Nuclear factor-Y–polycomb repressive complex2 dynamically orchestrates starch and seed storage protein biosynthesis in wheat

Abstract

The endosperm in cereal grains is instrumental in determining grain yield and seed quality, as it controls starch and seed storage protein (SSP) production. In this study, we identified a specific nuclear factor-Y (NF-Y) trimeric complex in wheat (Triticum aestivum L.), consisting of TaNF-YA3-D, TaNF-YB7-B, and TaNF-YC6-B, and exhibiting robust expression within the endosperm during grain filling. Knockdown of either TaNF-YA3 or TaNF-YC6 led to reduced starch but increased gluten protein levels. TaNF-Y indirectly boosted starch biosynthesis genes by repressing TaNAC019, a repressor of cytosolic small ADP-glucose pyrophosphorylase 1a (TacAGPS1a), sucrose synthase 2 (TaSuS2), and other genes involved in starch biosynthesis. Conversely, TaNF-Y directly inhibited the expression of Gliadin-γ-700 (TaGli-γ-700) and low molecular weight-400 (TaLMW-400). Furthermore, TaNF-Y components interacted with SWINGER (TaSWN), the histone methyltransferase subunit of Polycomb repressive complex 2 (PRC2), to repress TaNAC019, TaGli-γ-700, and TaLMW-400 expression through trimethylation of histone H3 at lysine 27 (H3K27me3) modifications. Notably, weak mutation of FERTILIZATION INDEPENDENT ENDOSPERM (TaFIE), a core PRC2 subunit, reduced starch but elevated gliadin and LMW-GS contents. Intriguingly, sequence variation within the TaNF-YB7-B coding region was linked to differences in starch and SSP content. Distinct TaNF-YB7-B haplotypes affect its interaction with TaSWN-B, influencing the repression of targets like TaNAC019 and TaGli-γ-700. Our findings illuminate the intricate molecular mechanisms governing TaNF-Y–PRC2-mediated epigenetic regulation for wheat endosperm development. Manipulating the TaNF-Y complex holds potential for optimizing grain yield and enhancing grain quality.

A protein complex intricately orchestrates the temporal expression patterns of starch biosynthesis genes and seed storage protein-encoding genes through epigenetic regulation in wheat.

Introduction

Wheat (Triticum aestivum L.) stands as one of the world's foremost staple crops, fulfilling a substantial portion of the global dietary requirements (Shewry and Hey 2015). Securing robust grain yield and exceptional quality in wheat is crucial for food security (Xiao et al. 2022). The primary goal of enhancing crops lies in optimizing grain yield and quality, which hinges on seed storage protein (SSP) and starch accumulation within the endosperm.

The composition of gluten proteins, including high molecular weight glutenin subunit (HMW-GS), low molecular weight glutenin subunit (LMW-GS), and gliadins, substantially impacts the dough quality and end-use characteristics of wheat flour (Liu et al. 2005; Dong et al. 2010; Biesiekierski 2017). Regulating gluten protein accumulation involves a complex interplay among a network of genes and transcription factors (TFs). Additionally, TFs like Dof (DNA-binding with one finger) proteins and bZIP (basic leucine zipper) factors, among other TFs, contribute to modulating gluten protein synthesis, ensuring precise spatiotemporal control during seed development.

The pathway starch biosynthesis is intricate, governed by a series of pivotal enzymes. Notable among these are UDP-glucose pyrophosphorylase (UGPase), cytosolic ADP-glucose pyrophosphorylase (cAGPase), granule-bound starch synthases (GBSS), starch synthases (SS), Sucrose synthase (SuS) and starch branching enzymes (SBE), each contributing to the orchestration of starch synthesis (Liu et al. 2013; Pfister and Zeeman 2016). ADP-glucose, the essential substrate for starch biosynthesis, enters the amyloplast via the ADP-glucose transporter named BRITTLE1 (BT1) (Bahaji et al. 2014). Starch synthesized as semicrystalline granules in wheat endosperm exhibits distinct A- and B-type granules. A-type granules initiate around 6 to 8 days after pollination (DAP) in wheat and grow throughout development, while B-type granules initiate later, around 15 to 20 DAP. Key regulators for A-type granules in wheat include STARCH SYNTHASE4 (SS4) and B-GRANULE CONTENT1 (BGC1), while B-type granules are influenced by BGC1 and plastidial alpha-glucan phosphorylase (PHS1) (Chia et al. 2020; Hawkins et al. 2021; Kamble et al. 2023). Concurrently, enzymes involved in starch degradation, such as amylases, hold crucial sway over starch content and quality during the grain-filling phase (Subasinghe et al. 2014; Cuesta-Seijo et al. 2017).

Several TFs regulate starch biosynthesis in wheat, including TabZIP28 (Wheat basic leucine zipper family TF 28), TaRSR1 (Wheat homolog of Rice Starch Regulator1), and TaNAC019 (Wheat NAM [No Apical Meristem]–ATAF [Arabidopsis Transcription Activation Factor]–CUC [Cup-shaped Cotyledons] family TF 019) (Xiao et al. 2022). Knockdown of TabZIP28 reduces cytoplasmic AGPases, decreasing total starch content by approximately 4% (Song et al. 2020). TaRSR1 temporally regulates starch biosynthesis-related enzymes (Liu et al. 2016). Liu et al. (2020) reported that overexpressing TaNAC019 reduces grain weight and starch content, but another study suggests it positively regulates starch synthesis and storage protein accumulation (Gao et al. 2021). The exact role of TaNAC019 in wheat starch synthesis and gluten accumulation therefore remains uncertain.

The NF-Y (nuclear factor Y) complex, comprising NF-YA, NF-YB, and NF-YC subunits, is a conserved transcription factor crucial for seed development in different plant species (Petroni et al. 2012). This trimeric complex binds to the CCAAT-box cis-element in target genes’ promoters (Laloum et al. 2013; Gnesutta et al. 2017; Nardone et al. 2017). NF-YB and NF-YC subunits feature conserved histone-fold (HF) domains resembling histone 2B (H2B) and histone 2A (H2A), respectively. In yeast and animals, NF-Y modifies chromatin structure by replacing H2A-H2B or introducing histone modifications like methylation and acetylation (Luger et al. 1997; Dolfini et al. 2012). NF-Y's influence on embryo development, seed filling, and starch synthesis in rice (Oryza sativa) is well-documented (Xu et al. 2016; Bello et al. 2019; Feng et al. 2022). In maize (Zea mays), it regulates endosperm development, impacting grain size and weight (Zhang et al. 2022). While NF-Y's specific role in wheat seed development is still under investigation, early findings suggest its involvement in grain filling and seed quality regulation (Yadav et al. 2015; Liu et al. 2023). How NF-Y coordinates starch and/or storage protein in cereals remains unclear.

Epigenetic regulation, including histone modifications, plays a crucial role in seed development, especially in early endosperm stages (Ding et al. 2022). POLYCOMB REPRESSIVE COMPLEX2 (PRC2) and its catalyzed mark, trimethylation of histone H3 at lysine 27 (H3K27me3), are key players in this process. They dynamically regulate gene expression during the initiation of endosperm proliferation and cellularization, ensuring proper development and contributing to grain filling and yield. This mechanism is well-studied across various plant species like rice, maize, wheat, and Arabidopsis thaliana (Gutierrez-Marcos et al. 2013; Zhang et al. 2018; Ni et al. 2019; Cheng et al. 2020, 2021; Tonosaki et al. 2021; Zhang et al. 2023a, 2023b). However, their role in later endosperm processes such as starch biosynthesis and gluten protein accumulation, particularly in wheat, remains largely unexplored.

Here, we explored the function of an endosperm-expressed TaNF-Y complex in regulating starch biosynthesis and gluten protein accumulation in wheat. Our findings demonstrate that this regulation involves TaNF-Y mediated recruitment of PRC2 and direct deposition of H3K27me3 on gluten protein coding genes, along with indirect modulation of starch biosynthesis by repression of TaNAC019. Moreover, natural variations in the TaNF-YB7-B impact starch and protein content by affecting its interaction with PRC2 and altering the expression of targets genes.

Results

Presence of endosperm highly expressed TaNF-Y in wheat

During endosperm development, starch and seed storage protein (SSP) biosynthesis are temporally regulated (Wei et al. 2009; Yin et al. 2012). Through a time-serial transcriptome analysis (Zhao et al. 2024), we observed an earlier peak in expression of starch biosynthesis related genes around 6 to 8 DAP, contrasting with the broader and later expression window of various SSP genes spanning from 6 to 16 DAP (Fig. 1A, B and Supplementary Data Set 1). We are intrigued by how this temporal expression pattern is orchestrated by various TFs. Notably, known regulators such as TaNAC019 and Wheat prolamin-box binding factor (WPBF) exhibit varied dynamic expression patterns throughout the developmental period from 0 DAP to 22 DAP (Fig. 1B). Additionally, a component of the TaNF-Y complex, TaNF-YA3, displays a transcriptional profile similar to starch biosynthesis genes, as indicated by the transcriptome data (Fig. 1B).

Figure 1.

Identification of endosperm-expressed trimeric TaNF-Y complex in wheat. A) The expression heatmap of starch and SSP related genes during endosperm development in wheat. The color of heatmap represents the normalized TPM value of genes expression. The normalized expression levels was indicated by a color bar. SSP, seed storage proteins; TPM, transcripts per kilobase of exon model per million mapped reads. DAP, days after pollination. B) The percentile transcript level of TaNF-YA3, TaPBF, TaNAC019 and a full set of starch synthesis genes (Starch), and SSP synthesis genes (SSP) during the course of endosperm development. The relative expression of each gene compared to TaActin was calculated. Reported as mean ± SD of three biological replicates. C) The expression heatmap of TaNF-Y components genes during endosperm development in wheat. The normalized expression levels was indicated by a color bar. D) Mass spectrometric (MS) identification of TaNF-YA3 interacting proteins from the 10 DAP endosperm. IgG was used as control (CK). Two biological repeats (Rep1 and Rep2) were performed. Total numbers of identified unique peptides for each protein are indicated. E) Y2H assay showing the interaction between TaNF-YA3-D and TaNF-YC6-B, TaNF-YB7-B and TaNF-YC6-B. SD-L/-T/-H, (SD-Leu/-Trp/-His, 0.5 mm 3AT); SD-L/-T, (SD-Leu/-Trp). Ad, pGADT7; BD, pGBKT7; SD, Synthetic Dropout Medium; 3 AT, 3-amino-1,2,4-triazole. F) BiFC assay of the interaction between TaNF-YA3-D and TaNF-YC6-B, TaNF-YB7-B and TaNF-YC6-B in Nicotiana benthamiana leaves. H2B-mCherry was used as nuclei marker. Scale bar, 0.5 µm. G) GST pull-down assay showing interaction between TaNF-YA3-D, TaNF-YB7-B, and TaNF-YC6-B.

