“Yu882” (Zea mays L., inbred line) was germinated for 5 days, and the seedlings were transferred to half-strength modified Hoagland's nutrient solution with growth conditions at 28°C in a 14/10-h photoperiod. Three-leaf-stage maize were transferred into nutrient solution containing 20% polyethylene glycol 6000 (PEG). Leaves of PEG stress and non-stress were harvested at 60 and 96 h and after 3 days recovery (denoted as T60, T96, TR3d and CK60, CK 96, CK R3d, respectively). The samples at these time points were subjected to RNA sequencing (RNA-Seq). The second experiment, we transferred seedings of three-leaf stage to nutrient solution containing 20% PEG ₆₀₀₀, NaCl (200 mmol/L), ABA (5 μmol/L) and 37°C conditions. The leaves of maize were collected at 0, 4, 12, 24, 36 h and stored at −80°C. Each treatment had three biological replicates.

The plants of maize inbred line Va35 were grown in an artificial climate incubator with 20°C/18°C (photoperiod: 16 h light/8 h dark). Tobacco was grown in an artificial climate chamber at 24°C/22°C (photoperiod: 14 h light /10 h dark).

Total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). A Qubit Fluorometer (Invitrogen) was used to measure RNA concentration, and an Agilent 2100 Bioanalyzer (Agilent, Palo Alto, California, USA) was used to verify RNA quality. The minimum RNA integration value was 8. After total RNA is extracted using TRIzol, RNA molecules with a size of 18–30 nt can be enriched by polyacrylamide gel electrophoresis (PAGE).

Use the Illumina HiSeq™2500 platform to reverse transcription of the enriched mRNA, amplify and sequence by polymerase chain reaction. High-quality clean reads are obtained by removing those containing adaptors and those containing more than 10% of unknown nucleotides (N) and more than 50% of low-quality (Q value ≤ 20) bases. Lastly, ribosome RNA (rRNA) mapped reads were removed through mapping reads to the rRNA database using the short reads alignment tool Bowtie2. The remaining reads are used for transcriptome assembly, and the raw data is compared with the maize inbred line B73V3 database and gene annotation. Use Blast2GO to analyze the pathway of these genes.

To identify differentially expressed genes (DEGs) samples or groups, the R package (http://www.r-project.org/) was used. We identified genes with |log2 (fold change)| ≥ 2 and a false discovery rate (FDR) <0.05 in a comparison as significant differentially expressed genes. Use R language heat map software for differential gene expression enrichment. Calculate the correlation coefficient between genes based on the expression of genes. We imported a correlation coefficient greater than 0.75 into Cytoscape software 3.71 version [ion3.7.1 (https://cytoscape.org)] with to make gene co-expression network to screen key node genes.

Total RNA was extracted using Total RNA Extraction Reagent (YEASEN, Shanghai, China). Hifair® II 1st Strand cDNA Synthesis SuperMix (YEASEN, Shanghai, China) was applied to synthesize first-strand cDNA. The quantitative real-time PCR (qRT-PCR) was carried out using hieff® qPCR SYBR® Green Master Mix (YEASEN, Shanghai, China). According to the corresponding sequences, we designed gene-specific primers using Vector NTI11.5, as shown in Supplementary Table 2. Each gene was analyzed with three technical replicates. The relative expression level (2^−ΔΔCt0h) in the control plants was normalized to 1.

The CDS of ZmbZIP33 was amplified by PCR using a pair of gene-specific primers (Supplementary Table 2), and then ligated to pGBKT7 vector digested by EcoRI/BamHI. PGBKT7-ZmbZIP33 construct, empty pGBKT7 vector (negative control) and pGBKT7-53 plus pGADT7-T (positive control) were transformed into yeast strain AH109 by lithium acetate-mediated method, respectively. The transformed yeast cells were examined on SD/-Trp and SD/-Trp/-His/-Ade/X-a-Gal medium at 30°C for 5 days.

