Multi-omics analysis reveals the pivotal role of phytohormone homeostasis in regulating maize grain water content

Abstract

Grain water content (GWC) is a key determinant for mechanical harvesting of maize (Zea mays). In our previous research, we identified a quantitative trait locus, qGWC1, associated with GWC in maize. Here, we examined near-isogenic lines (NILs) NILL and NILH that differed at the qGWC1 locus. Lower GWC in NILL was primarily attributed to reduced grain water weight (GWW) and smaller fresh grain size, rather than the accumulation of dry matter. The difference in GWC between the NILs became more pronounced approximately 35 d after pollination (DAP), arising from a faster dehydration rate in NILL . Through an integrated analysis of the transcriptome, proteome, and metabolome, coupled with an examination of hormones and their derivatives, we detected a marked decrease in JA, along with an increase in cytokinin, storage forms of IAA (IAA-Glu, IAA-ASP), and IAA precursor IPA in immature NILL kernels. During kernel development, genes associated with sucrose synthases, starch biosynthesis, and zein production in NILL , exhibited an initial up-regulation followed by a gradual down-regulation, compared to those in NILH. This discovery highlights the crucial role of phytohormone homeostasis and genes related to kernel development in balancing GWC and dry matter accumulation in maize kernels.

Headnote

ABSTRACT

Grain water content (GWC) is a key determinant for mechanical harvesting of maize (Zea mays). In our previous research, we identified a quantitative trait locus, qGWC1, associated with GWC in maize. Here, we examined near-isogenic lines (NILs) NILL and NILH that differed at the qGWC1 locus. Lower GWC in NILL was primarily attributed to reduced grain water weight (GWW) and smaller fresh grain size, rather than the accumulation of dry matter. The difference in GWC between the NILs became more pronounced approximately 35 d after pollination (DAP), arising from a faster dehydration rate in NILL . Through an integrated analysis of the transcriptome, proteome, and metabolome, coupled with an examination of hormones and their derivatives, we detected a marked decrease in JA, along with an increase in cytokinin, storage forms of IAA (IAA-Glu, IAA-ASP), and IAA precursor IPA in immature NILL kernels. During kernel development, genes associated with sucrose synthases, starch biosynthesis, and zein production in NILL , exhibited an initial up-regulation followed by a gradual down-regulation, compared to those in NILH. This discovery highlights the crucial role of phytohormone homeostasis and genes related to kernel development in balancing GWC and dry matter accumulation in maize kernels.

Keywords:

Grain water content

Maize kernel

Phytohormone homeostasis

1. Introduction

Modern agricultural production is increasingly shifting from labor-intensive practices to machinery, primarily due to greater efficiency and cost-effectiveness. However, maize production in China is not yet fully mechanized with mechanical harvesting of maize grain accounting for only 5% of all harvesting methods [1]. This limitation is largely attributed to challenges posed by high grain water content (GWC). Research indicates that the GWC of leading maize cultivars in China reaches 30%–40% at harvest time [2], a level considerably higher than the recommended 27% suitable for mechanical harvesting [3]. Excessively high GWC can also delay harvesting time, affecting the production of subsequent crops. Additionally, ears with high GWC are vulnerable to pests and birds, more susceptible to ear/grain rot, and prone to mildew during storage and transportation. Therefore, reducing GWC has become a pivotal objective in maize production.

GWC is a complex quantitative trait controlled by multiple genes. Identifying quantitative trait loci (QTL) associated with low GWC can aid in genetically reducing the GWC in maize, thereby facilitating mechanical harvesting. Over the past decade, numerous QTL related to GWC have been identified through both linkage and genome-wide association mappings [4–16]. However, none of these QTL has yet been proven optimal for molecular breeding efforts aimed at developing low-GWC maize varieties. Therefore, a more in-depth understanding of the physiological processes affecting GWC, such as grain development, maturation, and dehydration, is crucial for achieving breeding objectives.

Grain development can be categorized into three phases: the lag phase, the effective grain-filling phase, and the mature drying phase [17]. During the lag period, grain water weight (GWW) increases rapidly with minimal accumulation of dry matter. This stage is critical for determining grain size, as well as the subsequent grain growth rate [18]. In the effective grain-filling phase, GWW continues to increase, and dry matter accumulation rapidly accelerates. Kernel water status is primarily determined by internal metabolic activities associated with the deposition of storage products [19]. The replacement of water by these products (primarily starch granules) during the grain filling phase causes a progressive desiccation of the endosperm [20]. During the mature drying phase, dry matter accumulation ceases, and dry weight remains constant, reaching maximum grain yield, while GWC continues to decline [21].

Phytohormones play multifaceted roles in regulating various stages of plant growth and development. Jasmonic acid (JA) is crucial for coping with stress and maintaining a balance between plant growth and defense [22]. Under drought stress, JA helps to minimize water loss by regulating stomatal aperture and orchestrating the metabolism of reactive oxygen species to improve drought resilience. Both endogenous and exogenous JA in plants contribute to drought resistance [23]. For instance, in the drought-tolerant species Prunus armeniaca, endogenous JA levels temporarily increase following drought stress, stimulating leaf senescence, which in turn helps to prevent excessive water loss [24]. Soybeans sprayed with methyl jasmonate (MeJA) showed a significant increase in leaf water potential, thereby enhancing their drought resistance [25].

Abscisic Acid (ABA) is another critical regulator of diverse physiological processes, including seed maturation, source-sink transport, and drought response. Different strategies deployed to either avoid or tolerate dehydration may involve the regulation of stress-responsive gene expression through the ABA and other signaling pathways [26]. Aquaporins, including the plasma membrane PIP and tonoplast TIP proteins, are a category of channel proteins found in multiple tissues, responsible for transporting water and neutral substances [27]. The regulation of aquaporin activity by ABA signaling contributes in adjusting water flux into and out of the plant [28,29]. In maize, white endosperm kernels exhibited higher GWC than a yellow (y1) mutant. This difference was attributed to the y1 mutation, which causes a deficiency in phytoene synthase, an enzyme essential for the biosynthesis of carotenoids and ABA, consistent with a significant role of ABA in regulation of GWC [30,31]. Additionally, the accumulation of osmocompatible solutes, along with regulated biosynthesis of dehydrins and late embryogenesis abundant (LEA) proteins, plays an important role in both water retention and protection of proteins and membranes during stress conditions [32,33].

In this study, we developed a pair of near-isogenic lines (NILs) that differed at the qGWC1 locus, aiming to investigate dynamic changes in GWC during the grain filling phase. We conducted multi-omics reprogramming, including transcriptome, proteome, and metabolome analyses on the NILs. Our multi-omics analysis revealed that phytohormone homeostasis plays a pivotal role in regulating GWC. This research provides new insights into the genetic control and molecular mechanisms underlying GWC, offering information that could applied in breeding maize varieties with low GWC.

