Sucrose and ABA regulate starch biosynthesis in maize through a novel transcription factor, ZmEREB156

Sucrose is not only the carbon source for starch synthesis, but also a signal molecule. Alone or in coordination with ABA, it can regulate the expression of genes involved in starch synthesis. To investigate the molecular mechanisms underlying this effect, maize endosperms were collected from Zea mays L. B73 inbred line 10 d after pollination and treated with sucrose, ABA, or sucrose plus ABA at 28 °C in the dark for 24 h. RNA-sequence analysis of the maize endosperm transcriptome revealed 47 candidate transcription factors among the differentially expressed genes. We therefore speculate that starch synthetic gene expression is regulated by transcription factors induced by the combination of sucrose and ABA. ZmEREB156, a candidate transcription factor, is induced by sucrose plus ABA and is involved in starch biosynthesis. The ZmEREB156-GFP-fused protein was localized in the nuclei of onion epidermal cells, and ZmEREB156 protein possessed strong transcriptional activation activity. Promoter activity of the starch-related genes Zmsh2 and ZmSSIIIa increased after overexpression of ZmEREB156 in maize endosperm. ZmEREB156 could bind to the ZmSSIIIa promoter but not the Zmsh2 promoter in a yeast one-hybrid system. Thus, ZmEREB156 positively modulates starch biosynthetic gene ZmSSIIIa via the synergistic effect of sucrose and ABA.

Scientific RepoRts | 6:27590 | DOI: 10.1038/srep27590 SP8a element in the iso1 promoter 16 . These reports confirm the importance of sucrose as a signalling molecule, but the molecular mechanism is not fully understood.
In a previous study, we found that sucrose combined with abscisic acid (ABA) synergistically influenced expression of 15 starch biosynthetic genes in maize endosperm 17 . Few reports indicate that sucrose induces starch biosynthetic gene expression by interacting with ABA signalling pathways [18][19][20][21] . AtAPL3 and OsAPL3 expression increased in response to exogenous application of sucrose in Arabidopsis leaves and cultured rice cells, respectively 20,21 . In addition, their expression was further enhanced by co-treatment with ABA. However, the molecular mechanisms by which starch biosynthetic genes in maize endosperm are regulated by interaction with sucrose and ABA remain unclear. In the present study, maize endosperms 10 DAP (days after pollination) were treated with sucrose, ABA, or both, and then analysed by RNA sequencing (RNA-seq). Analysis of these gene sets identified different treatments of gene expression, including hundreds of transcription factor genes. We found that some transcription factor genes were affected synergistically by Sucrose and ABA. We hypothesized that at least one transcription factor gene would be involved in maize endosperm starch synthesis by synergistic effect of sucrose combined with ABA. The results lay a foundation for understanding the underlying mechanisms that control seed yield and quality.

RNA sequencing and data analysis.
To analyse global gene expression in maize endosperm in response to sucrose or/and ABA signalling, maize endosperms were collected ten DAP and treated with sucrose (Suc), abscisic acid (ABA), or both (Suc + ABA). Mannitol was added to the samples without sugars as an osmotic control. Four cDNA libraries were constructed from total RNAs extracted and analysed sequences on the Illumina HiSeq TM 2000. After quality control, approximately 214 million valid reads and roughly 19.3 Gb of nucleotides were obtained. The generated reads were then aligned to the maize reference gene set based on B73 genome (release 5b.60) by applying the SOAPaligner/SOAP2 programs 22 . Sample data from the four libraries were summarized in Table 1. About 62% of the reads from each sample perfectly matched the gene set, [15][16].8% of the reads mapped to the gene set with no more than five aligned positions, and about 74-78% of the RNA-seq reads mapped to a unique position in the gene set.
Since reference genes have different lengths, the read location on each gene is standardized to a relative position (which is calculated as the ratio between read location on the gene and the gene length), and then the number of reads in each relative position is counted. The reads on the reference genes of our libraries were evenly distributed (Supplementary Figure S1), indicating that the randomness of the reads was satisfactory. As shown in Supplementary Figure S2, the distribution pattern of unique reads over different read abundance categories was similar for all four RNA-seq libraries.

