Integrated physiological and transcriptomic measurements reveal changes in the accumulation and partitioning of storage reserves after prevention of pollination in maize Authors: Lei Tian, Pei Li, Yangguo Liu, Shanshan Liu, Jun Zhang, Mengzhen Xu, Shijin Xu, Mingna Li, Zhenfei Zhang, Wenheng Dong, Jiazhen Mu, Xia Liu, Yanhui Chen, Lixia Ku, Liancheng Wu Published: 2026 ABSTRACT AbstractBackgroundReduced sink demand, such as that caused by grain or ear removal, is known to accelerate senescence in maize; nonetheless, how diminished sink strength mechanistically triggers senescence and disrupts grain development remains poorly understood. In particular, the transcriptional and metabolic regulatory networks that link reduced sink demand to the onset of senescence have not been clearly defined.ResultsWe grew the maize inbred line 08LF, which exhibits early senescence in response to grain removal, under non-pollinated (NONPOL) and normally pollinated (POL) conditions. We then collected phenotypic, physiological, and transcriptomic data from developing grains (6 to 24 days after silking) to gain insight into the molecular mechanisms responsible for the changes induced under NONPOL conditions. Pollination prevention induced early senescence, and major transcriptomic divergence between POL and NONPOL grains occurred between 14 and 18 DAS, indicating a key time window associated with the onset of senescence under reduced sink demand. Key pathways involved in starch, sucrose, and trehalose metabolism were disrupted in nonpollinated grains, and gene co-expression network analysis identified five co-expressed modules linked to energy metabolism, photosynthesis, and stress responses. These modules were correlated with changes in sugar accumulation and oxidative stress indices. Importantly, 20 hub genes, including genes encoding MYB transcription factors and heat shock proteins, were identified as potential regulators of the senescence process triggered by reduced sink demand. These genes may represent key components for dissecting the regulatory mechanisms that underlie senescence triggered by reduced sink demand in maize.ConclusionsThis study links reduced sink demand to coordinated metabolic and transcriptional changes associated with senescence, providing novel insights and genetic resources for understanding maize grain development under stress.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12870-026-09270-7.Keywords: Maize, Senescence, Grain development, Pollination prevention, Transcriptomics Introduction Grain development is a complex process regulated by an intricate transcriptional regulatory network that begins with fertilization and concludes with the production of the mature grain body. A comprehensive understanding of the molecular mechanisms that regulate various aspects of grain development is needed to facilitate the development of new cereal crop varieties. However, although the biological pathways and molecular networks involved in seed development have been studied extensively in Arabidopsis (Arabidopsis thaliana; [1–5]), less is known about these molecular mechanisms in crop plants. Maize (Zea mays L.) is among the most important sources of food, feed, and biofuel materials [6], serving as an excellent model for crop research on grain development. Understanding spatial and temporal patterns of gene expression during seed development is helpful for the genetic improvement of this important crop. In recent years, a number of transcriptome profiling studies have characterized gene expression and cellular processes during grain development in maize [7–11]. Maize grain development involves multiple biological processes and signaling pathways, including photoassimilate supply, carbohydrate partitioning, and storage compound accumulation [9, 12–14], as well as hormone signal transduction and epigenetic imprinting [13, 15–17]. Recent gene cloning and functional analyses have demonstrated that maize kernel development is governed by multilayer regulatory networks. A substantial number of cloned kernel mutants carry mutations in pentatricopeptide repeat (PPR) proteins involved in organellar RNA processing, highlighting the importance of mitochondrial and plastid function for seed development [18, 19]. In addition, transcription factor networks centered on OPAQUE2 (O2) coordinate zein biosynthesis and starch metabolism during endosperm filling [20, 21]. More recent studies have revealed chromatin-associated regulators and multilayer transcriptional modules that fine-tune storage protein accumulation and endosperm development [22, 23]. Together, these findings indicate that kernel development is a tightly regulated, energy-demanding process, which requires coordinated control of gene expression, RNA processing, and assimilate allocation. Because grain is a major sink for photoassimilates, its loss or developmental failure strongly affects metabolic processes throughout the whole plant, including photosynthesis, carbohydrate metabolism, and source–sink relationships [24–26], ultimately affecting final maize yields. Grain removal by the experimental prevention of pollination after silking, achieved by controlled manual treatment, alters the source-to-sink ratio, increasing the contents of soluble sugars, starch, and nitrogen in leaves compared with those of normally pollinated plants and delaying or promoting the process of leaf senescence. Some maize inbred lines, such as B73 and HCL645, exhibit accelerated senescence upon grain removal compared with normally pollinated plants, whereas other cultivars, such as Mo17 and Chang 7 − 2, show no significant differences in senescence rate [27, 28]. Transcriptomic changes in B73 leaves under non-pollinated (NONPOL) conditions are similar to those observed in Arabidopsis under sugar-induced senescence, indicating that premature leaf senescence in B73 may be reduced by limiting excessive sugar accumulation. It has also been demonstrated that the efficiency of sugar partitioning and sugar-transport signaling are key factors in the initiation of leaf senescence when pollination is prevented [27, 29]. However, the genetic and molecular determinants of maize responses under NONPOL conditions, as well as the regulatory mechanisms and interaction networks involved in its leaf senescence, have not been fully characterized. In previous studies, we reported that the maize inbred line 08LF, which was derived from American germplasm with good lodging resistance, high yield potential, rapid grain dehydration rate, and suitability for mechanical harvesting, exhibits premature senescence when pollination is prevented [28]. This distinct and stable phenotype makes 08LF an appropriate model for dissecting the physiological and transcriptional basis of sink-mediated senescence. Here, we collected phenotypic, physiological, metabolite, and transcriptome data from developing grains (6 to 24 days after silking, DAS) of field-grown 08LF maize plants, with and without pollination prevention, to gain insight into the genetic and molecular mechanisms of grain development during natural senescence and senescence induced under NONPOL conditions. Materials and methods Plant materials, growing conditions, and sampling details The maize (Zea mays L. ssp. mays) inbred line 08LF was selected from an American germplasm as an elite inbred line with high combining ability, rapid water loss rate, and precocity; it is known to undergo early leaf senescence when pollination is prevented. During the summer of 2015, 08LF plants were grown in the field at the experimental farms of Henan Agricultural University, Zhengzhou, China. The plants were hand-planted in 4-m long rows with a row spacing of 66 cm and a plant spacing of 30 cm [28]. All ears of the experimental plants were covered with shoot bags before silking, then checked every day to record the dates of silking and pollination. Plants that silked on the same day were selected and tagged. Shoot bags were then removed from plants in alternating rows at 6 DAS, and these plants were artificially pollinated to serve as the normally pollinated (POL) treatment, whereas bags were retained on plants in the other rows to maintain the non-pollinated (NONPOL) treatment. For sampling, kernels from six ears collected at 9:00 am from six independent POL and NONPOL plants in each block were pooled to generate one biological replicate. In total, six biological replicates from the different plots (36 ears in total) were collected at 6, 10, 14, 18, 21, and 24 DAS. The 6 DAS time point represents the common pre-treatment stage before the