The complete open reading frames (ORF) of ZmbZIP33, ZmSRK2E and ZmPP2C7 were PCR-amplified, and then constructed into pMDC83-GFP vector. The primers are shown in Supplementary Table 2. The constructed ZmbZIP33-GFP, ZmSRK2E-GFP, ZmPP2C7-GFP, and pMDC83-GFP were transferred into Nicotiana benthamiana leaves by Agrobacterium infection. Two days later, the fluorescence of tobacco leaves was observed under a confocal microscope of Zeiss LSM700 (Zeiss, Jena, Germany).

The ORF of ZmbZIP33 and ZmPP2C7, ZmSRK2E and ZmPYL10 genes were cloned into vectors pGBKT7 and pGADT7, respectively. Subsequently, the pGBKT7-ZmPP2C7+pGADT7-ZmPYL10,PGBKT7-ZmPP2C7+pGADT7-ZmSRK2E, pGBKT7-ZmbZIP33+pGADT7-ZmSRK2E and pGADT7-T+ pGBKT7-Lam (negative control), pGADT7-T+pGBKT7-53 (positive control) were introduced into the yeast strain using the high-efficiency lithium acetate transformation procedure (Xiao et al., 2011). The primers are shown in Supplementary Table 2.

The ORF of ZmbZIP33 and ZmPP2C7, ZmSRK2E and ZmPYL10 genes were cloned into vectors of pSAT4-nEYFP and pSAT4-cEYFP, respectively. The pSAT4-nEYFP-ZmPP2C7 +pSAT4-cEYFP-ZmPYL10,pSAT4-nEYFP-ZmPP2C7+pSAT4-cEYFP-ZmSRK2E, pSAT4-nEYFP-ZmbZIP33+pSAT4-cEYFP-ZmSRK2E, pSAT4-cEYFP-ZmPYL10+nEYFP, pSAT4-nEYFP-ZmPP2C7+cEYFP,pSAT4-cEYFP-ZmSRK2E+nEYFP and pSAT4-nEYFP-ZmbZIP33+cEYFP were then introduced into Nicotiana benthamiana leaves by Agrobacterium infection, respectively. Two days later, SP5 Meta confocal laser microscope (Leica, Germany) was used to observe the fluorescence signal, and yellow fluorescence signal represents an interaction between the two proteins. The primers are shown in Supplementary Table 2.

The ZmbZIP33 was cloned into pFGC5941 vector (with bar label) using specific primers (Supplementary Table 2), and then introduced into Arabidopsis by agrobacterium. Three independent homozygous lines, L1, L2, and L3 were selected. Transgenic Arabidopsis line with overexpression of ZmbZIP33 and Wild type (WT,Col) seedlings grown on 1/2 MS culture medium (MS) for 1 week. Then transferred to MS medium with 150 and 250 mM of mannitol. One week later, the root length of transgenic lines and Col (WT) Arabidopsis was observed. Simultaneously, transgenic lines and Col seedlings were transplanted into the soil with portions of sand, compost, and perlite (3: 3: 1, respectively) for 20 days in a climate chamber at 22°C with 60% of relative humidity in a 16-h/8-h day/night cycle. Drought stress was applied to transgenic lines and Col plants for 4 and 6 days, then watered for 5 days.

The specific fragments of 180 bp in the ORF region of ZmbZIP33 were selected for designing primers (Supplementary Table 2). Virus-induced gene silencing (VIGS) was used to silence ZmbZIP33.BMV-GFP is the vector for transiently silencing gene expression. The ZmbZIP33 gene was isolated from Va35 maize and loaded into the BMV-GFP vector to obtain the ZmbZIP33-BMV fusion. Then, BMV-ZmbZIP33 and empty BMV-GFP were introduced into Agrobacterium tumefactions strain GV3101. The Agrobacterium liquid of BMV-ZmbZIP33 and empty BMV-GFP were mechanically inoculated onto the leaves of A. tumefaciens-infiltrated N. benthamiana plants at 24 h post-infiltration, as described previously(Chen et al., 2017). Three days later, tobacco leaves were taken for virus extraction according to a previous method (Chen et al., 2017). The extracted BMV-ZmbZIP33 and BMV-GFP tobacco virus were infected into a second leaf of Va35 maize,and cultured in a 20°C light incubator for 6 days with a light-transmitting plastic cover. The plastic cover was subsequently removed and placed in a 22°C light incubator. BMV-GFP serves as a wild-type control, hereinafter referred to as BMV. The expressions of ZmbZIP33 were detected 7 days after virus infection on ZmbZIP33 mutant (BMV-ZmbZIP33) and BMV (Supplementary Figure 4B), and BMV-ZmbZIP33 and BMV plants were irrigated with 40% of PEG for 0, 4, and 6 days, respectively, and then watered for 4 days.

