Drought stress decreased the net photosynthesis rate and caused abnormal development
As elucidated in maize development, the ear at the V9 stage was undergoing primordium development, with the primordia determining the kernel rows. Kernel rows first initiate as “ridges” of cells that eventually differentiate into pairs of rows [4, 6]. Many potential spikelets were developing, and the number of kernel rows was determined at this stage. As shown in Additional file 1: Figure S1A, the average length of the ear was 2.55 cm under normal conditions and 1.95 cm when subjected to 5 days of drought stress. When the subsequent primordia were observed by scanning electron microscopy (SEM), their growth and development were delayed in plants subjected to drought stress, especially the upper part of the ear. For ears under control conditions, all the following primordia grow well to the tip of the ear, while in the ears subjected to drought, the development of subsequent primordia on the top part failed (Fig. 1b and c). As reflected in the mature ear (Fig. 1d and Additional file 1: Table S1), the ear length under drought stress conditions was significantly shorter than in the control (10.01 cm for drought stress and 12.36 cm for the control, grain numbers per row of 17.33 and 22.33, respectively), and the row number of ears was 12 under drought and 14 in the normal control. This showed that drought stress at the V9-V10 stage caused a reduced row number and no kernels on the upper part of the mature ear due to the disturbance of primordium development by drought stress.
Another stage that is sensitive to drought stress in maize is the grain fill period after pollination. During DAP1–4, the endosperm is composed of a large coenocyte in which nuclear divisions occur without cellularization [42, 43]. At approximately 6DAP, the endosperm differentiates into starchy endosperm, the basal endosperm transfer layer, the aleurone layer, and the embryo-surrounding region [43,44,45,46]. Five days drought stress treatment was performed after pollination for 24 h, and the kernels after the treatment were collected for morphological analysis. Paraffin sections indicated that whole kernel development was delayed, and the development of the basal endosperm transfer layer and embryo was particularly dramatically arrested by drought stress (Fig. 1e and f). Following maturation, the 100-grain dry weight was 20.33 g and 17.38 g, and kernel weight per ear was 59.75 g and 50.38 g under normal and drought stress conditions, respectively (Fig. 1g and Additional file 1: Table S1). As shown in Fig. 1g, there was difference in row number and slightly shorter ear length, but smaller kernels were observed between the plants subjected to drought stress and the control. The reduced grain weight by drought at this stage was due to the decrease in kernel size.
The ear leaf contributes greatly to the “source” for ear and kernel development as a photosynthesis factory. Here, the performance of the ear leaf at the 5DAP stage with and without drought stress treatment was observed. As shown in Fig. 1h, drought stress treatment led to a significant decline of the net photosynthetic rate (27.87 to 18.73 μmol CO2m−2S− 1, normal control to drought stress) and relative water content (RWC, 89 to 74%), while the soluble sugar contents (8.18 to 11.74 mg/g) and solute potential (− 6.34 Mpa to − 5.00 Mpa) were significantly increased. The analysis showed that the capability for photosynthesis was significantly arrested by drought stress treatment, and the “source” was reduced. Ultimately, the yield per plot was 2.31 kg under normal conditions and 1.57 kg (drought at V9) or 2.01 kg (drought at 5DAP) under drought stress (Fig. 1g and Additional file 1: Table S1). The yield loss during the treatment was due to the reduced “source” (soluble sugar content in leaves) and the “sink” (young ear or kernels).
Overall differentially expressed genes (DEGs) in three organs in response to drought stress
Young ear at the V9 stage and ear leaf and kernel at 5DAP from plants subjected to drought stress and grown under control conditions were used for RNA sequencing to identify the DEGs and pathways in response to drought stress (GEO Submission (GSE132113)). The total reads, base pairs, mapped reads, expressed genes and transcripts were listed in Additional file 1: Table S2. More expressed transcripts were identified in the ear and kernel compared to ear leaf. The latter consisted of differentiated and terminal tissues. As summarized in Fig. 2a, compared with the controls, 1136 upregulated and 689 downregulated genes were observed in the ear, 2357 upregulated and 1402 downregulated genes in the ear leaf, and 2627 upregulated and 3565 downregulated genes in the kernel. When comparing the DEGs in different tissues, 292 genes showed consistent changes in the three tissues. The kernel shared 662 DEGs with the ear and 1014 DEGs with the ear leaf, while only 141 DEGs were shared in the ear and ear leaf (Fig. 2b and c). To confirm the results, 22 genes with different transcript abundances were validated by real-time RT-PCR (Additional file 1: Figure S2). The expression of these genes showed good consistency between the two detection methods.