2. Materials and methods

2.1. Plant materials and phenotypic investigation

The parental inbred lines SN80044 and SN80007 were developed by Prof. Baoshen Liu, Shandong Agricultural University. The BC1F6 and BC1F7 populations were planted at Ledong (N18°60', E108°78',) in Hainan province in 2017 and at Tai'an in Shandong (N36°11′, E117°06′) in 2018. BC1Fn lines were developed from 134 recombinant inbred lines (RILs) backcrossed to SN80007, as described previously [10]. The NILs were planted at Beijing (39.9°N, 116.4°E), during the summers of 2019, 2020, and 2021, and at Ledong, during the winter of 2020. Seventeen kernels were planted per 4 m row (0.25 m between adjacent plants) with a row spacing of 0.50 m.

For QTL mapping, GWC was investigated as previously described [10]. The NILs (449 individuals each), were grown in Beijing during the summer of 2020 to investigate dynamic changes in GWC, grain drying rate (GDR), grain fresh weight (GFW), grain dry weight (GDW), and GWW throughout grain development. Approximately one hundred kernels were sampled, and fresh weight was immediately measured, followed by oven drying to a constant dry weight. GWC and GDR were calculated using the following formulas: GWC (%) = (fresh weight dry weight)/fresh weight x 100%; GDR (%) = (GWC (early sampling) - GWC (later sampling))/days. 100-GFW (per 100 kernels), 100-GDW, and 100-GWW were calculated with the following formulae: 100-GFW = fresh weight/number of kernels; 100-GDW = dry weight/number of kernels × 100; 100-GWW = 100-GFW - 100-GDW × 100. Agronomic traits (e.g., - ear length, ear stalk length, ear diameter, ear shaft diameter, number of ear leaves, male spike length, number of bracts, angle between the rod and stem, kernel row number, kernel number per row, kernel length, kernel width, plant height, ear height, and tassel branch number) were investigated in the NILs at 45 days after pollination (DAP). Days to silking and anthesis were recorded for each plant within each NIL. The volume of the water displaced by a set number of grain was used as a measure of grain size; i.e., grain size = displaced volume/number of kernels.

2.2. RNA-seq

A GWW difference in NILs were planted in Beijing in 2020 became apparent at 20 DAP and widened thereafter until harvest. Given that changes at the molecular level may precede changes in phenotype, we harvested grains from NILH and NILL plants at 10, 25, and 40 DAP. For each time point, three biological replicates were conducted, each consisting of five independent plants. Approximately 100 kernels were removed from the five ears and pooled into one sample, which was then flash-frozen in liquid nitrogen and stored at -80 °C. In total, 18 grain samples were crushed into small particles in liquid nitrogen using a mortar and pestle and mixed well for further RNA, metabolite, or protein extraction. Total RNA was extracted with a PLANT pure plant RNA Kit (Aidlab, Beijing), following the manufacturer's protocol. RNA-seq was performed by Novogene Bioinformatics Technology (Beijing). For library construction, 100 ng of poly(A) messenger RNA was fragmented as recommended by the manufacturer (Illumina, TruSeq RNA Library Prep Kit v2) and sequenced on an Illumina Hiseq 3000. Clean reads were then mapped to the masked maize genome in the Ensembl Zea_mays, AGPv4 database (https://plants.ensembl.org/Zea_mays/Info/Index). Gene expression calculation and normalization were based on FPKM (fragments per kilobase of exon model per million mapped reads) using Cufflinks, v2.1.1.

Differential expression analysis between the two groups (with three biological replicates per condition) was performed using the DESeq2 R package (v1.20.0) [34]. Differentially expressed genes (DEGs) were identified by using DESeq2, with an absolute |log2(-Fold Change)| > 0 and false discovery rate < 0.05. Gene Ontology (GO) enrichment analysis of DEGs was implemented by the clusterProfiler R package [35], where gene length bias was corrected. GO terms with a corrected P-value < 0.05 were considered significantly enriched by DEGs. The clusterProfiler R package was also used to statistically test the enrichment of DEGs in Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways [36].

2.3. Proteome analyses

The sample used for proteome analysis was the same as that used in the RNA-seq analysis; 18-grain samples were ground into a fine powder in liquid nitrogen using a mortar and pestle. The powder was suspended in 10x lysis buffer, containing 8 mol L 1 urea, 30 mmol L*¹ HEPES, 1 mmol L*¹ polyvinylpolypyrrolidone (PMSF), 2 mmol L*¹ EDTA, and 10 mmol L*¹ dithiothreitol (DTT), and then sonicated for 5 min. The material was centrifuged at 14,000xg for 20 min at 4 °C, and the supernatant was collected. Then, a four volumes of pre-chilled acetone was added, mixed by inversion, and left to precipitate overnight at 4 °C. The mixture was centrifuged at 14,000×g for 10 min, the pellet collected, resuspended in 1 mL of cold acetone, and centrifuged again at 14,000×g for 10 min. The supernatant was removed, the pellet air-dried, and reconstituted with lysate. After another centrifugation at 14,000×g for 20 min, the supernatant was collected and set aside. The protein concentration was determined using the Bradford assay (Bio-Rad, Hercules, CA, USA), with BSA as a standard [37]. Subsequently, 10 µg of each sample was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The extract was reduced with dithiothreitol at 60 °C for 1 h, then iodoacetamide was added to alkylate cysteine, and the mixture was incubated in darkness at 20 °C for 1 h. The protein was diluted with NH4HCO3 and digested with trypsin at 37 °C for 16 h at a protein/trypsin 50 (w/w) ratio. Acidified peptides with 10% (v/v) formic acid were desalted by reversed-phase extraction with the tip of a C18 ZipTip pipette, and then suspended in 0.1% formic acid (FA) for high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS).

Label-free analysis was conducted using HPLC-MS/MS coupled to a Q Exactive mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). Samples were injected using an autosampler and pre-enriched on a homemade C18 trap column. Thereafter, they were analyzed on a self-made analytical column, with solvents A and B as the mobile phases. The isolated peptide fragments were identified using Q Exactive HF MS/MS. Sequencing was performed using BMKcloud (Beijing). Protein identification and label-free quantification were performed with Proteome Finder (v2.1.0.81). Similarity searches were performed on the human UniProt Forward Database (UP000005640), with a cutoff of a 1.5-fold change in expression (P < 0.05) set to identify differentially expressed proteins (DEPs).

2.4. Metabolite analyses

The sample used for UHPLC-QTOF-MS was the same as that used in the RNA-seq analysis. Grain samples were ground into a fine powder in liquid nitrogen using a mortar and pestle. 50 mg of each sample was combined with 1000 µL of an extract containing an internal standard (1000:2) (methanol:acetonitrile:water volume ratio = 2:2:1, with an internal standard concentration of 2 mg L*¹), then vortexed and mixed for 30 sec and processed with a 45 Hz grinder for 10 min, sonicated in an ice-water bath for 10 min, and then allowed to stand for 1 h at -20 °C. The samples were centrifuged at 12,000 r min for 15 min at 4 °C, and 500 µL of the supernatant was removed and dried in a vacuum concentrator. The dried metabolites were reconstituted by adding 160 µL of the extract (acetonitrile:water volume ratio = 1:1), followed by vortexing for 30 sec and sonicating in an ice-water bath for 10 min. The samples were then centrifuged at 12,000 r min*¹ at 4 °C for 15 min. Thereafter, 120 µL of the supernatant was transferred into a 2 mL injection bottles, and 10 µL from each sample was mixed into a Quality Control (QC) sample for testing.