Statistical analyses of differentially expressed genes (DEGs).
Uniquely mapped reads were used to estimate normalized transcription level as reads per kilobase of transcript per million mapped reads (RPKM). To identify genes displaying significant expression changes during Suc, ABA, or Suc + ABA treatment, DEGs (FDR < 0.01, |log 2 Ratio| ≥ 1) were analysed by comparing the Suc, ABA, and Suc + ABA libraries with the control library, (i.e. Control-vs.-Suc, Control-vs.-ABA, Control-vs.-Suc + ABA). RNA sequence data analyses revealed that, when compared to the control, there were 2838 genes that were significantly differentially expressed in the Suc-treated endosperm (Fig. 1A, Supplementary Table S1). Among them, 1201 genes were up-regulated and 1637 genes were down-regulated (Fig. 1B, Supplementary Table S1). In the ABA-treated endosperm, 9396 genes were significantly differential expressed, with 3181 up-regulated and 6215 down-regulated genes (Fig. 1A,B, Supplementary  Table S1). In the Suc + ABA-treated endosperm, there were 4537 significantly differentially expressed genes, including 1661 up-regulated and 2876 down-regulated genes (Fig. 1A,B, Supplementary Table S1).
The hierarchical heatmap display revealed that maize endosperm treated with a combination of exogenous Suc and ABA had a different transcript pattern from that treated with Suc or ABA alone (Fig. 1C, Supplementary Table S1). As shown in the Venn diagrams in Fig. 1A, there was an overlap of 1504 genes between Suc and ABA groups, 1426 genes between Suc and Suc + ABA, and 3111 genes between ABA and Suc + ABA (Fig. 1A). There were only 897 significantly differential genes that were commonly shared in Suc, ABA and Suc + ABA groups (Fig. 1A). In addition, there were 805, 6275 and 1494 genes that were specific to the Suc, ABA and Suc + ABA  Total base pairs  4,752,209,520  4,902,232,680  4,752,209,520  4,902,232,680   Total mapped reads  40,845,708  44,414,887  41,222,336  41,128,823   Perfect match  32,907,332  35,290,266  32,654,624  32,715,975   < = 5 bp Mismatch  7,938,376  9,124,621  8,567,712  8,412,848   Unique match  39,194,164  42,663,993  39,502,779  39,401,228   Multi-position match  1,651, group respectively (Fig. 1A). These data suggested that the expression pattern in response to the combination of Suc and ABA was not just an additive or subtractive algorithm of those responding to Suc or ABA alone.
Gene ontology (GO) and pathway enrichment analysis of DEGs. To further understand the function of these DEGs, gene ontology (GO) term enrichment analysis (P < 0.05) was performed (Supplementary Figure S3  and Supplementary Table S2). For DEGs in the Control-vs.-Suc + ABA comparison, the most significantly enriched GO terms were ' cellular lipid metabolic process' (p = 6.25E-09) in the biological process (BP) group, 'catalytic activity' (p = 9. sucrose. In addition, their expression was further enhanced by co-treatment with ABA. For example, AGPS2 and PHOL were up-regulated by Suc, and SSI, SSIIIa, and bt1 were noticeably up-regulated by the combined action of Suc and ABA. We inferred from these results that sucrose and ABA also co-regulate the gene expression of starch synthesis genes in the developing endosperm of maize, which is consistent with the findings of Chen et al. 17 . In our RNA-seq analysis, 741 DEGs were annotated as TF genes and classified into 50 TF families according to the criteria of PlantTFDB 3.0 23 Table S5). Fifty-seven TFs were co-up-regulated or co-down-regulated among all three libraries.