Gene Ontology (GO) terms for response to light intensity, response to stress (drought, heat and oxidative stress), and regulation of the ABA-mediated signaling pathway were significantly enriched (Fig. 2d). Common drought response pathways identified in the three organs included the ABA-dependent signaling pathway, osmotic protective substance synthesis, protein folding and NAC TF (transcription factor)-mediated drought response (Fig. 2e, Additional file 2: Table S3). In all three organs, genes encoding the enzymes involved in the synthesis of sucrose, trehalose, raffinose and proline and genes encoding peroxidase, thioredoxin and oxidative stress, which contribute to ROS scavenging, were also largely induced by drought stress (Additional file 2: Table S3). Late embryogenesis abundant and some other drought-induced protein families were also accumulated (Additional file 2: Table S3), which could facilitate the adaptation of plants to water deficit. The other significantly changed process was the protein folding process, including the dramatic induction of heat shock protein (HSP). As shown in Additional file 2: Table S3, genes encoding HSP 101, HSP 70 and small HSPs were dramatically induced by drought stress, especially the small HSPs (five HSP20, two HSP17.6, and one DNAJ-like 20). Two heat shock TFs were also largely induced: one was the homolog of AtHSFB2A (GRMZM2G098696, 8.11-fold (log 2 value was 3.02, both log 2 value and ratio in the supplemental tables and log 2 value in the figures) in the ear, 5.78-fold in the ear leaf and 3.81-fold in the kernel), and the other was the homolog of AtHSFC1 (GRMZM2G105348, 81.91-fold in the ear, 2.73-fold in the ear leaf and 97.21-fold in the kernel). AtHSFB2A was a key component of heat stress signaling [47, 48]. These heat shock TFs may act as a connection between drought stress and cytosolic protein folding. Metabolic adjustment, osmotic adjustment, and protein folding acclimation were observed in all organs after drought treatment, and therefore, we deduced that a common mechanism regulates these responses.
To better understand the coordinate regulation of genes in all the tissues, we analyzed the upstream regions of the 292 genes for common sequence motifs using PromZea. Thirteen candidate motifs were identified enriched in the promoters of these genes (Fig. 2f), 6 of them with the core ABA-responsive element (ABRE) binding site (ACGTG) and 5 with identified NAC TFs binding site. That means two candidate regulatory systems were played roles in all the tissues: ABA-dependent and NAC-mediated stress response pathways. As shown in Additional file 2: Table S3, the ABA biosynthesis genes β-Ohase 1 (GRMZM2G382534, 4.84-, 2.49- and 42.40-fold in the ear, kernel and ear leaf respectively) and NCED9 (GRMZM2G014392, 4.33-, 4.28-, and 3.57-fold in the ear, kernel and ear leaf respectively) were induced by drought stress. Interestingly, 12 ABA-induced PP2Cs were also significantly upregulated by drought stress, including four HAI (Highly ABA-Induced) homologs and eight HAB (Hypersensitive to ABA) homologs, which work as positive and negative regulator of ABA signaling [49, 50]. Twenty TFs were identified as differentially expressed in all three organs. Five NAC TFs were significantly induced, including three ATAF2 homologs (GRMZM2G347043, 5.95-, 35.21- and 8.39-fold upregulated, GRMZM2G336533, 30.40-, 17.16- and 13.36-fold upregulated and GRMZM2G123667, 10.46-, 2.50- and 3.61-fold upregulated in the ear, kernel and ear leaf respectively), one NAC002/ATAF1 homolog (GRMZM2G014653, 5.97-, 16.56 and 3.13-fold upregulated in the ear, kernel and ear leaf) and one NAC047 (GRMZM2G134073, 2.23- and 2.33-fold upregulated in the kernel and ear leaf). All these NAC TFs have been reported to be involved in the stress response and development [51, 52]. Overexpression of NAC002/ATAF1 activates TREHALASE1 expression and leads to reduced trehalose-6-phosphate levels and a reduced sugar starvation metabolome [53]. In addition, ATAF2 regulates auxin biosynthesis [54] and BR catabolism [55], while NAC002/ATAF1 regulates ABA biosynthesis [56] in Arabidopsis. The upregulation of ATAF1/2 homolog genes in the maize response to drought stress may not only be involved in the sugar starvation response but also impact plant hormone levels and metabolic adjustment.