Metabolomic analysis was performed using a Waters Acquity IClass PLUS ultra-high-performance liquid chromatography (UHPLC) system, coupled with a Waters Xevo G2-XS QT high-resolution mass spectrometer using a Waters Acquity UPLC HSS T3 column (1.8 µm, 2.1 x 100 mm). Raw data collected using MassLynx v4.2 software (Waters, Milford, MA, USA) were processed with Progenesis QI software (Waters) for peak extraction, alignment, and other data processing operations. Identification of compounds was based on the online METLIN database through Progenesis QI software and Biomark's self-built library, ensuring that theoretical fragment identification and mass deviation were within 0.01% [38]. The raw peak area information was normalized to the total peak area for subsequent analysis. Principal component analysis (PCA) and Spearman correlation analysis were used to assess the repeatability of samples within groups and the quality of control samples. The identified compounds were classified and mapped to pathways using the KEGG, HMDB, and lipid-maps data-bases. A t-test was used to determine the significance for each compound. The R language package "ropls" (version 3.16, Metabo-HUB, Paris, France) was used to perform orthogonal partial least squares discriminant analysis (OPLS-DA) modeling, including 200 permutation tests to verify the model's reliability [39]. The VIP (Variable Importance in Projection) value of the model was calculated using multiple cross-validation. Differentially expressed metabolites (DEMS) were screened using a combination of P-value and the VIP value from the OPLS-DA model, with the criteria being a P-value < 0.05 and VIP > 1.

2.5. Determination of endogenous phytohormone contents

Using the same sampling method as for the RNA extraction, approximately 2 g of small particles of 10 DAP grains from NILS were collected and ground into a fine powder in liquid nitrogen, and stored at -80 °C. Phytohormone contents of were determined by MetWare (Wuhan) using the AB Sciex QTRAP 6500 LC-MS/MS platform (UPLC, ExionLC AD, https://sciex.com.cn/; MS, Applied Biosystems 6500; Triple Quadrupole, https://sciex.com.cn/). Three biological replicates were conducted for each sample.

2.6. Exogenous application of hormones

Seeds from NIL and NILH were germinated in square pots filled with vermiculite soil and placed in a greenhouse. They were kept at 26 °C with a 16-h white light/8-h darkness cycle for 5 d, after which the plants were uniformly sprayed with one of the following hormones: MeJA (100 µmol L*¹, 0.1% Tween-20), ΑΒΑ (10 µmol L*1, 0.1% Tween-20), or NAA (1 µmol L1, 0.1% Tween-20), or 0.1% Tween-20 as a control. Plant heights were measured at 5- and 10-d post treatment; and leaf water loss rates, above-ground biomass, and root length were measured 10 d after treatment.

2.7. Primers

All the primers used in this study are listed in Table S2.

3. Results

3.1. The qGWC1L allele reduces GWC

A semi-dominant QTL, named qGWC1, affecting GWC was previously identified on chromosome 1. The allele from inbred line SN80007, qGWC1L , exhibited reduced GWC compared to the allele qGWC1H from the inbred line SN80044 [10]. Following this discovery, we proceeded with the fine-mapping of qGWC1 using a sequential recombinant-derived progeny strategy [40]. Analysis of ten BC1F6 populations, comprising 703 individuals, we narrowed the qGWC1 interval to a 4.58 Mb (Fig. 1A). Nineteen recombinant individuals in these BC1F6 populations were self-pollinated to generate BC1F7 populations, totaling 1890 individuals. This enabled further refinement of the qGWC1 interval to 3.36 Mb (Fig. 1B). We calculated the average GWC for each of the three genotypes, qGWC1H/qGWC1H, qGWC1H/qGWC1L , and qGWC1L /qGWC1L , in the segregating populations. Compared to the qGWC1H/qGWC1H genotype, the qGWC1L /qGWC1L genotype had a significant reduction in GWC of 1.51% in the BC1F6 population and a 1.44% reduction in the BC1F7 population. The qGWC1H/qGWC1L exhibited no significant difference in GWC from qGWC1H/qGWC1H (Fig. 1C). The qGWC1L allele thus demonstrated a recessive inheritance pattern in reducing GWC.

From BC1F7 recombinant genotypes I and II, we developed two pairs of NILs: NILL /NILH and NILL'/NILH'. These NILs, along with their parents (SN80007 and SN80044), were genotyped using the maize M6H60K chip. The genomic identity between NILL and NILH was 97.9%, and between NILL' and NILH' was 95.5%. The low-GWC NILs, NILL and NILL', shared the qGWC1L region but differed in a few places (Fig. S1A, B). The qGWC1L region shared between NILL and NILL', estimated at 3.14 Mb based on the Maize6H-60 K SNP array chip [41], encompassed 40 genes according to the B73 RefGen_v4 reference genome [42] (Fig. S1C; Table S1).

Field trials were conducted in four environments including the summers of 2019, 2020, and 2021 at Beijing, and winter of 2020 in Hainan province. At maturity, 45 DAP at Beijing, and 42 DAP at Hainan, we observed differences in GWC between the two NILs. For NILL , the GWC was 34.7% ± 1.8%, 32.8% ± 3.7%, 41.4% ± 2.8%, and 40.4% ± 1.8%, respectively; and, for NILH, the corresponding GWC values were 36.1% ± 2.0%, 36.2% ± 3.0%, 43.4% ± 2.3%, and 44.7% ± 1.9%. On average, NILL exhibited GWC reductions of 1.4%, 3.4%, 2.0%, and 4.3% compared to NILH across these four field trials. Additionally, at 66 DAP during summer 2021 at Beijing, the GWC was 19.5% ± 2.8% for NILL and 22.7% ± 3% for NILH, a difference of 3.2% (Fig. 1D). We also analyzed the GWC for the other NIL pair. Compared to NILH', NILL' decreased GWC by 1.2%, 3.8%, 4.3%, and 2.5% across the four field trails. In the post-harvest stage at 66 DAP in Beijing, NILL' displayed a significant reduction in GWC of 1.7% compared to NILH' (Fig. 1E).

Investigation of various agronomic traits in the NILs detected no significant differences in either ear-related traits (e.g., ear length, ear stalk length, ear diameter, ear shaft diameter, number of ear leaves, male spike length, number of bracts, and the angle between rod and stem) or kernel-related traits (e.g., kernel row number, kernel number per row, kernel length, and kernel width) (Fig. S2A–L). Interestingly, NILL flowered 0.7–1.5 d earlier and exhibited slight reductions in plant height, ear height, and tassel branch number compared to NILH (Fig. S2M–Q).

