Dynamic chromatin regulatory programs of sucrose and citric acid metabolism during fruit ripening in Citrus

Abstract

Fruit ripening is a complex, tightly regulated process that affects fruit flavor, nutritional value, and shelf life. Citrus is a typical non-climacteric fruit, as citrus fruits ripen independently of ethylene. However, the precise regulatory mechanism underlying this process remains to be elucidated. To dissect the epigenetic and transcriptional basis of citrus ripening, we performed integrative analyses of genome-wide histone modifications (H3K27ac, H3K4me3, and H3K27me3), chromatin accessibility, and transcriptome profiles throughout fruit ripening in Citrus reticulata. We constructed a transcriptional regulatory network encompassing 52 key transcription factors that orchestrate citrus fruit ripening. Further, we identified the key role of abscisic acid (ABA) in modulating sucrose and citric acid metabolism and uncovered the underlying transcriptional and epigenetic regulatory network during fruit ripening in C. reticulata. The findings demonstrate that CrHSFA6B and CrbZIP5b interact with the promoters of CrACO1 (Aconitase 1) and CrACO3 and CrSPSF1 (Sucrose Phosphate Synthase F1), regulating their expression to affect the sucrose and citric acid metabolism and fruit ripening. This study provides insights into the epigenomic dynamics of citrus fruit ripening and identifies a putative regulatory cascade centered on ABA, CrbZIP5b, and CrHSFA6B that modulates key processes in citrus ripening.

A transcriptional regulatory network of the citrus ripening process is established based on the histone modifications, chromatin accessibility, and transcriptome profiles. Two transcription factors activate sucrose biosynthesis and citric acid degradation genes for turning the fruit from sour to sweet.

Graphical Abstract

Graphical Abstract.

Introduction

Fruit ripening is a highly coordinated and intricate physiological process that involves a series of metabolic and structural changes, which ultimately shape the fruit's quality attributes, including pigmentation, flavor, aroma, texture, and nutritional content (Wang et al. 2023b). This process varies significantly between climacteric and non-climacteric fruits. Climacteric fruits, such as bananas (Musa paradisiaca), apples (Malus domestica), mangoes (Mangifera indica), and tomatoes (Solanum lycopersicum), display a burst in both respiration and ethylene production at the onset of ripening. Meanwhile, non-climacteric fruits, including citrus (Citrus reticulata), grapes (Vitis vinifera), and strawberries (Fragaria ananassa), lack such a substantial burst (Cherian et al. 2014; Cao et al. 2024b).

As the largest fruit crop worldwide, citrus is well known for its abundant nutrient materials and excellent flavor quality (Lü et al. 2018). Citrus fruits are typical hesperidia, with leathery peel surrounding the edible fruit portion, and have non-climacteric fruit ripening characteristics (Strazzer et al. 2019; Zhu et al. 2023 ). The development of citrus fruits can be divided into three stages: cell division stage, expansion stage, and ripening stage (Paul et al. 2012). The dynamic balance between soluble sugar accumulation and organic acid degradation plays a central role in defining the fruit maturity and taste profile, directly affecting consumer acceptance and marketability (Feng et al. 2021; Mao et al. 2024). Hence, knowledge of the molecular regulation of sucrose and citric acid metabolic processes during fruit ripening is essential for genetic improvement and promotion of fruit quality in citrus.

Key phytohormones such as ethylene, abscisic acid (ABA), and auxin play central roles in regulating the metabolic pathways during citrus fruit ripening. Auxin is known to play a pivotal role in the regulation of fruit set and development in strawberry and tomato through the precise spatial and temporal control of auxin distribution and signaling (Kang et al. 2013; Hu et al. 2023). Ethylene is essential for initiating and regulating the ripening process in climacteric fruits such as tomato, banana, and apple. The mechanisms of fruit ripening mediated by ethylene in climacteric fruits have been well studied (Li et al. 2020; Fenn and Giovannoni 2021). In contrast, ABA is considered to be the primary phytohormone that regulates fruit ripening in non-climacteric fruits. The ABA content is increased dramatically during fruit ripening processes in citrus and strawberry (Feng et al. 2021; Zhang et al. 2023). However, the mechanisms determining fruit ripening and quality by ABA in non-climacteric fruits remain to be elucidated.

Epigenetics, including DNA methylation, histone modifications, and chromatin remodeling, has been shown to modify chromatin structure and thus affect gene transcription in response to developmental and environmental signals (Tang et al. 2020; He et al. 2022; Liu et al. 2022). For example, in wheat open chromatin, H3K4me3, H3K9ac, and H3K27ac are all associated with active gene transcription, whereas H3K27me3 is enriched in repressed genes (Liu et al. 2023, 2024; He et al. 2024). Recently, studies have revealed that epigenetics play a key role in the fruit ripening network of climacteric fruits, such as tomato and banana (Lü et al. 2018; Ding et al. 2022; Li et al. 2024b). However, the dynamic changes of chromatin accessibility and histone modifications, and how it affects gene transcription during citrus fruit ripening processes, remain largely unknown.

Here, we perform expression analysis and describe the dynamics of chromatin accessibility and histone modifications (H3K27ac, H3K4me3, and H3K27me3) in citrus fruit at three developmental stages (120, 150, and 180 d, days after flowering). By integrating the information of transcriptomics, epigenomics (ATAC, CUT&Tag), TF footprint, and metabolomics, we construct gene regulatory networks and uncover hub genes to control citrus fruit ripening. Moreover, our findings unveil an ABA-CrbZIP5b-CrHSFA6B-centered regulatory cascade that governs sucrose accumulation and citric acid degradation during the citrus fruit ripening process, providing potential targets for optimizing fruit flavor and nutritional quality.

Results

Six citrus cultivars exhibit similar ripening dynamics in terms of sucrose accumulation, citric acid degradation, and hormone fluctuations

Six widely cultivated citrus, Fortunella margarita (kumquat “YPJG”), C. reticulata (ponkan “EG” and mandarin “DF4”), C. sinensis (sweet orange “AL” and navel orange “F72”), and C. paradisi (grapefruit “Cocktail”), were selected to assess ripening indicators, including sucrose and citric acid contents (Fig. 1 and Table S1). Citrus fruit development in EG (C. reticulata Blanco cv. Ponkan) can be divided into three distinct stages: cell division (0 to 60 days after flowering, d), cell expansion (60 to 120 d), and initiation of fruit ripening (120 to 180 d) (Mao et al. 2024). During the early cell division phase, fruit size increases gradually, and citric acid begins to accumulate (Fig. 1b and Figure S1a). In the subsequent cell expansion stage, fruit size and weight grow rapidly, and citric acid content peaks around 90 d (Fig. 1b and Figure S1a). The final ripening stage is marked by a decline in citric acid level and the accumulation of sucrose, indicating a metabolic shift toward ripening. All 6 selected citrus cultivars exhibited similar maturity characteristics, suggesting that the initiation of the ripening process of citrus fruits is regulated by a conserved internal mechanism.

Figure 1.

Citric acid and sucrose contents during citrus fruit development stages. (a) Phylogenetic tree of representative citrus species. A phylogenetic tree is constructed using the maximum likelihood method. The citrus cultivars used for Figures 1B and C were highlighted in red font with fruit photos. Single-nucleotide polymorphisms (SNPs) in the regions of conserved single-copy genes from 12 representative accessions within the Rutaceae family are used. The images of citrus fruit were digitally extracted for illustrative purposes and are not shown to scale. (b) The citric acid and sucrose contents in six citrus cultivars during the whole development stage. EG ponkan (Citrus reticulata); DF4 mandarin (Citrus unshiu), data from Chen et al. (2023); F72 navel orange (Citrus sinensis), data from Feng et al. (2021); AL sweet orange (Citrus sinensis), data from Mao et al. (2024); YPJG Kumquat (Fortunella margarita), data from Wei et al. (2021); Cocktail grapefruit (Citrus paradisi). d: Days After Flowering.