In hexaploid wheat, we identified a total of 19 NF-YA, 36 NF-YB, and 18 NF-YC subunits through sequence similarity alignment (Supplementary Fig. S1A, Supplementary File 1 to S3). These coding triads generally exhibit similar transcriptional patterns across different tissues, showcasing balanced expression levels (Supplementary Fig. S1, B and C). Notably, certain TaNF-Y coding genes exhibit high expression during grain and spike development (Supplementary Fig. S1B). Specifically, TaNF-YA3-D (TraesCS4D02G289600), TaNF-YB7-B (TraesCS7B02G121900), and TaNF-YC6-B (TraesCS6B02G185700) are specifically expressed during 4 to 12 DAP of seed development (Fig. 1C), with no reported functional information regarding their putatively orthologues in other cereals (Xu et al. 2016; Zhang et al. 2022). To investigate whether TaNF-YA3-D, TaNF-YB7-B, and TaNF-YC6-B form a complex in vivo, we conducted immunoprecipitation coupled with mass spectrometry (IP-MS) using proteins purified with the TaNF-YA3 specific antibody from 10 DAP wheat endosperm. Mass spectrometry (MS) analysis detected peptides of TaNF-YB7-B/D and TaNF-YC6-B/D, indicating the presence of a complex comprising TaNF-YA3-D, TaNF-YB7-B/D, and TaNF-YC6- B/D endogenously (Fig. 1D, Supplementary Fig. S1, D to F, Supplementary Data Set 2). The interaction between TaNF-YA3-D/TaNF-YC6-B, and TaNF-YB7-B/TaNF-YC6-B within nuclei was further confirmed through yeast two-hybridization (Y2H) (Fig. 1E) and bi-fluorescence complementation (BiFC) assays (Fig. 1F), aligning with their subcellular localization (Supplementary Fig. S1G). Contrary to findings in other species (Hou et al. 2014), no direct interaction between TaNF-YA3-D and TaNF-YB7-B was observed (Fig. 1G). Remarkably, the addition of TaNF-YC6-B facilitated the interaction between TaNF-YA3-D and TaNF-YB7-B via an in vitro pull-down assay (Fig. 1G), suggesting their potential to form a trimeric complex. Thus, we identified a dynamically expressed TaNF-Y complex in the endosperm of hexaploid wheat.

TaNF-Y affects starch synthesis and gluten protein composition

Wheat grain filling begins during 6 to 12 DAP, characterized by the synthesis of starch and storage proteins (Zhang et al. 2021). Given TaNF-Y's spatiotemporal expression in endosperm during this stage, we speculate its role in regulating starch and SSP biosynthesis. To investigate, we generated multiple knockdown lines of TaNF-YA3 and TaNF-YC6 using an RNA interference (RNAi) strategy, with varying dosages (Supplementary Fig. S2, A to C) (Borrill et al. 2015; Sestili et al. 2019). The expression levels of other members in the NF-YA and NF-YC families remained largely unchanged in the TaNF-Y-RNAi lines (Supplementary Fig. S2D). In the T₂ generations, we observed a significant reduction in grain length (GL), grain width (GW), and thousand-grain weight (TGW) in multiple TaNF-YA3-RNAi and TaNF-YC6-RNAi lines (Fig. 2, A to C). As starch is a primary storage material in seeds (Toepfer et al. 1972), we further assessed starch content. Consistent with the smaller seed size, both total starch and amylose content were significantly reduced in TaNF-YA3-RNAi and TaNF-YC6-RNAi lines (Fig. 2, D and E). Coomassie brilliant blue staining of semi-thin sections revealed fewer starch granules (SG) in TaNF-YA3-RNAi and TaNF-YC6-RNAi lines compared to Fielder, a wheat variety easily genetically transformable, which serves as a “wild-type” for controlling transgenic wheat lines (Fig. 2F). Scanning electron microscopy (SEM) revealed smaller size of A-type SGs and a reduced number of B-type SGs in TaNF-YA3-RNAi and TaNF-YC6-RNAi lines compared to Fielder (Supplementary Fig. S3, A to C). RT-qPCR analysis indicated reduced transcription levels of both TaBGC1 and TaSS4 in TaNF-YA3-RNAi and TaNF-YC6-RNAi lines compared to Fielder at 6 DAP (Supplementary Fig. S3D). At 15 DAP, TaPHS1 expression decreased in both lines, while TaBGC1 decreased only in TaNF-YA3-RNAi (Supplementary Fig. S3E). These results suggest that both TaNF-YA3 and TaNF-YC6 regulate A- and B-type starch granules initiation by positively regulating TaBGC1, TaSS4 and TaPHS1 with functional differentiation between TaNF-YA3 and TaNF-YC6. TaNF-YC6's regulation of TaBGC1 is temporally restricted to early grain development but still influences B-type granule initiation through TaPHS1. This aligns with the reduced number of B-type granules in both TaNF-YA3-RNAi and TaNF-YC6-RNAi lines.

Figure 2.

Knockdown of TaNF-YA3 or TaNF-YC6 influences starch biosynthesis and SSP composition. A) Grain morphology of TaNF-YA3 and TaNF-YC6 multiple knockdown lines, and the control transgene-null line (Fielder). Scale bar, 1 cm. B–F) Quantification of grain length (GL) and grain width (GW) (B), thousand grain weight (TGW) (C), total starch content (D), amylose content (E), and starch granules (F) in TaNF-YA3-RNAi, TaNF-YC6-RNAi lines and Fielder. Grain morphology traits B, C) were determined using a camera-assisted phenotyping system with six biological replicates, each with approximately 10 g seeds. The percentage of starch content in grain D, E) represents mg per 100 mg dry wheat flour. The starch granules were stained with Coomassie Brilliant Blue in semi-thin section of 10 DAP seeds (F). The triangles represent starch granules, the number was manually calculated per 75 μm². Scale bar, 100 µm. The data are shown as mean ± SD (n = 10). Statistical significance was determined by Student's t test. *, P < 0.05; **, P < 0.01. DAP, days after pollination; SG, starch granules. G–I) Quantification of total protein content (G), HMW-GS, LMW-GS (H), and gliadins contents (I) in TaNF-YA3-RNAi, TaNF-YC6-RNAi lines and Fielder. The percentage of protein content in grain was estimated by Kjeldahl method, which represents mg per 100 mg dry wheat flour. The data are shown as mean ± SD (n = 3). Statistical significance was determined by Student's t test. *, P < 0.05; **, P < 0.01; ns, no significant difference. HMW-GS, high molecular weight glutenin subunit; LMW-GS, low molecular weight glutenin subunit.

Moreover, the knockdown of TaNF-YA3 and TaNF-YC6 function was associated with an increase in total protein content (Fig. 2G). We also performed reverse-phase high-performance liquid chromatography (RP-HPLC) to measure the content of HMW-GS and LMW-GS in TaNF-Y knockdown lines. The levels of HMW-GS were similar in the TaNF-YA3-RNAi and TaNF-YC6-RNAi lines compared to Fielder (Fig. 2H). Whereas, the LMW-GS levels were increased only in TaNF-YA3-RNAi-1 and TaNF-YC6-RNAi-1 (Fig. 2H). Conversely, the levels of γ and α/β gliadin subunits (Gli-γ and Gli-α/β) increased in multiple TaNF-YA3-RNAi and TaNF-YC6-RNAi lines compared to Fielder (Fig. 2I). Additionally, the plant height decreased in TaNF-YA3-RNAi and TaNF-YC6-RNAi plants relative to Fielder, while the tiller number was slightly reduced only in TaNF-YC6-RNAi-1 (Supplementary Fig. S2, E to G). However, transgene-free lines did not differ from Fielder in TaNF-Y expression or grain size (Supplementary Fig. S2, H to J). Thus, TaNF-Y is involved in regulating starch synthesis and gluten protein composition during endosperm development.

TaNF-Y directly represses the transcription of gluten coding genes

To elucidate the impact of TaNF-Y on starch biosynthesis and gluten protein composition, we conducted transcriptome analysis comparing the TaNF-YA3-RNAi-1 and TaNF-YC6-RNAi-1 lines to Fielder at 10 DAP, with three biological replicates (Supplementary Fig. S4A). We identified 5,714 and 3,989 differentially expressed genes (DEGs) in the TaNF-YA3-RNAi-1 and TaNF-YC6-RNAi-1 lines relative to Fielder, respectively (Supplementary Fig. S4B and Supplementary Data Set 3). A significant overlap of DEGs emerged, highlighting the functional interplay between TaNF-YA3 and TaNF-YC6 (Fig. 3A). We focused on these overlapped DEGs for further analysis (Supplementary Data Set 3), where 988 genes were up-regulated. Considering the elevated levels of gliadins and LMW-GS proteins in TaNF-YA3 and TaNF-YC6 knockdown lines, we scrutinized storage protein coding genes including gliadins, HMW-GS and LMW-GS, finding their expression was increased in both TaNF-YA3-RNAi-1 and TaNF-YC6 RNAi-1 (Fig. 3B). This finding was further corroborated individually through RT-qPCR across multiple TaNF-YA3-RNAi and TaNF-YC6-RNAi lines (Fig. 3D). Additionally, among 107 expressed starch-related genes with TPM ≥ 1 (transcripts per kilobase of exon model per million mapped reads) in at least one sample, 64 genes were down-regulated (such as genes encoding UGPase, cAGPase, SSII, SSIII, SuS, and BT1), while 43 genes up-regulated in TaNF-Y-RNAi lines such as genes encoding Isoamylase 1 (ISA1), pullulanase (PULL) and Sucrose phosphate synthase (SPS) (Fig. 3C). Noteworthy among the 970 down-regulated genes were cytosolic small ADP-glucose pyrophosphorylase 1a (TacAGPS1a), cytosolic large ADP-glucose pyrophosphorylase 1 (TacAGPL1), and TaGBSSII, with a significant enrichment in Gene Ontology (GO) terms associated with starch biosynthetic and metabolic process (Fig. 3, A and C and Supplementary Fig. S4C). RT-qPCR validation confirmed the down-regulation of these genes in both TaNF-YA3-RNAi and TaNF-YC6-RNAi lines (Fig. 3D).

Figure 3.