divergence of NONPOL and POL treatments. Kernels were collected from the middle of each ear by manual dissection, and each replicate sample was prepared by mixing an equal number of kernels from six ears. Three biological replicates from each kernel developmental stage were frozen immediately in liquid nitrogen and used for RNA extraction and biochemical measurements, including measurements of antioxidant enzyme activities and metabolites. Another three samples from each stage were used for measurements of kernel weight (both fresh and dry) and water content. All measurements were performed using three independent biological replicates. Kernel weight, antioxidant enzyme activities, and metabolite contents Grains were weighed immediately after harvest (fresh weight) and after drying at 65 °C for 72 h (dry weight), and water content was calculated. A commercial assay kit was used to measure the starch content of the insoluble carbohydrate fraction following the manufacturer’s instructions (R-Bio-pharm, Darmstadt, Germany). A K-SUFRG kit was used to determine the sucrose content of the soluble carbohydrate fraction (Megazyme, Bray, Ireland) as described previously. Glucose content was measured by the glucose oxidase/peroxidase method using the K-GLUC kit (Megazyme) according to the manufacturer’s instructions with some modifications [30]. Trehalose (TRE) content was measured as described by [31]. Crude protein extraction, measurement of malondialdehyde (MDA) content, and assays for the activities of superoxide dismutase (SOD) and catalase (CAT) were performed as described previously. Measurements for all physiological indicators (starch, sucrose, glucose, trehalose, MDA, H2O2 contents, and SOD and CAT activities) were performed on three independent biological replicate samples. Three technical replicates were conducted for each sample. Total soluble protein was quantified using the Bradford method, and metabolite contents were normalized to protein concentration. All measurements were conducted using samples collected from multiple organs, including the blade of the ear leaf (ELB), the blade of the first leaf above the ear leaf (ULB), the blade of the first leaf below the ear leaf (LLB), the sheath of the ear leaf (ELS), the first sheath above the ear leaf (ULS), the first sheath below the ear leaf (LLS), the internode above the ear-bearing node (EI), the first internode above the ear-bearing internode (UI), the first internode below the ear-bearing internode (LI), and developing grains. Histological analysis Immature 08LF kernels from POL and NONPOL plants sampled at each stage (from 6 to 24 DAS) were fixed in FAA solution (63% ethanol, 5% acetic acid, and 2% formaldehyde) for 10 min at room temperature. After dehydration in a graded ethanol series (50%, 70%, 85%, 95%, and 100% ethanol), the grains were embedded in Technovit 7100 (Haereus Kluver) according to the manufacturer’s instructions. Thin Sect. (5 μm) were stained with Safranin O-Fast Green and observed by light microscopy as described [32]. RNA isolation, library construction, and transcriptome sequencing RNA was extracted from developing kernels at 6, 10, 14, 18, 21, and 24 DAS from both NONPOL and POL plants, with three biological replicates per time point, using the RNAprep Pure Plant Kit (TIANGEN). The 6 DAS samples represent a shared pre-treatment stage prior to the manipulation of pollination. RNA-seq libraries were constructed using the Illumina TruSeq RNA Library Preparation Kit (Illumina, San Diego, USA). Single-end sequencing was performed on the Illumina HiSeq 2000 platform. RNA libraries from all samples were constructed simultaneously and sequenced in the same Illumina sequencing batch. The raw read data are available at the National Center for Biotechnology Information Sequence Read Archive (https://www.ncbi.nlm.nih.gov/sra; accession number PRJNA1347479). Weighted Gene Co-expression Network Analysis (WGCNA) [33] was used to identify modules of highly correlated genes on the basis of the RNA-seq expression data. Cytoscape version 3.0.2 was used for network visualization and enrichment with various data, such as differential gene expression levels. Identification and functional annotations of DEGs An in-house Perl script was used to remove single-end reads with > 5% ambiguous nucleotides (Ns) and reads for which > 10% bases had a Phred quality score of < 20 [34]. The remaining high-quality reads from each sample were mapped to the Maize B73 RefGen_v3 sequence (http://www.maizegdb.org) using TopHat version 2.0.6. DEGs were identified using a published pipeline [35]: transcripts were assembled using the Cufflinks pipeline, and DEGs were identified using Cuffdiff version 2.0.1 [36]. Genes with a fold change > 1.5 or < 0.7 and a false discovery rate (FDR)-adjusted P value < 0.05 were considered to be DEGs. Bioinformatics analysis Principal component analysis (PCA) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were performed using OmicShare tools (www.omicshare.com/tools) with default parameters. Gene Ontology (GO) enrichment analysis of genes from co-expression modules was performed using tools available at http://www.geneontology.org/. Pearson correlation coefficients were calculated to assess the relationships between hub gene expression and physiological traits. The corresponding P values were adjusted for multiple comparisons using the Benjamini–Hochberg FDR correction. Reverse transcription quantitative PCR (RT–qPCR) validation Total RNA was extracted from three biological replicates of developing kernel tissues from POL and NONPOL plants using TRIzol Reagent (Invitrogen, Carlsbad, CA, USA). RNA integrity was verified by 1% agarose gel electrophoresis, and RNA was reverse transcribed using HiScript Q RT SuperMix for qPCR (Vazyme, Nanjing, China) in a final volume of 25 µL according to the manufacturer’s instructions. Real-time PCR was performed using a LightCycler 480 II Real-Time PCR detection system (Roche) with SYBR Green I. Each reaction included 0.5 µL of the RT reaction product, 0.6 µL of each primer (forward and reverse), 1 µL of 20× SYBR Green I dye, 12.5 µL of premix (BioTeke, Beijing, China), and 9.8 µL of nuclease-free water. All RT–qPCR reactions were performed in a 96-well plate at 95 °C for 3 min, followed by 40 cycles of 95 °C for 10 s, 60 °C for 20 s, and 72 °C for 20 s. All reactions were performed in triplicate. The 18S gene was used as the endogenous control. The specificity of each primer pair was verified by agarose gel electrophoresis and melting curve analysis. The 2−ΔΔCt method was used to analyze the relative abundance of each cDNA [37]. Statistical analysis All data are presented as mean ± standard deviation (SD) based on three independent biological replicates. Statistical analyses of physiological and biochemical data were performed using R software. Differences between POL and NONPOL treatments were evaluated at each developmental stage using Student’s t-test. A significance threshold of P < 0.05 was used. Pearson’s correlation coefficients were calculated to assess the relationships between gene expression levels and physiological traits. Corresponding P values were adjusted for multiple comparisons using the Benjamini–Hochberg method to control the FDR, and adjusted P < 0.05 was considered statistically significant. Results Physiological characteristics of developing maize grains under normally pollinated and non-pollinated conditions Pollination prevention clearly accelerated leaf senescence in 08LF plants (Fig. 1A), consistent with our previous study [28]. As expected, grains failed to develop normally when pollination was prevented. In normally pollinated (POL) plants, both fresh and dry grain weight increased slowly from 10 to 14 DAS, then more rapidly to 39 DAS. After 39 DAS, dry weight (average 50-grain maximum weight 11.59 ± 0.57 g) continued to increase until 45 DAS, whereas fresh weight (average 50-grain maximum weight 18.64 ± 0.93 g) declined. Grain moisture content (which was as high as 89.6% at 10 DAS) decreased rapidly from 18 to 24 DAS, then more slowly after 27 DAS to a final value of 34.14% (Fig. 1B). In the NONPOL treatment, grain dry weight (maximum value 0.10 g at 18 DAS) was extremely low and increased slightly over time, whereas fresh weight (average 50-grain maximum weight 0.75 ± 0.04 g) and water content (as high as 88.00% at 14 DAS, final value of 76.56%) increased from 6 to 14 DAS and decreased thereafter (Fig. 1C). Grain weights and water contents differed significantly between the POL and NONPOL treatments. These results indicate that different