WRKY Genes Improve Drought Tolerance in Arachis duranensis

WRKY transcription factor participates in plant growth and development and response to biotic and abiotic stresses. Arachis duranensis, a turfgrass, has high drought tolerance, yet little is known about AdWRKYs response to drought stress in A. duranensis. In this study, RNA-seq identified five AdWRKYs, including AdWRKY18, AdWRKY40, AdWRKY42, AdWRKY56, and AdWRKY64, which were upregulated under drought stress. Orthologous relationships between AdWRKYs and Arabidopsis WRKY were determined to predict the regulatory networks of the five AdWRKYs based on AtWRKYs. Additionally, protein–protein interactions were predicted using differentially expressed proteins from RNA-seq. The quantitative real-time PCR (qRT-PCR) results showed that AdWRKY40 was upregulated, while AdWRKY42, AdWRKY56, and AdWRKY64 were downregulated at different time-points under drought stress. The predicted regulatory networks showed that AdWRKY40 activates COR47, RD21, and RD29A expression under drought stress. Besides, AdWRKY56 regulated CesA8 under drought stress. Aradu.YIQ80 (NAC019) interacted with AdWRKY40, AdWRKY42, AdWRKY56, and AdWRKY64, while Aradu.Z5H58 (NAC055) interacted with AdWRKY42 and AdWRKY64 under drought stress. This study used Arabidopsis to assess AdWRKYs function and regulatory networks, providing a basis for understanding drought tolerance in A. duranensis.

Arachis duranensis, a turfgrass, has strong drought tolerance (Leal-Bertioli et al., 2012. The genome sequences of A. duranensis and drought-related RNA-seq datasets are available in the PeanutBase database (Brasileiro et al., 2015;Bertioli et al., 2016;Dash et al., 2016). A previous study identified 75 WRKYs (AdWRKYs) in A. duranensis (Song et al., 2016b), providing a basis for identifying AdWRKYs involved in drought stress response. In this RNA-seq based study, five AdWRKYs (AdWRKY18, AdWRKY40, AdWRKY42, AdWRKY56, and AdWRKY64) were differentially expressed under drought stress. The regulatory networks of AdWRKYs were then determined and verified. This work provides a theoretical basis for further analysis of the function of AdWRKYs.

Identification of AdWRKYs Involved in Drought Stress Response
The transcriptomes of A. duranensis under drought stress and normal growth conditions were sequenced and de novo assembled to detect differentially expressed genes (DEGs) under normal growth and drought stress conditions in 2015 (Brasileiro et al., 2015). In 2016, the RNA-seq was re-assembled using the A. duranensis genome as the reference, and the updated RNA-seq data were released in the PeanutBase database (Dash et al., 2016). The DEGs were identified between drought and control using the edgeR program (Robinson et al., 2010). Genes with log 2 (Foldchange) > 2 or <−2 at FDR < 0.05 were considered differentially expressed (Brasileiro et al., 2015;Dash et al., 2016).
A previous study identified WRKYs in A. duranensis (Song et al., 2016b). This study extracted the differentially expressed AdWRKY genes in the abovementioned RNA-seq datasets. The differentially expressed AdWRKY features and subcellular localization were predicted using ExPASy (Gasteiger et al., 2003) and Plant-mPLoc (Chou and Shen, 2010) with default parameters.

Phylogenetic Relationship Analysis
Phylogenetic trees were constructed using AdWRKYs and Arabidopsis WRKY (AtWRKY) to reveal their orthologous relationship. AtWRKYs were obtained from a public database (The Arabidopsis Information Resource, https://www.arabidopsis. org/index.jsp). Multiple sequence alignments were conducted using the MAFFT program (Katoh and Standley, 2013). The ProtTest program was used to estimate the best-fit model for constructing phylogenetic trees (Darriba et al., 2011). The maximum likelihood (ML) of the trees was determined using the IQ-tree program (Nguyen et al., 2015).

Cis-Acting Element Analysis
WRKY regulates downstream gene expression by binding to cis-acting elements, such as W-box (C/TTGACC/T), WT-box (GACTTT), WK-box (TTTTCCAC), PRE (TACTGCGCTTAGT), and SURE (TAAAGA TTACTAATAGGAA; Rinerson et al., 2015;Chen et al., 2019). Herein, the TBtools program was used to extract the 2-kb sequences upstream of the start codon of the predicted genes to identify AdWRKYs genes involved in regulation (Chen et al., 2020). PlantCARE (Rombauts et al., 1999) was used to predict the binding sites of the WRKY transcription factor.

Protein Interaction Analysis
WRKYs interact with other proteins involved in plant development and stress response (Hu et al., 2013;Chen et al., 2017b;Zhang et al., 2018). The protein interaction of AdWRKYs and their differentially expressed genes were predicted using the STRING public database 1 with Arabidopsis protein sequences as the reference.