The activities of superoxide dismutase (SOD) and peroxidase (POD) were measured according to Azevedo et al. (1998). Chlorophyll content was measured according to Lakra et al. (2015). ABA content was measured using an enzyme-linked immunosorbent assay (ELISA) under drought stress and rewatering, as previously described (Yang et al., 2001). Leaves at three-leaf stage of BMV-ZmbZIP33 and BMV were detached, and the water loss rates of each lines were calculated within 5 h.

Three independent homozygous lines, L1, L2, and L3 were selected. Transgenic lines and WT (Col) seedlings were grown on MS medium with 0, 0.5 and 1 μmol/L of ABA. Ten days later, the sensitivities of transgenic lines and Col to ABA were observed.

BMV-ZmbZIP33 and BMV seedings were germinated for 5 days, and the seedlings were transferred to half-strength modified Hoagland's nutrient solution. Three-leaf-stage maize were transferred to nutrient solution containing 0, 1, and 5 μmol/L ABA. Five days later, the sensitivities of BMV-ZmbZIP33 and BMV to ABA were observed.

BMV-ZmbZIP33 and BMV plants were irrigated with 40% of PEG for 0,4 and 6 days, respectively, and then watered for 4 days. The leaves stomata image of BMV-ZmbZIP33 and BMV were obtained by scanning electron microscopy (XL-30ESEM, PHILIPS, Netherland).

When using 20% Polyethylene glycol 6000 (PEG) to simulate drought stress, it was found that compared with CK (control), the maize leaves and particularly the tips began to wilt after 60 h. After 96 h, the leaves were severely and the leaf tips had begun to die. After 3 days of rewatering, the leaves gradually expanded, however, the growth was worse than that of CK (Supplementary Figure 1), indicating that the maize had a partial compensation effect after rewatering. According to the phenotypic results, we selected 60 h, 96 h, and 3 days leaves for high-throughput sequencing (RNA-Seq). The sequencing data has been published in NCBI, and the accession number is PRJNA477643 (https://www.ncbi.nlm.nih.gov/sra/?term=SRA729707).

Functional annotation was performed on the DEGs obtained by RNA-Seq, and MYB/MYC, bZIP, NAC, HD-ZIP and DREB families transcription factors (TFs) were enriched (Supplementary Figure 3). All TFs except ZmbZIP35, ZmMYB1, ZmMYB2, ZmMYB11, ZmNAC6, ZmNAC7, ZmNAC12, ZmNAC16, and ZmNAC20 increased within 96 h of drought stress, however their expression levels decreased after 3 days of rewatering compared with CK (Figure 1, Supplementary Table 1).

By analyzing the correlation coefficients between the expression levels of these TFs, a co-expression network was constructed. The core genes that respond to drought stress and rewatering were further screened by the co-expression network. As shown in fig.2, ZmbZIP33, ZmbZIP15, and ZmbZIP19 had a higher correlation with more TFs, indicating that these three genes are at the core of the co-expression network. We speculate that these three genes, especially ZmbZIP33, play an important role in response to drought stress and rewatering. The expression of ZmbZIP33 was significantly up-regulated at 60 and 96 h under drought stress, and down-regulated after rewatering. It is worthwhile to further explore the drought tolerance and mechanism of ZmbZIP33.