3.2. Dynamic changes in GWC in the NILs during grain development

To further investigate the impact of qGWC1 on GWC, we measured GWC in the NILs at 11 time points from 10 to 60 DAP, i.e., at five-day intervals. NILL consistently exhibited lower GWC than NILH (Fig. 2A). Apart from test points 25, 35, and 45 DAP, the 100-GWW of NILL was significantly lower than that of NILH (Fig. 2B) whereas there was no significant difference in 100-GDW between NILL and NILH across all test points (Fig. 2C). The 100GFW in NILL was consistently lower than in NILH across all test points, although many of these differences were not statistically significant (Fig. 2D). Furthermore, differences in GWC between the contrasting NILs remained stable during the early stage of grain development but widened significantly after 35 DAP. In contrast, the differences in GWW between the NILs increased slowly after 35 DAP (Fig. 2E). Both NILs exhibited a decreased GDR before 35 DAP and an increased GDR after 35 DAP. Notably, NILL had a relatively higher GDR than NILH during the latter grain-filling stage (Fig. 2F). These findings suggested that the accelerated decline in GWW in NILL was due to a faster rate of physical dehydration during the later grain-filling stage.

3.3. GWC is closely associated with GWW and fresh grain size

Since GWC is calculated by dividing GWW by the GFW (the sum of GWW and GDW) we aimed to clarify the relationships among GWW, GDW, and GWC. In our analysis of individual plants within the NILs, we observed a pronounced correlation between GWC and GWW (Fig. 3A). A similar pattern of correlation was evident in the data from 1635 BC1F12 individual plants (Fig. 3B). However, there was a weak correlation between GWC and GDW in the NIL population, and even a slightly negative correlation in the BC1F12 plants (Fig. 3C, D). These observations imply that the qGWC1-mediated GWC has a stronger association with water weight than with dry matter accumulation.

Given that qGWC1L reduced GWW well before reaching maximum grain size, we sought to determine whether qGWC1L could also affect the grain size. At 40 DAP, using the displaced volume method, we determined that NILL indeed had a smaller fresh grain size than NILH (Fig. 3E). Smaller grains have a larger specific surface area, allowing for easier water escape from the grain to the surrounding environment. This may also account for the widening discrepancy in GWW between the two NILs during the later grainfilling stages (Fig. 2B).

Derived from the aforementioned results, our model suggests that the qGWC1L allele primarily reduces both GWC and grain size by decreasing water content, while not affecting dry matter accumulation. The reduced grain size in NILL accelerates water loss, which further amplifies the GWW difference between the NILs. This increased GWW difference led to a more widened GWC difference during the later grain-filling stage (Fig. 3F).

3.4. Reprogramming of the transcriptome between the NILs

To elucidate the molecular mechanism underlying the reduced GWC in NILL , we performed global transcript profiling; 18-grain samples from each of NILL and NILH at 10, 25, and 40 DAP were harvested. PCA analysis revealed a high degree of similarity among the biological replicates. However, significant discrepancies were observed across the three different stages (Fig. S3A); 614, 147, and 236 DEGs were identified at 10, 25, and 40 DAP, respectively. The low overlap of DEGs across the three stages suggests that qGWC1 regulates different genes at different stages to influence variation in GWCs (Fig. 4A). In the 10D (DAP)_L (NILL ) vs. H (NILH) comparison, about 80% of the DEGs were down-regulated (Fig. 4B). However, this trend was not observed at 25 and 40 DAP.

In the 10D_L vs. H comparison, GO analysis revealed a strong enrichment of DEGs in various categories, including "sequencespecific DNA binding", "calcium ion binding", and "nutrient reservoir activity (Fig. S4A). In the 25D_L vs. H comparison, the enriched categories for DEGs primarily included "nutrient reservoir activity", "multi-multicellular organism process", among others (Fig. S4B). In the 40D_L vs. H comparison, DEGs were mainly enriched in "heme binding", "nutrient reservoir activity", and others (Fig. S4C). DEGs associated with "nutrient reservoir activity" were enriched across all three stages. A KEGG enrichment analysis was performed to gain further insights into the roles of qGWC1regulated DEGs. This showed a significant enrichment of DEGs in the 10D_L vs. H comparison in various pathways, including "plant hormone signal transduction", "plant-pathogen interaction", and "MAPK signaling pathway-plant (Fig. S5A). In the 25D_L vs. H comparison, there was enrichment in "sulfur metabolism", "benzoxazinoid biosynthesis", among others (Fig. S5B). The 40D_L vs. H comparison highlighted enrichments in "phenylpropanoid biosynthesis" and "flavonoid biosynthesis", among others (Fig. S5C).

Our findings revealed that the majority of genes involved in hormone signaling pathways, especially those related to JA, ABA, indole-3-acetic acid (IAA), and ethylene (EH), were downregulated during the early stage but were considerably upregulated at the late-grain filling stages. Exceptions to this trend were JAZ8, GH3.1 and GH3.8 and bzip7 (Fig. 4C). A similar pattern was evident for genes associated with hormone biosynthesis, with notable exceptions being Tar1 and Tar3 in tryptophan metabolism (Fig. 4D). These findings suggest that qGWC1 may influence GWC by regulating hormonal signaling and biosynthesis.

In the 10D_L vs. H comparison, DEGs encoding proteins involved in desiccation, such as the LEA group 3 and Dehydrins 1/15, were down-regulated; however, all these DEGs were upregulated in the 25D_L vs. H and 40D_L vs. H comparisons (Fig. 4E). In contrast, DEGs encoding aquaporin TIP3.1, which facilitates the transport of water and small neutral solutes across cell membranes [27], were up-regulated in the 10D_L vs. H comparison but down-regulated in the 40D_L vs. H comparison (Fig. 4E). These expression patterns may partly contribute to the reduced GWC in NILL during the early stage of grain development. Furthermore, expansin-related genes, known for their role in cell-wall relaxation, cell expansion, and water accumulation [43,44], were down-regulated in the 10D_L vs. H comparison. There was no significant change or only slight increases in later two comparisons (Fig. 4E). This may partly explain the smaller fresh grain size observed in the NILL line.

In other pathways, DEGs encoding sucrose synthases (Sh1, Sus1, Sus2, and Sus3) were up-regulated in the 10D_L vs. H comparison, showed no significant changes in the 25D_L vs. H comparison, and were significantly down-regulated in the 40D_L vs. H comparison (Fig. 4F). A similar pattern was observed for DEGs encoding pyruvate orthophosphate dikinase (PPDK) 1/2, crucial components of the starch biosynthetic enzyme complex (Fig. 4F). The upregulation of these genes during the early grain-filling phase in NILL might have caused an increase in starch content compared to NILH, potentially increasing the osmotic potential of the grains, which in turn could result in reduced water influx [45]. Zeins, as hydrophobic proteins, are the primary storage proteins in the endosperm. Numerous zein genes, including those encoding 22KD alpha zein, 19-KD alpha zein, 50-kD gamma zein, 27-kD gamma zein, 16-kD gamma zein, Zein-alpha 19B1, Z1A alpha zein protein, Zein protein3, Floury2, Zein-alpha ZA1/M1, and Zein-alpha 19D1, exhibited similar expression patterns across the three comparisons. That is, they were up-regulated in the 10D_L vs. H comparison, showed no significant changes in the 25D_L vs. H comparison, and were significantly down-regulated in the 40D_L vs. H comparison (Fig. 4G). Similarly, upregulation of these zein genes during the early grain-filling phase might have caused an increase in zein proteins in NILL , potentially leading to reduced GWC.