Quantitative Real-Time PCR Analyses
Quantitative real-time PCR (qRT-PCR) analyses were used to verify the drought-tolerance function of the abovementioned genes. Briefly, the A. duranensis seeds were sterilized and germinated on wet filter paper at 28°C. The seedlings were then transferred to the Hoagland solution. Four-leaf plants were treated with 10% (w/v) PEG6000. The leaves were collected after 0, 6, 12, 24, 36, and 48 h of treatments. The control was sampled from 0 h. Three biological replicates were used.
Plant RNA Extraction Kit (TaKaRa, Dalian, China) was used to extract total RNA. The RNA (1 μg) was used for cDNAs synthesis via Reverse Transcriptase M-MLV System (TaKaRa, Dalian, China). The primers were designed by Beacon Designer 8 (Supplementary Table S1). The primers were specifically for amplification since WRKY sequences are conserved (Zhang et al., 2020a). qRT-PCR was performed using TB green premix ex Taq II (TaKaRa, Dalian, China) on the CFX96 real-time PCR machine (Bio-Rad, CA, United States) with UBI2 as the reference gene (Morgante et al., 2011). The PCR conditions included: 95°C denaturation for 30 s, followed by 40 cycles at 95°C for 5 s and 60°C for 45 s. A melting curve analysis was performed at the end of the PCR running end. The 2 -ΔΔCt method was used for quantification (Livak and Schmittgen, 2001).
The ML trees were constructed using WRKY domains and WRKY full-length proteins to reveal the orthologous relationships between AdWRKYs and AtWRKYs. The best-fit models were JTT + I + G and VT + I + G + F. The phylogenetic tree showed that AdWRKY42 had homology with AtWRKY40 ( Figure 1B;  Supplementary Figures S1, S2). The other four AdWRKYs had different topological structures between the WRKY domain tree and the WRKY full-length protein tree ( Figure 1B;  Supplementary Figures S1, S2). Altogether, the following homologous relationships were obtained based on two ML 1 https://string-db.org trees: AdWRKY18 with AtWRKY45 and AtWRKY75; AdWRKY40 with AtWRKY8, AtWRKY28, and AtWRKY71; AdWRKY42 with AtWRKY40; AdWRKY56 with AtWRKY33 and AtWRKY25; and AdWRKY64 with AdWRKY42, AtWRKY40, AtWRKY60, and AtWRKY18.
Experiments have shown the functions of many AtWRKYs. AtWRKY45 overexpression alleviates phosphate starvation and promotes leaf senescence in Arabidopsis (Wang et al., 2014;Chen et al., 2017b). Overexpression of AtWRKY75 promotes leaf senescence and flowering in Arabidopsis (Guo et al., 2017;Zhang et al., 2018). A recent study showed that PtrWRKY75, which is orthologous with AtWRKY75, improves drought tolerance in polar (Zhang et al., 2020b), indicating that AdWRKY18 may confer drought tolerance traits.
Overexpression of AtWRKY8 improves salt stress (Hu et al., 2013). Besides, AtWRKY28 and AtbHLH17 overexpression enhance drought and salt tolerance in Arabidopsis (Babitha et al., 2013). These results indicate that AdWRKY40 is tolerant to drought and salt stresses.
Therefore, AdWRKY18, AdWRKY40, AdWRKY42, AdWRKY56, and AdWRKY64 have potential functions in drought stress response based on the abovementioned homologous relationships.
AtCesA8, a cellulose synthase catalytic subunit, plays a crucial role in cellulose synthesis in the secondary cell wall (Taylor et al., 2000(Taylor et al., , 2003. AtCesA8 mutant accumulates high ABA content, thus reducing the expression of stress-related genes (Chen et al., 2005). AtCesA8 negatively regulates drought stress (Chen et al., 2005;Wang et al., 2013). AtWRKY33 decreases gene expression by binding to CesA8 W-box element, thus increasing drought tolerance . Herein, AdWRKY56 had homology with AtWRKY33, indicating that AdWRKY56 can enhance drought tolerance by controlling CesA8. Therefore, these results suggest that AdWRKY18, AdWRKY40, and AdWRKY56 improve drought tolerance by regulating the expression of downstream genes. WRKYs control downstream genes by binding to W-box, WT-box, WK-box, PRE, and SUR cis-acting elements (Sun et al., 2003;van Verk et al., 2008;Hu et al., 2013;Xiao et al., 2013;Machens et al., 2014;Rinerson et al., 2015;Chen et al., 2019   Subcellular localization showed that AtPAL1 and Aradu. NNP8F are located in the cytoplasm, AtRD21 and Aradu.08WSJ are located in the vacuole, AtCesA8 and Aradu.UPY7V are located in the chloroplast, and others are located in the nucleus (Supplementary Table S2). A previous study revealed that AtWRKY40 moves from the nucleus to the cytoplasm to control downstream genes (Rushton et al., 2012). Similarly, AdWRKYs can potentially move from the nucleus to other organelles to regulate Aradu.NNP8F, Aradu.08WSJ, and Aradu.UPY7V is located outside the nucleus.
The five AdWRKYs interacted with common proteins and specific proteins (Figure 3B) Table S4). AtNAC019 improves drought stress in Arabidopsis by activating ERD1 expression (Tran et al., 2004). AtWRKY1 enhances drought stress or ABA treatment by binding to the W-box cis-acting element of AtMYB2 (Qiao et al., 2016). AtNAC055 overexpression increases drought tolerance, and the AtNAC055 mutant has a decreased drought tolerance (Fu et al., 2018). Drought stress induces AtABR1 expression. However, mannitol stress can decrease AtABR1 expression, thus, reducing seed germination (Pandey et al., 2005).