TaNF-Y directly represses the transcription of gluten coding genes to regulate SSP composition. A) The Venn-diagram showing the overlapping of up-regulated (lower panel) and down-regulated (upper panel) DEGs (compared to Fielder) between TaNF-YA3-RNAi-1 and TaNF-YC6-RNAi-1. DEGs, differential expressed genes. P value is calculated by fisher exact test. B) Heatmap showing the expression level of SSP coding genes in TaNF-YA3-RNAi-1, TaNF-YC6-RNAi-1 and Fielder. The normalized expression levels was indicated by a color bar. SSP, seed storage proteins. C) Heatmap showing the expression level of the full sets of starch biosynthesis genes in TaNF-YA3-RNAi-1, TaNF-YC6-RNAi-1 and Fielder. The normalized expression levels was indicated by a color bar. D) The relative expression level of TaGli-α, TaGli-γ-700, TaLMW-400, TacAGPL1, TacAGPS1a and TaGBSSII in the TaNF-YA3-RNAi and TaNF-YC6-RNAi lines as compared to Fielder, estimated by RT-qPCR and normalized to TaActin, the level in Fielder was set as 1. Data were shown as mean ± SD of three biological replicates. Statistical significance was determined between transgene lines with Fielder by Student's t test. *, P < 0.05; **, P < 0.01. E) The diagram illustrates DEGs potentially regulated by TaNF-Y. The 1,958 DEGs were identified in both TaNF-YA3 and TaNF-YC6-RNAi lines; The 960 potential targets of TaNF-Y were identified through ATAC-seq and motif analysis; Among them, 56 genes were related to SSP and starch biosynthesis. Up or Down denote the direction of expression level changes in TaNF-Y knockdown lines compared to Fielder. F) IGV screenshot shows the accessible chromatin region of TaGli-γ-700 and TaLMW-400 as indicated by ATAC-seq. The CCAAT motif was indicated with vertical dash line and represented the position of TaNF-Y binding on the promoter of targets. Data is from endosperm tissue at 12 DAP. DAP, days after pollination. FPKM, fragments per kilobase of exon model per million mapped fragments. G) The EMSA validated TaNF-Y complex (formed by co-expressing TaNF-YA3-D-His, GST-TaNF-YB7-B, and MBP-TaNF-YC6-B) directly interact with promoter containing CCAAT motifs of TaGli-γ-700 and TaLMW-400. In the assay, “+” and “-” denote the presence and absence of the probe or protein, respectively. The term “Shift” denotes the protein-bound DNA probe bands, while “Free probe” indicates the unbound DNA probe bands. H) The dual luciferase reporter assay showed TaNF-Y complex suppress the expression of TaGli-γ-700 and TaLMW-400. The Relative LUC/REN value represents the ratio of the signal detected for firefly luciferase (LUC) to Renilla reniformis luciferase (REN) activity. The data presented are the mean ± SD of six biological replicates. Statistical significance was determined by Student's t test. **, P < 0.01.

Our next objective was to identify potential direct targets of TaNF-Y among the DEGs. Leveraging previously generated chromatin accessibility data from the endosperm at 6, 8, and 12 DAP (Zhao et al. 2024), we searched for DEGs containing CCAAT motifs within proximal open chromatin regions. This analysis identified 960 genes (Supplementary Data Set 4), comprising 514 up-regulated and 446 down-regulated in TaNF-Y-RNAi lines. Notably, among these were 41 gluten protein coding genes (all up-regulated genes, such as Gliadin γ-subunit 700 (TaGli-γ-700) and low molecular weight-400 (TaLMW-400)), along with 15 starch biosynthesis genes (six up-regulated and nine down-regulated) (Fig. 3, E and F). We further validated the direct binding of TaNF-Y to the promoters of specific gluten coding genes, such as TaGli-γ-700 and TaLMW-400, through Electrophoretic Mobility Shift Assay (EMSA) (Fig. 3G). Interestingly, TaNF-YA3-D alone did not bind to the promoters of TaGli-γ-700 and TaLMW-400, nor did other components like TaNF-YB7-B and TaNF-YC6-B (Supplementary Fig. S5A), consistent with previous findings in A. thaliana (Siriwardana et al. 2016). However, all three components of the TaNF-Y trimeric were required for binding to the promoters of TaGli-γ-700 and TaLMW-400 (Fig. 3G), underscoring the importance of the TaNF-Y complex's integrity. Furthermore, TaNF-YA3-D and TaNF-YC6-B exhibited transcriptional repressive activity in the dual-luciferase (LUC) reporter assay in wheat protoplast (Supplementary Fig. S5B). The LUC reporter assay in Nicotiana benthamiana leaves further confirmed the repression of TaNF-Y complex on TaGli-γ-700 and TaLMW-400 (Fig. 3H). Thus, TaNF-Y directly represses gluten coding genes to modulate the composition of SSP in wheat.

Although a few starch biosynthesis genes were identified as DEGs with CCAAT motifs in their open promoter regions (Fig. 3E), we did not observe a shift bond in the EMSA for certain genes like TacAGPS1a (Supplementary Fig. S5C). Moreover, starch biosynthesis genes generally exhibited down-regulation in TaNF-YA3 and TaNF-YC6 attenuation lines (Fig. 3C), and TaNF-Y is likely acting as a repressor of the targets (Supplementary Fig. S5B). These findings suggest that TaNF-Y may function distinctly in regulating the transcription of starch biosynthesis and SSP coding genes.

TaNF-Y indirectly promotes starch biosynthesis by repressing TaNAC019

We speculate that TaNF-Y may facilitate starch biosynthesis indirectly by inhibiting certain repressors of starch biosynthesis genes. One potential targets of TaNF-Y is TaNAC019, which has been reported to negatively regulate starch biosynthesis in wheat (Liu et al. 2020), although another study shows slight positive regulation of starch biosynthesis by TaNAC019 at 24 DAP (Gao et al. 2021). Interestingly, TaNAC019-A/B/D show lagged activation pattern following the decline of TaNF-YA3-B/D during endosperm development (Supplementary Fig. S6A). Furthermore, TaNAC019-A/B/D are up-regulated in either TaNF-YA3-RNAi or TaNF-YC6-RNAi lines at 10 DAP (Fig. 4A). Additionally, the CCAAT motif is present in the proximal open chromatin region of TaNAC019-B1 at 12 DAP (Supplementary Fig. S6B). We further confirmed the directly binding of TaNF-Y to the promoter of TaNAC019-B1 via EMSA (Fig. 4B), as well as the repression of TaNF-Y on TaNAC019-B1 using a LUC reporter assay in Nicotiana benthamiana leaves (Fig. 4C). Importantly, this binding and transcriptional repression are dependent on the presence of CCAAT motif (Fig. 4, B and C). Therefore, TaNAC019-B1 is indeed a direct inhibited target of TaNF-Y.

Figure 4.

TaNF-Y directly inhibits TaNAC019 expression and indirectly promotes starch biosynthesis. A) The relative expression level of TaNAC019 triads in Fielder and TaNF-YA3-RNAi-1 and TaNF-YC6-RNAi-1 lines. The results are presented as the mean ± SD of three biological replicates. Statistical significance was determined by Student's t test. **, P < 0.01. B) The EMSA showing direct binding of TaNF-Y complex (TaNF-YA3-D-His/GST-TaNF-YB7-B/MBP-TaNF-YC6-B) to TaNAC019-B1 promoter sequence. In the assay, “+” and “−” denote the presence and absence of the probe or protein, respectively. The term “Shift” denotes the protein-bound DNA probe bands, while “Free probe” indicates the unbound DNA probe bands. C) Dual luciferase reporter assay validation of transcriptional inhibition of TaNF-Y complex to TaNAC019-B1. Mutation of the CCAAT motifs (TaNAC019-B1^mu-Promoter) was introduced in the promoter region of TaNAC019-B1. The data presented are the mean ± SD of six biological replicates. Statistical significance was determined by Student's t test. **, P < 0.01; ns, no significance. D) Quantification of starch granules in Fielder and Tanac019-cr seeds at 6 DAP. The semi-thin sections were stained with Coomassie Brilliant Blue, the triangles represent starch granules, the number was manually calculated per 75 μm², scale bar, 100 µm. The data are shown as mean ± SD (n = 10). Statistical significance was assessed using Student's t test. *, P < 0.05. SG, starch granules. E) Heatmap showing the expression level of the full set of starch biosynthesis genes of developing endosperm at 6 DAP in Tanac019-cr and Fielder. The normalized expression levels was indicated by a color bar. DAP, days after pollination. F) The relative expression level of TacAGPL1, TacAGPS1a, TaGBSSII, TaSSIIIa, and TaGBSSII in the Tanac019-cr as compared to Fielder, estimated by RT-qPCR and normalized to TaActin. Data were shown as mean ± SD of three biological replicates. Statistical significance was determined by Student's t test. *, P < 0.05; **, P < 0.01. G) The EMSA showed TaNAC019-B1 directly bind to the TacAGPS1a and TaSuS2 promoter. In the assay, “+” and “−” denote the presence and absence of the probe or protein, respectively. The term “Shift” denotes the protein-bound DNA probe bands, while “Free probe” indicates the unbound DNA probe bands. H) Dual luciferase reporter assay validation of transcriptional inhibition of TaNAC019-B1 to TacAGPS1a and TaSuS2. The data presented are the mean ± SD of six biological replicates. Statistical significance was determined by Student's t test. **, P < 0.01. I) The expression pattern of the TaNAC019-B1, TaGli-γ-700 and TaLMW-400 during endosperm development in TaNF-YA3-RNAi-1, TaNF-YC6-RNAi-1 and Fielder. The vertical dash line indicates the temporal peak expression of each gene during endosperm development in Fielder. Data were shown as mean ± SD of three biological replicates. Statistical significance was determined by Student's t test. *, P < 0.05; **, P < 0.01.

We thoroughly examined the starch contents in the Tanac019-cr mutant, a knock-out mutant in the Fielder variety background created via Clustered regularly interspaced palindromic repeats (CRISPR)/Cas9 genome editing (Gao et al. 2021), and compared it to Fielder. Using semi-thin sections, we observed that the Tanac019-cr mutant exhibited a higher starch content than Fielder during the early stages of endosperm development, such as at 6 DAP (Fig. 4D). We further performed RNA-seq analysis on the 6 DAP endosperm of Tanac019-cr and Fielder, and a total of 1,559 DEGs, including 934 up- and 625 down-regulated genes, were identified (Supplementary Fig. S6C and Supplementary Data Set 5). A comprehensive GO enrichment analysis of the up-regulated DEGs in the Tanac019-cr mutant revealed significant enrichment in pathways related to starch synthesis (Supplementary Fig. S6D). Moreover, several starch synthesis genes were up-regulated in Tanac019-cr, including TacAGPS1a, TacAGPL1, TaGBSSII, TaSSIIIa, and TaSuS2 (Figure 4E), which was further validated by RT-qPCR (Fig. 4F). These findings suggest that TaNAC019 may repress the expression of these genes, as previously reported (Liu et al. 2020). Taken together, these results indicate that TaNF-Y could regulate the expression of starch biosynthesis genes by inhibiting TaNAC019.