Among the genes co-regulated among all three libraries, the NAC and bZIP TF family were highly induced. However, substantial numbers were regulated exclusively in Suc library (52) Table S6). The AP2/EREBP superfamily is composed of the AP2, ERF, and RAV families. The ERF family includes the ERF and CBF/DREB subfamilies, which are involved in plant responses to abiotic stress 24 . In our study, two starch synthesis genes (GRMZM2G429899, ZmAGPL1 and GRMZM2G141399, ZmSSIIIa) and a transcription factor from the AP2/EREBP superfamily (GRMZM2G421033, ZmEREB156) were chosen for further analysis. AGPase catalyses the first key regulatory step in the starch biosynthetic pathways present in all higher plants that produce ADP-Glc and pyrophosphate (PPi) from Glc-1-P and ATP 3 . The maize AGPL1 gene (shrunken-2, sh2) encodes the large subunit of the rate-limiting starch biosynthetic enzyme, ADP-glucose pyrophosphorylase 25 . A sh2 transgene increased maize yield by acting in maternal tissues to increase the frequency of seed development and played an important role only in the endosperm 26,27 . The maize ZmSSIIIa gene is one of the most important starch synthase genes and is mainly expressed in maize endosperm during starch biosynthesis. Mutations in maize eliminating SSIIIa (du1) produce an endosperm with a glassy, dull appearance 2,3 . For analysis of the cis-elements in promoters, the 2000 nucleotide sequences upstream of the transcription initiation sites of maize starch genes were used to search the PLACE (Plant Cis-acting Regulatory DNA Elements) database available online (http://www.dnaaffrc.go.jp/PLACE/signalscan). It was determined that cis-elements including the TATA-box, CAAT-box, G-box, ABRE, GARE, and Skn-1-motif (cis-acting regulatory element required for endosperm expression) were found in promoters. In addition, the cis-element SURE1 was rich in the two promoters.

Verification of DEGs using qPCR.
To confirm the reliability of the RNA-seq results, qPCR was conducted for 20 DEGs with high or low expression levels. Additionally, qPCR was used to measure eight genes related to starch synthesis. Our RNA-seq data appeared to be reliable via RNA-seq analyses subsequently being confirmed by qPCR (Fig. 3).

Transcriptional activation and Subcellular localization of ZmEREB156. To ascertain whether
ZmEREB156 has potential transcriptional activation ability, we fused the encoding region of ZmEREB156 in frame to the GAL4 DNA binding domain in the pGBKT7 vector, and transformed the resulting construct into yeast strain AH109. The results showed that cells transformed with pGBKT7-ZmEREB156 or pGBKT7-GAL4 AD (as a positive control) grew well on synthetic SD-Trp-His-Ura selection media and turned blue in X-α -gal indicator (Fig. 4A), demonstrating that ZmEREB156 was a transcriptional activator in yeast.
To determine the sub-cellular localization of the ZmEREB156 protein, the 35S:eGFP-ZmEREB156 construct and an empty vector were introduced into onion epidermal cells via particle bombardment. GFP fluorescence Overexpression of ZmEREB156 enhances the promoter activity of starch genes in maize endosperms. To determine whether ZmEREB156, in response to sucrose and/or ABA, was involved in the induction of the two starch genes (Zmsh2 and ZmSSIIIa) in vivo, the ZmEREB156 coding sequence was amplified by primer P156 (Supplementary Table S8) and cloned into an expression vector (pBIUbi-221) driven by the maize ubiquitin promoter (pUbi:ZmEREB156), which served as an effector construct (Fig. 6A). Promoters of two possible target genes were amplified from the genomic DNA of B73 by PCR using primers P1 and P2 (Supplementary  Table S8) and cloned into the reporter construct (pPromoter:Luc).
The effector constructs were co-bombarded into maize endosperm with the respective reporter constructs. As shown in Fig. 5, the luciferase (LUC) activity driven by the psh2:Luc construct in the ZmEREB156-overexpressing endosperms was ~8.6× higher than that of endosperms not expressing ZmEREB156. LUC activity also increased about 3.4× in the pSSIIIa:Luc construct-bombarded endosperms. These findings indicate that Zmsh2 and ZmSSIIIa were regulated by ZmEREB156, suggesting that the positive involvement of ZmEREB156 in response to Suc may be mediated by an ABA-dependent signalling pathway.