Verification of Drought Tolerance of AdWRKYs and Their Regulatory Genes Using qRT-PCR
This study identified five differentially expressed AdWRKYs using RNA-seq and their regulatory networks based on Arabidopsis. qRT-PCR was used to assess the drought stress response of the five AdWRKYs and their regulatory genes. AdWRKY42, AdWRKY56, and AdWRKY64 genes were downregulated at 6, 12, 24, 36, and 48 h (Figure 5). AdWRKY40 was upregulated at 6, 12, and 36 h, and downregulated at 48 h (Figure 5). AdWRKY18 was downregulated at 24 and 48 h and upregulated at 36 h ( Figure 5). The five AdWRKYs were downregulated at 48 h. AdWRKY40 had similar qRT-PCR and RNA-seq results. The differential expression of AdWRKY42, AdWRKY56, and AdWRKY64 contradicted the RNA-seq results. The possible reason is different drought treatment and cultural environments between qRT-PCR and RNA-seq. The RNA-seq data were produced from the samples under natural drought and normal growth conditions (Brasileiro et al., 2015), while qRT-PCR was analyzed under five drought-stress time points using 10% (w/v) PEG6000 treatments.
Quantitative real-time PCR showed that AdCOR47 and AdCesA8 genes were upregulated and downregulated, respectively, at five time-points (Figure 6). Similarly, previous studies showed that COR47 is positively regulated in Oryza sativa under drought stress, while CesA8 is negatively regulated in A. thaliana under drought stress (Song et al., 2010;Wang et al., 2013). Herein, AdPAL1 and AdRD29A were downregulated at 6, 12, 24, and 36 h (Figure 6). However, PtrPAL1 and AtRD29A are positively regulated under drought stress (Hu et al., 2013;Zhang et al., 2020b). WRKYs directly regulate PtrPAL1 and RD29A by binding to W-box elements. Although AdPAL1 and AdRD29A do not have W-box elements, they were upregulated at 48 h (Figure 6). AdRD21 also lacked the W-box element and was upregulated at five time-points. Similarly, a previous study showed that AtRD21 is positively regulated under drought stress (Narusaka et al., 2003;Hua et al., 2006;Song et al., 2010). Therefore, AdWRKY40 is involved in regulating different regulatory networks of downstream genes AdRD21 and AdRD29A.
Quantitative real-time PCR was used to assess the drought stress response of the genes that translated proteins-AdWRKYs interaction. This study found that AdNAC019 and AdNAC055 were differentially expressed after drought stress. However, other genes were not differentially expressed relative to the control group. AdNAC019 interacted with the five AdWRKYs, while AdNAC055 interacted with only AdWRKY42 and AdWRKY64. AdNAC019 was downregulated at the five time-points (Figure 6). AdNAC055 was upregulated at 6 and 36 h and downregulated at 12, 24, and 48 h (Figure 7). However, NAC019 and NAC055 positively regulate drought stress in Arabidopsis (Tran et al., 2004;Fu et al., 2018). Similarly, AdWRKY42, AdWRKY56, and AdWRKY64 were downregulated after drought stress, different from the expression patterns in Arabidopsis. These results indicate that Arachis and Arabidopsis have different regulatory networks under drought stress.