We further explored how TaNAC019 affects the transcription of starch synthesis genes. Using an EMSA, we confirmed that TaNAC019-B1 can bind to the promoters of TacAGPS1a and TaSuS2 (Fig. 4G). Additionally, results from LUC assay showed that TaNAC019-B1 represses the transcription of TacAGPS1a and TaSuS2 (Fig. 4H). Therefore, TaNAC019 directly represses the expression of starch biosynthesis genes, such as TacAGPS1a and TaSuS2, and thus negatively regulates starch biosynthesis, particularly during the early stages of endosperm development.

Notably, a time-course RT-qPCR assay revealed an early shift in the peak expression time points of SSP coding genes, such as TaGli-γ-700 and TaLMW-400, as well as TaNAC019-B1, in the TaNF-YA3-RNAi-1 and TaNF-YC6-RNAi-1 lines, along with elevated expression levels (Fig. 4I). This further indicates that the TaNF-Y complex regulates not only affects the expression levels but also the temporal patterns of its direct targets during endosperm development.

TaNF-Y interacts with PRC2 and mediates deposition of H3K27me3 at TaNAC019 and gluten coding genes

Furthermore, we aimed to investigate how TaNF-Y functions in repressing gene expression. Interestingly, we observed reverse trends between the accumulation of the H3K27me3 mark at TaNAC019-B1 and gluten coding genes, such as TaGli-γ-700 and TaLMW-400, with their expression levels during different endosperm developmental stages (Fig. 5A). In Arabidopsis, the NF-Y complex was reported to interact with PRC2, the writer complex for H3K27me3, to mediate gene silencing in the context of flowering regulation (Liu et al. 2018b). Several components of the PRC2 complex, including SWINGER (TaSWN) and FERTILIZATION INDEPENDENT ENDOSPERM (TaFIE), are highly expressed during endosperm development (Supplementary Fig. S7A). We investigated the potential protein interaction between PRC2 and the TaNF-Y complex in wheat using Y2H, and BiFC assays (Fig. 5, B and C). TaSWN-B-N mainly interacts with TaNF-YC6-B and TaNF-YB7-B in the nucleus (Fig. 5, B and C). Additionally, the conserved HF domain of TaNF-YB7-B and TaNF-YC6-B contributes significantly to the interaction with TaSWN-B-N. However, the N-terminal (Nt) of TaNF-YB7-B and TaNF-YC6-B can suppress this interaction, which is consistent with the results of Liu et al. (Liu et al. 2018b) (Supplementary Fig. S7B).

Figure 5.

The TaNF-Y interacts with PRC2 to facilitate H3K27me3 deposition for inhibiting TaNAC019-B1 and SSP coding genes. A) IGV screenshot shows the dynamic change of H3K27me3 and the mRNA level of TaNAC019-B1, TaGli-γ-700 and TaLMW-400 during endosperm development. The CCAAT motif was indicated with vertical dash line. The mRNA level was shown as mean ± SD of three biological replicates. DAP, days after pollination; TPM, transcripts per kilobase of exon model per million mapped reads. FPKM, fragments per kilobase of exon model per million mapped fragments. B) Y2H assay showing the interaction between TaNF-YB7-B/TaNF-YC6-Band TaSWN-B-N. SD-L/-T/-H, (SD-Leu/-Trp/-His, 0.5 mm 3 AT); SD-L/-T, (SD-Leu/-Trp). Ad, pGADT7; BD, pGBKT7; SD, Synthetic Dropout Medium; 3 AT, 3-amino-1,2,4-triazole. C) BiFC assay showing the interaction between TaNF-YB7-B/TaNF-YC6-B and TaSWN-B-N in Nicotiana benthamiana leaves. H2B-mCherry was used as nuclei marker. Scale bar, 0.5 µm. D) The meta profile of global H3K27me3 levels in TaNF-YC6-RNAi-1 and Fielder. TSS, transcription start site; TES, transcription end site. E) The proportion of genes containing CCAAT motifs in accessible chromatin regions exhibiting up or down-regulated H3K27me3 (compared to Fielder) in the TaNF-YC6-RNAi-1 line. F) The IGV screenshot showing the H3K27me3 level at TaNAC019-B1, TaGli-γ-700, and TaLMW-400 loci in Fielder and TaNF-YC6-RNAi-1. The CCAAT motif was indicated with vertical dash line. G) CUT&Tag-qPCR was employed to verify H3K27me3 change at the TaNAC019-B1, TaGli-γ-700, and TaLMW-400 loci in TaNF-YA3-RNAi-1, TaNF-YC6-RNAi-1, and Fielder. The CUT&Tag-qPCR targeting genomic regions for each gene were indicated by P1-P4. The H3K27me3 fold change values, normalized to input, were assessed. The horizontal dash line represents relative enrichment equal to 1, as relative enrichment > 1 indicates enrichment of binding, while relative enrichment < 1 indicates relative depletion. TaActin served as a negative control. Data are presented as mean ± SD (n = 3), and statistical significance was determined using Student's t test. *, P < 0.05; **, P < 0.01.

To determine whether such interactions influence the deposition of H3K27me3 at TaNF-Y targets, we conducted Cleavage Under Targets & Tagmentation (Cut&Tag) (Zhao et al. 2023) to obtain whole-genome-wide H3K27me3 profiles in TaNF-YC6-RNAi-1 line and Fielder, given that TaNF-YC6-B interacts with TaSWN-B. We observed a general reduction in H3K27me3 peak coverage and numbers in TaNF-YC6-RNAi-1 compared to Fielder (Supplementary Fig. S8). The mega profile of H3K27me3 at gene body regions showed a decline in TaNF-YC6-RNAi-1 relative to Fielder (Fig. 5D). Genes containing CCAAT motifs exhibited a dramatic reduction in H3K27me3 levels in TaNF-YC6-RNAi-1 (Fig. 5E and Supplementary Data Set 6), including TaNAC019-B1, TaGli-γ-700, and TaLMW-400 (Fig. 5F).

We further performed CUT&Tag-qPCR to quantitatively measure the accumulation of H3K27me3 at TaNAC019-B1, TaGli-γ-700, and TaLMW-400 loci, using endosperm tissue from TaNF-YA3-RNAi and TaNF-YC6-RNAi lines, together with Fielder. A significant reduction in H3K27me3 levels at the regulatory regions and gene bodies of these genes was observed in TaNF-YA3-RNAi-1 and TaNF-YC6-RNAi-1 compared to Fielder (Fig. 5G). These findings suggest that the TaNF-Y complex interacts with PRC2 components and mediates the deposition of H3K27me3 at TaNAC019-B1 and gliadin coding genes, thereby properly regulating their temporal expression patterns during endosperm development.

Mutation of PRC2 influences starch synthesis and gluten protein composition

Considering the significance of H3K27me3 in silencing the expression of TaNAC019-B1 and gliadin coding genes, we conducted further investigations on the endosperm developmental defects using a mutant line for TaFIE, a core PRC2 component. Since PRC2-mediated deposition of H3K27me3 is crucial for wheat embryo development (Zhao et al. 2023), we selected a weak mutant of TaFIE, which carries frameshift mutations in the A and D subgenomes and a two-amino acids deletion in the B subgenome, for further analysis as it produces fertile seeds.

The Tafie-C87 line displayed altered seed size, with decreased GW but increased GL, and a significant reduction in TGW compared to the wild-type KN199 (Fig. 6, A and B). Consistent with the shriveled seed phenotype, the Tafie-C87 line exhibited a significant decrease in total starch content (Fig. 6C) and SG number, as indicated by coomassie brilliant blue staining of semi-thin sections (Fig. 6D and Supplementary Fig. S9A). SEM analysis also revealed a decrease in the diameter of A-type SGs and the number of B-type SGs in mature seeds of Tafie-C87 (Supplementary Fig. S9, B and C), consistent with the down-regulated expression of TaBGC1 and TaPHS1 in Tafie-C87 compared with KN199 at 6 DAP and 15 DAP, respectively (Supplementary Fig. S9D). The slight decrease in gene expression, particularly TaPHS1, observed in the Tafie-C87 mutant compared to TaNF-Y-RNAi lines, aligns with its modest reduction in the number of B-type starch granules. Additionally, the total protein content increased in Tafie-C87 (Fig. 6E). While the level of HMW-GS and LMW-GS were similar to KN199, only the gliadin contents were significantly increased in Tafie-C87 (Fig. 6, F and G).

Figure 6.

Mutation of PRC2 component TaFIE influences starch synthesis and SSP composition. A–G) Grain morphology (A) and quantification of grain size-related traits (B), starch content (C), histological section (D), total protein content (E), gliadins content (F), and GSs content (G) were compared between KN199 and Tafie-C87. Histological analysis and starch granules staining per 75 μm² in panel (D) were conducted using three grains at 15 DAP. Approximately 10 g seeds were used for evaluation of grain morphology traits. Data were shown as mean ± SD (for panel B, n = 6; for panel C and panel E–G, n = 3). Statistical significance was determined by Student's t test. *, P < 0.05; **, P < 0.01, ns, no significance. Starch granules were indicated by triangles; scale bar, 100 µm. %, mg per 100 mg dry wheat flour. H) Volcano plot showing the up- and down-regulated genes in Tafie-C87 as compared to KN199. Known genes are highlighted. DEGs, differential expressed genes; FDR, false discovery rate; FC, fold change. SG, starch granules. I) GO enrichment analysis of down-regulated genes in the Tafie-C87 as compared to KN199. J–L) The relative expression level of TaNAC019(J), gliadins coding genes (K), and starch synthesis related genes (L) in Tafie-C87 and KN199. The relative expression level of each gene was estimated by RT-qPCR using endosperm of grains at 10 days after pollination. Data were shown as mean ± SD of three biological replicates. Statistical significance was assessed by Student's t test. *, P < 0.05; **, P < 0.01.

Consistent with the phenotypical defects, RNA-seq analysis of developing endosperm of KN199 and Tafie-C87 revealed numerous altered gene expression, including the up-regulation of TaNAC019-B1, TaGli-γ-700, TaGli-α-856 (Fig. 6H, Supplementary Data Set 7). Interestingly, among the 107 expressed starch-related genes, 78 were down-regulated in Tafie-C87, including UGPase, cAGPase, Phosphoglucose isomerase (PGI), BT1, and SSIII, while 29 were up-regulated, such as SPS, glucose-6-phosphate/phosphate translocator (GPT), and Phosphoglucomutase (PGM) (Fig. 3C). Additionally, a substantial number of starch degradation genes were down-regulated in Tafie-C87 mutant compared to TaNF-Y-RNAi lines (Supplementary Fig. S10). The down-regulated genes in Tafie-C87 mutant were enriched in starch biosynthesis and catabolic process (Fig. 6I). RT-qPCR assays confirmed the up-regulation of TaNAC019-A/B/D, and SSP-related genes (TaGli-ω-063, TaGli-α-856 and TaGli-γ-700), as well as the down-regulation of starch biosynthesis genes (TacAGPL1, TacAGPS1a, TaGBSSII, and TaSSIIIa) in Tafie-C87 compared to KN199 (Fig. 6, J to L). These findings indicate that despite potential variances from TaNF-Y, PRC2 plays a role in regulating starch synthesis and gluten protein composition through transcriptional control.