Based on the results in Figures 1, 2, we speculate that ZmbZIP33 is a key gene in response to drought stress and rewatering. Using qRT-PCR to further analyze the expression patterns of ZmbZIP33 in response to drought and other abiotic stresses and ABA treatment. As shown in Figure 4, expressions of ZmbZIP33 in leaves were significantly up-regulated by PEG, NaCl, 37°C and ABA. ZmbZIP33 gene is strongly induced by PEG and reached a maximum 24 h, following this, the expression level dropped sharply 24 h after rewatering (Figure 3A). By contrast, ZmbZIP33 transcripts declined to their highest at 12 h following NaCl treatment (Figure 3B). Furthermore, it was found that ZmbZIP33 mRNA accumulates at 4 h and reached its highest level at 12 h following 37°C and ABA treatments (Figures 3C,D).

The CDS of ZmbZIP33 (GenBank accession number: ACF82064.1) was isolated using reverse-transcription PCR (RT-PCR) from the cDNA library of the “Yu 882” maize inbred line. The nucleotide sequence of ZmbZIP33 contains a 1,071 bp open reading frame that encodes a protein of 356 amino acids including a bZIP_plant_BZIP46 domain (Figure 4A), which is a DNA-binding and dimerization domain. ZmbZIP33 encodes ABSCISIC ACID-INSENSITIVE 5-like protein 5. This family protein is mainly involved in ABA-activated signaling pathway and positive regulation of transcription. A protein phylogenetic analysis showed that ZmbZIP33 is closely related to SbABI1, SitbZIP68 and OsABA1 (Figure 4B). Studies have shown that SbABI1 and OsABA1 proteins respond to abiotic stress by participating in the ABA pathway (Kagaya et al., 2002; Kobayashi et al., 2005). The present study have confirmed that ZmbZIP33 actively responded to exogenous ABA treatment. Therefore, ZmbZIP33 gene may also have similar functions to homologous genes, responding to drought stress and relying on ABA pathways.

First, we verified whether ZmbZIP33 has transcriptional activity. The open-reading frame of ZmbZIP33 was fused with the GAL4 DNA-binding domain of the pGBKT7 vector. The pGBKT7-ZmbZIP33 construct was transformed into yeast strain AH109. As shown in Figure 4C, the yeast cells harboring negative control grew on the SD/-Trp medium only and did not exhibit α-galactosidase activity. In contrast, the transformants containing pGBKT7-ZmbZIP33 and the positive control grew well on the SD/-Trp medium and SD/-Trp/-His/-Ade/X-α-Gal medium. These results suggested that ZmbZIP33 had transcriptional activity in yeast.

The coding sequence of ZmbZIP33 was fused with the CaMV 35S promoter and T3 generation transgenic lines were obtained after the resulting construct was transfected into Arabidopsis. The expression level of ZmbZIP33 in overexpression lines were significantly higher than that of Col (WT) (Supplementary Figure 4A). The segregation of empty vector and bZIP33 transgenes and test goodness of fit for Medellian patterns were also investigated (Supplementary Table 3). The results showed that three transgenic Arabidopsis lines of ZmbZIP33 (L1–L3) were successfully selected.

Col and transgenic Arabidopsis lines were treated with 150 and 250 mM of mannitol to simulate drought stress (Figure 5A). There was no difference between the phenotype of Col and transgenic Arabidopsis before drought treatment. Following 1 week of treatment, it was found that the roots of the transgenic lines were longer than those of the Col at both mannitol concentrations (Figures 5B,C), indicating that ZmbZIP33 overexpression increased drought resistance of Arabidopsis. After 4 and 6 days of drought exposure, the transgenic Arabidopsis lines grew better than Col. After rewatering for 5 days, both Col and transgenic Arabidopsis lines recovered, and the transgenic Arabidopsis lines recovered better than the Col (Figure 5D). The leaves chlorophyll contents of the leaves of L1–L3 transgenic lines were significantly higher than that of Col after drought stress and rewatering (Figure 5E). This indicated that the overexpression of ZmbZIP33 improved drought resistance and recovery ability of Arabidopsis. In addition, we found that the difference in ABA content between Col and L1–L3 transgenic lines was small before stress (0 DOD), however after 4 and 6 days of drought stress, the ABA contents in transgenic lines were significantly higher than Col. Interestingly, after 5 days of rewatering, the ABA contents in L1–L3 transgenic lines dropped rapidly and approached the level found in the Col (Figure 5F).