Effects of drought stress on young ear development and cell division based on transcriptome analysis
As shown in Fig. 1, drought stress at the V9 stage significantly inhibited primordium development, which contributed to the spikelets and determined the number of kernels and the ear weight. Transcriptome analysis showed that the expression of genes involved in nucleosome assembly, cell division and growth for the formation of anatomical boundaries were significantly altered by drought stress (Fig. 3, Additional file 2: Table S4). Eleven genes encoding histone 3, 12 genes encoding histone 4, 2 genes encoding histone 2, and 4 genes encoding histone 1 were significantly downregulated by drought stress (Fig. 3a). The downregulation of CYC1; 2, CYC2; 3, CYC3; 1 and CYC3; 4 indicated a reduction of cell division (Fig. 3b, Additional file 2: Table S4). Identification of the regulator in this process is merited to modify the maize yield under drought stress conditions. By comparison, auxin was the most altered hormone signaling cascade, as demonstrated by the inhibited expression of two AUX1-like auxin influx transporters (GRMZM2G067022, 0.25-fold and GRMZM2G149481, 0.48-fold), the downregulation of ABP1 (auxin-binding protein 1, GRMZM2G078508, 0.49-fold), and the upregulated expression of one auxin efflux transporter (GRMZM2G050089, 2.12-fold), one IAA7/AXR2 (GRMZM2G079200, 3.55-fold), two IAA3/SHY2 (GRMZM2G115357, 6.97-fold and GRMZM2G152796, 5.99-fold) and some other auxin response genes such as GH3 and SAUR-like auxin-responsive family genes (Fig. 3d). Both the axr2–1 and axr3–1 exhibited strong insensitivity to ABA for embryonic axis elongation [57,58,59] . IAA3/SHY2, AUX1/LAX3 are required for auxin signaling that activates LBD16/ASL18 and LBD18/ASL20 to control lateral root development [60, 61]. Analysis of the DEGs in auxin signaling pathways revealed that these IAA-modulated developmental processes were altered in drought stress, which induced growth retardation. The GO terms relative to development and growth were also significantly different between control and drought conditions (Fig. 3c). The decreased cell division and growth due to drought stress caused a change in meristem activity and led to modified structural development.
Here, the significant upregulation of four NAC081/ATAF2 (GRMZM2G347043, 5.95-fold, GRMZM2G123667, 10.36-fold, GRMZM2G068973, 14.56-fold and GRMZM2G336533, 30.40-fold) and one NAC002/ATAF1 (GRMZM2G014653, 5.97-fold) and the downregulation of the homologs of NAC031/CUC3 (GRMZM2G430522, 0.45-fold), NAC098/CUC2 (GRMZM2G139700, 0.42-fold) and NAC047 (GRMZM2G134073, 0.40-fold) in response to drought stress treatment were observed compared with the ears under normal conditions (Additional file 2: Table S4). CUC2 and CUC3 have been reported to participate in the regulation of shoot meristem boundary and formation and subsequent development [62, 63]. ATAF1/2 have been reported not only as the central regulator of plant defense and light-mediated seedling development but also as a regulator of hormone metabolism, such as BR, auxin and ABA [54,55,56]. The downregulation of homologs of NAC031/CUC3, NAC098/CUC2 and NAC047 in response to drought stress treatment undoubtedly retarded ear development. The collaborative regulation of auxin, ABA and NAC was involved in the responses to drought stress in ear development in maize.
Effects of drought stress on kernel development and yield
As shown in Fig. 1, the significant developmental change caused by drought at the 5DAP stage was a delay or abnormal endosperm differentiation (Fig. 1), which ultimately led to a small kernel size. GO terms related to metabolic process, cell wall biosynthesis process, cytokinesis, epidermal morphogenesis and organ development were enriched in response to drought stress at this stage (Fig. 4a, Additional file 2: Table S5). A significant change was observed in carbohydrate metabolism pathways. Glycolysis and TCA pathway activities were significantly reduced; however, the synthesis of trehalose, which aids in the osmotic adjustment response to stress, was significantly increased (Fig. 4b). Amylose synthesis was decelerated (Fig. 4c), and disaccharide and polysaccharide utilization was accelerated, potentially due to the reduced supply of photosynthate. Moreover, under unfavorable conditions, the kernel attempted to use more carbohydrates to cope with the unfavorable environment instead of undergoing rapid cell divisions and cellularization.