Natural variation of TaNF-Y component is associated with starch and protein contents for breeding selection

Grain size and end-use quality are crucial traits that have been subject to domestication and breeding selection (Michel et al. 2019). To assess the contribution of the TaNF-Y components in breeding selection, we conducted natural variation analysis in the genic region of TaNF-YA, TaNF-YB, and TaNF-YC, utilizing the genome sequencing data from the Watkins collection (Cheng et al. 2023). In general, we observed variation in four TaNF-YAs, eight TaNF-YBs, and four TaNF-YCs, associated with seed developmental traits (Supplementary Data Set 8). Among these, nine genes are linked to TGW, four genes to grain protein content, and six genes to grain starch content (Supplementary Data Set 8). In particular, we identified six Single nucleotide polymorphisms (SNPs) in the coding region and downstream of TaNF-YB7-B (TraesCS7B02G121900), defining two haplotypes (Hap1 and Hap2) with the frequency of 43.4% and 56.6%, respectively (Fig. 7A). Notably, significant differences in grain protein and starch content were observed between TaNF-YB7-B-Hap1 and TaNF-YB7-B-Hap2 (Fig. 7B), with TaNF-YB7-B-Hap2 showing higher grain starch content but lower protein content. Interestingly, the frequency of TaNF-YB7-B-Hap2 in modern cultivars (92.5%) was significantly higher than that in landraces (47.9%) (Fig. 7C), indicating that TaNF-YB7-B-Hap2 was positively selected during modern wheat breeding for its favorable grain starch characteristics.

Figure 7.

Natural variation of TaNF-YB7-B influence starch and protein contents. A) Schematic diagram illustrating haplotypes (Hap1 and Hap2) of TaNF-YB7-B in the wheat population, including 768 Watkins landraces and 186 modern cultivars. B) Comparison of TGW, grain protein content, and starch content between TaNF-YB7-B-Hap1 and Hap2. Box-plot elements are defined as: center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; points, outliers. Statistical significance was determined using the Wilcoxon rank-sum test. The numbers in parentheses represent the sample size used for statistical analysis. C) Distribution of different haplotypes of TaNF-YB7-B in landraces and modern cultivars. “n” indicates the sample size. D) Y2H demonstrates the interaction intensity difference between TaNF-YC6-B, TaSWN-B-N, and TaNF-YB7-B-Hap1, TaNF-YB7-B-Hap2, respectively. SD-L/-T/-H, (SD-Leu/-Trp/-His, 0.5 mm 3 AT); SD-L/-T, (SD-Leu/-Trp). Ad, pGADT7; BD, pGBKT7; SD, Synthetic Dropout Medium; 3 AT, 3-amino-1,2,4-triazole. E) The Split-LUC assay measured the interaction intensity difference between TaNF-YC6-B, TaSWN-B-N, TaNF-YB7-B-Hap1, and TaNF-YB7-B-Hap2. the numbers in the circles representing different plasmid combinations. Values represent means ± SD (n = 6), and different letters above each bar indicate significant differences based on one-way ANOVA and Tukey's multiple comparison test at P < 0.05. F) Dual luciferase reporter assay to assess the transcriptional inhibition of different haplotypes of TaNF-YB7-B to TaNAC019-B1, and TaGli-γ-700. The data are shown as mean ± SD (n = 6). Different letters above each bar indicate significant differences by one-way ANOVA and Tukey's multiple comparison test at P < 0.05. G) The expression level of TaNAC019-B1, and TaGli-γ-700 in wheat groups based on TaNF-YB7-B two different haplotypes. The data are shown as mean ± SD. Statistical significance was determined by Student's t test (n = 20 for both Hap1 and Hap 2). TPM, transcripts per kilobase of exon model per million mapped reads.

To investigate how DNA variations within the TaNF-YB7-B coding region contribute to functional diversity, we conducted interaction tests among TaNF-Y trimeric components and its interaction with PRC2 using different coding sequences of TaNF-YB7-B (Fig. 7D). Y2H revealed that TaNF-YB7-B-Hap2 exhibiting higher interaction intensity with TaNF-YC6-B and also interacted more strongly with TaSWN-B than TaNF-YB7-B-Hap1 (Fig. 7D). The higher interaction intensity between TaNF-YB7-B-Hap2 and TaNF-YC6-B, TaSWN-B-N was further confirmed by split luciferase complementation (split-LUC) assay (Fig. 7E). Since TaNF-Y interacts with PRC2 to repress gene expression (Fig. 5) and all three components of TaNF-Y are required for the binding of TaNF-Y to targets (Fig. 4), we evaluated the effect of different coding haplotypes of TaNF-YB7-B on the transcriptional regulation using a LUC reporter assay (Fig. 7F). The results showed that the coding variations in TaNF-YB7-B indeed affect repressive transcriptional activity. The TaNF-YB7-B-Hap2 variant exhibited a higher repression effect on TaNF-Y targets, such as TaNAC019-B1, and TaGli-γ-700 (Fig. 7F). Consistent with these findings, TaNAC019-B1 and TaGli-γ-700 showed significantly higher expression levels in developing grain (20 DAP) of accessions with TaNF-YB7-B-Hap1 compared to TaNF-YB7-B-Hap2 (Fig. 7G), according to population-wide transcriptome (Zhao et al. 2024). This further validates that TaNF-YB7-B-Hap2 increases starch and decreases protein content in endosperm by repressing the expression of TaNAC019-B1 and TaGli-γ-700.

Thus, natural variations in TaNF-Y components indeed play a crucial role in influencing starch biosynthesis and gluten protein composition by modulating transcriptional regulation, and TaNF-YB7-B-Hap2 has been selected during modern breeding process, possibly due to its favorable grain starch content.

Discussion

Balancing high grain yield and superior seed quality remains a persistent challenge in crop improvement endeavors (Michel et al. 2019). This challenge likely arises from the intricate regulation of the production of starch and SSP (Pleijel and Uddling 2012). Achieving this balance necessitates a deep understanding of the molecular mechanisms governing starch and SSP biosynthesis and the identification of regulatory factors that can be manipulated to separately optimize both aspects.

Temporal expression of starch and SSP biosynthesis genes: epigenetic regulation and key TFs

The proper development of starch biosynthesis and gluten protein accumulation in the endosperm is notably reflected in the distinct temporal expression patterns of related genes (Fig. 1). However, the mechanisms underlying the generation of these expression patterns remain largely enigmatic, although several upstream TFs have been identified. A notable observation is the presence of a high level of repressive histone modification, H3K27me3, surrounding SSP-encoding genes before their activation at 6 DAP. Subsequently, as SSP activation occurs during later endosperm development (8 to 16 DAP), H3K27me3 levels decline (Fig. 5). While the causal relationship between H3K27me3 and transcriptional regulation remains elusive, we did observe the activation of key SSP coding genes in a partially TaFIE (PRC2 component) -knockout mutant line (Fig. 6). Notably, the Tafie-C87 mutant line exhibited elevated gliadins and total protein content. Conversely, starch biosynthesis-related genes displayed minimal H3K27me3 coverage throughout endosperm development, indicating a lesser influence of histone modification (Zhao et al. 2024). Instead, TaNAC019, a known repressor of starch biosynthesis (Liu et al. 2020), exhibited an expression pattern opposite to that of starch biosynthetic genes. Our study further validated the direct repression of starch biosynthetic genes by TaNAC019 (Fig. 4). Interestingly, TaNAC019-B1 appeared to be a suitable target for H3K27me3, as H3K27me3 levels declined after 8 DAP while TaNAC019-B1 expression gradually increased (Fig. 5). Similar to SSP coding genes, TaNAC019-B1 was up-regulated in the Tafie-C87 mutant line (Fig. 6).

These findings suggested an association between epigenetic modifications, particularly H3K27me3, and the temporal expression patterns of SSP and starch-related genes, although the precise influence may be direct or indirect. This discovery expands our understanding of the role of PRC2-H3K27me3 in regulating later endosperm development, in addition to its known involvement in the initiation of endosperm proliferation and cellularization (Xiao and Wagner 2015; Zhang et al. 2023a, 2023b, 2023c).

The multifaceted role of the NF-Y complex in starch and SSP biosynthesis

The NF-Y complex has been recognized for its involvement in seed development across different plant species primarily through its transcription factor activity (Laloum et al. 2013). Among the three subunits of the NF-Y complex, NF-YA typically mediates binding to the CCAAT motif, while NF-YB or NF-YC may interact with epigenetic modification factors, playing a crucial role in its function. For instance, NF-YC regulates photomorphogenesis and flowering transition by interacting with the histone deacetylase 15 (HDA15) and histone methyltransferase CURLY LEAF (CLF), respectively (Tang et al. 2017; Liu et al. 2018a, 2018b). In our study, we identified an endosperm core TaNF-Y complex that functions as a repressor by interacting with TaSWN-B and facilitating the deposition of H3K27me3 at target loci (Fig. 5). TaNF-Y directly binds to SSP-encoding genes, ensuring specific enrichment of H3K27me3 by recruiting PRC2 to maintain a repressed transcriptional state of SSP coding genes during early endosperm development. This repressive chromatin status diminishes as TaNF-Y complex levels decline with endosperm development progression, aligning with the activation of SSP coding genes at later stages. Given the challenge of achieving significant phenotypic changes through individual overexpression of TaNF-Y trimer components, as their functionality depends on coordinated interplay (Figs. 3G and 4G), we opted to create knockdown lines for TaNF-YA3 and TaNF-YC6 instead. The transcript levels of SSP synthesis genes consistently increased in TaNF-YA3 and TaNF-YC6 RNAi lines. However, the content of HMW-GS remained unaffected, while LMW-GS increased in only one transgenic line. This discrepancy between gene transcripts and SSP content in TaNF-Y-RNAi lines might be attributed to a protein turnover pattern, indicating variations in protein abundance independent of gene expression (Cao et al. 2022). Notably, not only did TaGli-γ-700 and TaLMW-400 expression levels increase, but their peak expression times also advanced by several days (Fig. 4).

