The bZIP Transcription Factor GmbZIP15 Negatively Regulates Salt- and Drought-Stress Responses in Soybean

Soybean (Glycine max), as an important oilseed crop, is constantly threatened by abiotic stress, including that caused by salinity and drought. bZIP transcription factors (TFs) are one of the largest TF families and have been shown to be associated with various environmental-stress tolerances among species; however, their function in abiotic-stress response in soybean remains poorly understood. Here, we characterized the roles of soybean transcription factor GmbZIP15 in response to abiotic stresses. The transcript level of GmbZIP15 was suppressed under salt- and drought-stress conditions. Overexpression of GmbZIP15 in soybean resulted in hypersensitivity to abiotic stress compared with wild-type (WT) plants, which was associated with lower transcript levels of stress-responsive genes involved in both abscisic acid (ABA)-dependent and ABA-independent pathways, defective stomatal aperture regulation, and reduced antioxidant enzyme activities. Furthermore, plants expressing a functional repressor form of GmbZIP15 exhibited drought-stress resistance similar to WT. RNA-seq and qRT-PCR analyses revealed that GmbZIP15 positively regulates GmSAHH1 expression and negatively regulates GmWRKY12 and GmABF1 expression in response to abiotic stress. Overall, these data indicate that GmbZIP15 functions as a negative regulator in response to salt and drought stresses.


Introduction
As a result of their sessile nature, plants are subject to variable biotic and abiotic-stress conditions. As the most pertinent abiotic-stress conditions, drought and salinity threaten the growth and productivity of crops. Plants respond and adapt to these stress conditions by activating stress-related pathways, which comprise signal perception and transduction, regulation of gene expression, and biochemical and physiological responses [1]. Signaling pathways, including those involving various phytohormones, multiple secondary metabolism processes, and reactive oxygen species (ROS), are crucial for plant survival under environmental-stress conditions [2][3][4].

GmbZIP15 Expression Pattern in Response to Abiotic-Stress Conditions
We previously identified 160 bZIP genes from soybean and characterized their expression in response to abiotic stress [24]. Among these genes, the expression of GmbZIP15 was suppressed by drought and flooding stress and it was therefore selected for further investigation. To validate the response of GmbZIP15 to abiotic-stress conditions, we generated soybean plants overexpressing GmbZIP15 (OX-GmbZIP15), and two lines (OE-8, OE-16) with higher expression level were selected for further research ( Figure S1). The qRT-PCR analysis was used to determine GmbZIP15 expression patterns in 2-week-old wild-type (WT) and two overexpression lines under salt and drought treatments. Under normal conditions, GmbZIP15 expression level in OX-GmbZIP15 plants was clearly higher than that in WT; however, GmbZIP15 expression in OX-GmbZIP15 plants treated with NaCl and mannitol sharply decreased, but was higher than WT plants by 12 h and was nearly undetectable by 24 h post-treatment ( Figure S2A,B). In addition, through the GUS staining of 1-week-old pGmbZIP15:GUS transgenic Arabidopsis seedlings grown on media supplemented with 100 mM NaCl or 200 mM mannitol, GmbZIP15 promoter activity was observed to be significantly decreased in cotyledons and true leaves under NaCl and mannitol conditions compared to normal conditions ( Figure S2C), which was consistent with the results of qRT-PCR ( Figure S1A,B). These results suggest that the expression of GmbZIP15 is suppressed by abiotic stress.