storage-reserve processes occurred during early grain development in NONPOL and POL plants. Phenotypes of 08LF plants under normally pollinated (POL) and non-pollinated (NONPOL) conditions. A Phenotypic observations of premature senescence in whole NONPOL and POL plants at 10, 14, 18, 21, 24, and 27 DAS. NONPOL plants are shown on the left, and POL plants are shown on the right. Bar = 1 cm. B-C Water content (green), fresh weight (red), and dry weight (yellow) of developing grains (6 to 51 DAS) from POL (B) and NONPOL (C) plants. Error bars represent the SDs of three independent biological replicates. The histodifferentiation (I), seed-filling (II), and desiccation (III) phases are indicated at the bottom. D-E Phenotypes of ears (D) and grains (E) at early developmental stages (from 6 DAS to 27 DAS) in POL and NONPOL plants. POL grains are shown in the upper panel and NONPOL grains in the lower panel. F Cellular observations of NONPOL and POL grains from 6 DAS to 27 DAS. Per: pericarp; end: endosperm; emb: embryo; sg: starch granules; BETL: basal endosperm transfer layer To investigate phenotypic differences between the two treatments during early grain development, we examined whole kernels from the middle sections of ears at 6, 10, 14, 18, 21, 24, and 27 DAS (Fig. 1D–E). Kernels were also fixed, sectioned, and stained with Safranin O-Fast Green to visualize developmental changes at the histological level. In POL plants, the endosperm increased dramatically in size from 6 to 10 DAS; amyloplasts became visible, and the basal endosperm transfer layer and aleurone cells differentiated. The embryo was between the pre-embryonic and sheath-leaf stages, and its area gradually increased. The area of the embryo increased markedly from 10 to 14 DAS, and the scutellum, coleoptile, radicle sheath, and other structures differentiated (Fig. 1F). The endosperm gradually changed from a liquid state to a solid state until 24 DAS, at which time the embryonic leaves had clearly differentiated. By contrast, double fertilization did not occur within the ovaries of NONPOL plants, and no endosperm or embryo structures formed. The ovary increased in volume only slightly between 6 and 10 DAS, then began to collapse and ultimately die (Fig. 1F). These results confirmed that pollination prevention halted the normal process of early grain development. Carbohydrate levels and stress-related indices associated with premature senescence induced by pollination prevention To examine the partitioning of major carbohydrates to other maize organs, as secondary sinks, in the absence of the primary grain sink, we measured the levels of sucrose, glucose, starch, and trehalose in ten organs of POL and NONPOL plants (Figure S1). Starch, trehalose, and glucose contents in grains were significantly higher in POL than in NONPOL plants from 14 DAS onward (Fig. 2). By contrast, levels of several carbohydrates in leaf, sheath, and internode tissues were higher in NONPOL than in POL plants on one or more sampling dates. For example, starch contents in internodes (UI, EI, and LI) were significantly higher in NONPOL plants than in POL plants, but there were no significant differences in the starch contents of leaf sheaths (ULS, ELS, and LLS) (Fig. 2A). Sucrose contents in the ULS and ELS and glucose contents in the leaf blades were significantly higher in NONPOL plants than in POL plants (Fig. 2B–C). The content of trehalose was higher in POL plants after 10 DAS; although trehalose content was generally lower in most NONPOL plant organs, it was significantly higher in the EI and ELS of NONPOL plants (Fig. 2D). These results indicate that pollination prevention altered sugar partitioning and translocation, which coincided with the onset of premature senescence. Carbohydrate levels and partitioning are affected by pollination prevention. A-D Contents of starch A, sucrose B, glucose C, and trehalose (D) from 6 DAS to 24 DAS in different organs of NONPOL and POL plants: the blade of the ear leaf (ELB), the blade of the first leaf above the ear leaf (ULB), the blade of the first leaf below the ear leaf (LLB), the sheath of the ear leaf (ELS), the first sheath above the ear leaf (ULS), the first sheath below the ear leaf (LLS), the internode above the primary ear-bearing node (EI), the first internode above the EBI (UI), the first internode below the EBI (LI), and the grain. EBI: ear-bearing internode. Green and red lines indicate NONPOL and POL plants, respectively. Metabolite levels were determined by ELISA and normalized to total soluble protein concentration. Data are presented as mean ± SD of three independent biological replicates. Statistical significance between POL and NONPOL treatments at each developmental stage was determined using Student’s t-test (P < 0.05). Asterisks indicate significant differences: *P < 0.05, **P < 0.01, and ***P < 0.001. E-F Pearson’s correlations between grains and other organs based on the physiological data from POL (E) and NONPOL (F) plants. P values were adjusted using the Benjamini–Hochberg method We also measured the activity of the antioxidant enzyme SOD and the contents of the stress-related metabolites H2O2 and MDA in these organs (Figure S2). H₂O₂ and MDA contents were generally higher in NONPOL plants, particularly in leaf blades and grains. By contrast, SOD activity was higher before 18 DAS in NONPOL plants (except in grains) but became higher in POL plants at later stages (Figure S2). These physiological changes were consistent with the progression of premature senescence in NONPOL plants. Pearson correlation analysis among these organs using all the physiological data showed distinct patterns between POL and NONPOL plants. In POL plants, grain showed the highest correlation with the UI (r = 0.69) and LI (r = 0.66) (Fig. 2E), whereas in NONPOL plants, the grain showed the highest correlation with the ULS (r = 0.75) and ELS (r = 0.68) (Fig. 2F). These results provide further evidence that pollination prevention altered the translocation and partitioning of sugars and the occurrence of oxidative stress during grain development. Global transcriptome analysis of whole maize grains from POL and NONPOL plants To investigate the genetic control of grain development under POL and NONPOL conditions, we performed RNA-seq of whole grains from the inbred line 08LF. Grains were collected from POL plants and NONPOL plants at 6, 10, 14, 18, 21, and 24 DAS. Notably, 6 DAS corresponded to approximately 0 days after pollination in POL plants, as artificial pollination was performed at 6 DAS. The 6 DAS samples represent the pre-treatment stage prior to the differentiation of NONPOL and POL treatments. Approximately 7.5–8.4 million clean Illumina reads were obtained from each library and mapped to the maize B73 reference genome. An average of 70.9% of reads were uniquely mapped (Fig. 3A). Expression values of the three biological replicates from each treatment at each time point were highly correlated (R2 = 0.86 to 0.99; Figure S3), highlighting the reproducibility of the data. Transcriptome analysis of early maize grain development 6 DAS (pre-treated) to 24 DAS under non-pollinated (NONPOL) and normally pollinated (POL) conditions. A Summary of read analysis. Samples labeled “N” represent NONPOL grains, whereas samples labeled “Y” represent POL grains. B PCA of transcriptome data from NONPOL and POL plants at six time points. C-D Venn diagram of expressed gene numbers in grains of NONPOL (C) and POL (D) plants at six developmental time points. E-F Numbers of shared and unique preferentially expressed TF genes in NONPOL (E) and POL (F) grains at each stage of development. G Percentages of genes expressed at different levels (based on RPKM) in NONPOL and POL grain from 6 DAS to 24 DAS. From left to right, the four subpanels represent genes with RPKM values of 0–1, 1–10, 10–100, and > 100, respectively. The 6 DAS samples correspond to the stage prior to treatment differentiation To investigate global differences in transcriptome dynamics during grain development under POL and NONPOL conditions, we performed PCA of samples from the different developmental stages (Fig. 3B). During early grain development (6, 10, and 14 DAS), POL and NONPOL samples were more closely clustered along the first two PC axes, indicating that their transcriptional programs were relatively similar. However, the POL and NONPOL samples were clearly separated from 18 DAS onwards (Fig. 3B), suggesting that a major divergence in the transcriptomes of POL and NONPOL plants occurred between 14 and 18 DAS that determined their