NF-Y facilitates PRC2-mediated H3K27me3-based gene silencing during endosperm development (Fig. 5), but starch synthesis genes shows a weak correlation with H3K27me3 levels (Zhao et al. 2024), indicating a nuanced regulatory network. Our transcriptome analysis contradicts the expected up-regulation in TaNF-Y-RNAi lines (Supplementary Fig. S5B), hinting at indirect regulatory mechanisms. Indeed, we found that the TaNF-Y complex binds to TaNAC019-B1's promoter, recruits H3K27me3 modifications, and inhibits its expression, suggesting a positive role of TaNF-Y in starch synthesis by suppressing TaNAC019-B1 (Figs. 4 and 5). Thus, TaNF-Y directly inhibited the expression of TaNAC019-B1, which displayed a “perfect” lagged activation upon the decline of TaNF-Y expression (Fig. 1) and its expression elevated with an earlier peak when TaNF-Y was knocked down (Fig. 4). However, comprehensive analysis of all genes in starch biosynthesis and starch degradation revealed that while TaNF-Y and TaFIE share common functionality in the regulation of starch synthesis, there are also some aspects where they differ, particularly in starch degradation (Fig. 3C, Supplementary Fig. S10). In summary, our study identifies a core TaNF-Y complex that plays a pivotal role in modulating the levels of protein and starch content. By integrating both direct and indirect pathways, TaNF-Y is intricately involved in the regulation of both starch and SSP in wheat, via PRC2-mediated H3K27me3-based silencing on TaNAC019-B1 and gliadin coding genes.

Decoupling starch and SSP biosynthesis for yield and quality improvements

Natural variations within the components of the TaNF-Y complex present promising opportunities for discovering elite alleles. Indeed, we identified two amino acid variations within the coding region of TaNF-YB7-B, which influence the formation of the TaNF-Y trimeric complex and its interaction with TaSWN-B. Consequently, different haplotypes of TaNF-Y exhibit varied transcriptional repressive activities towards downstream targets, such as TaNAC019-B1 and SSP coding genes. This variability is linked to altered starch and SSP content, ultimately influencing grain yield and seed quality (Fig. 7). Comparing with knockdown lines, natural variations show comparable genetic effects on SSP-related traits (3.38% vs 3.36% to 7.45%) much weaker effects on starch-related traits (1.11% vs 6.51% to 8.63%) (Figs. 6C and 7B). TaNF-YB7-B-Hap2, with higher starch content but a lower protein content, is more common in modern cultivars than in landraces (Supplementary Fig. S11). Previous research has highlighted the selection of TaNAC019-B1-Hap2 during modern wheat breeding (Liu et al. 2020; Gao et al. 2021). Interestingly, TaNAC019-B1-Hap2 and TaNF-YB7-B-Hap2, both linked to higher starch content but lower SSP content, were positively selected together during wheat breeding process (Supplementary Fig. S10, Supplementary Data Set 9), indicating a prioritization of yield over SSP content in modern wheat breeding. This haplotype analysis underscores their significant roles in shaping grain traits and their joint positive selection in modern wheat breeding.

The distinct regulation pathways for starch and SSP by the TaNF-Y-PRC2 module provide a unique avenue for decoupling these two processes. By manipulating the temporal expression patterns of TaNF-Y, or using precise genome editing to modify the binding sites on TaNAC019 that interact with TaNF-Y, we could specifically modify NF-Y's regulation of TaNAC019 without affecting its inhibition of SSP’ expression—allowing us to fine-tune its regulatory effects on starch and SSP independently. Further analyses may uncover additional variations of TaNF-Y that specifically influence either starch or SSP biosynthesis. With the advances in molecular breeding techniques, including genome editing, these findings open promising avenues for the development of crop varieties that simultaneously enhance grain yield and seed quality.

Conflicting reports exist regarding TaNAC019's role in regulating starch biosynthesis (Liu et al. 2020; Gao et al. 2021). Liu et al. (2020) noted decreased starch content, grain size, and weight upon over expression of TaNAC019-A1, while Gao et al. (2021) observed similar reductions in a TaNAC019 knockout mutant. Our study, which combines morphological observations, transcriptome analyses, and regulatory investigations, indicates that TaNAC019 inhibits starch biosynthesis, particularly during early endosperm development (∼6 DAP). We found that TaNAC019 can directly repress key starch biosynthesis enzyme genes like TacAGPS1a and TaSuS2. In summary, our data support Liu's conclusion. The differing phenotypes of TaNAC019 reported by Gao et al. (2021) and Liu et al. (2020), and our study suggest that TaNAC019 regulates a complex starch synthesis mechanism. This complexity might be linked to its proper spatiotemporal expression, which requires further investigation.

Our study identifies a core TaNF-Y complex that modulates protein and starch levels, finely tuning their synthesis without completely halting it, crucial for practical applications in cereal crops. The regulation of TaNF-Y is associated with specific PRC2-mediated deposition of H3K27me3 modifications on key factors associated with SSP biosynthesis and the critical TF, TaNAC019, which regulates key enzymes involved in starch biosynthesis. Importantly, the natural variation of the TaNF-Y component TaNF-YB7-B associated with starch and protein content in a panel of elite cultivars in China and was selected during breeding process (Supplementary Fig. S12).

Materials and methods

Plant materials and growth conditions

To generate the TaNF-YA3-RNAi and TaNF-YC6-RNAi transgenic lines in the wheat (Triticum aestivum L.) cultivar Fielder, the conserved fragments of TaNF-YA3 (380 bp), TaNF-YC6 (300 bp) were cloned into the NotI/AscI-digested PC414C and inserted into the RNAi vector PC336 as inverted repeats to create the TaNF-YA3-RNAi and TaNF-YC6-RNAi recombinant vectors using Gateway technology (Wang et al. 2022). The recombinant constructs were transformed into Agrobacterium tumefaciens strain EHA105. The transgenic wheat lines were generated through Agrobacterium-mediated infiltration of immature wheat embryos as previously described (Ishida et al. 2015). The resulting T₀ to T₂ transgenic lines were selected via PCR using ubiquitin promoter-specific primers (Supplementary Data Set 10). For the TaNAC019 mutant, we used the Tanac019-cr line as previously described (Gao et al. 2021). For the Tafie mutant, we use the Tafie-C87 line as previously described (Zhao et al. 2023). All wheat plants were grown in the experimental field of Beijing (39°55′ N, 116°23′ E) and in a greenhouse under long-day conditions (16 h light–8 h dark cycles). Nicotiana benthamiana was grown in a greenhouse at 22 °C under a 16 h light and 8 h darkness photoperiod.

Yeast two-hybrid assay

Yeast two-hybrid assays were performed as described in the Frozen-EZ Yeast Transformation II (T2001, Orange). For the interaction of TaNF-YA3-D, TaNF-YB7-B and TaNF-YC6-B, the coding region sequence of TaNF-YA3-D, TaNF-YB7-B were introduced into the EcoRI and XhoI-digested prey vector (pGADT7). To construct the bait vector, we ligated the full-length CDSs of TaNF-YC6-B into the EcoRI and PstI-digested bait vector (pGBKT7). For the interaction of TaNF-YB7-B, TaNF-YC6-B and truncated version with TaSWN-N, we ligated the full-length CDSs of TaNF-YB7-B, TaNF-YC6-B, -Nt, -ΔN, -HF, -Ct, and -ΔC into the EcoRI and XhoI-digested prey vector (pGADT7), vector and TaNWN-B-N into the EcoRI and PstI-digested bait vector (pGBKT7). For the interaction of TaNF-YC6-B, TaSWN-B-N with TaNF-YB7-B-Hap1 and TaNF-YB7-B-Hap2, respectively, we cloned the CDS of TaNF-YC6-B and TaSWN-B-N into the EcoRI and XhoI-digested prey vector (pGADT7), TaNF-YB7-B-Hap1 and TaNF-YB7-B-Hap2 into the EcoRI and PstI-digested bait vector (pGBKT7), and selected on DDO (Synthetic Dropout Medium/-Tryptophan-Leucine, 630417, Clontech) and TDO (Synthetic Dropout Medium/-Tryptophan-Histone-Leucine, 630419, Clontech) media with corresponding concentration of 3-AT (3-Amino-1,2,4-triazole, A8056-10G, Sigma). The empty pGADT7 or empty pGBKT7 used as a negative control. The primers are listed in Supplementary Data Set 10.

Bimolecular fluorescence complementation (BiFC) assay

For the construction of BiFC vectors, pSCY-NE(R)-nCFP is the vector for N-terminal fusion to cyan fluorescent protein (nCFP), and pSCY-Ce(R)-cCFP is the vector for C-terminal fusion to CFP (cCFP). For the interaction of TaNF-YA3-D, TaNF-YB7-B and TaNF-YC6-B, the full-length CDS of TaNF-YA3-D, and TaNF-YB7-B were cloned into BamHI and XhoI-digested pSCY-NE(R)-nCFP. The full-length CDS of TaNF-YC6-B was cloned into BamHI and XhoI-digested pSCY-Ce(R)-cCFP. For the interaction of TaSWN-B-N with TaNF-YB7-B and TaNF-YC6-B, the N-terminal of TaNWN-B (1,050 bp) was cloned into BamHI and XhoI-digested pSCY-NE(R)-nCFP. The CDS of TaNF-YB7-B and TaNF-YC6-B were cloned into BamHI and XhoI-digested pSCY-Ce(R)-cCFP. Young leaves of 4-week-old Nicotiana benthamiana plants were co-infiltrated with Agrobacterium (strain GV3101) harbouring different combinations of these plasmids. Nicotiana benthamiana plants grown in long-day conditions for 48 h after infiltration, the CFP signals were detected by a confocal microscope (LSM980, Carl Zeiss). H2B-mCherry was used as a cell nucleus marker. The fluorescence was observed by the Carl Zeiss LSM980 system with a 405-nm laser (12% intensity), a collection bandwidth of 454 to 581 nm and a detector gain of 650 V for CFP observation; with a 561-nm laser (8% intensity), a collection bandwidth of 580 to 610 nm and a detector gain of 710 V for H2B-mCherry observation. The primers are listed in Data Set S10.

Phylogenetic analysis

We obtained NF-YA, NF-YB, and NF-YC subunit protein sequences of A. thaliana, O. sativa and T. aestivum from Plant Transcription Factor Database (PlantTFDB) (version 5.0, http://planttfdb.gao-lab.org/), We aligned the amino acid sequences with ClustalW and constructed a phylogenetic tree using the neighbor-joining method in MEGA 11 software with default parameters, The evolutionary distances were calculated using the Poisson model. The phylogeny test was computed using bootstrap method with 10,000 replications. Protein sequences, the alignment, and the Newick-format tree can be found in Supplementary Files S1 to S3.

