG proteins, including 89 protein kinases, 55 transcription factors, and 76 UPS enzymes are represented (Fig. 4a, Additional file 5: Figure S4 and Additional file 2: Table S7). In the salt-responsive network, 204 DEG proteins, including 85 protein kinases, 42 transcription factors, and 77 UPS enzymes are represented (Fig. 4b, Additional file 6: Figure S5 and Additional file 2: Table S7). These highly integrated networks incorporated both conserved and novel gene products associated with exposure to osmotic stress and salt conditions in Z. xanthoxylum. For example, Unige-ne22615_All matches to Arabidopsis MKK9, a known MAPK determinant of salt tolerance [61]. When protein kinases, transcription factors, and UPS enzymes directly related to MKK9 are examined, these proteins together with relationships among them form a sub-network showing predicted regulatory events surrounding Unige-ne22615_All in response to osmotic-stress and salt treatment of Z. xanthoxylum, respectively ( Fig. 4c and d). On the other hand, Unigene57160_All blast hits to functionally unresolved Arabidopsis transcription factor gene MYB40. Network analysis between MYB40 and protein kinase, transcription factors, and UPS enzymes reveals potentially novel regulatory events ( Fig. 4e and f ). For example, MYB40 may regulate plant response to osmotic and salt stress by interacting with GATA transcription factors, regulating the transcription level of GATA genes, or GATAs may function upstream of MYB40 ( Fig. 4e and f). These relationships provide a theoretical basis for investigating MYB40 function in stress tolerance. Network analysis also reveals an extensive ABA signaling network in Z. xanthoxylum roots in response to osmotic-stress and salt treatments (Fig. 5 and Additional file 2: Table S1). Under osmotic stress, the majority of ABA pathway genes are up-regulated, including 2 ABA biosynthesis and catabolism genes, 1 transport gene, 1 receptor gene, 1 SnRK gene, 1 ABI transcription factor gene, and various other ABA responsive and UPS genes (Fig. 5b). In total, 30 interrelated proteins shed light on a complex ABA-dependent network ( Fig. 5a and b). Analysis of conserved and novel gene products within these networks provides a foundation for future study (Figs. 4 and 5 and Additional file 2: Table S7).