3.5. Proteome and metabolome reprogramming between the two NILs

We extracted total proteins from the same grain samples used in the earlier RNA-seq analysis and assessed these proteins and their expression changes using iTRAQ-based proteomics with three biological replicates. PCA analysis was employed to evaluate protein expression levels for each replicate (Fig. S6A). When the relative expression changes of all 98 identified DEPs across the three time points were visualized in a Venn diagram, with no DEP overlapping across all three time points (Fig. S6B). KEGG annotation of the DEPs categorized them by pathway type, revealing that "metabolic pathway" and "biosynthesis of secondary metabolites" had the highest number of DEPs at all time points (Fig. S6C–E). Of note, DEPs OPR5 and OPR7 were associated with JA biosynthesis (Fig. S6F). Eight genes that simultaneously displayed both DEGs and DEPs are prime candidates for further investigation (Fig. S6G–I). Although they are not located within the qGWC1 interval, they may be important genes regulated by the causative gene in qGWC1 interval. In particular, OPR5 may affect the balance of hormone signaling by regulating JA content, and ultimately regulate GWC.

Grain samples from the previously described RNA-seq study were subjected to metabolome analysis. Using UHPLC-QTOF-MS [46], we detected 1722 metabolites. The metabolite expression levels for each replicate were assessed by PCA (Fig. S7A). The Venn plots illustrated the relative changes of all DEMs, with no DEMs overlapping across the three time points (Fig. S7B). We examined metabolites potentially involved in regulating osmotic pressure. Some metabolites, such as ThrPheArgAla and sucrose6phosphate, were down-regulated in the 10D_L vs. H comparison and progressively up-regulated from the 25D_L vs. H to 40D_L vs. H comparisons (Fig. S7C). Others, including Thr-Phe-Phe-Trp, DMannitol, and D-Mannitol 1-phosphate, were markedly upregulated in the early low-GWC grains. Specifically, D-Mannitol consistently showed up-regulation, whereas the other two metabolites demonstrated significant down-regulation by the 40D_L vs. H comparison (Fig. S7C). Many DEMs were identified as related to the hormone biosynthesis and signaling pathways (Fig. S7D).

3.6. Orchestration of phytohormones plays a key role in balancing water and dry matter accumulation in maize grains

The GWC differences between the NILs were evident from the early grain development stage, emphasizing the importance of early-stage DEGs enriched in the plant hormone pathway (Fig. S5A). Moreover, analyses of both the proteome and metabolome revealed several DEPs and DEMs related to hormone biosynthesis and signaling pathways (Figs. S6C–F, S7D). These findings suggest that the qGWC1-mediated reduction of GWC may be intricately linked to the biosynthesis and signaling pathways of plant hormones.

To investigate potential crosstalk among hormones that regulate GWC at the early stage, we first quantified the contents of 46 hormones and their derivatives using liquid chromatography tandem mass spectrometry (LC-MS/MS). In low-GWC NILL grains, the levels of JA and its active isoforms, JA-Ile and JA-Val, were dramatically decreased, accompanied by a reduction in OPC-4 precursor (Fig. 5A). Consistent with the reduced levels of JA, the transcript levels of ZmLOX6 and ZmOPR6, which encode enzymes involved in JA biosynthesis, were significantly reduced in NILL grains (Fig. 4D), suggesting that qGWC1 affects JA biosynthesis. Likewise, genes involved in the JA signaling pathways (JAZ6, JAZ1, JAZ21, and MYC7) were strongly downregulated in NILL grains (Fig. 4C). There was a significant reduction in transcript levels of ZmCKX10 in NILL grains. This gene encodes an enzyme responsible for irreversible degradation of cytokinins [47,48] (Fig. 4D). Most cytokinins and their derivatives, including cis-zeatin (cZ), trans-zeatin (tZ), kinetin (K), cis-zeatin riboside (cZR), trans-zeatin riboside (tZR), transzeatin-O-glucoside (tZOG), N6-isopentenyladenosine (IPR), and N6-isopentenyl-adenine-7-glucoside (IP7G), 4-[[(9-beta-D-gluco pyranosyl9Hpurin6yl)amino]methyl]phenol (pT9G), 2methylthio-cis-zeatin riboside (2MeScZR), were more abundant in NILL grains. However, the levels of isopentenyladenine (IP) and dihydrozeatin ribonucleoside (DHZR) were significantly lower in NILL compared to NILH grains (Fig. 5B). Levels of other cytokinin metabolites did not differ between the two lines (Table S3).

In immature NILL grains, storage forms of IAA, including IAAglutamic acid (IAA-Glu), IAA-aspartic acid (IAA-ASP), and IAAvaline methyl ester (IAA-Val-Me) [49], along with the IAA precursor 3-Indolepropionic acid (IPA), accumulated in higher quantities, whereas the inactive form OxIAA remained at a low level (Fig. 5C). Regarding the two IAA precursors, the quantity of IPA was about 10 times more than that of 3-indoleacetonitrile (IAN). Moreover, the IPA level was higher in NILL than in NILH, whereas the IAN level was lower in NILL than in NILH (Fig. 5C). Consistent with the increased level of IPA, the auxin biosynthesis gene Tar1, responsible for introducing additional L-tryptophan into the IPA pathway to produce IAA [50], was upregulated in NILL grains (Fig. 4D).

ABA-glucosyl ester (ABA-GE), serving as a storage and transport form of ABA, accumulated at high levels in NILL grains, but was almost undetectable in NILH grains. However, the level of ABA itself did not differ significantly (Table S3). Given that ABA biosynthesis primarily occurs in phloem companion cells [51], ABA must first be loaded into the leaf phloem and then unloaded into the caryopses [52]. Corresponding to the increased levels of ABA-GE, the transcript levels of four PP2Cs (negative regulators of ABA signaling) were significantly reduced in NILL grains (Fig. 4C). Additionally, gibberellin A9 (GA9), a precursor of the bioactive gibberellins of GA1/GA4, was up-regulated in NILL grains, whereas other hormones and derivatives, such as GA20, ethylene, salicylic acid, and unicornolide, showed no significant differences (Table S3). The up-regulation of auxins, cytokinins, and ABA-GE hormones and derivatives may enhance water-use efficiency during the grain filling phase, leading to lower water accumulation in NILL without affecting the rate of grain filling. These findings suggest that qGWC1 plays a pivotal role in coordinating the crosstalk among various plant hormones. Thus, orchestration of plant hormones could balance water content and dry matter accumulation in maize grains.

3.7. Responses of the NILs to phytohormone treatments

Exogenous applications of MeJA, ABA, and naphthaleneacetic acid (NAA) indicated that MeJA-induced inhibition of seedling growth was more pronounced in NILL plants than in NILH plants (Fig. 6A–C). This suggested a heightened response of qGWC1L to MeJA, implying a potential direct role for qGWC1L in JA biosynthesis or JA signaling pathway. However, the exact relationship between JA signaling and variations in GWC remains uncertain.