subsequent developmental trajectories. A total of 29,389 and 29,664 expressed genes were identified in grains from NONPOL and POL plants, respectively (Fig. 3C–D), 27,758 and 28,209 of which were expressed in all samples of the respective treatments. Approximately 521, 315, 221, 219, 224, and 131 genes were expressed specifically at different time points in NONPOL grains (Fig. 3C), and 293, 354, 226, 217, 140, and 225 genes were expressed specifically in POL grains (Fig. 3D). Furthermore, 1219–2821 genes were specifically expressed at each grain developmental stage between the two pollination treatments (Figure S4). The proportion of expressed transcription factor (TF) genes closely tracked the total number of expressed genes (Fig. 3E–F). The proportion of genes with low (0 < RPKM < 1) and medium expression (1 ≤ RPKM < 10) was higher in POL grains than in NONPOL grains, but the percentage of genes with high expression (10 ≤ RPKM < 100) was similar between the two pollination treatments before 14 DAS. After 14 DAS, the proportion of genes with high expression was greater in NONPOL grains. Interestingly, the proportion of genes with very high expression (RPKM ≥ 100) increased sharply at 10 and 14 DAS in POL grains, then declined lower than that in NONPOL plants after 18 DAS (Fig. 3G), which is defined as the early stage of grain morphogenesis and milk ripening. These highly expressed genes may play an important role in early grain development and will be the focus of study for future research on the grain-development process. To validate the RNA-seq results, we performed RT–qPCR on a subset of 25 genes using gene-specific primers (Table S1). The expression levels measured by RT–qPCR were consistent with those obtained by RNA-seq, with R2 values of 0.71–0.91 (Figure S5). Taken together, our RNA-seq analysis indicated that we had obtained sufficient transcriptome coverage for subsequent analysis of gene co-expression networks and also suggested that pollination prevention was associated with a genome-wide transcriptional change at around 14 DAS relative to pollinated plants. Identification of up- and downregulated genes during seed development under POL and NONPOL conditions To explore temporal changes in gene expression during grain development, we identified differentially expressed genes (DEGs) between subsequent time points under POL and NONPOL conditions. We identified 3936, 4006, 2071, 5045, and 4902 such DEGs in POL grains (Fig. 4A; Table S2) and 1657/1409, 1819/1821, 3215/2366, 2906/2378, and 2926/2420 (upregulated/downregulated) DEGs in NONPOL grains (Fig. 4B; Table S2) across five developmental stages. DEGs identified in NONPOL and POL plants. A-B Numbers of upregulated and downregulated genes in POL (A) and NONPOL (B) plants at different time points. C Heatmap of differentially expressed genes related to sugar synthesis and metabolism. D Heatmap of differentially expressed genes related to zein synthesis and metabolism. 6 DAS represents the pre-treatment stage prior to the differentiation of NONPOL and POL treatments Because the level of storage reserves was one of the main differences between POL and NONPOL grains, we specifically examined DEGs related to major storage compounds, including genes related to sugar synthesis and zein (Fig. 4C–D, Tables S3–S8). GO and KEGG enrichment analyses supported the functional specialization of these gene sets. For the sugar-synthesis-related DEGs shown in Fig. 4C (Table S3), GO enrichment analysis revealed significant overrepresentation of terms associated with carbohydrate metabolic processes, including carbohydrate metabolic process, cellular carbohydrate metabolic process, cellular carbohydrate biosynthetic process, disaccharide metabolic process, and polysaccharide metabolic process (Table S4). Consistent with these results, KEGG pathway analysis indicated strong enrichment in starch and sucrose metabolism (ko00500), together with broader categories such as metabolic pathways and biosynthesis of secondary metabolites, highlighting active carbohydrate biosynthesis and metabolic reprogramming in POL grains (Table S5). By contrast, the zein-related gene set displayed a distinct enrichment pattern (Fig. 4D; Table S6). GO analysis showed significant enrichment in regulatory processes, particularly regulation of transcription, DNA-templated, regulation of RNA biosynthetic process, and regulation of nucleic acid-templated transcription. Notably, the term regulation of starch biosynthetic process was also enriched, suggesting potential coordination between protein and starch accumulation (Table S7). KEGG analysis identified enrichment in signaling-related pathways, including MAPK signaling pathway–plant and plant–pathogen interaction (Table S8), implying that transcriptional regulation and signaling cascades may contribute to the differential expression of storage protein genes between POL and NONPOL grains. Genes upregulated in POL grains at 10 DAS included mn1 (miniature seed1), incw5 (invertase cell wall5), ivr1 (invertase1), trps12 (trehalose-6-phosphate synthase 12), agp2 (ADP glucose pyrophosphorylase2), sbe3 (starch branching enzyme3), sps1 (sucrose phosphate synthase1), sps2, hex8 (hexokinase8), and GRMZM5G828987, GRMZM2G133966, GRMZM2G136106, GRMZM2G071339, AC217401.3_FG001, GRMZM2G376416, GRMZM2G463871, GRMZM2G111998, GRMZM2G019411, GRMZM2G120779, GRMZM2G110558, GRMZM2G017186, and GRMZM2G007939. Downregulated genes included pho2 (starch phoshorylase2), glu1 (beta glucosidase1), tre1 (trehalase1), trps5 (trehalose-6-phosphate synthase5), trps9, trps10, trps13, trpp1 (trehalose-6-phosphate phosphatase1), trpp4, trpp7, trpp10, and GRMZM2G069024. Although the specific genes differed, the numbers of up- and downregulated DEGs in POL grains at 14 DAS were similar to those at 10 DAS. This period (from 10 to 14 DAS) coincides with the morphological formation period of 08LF, when key structures such as the basal endosperm cell transfer layer are formed. The genes upregulated and downregulated at 10 and 14 DAS may therefore be involved in regulating the seed morphology of 08LF. Genes upregulated in POL grains from 18 to 21 DAS included GRMZM2G428518, GRMZM2G037411, pho1 (starch phosphorylase1), ss1 (starch synthase I), sh2 (shrunken2), bt2 (Brittle endosperm2), agpsl1 (ADP glucose pyrophosphorylase small subunit leaf1), GRMZM2G013811, su2 (sugary2), agpll1 (ADP glucose pyrophosphorylase large subunit leaf1), ae1 (amylose extender1), and sbe1 (starch branching enzyme1); downregulated genes included trps5, trps13, trpp1, trpp4, trpp7, trpp10, tre1, glu1, glu2, pho2, GRMZM2G069024, GRMZM5G810727, AC234160.1_FG003, GRMZM2G148176, GRMZM2G175499, GRMZM2G130046, and GRMZM2G120962 (Table S3). The upregulated genes were mainly involved in starch biosynthesis, and the downregulated genes were mainly involved in trehalose biosynthesis. This result was consistent with the histological analysis of 08LF grains, which indicated that significant starch biosynthesis and accumulation were initiated during this period. Genes upregulated in NONPOL grains at 10 DAS included GRMZM2G148176, AC234160.1_FG003, GRMZM2G069024, GRMZM5G810727, GRMZM2G175499, GRMZM2G130046, and GRMZM2G019411, as well as genes involved in the trehalose synthesis pathway (trps13). Genes upregulated at 14 DAS included GRMZM2G069024 and GRMZM5G810727, as well as trps5, trpp1, trpp4, trpp7, and trpp10 in the trehalose biosynthesis pathway. Upregulated genes at 18 DAS included ivr1, hex7, GRMZM2G103055, GRMZM2G015804, GRMZM2G110558, GRMZM2G017186, GRMZM2G007939, GRMZM2G055699, GRMZM2G108133, and multiple trehalose synthesis genes (trps4, trps9, trps10, trps13, trpp1, trpp4, trpp6, and trpp10). The trehalose synthesis genes trps4, trps9, trpp6, hex7, GRMZM2G055699, GRMZM2G108133, and GRMZM2G110558 were also upregulated from 14 to 21 DAS in NONPOL grains but were downregulated in POL grains (Table S3). These results suggest that trehalose biosynthesis genes may be involved in abnormal ovary development under NONPOL conditions. Overall, these findings indicate significant differences in the overall expression patterns of genes related to storage reserve metabolism in the seeds and ovaries of 08LF maize under POL and NONPOL conditions. Starch, storage protein, and oil are the three main storage components of corn kernels, and zein is the most abundant kernel storage protein. Zein genes differed in expression in the seeds and ovaries of POL and NONPOL grains and could be divided into three categories: (1) genes sharply upregulated after 21 DAS in POL grains, including GRMZM2G100018 (dzs18, δ-zein), GRMZM2G088365, GRMZM2G346897, GRMZM2G138689 (gz50; gutelin, 50-kD γ-zein), GRMZM2G397687 (z1C2 [floury2], α-zein), GRMZM2G094532 (floury1, zein protein-body ER membrane protein), GRMZM2G026939, GRMZM2G025763, GRMZM2G059620, GRMZM2G353272 (z1A1_6 [floury4]); (2) genes upregulated in POL grains and downregulated in NONPOL grains from 10 to 21 DAS, including GRMZM2G125931, GRMZM2G099577 (MADS47), GRMZM2G075039, and GRMZM2G075077; and (3) genes downregulated in POL grains but upregulated in NONPOL grains from 10 to 21 DAS, including GRMZM2G016150 (ohp1), GRMZM2G446234, and GRMZM2G134976 (Table S6). These results suggest that zein synthesis increased sharply at 21 DAS, when 08LF grains entered the milky stage, and that this period was critical for zein accumulation. Identification of gene modules associated with grain developmental events under POL and NONPOL conditions To investigate differences in gene regulatory networks between the two pollination treatments during grain development, we identified co-expressed gene sets by weighted gene co-expression network analysis (WGCNA). Power 16, the lowest power for which the scale-free topology fit index reached 0.90, was used to produce a hierarchically clustered tree (Fig. 5A), and the Dynamic Tree Cut and Merged Dynamic methods were used to identify 14 distinct gene modules in the tree (Fig. 5B). GO enrichment analysis of each module highlighted key biological processes and molecular functions in the set of co-expressed genes and corroborated the results of our differential gene expression analysis. In addition to GO enrichment, KEGG pathway analysis further supported the functional classification of these modules (Table S10). The enriched pathways were largely associated with central metabolic processes and signal transduction, indicating that the trait-associated modules represent coordinated regulatory networks involved in carbon metabolism, amino acid metabolism, photosynthesis, and stress-related signaling pathways during grain development. Most of the modules reflected distinct grain developmental processes between the two pollination treatments (Fig. 5C). WGCNA of grain samples from NONPOL and POL plants revealed gene modules associated with grain developmental events at different time points. A Scale-free topology index and mean connectivity were used to determine the soft threshold. B Hierarchically clustered dendrogram of NONPOL and POL grain samples showing co-expression modules identified by performing WGCNA on the 9,000 most temporally regulated genes. Modules corresponding to branches are labeled with colors indicated by the color bands below the tree. With 0.25 threshold merging, 14 modules were generated. C GO terms associated with each of the 14 modules and average expression levels of the 30 genes with the highest module membership within each module at all time points for NONPOL and POL grain samples Expression of genes related to energy-reserve metabolism and starch/glycogen biosynthesis and metabolism (the black module) initially increased and then decreased slightly at 21 and 24 DAS in POL grains. Consistent with these results, KEGG analysis showed that the black module was significantly enriched in starch and sucrose metabolism, glycolysis/gluconeogenesis, and carbon metabolism, further confirming its central role in energy-reserve accumulation and carbohydrate biosynthesis during grain development. By contrast, expression of these genes decreased before 10 DAS in NONPOL grains, then increased slowly from 14 DAS and remained lower than that in POL grains through 24 DAS. The pink module was enriched in genes associated with the terms RNA polymerase complex, RNA polymerase II, catalytic complex, transferase complex, nuclear DNA-directed RNA polymerase complex, DNA-directed RNA polymerase complex, nucleoplasm part, transferase complex, and nucleoplasm. In POL grains, expression of the pink-module genes remained unchanged from 6 to 10 DAS, increased at 14 DAS, and decreased thereafter until 24 DAS. In NONPOL grains, the expression pattern of pink-module genes was similar to that of black-module genes. Genes in the magenta module were enriched in the terms carbohydrate binding, endopeptidase inhibitor/regulator activity, peptidase inhibitor/regulator activity, serine-type endopeptidase inhibitor activity, nutrient reservoir activity, enzyme inhibitor activity, and molecular function regulator. These genes were dramatically upregulated in POL grains until 24 DAS (with a slight decline at 14 DAS) but showed a fluctuating expression pattern in NONPOL grains. Genes in the brown module were related to aminoglycan metabolic/catabolic process, chitin metabolic/catabolic process, amino sugar metabolic/catabolic process, glucosamine-containing compound metabolic/catabolic process, and cell wall macromolecule metabolic/catabolic process. In POL grains, their expression rose sharply at 10 DAS and then dropped rapidly through 24 DAS, but in NONPOL grains, their expression increased gradually through 24 DAS. In line with the GO results, KEGG pathways enriched in the brown module included amino sugar and nucleotide sugar metabolism and plant–pathogen interaction, suggesting coordinated regulation of cell wall remodeling and stress-responsive processes. Genes in the turquoise module (enriched in translation, peptide biosynthetic process, ribonucleoprotein complex biogenesis/assembly/subunit organization, cellular protein-containing complex assembly, protein-containing complex assembly/subunit organization, protein folding, and cellular respiration), the greenyellow module (enriched in sequence-specific DNA binding and carbohydrate phosphatase activity), and the purple module (enriched in endomembrane system) showed an expression pattern similar to that of brown-module genes in POL grains, with the greenyellow module exhibiting a delayed peak at 14 DAS. However, these genes behaved differently in NONPOL grains: expression of genes in the green-yellow module increased until 18 DAS and then declined from 21 DAS onward, and expression of genes in the turquoise module was dramatically downregulated after 6 DAS and remained low until 18 DAS, rising slightly after 21 DAS. Genes in the tan module were enriched in protein folding, response to heat, response to temperature stimulus, and response to abiotic stimulus; they showed similar expression patterns in POL and NONPOL grains before 18 DAS. After 18 DAS, their expression remained low in POL grains but was dramatically upregulated in NONPOL grains. Genes in the red module were enriched in cell wall organization or biogenesis, external encapsulating structure organization, cellular glucan metabolic process, glucan metabolic process, polysaccharide metabolic process, cellular polysaccharide metabolic process, and cellulose biosynthetic process, and genes in the yellow module were enriched in photosynthesis/light harvesting, photosynthesis, protein–chromophore linkage, photosynthesis/light reaction, response to abiotic stimulus, and response to temperature stimulus. KEGG pathways enriched in the yellow module included photosynthesis–antenna proteins and protein processing in the endoplasmic reticulum, indicating coordinated regulation of light-related functions and protein maturation. Expression of genes in both of these modules remained low throughout 24 DAS in POL grains. By contrast, expression of red-module genes initially declined and then increased after 18 DAS in the NONPOL grains, whereas expression of yellow-module genes rose sharply at 10 DAS and then dropped rapidly, remaining low through 24 DAS. Genes in the green module were enriched in cell communication, signal transduction, signaling, and fatty acid metabolic process; their expression was similar in POL and NONPOL grains, declining continuously after 6 DAS. Similarly, expression of genes in the salmon module (enriched in the term molecular function regulator) also decreased after 6 DAS but was higher at 10 and 18 DAS in POL grains than in NONPOL grains. Genes in the blue module were enriched in phosphorelay sensor kinase activity, protein serine/threonine phosphatase activity, protein histidine kinase activity, and phosphotransferase activity. KEGG analysis further revealed enrichment in amino acid degradation pathways, including valine, leucine, and isoleucine degradation, supporting a role for this module in amino acid turnover and associated signaling processes. Their expression was similar to that of salmon-module genes in POL grains but fluctuated (“down-up-down”) in NONPOL grains, with a sharp peak at 14 DAS. No GO