Validation of stress-related functions for novel ZxNAC regulatory genes
To further analyze the biological relevance of our dataset, novel Z. xanothxylum regulatory DEGs from our study were tested as potential osmotic and salt stress determinants in Arabidopsis. Eight candidate genes were selected for functional characterization including two protein kinases genes [Unigene9800_All (ZxSERK1) and CL11556.Contig3_All   Figure S4 and Additional file 6: Figure S5. c,d predicted MKK9 (Unigene22615_All) module showing regulatory proteins and relationships among them (nodes and edges) for Z. xanthoxylum roots in response to (c) osmotic stress and (d) salt, respectively. e, f predicted MYB40 (Unigene57160_All) module showing regulatory proteins and relationships among them (nodes and edges) for Z. xanthoxylum roots in response to (e) osmotic stress and (f) salt, respectively. Green nodes and edges (lines), kinases and interactions among them; magenta nodes and edges, transcription factors and interactions among them; blue nodes and edges, UPS enzymes and interactions among them; gray edges, interactions among kinases, transcription factors, and UPS enzymes except those among the same category of proteins after their putative orthologs in Arabidopsis. Independent qRT-PCR analysis confirmed that both genes are highly induced in osmotic-stressed and salt-treated Z. xanthoxylum roots (Additional file 7: Figure S6a, b and Additional file 2: Table S6). Transgenic Arabidopsis plants expressing these genes under the control of a strong constitutive Cauliflower Mosaic Virus 35S promoter were next generated. Two independent transgenic lines per construct were selected for phenotypic analyses under normal, drought, and salt-stress conditions (Additional file 7: Figure S6c, d). The drought tolerance of wild type Col-0 and transgenic lines was first compared by withholding water for 7 days [62]. Under this treatment, wild-type plants began to wilt but the ZxNAC083 and ZxNAC035 transgenic plants grew well (Additional file 8: Figure S7). To further analyze growth and physiological parameters, period drought stress was applied by withholding water for 2 weeks followed by normal watering for 7 days [63]. The wild-type treated Arabidopsis plants had etiolated and wilted leaves, whereas the transgenic lines grew well ( Fig. 6a, b, g, and h). To quantify effects on growth, the dry weight of stems was measured. Under normal watering conditions, transgenic plants had slightly shorter stems with a lower dry weight compared to wild-type plants (Fig. 6a, c, g, and i). However, wildtype plant growth was more inhibited by drought compared to transgenic plants (Fig. 6b, c, h, and i). The dry weight of stems from transgenic plants under control and drought stress conditions was similar (Fig. 6c and i). Physiological parameters including relative water content, chlorophyll content, and net photosynthetic rate were also measured. Under well-watered conditions, wild-type and transgenic plants exhibited no significant differences ( Fig. 6d and j). After period drought treatment, the relative water content of transgenic plants was higher than wild-type Arabidopsis, suggesting that ZxNAC083 and ZxNAC035 can regulate osmotic homeostasis in plants under drought conditions (Fig. 6d and j). Water deficit can lead to a decrease in chlorophyll content thus lowering net photosynthetic rate [63,64]. The chlorophyll content of transgenic plants was maintained at Fig. 5 Module for ABA signaling pathway in osmotic stress responses of Z. xanthoxylum roots. a Functional network based on known Arabidopsis ABA-dependent signaling components, cross-referenced against the complete DEG functional network. Node colors represent different classes of signaling components: orange, ABA biosynthesis; red, receptors; yellow-green, PP2Cs, green, kinases; dark magenta, transporters; magenta, transcription factors; and blue, UPS enzymes. Border colors of nodes represent expression as log2 fold-change. Magenta, increased expression; the deeper the color, the higher the expression. Green, decreased expression; the deeper the color, the lower the expression. Edge (line) colors, interactions of proteins in the core ABA signaling pathway (red), kinases (green), UPS enzymes (blue), transcription factors (magenta), and proteins except in the same class (cyan). Thickness of edges, normalized link closeness; the thicker the edge, the closer the link. b ABA signaling and biosynthetic pathway components. Shading, colors represent different classes of signaling components: orange, ABA biosynthesis; red, receptors; yellow-green, PP2Cs, green, kinases; dark magenta, transporters; magenta, transcription factors; and blue, UPS enzymes Data are mean ± SD of three replicates. c,i Asterisks, significant difference. p < 0.05 (t-test). (d, e, f, j, k, l) Asterisks, significant difference. p < 0.05 (Duncan's test) more normal levels compared to Col-0 plants ( Fig. 6e  and k). Transgenic plants also showed a higher net photosynthetic rate compared to Col-0 under drought treatment ( Fig. 6f and l).
Finally, we compared the salinity tolerance of wild type and transgenic lines. Plants were irrigated with 100 mM NaCl solution for 2 weeks. After treatment, the rosette leaves of Col-0 were faded and withered whereas transgenic plants grew well (Fig. 7b and h). Col-0 plants had longer stems with a greater dry weight compared to transgenic plants under normal conditions ( Fig. 7a and  g). However, wild-type plant growth was more severely inhibited by salt stress compared to transgenic plants ( Fig. 7c and i). Physiological parameters including relative water content, chlorophyll content and net photosynthetic rate were also measured. Salinity can decrease soil water potential leading to physiological drought. While the relative water content of wild-type and transgenic plants was similar under normal conditions, ZxNAC083 and ZxNAC035 overexpression lines showed significantly higher relative water content compared to wild-type plants after salt treatment ( Fig. 7d and j). High relative water content improves chlorophyll stability and photosynthetic system integrity under physiological drought. Under salt stress, both chlorophyll content and net photsynthetic rate of ZxNAC83 and ZxNAC035 transgenic lines were significantly higher than Col-0 wild-type showing a protective effect (Fig. 7e, f, k, and l).
These combined data demonstrate that overexpression of ZxNAC083 and ZxNAC035 enhances Arabidopsis drought and salt tolerance. These data provide powerful evidence that regulatory genes identified in this study include authentic osmotic and salt stress determinants in Z. xanthoxylum. Many of these genes have unresolved functions in stress tolerance.

Conclusions
This study provides a first comprehensive analysis of regulatory gene transcripts in Z. xanthoxylum under osmotic stress and salt treatments. These data reveal a complex network of conserved and unique regulatory genes induced in Z. xanthoxylum roots with potential roles in abiotic stress tolerance. Functional analysis of these genes is a rich resource for crop improvement.