GmbZIP15 Negatively Regulates Salt and Drought Tolerance in Soybean
To investigate the role of GmbZIP15 in plant response to salt stress, transgenic soybean plants carrying functional repression of GmbZIP15 (35S:GmbZIP15-SRDX) were obtained and two lines (SRDX-15, SRDX-21) with higher expression levels were selected for further research ( Figure S1). WT, OX-GmbZIP15 and 35S:GmbZIP15-SRDX seedlings were treated with 200 mM NaCl. Following 2 weeks of salt treatment, WT and 35S:GmbZIP15-SRDX soybean seedlings exhibited a comparable degree of leaf shedding, whereas OX-GmbZIP15 plants displayed a severe, almost lethal wilt phenotype ( Figure 1A). These results suggest that overexpression of GmbZIP15 in soybean causes sensitivity to salt stress.
To test the function of GmbZIP15 in plant drought responses, WT, OX-GmbZIP15 and 35S:GmbZIP15-SRDX seedlings were withheld water for 2 weeks. Compared to WT and 35S:GmbZIP15-SRDX seedlings, OX-GmbZIP15 seedlings were severely wilted and almost all leaves displayed a considerable dehydration phenotype (Figure 2A). To test whether the dehydration phenotype could be rescued, we rewatered the drought-treated plants for 3 days. Although there was slight shedding of the old leaves of WT and 35S:GmbZIP15-SRDX seedlings after rewatering, there was vigorous new leaf growth, suggesting that WT and 35S:GmbZIP15-SRDX plants can recover from such a dehydration phenotype. In contrast, OX-GmbZIP15 seedlings did not display growth recovery after rewatering. Combined, these results further support that GmbZIP15 acts as a potential negative regulator of abiotic-stress response.
In addition, the transcription levels of drought-/salt-tolerance marker genes, including GmDREBb, GmMYB118, and GmWRKY28, were evaluated by qRT-PCR in WT and OX-GmbZIP15 plants. Under mock conditions, the expression levels of each marker were considerably higher in OX-GmbZIP15 plants compared to WT plants ( Figure S3A,B). However, under salt stress, the expression levels of GmDREBb, GmMYB118, and GmWRKY28 dramatically decreased in OX-GmbZIP15 plants compared with that in WT plants ( Figure S3A). Similarly, under drought stress, the expression levels of GmDREBb, GmMYB118, and GmWRKY28 genes in OX-GmbZIP15 plants also decreased significantly compared with that in WT plants ( Figure S3B). These results indicate that GmbZIP15-overexpressing soybean plants are hypersensitive to abiotic stress. To test the function of GmbZIP15 in plant drought responses, WT, OX-GmbZIP15 and 35S:GmbZIP15-SRDX seedlings were withheld water for 2 weeks. Compared to WT and 35S:GmbZIP15-SRDX seedlings, OX-GmbZIP15 seedlings were severely wilted and almost all leaves displayed a considerable dehydration phenotype (Figure 2A). To test whether the dehydration phenotype could be rescued, we rewatered the drought-treated plants for 3 days. Although there was slight shedding of the old leaves of WT and 35S:GmbZIP15-SRDX seedlings after rewatering, there was vigorous new leaf growth, suggesting that WT and 35S:GmbZIP15-SRDX plants can recover from such a dehydration phenotype. In contrast, OX-GmbZIP15 seedlings did not display growth recovery after rewatering. Combined, these results further support that GmbZIP15 acts as a potential negative regulator of abiotic-stress response.  In addition, the transcription levels of drought-/salt-tolerance marker genes, including GmDREBb, GmMYB118, and GmWRKY28, were evaluated by qRT-PCR in WT and OX-GmbZIP15 plants. Under mock conditions, the expression levels of each marker were considerably higher in OX-GmbZIP15 plants compared to WT plants ( Figure S3A,B). However, under salt stress, the expression levels of GmDREBb, GmMYB118, and GmWRKY28 dramatically decreased in OX-GmbZIP15 plants compared with that in WT plants ( Figure S3A). Similarly, under drought stress, the expression levels of GmDREBb, GmMYB118, and GmWRKY28 genes in OX-GmbZIP15 plants also decreased significantly compared with that in WT plants ( Figure S3B). These results indicate that GmbZIP15-overexpressing soybean plants are hypersensitive to abiotic stress.

GmbZIP15 Depresses the ROS Scavenging Ability of Soybean
Abiotic stress can lead to damage to plant cells via oxidative stress involving the generation of ROS [31]. Diaminobenzidine (DAB) staining showed that H 2 O 2 levels were largely increased in the leaves of OX-GmbZIP15 soybean plants compared with WT plants as indicated by the larger amount of reddish-brown precipitate observed following the treatment with 200 mM NaCl and 300 mM mannitol ( Figures 1B and 2B). By contrast, the H 2 O 2 contents in 35S:GmbZIP15-SRDX soybean plants were comparable with that in WT plants under salt-or drought-stress conditions ( Figures 1B and 2B). We further investigated whether altered H 2 O 2 contents reflected altered ROS-scavenging capability in these plants. For this, the activities of the two main antioxidant enzymes involved in ROS scavenging, namely, peroxidase (POD) and catalase (CAT), were determined in WT, OX-GmbZIP15, and 35S:GmbZIP15-SRDX soybean seedlings before and after 24-h treatment with 200 mM NaCl and 300 mM mannitol. Before stress treatment, OX-GmbZIP15 soybean plant showed higher activities of POD and CAT than WT plants ( Figure S4A,B), whereas following salt-and drought-stress treatments, there was a marked decrease in POD and CAT activities in OX-GmbZIP15 soybean plant compared to WT plants ( Figure S4A,B). These results indicate that ROS scavenging was depressed in GmbZIP15-overexpressing soybean plants upon abiotic stress.