Virus-induced gene silencing (VIGS) was used to silence ZmbZIP33 in order to study the drought resistance and recovery ability of maize (Figure 4B). Next, BMV and BMV-ZmbZIP33 mutant were subjected to drought stress for 4 and 6 days, respectively. It was found that BMV and BMV-ZmbZIP33 leaves became curled up, wilted and lost their green color, and gradually the leaf tips withered. Additionally, the BMV seedlings grew better than the BMV-ZmbZIP33 seedlings. Following 4 days of rewatering, BMV resumed growth, but BMV-ZmbZIP33 growth did not return to normal level (Figure 6A). The chlorophyll content of BMV leaves was significantly higher than that of BMV-ZmbZIP33 leaves after drought stress and rewatering (Figure 6B). After drought stress, the activity of POD and SOD in BMV and BMV-ZmbZIP33 leaves increased and these effects were more pronounced for BMV than for BMV-ZmbZIP33 (Figures 6C,D), resulting in a decrease in drought resistance and recovery ability of maize.

The accumulation of ABA content promotes stomatal closure and reduces water evaporation. To further study the role of ZmbZIP33 in regulating water saving, we tested ABA content, stomatal closure and water loss tests under drought stress and rewatering in BMV and BMV-ZmbZIP33.ABA concentration was below detection level before drought in BMV and BMV-ZmbZIP33,whereas, in response to drought, ABA content significantly increased in both genotypes and decreased after rewatering. Interestingly, ABA concentration in BMV was significantly higher than that in BMV-ZmbZIP33 under drought stress (Figure 6E). Measurements of stomatal movements in the leaves showed that completely closed stomata number were less and completely opened stomata number were more in BMV-ZmbZIP33 compared with BMV plants (Figure 6F). Measurements of the water loss rate of leave showed that BMV-ZmbZIP33 plants lost more water than BMV plants (Figure 6G). These results suggested that silencing the expression of ZmbZIP33 resulted in an increased water loss rate and reduced drought resistance of the BMV-ZmbZIP33 due to increased stomatal opening. The results suggested that silencing the expression of ZmbZIP33 hinders ABA accumulation and increases stomatal opening, leading to an increased water rate loss, ultimately, reducing drought resistance.

To investigate whether ZmbZIP33 is involved in the ABA signaling pathway, transgenic and WT Arabidopsis lines were treated with ABA. When germinated and grown on MS medium, ZmbZIP33 transgenic plants showed no difference in growth as compared to Col plants. However, when cultured on MS medium with 0, 0.5, and 1 μmol/L ABA for ten days, seed germination and primary root growth of ZmbZIP33 transgenic plants (L1–L3) was severely inhibited, compared to Col (Figure 7A). The root lengths of L1, L2, and L3 transgenic lines were significantly less than those of Col at 40 – 60% for plants grown on MS medium containing 0.5 and 1 μM ABA, respectively (Figure 7B).

Sensitivity to ABA of maize lines with silenced ZmbZIP33 was also investigated at 0, 1, and 5 μmol/L ABA for 5 days. As shown in Figure 7C, the root growth of BMV was greatly inhibited for the 1 and 5 μmol/L ABA treatments, but not for the BMV-ZmbZIP33. It indicated that silencing the expression of ZmbZIP33 (BMV-ZmbZIP33) plants displayed ABA insensitivity. We speculated that activation of ZmbZIP33 is important for its participation in ABA signal transduction.