Given the pivotal role of plant hormones in regulating fruit ripening, further analysis was conducted to detect the changes in the content of various hormones in citrus fruits. As shown in Figure S1b, ethylene (ACC) levels remain generally low during citrus fruit development, with only a transient increase around 90 to 120 d before declining again, further reinforcing the non-climacteric nature of citrus fruit ripening. Auxins and cytokinins exhibit high activity during the early stages of development, facilitating fruit growth-related processes. In contrast, ABA levels significantly increase after 150 d, coinciding with the initiation of ripening. The pronounced elevation in ABA concentration, together with the observed metabolic changes, clearly demarcates the transition into the initiation of the fruit ripening stage and underscores the critical role of ABA in regulating citrus fruit ripening.

A transcriptional and epigenomic atlas of citrus fruit ripening

To investigate the epigenomic and transcriptional dynamics during citrus fruit ripening, we performed RNA-seq on developing juice sacs of Citrus reticulata (ponkan “EG”) across five developmental stages: 60, 90, 120, 150, and 180 d (Fig. 2a and Table S2). Phenotyping analysis revealed a crucial ripening phase occurring between 120 and 180 d, which was characterized by rapid fluctuations in sucrose and citric acid concentrations. To explore chromatin regulation during this stage, we conducted ATAC-seq and CUT&Tag profiling targeting H3K27ac, H3K4me3, and H3K27me3 at three key time points: 120, 150, and 180 d (Fig. 2a and Table S2). All datasets showed high quality and reproducibility, with strong Pearson correlations (r ≥ 0.96) and enrichment scores (FRiP ≥ 0.3) (Figure S2a, b, and Table S3). Principal component analysis (PCA) of RNA-seq data revealed clear transcriptional reprogramming during fruit ripening, with samples showing a continuous developmental trajectory from 60 to 180 d (Fig. 2b, Figure S3, and Table S4). Similarly, ATAC-seq and CUT&Tag data showed distinct separation among the three time points (120, 150, and 180 d) (Fig. 2b and Figure S2c), indicating dynamic changes in chromatin accessibility and histone modification throughout the ripening process. Most chromatin accessibility and histone modification peaks were enriched in promoter and coding sequence regions (Fig. 2c). The peaks of open chromatin and H3K27ac were mainly located near the transcription start site (TSS), whereas H3K27me3 and H3K4me3 were spread throughout the gene (Fig. 2d). In summary, we found ATAC-seq peaks enriched at promoters, reflecting sites accessible for transcription initiation, while histone modifications across gene bodies regulate transcriptional efficiency and stability. These layers may act together, altering chromatin accessibility, to ensure precise control of gene expression during citrus fruit ripening. During citrus fruit ripening, a progressive increase in the number and proportion of differentially accessible chromatin regions and activating histone marks (H3K27ac and H3K4me3) was observed, particularly from 120 to 180 d, indicating enhanced transcriptional activity (Fig. 2e). Highly expressed genes are generally associated with open chromatin and enriched in H3K27ac and H3K4me3 but depleted in H3K27me3 (Figure S4a). In contrast, low or non-expressed genes show reduced chromatin accessibility and are marked by H3K27me3 (Figure S4a). Differential analysis revealed numerous changes in chromatin accessibility and histone modification peaks across developmental stages, with the greatest differences observed between 120 and 180 d (Fig. 2e, Figure S4b and Supplementary Tables S5–S8), which coincided with marked differences in sucrose and citric acid levels, suggesting that dynamic chromatin accessibility and histone modifications are involved in regulating sucrose and citric acid metabolism during citrus fruit ripening.

Figure 2.

A transcriptional and epigenomic atlas of fruit ripening in C. reticulata.  (a) Overview of the developmental stages of C. reticulata fruit sampled for transcriptomic (RNA-seq) and epigenomic (ATAC-seq and CUT&Tag for H3K27ac, H3K4me3, and H3K27me3) profiling. Representative cross-sections of fruits collected at five time points (60, 80, 120, 150, and 180 d) are shown. Scale bars = 2 cm. The images are digitally extracted for comparison. (b) Principal component analysis (PCA) of RNA-Seq (left) and ATAC-seq (right). Each point represents an individual biological replicate. The black lines indicated the fruit developmental stages of C. reticulata. (c) Genomic distribution of peaks identified from ATAC-seq and CUT&Tag datasets for histone modifications. Peaks were annotated to promoter regions (± 2 kb from transcription start sites), gene bodies, or distal intergenic regions. (d) Epigenetic profiles on gene sets of C. reticulata fruits at different developmental stages. TSS, transcription start site; TES, transcription end site. (e) Number and proportion of differentially accessible chromatin regions (ATAC-seq) and differentially enriched histone modification peaks (H3K27ac, H3K4me3, and H3K27me3) across fruit developmental stages, identified using a fold change ≥ 1.5 and P < 0.05. (f) K-means clustering of RNA-seq data reveals six major gene expression modules (C1–C6) with distinct temporal expression patterns during fruit development in C. reticulata. (g) Transcription factor (TF) family enrichment analysis for each expression cluster. Enrichment was calculated using Fisher's exact test with Benjamini–Hochberg correction (Padj < 0.05). Font size reflects enrichment significance (−log₁₀(P_adj)). (h) Gene Ontology (GO) enrichment analysis of biological processes for clusters C1 and C2. The x-axis represents the number of enriched genes associated with each GO term.

Based on the RNA-seq data, a total of 18,025 high-confidence genes were identified across the five developmental stages, with expression detected during at least one stage (TPM > 0.5). These genes were subsequently categorized into 6 clusters through K-means clustering analysis (Fig. 2f and Table S9). Notably, genes in C5 and C6, such as members of the Dof, GRAS, MYB, ARF, and HD-ZIP families, are upregulated during early development stages (Fig. 2g and Table S10), indicating their role in initiating fruit development processes. Meanwhile, genes in C1 and C2, enriched for NAC, WRKY, ERF, and HSF transcription families, showed strong activation in the late ripening stages, suggesting their involvement in fruit ripening (Fig. 2g and Table S10). Gene ontology (GO) enrichment analysis of C1 and C2 clusters further emphasized that they were particularly enriched in categories related to response to hormones, carbohydrate metabolic processes, and tricarboxylic acid cycle (Fig. 2h), which are critical for the coordination of citrus fruit ripening.

Previous studies have demonstrated that CsPH5 and CsPH4 genes are involved in citric acid metabolism during citrus fruit development, primarily regulating citric acid synthesis and accumulation (Strazzer et al. 2019; Huang et al. 2023). Their expression levels have been shown to correlate strongly with the concentration of citric acid in the citrus fruit. However, our study found that the expression of these genes, along with chromatin accessibility and histone modifications, exhibited weak correlations with changes in citric acid content during the initiation of the fruit ripening stage (Figure S5). This suggests that these two genes play a role in the accumulation of organic acids in early citrus fruit development; however, the mechanism underlying the decline in citric acid content during the maturation phase warrants further investigation.

Coordinate regulation of epigenomes and transcriptomes in response to fruit ripening.