Changes of Stomatal Aperture in GmbZIP15 Transgenic Soybean Plants during Abiotic-Stress Conditions
Abiotic stress usually leads to a reduction in plant water loss through the regulation of the stomata aperture [32]. Thus, we analyzed stomatal regulation and its possible association with the stress-sensitive phenotype of GmbZIP15 transgenic soybean plants. The stomatal apertures (width/length, W/L) of WT, OX-GmbZIP15, and 35S:GmbZIP15-SRDX soybean plants were measured under control and abiotic-stress conditions. As shown in Figure 3, WT and transgenic soybean plants displayed similar stomatal apertures under control conditions. Moreover, following 200 mM NaCl and 300 mM mannitol treatments, there were no obvious differences in stomatal apertures between 35S:GmbZIP15-SRDX and WT plants; however, a greater stomatal aperture was observed in OX-GmbZIP15 plants compared to WT and 35S:GmbZIP15-SRDX plants ( Figure 3A,B). These data show that there is defective stomatal closure in OX-GmbZIP15 plants upon abiotic stress.
of reddish-brown precipitate observed following the treatment with 200 mM NaCl and 300 mM mannitol ( Figures 1B and 2B). By contrast, the H2O2 contents in 35S:GmbZIP15-SRDX soybean plants were comparable with that in WT plants under salt-or drought-stress conditions ( Figures 1B and  2B). We further investigated whether altered H2O2 contents reflected altered ROS-scavenging capability in these plants. For this, the activities of the two main antioxidant enzymes involved in ROS scavenging, namely, peroxidase (POD) and catalase (CAT), were determined in WT, OX-GmbZIP15, and 35S:GmbZIP15-SRDX soybean seedlings before and after 24-h treatment with 200 mM NaCl and 300 mM mannitol. Before stress treatment, OX-GmbZIP15 soybean plant showed higher activities of POD and CAT than WT plants ( Figure S4A,B), whereas following salt-and drought-stress treatments, there was a marked decrease in POD and CAT activities in OX-GmbZIP15 soybean plant compared to WT plants ( Figure S4A,B). These results indicate that ROS scavenging was depressed in GmbZIP15-overexpressing soybean plants upon abiotic stress.

Changes of Stomatal Aperture in GmbZIP15 Transgenic Soybean Plants during Abiotic-Stress Conditions
Abiotic stress usually leads to a reduction in plant water loss through the regulation of the stomata aperture [32]. Thus, we analyzed stomatal regulation and its possible association with the stress-sensitive phenotype of GmbZIP15 transgenic soybean plants. The stomatal apertures (width/length, W/L) of WT, OX-GmbZIP15, and 35S:GmbZIP15-SRDX soybean plants were measured under control and abiotic-stress conditions. As shown in Figure 3, WT and transgenic soybean plants displayed similar stomatal apertures under control conditions. Moreover, following 200 mM NaCl and 300 mM mannitol treatments, there were no obvious differences in stomatal apertures between 35S:GmbZIP15-SRDX and WT plants; however, a greater stomatal aperture was observed in OX-GmbZIP15 plants compared to WT and 35S:GmbZIP15-SRDX plants ( Figure 3A,B). These data show that there is defective stomatal closure in OX-GmbZIP15 plants upon abiotic stress.  Errors bars indicate ± SD of three biological replicates. Significant differences between samples labeled a, b, and c were determined by one-way ANOVA, p < 0.05.