terms were significantly enriched in the grey module, and the biological significance of changes in grey-module gene expression remains to be determined. Identification of gene modules related to physiological characteristics during grain development We next investigated the relationships between individual gene co-expression modules and physiological characteristics by correlating the eigengenes of each module with physiological indices of whole maize grains under the two pollination treatments. As expected, four modules (black, salmon, blue, and brown) exhibited high correlations with the two pollination treatments (|r| ≥ 0.70, P < 1e–3; Fig. 6A). The black and salmon modules were negatively correlated with normal pollination (r < − 0.5, P < 10− 2), whereas the blue and brown modules were positively correlated with normal pollination (Fig. 6A). The NY column in Fig. 6A represents the pollination treatment (POL vs. NONPOL), and these strong correlations indicate coordinated transcriptional responses to pollination status. Eigengene expression patterns of these modules further revealed distinct divergence between POL and NONPOL samples after treatment separation (post-6 DAS), demonstrating dynamic expression shifts associated with pollination prevention. Carbohydrate contents and stress-related indices were significantly correlated with three modules (yellow, brown, and salmon). The yellow module, which contained 721 genes, was correlated with sucrose content, glucose content, and SOD activity across all developmental stages. The brown module (1670 genes) was highly correlated with starch content and H2O2 content, whereas the salmon module (56 genes) was negatively correlated with these parameters (Fig. 6A, Table S9). Furthermore, seven of the 14 co-expression modules were highly correlated with individual samples (|r| ≥ 0.70, P < 1e − 3), e.g., the green, red, and salmon modules with N6N, the black module with N18N, the magenta module with N24N, the yellow module with Y10Y, and the tan module with Y14Y (Fig. 6A). In summary, expression of genes in the black, blue, brown, salmon, and yellow modules appeared to be most closely associated with physiological differences between NONPOL and POL grains. Module–trait analysis and hub genes of modules of interest in NONPOL and POL grains at different developmental stages. A Module–trait analysis of WGCNA modules in NONPOL and POL grains at different developmental stages. The heatmap shows correlation coefficients (Pearson’s r) between module eigengenes and physiological traits. P-values were adjusted for multiple testing using the Benjamini–Hochberg procedure to control the false discovery rate (FDR). Cells marked with an asterisk indicate correlations with adjusted P < 0.05. Modules showing significant correlations with NY reflect coordinated transcriptional responses to pollination status. ME represent modules identified by WGCNA. 6 DAS represents the pre-treatment stage prior to the differentiation of NONPOL and POL treatments. B Hub genes identified in modules of interest (black, blue, brown, salmon, and yellow). C Pearson’s correlations between hub genes in the black modules and physiological traits We next identified hub genes of the co-expression modules. We modeled a network view of the modules using the cytoHubba plugin with TOM ≥ 0.25 and identified 20 major hub genes (Fig. 6B). Detailed gene IDs and functional annotations of the hub genes shown in Fig. 6B are listed in Table S11. Five hub genes were detected in the black module: two MYB transcription factor genes (GRMZM2G137974 and GRMZM2G369799), two Dull endosperm1 genes (GRMZM2G141399 and GRMZM2G021578), and one Legumin 1 gene (GRMZM2G174883). Three hub genes were detected in the brown module (GRMZM2G037015 [PIF/Ping-Pong transposase family protein], GRMZM2G117007 [Zinc finger CCCH domain-containing protein 29], and GRMZM2G445634 [ZIM-family protein, ZIM16]); four in the blue module (AF546188.1_FG007 [19 kDa alpha-zein 19C2, de30], GRMZM2G026939 [Z1A alpha zein protein], GRMZM2G059620 [Zein-alpha 19B1, uaz129], and GRMZM2G353268 [Zein-alpha A30]); three in the salmon module (GRMZM2G323622 [kinesin motor protein-related], GRMZM2G147500 [Afadin/alpha-actinin-binding protein], and GRMZM2G096470 [TBC domain containing protein]); five in the yellow module (GRMZM2G335242 [Class I heat shock protein 3], GRMZM2G422240 [heat shock protein 17.2, hsp17.2], GRMZM2G059836 [farnesylated protein 2], GRMZM2G031938 [MATE efflux family protein], and GRMZM2G456618 [uncharacterized protein]) (Fig. 6B, Table S11). These hub genes represent candidate regulatory genes that could serve as study targets for grain development to alter specific gene programs after different pollination treatments. To further explore the regulatory mechanisms of these hub genes in leaf and grain development, we correlated the transcriptomic and physiological data from leaves with these gene expression patterns in seeds (Fig. 4, Figure S2). Hub genes of the black and blue modules were highly correlated with starch content in leaves (r = 0.78–0.93) (Fig. 6C, Figure S6A), whereas those of the brown module were negatively correlated with glucose content (r = − 0.53 to − 0.48) and positively correlated with H2O2 activity (GRMZM2G445634, r = 0.56) (Figure S6B). Hub genes of the other two modules, yellow and salmon, were mainly correlated with stress-related indices (Figure S6C–D). In the yellow module, the hub gene GRMZM2G031938 was positively correlated with MDA content (r = 0.54), whereas GRMZM2G059836, GRMZM2G422240, and GRMZM2G456618 were correlated with SOD activity. The salmon-module hub genes were negatively correlated with MDA content (GRMZM2G096470, r = − 0.53; GRMZM2G323622, r = − 0.58). Most of the hub genes (15/20) were highly correlated with each other (0.87–0.99 for the black module, 0.87–0.97 for the brown module, 0.92–0.98 for the salmon module, and 0.99–1.0 for the blue module); the exception was the yellow module (r = 0.44–0.97) (Fig. 6C, Figure S6). These correlations suggest coordinated expression patterns between sink and source tissues under different pollination treatments. Discussion Pollination prevention disrupts sink establishment and alters reserve accumulation dynamics The establishment of a functional grain sink is essential for the coordinated accumulation of storage reserves during maize kernel development. The closely coordinated process of maize kernel development is divided into three main stages: the early developmental stage, the filling stage, and the dehydration stage. The early developmental stage takes place over the first 15 days after pollination, when the embryo and endosperm rapidly develop from the original zygote and endosperm nuclei to produce highly differentiated tissues [38]. In this study, the dry weight of POL grains increased slowly from 6 to 10 DAS, followed by a more rapid increase from 14 DAS to 39 DAS, reflecting the progressive establishment of sink strength and storage reserve deposition. Although the dry weight of NONPOL grains remained extremely low, a peak was observed at 14 DAS; we speculate that the increase in dry weight of NONPOL grains may simply reflect the growth of the ovaries. Our observations suggested that major differentiation changes during the early developmental stage occurred at around 14 DAS in 08LF grains. This conclusion was supported by the results of PCA analysis, in which the transcriptomic profiles of POL and NONPOL grains were relatively similar from 7 to 14 DAS, then diverged at 18 DAS, indicating that both NONPOL and POL ovaries probably underwent normal cell division and differentiation. The grain filling stage, when the endosperm and embryo begin to accumulate storage reserves, occurs from about 12 to 45 days after pollination [39]. During this stage, the developing maize grain is the main sink for photosynthetic assimilates in the maize plant, and the absence of this strong sink is closely associated with the striking differences observed between NONPOL and POL grains from 14 DAS onward, highlighting the impact of pollination prevention on storage reserve accumulation and assimilate partitioning. In our previous study, leaf senescence, characterized by a slight red coloration of the veins, began in NONPOL plants at 18 DAS [28], consistent with the occurrence of early senescence induced by grain removal in pollination-prevention treatments, and further supporting the disruption of whole-plant source–sink balance. Transcriptional modules underlying altered reserve biosynthesis and partitioning Although effects of various biological