ZmEREB156 binds specifically to the promoters of ZmSSIIIa in yeast. To confirm whether the ZmEREB156 protein could bind to the promoter region of the two starch genes, a yeast one-hybrid assay was performed. The pGADT7-Rec2-ZmEREB156 plasmid (containing the putative DNA-binding domain of the two transcription factors fused to the GAL4 activation domain) and the construct pHIS-cis (promoters of the two putative target genes) were co-transformed into yeast strain Y187 (Fig. 6A). As indicated by the activation of the reporter genes, ZmEREB156 can bind to the promoters of ZmSSIIIa gene (Fig. 6B). These results implied that ZmEREB156 has DNA binding activity and may directly regulate the expressions of the target gene ZmSSIIIa; however, ZmEREB156 may indirectly regulate the expressions of the target gene Zmsh2.

Discussion
In other plants, sucrose signals are central in determining plant growth and development by interacting with Sucrose and ABA signalling pathways 7,28,29 . However, the molecular mechanism of these interactions is complicated and still unclear. To date, no studies on the possible interactions between ABA signalling and sucrose have been reported in starch biosynthesis of maize endosperm. Sugars regulate cellular activity at multiple levels, from transcription and translation to protein stability and activity 30 . The identification and understanding of transcriptional regulatory networks is a major challenge in biology. It has been observed that sucrose affects gene expression through the regulation of other transcription factors, such as bZIP11, MYB75/PAP1 and WRKY. Increasingly, computational methods are being used that identify relationships between gene expression patterns and putative regulatory promoters or sequences in promoter regions. Information on the response of regulatory promoter or promoter elements controlling gene expression to sugar and ABA allows transcriptional networks to be understood at a molecular level.
RNA-seq analysis was used to monitor global transcriptional changes at the 10 DAP developmental stage of endosperm in B73 maize under exogenous Suc, ABA, and Suc plus ABA, which enabled comprehensive analysis of differential transcriptional genes between different treatments (Fig. 1A and Supplementary Table S1). The DEGs were mainly involved in carbohydrate metabolism, hormone signal transduction, brassinosteroid biosynthesis, and fatty acid biosynthesis pathways, which are all related to endosperm development in maize. Expression Full-length promoters of Zmsh2 and ZmSSIIIa were constructed into the report vector, and ZmEREB156 was cloned into the effect vector, respectively; (B) The reporter construct, pSSIIIa:Luc and psh2:Luc, and the internal control construct, pUbi:Gus, were co-bombarded into 10 DAP maize endosperm either with (+ ) or without (− ) the effector construct (pUbi:ZmEREB156). LUC activity was normalized to GUS activity in every independent transformation. The data are given as the means ± SE of three replicates. The significance of the difference between the − ZmEREB156 and + ZmEREB156 conditions was analysed using a one-sided paired t-test (* * P < 0.01).
Scientific RepoRts | 6:27590 | DOI: 10.1038/srep27590 of 20 genes from the RNA-seq result was confirmed by correlation with qPCR. In addition, 47 candidate transcription factors were selected, based on their relatively high expression levels in the libraries that hold promise for future studies on molecular mechanisms underlying endosperm development in maize. The maize endosperm treated with a combination of exogenous Suc and ABA had a different transcript pattern from that treated with Suc or ABA alone (Fig. 1C and Supplementary Table S1). Approximately 12% of differentially expressed genes responding to the combination were not found in the differentially expressed genes responding to Suc or ABA. For example, ZmEREB156 was up-regulated specifically by the combination but neither by Suc nor ABA. In contrast, ZmbHLH171 was specially down-regulated by the combination (Supplementary  Table S1). In addition, there were 805, 6275 and 1494 genes that were specific to the Suc, ABA and Suc + ABA group respectively and the total number of genes exclusively up-or down-regulated by treatment of Suc or Suc + ABA is less than the number of genes exclusively regulated by ABA alone (Fig. 1A,B). These data suggested that the interaction between Suc and ABA was complicate and there might exist synergistic effect and antagonistic effect between sucrose and ABA response. Many reports have shown that sucrose up-regulates diverse genes related to the starch biosynthetic pathway, such as those that encode specific subunits of ADP-Glc pyrophosphorylase (AGPase) in different species 11,12,31 . In our study, most identified starch synthetic genes were induced by Suc or ABA, for example, AGPS2 and PHOL were up-regulated by Suc, and SSI, SSIIIa, and bt1 were noticeably up-regulated by the combined action of Suc and ABA (Fig. 3). Our results were consistent with Chen et al. 17 . These results suggest the existence of crosstalk between the sucrose and ABA signalling pathways in the starch biosynthetic pathway. Studies have suggested that sugars and ABA often have synergistic effects and antagonistic effects on diverse developmental processes in plants [32][33][34] . However, we mainly focus on the molecular mechanism of the synergistic effect between sucrose and ABA signalling involving in starch biosynthesis of maize endosperm in this paper and we will pay more attention to the antagonism effect on endosperm or seed development in maize for future research.