Conservation of GmbZIP15-Mediated Abiotic-Stress Responses in Soybean and Arabidopsis
To further investigate the function of GmbZIP15 in response to abiotic stress, OX-GmbZIP15 and 35S:GmbZIP15-SRDX transgenic Arabidopsis plants were generated and two lines of each with higher expression levels were selected for further research ( Figure 4B). Five-week-old soil-grown WT, OX-GmbZIP15 and 35S:GmbZIP15-SRDX transgenic Arabidopsis plant were watered with 150 mM NaCl. After 2 weeks, approximately 80% of WT (n = 50) and 35S:GmbZIP15-SRDX plants (n = 50) remained viable, while nearly 90% of OX-GmbZIP15 plants (n = 50) died ( Figure S5A). These results indicate that, similar to OX-GmbZIP15 soybean plants, OX-GmbZIP15 Arabidopsis plants were also sensitive to salt stress. Similarly, following a 2-week dehydration treatment of WT, OX-GmbZIP15, and 35S:GmbZIP15-SRDX transgenic Arabidopsis plants, approximately 85% of WT (n = 50) and 35S:GmbZIP15-SRDX (n = 50) plants showed some degree of wilting phenotype but remained alive, whereas approximately 90% of the OX-GmbZIP15 plants (n = 50) displayed a severe, near lethal wilting phenotype ( Figure S5B). These results suggest that OX-GmbZIP15 overexpression causes similar drought-and salt-stress sensitivity in both Arabidopsis and soybean.
In addition, seed germination efficiency of WT, OX-GmbZIP15, and 35S:GmbZIP15-SRDX Arabidopsis lines were evaluated under control and drought-and salt-stress conditions. For this, seeds of each line were germinated on 1/2 Murashige and Skoog Medium (MS) with or without 150 mM NaCl or 300 mM mannitol; the growth of two OX-GmbZIP15 lines were inhibited severely ( Figure 4A). In addition, the cotyledon greening rate was much lower in OX-GmbZIP15 compared to WT and 35S:GmbZIP15-SRDX transgenic Arabidopsis plants ( Figure 4C). Taken together, these results indicate that GmbZIP15 plays a conserved role in drought-and salt-stress responses in both soybean and Arabidopsis.
To further understand the causal factors behind the drought-and salt-stress hypersensitivity of GmbZIP15-overexpressing Arabidopsis plants, we assayed expression levels of several known abiotic-stress-responsive genes in WT and OX-GmbZIP15 transgenic Arabidopsis plants under control and drought-and salt-stress conditions. Transcript levels of each of the analyzed genes, including AtWRKY33, AtCOR6-6, AtDREB2A, and AtRD29A, were suppressed in OX-GmbZIP15 plants compared with WT plants under normal conditions ( Figure 4D,E). Under salt stress, the expression levels of AtCOR6-6, AtDREB2A, and AtRD29A were increased in both WT and OX-GmbZIP15 plants following salt-stress treatment, although the magnitude of expression induction was much lower in OX-GmbZIP15 plants ( Figure 4D). Similar patterns of repressed expression of abiotic stress-responsive genes were detected in OX-GmbZIP15 plants following drought-stress treatment. The expression levels of AtWRKY33, AtDREB2A, and AtRD29A were induced in both WT and OX-GmbZIP15 plants by drought stress, but to a smaller extent in OX-GmbZIP15 plants ( Figure 4E). These results indicate that the drought-and salt-stress hypersensitivity caused by GmbZIP15 overexpression in Arabidopsis may be due to the repressed expression of drought-and salt-responsive genes.
AtbZIP60 is the homologue of GmbZIP15 in Arabidopsis. To investigate the role of AtbZIP60 in response to drought and salt stress, we analyzed the growth of the bzip60 mutant (SALK_050203C) under control and drought-and salt-stress conditions. The results showed that the growth of the bzip60 mutant was significantly repressed as compared to that in WT plants after 150 mM NaCl or 300 mM mannitol treatment ( Figure 4A), which was accompanied by a lower cotyledon greening rate ( Figure 4C), suggesting that these mutants are sensitive to drought and salt stress. These results agree with the previous findings that overexpression of AtbZIP60 enhances salt, drought, and cold tolerance in rice [33]. Moreover, these data suggest that the roles of GmbZIP15 and AtbZIP60 in response to abiotic stress have diversified.  , and AtRD29A were quantified by qRT-PCR assays after 150 mM NaCl treatment for 0.12, and 24 h (D). Gene expression levels of AtWRKY33, AtDREB2A, and AtRD29A were quantified by qRT-PCR assays after 300 mM mannitol treatment for 0.12, and 24 h (E). Errors bars indicate ± SD of three biological replicates. Significant differences between samples labeled a, b, and c were determined by one-way ANOVA, p < 0.05.

Transcriptomic Analysis of OX-GmbZIP15 Transgenic Soybean Plants in Response to Salt and Drought Stress
To further reveal the molecular mechanism behind the abiotic-stress sensitivity caused by GmbZIP15 overexpression, we conducted RNA sequencing (RNA-seq) analysis using OX-GmbZIP15-16 (OE) and WT soybean plants grown under either control conditions (mock treated) or treated with NaCl or mannitol. Three biological replicates were collected for each sample. In OE and WT plants, 2229 and 1693 differentially expressed genes (DEGs) were detected respectively, under salt-stress conditions compared to control conditions (fold change: ≥ 2 and p ≥ 0.05) ( Figure 5A). Moreover, 8917 and 4811 DEGs were detected in OE and WT plants, respectively, under drought-stress conditions compared to control conditions ( Figure 5A, Table S1). Thus, there were more DEGs induced by both salt and drought stress in OE plants than in WT plants, indicating that GmbZIP15 is responsible for gene expression changes upon salt and drought stress. In addition, we detected 1361 common DEGs (695 upregulated and 546 downregulated) in OE plants upon salt and drought stress ( Figure 5B). In order to characterize the DEGs downregulated in the OE line upon salt and drought stresses, we studied their gene annotation (GO) term enrichment compared to that for untreated OE plants. As shown in Figure 5, a number of metabolic processes were enriched in both salt-stress and drought-stress downregulated gene sets in OE plants, such as response to stimuli, oxidation-reduction reactions, photosynthesis, hydrolase activity, phenylalanine biosynthesis, and some secondary metabolism processes ( Figure 5C-F). These results imply that the above metabolic pathways are repressed in GmbZIP15-overexpressing soybean plants under abiotic stress.