During kernel development, phytohormones play important roles, especially auxin signaling. As shown in Fig. 5 and Additional file 2: Table S5, auxin polar transport was reduced by drought stress because of the downregulation of four AUX1/LAX genes (AUX1/GRMZM2G127949, 0.22-fold, LAX1/GRMZM2G067022, 0.05-fold, LAX1/GRMZM2G045057, 0.02-fold, and LAX2/GRMZM2G045057, 0.12-fold) and one PIN1 gene. ZmPIN1-mediated auxin fluxes facilitate auxin accumulation and promote kernel differentiation [64]. In this study, both auxin influx and efflux were affected by drought stress. Local auxin biosynthesis was promoted, especially the conversion of tryptophan to indole-3-acetaldehyde and then to indoleacetate (IAA) (Fig. 5a), and the input was reduced. Together with the change in the local auxin concentration, auxin signaling was also altered. As shown in Fig. 5a and Additional file 2: Table S5, IAA2 (3.95-fold), IAA3 (3.28-fold), IAA7 (10.52-fold), IAA8 (2.58-fold), IAA13 (2.90-fold) and IAA26 (13.06-fold) were upregulated and two IAA16 (0.19-fold and 0.12-fold), one IAA27 (0.47-fold) and one IAA3 (0.32-fold) were downregulated in the kernel in response to drought stress treatment compared with the control. Accompanying AUX/IAA, one ARF1 (0.28-fold), three ARF3 (0.19-, 0.36- and 0.07-fold), one ARF6 (0.36-fold), one ARF11 (0.28-fold) and an IBR1 (0.49-fold) were downregulated. The degradation of Aux/IAA proteins could free ARFs, and the latter could activate or repress auxin response gene expression directly. Auxin response genes such as SAUR and GH3 were differentially expressed compared with the control in the kernel under drought stress conditions. The differential expression of genes involved in auxin transport, local biosynthesis and signaling pathway may modify metabolism to establish a new hemostasis in kernels subjected to drought stress at this stage.
In addition to auxin, ethylene, BR and cytokine signaling were also changed to cope with the stress environment and to adjust growth or development. As shown in Fig. 5b, two EIN3 genes were upregulated (GRMZM2G033570, 4.88-fold and GRMZM2G317584, 3.62-fold) and their downstream RAP/ERF TFs showed varied expression levels. RAP/ERF TFs can help with the stress tolerance of plants. Compared with ethylene, BR and cytokines seem to function in growth and developmental regulation to acclimate to drought stress (Fig. 5c-d). Genes encoding BRI1 (GRMZM2G048294, 0.47-fold downregulated) and BSK2 (GRMZM2G169080, 0.32-fold and GRMZM2G054634, 0.23-fold downregulated), which were considered as main target of BR to regulate plant growth were differentially expressed. Their targets TCH4 (GRMZM2G119783, 0.20-fold), which contributes to cell elongation, and two CYCD3 (GRMZM2G107377, 0.12-fold and GRMZM2G161382, 0.18-fold), which contribute to cell division, were inhibited. Interestingly for cytokine signaling, five A-ARR were identified as downregulated (GRMZM2G096171, 0.34-fold, GRMZM2G179827, 0.12-fold, GRMZM2G129954, 0.17-fold, GRMZM2G319187, 0.45-fold, GRMZM2G148056, 0.24-fold), and three B-ARR were upregulated (GRMZM2G177220, 5.66-fold, GRMZM2G100318, 2.20-fold and GRMZM2G060485, 2.55-fold). B-type ARRs, ARR1, ARR10, and ARR12 redundantly act as negative regulators of drought responses in both ABA-dependent and -independent pathways [65] and mediate cytokinin signaling to WUS for the maintenance of stem cells [66]. Based on these findings, plant hormones play important roles in the adaptation of metabolism in kernels subjected to drought stress, including reduced cell division, expansion, differentiation and embryo development at this stage.