ZmEREB156, which belongs to the AP2/EREBP transcription factor family, was induced by the combined action of Suc and ABA. The AP2/EREBP gene family has divided into four subfamilies, AP2, RAV, dehydration-responsive element binding protein (DREB), and ERF (ethylene-responsive element binding factor). The RAV, DREB, and ERF subfamilies are of particular interest due to their involvement in plant responses to stress. For example, the AP2/EREBP transcription factor WRINKLED1 (WRI1) is involved in the regulation of seed storage metabolism in Arabidopsis 24 . In addition, Xiao found that OsERF2 affected the accumulation of sucrose and UDPG by mediating expression of key genes involved in sucrose metabolism and hormone signalling pathways 35 . In our study, Zmsh2 and ZmSSIIIa promoter activity increased about 8.6 times and 3.4 times, respectively, after overexpression of ZmEREB156. Interestingly, ZmEREB156 could bind to the promoters of ZmSSIIIa and but not bind to the promoter of Zmsh2 (Figs 5 and 6). These findings indicate that ZmSSIIIa was under the regulation of ZmEREB156, suggesting that the positive involvement of ZmEREB156 in the responses to Suc may be mediated by an ABA-dependent signalling pathway and the interaction between Suc and ABA was synergistic effect. It also suggests that ZmEREB156 might indirectly regulate the expressions of the target gene Zmsh2, and Zmsh2 might be co-regulated by ZmEREB156 with other transcription factors co-mediated by Suc/ABA signalling. A similar conclusion was reported by Grimault 36 . For example WRI1, which encodes the main regulator of lipid biosynthesis in the seed, is a direct target of LEC2. However, LEC2 also acts directly on genes involved in reserve accumulation including OLE1, encoding an oleosin, and At2S1-S4 and 2S-like, which encode seed storage proteins. Direct targeting by LEC2 is mediated by its B3 domain, which binds specifically to RY-motifs such as CATGCA. It was also deduced that ZmEREB156 plays a distinctive role in regulating the co-expression of several starch synthesis genes in maize endosperm through sucrose and ABA signalling. To date, demonstration of protein-protein interactions between starch biosynthetic enzymes has been confined to endosperm of cereal species. Some rice starch biosynthetic isozymes are physically associated with each other and form active protein complexes 37 . Abe showed that there is a close interaction among SSI, BEI, and BEIIb during amylopectin biosynthesis in rice endosperm 38 . This means that the requirement for multiple, synergistic signals of Suc and/or ABA is probably correlated with the complexity of the mechanism underlying starch synthesis pathway in maize endosperm. Further investigation is needed into the integral regulatory network of ZmEREB156 at the molecular level and the functional linkage of related maize AP2/EREBP factors that co-regulate starch synthesis in maize endosperm.
Production of stable, fertile transgenic lines is an essential technique in gene function studies; however, transformation in maize is difficult, with low transformation rates and long breeding times. Therefore, using transient transformation approaches to study gene function is an alternative method 39,40 . Because the ZmEREB156 gene was expressed in maize endosperm, 10 DAP endosperm was used as the receptor of transient expression, which was conducted in order to determine the regulatory effect of the ZmEREB156 protein on the expression of the starch biosynthesis gene. Transient expression in maize endosperms is a good method for identifying the roles of target genes expressed in endosperm and has been used successfully in our previous studies [41][42][43] . In conclusion, we speculate that the ZmEREB156 gene might play a role in the starch biosynthesis via the synergistic effect of sucrose and ABA based on a series of experimental validation.