Subcellular localization

The full-length open-reading frames (ORFs) of TaNF-YA3-D, TaNF-YB7-B and TaNF-YC6-B were amplified using specific primers and cloned in the BamHI and SalI-digested pBI221-EGFP vector containing the CaMV 35S Promoter. The resulting construct p35S-TaNF-YA3-D-EGFP, p35S-TaNF-YB7-B-EGFP, p35S-TaNF-YC6-B-EGFP and the p35S- EGFP served as a control were separately introduced into wheat protoplasts. The transformed protoplasts were cultured at 22 °C under darkness for 18 h. H2B-mCherry was used as a cell nucleus marker. The fluorescence was observed by the Carl Zeiss LSM980 system with a 488-nm laser (8% intensity), a collection bandwidth of 510 to 530 nm and a detector gain of 650 V for GFP observation; with a 561-nm laser (8% intensity), a collection bandwidth of 580 to 610 nm and a detector gain of 710 V for H2B-mCherry observation. The primers are listed in Data Set S10.

Phenotypic analysis

The seed-related phenotypes including TGW, grain length and width were determined using a camera-assisted phenotyping system with six biological replicates, each with approximately 10 g seeds. (Wanshen Detection Technology Co., Ltd). To determine total starch content, 100 mg flour was used and measured using Megazyme Total Starch Assay Kit (K-TSTA, Irishtown) according to the manufacturer's instructions (Botticella et al. 2018). The content of amylose was determined according to the Chinese national standard method (GB/T 15683-2008). The contents of HMW-GSs, LMW-GSs, and gliadins were detected by RP-HPLC as previously described (Gao et al. 2021). Total protein content was evaluated using the voluntary standard method of the National Standard of China 5506.4-2008 (GB/T5506.4-2008). Each sample was tested in three biological replications.

Semi-thin sections and scanning electron microscopy

For Fielder, TaNF-YA3-RNAi, TaNF-YC6-RNAi, KN199 and Tafie-C87, seeds at 10 DAP were sampled from spikes. For Tanac019-cr mutant, seeds at 6 DAP were selected. Seeds were fixed in FAA solution (63% v/v ethanol, 5% v/v acetic acid, 2% v/v formaldehyde) under vacuum for 2 h and stored at 4 °C for use. The samples were dehydrated through a graded ethanol series and embedded in Technovit 7100 resin (K010107, Kulzer) according to the manufacturer's instructions. Sections were cut with 2-µm thick using a microtome (UC7&2265, Leica). To visualize starch granules and protein bodies, the sections were stained with 0.1% w/v coomassie brilliant blue R-250 (0472, LABLEAD) (Tonosaki et al. 2021). Images of stained sections were captured using a Microscope (DP74, OLYMPUS). Endosperm starch granules number was manually calculated for three seeds and the differences between Fielder TaNF-YA3-RNAi-1, TaNF-YC6-RNAi-1, Tanac019-cr, KN199, and Tafie-C87 were by one-way ANOVA. The data are shown as mean ± SD (n = 10).

For scanning electron microscopy observation, the grains were dried at 37 °C for 1 week after harvest. The seeds of TaNF-YA3-RNAi, TaNF-YC6-RNAi, Fielder, Tafie-C87, and KN199 were broken transversely, and the ruptured surfaces were coated with gold. A scanning electron microscope (Crossbeam 340 & VCT500, Carl Zeiss) was employed to observe the samples. The data are shown as mean ± SD (n = 10).

RNA extraction, RT-qPCR analysis, and RNA-Seq

Total RNAs were extracted from endosperm at 4 DAP, 8 DAP, 10 DAP, 12 DAP, 16 DAP, 20 DAP, 24 DAP of Fielder, TaNF-YA3-RNAi-1, TaNF-YC6-RNAi-1, 10 DAP of KN199 and Tafie-C87, 6 DAP of Fielder and Tanac019-cr using the Quick RNA Isolation Kit with on-column DNaseI digestion (0416-50, Huayueyang) according to the manufacturer's protocol. First-strand cDNAs were generated using a reverse transcription kit (KR116-02, TRANSGEN). Subsequent RT-qPCR assays were performed using the SYBR Green PCR Master Mix (Q121-02/03, Vazyme Biotech). with each RT-qPCR assay being replicated at least three biological replicates and three technical replicates. Wheat Actin gene (TaActin, TraesCS5A02G124300) was used as an internal reference. Relevant primer sequences are given in Supplementary Data Set 10. For RNA sequencing, oligo (dT) was used for enriching the mRNA from total RNA and then fragmentation and random primer was used for reverse transcript process. Sequencing is performed via the Illumina NovaSeq platform by Annoroad Gene Technology (Li et al. 2022).Clean reads were aligned to IWGSC RefSeq v1.1 using hisat2 (v2.0.5) (International Wheat Genome Sequencing 2018; Kim et al. 2019). Transcripts per kilobase of exon model per million mapped reads (TPM) was used for estimating gene expression levels. Absolute value of Log₂(Fold Change) ≥ 1 and FDR ≤ 0.05 were considered to be differentially expressed genes.

Transcriptional activity analysis

To test the transcription activity, the coding sequence of TaNF-YA3-D and TaNF-YC6-B were cloned into the BamHI and SalI-digested 35S-GAL4-BD to generate effectors. Virion protein 16 (VP16) contain an acidic transcriptional activation domain was used as a control to identify the effect of TaNF-YA3-D and TaNF-YC6-B on transcriptional activity driven by the UAS module in the reporter vector 35S-LUC. The reference vector pRTL containing a 35S promoter-driven REN was used as an internal control. Each effector construct together with the reporter and reference vectors were co-transformed into wheat protoplasts (Bart et al. 2006; Yoo et al. 2007). The GLOMA 20/20 LUMINOMETER detector (GLOMA 20/20, Promega) was used to measure fluorescence signals. Relative LUC activity was calculated by the ratio of LUC/REN. Each sample was tested in three biological replications. The primers are listed in Supplementary Data Set 10.

Dual-luciferase reporter assay

To generate pTaNAC019-B1-LUC, pTaGli-γ-700-LUC, pTaLMW-400-LUC, pTacAGPS1a-LUC, and pTaSuS2-LUC constructs, we amplified 2-Kb promoter fragments upstream of each gene from CS and ligated them with the EcoRI-digested CP461-LUC as the reporter vector. The p35S-TaNF-YA3-D-GFP, p35S-TaNF-YB7-B-GFP, p35S-TaNF-YC6-B-GFP, and p35S-TaNAC019-B1-GFP constructs were used as effectors and these plasmids were transformed into GV3101. Then these strains were injected into Nicotiana benthamiana leaves in different combinations. Dual luciferase assay reagents (VPE1910, Promega) with the Renilla luciferase gene as an internal control were used for luciferase imaging. The Dual-Luciferase Reporter Assay System kit (E2940, Promega) was used to quantify fluorescence signals. Relative LUC activity was calculated by the ratio of LUC/REN. Each sample was tested in three biological replications. The relevant primers are listed in Supplementary Data Set 10.

Affinity purification and mass spectrometry

To identify TaNF-YA3 interaction proteins, we performed immunoaffinity purification as previously described by Tan et al. (2022) with minor modifications. Briefly, 1 g whole-grain flour from 10 DAP grains of Fielder were harvested and ground in liquid nitrogen and quickly placed in 3 mL pre-cooled IP buffer (50 mm Tris-HCl, pH 7.5, 150 mm NaCl, 5 mm pH 8.0 EDTA, 0.2% v/v NP-40, 0.1% v/v TritonX-100, 1 mm PMSF, 1× protease inhibitor cocktail from Roche, Basel, Switzerland). Samples were further homogenized to be free of clumps and lysed for 30 min at 4 °C and then centrifuged at 5,000 g and 4 °C for 30 min. 50 μL TaNF-YA3 antibody (made in Abconal) was pre-bound to 100 μL protein G beads for 2 h at 4 °C. The supernatant was incubated with 150 μL of anti-TaNF-YA3 Protein G beads for 2 h at 4 °C. Protein-bound beads were washed five times with 1.5 mL IP buffer (5 min spin at 4 °C) and finally with IP buffer without NP-40. The beads were sent to Institute of Biophysics, Chinese Academy of Sciences (Beijing, China) for mass spectrometry analysis. Two independent replicates were performed. The Anti-IgG (AC005, ABclonal) was used as a negative control.

Protein expression and purification

We cloned the TaNF-YA3-D coding sequences into the SalI and NotI-digested pET-32a vector, TaNF-YB7-B coding sequences into the BamHI and XhoI-digested pGEX-4T-1, TaNF-YC6-B and TaNAC019-B1 coding sequence into the EcoRI and PstI-digested pMAL-c5X-MBP. The recombinant proteins were expressed in Escherichia coli BL21 (CW08095, CWBIO). Induction was performed by adding IPTG to a final concentration of 0.5 mm and cells were cultured at 28 °C for an additional 16 h. Recombinant proteins were purified using the His-tag Protein Purification Kit (30210, QIAGEN), GST-tag Protein Purification Kit (6555-10, BioVision) and MBP-tag Protein Purification Kit (SA0077100, Smart-Lifesciences), respectively, and were later used for pull-down and EMSA. The primers are listed in Supplementary Data Set 10.

Pull-down assay and Immunoblot analysis

The recombinant GST-TaNF-YB7-B protein was purified and immobilized on Glutathione Sepharose beads (6555-10, BioVision) following the manufacturer's instructions. The beads were divided into two equal aliquots and incubated with the same amount of TaNF-YA3-D-His protein lysate, together with MBP-TaNF-YC6-B or with MBP for 2 h at 4 °C. The beads were subsequently washed five times with PBST buffer, followed by elution with 100 μL of elution buffer (50 mm Tris-HCl, 10 mm pH 8.0 reduced glutathione). Supernatants were resolved by 12% SDS-PAGE and subjected to immunoblotting using anti-GST (G1001, LABLEAD, 1:2,500), anti-His (BE2019-100, EASYBIO, 1:2,500), anti-MBP (66003-1-Ig, Proteintech, 1:2,500).