We further investigated the regulatory networks associated with the citrus fruit ripening transition from 120 to 180 d, with a particular focus on how chromatin accessibility and histone modifications shape gene expression. As expected, changes in gene expression between 120 and 180 d were positively correlated with chromatin accessibility, H3K27ac and H3K4me3 levels, and negatively correlated with changes in H3K27me3 (Fig. 3a). Notably, several genes related to citrus fruit ripening, such as CrACO3 (involved in citric acid catabolism), CrSPSF1 (involved in sucrose biosynthesis) and CrPSY (involved in carotenoids biosynthesis) were regulated by chromatin accessibility and histone modifications, indicating a potential role of epigenetic modifications during fruit ripening (Fig. 3b). In total, 4,108 upregulated and 4,651 downregulated expressed genes were identified between 120 and 180 d fruit samples (Fig. 3c and Table S11). Among the upregulated genes, 3,062 were categorized into Cluster 1 and Cluster 2 (Fig. 3d and Table S12), characterized by a substantial increase in expression during 120 to 180 d. We assessed the Pearson correlation coefficient (PCC) between transcriptional patterns and various chromatin features from 120 to 180 d (Fig. 3e). Genes were considered as being regulated by chromatin accessibility or histone modifications if their PCC values exceeded 0.7 (or were less than −0.7 for H3K27me3). The findings revealed that 96.24% of the differentially expressed genes (DEGs) exhibited at least one ATAC-seq peak within their promoter or genic regions, with 71.91% being regulated by chromatin accessibility (Fig. 3e and Table S13). Regarding histone modifications, 57.64%, 47.22%, and 70.71% of DEGs associated with H3K27ac, H3K4me3, and H3K27me3, respectively, exhibited at least one peak within the promoter or genic regions. Of these, 44.84%, 25.05%, and 57.81% were regulated by the corresponding histone modifications (Fig. 3e and Table S13). Collectively, 2,018 DEGs were identified as being regulated either by open chromatin or by at least one of the histone modifications H3K27ac, H3K4me3, or H3K27me3 (Fig. 3e, 3f, and Table S14). Notably, approximately 76% of these DEGs were classified into Cluster 2 (Table S14), and these DEGs showed a consistent trend of increased expression from 120 to 180 d, accompanied by enhanced chromatin openness, increased levels of H3K27ac and H3K4me3 modifications, and decreased levels of H3K27me3 modifications (Fig. 3f).

Figure 3.

Chromatin dynamics coordinate transcription profile during fruit ripening in C. reticulata.  (a) Pearson correlation between differential gene expression and chromatin features—chromatin accessibility (ATAC), H3K27ac, H3K4me3, and H3K27me3—between 180 and 120 d of fruit in C. reticulata. The differentially expressed genes were categorized into 100 groups according to the fold change in gene expression. P values were determined by two-sided Pearson correlation coefficient analysis. (b) Dynamic transcription (top graphs) and epigenetic modification tracks for CrACO3, CrSPSF1, and CrPSY. ACO3, aconitase3; SPSF1, sucrose phosphate synthetase F1; PSY, phytoene synthase. Transcription data shown as mean ± SD of 3 biological replicates. TPM, transcripts per million. (c) Upset plots showing overlaps of differentially expressed genes (DEGs). The plots display the numbers of upregulated (left) and downregulated (right) DEGs across three pairwise comparisons: 120 vs 150 d, 150 vs 180 d, and 120 vs 180 d. Vertical bars represent the number of DEGs in each intersection, while connected dots indicate the specific comparisons contributing to each overlap. (d) Venn diagram showing the overlap between upregulated DEGs (from C) and genes assigned to expression clusters C1 and C2 based on the EG (C. reticulata) RNA-seq dataset. (e) Identification of DEGs regulated by chromatin accessibility or histone modifications. The Upset plot illustrates the distribution and overlap of DEGs across chromatin accessibility (ATAC) and histone modification states. The top bar represents the number of DEGs associated with individual chromatin features (ATAC or specific histone modifications), while the connected matrix and corresponding vertical bars indicate the number of DEGs within each intersection of chromatin feature combinations. “With ATAC” indicates DEGs containing at least one ATAC in the promoter or gene body regions. DEGs were defined as epigenetically regulated if their expression levels significantly correlated with chromatin features: ATAC.reg, H3K27ac.reg, and H3K4me3.reg (PCC > 0.7), or H3K27me3.reg (PCC < −0.7). Among the ATAC.reg genes, two regulatory groups were further defined as C1# and C2#. (f) The heat map shows transcription, chromatin accessibility, and histone modification dynamics of DEGs in C1# and C2#.

As shown in Fig. 1, the citric acid degradation exhibited similar patterns between navel orange (F72, C. sinensis) and ponkan (EG, C. reticulata), characterized by a sharp decline in citric acid content between 120 and 180 d. Transcriptomic analysis across developmental stages revealed highly similar transcriptional profiles between these two cultivars, particularly at 120 and 180 d, suggesting conserved transcriptional regulatory mechanisms during the late stages of fruit ripening (Figure S6a and b) (Feng et al. 2021). We identified 1,171 genes that are regulated by ATAC-seq and at least one histone modification (H3K27ac, H3K4me3, or H3K27me3) and are also highly expressed between 120 and 180 d (Fig. 3f, Figure S6c, and Supplementary Table S15). These genes were primarily involved in organic acid metabolism, carbohydrate biosynthesis processes, and stress responses (Figure S7a), highlighting the metabolic and regulatory shifts associated with fruit ripening. Among them, HSF and the bZIP TF families are enriched (Figure S7b), which indicates that HSFs and bZIPs may play vital roles in the transcriptional reprogramming during citrus fruit ripening, potentially contributing to hormone or environmental response pathways that are critical for the final ripening phase.

Gene regulatory networks identified CrHSFA6B and CrbZIP5b as hub regulators controlling fruit ripening

TF footprint analysis in open chromatin regions (OCRs) enables a comprehensive genome-wide prediction of TF binding. We conducted ATAC-seq to identify OCRs, and an average of 109,508 high signal-to-noise OCRs across the three stages of fruit development were detected (Table S15). By integrating gene co-expression data with cis-motifs and TF footprints identified in the OCRs, we constructed a transcriptional regulatory network (TRN) controlling fruit development. Finally, a network containing 1,822 nodes and 64,916 edges mediated by 52 TFs was obtained (Fig. 4a and Table S16). To capture the core networks, transcription factors such as CrDREB2C, CrHSFA6B, CrbZIP5b, CrDof2.4, and CrESE3 were identified among the top 20 hub genes (Table S17). Among the 20 identified hub genes, 95% are marked by the activating histone modification H3K27ac, except for CrDREB2C (Fig. 4b). Furthermore, 60% of these hub genes lack the repressive chromatin mark H3K27me3 (Fig. 4b), underscoring the pivotal role of activating chromatin modifications in regulating hub transcription factors. Here, we extract networks of CrHSFA6B (Table S18) and CrbZIP5b (Table S19) to understand their mechanisms for fruit quality formation. Target genes of CrHSFA6B, which are enriched in response to heat and organic acid metabolic processes such as CrHSP70 and CrACO1/CrACO3, which are related to heat stress and citric acid degradation (Fig. 4c). Target genes of CrbZIP5 are related to response to heat, abscisic acid, and saccharide biosynthetic process, such as CrHSFA2/CrHSFA4A/CrHSFA6B, CrHVA22E, and CrSPSF1/CrSDH2-2/CrPFK3 (Fig. 4d).

Figure 4.

Gene regulatory network controlling citrus fruit ripening. (a) Overview of the transcriptional regulatory network (TRN) mediated by 52 core transcription factors (TFs) during citrus fruit ripening. The network comprises 1,822 nodes and 64,916 edges. The circle represents genes in the network, and the edges represent the regulatory relationships between genes. The edges are represented by gray lines. A Circular Layout is used to arrange the points into circles. (b) TRN of key hub genes associated with citrus fruit ripening. Circle size indicates the hub gene ranking, and circle color represents the type of histone modification present at each gene. (c) Predicted TRN of the hub transcription factor CrHSFA6B (top) and GO enrichment analysis of its predicted target genes (bottom). (d) Predicted TRN of CrbZIP5b (top) and GO enrichment analysis of its target genes (bottom). (e) Comparison of hub TFs regulated genes from TRN with DAP-seq or ChIP-seq.