NILL plants exhibited a higher rate of leaf water loss compared to NILH. After ABA treatment, the difference in rates of leaf water loss between NILH and NILL slightly widened compared to the control (Fig. 6D). These results further confirmed the close relationship between endogenous ABA content and GWC, possibly explaining the lower water accumulation and faster water loss in low-GWC grains.

When treated with NAA or water, NILL showed no significant differences in above-ground biomass and root length compared to NILH (Fig. 6E, F) suggesting that qGWC1 may be indirectly involved in auxin biosynthesis and signaling through phytohormone signaling crosstalk.

4. Discussion

The qGWC1L allele reduces the rate of water accumulation, leading to reduced water retention throughout the growth period and ultimately lowering GWC. Given that grain dry weight inversely affects GWC, we hypothesize the existence of QTL capable of reducing GWC by increasing dry matter accumulation. To identify such QTL, we conducted phenotyping studies focusing on both GWC and GDW. This strategy enables us to select QTL that have a positive impact on yield while also reducing GWC. Given that GWC is a composite trait, the use of grain dry weight and water weight as auxiliary phenotypes in QTL mapping for GWC will enable more accurate and efficient cloning and analysis of genes related to GWC.

As a complex trait, the change of GWC may involve multiple pathways at different stages of grain development. Our results indicate that gene expression was active during the early stage of grain development, but was inactive at later stages (Fig. S3B). The DEGs in early grain development were mainly enriched in the plant hormone signaling pathway (Fig. S5A). These differential signals were transmitted downward, regulating the grain development mode and some metabolic pathways, finally causing differences in GWC. During the late stage of grain development, DEGs were primarily enriched in the phenylpropanoid and flavonoid biosynthesis pathways (Fig. S5C). Expression of genes related to flavonoid synthesis was significantly up-regulated in the low GWC NILs, and may contribute to the dehydration tolerance of low-GWC grains. Genes from different metabolic pathways showed inconsistent trends across the three stages. For example, a DEG encoding proteins involved in desiccation response was down-regulated in the early stage but up-regulated in the later stages (Fig. 4E). In contrast, genes related to starch and protein synthesis were up-regulated at the early stage but down-regulated in the later stages (Fig. 4F, G). This suggests that qGWC1L activates nutrient reservoir activity genes in the early stages to facilitate water accumulation and initiate grain filling. Later in the grain filling process, the expressions of stress-responsive genes, such as those related to water deficit tolerance, increase to facilitate faster grain dehydration. Although the expression pattern of the active genes in the nutrient reservoir activity was altered, it may not affect the eventual accumulation of dry matter. This is consistent with the result that qGWC1 regulates GWC without affecting dry matter accumulation (Figs. 1D, 2C).

Our findings indicate that qGWC1 plays a pivotal role in modulating GWC by affecting both early water accumulation and later water loss. qGWC1 also regulated the balance among various hormones, such as JA, auxin, CK, and ABA. For instance, in early lowGWC grains, we observed a decrease in JA content, accompanied by increases in auxins, cytokinins, and ABA-GE levels. Our results demonstrated that qGWC1L enhanced the response to MeJAinduced inhibition of seedling growth. While leaf water loss rate in NILH was insensitive to exogenous ABA treatment, it increased in NILL in response to ABA treatment. Both NILH and NILL exhibited significant increases in leaf water loss when treated with JA (Fig. S8). These findings are valuable for exploring the crosstalk between plant hormones and their impact on changes in GWC.

Combining the hormone content with multi-omics reprogramming in the NILs, we believe that the decrease in JA content may be a contributing factor to low-GWC grains. ABA, as a key hormone affecting grain ripening and dehydration may affect downstream genes responsible for grain maturation [53]. Overall, these changes would affect grain composition and hydrophilicity, finally influencing water accumulation and dissipation. By identifying and modulating these key components, we could potentially exert significant control over GWC. Our study shows that qGWC1 fine-tunes GWC through a nuanced hormone signaling network, thereby ensuring a balance between water and dry matter accumulation in maize grains. Low GWC limits initial grain size, thereby limiting the ability to hold water while not reducing dry matter accumulation. This suggests that hormonal crosstalk improves water use efficiency during the grain filling process, allowing normal dry matter accumulation even in the context of limited water supply. This effect was evident in the immature low GWC grains, where genes related to dry matter accumulation showed higher expressions in NILL compared to NILH. Understanding this mechanism could enable decoupling of early-stage water retention from water-use efficiency during the filling period. This would allow for more dry matter accumulation when water is available, thus increasing overall yield.

The parental line SN80007 originates from tropical germplasm, whereas the other parent, SN80044, is derived from temperate germplasm [10]. The ability of qGWC1L to reduce GWC is attributed to the natural genetic variation inherent in SN80007. Given that global maize production relies primarily on germplasm derived from temperate US germplasm sources [54], leveraging the advantageous natural allele present in the tropical SN80007 germplasm offers a viable strategy for improving existing germplasm. Throughout the history of maize breeding, both natural selection and artificial domestication played roles in the positive selection of beneficial genetic variants and the elimination of deleterious ones. Phenotypic variations arising from natural variation often confer advantages in performance and adaptability [55]. Therefore, utilizing natural variation as the genetic basis to analyze the regulatory network controlling GWC is crucial for genetic improvement of varieties with low GWC. In the current study, we found that NILL reduced GWC at maturity by 3.11%, without incurring any yield penalty (Figs. 1D, 2C). This makes NILL an exceptional donor to introduce qGWC1 into Zheng 58 and Chang7-2, the parental lines of Zhengdan 958, the most widely cultivated maize hybrid in China. We predict that an improved version of Zhengdan 958, featuring reduced GWC, will be more amenable to mechanical harvesting [3].

CRediT authorship contribution statement

Yuanliang Liu: Methodology, Investigation, Data curation, Formal analysis, Writing – original draft. Manman Li: Investigation. Jianju Liu: Methodology, Investigation. Suining Deng: Methodology. Yan Zhang: Resources. Yuanfeng Xia: Resources. Baoshen Liu: Resources, Supervision, Project administration. Mingliang Xu: Conceptualization, Project administration, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

We thank Xu laboratory members for helpful discussions. We thank the Center for Crop Functional Genomics and Molecular Breeding of China Agricultural University for providing transgenic technology support. This research was supported by the Jiangsu province Seed Industry Revitalization project [JBGS(2021)002] and Beijing Germplasm Creation and Variety Selection and Breeding Joint Project [NY2023-180].

Appendix A. Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2024.05.009.

Sidebar

ARTICLE INFO

Article history:

Received 25 February 2024

Revised 12 April 2024

Accepted 14 May 2024

Available online 7 June 2024

Footnote

* Corresponding authors.

E-mail addresses: [email protected] (B. Liu), [email protected] (M. Xu).

References

References

[1] T. Cui, C.L. Fan, D.X. Zhang, L. Yang, Y.B. Li, H.H. Zhao, Research progress of maize mechanized harvesting technology, Trans. Chin. Soc. Agric. Mach. 50 (2019) 1–13 (in Chinese with English abstract).