Materials and Methods
Plant materials and growth conditions. Zea mays L. B73 inbred seedlings were grown in the field under recommended agronomic guidelines and self-pollinated. When silks emerged, strict self-pollinations were performed between 11:00 and 13:00 every day.
Ten constructed based on total RNAs extracted from the maize endosperms to use in RNA sequencing (RNA-seq) analysis. The optimum concentration of each treatment was defined as that at which the gene demonstrated a response to exogenous sucrose and ABA 44 . The timing of maize endosperm collection was based on previous reports that the developmental steps associated with endosperm function, including the activation of storage product synthesis and deposition, initiate at 8 to 10 DAP [45][46][47][48] . Furthermore, genes associated with energy reserve and carbohydrate biosynthesis are up-regulated at 10 DAP according to Transcriptome Sequencing 49 . We therefore selected 10 DAP maize endosperms for our experiments.
RNA extraction and library preparation for RNA-seq. Total RNA was extracted using TRIzol reagent (Invitrogen) and purified using the RNA easy Mini RNA kit (Qiagen). RNA quality and quantity were verified using a Nano Drop 1000 spectrophotometer and an Agilent 2100 Bioanalyzer prior to further processing at the Beijing Genomics Institute (BGI; Shenzhen, China). On-column DNase digestion was performed according to the manufacturer's protocol. The total RNA was treated with DNase I prior to library construction, and poly-(A) mRNA was purified with Magnetic Oligo d (T) beads (Dynabeads). The mRNA was fragmented by treatment with divalent cations and heat. First-strand cDNA was transcribed from the cleaved RNA fragments using reverse transcriptase and random hexamer primers; second-strand cDNA synthesis followed, using DNA polymerase I and RNaseH. The double-stranded cDNA was subjected to end-repair using T4 DNA polymerase, the Klenow fragment, and T4 polynucleotide kinase. This was followed by a single < A> base addition using Klenow 3′ to 5′ exo-polymerase and ligation to an adaptor or index adaptor using T4 DNA ligase. Adaptor-ligated fragments were separated by size on an agarose gel, and cDNA fragments of the desired range (200 ± 25 bp) were excised from the gel. PCR was performed to selectively enrich and amplify the cDNA fragments. After validation with an Agilent 2100 Bioanalyzer and ABI StepOnePlus Real-Time PCR System, the cDNA library was sequenced on a flow cell using the Illumina HiSeq TM 2000 sequencing platform. The sequence data were deposited in the NCBI Sequence Read Archive (http://www.ncbi.nlm.nih.gov/Traces/sra) under accession number SRP068962.
Statistical and functional analyses of DEGs. Clean reads were obtained after filtering the adaptor, empty, low-quality, and one-copy tags from raw reads. Next, they were aligned to the maize reference gene set based on B73 genome (release 5b.60) by applying the SOAPaligner/SOAP2 programs 22  Gene ontology (GO) and pathway enrichment analysis of DEGs. DEGs were subjected to GO functional enrichment and KEGG pathway analysis. Gene ontology annotation was carried out using Blast2GO software (v.2.5.0) 52 . GO enrichment analysis used hypergeometric test to find significantly enriched GO terms in the input list of DEGs, based on GO::TermFinder 53,54 . KAAS (KEGG Automatic Annotation Server) provides functional annotation of genes by BLAST comparisons against the manually curated KEGG genes database. The result contains KO (KEGG Orthology) assignments and automatically generated KEGG pathways.
Verification of changes in gene expression using qPCR. To confirm the reliability of the RNA-seq results, qPCR was conducted using the iQ5 real-time PCR detection system (Bio-Rad, CA, USA) and SYBR Green PCR Master Mix (Takara, Dalian, China). Twenty-three DEGs were tested using primer sequences listed in Supplementary Table S7. Sample collection and treatment were as described under "Plant materials and growth conditions", this paper.