Interestingly, our analysis revealed a regulatory interplay between CrbZIP5b and CrHSFA6B within the TRN (Fig. 4d). CrbZIP5b was found to activate CrHSFA6B as well as sucrose phosphate synthase (CrSPSF1), a key enzyme in sucrose metabolism (Fig. 4d). Concurrently, CrHSFA6B directly regulated genes involved in citric acid metabolism, including CrACO1 and CrACO3 (Fig. 4c). To validate our TRN, we performed DNA affinity purification sequencing (DAP-seq) for CrHSFA6B (Figure S8a). This approach identified putative CrHSFA6B binding sites, with a substantial proportion (34.72%) located in promoter regions, predominantly near the transcription start sites (TSSs) of annotated genes (Figure S8b). De novo motif discovery analysis revealed a consensus sequence for CrHSFA6B binding: GAANNTTCTAGA (Figure S8c). The 62.78%, 56.47%, and 57.89% of the regulatory interactions involving CrHSFA6B, CsDREB2C, and CrNAC25, respectively, in our constructed TRN were validated in vitro through DAP-seq or ChIP-seq (Fig. 4e), further validating the reliability of our network.

CrHSFA6B promotes citric acid degradation during citrus fruit ripening

The expression of CrHSFA6B was significantly up-regulated during the initiation of the fruit ripening stage (120– to 180 d) (Fig. 5a), accompanied by the increase of chromatin accessibility, H3K27ac, and H3K4me3 (Fig. 5b). Subcellular localization analysis (Figure S9) showed that CrHSFA6B protein is localized in the nucleus, supporting its role as a transcriptional regulator. Phylogenetic analysis revealed that CrHSFA6B was the Arabidopsis HSFA6B ortholog (Figure S10a). TRN network analysis showed that CrHSFA6B interacts with citric acid degradation-related genes (eg, CrACO1 and CrACO3) (Figs. 4c and 5c), which could be confirmed by DAP-seq of CrHSFA6B (Fig. 5d). Electrophoretic Mobility Shift Assays (EMSAs) demonstrated that CrHSFA6B could bind directly to the heat shock elements (HSEs) in the promoter regions of CrACO1 and CrACO3 (Fig. 5e). This binding was further validated by a luciferase (LUC) reporter assay (Fig. 5f and g), where co-transformation with CrHSFA6B effector and CrACO1/3 promoter (CrACO1pro, CrACO3pro) reporter constructs significantly enhanced LUC activity, confirming the activation of the CrACO1 and CrACO3 promoters by CrHSFA6B in vivo. To further investigate the function of CrHSFA6B, we generated overexpression (OE-CrHSFA6B) and RNA interference (RNAi-FmHSFA6B) constructs, which were transiently introduced into Fortunella margarita fruits for functional analysis (Fig. 5h). Additionally, we also generated stable overexpression lines (OE-CrHSFA6B) in C. unshiu callus (Figure S11a). RT-qPCR analysis revealed that expressions of Fm/CuACO1 and Fm/CuACO3 were markedly upregulated in OE-CrHSFA6B fruits and callus, whereas their expressions were significantly downregulated in RNAi-FmHSFA6B fruits (Fig. 5i and Figure S11b). Consistent with this result, the citric acid content was significantly lower in OE-CrHSFA6B and significantly higher in RNAi-FmHSFA6B compared with control (Empty vector, EV) (Fig. 5j and Figure S11d). These findings suggest that CrHSFA6B plays a role in regulating the expression of ACO genes, thereby participating in the citric acid degradation process during the citrus fruit ripening.

Figure 5.

CrHSFA6B promoted citric acid degradation during citrus fruit ripening. (A, B) Transcription (a) and epigenetic modification tracks (b) for HSFA6B. Transcription data were shown as mean ± SD of 3 biological replicates. (c) Schematic representation of the citric acid degradation pathway. ACO, aconitase; IDH, isocitrate dehydrogenase. (d) DAP-seq analysis testing binding of CrHSFA6B to the HSE motif on the promoters of CrACO1/3. (e) Electrophoretic mobility shift assay (EMSA) to test CrHSFA6B binding to the HSE motifs in the CrACO1 and CrACO3 promoters. “+” and “−” indicate the presence or absence of the indicated proteins and probes. FAM-P, 5′-FAM-labeled probe; Cold-P, unlabeled competitor probe; Mut-P, unlabeled mutated probe. (f-g) Dual-luciferase assay in N. benthamiana leaves testing CrHSFA6B activation of CrACO1 and CrACO3 expression. Schematic diagrams of the reporter and effector constructs (f). The LUC/REN ratio in OE-GFP was used as the control (g) (Student's t-test, Mean ± SD, n = 5). (h) Transient overexpression of CrHSFA6B and silencing of FmHSFA6B in F. margarita fruits. Scale bars = 0.5 cm. (i-j) Relative expression levels of CrHSFA6B, CrACO1, and CrACO3  (I), and quantification of citric acid content (j) in transgenic F. margarita fruit. Each biological replicate represents a pooled sample of ten independently infiltrated fruit segments. EV, empty vector control. (Student's t-test, Mean ± SD, n = 3 to 4).

CrbZIP5b promotes sucrose accumulation during citrus fruit ripening

RNA-seq data revealed that the expression of CrbZIP5b was increased progressively from 120 to 180 d (Fig. 6a), implying its active role during the later stages of fruit ripening. Chromatin accessibility and histone modification data showed a corresponding increase in ATAC-seq peaks and active histone marks (H3K27ac and H3K4me3), with a low level of the repressive H3K27me3 mark (Fig. 6b). Phylogenetic analysis revealed that CrbZIP5b was the Arabidopsis bZIP5 ortholog (Figure S10b). Moreover, the sequence and gene expression patterns of bZIP5 were conserved within citrus (Figure S12). TRN network analysis showed that CrbZIP5b interacts with CrSPSF1 (Fig. 4d), a key gene involved in sucrose biosynthesis (Fig. 6c). Further supporting this transcriptional regulatory pathway, EMSA results demonstrated that CrbZIP5b directly bound to the CATG motif in the promoter region of CrSPSF1 (Fig. 6d). A dual-luciferase (LUC) reporter assay showed that co-transformation with CrbZIP5b significantly activated the transcriptional activity of the CrSPSF1 promoter in vivo, leading to higher LUC activity in comparison to the EV control (Fig. 6e and f). CUT&Tag-qPCR analyses using citrus callus transfected with a 35S:CrbZIP5b-FLAG construct showed strong enrichment for the binding of CrbZIP5b to both P1 (4-fold) and P4 (5-fold) regions of the CrSPSF1 promoter but not to P0 region, which lacked a CATG motif binding site (Fig. 6g). To ascertain the function of CrbZIP5b in citrus fruit quality formation, we performed transient overexpression and silencing of CrbZIP5b in F. margarita fruit and generated stable overexpression lines in C. unshiu callus (Fig. 6h and Figure S11a). As expected, overexpression of CrbZIP5b led to a substantial increase in Fm/CuSPSF1 transcript level and a significantly higher sucrose content in citrus fruit and callus (Fig. 6i, j, and Figure S11c, e). In contrast, RNA interference of FmbZIP5b resulted in significantly decreased expression of FmSPSF1 and the sucrose content (Fig. 6i and j), suggesting that the bZIP5b-SPSF1 pathway plays a crucial role in enhancing sucrose accumulation during the ripening of citrus fruits.

Figure 6.

CrbZIP5b promoted sucrose accumulation during citrus fruit ripening. (a, b) Transcription and epigenetic modification tracks for CrbZIP5b. Transcription data shown as mean ± SD. of 3 biological replicates. (c) Schematic representation of sugar biosynthesis and metabolism pathways. FRK, Fructokinase; SPS, sucrose phosphate synthetase; SPP, Sucrose-phosphatase. (d) EMSA testing CrbZIP5b binding to the CATG motif on the CrSPSF1 promoter. +, presence of corresponding proteins and probes. -, absence of the corresponding proteins and probes. FAM-P, 5′-FAM-labeled probe; Cold-P, unlabeled competitor probes; Mut-P, unlabeled mutant probe. (e, f) Dual-luciferase assay in N. benthamiana leaves. Schematic diagrams of the reporter and effector constructs (E). The ratio of LUC/REN in the OE-GFP group was used as control (f) (Student's t-test, Mean ± SD, n = 5). (g) CUT&Tag-qPCR analysis. Anti-FLAG antibody was used to pull down DNA fragments from CrbZIP5b-3×FLAG transgenic citrus callus, followed by qPCR analysis using different primer pairs. (h) Transient overexpressing CrbZIP5b and silencing FmbZIP5b in F. margaritafruits. Scale bars = 0.5 cm. (i-j) Expression levels of CrbZIP5b and FmSPSF1 relative to FmActin7 (Cs_ont_1g004160) (I), and measurement of sucrose content (j) in transgenic F. margarita fruit. Each biological replicate represents a pooled sample of ten independently infiltrated fruit segments. EV, empty vector control. (Student's t-test, Mean ± SD, n = 3 to 4).