GmbZIP15 Regulates the Expression of GmSAHH1, GmABF1, and GmWRKY12 in Soybean in Response to Abiotic Stress
On the basis of the RNA-seq data, five genes (FPKM > 100) with higher expression levels in OE plants compared with WT plants under normal conditions were selected for further analysis. However, only three of these genes (GmSAHH1, GmWRKY12, and GmABF1) were cloned successfully. GmSAHH1 encodes a phosphate dehydrogenase, and expression of its Arabidopsis homologue (ATSAHH1; At4g13940) is detected in developing seeds and some anthers [34]. Moreover, the abundance of ATSAHH1 is reduced in protein extracts from salt-treated cells [35].

GmbZIP15 Regulates the Expression of GmSAHH1, GmABF1, and GmWRKY12 in Soybean in Response to Abiotic Stress
On the basis of the RNA-seq data, five genes (FPKM > 100) with higher expression levels in OE plants compared with WT plants under normal conditions were selected for further analysis. However, only three of these genes (GmSAHH1, GmWRKY12, and GmABF1) were cloned successfully. GmSAHH1 encodes a phosphate dehydrogenase, and expression of its Arabidopsis homologue (ATSAHH1; At4g13940) is detected in developing seeds and some anthers [34]. Moreover, the abundance of ATSAHH1 is reduced in protein extracts from salt-treated cells [35].
We performed a further qRT-PCR analysis to validate the RNA-seq data. Results consistently showed that the expression of GmSAHH1 was higher in OE than in WT soybean plants under normal conditions, but lower in OE than in WT plants under salt-and drought-stress conditions ( Figure S6A,B). This repressed expression of GmSAHH1 under abiotic stress was similar to that of GmbZIP15. To further investigate the biological function of GmSAHH1 expression changes under abiotic stress, GmSAHH1-overexpressing (OX-GmSAHH1) transgenic Arabidopsis plants were obtained, and two lines with higher expression levels were selected for further research ( Figure 6A) and then subjected to salt-and drought-stress treatments. Before treatment, no obvious morphological differences between 5-week-old WT and OX-GmSAHH1 plants were observed. By contrast, under salt and drought stress, OX-GmSAHH1 transgenic plants exhibited much more pronounced wilting compared with WT plants, which was almost lethal ( Figure S5A,B). Seedling growth was also significantly inhibited in OX-GmSAHH1 plants upon salt and drought stress ( Figure 6B). The similar salt-and drought-stress hypersensitivity of both GmbZIP15and GmSAHH1-overexpressing plants suggests that GmbZIP15-regulated responses to salt and drought stress are likely mediated by GmSAHH1 expression activation.
Previous work found that GmWRKY12 positively regulates drought-and salt-stress responses in association with ABA and salicylic acid (SA), and GmWRKY12 overexpression in soybean roots enhances soybean salt and drought tolerance [36]. In our RNA-seq data, GmWRKY12 exhibited higher expression in OE plants than in WT plants under normal conditions and displayed increased expression in OE plants under stress conditions than normal conditions. We conducted a further qRT-PCR analysis and found that the expression of GmWRKY12 was induced by salt and drought treatment in WT and OE plants ( Figure S6A,B). Moreover, we cloned an ABA-responsive gene (GmABF1) based on RNA-seq data, and its homologue in Arabidopsis (AtABF1, AT1G49720) is an ABA-dependent TF that regulates the expression of downstream ABA-inducible genes to improve plant drought resistance [10,37]. Our result demonstrated that its expression is induced by salt and drought stress ( Figure S6A,B). Furthermore, GmWRKY12-overexpressing (OX-GmWRKY12) and GmABF1-overexpressing (OX-GmABF1) seedlings and plants showed improved salt-and drought-stress tolerance compared to WT plants ( Figure 6B and Figure S6A,B), which agreed with the induced expression of GmWRKY12 and GmABF1 under salt and drought stress. These results suggest that GmbZIP15 regulates plant salt-and drought-stress responses partly through inhibiting the expression of GmWRKY12 and GmABF1 upon abiotic stress.
contrast, under salt and drought stress, OX-GmSAHH1 transgenic plants exhibited much more pronounced wilting compared with WT plants, which was almost lethal (Figure S5A,B). Seedling growth was also significantly inhibited in OX-GmSAHH1 plants upon salt and drought stress ( Figure 6B). The similar salt-and drought-stress hypersensitivity of both GmbZIP15-and GmSAHH1-overexpressing plants suggests that GmbZIP15-regulated responses to salt and drought stress are likely mediated by GmSAHH1 expression activation. Previous work found that GmWRKY12 positively regulates drought-and salt-stress responses in association with ABA and salicylic acid (SA), and GmWRKY12 overexpression in soybean roots enhances soybean salt and drought tolerance [36]. In our RNA-seq data, GmWRKY12 exhibited higher expression in OE plants than in WT plants under normal conditions and displayed increased expression in OE plants under stress conditions than normal conditions. We conducted a further qRT-PCR analysis and found that the expression of GmWRKY12 was induced by salt and drought treatment in WT and OE plants ( Figure S6A,B). Moreover, we cloned an ABA-responsive gene (GmABF1) based on RNA-seq data, and its homologue in Arabidopsis (AtABF1, AT1G49720) is an ABA-dependent TF that regulates the expression of downstream ABA-inducible genes to improve plant drought resistance [10,37]. Our result demonstrated that its expression is induced by salt and drought stress ( Figure S6A,B). Furthermore, GmWRKY12-overexpressing (OX-GmWRKY12) and GmABF1-overexpressing (OX-GmABF1) seedlings and plants showed improved salt-and drought-stress tolerance compared to WT plants ( Figures 6B and S6A,B), which agreed with the induced expression of GmWRKY12 and GmABF1 under salt and drought stress. These results