cues on grain development have been described [13, 39–41], the transcriptional regulatory mechanisms that underlie the response of maize grains to pollination-prevention treatments have rarely been reported. To investigate the transcriptional basis underlying altered storage reserve accumulation and partitioning, we used WGCNA to identify 14 gene co-expression modules from 31 maize grain samples. Most modules showed differences in gene expression patterns between NONPOL and POL grains, clearly highlighting the transcriptional changes caused by pollination prevention in maize kernels. Seven of the modules were significantly correlated with physiological characteristics, and we focused in particular on the black, blue, brown, salmon, and yellow modules, which were correlated with multiple physiological indices. GO enrichment analysis revealed that energy reserves, kinase activities, cell wall metabolic/catabolic process, photosynthesis, and abiotic stress were the top enriched terms in these five modules, respectively, indicating coordinated transcriptional regulation of pathways associated with reserve biosynthesis and metabolic reprogramming. WGCNA trait–module analysis provided additional evidence that the genes in these modules might play important roles in grain development, particularly during the transition from early development to active reserve deposition. Hub genes were identified in each module, including MYB transcription factor genes and genes encoding dull endosperm protein in the black module, genes encoding zein proteins in the salmon and blue modules, heat shock protein genes in the yellow module, and genes encoding a PIF/Ping-Pong transposase protein, zinc finger CCCH domain-containing protein 29, and ZIM16 in the brown module. Zein proteins constitute the major storage protein fraction in maize endosperm and play important roles in protein body formation and grain protein composition [42]. The transcriptional regulation of zein biosynthesis is tightly linked to endosperm development and storage reserve accumulation [43]. In this study, zein hub genes in the blue module showed much higher expression in POL grains than in NONPOL grains. This pattern is consistent with active storage protein deposition during normal grain filling. By contrast, reduced zein gene expression in NONPOL grains reflects impaired storage protein accumulation under weakened sink conditions. Although grain quality traits were not directly measured in this study, altered zein gene expression may influence endosperm protein composition and kernel characteristics when sink establishment is disrupted. These findings suggest that proper regulation of zein biosynthesis is closely associated with coordinated reserve accumulation during grain development. MYB TFs are expressed throughout plant development, including in senescent leaves, and can act as negative regulators of ABA response, stress, wounding, and senescence by maintaining growth through a homoeostatic function [44]. In Arabidopsis, AtMYB23 regulates epidermal cell development in both seeds and vegetative tissues, as confirmed by fusion with the EAR repression domain [45]. Two Du1 genes (GRMZM2G021578, GRMZM2G141399) have been reported to influence the structure of endosperm starch, and du1 mutations affect the activity of two enzymes involved in starch biosynthesis, starch synthase II and starch branching enzyme IIa [46]. In this study, two hub MYB TFs had significantly higher expression in POL grains than in NONPOL grains, suggesting that reduced expression of these genes may contribute to altered grain development under NONPOL conditions. The hub genes of the blue module encoded zein proteins, the major storage proteins in maize grain [43], which constitute an essential component of storage reserve deposition in developing kernels. Their expression was much higher in POL grain than in NONPOL grain owing to complete fertilization failure, further supporting the reduced capacity for storage reserve accumulation in NONPOL grains. Heat shock proteins (HSPs) are not only induced under thermal stress but also differentially regulated at different seed developmental stages, suggesting the importance of their chaperone roles during seed development [47, 48]. The most prominent developmental stage in plants during which HSP genes are expressed is embryogenesis. The functions of HSPs in the developing embryo are not well understood, despite the fact that their expression is a conserved feature of animal oogenesis and embryogenesis. In the present study, HSP gene expression was relatively stable in POL grains but increased sharply at 10 DAS and then declined rapidly in NONPOL grains. This observation suggests that HSP genes might play an important role in the early stages of grain development. One hub gene in the brown module encoded a ZIM-motif family protein with high sequence identity to Arabidopsis JAZ1 (AT1G19180), which is involved in JA signaling. Multiple developmental processes—such as germination, seedling development, root growth, flower development, grain development, tuber formation, and senescence—are regulated by JA/JA-Ile [28, 49]. Expression of the ZIM-motif protein gene was much higher in NONPOL grain than in POL grain, suggesting a possible role for JA in ovary senescence, especially at later stages of development. Such hormone-associated regulatory shifts may contribute to altered assimilate partitioning and the weakening of sink activity in NONPOL plants. The expression of hub genes identified in grains (sink tissues) also clearly differed between leaves (source tissues) of NONPOL and POL plants and was closely correlated with differences in leaf physiological characteristics between the two treatments. This coordinated transcriptional reprogramming between sink and source tissues supports the notion that pollination prevention disrupts whole-plant source–sink communication and storage reserve allocation. Further study of these hub genes may therefore contribute to understanding the molecular mechanisms involved in the plant source–sink regulatory network. It should be noted that the hub genes identified in this study were inferred from co-expression relationships and correlations with physiological traits. Although these findings are based on associative relationships, further functional validation through genetic or molecular approaches would help to clarify their precise regulatory roles in grain development and source–sink regulation. Carbohydrate redistribution and source–sink imbalance following pollination prevention The initiation and progression of senescence is controlled by multiple factors, including disease pressure, shading, extreme temperature, oxidative stress, and nutrient starvation [50–52]. Internal factors, including changes in phytohormone and sugar levels, as well as source–sink relationships, also affect the senescence program [26, 29, 50, 51, 53]. ABA, SA, auxins, cytokinins, and gibberellins have all been implicated in leaf senescence [26, 54, 55], and the direct application of sugars to leaf cells of Arabidopsis resulted in premature senescence [56, 57]. Likewise, early senescence in barley (Hordeum vulgare) resulted from hyperaccumulation of leaf carbohydrates due to stem girdling [58]. These findings collectively indicate that carbohydrate accumulation and partitioning are tightly linked to the regulation of senescence and source–sink balance. Despite advances in the study of hormones and sugar, the underlying mechanisms of source–sink relationships during senescence remain unclear. Leaf senescence can be induced by ear removal or pollination prevention in maize, but this effect is genotype specific, and these two treatments can either accelerate or reduce the rate of leaf senescence [27, 29, 59]. We previously reported that pollination prevention can induce premature leaf senescence in 08LF through the degradation of Chlorophyll-a [28], and early leaf senescence of 08LF was also associated with the excessive accumulation of sucrose in another report [60]. In the present study, the genes upregulated in POL grains were mainly associated with the processes of sucrose hydrolysis and starch synthesis, whereas those upregulated in NONPOL grains were mainly involved in trehalose biosynthesis, indicating a shift in carbohydrate metabolic pathways under NONPOL conditions (Fig. 7A). Trehalose metabolism has been widely reported to participate in plant stress responses and metabolic signaling pathways, particularly through the T6P–SnRK1 regulatory system that coordinates carbon status and stress