ABA promotes sucrose accumulation and citric acid degradation by activating CrbZIP5b during citrus fruit ripening

During citrus fruit ripening, sucrose and citric acid serve as two key metabolites, and their co-regulation is crucial for fruit flavor. The TRN results also support that CrbZIP5b potentially regulates CrHSFA6B (Fig. 4d). The EMSA experiment further confirmed that CrbZIP5b bound directly to the promoter region of CrHSFA6B (Fig. 7a), indicating that CrbZIP5b is involved in citric acid degradation via the heat shock factor pathway. This regulatory link was validated using a dual-luciferase reporter assay (Fig. 7b and c), where CrbZIP5b significantly activated the transcriptional activity of the CrHSFA6B promoter. CUT&Tag-qPCR analyses showed strong enrichment for the binding of CrbZIP5b to both P1 (5-fold) and P2 (4.6-fold) regions of the CrHSFA6B promoter but not to P0 region, which lacked a CATG motif binding site (Fig. 7d). RT-qPCR analysis revealed that the expression of FmHSFA6B, was markedly upregulated in OE-CrbZIP5b fruits, whereas it was significantly downregulated in RNAi- FmbZIP5b fruits (Fig. 7e and Figure S11c). The metabolic results showed that citric acid contents in overexpressed CrbZIP5b fruits and callus were lower than in the control, whereas CrbZIP5b interfering fruits had higher citric acid contents than the control (Fig. 7f and Figure S11f).

Figure 7.

ABA promoted sucrose accumulation and citric acid degradation by activating bZIP5b during citrus fruit ripening. (a) EMSA testing CrbZIP5b binding to the CrHSF6B promoter. +, presence of corresponding proteins and probes. -, absence of the corresponding proteins and probes. FAM-P, 5′-FAM-labeled probe; Cold-P, unlabeled competitor probes; Mut-P, unlabeled mutant probe. (b, c) Dual-luciferase assay in N. benthamiana leaves. Schematic diagrams of the reporter and effector constructs (B). The ratio of LUC/REN in the OE-GFP group was used as control (c) (Student's t-test, Mean ± SD, n = 5). (d) CUT&Tag-qPCR analysis. Anti-FLAG antibody was used to pull down DNA fragments from CrbZIP5b-3×FLAG transgenic C. unshiu callus, followed by qPCR analysis using different primer pairs (Student's t-test, Mean ± SD, n = 3). (e)  FmHSFA6B expression level in OE-CrbZIP5b and RNAi-FmbZIP5b fruit. (Student's t-test, Mean ± SD, n = 4). (f) Measurement of sucrose content in OE-CrbZIP5b and RNAi-CrbZIP5b fruits, each biological replicate represents a pooled sample of ten independently infiltrated fruit segments. EV, empty vector control. (Student's t-test, Mean ± SD, n = 3). (g) Temporal correlation of ABA, citric acid, and sucrose levels during fruit development (60 to 180 d), based on smoothed Z-score trends. (h) Phenotypes of citrus fruits (Fortunella margarita) following ABA injection at concentrations of 5, 10, or 50 μM. Scale bars = 2 cm. (i) Quantification of citric acid and sucrose contents in ABA-treated fruits. Each replicate represents a pooled sample from ten infiltrated fruit segments. (Student's t-test, Mean ± SD, n = 3). (j) Luciferase reporter assay confirmed ABA activation of the CrbZIP5b promoter in N. benthamiana leaves (Student's t-test, Mean ± SD, n = 5). (k) Relative expression levels of CrbZIP5b and CrHSFA6B in ABA-treated citrus fruits. Each replicate represents a pool of ten infiltrated fruit segments (Student's t-test, Mean ± SD, n = 4).

To determine whether these two core TFs can recruit histone-modifying enzymes (eg, HATs, HDACs) to modulate histone modification levels, a total of 13 histone-modifying enzymes (including HDACs, SRTs, and WDR5s) were identified to be differentially expressed during fruit development (Figure S13). Then, the potential interactions between CrHSFA6B/CrbZIP5b and these 13 candidate enzymes were tested using yeast two-hybrid (Y2H) and bimolecular fluorescence complementation (BiFC) assays. As shown in Figures S14 and S15, no direct protein–protein interactions were detected in either system.

Given the potentially important role of abscisic acid (ABA) in citrus fruit ripening (Figure S1), further analyses showed a dynamic relationship between ABA, citric acid, and sucrose contents over time. Specifically, ABA concentrations increased as citric acid levels decreased and sucrose accumulated from 120 to 180 d (Fig. 7g). Moreover, we found that the expression of ABA-related genes, including CrPP2Ca (Protein Phosphatase 2C a), CrUGT71Ba/b (UDP-Glycosyltransferase 71B a/b), CrPYL8 (Pyrabactin Resistance 1-Like 8), and CrNCED1/3 (9-cis-Epoxycarotenoid Dioxygenase 1/3), was gradually up-regulated during citrus fruit ripening (Figure S16). Notably, these genes showed increased chromatin accessibility and active histone modifications with fruit ripening (Figure S16), suggesting that the genes in ABA pathways were epigenetically regulated during citrus fruit ripening. ABA treatment experiments provided further evidence that ABA promoted citric acid degradation and sucrose accumulation (Fig. 7h and i), thereby enhancing the ripening process of citrus fruits.

Considering the significance of ABA and the core transcriptional role of CrbZIP5b in the citrus ripening process, we further analysed the correlation between ABA and the expression of CrbZIP5b. Luciferase reporter assays demonstrated that ABA directly activates the transcriptional activity of CrbZIP5b promoter (Fig. 7j). The gene expression analysis showed that ABA treatment in F. margarita enhanced the expression of FmbZIP5b and FmHSFA6B, which further resulted in significant upregulation of citric acid degradation genes (FmACO1, FmACO3) and sucrose biosynthesis gene (FmSPSF1) (Fig. 7k and Figure S17). As depicted in Fig. 8, the regulatory model of ABA-CrbZIP5b-CrHSFA6B regulatory cascade pathway elucidates the mechanism by which ABA activates CrbZIP5b, thereby modulating both sucrose accumulation and citric acid degradation during citrus fruit ripening. Specifically, CrbZIP5b enhances sucrose synthesis through the upregulation of CrSPSF1, a pivotal gene in sucrose biosynthesis, while concurrently regulating citric acid degradation via the activation of CrHSFA6B, which targets genes such as CrACO1 and CrACO3 that are involved in citric acid degradation.

Figure 8.

Proposed model for epigenetically mediated transcriptional regulation during citrus fruit ripening. Abscisic acid (ABA) plays a central role in coordinating transcriptional reprogramming through histone modifications and chromatin accessibility during citrus fruit ripening. As fruit transitions from the unripe (60 to 120 d) to the ripening stage (150 to 180 d), levels of the active histone marks H3K27ac and H3K4me3, along with chromatin accessibility, increase at the loci of CrHSFA6B, CrbZIP5b, and their target genes, while repressive H3K27me3 marks decrease. These epigenetic changes are accompanied by increased expression of CrHSFA6B and CrbZIP5b, elevated sucrose accumulation, and reduced citric acid content. Low ABA levels in early fruit development are associated with limited activation of target genes, whereas elevated ABA in late stages enhances CrbZIP5b activity and downstream transcriptional cascades, including CrSPSF1 and CrACO1/3.