Discussion
Abiotic stress, including salt and drought stress, has a considerable impact on the quality and yield of agricultural products. The important oilseed crop, soybean, is threatened by diverse categories of abiotic stress. Previous studies have indicated that bZIP TFs play diverse roles in response to various biotic and abiotic stress factors in different crop species, such as rice, soybean, rape, cotton, and maize [38][39][40][41]. In this study, a group-K bZIP TF, namely, GmbZIP15, was identified in soybean, and its functions in response to abiotic-stress conditions were analyzed in detail.
AtbZIP60, the Arabidopsis GmbZIP15 homologue, positively modulates plant responses to salt, cold, and abiotic-stress conditions [33]. Here, we found that GmbZIP15 acts as a negative regulator of abiotic-stress responses. The transcription of GmbZIP15 was suppressed by salt and drought stress ( Figure S1A,B) and GmbZIP15-overexpressing soybean displayed hypersensitivity to salt and drought stress (Figures 1A and 2A). Our study suggests that GmbZIP15 function in abiotic-stress responses differs from that of AtbZIP60, which is possibly due to the functional divergence of soybean and Arabidopsis during long-term evolution.
To adapt to abiotic stress, especially salt and drought stresses, the plants derive several strategies, including ion regulation and compartmentalization, induction of antioxidant enzymes, plant hormones and regulatory genes [42][43][44]. For example, the novel soybean regulatory gene GmTIP2;3 could effectively improve the tolerance of yeast to drought stress [43,45]. In addition, when under abiotic-stress conditions, plant endogenous ABA accumulates rapidly and activates the expression of stress-responsive genes, causing many physiological responses [37,46]. It has been demonstrated that ABA plays key roles in maintaining seed dormancy, inhibiting germination, and preventing seedling growth [47], and that abiotic stress is able to induce ABA biosynthesis and trigger ABA-dependent signaling pathways [48]. GmbZIP15 negatively regulates the expression of GmABF1, and the overexpression of GmABF1 in Arabidopsis confers increased resistance to salt-and drought-stress conditions ( Figure 6B). In addition, GmWRKY12 in association with ABA positively regulates drought-and salt-stress responses, and the overexpression of GmWRKY12 in soybean roots enhances plant salt and drought tolerance [36]. These results indicate a negative regulation of GmbZIP15 for ABA signaling which might via GmABF1 and GmWRKY12 in response to abiotic stress. In addition, the stress-responsive genes AtDREB2A and AtRD29A in OX-GmbZIP15 Arabidopsis plants ( Figure 4C,D) and GmDREBb ( Figure S2A,B) in OX-GmbZIP15 soybean plants exhibited lower transcript levels than those in WT plants under salt or drought conditions. AREBs and DREB are two groups of TFs that independently regulate the expression of genes involved in ABA-dependent and ABA-independent pathways [11]. The promoter regions of RD29 genes (RD29A and RD29B in Arabidopsis) are targeted by AREBs and DREBs; these genes encode hydrophilic proteins that endow plants with enhanced resistance to abiotic and cold stress [49]. Therefore, our study suggests that GmbZIP15 might act a negative regulator of plant drought-and salt-stress responses through ABA-dependent and ABA-independent pathways.
Abiotic stress potentially impairs plant cellular physiology and biochemistry via the excess generation of ROS [7,50,51]. For example, SlWRKY81 improved drought tolerance in tomato plants via the repression of SlP5CS1 transcription and thus reducing proline biosynthesis [52]. In this study, H 2 O 2 contents sharply increased in OX-GmbZIP15 plants compared to WT plants under droughtand salt-stress conditions ( Figures 1B and 2B). To control the level of ROS accumulation under stress conditions, plants have evolved a number of antioxidants, such as SOD, POD, and CAT, to scavenge ROS and to restore cellular redox homeostasis [53][54][55][56][57]. With the development of molecular biology, our understanding of molecular and physiology mechanisms is becoming clearer. Our results showed that the activities of POD and CAT were suppressed in OX-GmbZIP15 transgenic soybean plants under salt and drought stress ( Figure S2C,D), indicating compromised ROS scavenging capability in OX-GmbZIP15 plants in comparison with WT plants. Therefore, we hypothesize that GmbZIP15 plays a negative role in regulating these ROS-scavenging enzyme systems under abiotic stress.