[2] K.R. Wang, S.K. Li, Analysis of influencing factors on kernel dry-down-rate of maize hybrids, Sci. Agric. Sin. 50 (2017) 2027–2035 (in Chinese with English abstract).

[3] R.Z. Xie, X.P. Lei, K.R. Wang, Y.Q. Guo, Z.W. Cai, P. Hou, S.K. L, Research on corn mechanically harvesting grain quality in Huanghuaihai plain, Crops 2 (2014) 4.

[4] G. Zhou, Q. Zhu, Y. Mao, G. Chen, L. Xue, H. Lu, M. Shi, Z. Zhang, X. Song, H. Zhang, D. Hao, Multi-locus genome-wide association study and genomic selection of kernel moisture content at the harvest stage in maize, Front. Plant Sci. 12 (2021) 697688.

[5] J. Zhang, F.Q. Zhang, B.J. Tang, Y. Ding, L.K. Xia, J.S. Qi, X.Y. Mu, L.M. Gu, D.W. Lu, Y.H. Chen, Molecular mapping of quantitative trait loci for grain moisture at harvest and field grain drying rate in maize (Zea mays L.), Physiol. Plant. 169 (2020) 64–72.

[6] Z. Wang, X. Wang, L. Zhang, X. Liu, H. Di, T. Li, X. Jin, QTL underlying field grain drying rate after physiological maturity in maize (Zea mays L.), Euphytica 185 (2012) 521–528.

[7] R.G. Sala, F.H. Andrade, E.L. Camadro, J.C. Cerono, Quantitative trait loci for grain moisture at harvest and field grain drying rate in maize (Zea mays L.), Theor. Appl. Genet. 112 (2006) 462-471.

[8] J.Z. Qu, S.T. Xu, X.A. Gou, H. Zhang, Q. Cheng, X.Y. Wang, C. Ma, J.Q. Xue, Timeresolved multiomics analysis of the genetic regulation of maize kernel moisture, Crop J. 11 (2023) 247–257.

[9] R. Mihaljevic, C.C. Schön, H.F. Utz, A.E. Melchinger, Correlations and QTL correspondence between line per Se and testcross performance for agronomic traits in four populations of European maize, Crop Sci. 45 (2005) 114–122.

[10] J. Liu, H. Yu, Y. Liu, S. Deng, Q. Liu, B. Liu, M. Xu, Genetic dissection of grain water content and dehydration rate related to mechanical harvest in maize, BMC Plant Biol. 20 (2020) 118.

[11] Y.L. Li, Y. Dong, M. Yang, Q. Wang, Q. Shi, Q. Zhou, F. Deng, Z. Ma, D. Qiao, H. Xu, Physiology, QTL detection for grain water relations and genetic correlations with grain matter accumulation at four stages after pollination in maize, J. Plant Biochem. Physiol. 2 (2014) 1–9.

[12] W.Q. Li, Y.H. Yu, L.X. Wang, Y. Luo, Y. Peng, Y.C. Xu, X.G. Liu, S.S. Wu, L.M. Jian, J. T. Xu, Y.J. Xiao, J.B. Yan, The genetic architecture of the dynamic changes in grain moisture in maize, Plant Biotechnol. J. 19 (2021) 1195–1205.

[13] S. Li, C. Zhang, D. Yang, M. Lu, Y. Qian, F. Jin, X. Liu, Y. Wang, W. Liu, X. Li, Detection of QTNs for kernel moisture concentration and kernel dehydration rate before physiological maturity in maize using multi-locus GWAS, Sci. Rep. 11 (2021) 1764.

[14] Y. Dong, Z.Q. Feng, F. Ye, T. Li, G.L. Li, Z.S. Li, Y.C. Hao, X.H. Zhang, W.X. Liu, J.Q. Xue, S.T. Xu, Genome-wide association analysis for grain moisture content and dehydration rate on maize hybrids, Mol. Breed. 43 (2023) 5.

[15] V. Capelle, C. Remoué, L. Moreau, A. Reyss, A. Mahé, A. Massonneau, M. Falque, A. Charcosset, C. Thévenot, P. Rogowsky, S. Coursol, J.L. Prioul, QTLs and candidate genes for desiccation and abscisic acid content in maize kernels, BMC Plant Biol. 10 (2010) 22.

[16] D.F. Austin, M. Lee, L.R. Veldboom, A.R. Hallauer, Genetic mapping in maize with hybrid progeny across testers and generations: Grain yield and grain moisture, Crop Sci. 40 (2000) 30–39.

[17] J. Bewley, M. Black, Seeds: Physiology of Development and Germination, Springer, New York, NY, USA, 1994.

[18] R.J. Jones, B.M.N. Schreiber, J.A. Roessler, Kernel sink capacity in maize: genotypic and maternal regulation, Crop Sci. 36 (1996) 301–306.

[19] M.E. Westgate, J.S. Boyer, Water status of the developing grain of maize, Agron. J. 78 (1986) 714–719.

[20] D.B. Egli, D.M. TeKrony, Species differences in seed water status during seed maturation and germination, Seed Sci. Res. 7 (1997) 3–11.

[21] B.L. Gambín, L. Borrás, M.E. Otegui, Kernel weight dependence upon plant growth at different grain-filling stages in maize and sorghum, Aust. J. Agric. Res. 59 (2008) 280–290.

[22] H. Huang, B. Liu, L. Liu, S. Song, Jasmonate action in plant growth and development, J. Exp. Bot. 68 (2017) 1349–1359.

[23] J. Wang, L. Song, X. Gong, J. Xu, M. Li, Functions of jasmonic acid in plant regulation and response to abiotic stress, Int. J. Mol. Sci. 21 (2020) 1446.

[24] Y.X. Ge, L.J. Zhang, F.H. Li, Z.B. Chen, C. Wang, Y.C. Yao, Z.H. Han, J. Zhang, Z.S. Shi, Relationship between jasmonic acid accumulation and senescence in drought-stress, Afr. J. Agric. Res. 5 (2010) 1978–1983.

[25] H.I. Mohamed, H.H. Latif, Improvement of drought tolerance of soybean plants by using methyl jasmonate, Physiol. Mol. Biol. Plants 23 (2017) 545–556.

[26] J.K. Zhu, Salt and drought stress signal transduction in plants, Annu. Rev. Plant Biol. 53 (2002) 247–273.

[27] F. Chaumont, F. Barrieu, E. Wojcik, M.J. Chrispeels, R. Jung, Aquaporins constitute a large and highly divergent protein family in maize, Plant Physiol. 125 (2001) 1206–1215.

[28] A. Grondin, O. Rodrigues, L. Verdoucq, S. Merlot, N. Leonhardt, C. Maurel, Aquaporins contribute to ABAtriggered stomatal closure through OST1mediated phosphorylation, Plant Cell 27 (2015) 1945–1954.

[29] K. Prado, V. Cotelle, G. Li, J. Bellati, N. Tang, C. Tournaire-Roux, A. Martinière, V. Santoni, C. Maurel, Oscillating aquaporin phosphorylation and 14-3-3 proteins mediate the circadian regulation of leaf hydraulics, Plant Cell 31 (2019) 417429.