Opportunities and Challenges in AdWRKYs Study
Herein, five AdWRKYs were identified under drought stress based on previous studies and RNA-seq (Brasileiro et al., 2015;Song et al., 2016b). Subsequently, orthologous relationships between AdWRKYs and AtWRKYs were constructed. The regulatory networks of the five AdWRKYs were determined based on AtWRKYs and verified using qRT-PCR. The results showed that AdWRKY40 positively regulated drought stress, while AdWRKY42, AdWRKY56, and AdWRKY64 negatively regulated drought stress. Moreover, AdWRKY40 was upregulated under drought stress, confirming the orthologous WRKYs from Arabidopsis and Oryza (Narusaka et al., 2003;Hua et al., 2006;Song et al., 2010;Hu et al., 2013). However, the orthologous AdWRKY42, AdWRKY56, and AdWRKY64 showed the opposite results in Arabidopsis under drought stress. Besides, qRT-PCR and RNA-seq results showed opposite results in A. duranensis. Additionally, Arabidopsis had different regulatory networks between AdWRKY40 and its orthologs. AtWRKYs activate COR47, RD21, and RD29A (Song et al., 2010;Hu et al., 2013). AdWRKY40 control COR47 by potential binding to the W-box element. AdWRKY40 indirectly regulates RD21 and RD29A because they lack the W-box element.
Protein-protein analyses are different from transcriptional regulation. To the best of our knowledge, no study has shown that the abovementioned proteins interact with WRKYs in Arabidopsis under drought stress. However, protein-protein interaction has been predicted based on Arabidopsis protein in A. duranensis. This study found that the NAC transcription factor interacts with WRKY involved in the drought stress response of A. duranensis. MaNAC5 interacts with MaWRKY1 and MaWRKY2 to activate PR1-1, PR2, PR10c, and CHIL1 expressions in response to Colletotrichum musae infection in bananas (Shan et al., 2016). Moreover, the CitNAC62-CitWRKY1 interaction enhances CitAco3 expression, thus decreasing citric acid content . We hypothesize that AdWRKY40 and AdNAC interaction can control RD21 and RD29A expression. However, more experimental tests are needed to verify the hypothesis. Therefore, this study provides a model for studying gene function and a basis for revealing WRKY regulatory networks in A. duranensis under drought stress. FIGURE 5 | Differentially expressed AdWRKYs under drought stress detected by quantitative real-time PCR (qRT-PCR). Four-leaf plants were treated with 10% (w/v) PEG6000. The leaves were collected after 0, 6, 12, 24, 36, and 48 h of treatments. The control was sampled from 0 h. Three biological replicates were used. Asterisks * and ** indicate significant differences at 0.05 and 0.01 using t tests, respectively.
FIGURE 6 | Differential expression levels of AdWRKYs regulating downstream genes under drought stress, assessed using quantitative real-time PCR. Four-leaf plants were treated with 10% (w/v) PEG6000. The leaves were collected after 0, 6, 12, 24, 36, and 48 h of treatments. The control was sampled from 0 h. Three biological replicates were used. Asterisks * and ** indicate significant differences at 0.05 and 0.01 using t tests, respectively.

CONCLUSION
This study showed that AdWRKY18, AdWRKY40, AdWRKY42, AdWRKY56, and AdWRKY64 were differentially expressed under drought stress, but various regulatory networks were formed among the five AdWRKYs. AdWRKY18 was excluded from regulatory network analyses because it did not have the same differential expression pattern at two time-points. AdWRKY40 potentially regulates COR47 by binding the W-box element and indirectly regulates RD21 and RD29A under drought stress. AdWRKY56 controlled the CesA8 expression under drought stress (Figure 8). Protein-protein interaction results showed that AdNAC019 interacted with AdWRKY40, AdWRKY42, AdWRKY56, and AdWRKY64 under drought stress AdNAC055 interacted with AdWRKY42 and AdWRKY64 (Figure 8).

DATA AVAILABILITY STATEMENT
Publicly available datasets were analyzed in this study. This data can be found at: https://www.ncbi.nlm.nih.gov/, JZ390113 to JZ390862.

AUTHOR CONTRIBUTIONS
HS conceived and designed this research, analyzed data, and wrote the manuscript. YZ and PD analyzed data.
PD, FX, XZ, and HS evaluated and revised the manuscript. All authors contributed to the article and approved the submitted version.
FIGURE 7 | Differential expression levels of genes that are translated proteins interaction with AdWRKYs under drought stress detected via quantitative real-time PCR. Four-leaf plants were treated with 10% (w/v) PEG6000. The leaves were collected after 0, 6, 12, 24, 36, and 48 h of treatments. The control was sampled from 0 h. Three biological replicates were used. Asterisks * and ** indicate significant differences at 0.05 and 0.01 using t tests, respectively.   Supplementary Table S5 | Specific interaction between proteins and AdWRKYs.
Supplementary Figure S1 | The maximum likelihood tree constructed using WRKY domains.
Supplementary Figure S2 | The maximum likelihood tree constructed using full-length WRKY proteins.
Frontiers in Plant Science | www.frontiersin.org