Data acquisition and differential expression analysis
The transcriptomic data and digital gene expression libraries (C6, C24, D6, D24, S6, and S24) analyzed in this study were generated as previously described [20]. Seeds of Z. xanthoxylum used in this work were collected from wild plants in Alxa League of Inner-Mongolia Autonomous Region, China, and were stored in the Key Laboratory of Grassland Livestock Industry Innovation, Ministry of Agriculture and Rural Affairs, Lanzhou, China. The raw data are submitted to the NCBI Sequence Read Archive (PRJNA512400). Each transcriptomic library was constructed from seedling roots, using 5-6 seedlings per sample. Libraries C6 and C24 correspond to treatment with a mock solution (1/2 strength Hoagland nutrient solution) for 6 h and 24 h, respectively; D6 and D24 correspond to treatment with an osmotic stress solution (1/2 strength Hoagland nutrient solution with added sorbitol, equivalent to − 0.5 MPa osmotic potential, sorbitol concentration ≈ 165 mM) for 6 h and 24 h, respectively; and S6 and S4 correspond to treatment with a salt solution (1/2 strength Hoagland nutrient solution with 50 mM NaCl) for 6 h and 24 h, respectively. Libraries were constructed using a tag-based digital gene expression protocol [20]. The resulting libraries were sequenced in parallel using an Illumina HiSeq™ 2000 sequencing platform (BGI Shenzhen). Low quality tags from each library were removed (e.g. tags with unknown nucleotide "N", "empty tags" with no tag sequence between the adaptors, and tags with only one copy number). The resulting "clean" tags were mapped to our transcriptome reference database [20]. To establish differential expression, the number of clean tags for each gene was calculated and normalized to the number of transcripts per million clean tags [65]. To compare differences in gene expression, the tag frequency in each library was statistically analyzed according to Audic and Claverie [66]. The false discovery rate (FDR) was used to determine the threshold p-value in multiple tests [65,67]. Genes that displayed FDR<0.001 and absolute value of the log2 ratio>1 were selected as significantly different.
DEGs regulatory pathways analysis, hierarchical clustering, and validation by qRT-PCR The Blast2GO program (https://www.blast2go.com/) was employed to obtain Gene Ontology (GO) annotations for Unigenes. Regulatory pathways were investigated by matching Z. xanthophyllum genes to putative orthologs in the Kyoto Encyclopedia of Genes and Genomes (KEGG) protein database (www.genome.jp/dbget/). Hierchical cluster was performed using Heml software (http://hemi.biocuckoo.org/down.php). Default options were used and Pearson correlations were performed on gene expression data.
RNA sequencing data were verified by measuring the transcript abundance of randomly-selected predicted regulatory genes using total RNA extracted from treated roots treated as described above [20]. First-strand cDNA was synthesized from 2 μg of DNase-treated RNA according to the manufacturer's instructions (TaKaRa Biotechnology, China). qRT-PCR was performed in triplicate on three bioreplicates using Power SYBR™ Green Master Mix (TaKaRa Biotechnology, China) on a StepOne Real-Time PCR Thermocycler (Applied Biosystems). To normalize sample variance, ZxACTIN (GenBank accession no. EU019550) was used as the internal control gene. The relative expression level of each gene were determined using the 2 -ΔΔCt method [65]. Primer-BLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast) was used for primer design. A standard curve and efficiency test were performed for each set of primers (Additional file 2: Table S8).  c Dry weight of stems, d relative water content, e chlorophyll content, and (f) net photosynthetic rate. i, j, k, l Physiological parameters of Col-0, ZxNAC035 overexpression transgenic plants under normal conditions and salt stress. i Dry weight of stems, (j) relative water content, (k) chlorophyll content, and (l) net photosynthetic rate. Data are mean ± SD of three replicates. c, i Asterisks, significant difference. p < 0.05 (t-test). (d, e, f, j, k, l) Asterisks, significant difference. p < 0.05 (Duncan's test) against the reference Arabidopsis TAIR10 genome annotation (http://www.arabidopsis.org). Protein-protein interaction analysis of protein function, protein subcellular localization, and gene co-expression of putative Arabidopsis orthologs was carried out in STRING [68] (http://www.string-db.org/). STRING protein-protein interaction images were re-modulated in Cytoscape [69] (https://cytoscape.org/).

Generating overexpression lines
The coding regions of Unigene16368_All (ZxNAC083) and CL6534.Contig1_All (ZxNAC035) were amplified from seedling total RNA using a SMART RACE cDNA Amplification Kit (TaKaRa Biotechnology, China). The resulting cDNA products were cloned into pDONR™/Zeo using a Gateway® BP reaction (ThermoFisher Scientific, China) then inserted into binary vector pBIB-BASTA-35S-GWR-GFP using a Gateway® LR reaction (ThermoFisher Scientific, China). The resulting constructs were introduced into Agrobacterium tumefaciens strain GV3101 and used to transform wild-type Arabidopsis Col-0 plants by floral dipping (seeds of wild-type Arabidopsis Col-0 were obtained from Ministry of Education Key Laboratory of Cell Activities and Stress Adaptation, Lanzhou, China) [70]. Transgenic overexpression lines were validated using semi-quantitative RT-PCR to measure ZxNAC083 and ZxNAC035 transcript levels. AtACTIN2 was used as the internal control gene. Primer sequences are provided in Additional file 2: Table S8.

Phenotypic and physiological assessments of transgenic plants
Phenotypic assays were performed on Arabidopsis wild-type Col-0 and T3 homozygous transgenic overexpression lines using two independent lines per construct (ZxNAC083-OX48, OX65 and ZxNA-C035-OX57, OX59). 4-week-old plants grown on soil in pots (8 × 8 × 8 cm 3 ) were subjected to drought or salt stress treatments as described with minor changes [71]. For period drought stress treatments, water was withheld for 2 weeks followed by normal watering for 7 days to permit recovery [71]. Under normal conditions, each pot was given 150 mL of water every 2 days. For salt stress treatments, each pot was given 150 mL of 100 mM NaCl solution every 2 days for 2 weeks. The dry weight of stems (shoots excluding rosette leaves), relative water content, chlorophyll content, and net photosynthetic rate were determined as previously described [19]. Seedlings were grown in a greenhouse at 20-22°C with relative humidity 65-75% under long days (16 h light/8 h dark, photon flux density 100-120 μmol·m − 2 ·s − 1 ). .