adaptation [61]. Working model for the regulation of seed development and leaf senescence by source–sink communication. A Hub genes (shown in green font) identified from WGCNA analysis and previous studies are involved in sugar transport, hormone signaling, storage, and abiotic stress response and are thus likely to regulate senescence and seed development. Sucrose generated by photosynthesis during the reproductive phase of the maize plant is destined for the primary sink (grain). When sugar demand by the primary sink ceases after the completion of grain fill (indicated by a thin yellow line), some of the hub genes may be differentially expressed (up- and downregulated genes are represented by red arrows). The level of storage reserve accumulation (shown in the dark or light blue box) is likely important for the regulation of senescence or seed development. B Diagram of storage reserve translocation induced by pollination prevention Several trehalose pathway genes, including trps5, trps9, trps10, trps13, trpp1, and trpp4, were consistently upregulated in NONPOL grains between 14 and 21 DAS. These genes encode trehalose-6-phosphate synthases (TPSs) and trehalose-6-phosphate phosphatases (TPPs), which regulate the levels of trehalose-6-phosphate (T6P), a key signaling metabolite in plants. T6P has been reported to function as an indicator of cellular carbon status and to interact with SnRK1-dependent signaling pathways that modulate developmental transitions and stress responses [62, 63]. Previous studies have shown that T6P signaling can also influence leaf senescence [64, 65]. We also identified a ZIM-motif family gene homologous to Arabidopsis JAZ1 as a hub gene with increased expression in NONPOL grains. The coordinated upregulation of trehalose pathway genes, together with altered expression of a JA-related regulatory gene, suggests that carbohydrate signaling and hormone-associated pathways may be concurrently reprogrammed under pollination-prevention conditions. Although our study does not provide direct functional validation, these transcriptional patterns suggest that trehalose metabolism may link altered carbon allocation with stress signaling during ovary degeneration and senescence. Further study of the functions and molecular mechanisms of these genes will contribute to our understanding of seed development and plant senescence. These transcriptional changes are consistent with the observed differences in storage reserve accumulation and support the notion that pollination prevention reshapes carbohydrate partitioning between source and sink tissues. Correlation analysis between maize grains and other organs based on all the physiological data revealed that grain was more highly correlated with EI and ELS in NONPOL plants than in POL plants, suggesting that pollination prevention may be associated with altered reserve translocation (Fig. 7B) and that an additional compensating mechanism might be induced. Such altered correlations are consistent with a reconfigured source–sink network, in which lower grain sink strength may redirect assimilates and disturb normal patterns of reserve allocation. It should be noted that the responses described here were characterized in a single inbred line. Because maize genotypes differ in their sensitivity to pollination prevention, further studies in diverse genetic backgrounds will be necessary to evaluate the generality of these regulatory patterns. By integrating physiological measurements with transcriptomic network analysis, this study links sink failure to coordinated metabolic, transcriptional, and systemic adjustments. The identified hub genes may represent regulatory nodes that connect reserve accumulation processes with stress and senescence pathways. This integrated view advances our understanding of how grain sink strength influences whole-plant carbon allocation and developmental stability. These findings may also have practical implications for maize improvement. Stable sink strength is closely associated with coordinated grain filling, carbon allocation efficiency, and senescence timing, all of which influence yield stability under variable field conditions. The hub genes identified in this study are associated with pathways involved in storage reserve biosynthesis, stress responses, and hormonal regulation, suggesting their potential relevance for breeding strategies aimed at maintaining balanced assimilate partitioning and delaying premature senescence. Such insights may contribute to improving grain filling stability, enhancing stress resilience, and supporting mechanical harvesting suitability through improved developmental synchronization. Conclusion This study shows that pollination prevention is associated with premature senescence in maize and coincides with a marked reduction in sink strength. A major divergence in transcriptional reprogramming was observed from 14 to 18 days after silking, corresponding to key physiological and developmental changes. We identified significant alterations in carbohydrate metabolism and elevated oxidative stress in NONPOL plants. Furthermore, gene co-expression analysis revealed key hub genes, including MYB and HSPs, that may play central roles in coordinating storage reserve accumulation and stress responses. Further functional studies of these genes will help to clarify the molecular mechanisms that underlie this regulatory network. On the basis of these findings, we propose a conceptual model in which reduced sink strength alters carbohydrate partitioning and oxidative stress, thereby promoting premature senescence in maize. In summary, this study provides new insight into sink-strength-associated senescence and offers valuable genetic targets for further investigation aimed at improving maize productivity. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Supplementary Material 5. Supplementary Material 6. Supplementary Material 7. Supplementary Material 8. Supplementary Material 9. Supplementary Material 10. Supplementary Material 11. Supplementary Material 12. Supplementary Material 13. Supplementary Material 14. Supplementary Material 15. Supplementary Material 16. Supplementary Material 17. Acknowledgements The Authors are thankful to Henan Agricultural University and Henan Academy of Agricultural Sciences for supporting this study. Abbreviations Non-pollinated Normally pollinated Days after silking Malondialdehyde Superoxide dismutase Catalase The blade of the ear leaf The blade of the first leaf above the ear leaf The blade of the first leaf below the ear leaf The sheath of the ear leaf The first sheath above the ear leaf The first sheath below the ear leaf The internode above the primary ear-bearing node Ear-bearing internode The first internode above the EBI The first internode below the EBI Authors' contributions L. Ku and L. Wu conceived and designed the project. Experiments were performed by L. Tian, P. Li, Y. Liu, S. Liu, J. Zhang, M. Xu, S. Xu, M. Li, Z. Zhang, J. Mu, and X. Liu. Data analysis was carried out by L. Tian, P. Li, Y. Liu, S. Liu, J. Zhang, M. Xu, S. Xu, M. Li, Z. Zhang, and W. Dong. The manuscript was written by L. Tian and P. Li. The project was supervised by Y. Chen, L. Ku, and L. Wu. All authors revised the manuscript and approved the final version. Funding This research was supported by the Science and Technology Innovation Fund of Henan Agricultural University (2023CXZX008) and the National Key Research and Development Program of China (2016YFD0101205-3). Data availability Sequence data that support the findings of this study have been deposited in the NCBI repository, with the primary accession code PRJNA1347479 (https://www.ncbi.nlm.nih.gov/sra/PRJNA1347479). Materials are available from the corresponding author upon reasonable request. Declarations Ethics approval and consent to participate The study does not involve any human participants or animals. The materials and methods comply with relevant institutional, national, and international guidelines and legislation. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Lei Tian and Pei Li contributed equally to this work. Contributor Information Yanhui Chen, Email: chy9890@163.com. Lixia Ku, Email: kulixia0371@163.com. Liancheng Wu, Email: wuliancheng998@henau.edu.cn. References Associated Data Data Availability Statement Sequence data that support the findings of this study have been deposited in the NCBI repository, with the primary accession code PRJNA1347479 (https://www.ncbi.nlm.nih.gov/sra/PRJNA1347479). Materials are available from the corresponding author upon reasonable request.