The PYR/PYL receptor, PP2C protein, and SnRK2 kinase are key members of this pathway and therefore we explored the position and interaction of these core components. Pathways analysis of transcriptome differential genes found that ZmPYL10, ZmPP2C7, ZmSRK2E, and ZmbZIP33 are enriched in the ABA signaling pathway. Therefore, these four genes were selected to explore the ABA signal pathways in ZmbZIP33 transgenic lines in response to drought stress and rewatering. The subcellular localization of ZmbZIP33-GFP and ZmPP2C7-GFP were determined by observation of green fluorescence, which was found in the nucleus only, indicating that ZmbZIP33 and ZmPP2C7 are located there. By contrast, green fluorescence from ZmSRK2E-GFP was observed in the plasma membrane, indicating that ZmSRK2E is located in the plasma membrane (Figure 8A). The results of Y2H (Figure 8B) showed that co-transformation of AT-ZmPYL10 and KT-ZmPP2C7, KT-ZmPP2C7 and AT-ZmSRK2E, AT-ZmSRK2E, and KT-ZmbZIP33 all showed blue spots, indicating an interaction within each of these pairs. BIFC results showed (Figure 8C) that co-transformation of ZmPYL10-cY and ZmPP2C7-nY, ZmPP2C7-nY and ZmSRK2E-cY, ZmSRK2E-cY and ZmbZIP33-nY showed strong yellow fluorescence signals in the nucleus, plasma membrane and plasma membrane, respectively (Figure 8C). This indicated that ZmPYL10 interacts with ZmPP2C7 in the nucleus, ZmPP2C7 interacts with ZmSRK2E in the plasma membrane, and ZmSRK2E interacts with ZmbZIP33 in the plasma membrane.

To further verify whether ZmbZIP33 responds to drought stress and rewatering through this pathway, we analyzed the expression of genes encoding these interacting proteins (Figure 8D). Expression levels were tested in BMV-ZmbZIP33 and BMV following 0 DOD, 4 DOD, and R4 D treatment as well as during 24 h treatments with PEG, rewatering, and ABA. It was found that under 0 DOD, 4 DOD, and R4 D treatments, the expression levels of ZmPYL10 and ZmSRK2E in the BMV-ZmbZIP33 were lower than those of BMV, while the opposite was ture for ZmPP2C7. However, the expressions of ZmPYL10 and ZmSRK2E in normal maize leaves were significantly up-regulated by 3–4 folds after drought stress and 24 h ABA treatment, while ZmPP2C7 was down-regulated. These results indicated that ZmbZIP33 may form oligomers with ZmSRK2E and ZmPP2C7 in response to drought stress.

Having shown that overexpression of ZmbZIP33 improved the sensitivity of Arabidopsis to ABA during seed germination/root growth and drought-tolerant and recovery ability. We further explored whether ZmbZIP33 responds to maize drought stress and rewatering through an ABA-dependent pathway. Therefore, we analyzed the expression of key genes for ABA biosynthesis in ZmbZIP33 transgenic line and control under drought stress and rewatering. 9-cis-epoxycarotenoid dioxygenase (NCED) and Zeaxanthin epoxidase (ZEP) encode key enzymes in ABA biosynthesis. Befoer drought stress (0 DOD) there were no significant differences between the expression level of NCED and ZEP in Col and ZmbZIP33 and in BMV and BMV-ZmbZIP33 (Figure 9). Drought stress induced expression of NCED and ZEP in both genotypes, the expression level began to decrease after rewatering. Compared with the control, the expression levels of NCED and ZEP were higher in overexpression ZmbZIP33 transgenic Arabidopsis under drought stress and rewatering. However, the expression of NCED and ZEP in BMV-ZmbZIP33 was significantly lower than that of BMV under drought stress and rewatering.

Yoshida et al. found that AREB1, AREB2, and ABF3, which encode ABSCISIC ACID-INSENSITIVE 5-like protein, are the main regulators, interacting with SnRK2.2 to regulate the expression of key genes in the ABA signaling pathway under drought stress, thereby improving the drought resistance of Arabidopsis (Yoshida et al., 2010). Kang et al. confirmed that the drought resistance of transgenic Arabidopsis plants overexpressing ABF3, AREB2, and ABF4 was significantly improved, while the three mutants of areb1, areb2, and abf3 were very sensitive to drought (Kang et al., 2002). Ying et al. found that overexpression of ZmbZIP72 can significantly enhance the tolerance of Arabidopsis to drought (Ying et al., 2012). Zong et al. found that OsbZIP23 plays an important role in rice drought resistance (Zong et al., 2016). Yang et al. found that overexpression of ZmbZIP17 increased the expression of ABA-responsive genes, thereby improving the drought resistance of Arabidopsis (Yang et al., 2013). In our study we found that overexpression ZmbZIP33 of Arabidopsis transgenic lines showed lower leaf wilting and longer roots under drought stress and rewatering compared to CK (Figures 5A–D). Growth of the root system is positively related to the ability of maizes to absorb and transport water. Therefore, overexpression of ZmbZIP33 in transgenic Arabidopsis enhanced drought resistance and recovery ability. However, the drought resistance and recovery ability of BMV-ZmbZIP33 mutant are reduced, indicating that ZmbZIP33 is a positive regulator of drought stress and rewatering (Figure 6A).