Discussion

Fruit ripening is a complex and finely tuned developmental process that determines citrus flavor, shelf life, and consumer preference. This study uncovers a coordinated epigenomic and transcriptional regulatory framework underlying citrus fruit ripening, highlighting the dynamic roles of chromatin accessibility and histone modifications. We identified CrHSFA6B and CrbZIP5b as central transcriptional regulators that control citric acid degradation and sucrose accumulation, respectively. Importantly, we demonstrate that abscisic acid (ABA) triggers a regulatory cascade ABA-CrbZIP5b-CrHSFA6B that links hormone signaling to metabolic reprogramming during ripening. These findings provide mechanistic insights into the epigenetic regulation of sucrose and citric acid metabolism and identify key regulatory nodes that can be leveraged for the genetic improvement of fruit flavor and quality in citrus and other non-climacteric fruit crops.

Dynamic epigenetic regulation during citrus fruit ripening

The critical stage of citrus fruit ripening occurs between 120 and 180 d, during which genes with dramatically changing expression levels are linked to the reduction of organic acids and the increase of soluble sugars in citrus fruit (Figs. 1 and 2, and Figure S1). During this period, citric acid levels declined while sucrose accumulated, transforming the fruit's taste profile from sour to sweet (Li et al. 2024a; Mao et al. 2024). Most interesting, 6 citrus cultivars (including C. sinensis “AL” and “F72', C. paradisi “Cocktail”, C. reticulata “EG”, C. unshiu “DF4”, and F. margarita “YPJG”) exhibited a similar trend, indicating this process was conserved in citrus (Feng et al. 2021; Wei et al. 2021; Chen et al. 2023; Mao et al. 2024). Notably, ABA played a central role in regulating citrus fruit ripening, where the increased ABA levels from 120 to 180 d coincided with a marked shift in the fruit's metabolic profiles. This hormonal shift was critical in non-climacteric ripening and aligns with observations from other non-climacteric fruits, such as grapes and strawberries, where ABA similarly modulated ripening processes (Jia et al. 2011; Perotti et al. 2023).

Chromatin accessibility and histone modifications play various roles in plant growth, development, biotic and abiotic stress responses (Ueda and Seki 2020; Ageeva-Kieferle et al. 2021; Liu et al. 2022; Pan et al. 2024). Recent advances have highlighted the crucial role of chromatin accessibility and histone modifications involved in the fruit ripening processes of various fruit crops (Lü et al. 2018; Li et al. 2024b; Pan et al. 2024). In this study, changes in gene expression were significantly positively correlated with the changes in chromatin open accessibility, H3K4me3, and H3K27ac, while negatively correlated with H3K27me3 (Fig. 3a), which were in line with the previous studies in wheat regeneration and endosperm development (Liu et al. 2023; He et al. 2024). In ripening fruits, dynamic changes in these epigenetic marks facilitate the activation or repression of ripening-related genes, including those involved in fruit color changes, soluble sugar accumulation, and fruit softening (Jia et al. 2023; Li et al. 2024a; Wang et al. 2024). We identified 4,108 upregulated and 4,651 downregulated expressed genes during the core citrus initiation of fruit ripening stage (120 to 180 d), and 71.91%, 44.84%, 25.05%, and 57.81% of them were regulated by chromatin accessibility, H3K27ac, H3K4me3, and H3K27me3, respectively (Fig. 3d and Tables S5–S8). The expression of fruit ripening-related genes revealed a sequential pattern during citrus fruit ripening, regulated collectively by chromatin accessibility, H3K27me3, H3K4me3, and H3K27ac (Fig. 3). These results highlighted the role of dynamic chromatin accessibility and histone modifications in fine-tuning the ripening process of citrus fruit.

We detected significant alterations in histone modifications and chromatin accessibility at key regulatory genes; however, many of these changes, while subtle in magnitude, were functionally meaningful, reflecting the gradual and quantitative nature of fruit development rather than abrupt on-off transitions. These incremental chromatin adjustments may act cumulatively, facilitating the coordinated activation or repression of target genes in response to developmental cues (Yang et al. 2020; Huang et al. 2025). DNA methylation may also contribute to this regulatory fine-tuning, adding an additional layer of control over gene expression during fruit ripening (Huang et al. 2019; Song et al. 2024).

TRN provides a robust approach to elucidate the regulatory mechanism involved in the citrus fruit ripening process

Transcription factors are crucial regulators of gene expression, controlling fruit development, ripening, and responses to environmental cues (Karlova et al. 2014; Cao et al. 2024a; Jia et al. 2024a; Yue et al. 2025). The transcriptional regulatory network (TRN) is an effective approach to identify transcription factors and their regulation mechanism in plant growth and development (Liu et al. 2023; He et al. 2024). However, little is known about the TRN involved in the fruit ripening process. In this study, TF footprints in open chromatin regions (OCRs) were analysed, and a TRN controlling fruit ripening had been constructed in citrus, a representative non-climacteric fruit. Totally, 52 core transcription factors and 1,822 downstream target genes were identified to be regulated by chromatin accessibility and histone modifications during fruit ripening, and DREB2C, HSFA6B, bZIP5b, DOF2.4, and ESE3 were identified among the top 20 hub TFs. Several fruit homologous genes of them, DREB2F, ESE3, and Dof2/6/15, were reported to regulate the quality formation and softening in the fruit ripening (Zhai et al. 2022; Fang et al. 2024; Zheng et al. 2024). Notably, 56.47%, 62.78%, and 57.89% of the regulatory interactions involving the top hub TFs, CrDREB2C, CrHSFA6B, and CrNAC25 in our constructed TRN were validated in vitro through DAP-seq or ChIP-seq (Fig. 4e). Taken together, our fruit ripening-related TRN provides a reliable and robust approach to dissect the fruit ripening process in citrus.

CrbZIP5 is the bridge between the regulatory cascades of sucrose and citric acid metabolism during citrus fruit ripening.

CsPH5 and CsPH4 are involved in citric acid biosynthesis and accumulation during citrus fruit development (Strazzer et al. 2019; Huang et al. 2023). However, the expression, chromatin accessibility, and histone modifications of CrPH5 and CrPH4 exhibited weak correlations with changes in citric acid content during fruit ripening (Figure S5). The acid content was determined by both biosynthesis and degradation of citric acid. Thus, the citric acid degradation process contributed more to achieving the better taste of citrus fruit than the biosynthesis process in this study. CrbZIP5b and CrHSFA6B were upstream regulators of CrSPSF1 and CrACO1/3, respectively, which were known to be key construct genes of sucrose synthesis and citric acid degradation during fruit ripening (Li et al. 2016, 2017; Wang et al. 2018; Zhang et al. 2022b). Overexpression and RNAi of target genes in citrus fruit further revealed the function of CrHSFA6B and CrbZIP5b in citric acid degradation and sucrose accumulation (Figs. 5 and 6), indicating the pivotal role of CrHSFA6B and CrbZIP5b in the regulatory network during citrus fruit ripening and their essential functions in sucrose and citric acid metabolism.