Total RNA was extracted using TRIzol (Invitrogen, CA, USA) and treated with DNase I (TaKaRa) to remove any genomic DNA. Reverse transcription-PCR (RT-PCR) was performed with 1 μ g of total RNA using the PrimeScript ™ RT reagent kit (TaKaRa). The specificity of qPCR primers was confirmed by melting curve analyses. The Ct value was determined using the instrument's software. The relative quantification of gene expression was monitored after normalization to 18S rRNA expression as an internal control. The relative transcription levels were calculated using the 2 -ΔΔCt method. The qPCR results were obtained from three biological replicates and three technical replicates for each gene and sample.

Subcellular localization of transcription factors.
The complete coding sequence of ZmEREB156, minus the stop codon, was ligated into the pCAMBIA2300-35S-eGFP vector after amplification by PCR using primer G156 (Supplementary Table S8) with BamHI and XbaI restriction sites to create the fusion construct (pCAM-BIA2300-ZmEREB156-eGFP). The empty pCAMBIA2300-35S-eGFP vector was used as a control. The fusion construct and the control were introduced into onion (Allium cepa L.) epidermal cells by particle bombardment using PDS-1000/He (Bio-Rad) 50 . The transformed cells were incubated on 1/2 MS medium for 24-48 h at 28 °C. The subcellular localization of GFP fusion proteins was visualized with a fluorescence microscope BX61 (Olympus, Japan).

Analysis of transcriptional activation in yeast cells.
A transcription activation assay was performed in yeast strain AH109 according to the Yeast Protocols Handbook (Clontech). The full-length coding region of ZmEREB156 was generated by PCR using primer BD156 (Supplementary Table S8). The PCR product was cloned into the pGBKT7 vector using EcoRI and BamHI sites and was named pGBKT7-ZmEREB156. Plasmids pGBKT7, pGBKT7-Lam, and pGBKT7-53 were used as negative controls. The pGBKT7-GAL4 AD was used as a positive control. These constructs were transformed into yeast strain AH109 by the lithium acetate method. After confirmation by screening on selective medium plates without tryptophan (SD/-Trp), the positive colonies were transferred onto SD/-Trp/-His/-Ura selective medium with or without X-α -D-Galactosidase (X-α -gal), and the yeast cells were photographed after incubating the plates at 30 °C for 3-4 d to evaluate transcription activity.

Particle bombardment and transient expression assays.
To investigate whether the candidate transcription factors induced by sucrose and/or ABA regulated the expression activity of endosperm starch synthesis genes, transient expression assays were performed. The ZmEREB156 coding sequence, spanning from ATG to the stop codon, was amplified by primer P156 (Supplementary Table S8) and cloned into an expression vector (pBIUbi-221) driven by the maize ubiquitin promoter (pUbi:ZmEREB156), which served as an effector construct (Fig. 5A). The upstream ~2000 bp promoters of two possible target genes (Zmsh2 and ZmSSIIIa) were amplified from the genomic DNA of B73 by PCR using primers P1 and P2 (Supplementary Table S8) and cloned into the reporter construct pPromoter:Luc (pZmsh2:Luc and pZmSSIIIa:Luc) (Fig. 5A). The pUbi:Gus was used as an internal construct (Fig. 5A). For the analysis of the reporter constructs, the molar ratio of the reporter construct to the internal construct plasmid DNA (pUbi:Gus) was 2:1. To analyse the effector constructs' effect on promoter expression, the reporter construct, effector construct, and internal construct were combined at a molar ratio of 2:2:1 and then co-bombarded into maize endosperm using a PDS-1000/He (Bio-Rad) 50 . Each independent experiment consisted of three replicates and was repeated 3-4 times with similar results. After particle bombardment, the medium plates containing bombarded endosperms were sealed and incubated at 28 °C in the dark for 24 h. The fluorescence and luminescence were determined using a Luminoskan Ascent (Thermo, IL, USA) 50 .