Previous studies showed that bZIP TFs function in many biotic and abiotic-stress responses in plants through regulating diverse biochemical and physiological pathways [12,23,39,58]. RNA-sequencing has been widely used to investigate the molecular processes related to adaptive responses to abiotic stresses and to identify stress-resistance candidate genes by analyzing differences in transcript abundance [44]. In our research, we found that the functional annotation of DEGs that were enriched in the set of downregulated genes in OE plants after salt and drought treatment compared to that in control conditions indicated that, under salt or drought-stress conditions, GmbZIP15-regulated genes were mainly involved in processes such as oxidoreductase activity, phenylalanine biosynthesis, phosphotransferase activity, and some secondary metabolism ( Figure 5). As an important polyphenolic secondary metabolite, isoflavones play a crucial role in plants facing diverse environmental-stress conditions [59][60][61]. PtSAP13, for example, enhances salt tolerance by upregulating the transcript level of stress-responsive genes and inducing multiple biological pathways, such as phenylalanine biosynthesis and dioxygenase activity [62], thus implying that the phenylalanine metabolism process is involved in GmbZIP15-regulated abiotic-stress responses ( Figure 5C,E). In addition, photosynthesis is essential for plant growth and is important for plants to maintain a balance between growth and stress responses [63,64]. For example, when cyanobacteria grow under stress conditions, photosynthesis-related genes are usually downregulated, whereas stress response-related genes are upregulated [65]. Water deficiency significantly affects photosynthetic characteristics. The drought-tolerant soybean cultivar displayed the maximum values of chlorophyll fluorescence (Fv/Fm, qP, φPSII, and ETR) [66]. When under environmental stress, plants will close their stomata and thus restrict the entry of CO 2 into the leaf and reduce the rate of photosynthesis [67]. In this study, many downregulated genes in OX-GmbZIP15 transgenic plants after drought and salt stress were associated with photosynthesis ( Figure 5C-F), suggesting that stress adaption was prioritized over photosynthesis; however, impaired stomatal aperture regulation in OX-GmbZIP15 affected plant survival. In addition, enzyme-catalyzed removal of ROS such as superoxide and H 2 O 2 are important in plant survival under stress conditions [54,63]. As observed here, those genes downregulated in response to abiotic stress represented antioxidant-related processes. Taken together, multiple metabolic pathways seem to be involved in the GmbZIP15-mediated abiotic-stress response network.
In summary, overexpression of GmbZIP15 in soybean resulted in hypersensitivity to salt and drought stresses compared with wild-type (WT) plants, which was associated with lower transcript levels of stress-responsive genes, defective stomatal aperture regulation, and reduced antioxidant enzyme activities. Furthermore, RNA-seq and qRT-PCR analyses revealed that GmbZIP15 positively regulates GmSAHH1 expression and negatively regulates GmWRKY12 and GmABF1 expression in response to salt and drought stresses (Figure 7). These data provided new information for understanding the function of GmbZIP15 and might facilitate the improvement of plant abiotic-stress tolerance through genetic manipulation in the future. deficiency significantly affects photosynthetic characteristics. The drought-tolerant soybean cultivar displayed the maximum values of chlorophyll fluorescence (Fv/Fm, qP, ϕPSII, and ETR) [66]. When under environmental stress, plants will close their stomata and thus restrict the entry of CO2 into the leaf and reduce the rate of photosynthesis [67]. In this study, many downregulated genes in OX-GmbZIP15 transgenic plants after drought and salt stress were associated with photosynthesis ( Figure 5C-F), suggesting that stress adaption was prioritized over photosynthesis; however, impaired stomatal aperture regulation in OX-GmbZIP15 affected plant survival. In addition, enzyme-catalyzed removal of ROS such as superoxide and H2O2 are important in plant survival under stress conditions [54,63]. As observed here, those genes downregulated in response to abiotic stress represented antioxidant-related processes. Taken together, multiple metabolic pathways seem to be involved in the GmbZIP15-mediated abiotic-stress response network.
In summary, overexpression of GmbZIP15 in soybean resulted in hypersensitivity to salt and drought stresses compared with wild-type (WT) plants, which was associated with lower transcript levels of stress-responsive genes, defective stomatal aperture regulation, and reduced antioxidant enzyme activities. Furthermore, RNA-seq and qRT-PCR analyses revealed that GmbZIP15 positively regulates GmSAHH1 expression and negatively regulates GmWRKY12 and GmABF1 expression in response to salt and drought stresses (Figure 7). These data provided new information for understanding the function of GmbZIP15 and might facilitate the improvement of plant abiotic-stress tolerance through genetic manipulation in the future.  A schematic model of GmbZIP15 mediated abiotic-stress tolerance in soybean. GmbZIP15 negatively modulates the abiotic-stress tolerance: GmbZIP15 positively regulates the expression of GmSAHH1 and negatively regulates the expression of GmWRKY12 and GmABF1 in response to abiotic stresses. The arrows indicate induction or positive modulation; the blunt-end arrows represent block or suppression.