In this study, we identified CrbZIP5b as a transcription factor that coordinates sucrose accumulation and citric acid degradation during citrus fruit ripening. CrbZIP5b functions as a central regulatory hub by directly binding to the promoter of CrHSFA6B and activating its expression, thereby promoting citric acid degradation while enhancing sucrose accumulation (Fig. 7). The high sequence conservation of CrbZIP5b among citrus cultivars (Figure S12a) suggests a conserved regulatory function within the genus. Members of the S1-bZIP subgroup, to which CrbZIP5b belongs, are established regulators of fruit ripening and quality formation across multiple species. In Solanum lycopersicum, overexpression of SlbZIP1 and SlbZIP2 significantly increased sucrose, glucose, and fructose contents in transgenic fruits (Sagor et al. 2016). Comparable functions have been reported for FvbZIP11 in Fragaria vesca (Zhang et al. 2022d), PbrbZIP15 in Pyrus bretschneideri (Jia et al. 2024b), PpybZIP43 in Pyrus pyrifolia (Zhang et al. 2022a), and PpbZIP18 in Prunus persica (Zhang et al. 2025), demonstrating a conserved role of S1-bZIP transcription factors in promoting sugar accumulation during fruit ripening. Notably, PpbZIP44 in P. persica acts as a dual regulator of sugar and organic acid metabolism by activating PpSDH9 and PpProDH1, thereby increasing fructose while reducing citrate and malate levels (Wang et al. 2023a). Collectively, these findings highlight the evolutionary conservation of S1-bZIP–mediated regulation of sugar–acid balance in fleshy fruits, with CrbZIP5b representing the key regulatory component in the Citrus genus.

Transcription factors are able to recruit histone-modifying enzymes to modulate the histone modification levels of the downstream target genes and thus regulate their expression levels (Li et al. 2025b). In this study, we also identified 13 histone-modifying enzymes (including HDACs, SRTs, and WDR5s) that were differentially expressed during fruit development (Figure S13). However, no direct protein-protein interactions were detected between CrHSFA6B/CrbZIP5b and 13 candidate histone-modifying enzymes. These results indicated that CrHSFA6B and CrbZIP5b were unlikely to regulate downstream target genes by directly recruiting histone deacetylases or acetyltransferases. Instead, their regulatory function may depend on cofactors that facilitate the recruitment of such chromatin-modifying proteins. For instance, the EAR motifs of MdZFP and SlERF.F12 transcription factors were known to mediate transcriptional repression through the recruitment of histone deacetylases via TPL (TOPLESS) proteins (Deng et al. 2022; Li et al. 2025a).

ABA is involved in the transcriptional and epigenetic regulation during citrus fruit ripening

ABA-induced ripening in both climacteric and non-climacteric fruits had evolutionary significance, enhancing fruit attractiveness to herbivores and thereby helping the dispersal of seeds (Leng et al. 2013; Fenn and Giovannoni 2021; Li et al. 2022). ABA was known to promote soluble sugar accumulation and organic acid degradation during fruit ripening in citrus, but its mechanism was largely unknown (Islam et al. 2015; Wang et al. 2016). Here, ABA levels were significantly increased in the late ripening stage, and the expression of ABA-related genes, including CrPP2Ca, CrUGT71Ba/b, CrPYL8, and CrNCED1/3, was gradually up-regulated with increased chromatin accessibility and active histone modifications during citrus fruit ripening (Figure S16). Recent studies have shown that SnRK2.6 was a key kinase in ABA signaling, and ABA-activated SmSnRK2.6 kinase phosphorylated SmbZIP5, enhancing its transactivation activity and stability in Salvia miltiorrhiza (Liu et al. 2025). In this study, we also found that ABA activated CrbZIP5b to regulate sucrose accumulation and citric acid degradation during citrus fruit ripening (Fig. 8), supporting that the ABA-mediated signaling was involved in the transcriptional and epigenetic regulation network of fruit ripening in citrus.

Taken together, these findings contributed to a broader understanding of sugar-acid dynamic regulation during ripening stages and provided effective strategies for controlling the fruit flavor and consumer preference of citrus. Further research would help to elucidate how these transcription factors coordinate the complex interactions between ABA signaling and the metabolic changes of sucrose and citric acid.

Methods

Plant materials and growth conditions

Ten-year-old “EG” (C. reticulata Blanco cv. Ponkan) trees from an orchard (Hubei Academy of Agricultural Sciences, Hubei Province, China) were used in the study. Fruits were sampled randomly from selected ten uniform trees on 60, 90, 120, 150, 180, and 210 Days after flowering (d). The size and weight of fruits on 10 trees were measured, and the average value was obtained. Twelve representative fruits were sampled from each tree at each developmental stage. After isolating the fruit tissues (juice sacs) by manual dissection, the samples were rapidly frozen in liquid nitrogen and stored at −80| °C. A portion of the samples was used for the determination of phytohormones, sucrose, and citric acid, and the other portion of the sample was used for RNA-seq, ATAC-seq, and CUT&Tag.

Plant materials used for RNA-seq, ATAC-seq, and CUT&Tag

Materials at 60, 90, 120, 150, 180, and 210 d were photographed with a digital camera (Canon 5D IV) for morphological observation. On the basis of the above observation, citrus materials for RNA-seq were collected at 60, 90, 120, 150, and 180 d, and samples of 120, 150, and 180 d were utilized for ATAC-seq and CUT&Tag. Three biological replicates were performed for RNA-seq and ATAC-seq, and two biological replicates were performed for CUT&Tag. For each biological replicate, approximately 10∼12 fruits were collected at different fruit development stages.

Sucrose and organic acid determination

The contents of sucrose and citric acid in 1 g of citrus juice sacs were determined using a gas chromatographic instrument (7890B; Agilent Technologies, Santa Clara, CA, USA), as previously described (Wang et al. 2025). Three independent extractions were performed per sample.

RNA-seq, ATAC-seq, and CUT&Tag experiment

Total RNAs were extracted using HiPure HP Plant RNA Mini Kit (Magen, R4165) according to the manufacturer's instructions. The products were sequenced using an Illumina NovaSeq 6000 by Personal Biotechnology (Shanghai, China), and RNA-seq data were obtained from three biological replicates. ATAC-seq and CUT&Tag assays were done following a previously described method (Liu et al. 2023; Song et al. 2025). Grind tissue into an extremely fine powder using liquid nitrogen. Add HBM buffer (25 mM Tris–HCl pH 7.5, 0.44 M sucrose, 10 mM MgCl₂, 10 mM β-mercaptoethanol, 0.1% Triton, and 2 mM spermine), mix thoroughly, and vortex at 100 rpm on ice for 15 min. Subsequently, pass the solution through 70-micron and 40-micron cell filters. After three consecutive washes, count the resulting crude nuclei on a hemocytometer (Countstar Rigel S3) using the AOPI or DAPI staining method. For ATAC-seq, the extracted nuclei (∼50,000 per reaction) were incubated with Tn5 transposase and tagmentation buffer at 37 °C for 30 min. DNA from ATAC was used to construct sequencing libraries following the protocol provided with the Hyperactive ATAC-Seq Library Prep Kit for Illumina (Vazyme Biotech, TD711). Finally, the Illumina Novaseq platform (Personal Biotechnology, Shanghai, China) was utilized for library sequencing. For CUT&Tag, the extracted nuclei (∼20,000 per reaction) were gently resuspended in 50 μl antibody buffer containing the corresponding antibody (H3K27me3, CST, C36B11, 1/50 dilution; H3K4me3, Abcam, ab8580, 1/50 dilution; H3K27ac, Abcam, ab4729, 1/50 dilution). Following an overnight incubation at 4 °C. DNA from CUT&Tag was used to construct sequencing libraries following the protocol provided with Hyperactive Universal CUT&Tag Assay Kit for Illumina Pro (Vazyme Biotech, TD904). Finally, the library was sequenced using an Illumina Novaseq platform by Personal Biotechnology (Shanghai, China). For CUT&Tag, qPCR was performed as described previously (Yu et al. 2024) using the ChamQ Universal SYBR qPCR Master Mix (Vazyme Biotech, Q711-02) by Roche Applied Biosystems. The relative expression levels were measured using the 2^−ΔΔCt analysis method. Primer sequences are listed in Table S20.