Chlorophyll is a pigment related to photosynthesis, which is responsible for the capture and transmission of light energy and the process of transforming it into chemical energy during photosynthesis. A change of chlorophyll content affects the intensity of photosynthesis. The results of this study showed that chlorophyll content of overexpressing ZmbZIP33 transgenic Arabidopsis leaves was significantly higher than that of wild-type (Figure 5E). However, after 4 and 6 days drought stress and 4 days rewatering, the leaves chlorophyll content of BMV-ZmbZIP33 was significantly lower than that of control (Figure 6B). The results showed that the ZmbZIP33 transgenic line affected the chlorophyll content of leaves under drought stress and rewatering.

Under drought stress, the accumulation of reactive oxygen species results in the peroxidation of cell membrane lipids, which leads to disruption of physiological function of maize and can even cause cell death. Antioxidant enzymes such as SOD and POD can scavenge reactive oxygen species, which is an important regulatory mechanism to resist drought stress. As the first line of defense of maize antioxidant system, SOD primarily catalyzes the dismutation of two superoxide free radicals to form oxygen and hydrogen peroxide. Following this, hydrogen peroxide reduction is catalyzed by a peroxidase which removes active oxygen. Collin et al. confirmed that increased expression of POD and SOD reduces drought-induced damages in barley (Collin et al., 2020). This study found that under drought stress and rewatering, lower SOD and POD activities were measured in the BMV-ZmbZIP33 mutant than BMV (Figures 6C,D). The results speculated that drought stress may cause more damage to the cell membrane of BMV-ZmbZIP33.

Moreover, we observed a higher ABA content in Arabidopsis transgenic lines than in Col under drought stress and rewatering, ABA content in Arabidopsis transgenic lines dropped rapidly and tended to the Col level (Figure 5F). However, ABA content in BMV-ZmbZIP33 mutant was significantly lower than that of BMV under drought stress, and after rewatering ABA content in BMV-ZmbZIP33 mutant declined rapidly to the Col level (Figure 6E). The results showed that ZmbZIP33 may affects ABA biosynthesis or accumulation in drought and rewatering. Another physiological trait that ensures a low water loss is stomatal closure in response to water scarcity. Stomatal movement is regulated by ABA content with high ABA content promoting stomata closure (Saito and Uozumi, 2019). Under drought stress, ABA content of BMV-ZmbZIP33 mutant was lower than that of BMV, and it was found to have a higher number of open stomata numbers. After rewatering, the difference of ABA content between BMV-ZmbZIP33 and BMV was small, the stomatal state was similar (Figure 6F). Stomata affected the physiological functions of transpiration and photosynthesis. In the transpiration process, it is the main factor that directly controls the water loss. This study found that under drought stress and after rewatering, BMV-ZmbZIP33 had a significantly increased water loss rate compared to BMV (Figure 6G). This showed that ZmbZIP33 controlled stomatal movement by regulating ABA content, which in turn affected water loss rate under drought stress and rewatering. In Arabidopsis, ABSCISIC ACID-INSENSITIVE 5-like protein, AtABF3 and AtABF1 take part in ABA signaling under abiotic stresses (Yoshida et al., 2015). In summary, these findings demonstrated the role of ZmbZIP33 in ABA biosynthesis and signal transduction regulation under drought stress and rewatering.