Effects of drought stress on ear leaf transcriptome and photosynthesis
Drought stress significantly reduced the photosynthesis capability of the ear leaf, which greatly contributes to the “source” for ear and kernel development as a photosynthesis factory in maize plants. In the ear leaf, GO terms related to photosynthesis were significantly enriched (Fig. 6a), with dramatically downregulated expression of photosynthesis protein and enzyme-coding genes. As shown in Fig. 6b and Additional file 2: Table S7, 21 genes involved in photosystem II, including 2 for PsbO, 4 for PsbP, 5 for PsbQ, 3 for PsbX, 2 for PsbY, 2 for Psb27 and 3 for Psb28, were downregulated in the leaf under drought stress conditions compared with the control. In the cytochrome b6/f complex, 7 genes in genes encoding PetA to D, showed significantly reduced expression levels, which was the FeS center and reduced electron transport. Twenty-four genes involved in photosystem I, including PsaA to H, PsaK, PsaL, PsaN and PsaO were downregulated. For photosynthetic electron transport, 1 gene encoding PetE, 6 genes encoding PetF and 4 genes encoding PetH/FNR were downregulated. However, 1 gene encoding PetF (GRMZM2G063126), that functions in distributing photosynthetic reducing power was dramatically induced 265-fold compared with the normal control conditions, which was the only upregulated gene in the photosynthesis systems. Eight genes in the ATPase complex were downregulated, including the genes for the gamma (1), delta (3 genes) beta (3) and b (1) subunits. Additionally, 19 genes encoding the antenna proteins were also downregulated. Plants under water deficit conditions exhibited a decline in photosynthetic rate through the control of stomatal closure by changing the turgor pressure in guard cells and activity of the PSII complex. When compare the candidate motifs in their promoter region, higher percentage of ABRE and NAC binding site were found compare to others, like MYB, ERF binding site and G-box (Fig. 6c). Figure 1 and Additional file 2: Table S7 show that the continuous decrease in net photosynthesis by drought stress in the ear leaf depends on transcription level regulation, which reduced the synthesis of some photosynthesis system proteins. This phenomenon resulted in a slow recovery for high rates of photosynthesis when drought stress was removed and led to a final yield loss compared with the normal conditions. Genes involved in photosynthesis and light harvest are listed in Additional file 2: Table S7, which implied a clear influence on the ear leaf by drought stress among the three organs.
As shown in Fig. 7 and Additional file 2: Table S7, four ABA receptor genes were differentially expressed, with downregulation of the homolog of PYR8 (GRMZM2G063882, 0.42-fold) and homologs of PYR2, 6 and 10 upregulated (GRMZM2G112538, 4.70-fold, GRMZM2G112488, 4.19-fold, GRMZM2G165567, 5.09-fold) in the ear leaf. As we mentioned previously, 9 ABA-induced PP2Cs were significantly upregulated by drought stress, including four HAI homologs that function in the negative regulation of ABA signaling and positive regulation of GA signaling and five HAB homologs genes that regulate activation of the Snf1-related kinase OST1 with ABA. Three SnRK2 genes showed differential expression, with two induced (GRMZM2G110922, 4.46-fold and GRMZM2G000278, 3.80-fold) and one downregulated (GRMZM2G110908, 0.42-fold) by drought stress. Additionally, the ABF TFs (ABF2/GRMZM2G479760, 2.38-fold and GBF4/GRMZM5G858197, 5.48-fold) were upregulated and then activated the expression of ABA response genes. Ethylene signaling was active since the genes encoding SIMKK (GRMZM5G834697, 3.5-fold), MPK6 (GRMZM2G002100, 2.59-fold), EIN3 (GRMZM2G040481, 64-fold) and EBF1 (GRMZM2G171616, 2.76-fold) were upregulated. For auxin, TIR1 was downregulated (GRMZM2G135978, 0.47-fold), two homologs of IAA26 were upregulated and one IAA27 was downregulated. For auxin response genes, six SAUR genes were dramatically induced, and another two were downregulated. GA signaling was inhibited with the upregulation of GID1C (GRMZM2G173630, 3.12-fold and GRMZM2G016605, 4.03-fold) and DELLA (GAI, GRMZM2G013016, 41.61-fold) and the downregulation of two PIF3 genes at a moderate level (GRMZM2G062541, 2.25-fold and GRMZM2G387528, 0.47-fold). For the cytokine pathway, two CRE1/WOL genes (AHK4/GRMZM5G833140, 0.19-fold, and AHK4/GRMZM2G155767, 0.24-fold) were markedly downregulated and another two members were upregulated (AHK4/GRMZM2G151223, 7.43-fold and AHK3/GRMZM2G423456, 2.64-fold), these genes encode cytokine-binding receptors that transduce signals across the plasma membrane. AHP4/GRMZM2G039246, which functions as a histidine-containing phosphotransferase factor, showed 0.25-fold downregulation. In comparison to B-ARR, which did not exhibit an obvious change, five A-ARR genes were dramatically downregulated in ear leaf after drought treatment: two ARR3 homologues (GRMZM2G179827, 0.02-fold, GRMZM2G096171, 0.03-fold), two ARR6 homologues (GRMZM2G040736, 0.001-fold and GRMZM2G392101, 0.03-fold and one ARR9 homologue (GRMZM2G129954, 0.13-fold). ABA, ethylene and cytokine signaling appeared to actively regulate the response to drought stress, but GA signaling was inactive during the acclimation to drought stress.