[30] B. Buckner, P.S. Miguel, D. Janick-Buckner, J.L. Bennetzen, The y1 gene of maize codes for phytoene synthase, Genetics 143 (1996) 479–488.

[31] M.S. Kang, M.S. Zuber, Combining ability for grain moisture, husk moisture, and maturity in maize with yellow and white endosperms, Crop Sci. 29 (1989) 689–692. [32] J. Ingram, D. Bartels, The molecular basis of dehydration tolerance in plants, Annu. Rev. Plant Physiol. Plant Mol. Biol. 47 (1996) 377–403.

[33] P.E. Verslues, M. Agarwal, S. Katiyar-Agarwal, J. Zhu, J.K. Zhu, Methods and concepts in quantifying resistance to drought, salt and freezing, abiotic stresses that affect plant water status, Plant J. 45 (2006) 523–539.

[34] M.I. Love, W. Huber, S. Anders, Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2, Genome Biol. 15 (2014) 550.

[35] M. Ashburner, C.A. Ball, J.A. Blake, D. Botstein, H. Butler, J.M. Cherry, A.P. Davis, K. Dolinski, S.S. Dwight, J.T. Eppig, M.A. Harris, D.P. Hill, L. Issel-Tarver, A. Kasarskis, S. Lewis, J.C. Matese, J.E. Richardson, M. Ringwald, G.M. Rubin, G. Sherlock, Gene ontology: tool for the unification of biology, the gene ontology consortium, Nat. Genet. 25 (2000) 25–29.

[36] M. Kanehisa, S. Goto, KEGG: Kyoto Encyclopedia of Genes and Genomes, Nucleic Acids Res. 28 (2000) 27–30.

[37] M.M. Bradford, A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding, Anal. Biochem. 72 (1976) 248–254.

[38] M. Sugimoto, M. Kawakami, M. Robert, T. Soga, M. Tomita, Bioinformatics tools for mass spectroscopy-based metabolomic data processing and analysis, Curr. Bioinform. 7 (2012) 96–108.

[39] J. Boccard, D.N. Rutledge, A consensus orthogonal partial least squares discriminant analysis (OPLS-DA) strategy for multiblock Omics data fusion, Anal. Chim. Acta 769 (2013) 30–39.

[40] Q. Yang, D. Zhang, M. Xu, A sequential quantitative trait locus fine-mapping strategy using recombinant-derived progeny, J. Integr. Plant Biol. 54 (2012) 228–237.

[41] H. Tian, Y. Yang, H. Yi, L. Xu, H. He, Y. Fan, L. Wang, J. Ge, Y. Liu, F. Wang, J. Zhao, New resources for genetic studies in maize (Zea mays L.): a genome-wide Maize6H-60K single nucleotide polymorphism array and its application, Plant J. 105 (2021) 1113–1122.

[42] Y. Jiao, P. Peluso, J. Shi, T. Liang, M.C. Stitzer, B. Wang, M.S. Campbell, J.C. Stein, X. Wei, C.S. Chin, K. Guill, M. Regulski, S. Kumari, A. Olson, J. Gent, K.L. Schneider, T.K. Wolfgruber, M.R. May, N.M. Springer, E. Antoniou, W.R. McCombie, G.G. Presting, M. McMullen, J. Ross-Ibarra, R.K. Dawe, A. Hastie, D.R. Rank, D. Ware, Improved maize reference genome with single-molecule technologies, Nature 546 (2017) 524–527.

[43] X.C. Lizana, R. Riegel, L.D. Gomez, J. Herrera, A. Isla, S.J. McQueen-Mason, D.F. Calderini, Expansins expression is associated with grain size dynamics in wheat (Triticum aestivum L.), J. Exp. Bot. 61 (2010) 1147–1157.

[44] J. Sampedro, D.J. Cosgrove, The expansin superfamily, Genome Biol. 6 (2005) 242.

[45] K. Koch, Sucrose metabolism: regulatory mechanisms and pivotal roles in sugar sensing and plant development, Curr. Opin. Plant Biol. 7 (2004) 235–246.

[46] J. Wang, T. Zhang, X. Shen, J. Liu, D. Zhao, Y. Sun, L. Wang, Y. Liu, X. Gong, Y. Liu, Z.J. Zhu, F. Xue, Serum metabolomics for early diagnosis of esophageal squamous cell carcinoma by UHPLC-QTOF/MS, Metabolomics 12 (2016) 116.

[47] T. Kakimoto, Identification of plant cytokinin biosynthetic enzymes as dimethylallyl diphosphate:ATP/ADP isopentenyltransferases, Plant Cell Physiol. 42 (2001) 677–685.

[48] K. Miyawaki, P. Tarkowski, M. Matsumoto-Kitano, T. Kato, S. Sato, D. Tarkowska, S. Tabata, G. Sandberg, T. Kakimoto, Roles of Arabidopsis ATP/ ADP isopentenyltransferases and tRNA isopentenyltransferases in cytokinin biosynthesis, Proc. Natl. Acad. Sci. U. S. A. 103 (2006) 16598–16603.

[49] K.I. Hayashi, K. Arai, Y. Aoi, Y. Tanaka, H. Hira, R. Guo, Y. Hu, C. Ge, Y. Zhao, H. Kasahara, K. Fukui, The main oxidative inactivation pathway of the plant hormone auxin, Nat. Commun. 12 (2021) 6752.

[50] A.N. Stepanova, J. Robertson-Hoyt, J. Yun, L.M. Benavente, D.Y. Xie, K. Dolezal, A. Schlereth, G. Jürgens, J.M. Alonso, TAA1-mediated auxin biosynthesis is essential for hormone crosstalk and plant development, Cell 133 (2008) 177191.

[51] T. Kuromori, E. Sugimoto, K. Shinozaki, Intertissue signal transfer of abscisic acid from vascular cells to guard cells, Plant Physiol. 164 (2014) 1587–1592.

[52] P. Qin, G. Zhang, B. Hu, J. Wu, W. Chen, Z. Ren, Y. Liu, J. Xie, H. Yuan, B. Tu, B. Ma, Y. Wang, L. Ye, L. Li, C. Xiang, S. Li, Leaf-derived ABA regulates rice seed development via a transporter-mediated and temperature-sensitive mechanism, Sci. Adv. 7 (2021) eabc8873.

[53] K. Chen, G.J. Li, R.A. Bressan, C.P. Song, J.K. Zhu, Y. Zhao, Abscisic acid dynamics, signaling, and functions in plants, J. Integr. Plant Biol. 62 (2020) 25–54.

[54] J.S. Smith, W. Trevisan, A. McCunn, W.E. Huffman, Global dependence on Corn Belt Dent maize germplasm: challenges and opportunities, Crop Sci. 62 (2022) 2039–2066.

[55] Y. Liang, H.J. Liu, J. Yan, F. Tian, Natural variation in crops: realized understanding, continuing promise, Annu. Rev. Plant Biol. 72 (2021) 357–385.