Electrophoretic mobility shift assay (EMSA)

EMSA analyses were performed essentially as previously described (Siriwardana et al. 2016). The double-stranded probes (labeled with biotin at their 5′-end) were annealed from complementary oligonucleotides by cooling from 100 °C to room temperature in annealing buffer (10 mm pH 7.5 to 8.0 Tris, 50 mm NaCl, 1 mm pH 8.0 EDTA). The sequences of the probes are listed in Supplementary Data Set 10. DNA binding reactions (20 μL reaction volume, 20 nm probe, 12 mm pH 8.0 Tris-HCl, 50 mm KCl, 62.5 mm NaCl, 0.5 mm pH 8.0 EDTA, 5 mm MgCl₂, 2.5 mm DTT, 0.2 mg/mL BSA, 5% v/v glycerol, 6.25 ng/μL poly dA-dT) were incubated TaNF-YA3-D-His/GST-TaNF-YB7-B/MBP-TaNF-YC6-B trimers, three recombinant proteins were pre-mixed in room temperature for 1 h, then added to DNA binding mixes. After 30 min incubation at 30 °C, binding reactions were loaded on 6% polyacrylamide gels and separated by electrophoresis in 0.5 x TBE. Transfer to a nylon membrane and detection of biotin-labeled DNA were performed using LightShift Chemiluminescent EMSA Kit according to the manufacturer's instructions (20148, Thermo Scientific).

Cleavage under targets & tagmentation (CUT&Tag)

CUT&Tag were performed using endosperm at 10 DAP from TaNF-YC6-RNAi-1 and Fielder plants. Two biological replicates were performed independently. The CUT&Tag experiment was performed exactly following protocols as previously described (Zhao et al. 2023). Finally, the purified PCR products were sequenced using an Illumina NovaSeq platform. For data analysis, the pipeline was largely based on the previous study (Zhang et al. 2023a, 2023b, 2023c). The low-quality reads of CUT&Tag library was filtered using fastp v0.20.0 (Chen et al. 2018). The cleaned reads were mapped to IWGSC RefSeq v1.1 (International Wheat Genome Sequencing 2018) using bwa mem algorithm 0.7.17 (Li and Durbin 2009). MACS2 v2.1.2 was used for peak calling. For H3K27me3 peak calling, “–keep-dup all -g 14600000000 –broad –broad-cutoff 0.05” were used. The peaks were further annotated to the wheat genome using ChIPseeker (Yu et al. 2015). The whole genome was divided into three regions: −3,000 bp of transcription start site (TSS) considered as promoter region, TSS to transcription end site (TES) considered as genic region, and other regions as distal region. The MAnorm package72 was used for the quantitative comparison of CUT&Tag signals between samples with the following criteria: |M value| > 1 and P < 0.05.CUT&Tag-qPCR was performed as previously described (Tian et al. 2023). Briefly, the DNA products of CUT&Tag were divided, 6 µL was used as “Input”, and the other 18 µL undergoing library PCR amplification and purification was used as “IP products.” Finally, the “Input” and “IP products” were dilute 30 times for qPCR assay. Each sample was tested in three biological replications. The primers for specific regions are provided in in Supplementary Data Set 10.

Firefly luciferase complementation imaging (LCI) assay

The coding sequence of TaNF-YB7-B-Hap1, TaNF-YB7-B-Hap2, TaNF-YC6-B and TaSWN-B-N were cloned into the KpnI and SalI-digested pCAMBIA1300-35S-Cluc-RBS or BamHI and SalI-digested pCAMBIA1300-35S-HA-Nluc-RBS vectors to generate fusion constructs (Liu et al. 2018a). Four different vectors (e.g. nLUC-TaNF-YB7-B-Hap1, nLUC-TaNF-YB7-B-Hap2, cLUC-TaNF-YC6-B and cLUC-TaSWN-B-N) enabling testing of protein-protein interaction, were cotransfected into Nicotiana benthamiana leaves epidermal cells by Agrobacterium-mediated infiltration. After 3 days of incubation, the injected leaves were sprayed with 1 mm luciferin (E2940, Promega) and the LUC signal was captured using a cooled CCD imaging apparatus (LB985, Berthold). Each assay was repeated at least three times. The Dual-Luciferase Reporter Assay System kit (E2940, Promega) was used to quantify Luminescence intensity. Each sample was tested in six biological replications. Relevant primer sequences are given in Supplementary Data Set 10.

Haplotype analysis of TaNF-Y

The single nucleotide polymorphisms (SNPs) and phenotypic data used for haplotype analysis were obtained from the publicly available database (https://opendata.earlham.ac.uk/wheat/under_license/toronto/WatSeq_2023-09-15_landrace_modern_Variation_Data/) (Cheng et al. 2023). The dataset contained genotype and partial phenotypic data for 827 Watkins landraces and 223 modern cultivars. The grain phenotype data (thousand-grain weight, grain protein content, and starch content) were collected at the John Innes Centre Field Experimental Station in 2014 (Supplementary Data Set 11). SNPs (MAF > 0.05) located in the gene upstream 2,000 bp, coding regions, and downstream 500 bp were extracted and subjected to haplotype analysis using the R package “geneHapR” v1.1.9 (Zhang et al. 2023c).

Statistics and data visualization

For quantitative results, including three biological replicates and at least three technical replicates, the data is presented in the form of mean ± standard deviation. The means of two samples were compared using Student's two-tailed t tests. Analysis of variance (one-way ANOVA) was conducted using default parameters in Graphpad Prism 8.0.2 software. Significant differences were determined by Student's t test or one-way ANOVA: *, P < 0.05 and **, P < 0.01. Detailed statistical analysis data are shown in Supplementary Data Set 12.

Accession Numbers

Sequence data from this article can be found in the EMBL library (http://plants.ensembl.org/index.html) under the following accession numbers: TaNF-YA3-D, TraesCS4D02G289600; TaNF-YB7-B, TraesCS7B02G121900; TaNF-YC6-B, TraesCS6B02G185700; TacAGPL1, TraesCS1A02G419600; TaGBSSII, TraesCS2A02G373600; TacAGPS1a, TraesCS7A02G287400; TaSSIIIa, TraesCS1A02G091500; TaSuS2, TraesCS2A02G168200; TaGli-α, TraesCS6B02G065856; TaGli-γ-700, TraesCS1D02G000700; TaLMW-400, TraesCS1D02G007400; TaNAC019-A1, TraesCS3A02G077900; TaNAC019-B1, TraesCS3B02G092800, TaNAC019-D1, TraesCS3D02G078500; and TaSWN-B, TraesCS4B02G181400. RNA-seq and CUT&Tag-seq data are available from the National Genomics Data Center (https://ngdc.cncb.ac.cn/) under accession number PRJCA022157.

Data availability

The data underlying this article are available in the article and in its online supplementary material.

Author contributions

J.X. designed and supervised the research, J.X. and J.-C.C. wrote the manuscript. J.-C.C. did most of the experiments; J.-C.C. and H.-R.L. did the GST pull down and EMSA; L.Z. Y.-J.L. and H.Z. performed bio-informatics analysis; D.-Z. W. and P.Z. did the haploid and selection analysis; X.-M. B. and X.-S.Z. provided Tafie-C87 line; C.-F.Y. measured the protein content; M.-X.Z. helped with endosperm sampling and section; Y.-Y.Y. provided Tanac019-cr transgenic wheat lines; D.-Z.W., D.S., X.-L.L., Y.-Y.Y., S.-B.X., X.-S.Z, and X-Y.Z. revised the manuscript; J.-C.C., L.Z., D.-Z.W., and J.X. prepared all the figures. All authors discussed the results and commented on the manuscript.

Supplementary data

The following materials are available in the online version of this article.

Supplementary Figure S1. Phylogenetic analysis and expression patterns of TaNF-Y components.

Supplementary Figure S2. Generation and validation of multiple TaNF-YA3 and TaNF-YC6 RNAi lines.

Supplementary Figure S3. Quantification of starch granules in TaNF-YA3-RNAi-1, TaNF-YC6-RNAi-1, and Fielder.

Supplementary Figure S4. Overview of RNA-seq data for TaNF-Y knockdown lines.

Supplementary Figure S5. DNA-binding capability and transcriptional activity of TaNF-Y components.

Supplementary Figure S6. The expression and chromatin accessibility pattern of TaNAC019.

Supplementary Figure S7. The expression pattern of PRC2 components and interaction between TaSWN and TaNF-Y.

Supplementary Figure S8. The change in H3K27me3 levels in TaNF-YC6-RNAi-1.

Supplementary Figure S9. Quantification of starch granules in Tafie-C87 and KN199.

Supplementary Figure S10. The expression pattern of starch degradation genes in TaNF-Y-RNAi lines and Tafie-C87 mutant.

Supplementary Figure S11. The frequency of different haplotype combinations of TaNF-YB7-B and TaNAC019-B1 in landraces and modern cultivars.

Supplementary Figure S12. Proposed model for direct and indirect regulation of starch biosynthesis and SSP composition by the TaNF-Y–PRC2 module.

Supplementary Data Set 1. The expression level of SSP and starch synthesis genes during wheat endosperm development.

Supplementary Data Set 2. List of TaNF-YA3 interaction proteins identified by IP-MS.

Supplementary Data Set 3. List of differentially expressed genes between TaNF-YA3-RNAi-1, TaNF-YC6-RNAi-1 and Fielder.

Supplementary Data Set 4. List of genes with CCAAT motif in the open promoter region that are differentially regulated in TaNF-Y-RNAi lines compared to Fielder.

Supplementary Data Set 5. List of differentially expressed genes between Tanac019-cr and Fielder.

Supplementary Data Set 6. List of genes with altered H3K27me3 levels in TaNF-YC6-RNAi-1 compared to KN199.

Supplementary Data Set 7. List of differentially expressed genes between Tafie-C87 and KN199.

Supplementary Data Set 8. Association of TaNF-YA/B/C gene variation with grain development traits.

Supplementary Data Set 9. The chi-square independence test for TaNAC019-B1-Hap2 and TaNF-YB7-B-Hap2.

Supplementary Data Set 10. Primers used in this study.

Supplementary Data Set 11. Detailed information of the wheat accessions used in this study.

Supplementary Data Set 12. Detailed statistical analysis in this study.

Supplementary File 1. Protein sequences for the phylogenetic tree shown in Supplementary Fig. S1.

Supplementary File 2. Protein sequence alignment for the phylogenetic tree shown in Supplementary Fig. S1.

Supplementary File 3. Newick format of the phylogenetic tree of Supplementary Fig. S1.

Funding

This research is supported by the National Key Research and Development Program of China (2021YFD1201500), Beijing Natural Sciences Foundation Outstanding Youth Project (JQ23026), the National Natural Sciences Foundation of China (31921005, 32000382), and the Major Basic Research Program of Shandong Natural Science Foundation (ZR2019ZD15).

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• SSP Gramene: AT2G17090

• SSP Araport: AT2G17090

• HDA15 Gramene: AT3G18520

• HDA15 Araport: AT3G18520

• ISA1 Gramene: AT2G39930

• ISA1 Araport: AT2G39930

• PHS1 Gramene: At3g29320

• PHS1 Araport: At3g29320

• HAP2 Gramene: AT4G11720

• HAP2 Araport: AT4G11720

• GPT Gramene: AT2G41490

• GPT Araport: AT2G41490

• grain size Planteome: TO:0000397