Vector Construction and Transformation
To generate the OX-GmbZIP15 construct, the GmbZIP15 (Glyma.02G161100) coding DNA sequence (CDS) was amplified and the PCR fragments were cloned into the pENTR/D-TOPO vector (Invitrogen, Carlsbad, CA, USA). The pENTR clones were recombined into the destination vector pGWB506 using LR Clonase II (Invitrogen). The resulting construct also contained the selectable marker BAR for glufosinate resistance [68].
35S: GmbZIP15-SRDX was generated by amplifying GmbZIP15 cDNA sequence and an SRDX motif was added to the end of the cDNA sequence (ctagatctggatctagaactccgtttgggtttcgcttaa). The PCR fragment was cloned into the pENTR/D-TOPO vector (Invitrogen), and the pENTR/D-TOPO clones were recombined into the destination vector pGWB506 using LR Clonase II (Invitrogen) [69]. The vectors OX-GmbZIP15 and 35S: GmbZIP15-SRDX were then transformed into soybean by agrobacterium-mediated method [70] and the soybean genotype C03-3 was used.

Plant Materials and Stress Treatments
Soybean plant seeds including WT (C03-3) and transgenic GmbZIP15 plants were grown for 15 days in pots containing nutritional soil and vermiculite in green house. The seedlings were then exposed to drought and salt stresses. For drought stress, the soybean seedlings were watered with 300 mM mannitol to induce the rapid drought stress. For salt treatment, the seedlings were transferred to 200 mM NaCl solution. All seedlings leaves were harvested at 0, 6, 12, and 24 h under stress conditions for RNA extraction.
Arabidopsis ecotypes Col-0 was used in this study. The T-DNA mutant Atbzip60 (SALK_050203C) was obtained from the Arabidopsis Biological Resource Centre (ABRC). All Seeds were germinated on 1/2 MS medium containing NaCl or mannitol, after vernalization at 4 • C for 3 days, the plates containing the seeds were placed in a growth chamber with temperature 22 • C, and a photoperiod of 16 h light/8 h dark.

Diaminobenzidine (DAB) Staining
Following previously described methods for hydrogen peroxide (H 2 O 2 ) detection [72], the soybean leaves after salt and drought treatment for 4 days were immediately vacuum-infiltrated for 20 min with Tris-HCl (pH 7.4) containing 1% (w/v) DAB. Thereafter, all the leaves were placed in light for 10 h then boiled for 20 min in 75% ethanol.

Determination of Stomatal Aperture
The fully expanded leaves of 2-week-old soybean plants were floated in the stomatal opening buffer with 30 mM KCl and 10 mM MES-KOH, pH 6.15 for 2 h under a cool white light, and then 200 mM NaCl or 300 mM mannitol were added to the opening buffer [22]. After 1 h, the subepidermal peels were stripped and used for stomatal aperture measurements under the microscope. In addition, different phytohormones were added to the opening buffer and the stomatal apertures were observed at different timepoints.

RNA Extraction and Quantitative qRT-PCR
Total RNA was extracted using Trizol (Invitrogen, Carlsbad, CA, USA) then reverse-transcribed using the PrimeScript RT-PCR kit (TaKaRa) [71]. The relative expression levels of selected genes were detected by qRT-PCR using Bio-Rad QRT-PCR system (Foster City, CA, USA) and SYBR Premix Ex Taq II (TaKaRa Perfect Real Time). The qRT-PCR program was 95 • C for 30 s; 40 cycles of 95 • C for 5 s and 60 • C for 34 s; and 95 • C for 15 s [68]. GmActin was used for normalization.

Determination of Antioxidant Enzyme Activity
The antioxidants including peroxidase (POD) and catalase (CAT) were extracted from approximately 0.1 g of soybean leaves using 1 mL extraction solution. The 2-week-old seedlings were treated with 200 mM NaCl or 300 mM mannitol for 24 h and then the leaves were harvested. The enzyme activities were measured according to the protocol from Solarbio Biochemical Assay Division.

RNA-Seq Data Analysis
Leaves of 2-week-old soybean plants including WT and OX-GmbZIP15-16 plants treated with 200 mM NaCl or 300 mM mannitol were harvested at 24 h for RNA-seq, and three biological replicates were analyzed. The libraries were constructed by BGI (Beijing Genomics Institute) then sequenced.

Conflicts of Interest:
The authors declare no conflict of interest.