Drought related ABA-dependent and ABA-independent pathways have been defined in plants. Most of the identified bZIP proteins were highly sensitive to ABA and can regulate abiotic stress response through ABA-dependent signaling pathway, e.g., AtAREB, AtABFs, OsbZIP33, and OsbZIP23 (Xiang et al., 2008; Chen et al., 2015; Wang et al., 2017). Kang et al. confirmed that the Arabidopsis plants overexpressing ABF3, AREB2, and ABF4 were significantly more sensitive to ABA, while the three mutants areb1, areb2, and abf3 were less sensitive to ABA (Kang et al., 2002). In wheat ABSCISIC ACID-INSENSITIVE 5-like protein, TaABI5 overexpression in Arabidopsis caused ABA-hypersensitive (Utsugi et al., 2020). In this study, we found that overexpression of ZmbZIP33 Arabidopsis plants were significantly sensitivity to ABA, but silencing ZmbZIP33 were insensitive to ABA (Figure 7). We speculated that activation of ZmbZIP33 may be important for its participation in ABA signaling transduction.

From the above results, it can be seen that ZmbZIP33 is not only significantly up-regulated after being induced by ABA induction, but also that the ZmbZIP33 overexpressing Arabidopsis was more sensitive to ABA than Col, while the ZmbZIP33-silent maize was not sensitive to ABA. In addition, this gene regulates the ABA synthesis process to participate in drought and rewatering. Therefore, we speculate that ZmbZIP33 responded to drought stress and rewatering through an ABA-dependent pathway. In this study, we found that ZmPYL10 interacted with ZmPP2C7 in the nucleus, ZmPP2C7 interacted with ZmSRK2E in the plasma membrane, and ZmSRK2E interacted with ZmbZIP33 in the plasma membrane (Figures 8A–C). We also analyzed the expression levels of ZmPYL10, ZmPP2C7, and ZmSRK2E in BMV and BMV-ZmbZIP33 mutant, and under drought stress, rewatering and ABA treatment. We found that under drought stress and rewatering, expression levels of ZmPYL10 and ZmSRK2E in BMV-ZmbZIP33 were lower than those of BMV, while opposite was found for ZmPP2C7. The expression of ZmPYL10 and ZmSRK2E were significantly up-regulated after drought stress and ABA treatment, while ZmPP2C7 was down-regulated. In the ABA signaling pathway, PYR/PYL is both an ABA receptor and suppressor of PP2C. PYR/PYL interacts with PP2Cs and inhibit phosphatase activity, allowing SnRK2s positively regulates the phosphorylation of downstream ABFs genes (Sakuma et al., 2007; Ma et al., 2009; Skubacz et al., 2016). In maize, ZmbZIP4 (ABSCISIC ACID-INSENSITIVE 5-like protein family member) is a phosphorylation substrate of ZmSRK2s and participates in an ABA-dependent signal pathway (Wang et al., 2018). Zong et al. have confirmed that OsABA1 (homologous protein of ZmbZIP33) can be phosphorylated by SAPK2 and SAPK6, and participates in an ABA-dependent signal pathway (Zong et al., 2016). In Barley, overexpression of HvABI5 increases HvSnRK2.1 expression, enabling the amplification of ABA signaling and activation of ABA-related genes. Under drought stress, HvABI5 directly upregulated of ABA synthesis, and ABA signaling via induction of HvPP2C4 and HvSnRK2.1 in response to drought stress (Collin et al., 2020).

9-cis-epoxycarotenoid dioxygenase (NCED) and Zeaxanthin epoxidase (ZEP) encoding key enzymes in ABA biosynthesis. In our study, NCED, which is a rate-limiting enzyme for ABA biosynthesis, showed a much higher expression in overexpression ZmbZIP33 transgenic line, compared to control, in response to drought stress and rewatering. However, the expression of NCED in BMV-ZmbZIP33 was significantly lower than that of BMV under drought stress and rewatering. The expression of ZEP is the same as NCED (Figure 9).

Therefore, we speculate that overexpression of ZmbZIP33 promotes ABA biosynthesis, which may induce ABA and ZmPYL10 to form a complex under drought stress. Subsequently, ZmbZIP33 may form oligomers with ZmSRK2E and ZmPP2C7, eventually inducing gene expression and stomata closure, thereby reducing water evaporation, and restoring growth in a short time. Our results provided a reliable reference for exploring how ZmPYL10, ZmPP2C7, ZmSRK2E, and ZmbZIP33 in response to drought stress through the ABA signaling pathway.