Plasmid construction and citrus transformation

Full-length coding sequence of CrHSFA6B or CrbZIP5b was cloned into pTOPO (Aidlab, CV17) and inserted into the Goldengate-compatible overexpression vector pGK1300-OE after being sequenced. The interference vector RNAi-CrHSFA6B and RNAi-CrbZIP5b were generated by cloning the CDS of CrHSFA6B or CrbZIP5b into pGK1300-RNAi, respectively. For transient transformation, the constructed vectors were transformed into Kumquat (Fortunella margarita) via Agrobacterium-mediated transformation (strain GV3101) following the described protocol (Gong et al. 2021). For stable transformation, the resultant recombinant vectors were transformed into A. tumefaciens (strain EHA105) and then transformed into the citrus callus according to a previously reported method (Wang et al. 2025). Primer sequences are listed in Table S20.

Phylogenetic analysis

To investigate the CrbZIP5b and CrHSFA6B transcription factors, protein sequences from Citrus sinensis and Arabidopsis thaliana were collected from publicly available databases: Citrus sinensis (SWOV, http://citrus.hzau.edu.cn), Arabidopsis thaliana (https://www.arabidopsis.org/). The maximum-likelihood tree was built by aligning amino acid sequences using Molecular Evolutionary Genetics Analysis (MEGA) software (version 11), and the bootstrap value scale of 1,000 trials was displayed at the branch points with 0.1 amino acid substitutions per site.

Quantitative polymerase chain reaction with reverse transcription (RT-PCR) analysis

Total RNA was extracted using HiPure HP Plant RNA Mini Kit (Magen, R4165). First-strand cDNA was synthesized using the HiScript II Q RT SuperMix for qPCR (Vazyme Biotech, R223-01) as recommended by the manufacturer and was stored at −20 °C. RT-qPCR was performed using the ChamQ Universal SYBR qPCR Master Mix (Vazyme Biotech, Q711-02) by Roche Applied Biosystems. Samples were normalized with CrActin7 expression. The relative expression levels were measured using the 2^−ΔΔCt analysis method. Primer sequences are listed in Table S20.

Luciferase reporter assay

The luciferase reporter assay was conducted in Nicotiana benthamiana leaves according to a previous report (Wang et al. 2025). The putative promoters (approximately 2.0 to 3.0 kb upstream of the ATG start codon) of CrACO1, CrACO3, and CrSPSF1 genes were amplified from ponkan genomic DNA and inserted into the cloning site of pGreen0800-LUC. The coding sequences of CrHSFA6B and CrbZIP5b were amplified and cloned into the effector vector (pGK1300-OE) under the control of the 35S promoter. All the primers used were listed in Table S20.

Subcellular localization

Subcellular localization was conducted in N. benthamiana leaves according to a previous report (Zhang et al. 2022c). The coding sequence (CDS) of CrHSFA6B and CrbZIP5b was cloned into the PC1300 vector. The 35S:AtH2B-RFP transgene was utilized as a nuclear marker. Fluorescence images were captured with a Leica TCS SP8 confocal laser scanning microscope (Leica Microsystems). For imaging, green fluorescent protein (GFP) and red fluorescent protein (RFP) were excited using an Argon laser at 488 and 583 nm, respectively. Primers used for cloning are listed in Table S20.

Protein purification and electrophoretic mobility shift assay

The coding sequence of CrHSFA6B and CrbZIP5b were cloned into the pET-M11 vector with a His-tag. The construct was transformed into E. coli BL21 (Rosetta), and protein expression was induced using 0.5 mM IPTG at 15 °C for 20 h. The expressed protein was purified following the manufacturer's instructions using the Ni-NTA His·Bind® Resin Kit (product no. 3421974). The EMSA was conducted as previously described (Wang et al. 2025), with modifications. Briefly, the His-tagged CrHSFA6B and CrbZIP5b proteins were expressed and purified as outlined above. The 5′ FAM-labeled oligonucleotide probes were synthesized and labeled by Shanghai Sangon Company (Shanghai, China). To perform binding reactions, the binding solution (0.1% Nonidet P-40, 1 mM benzamidine, 0.5 mM phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol, 50 μg·ml⁻¹ BSA and 100 ng·μl⁻¹ poly (dI-dC)), purified His-fused CrHSFA6B and CrbZIP5b and 1 μl of the 5′ FAM labeled probe (10 μmol·l⁻¹) were mixed together and incubated at 4 °C for 1 h. For the competition assays, the non-labeled probe was incubated with protein and binding buffer at 4 °C for 1 h. Next, 1 μl of the 5′-FAM-labeled probe (10 μmol·l⁻¹) was added and incubated at 4 °C for 1 h. The samples were loaded onto a prerun 6% polyacrylamide gel, and electrophoresis was conducted at 4 °C in 0.5× Tris-Borate-EDTA buffer for 1 hour in the dark. Gel images were captured using an Amersham Imager 600 (GE Healthcare, Tokyo, Japan). The sequence information of the probes is provided in Table S20.

RNA-seq data analysis

Raw reads were processed using fastp (v0.23.1) with the “–detect_ adapter_for_pe” parameter to remove adapters, trim low-quality bases, and filter bad reads (Chen et al. 2018). Clean reads were aligned to the citrus reference genome (Citrus sinensis v3.0, downloaded from http://citrus.hzau.edu.cn/) using STAR (version 2.4.2a) (Dobin et al. 2013). FeatureCounts (version 2.0.1) was used to count the number of paired mappings reads that overlapped each annotated gene (Citrus sinensis v3.0) (Liao et al. 2014). The count matrices were used as inputs to identify differentially expressed genes using the R package DESeq2 (v1.28.1) (Love et al. 2014), with thresholds of Padj < 0.05 and abs (log₂FoldChange) > 0.585. Transcripts per kilobase million (TPM) values generated from the counts matrix were used to characterize gene expression, and for PCA and hierarchical clustering analysis. K-means cluster analysis of genes was performed in R (v4.0.2) using Z-scaled TPM, and the gene expression heat map was displayed using the R package ComplexHeatmap(v2.4.3) (Gu et al. 2016). GO analysis was performed in AgriGO (http://systemsbiology.cau.edu.cn/agriGOv2/). The identification and family classification of TFs were predicted on the PlantTFdb website (http://planttfdb.gao-lab.org/) (Tian et al. 2020). Enrichment analysis of TF families for DEGs was achieved using the enricher function in the R package clusterProlifer (v3.16.1) (Wu et al. 2021).

CUT&Tag and ATAC-seq data analysis

Raw reads were processed using fastp (v0.23.1) with the “–detect_ adapter_for_pe” parameter to remove adapters, trim low-quality bases, and filter bad reads (Chen et al. 2018). Clean reads were mapped using Bowtie2(v2.4.1) on the C. sinensis v3.0 genome (http://citrus.hzau.edu.cn/) using the following parameters: –no-mixed –no-discordant -X 2000 (Langmead and Salzberg 2012). The aligned reads were sorted and filtered using samtools (v1.5) with the “-b -h -f 3 -F 4 -F 8 -F 256 -F 1024 -F 2048 -q 15” parameter (Danecek et al. 2021). Picard (v2.23.3) (https://broadinstitute.github.io/picard/) was used to remove the duplicated reads. The de-duplicated bam files from three biological replicates were merged using samtools (v1.5) and converted to BPM (Bins per million) normalized bigwig files with 1 bp bin size using deepTools (v3.5.1) (Ramírez et al. 2016), with the parameters “-bs 10 –effectiveGenomeSize 3.36e⁸ normalizeUsing RPKM –scaleFactor 1. The bigwig files were visualized using deeptools (v3.5.0) and IGV (v2.8.0.01) (Thorvaldsdóttir et al. 2013).

Peak calling was done using merged BAM files in MACS2 (v2.2.7.1) (Zhang et al. 2008). For ATAC-seq data, the parameter of peak calling using macs2 was “-q 0.05 -f BAMPE–nomodel–extsize 150–shift −75 -g 3.36e⁸. For narrow peaks of H3K27ac and H3K4me3, the peak calling parameter was “-p 1e-3 -f BAMPE -g 3.36e⁸ –keep-dup all”. For broad peaks of H3K27me3, the peak calling parameter was “-f BAMPE -g 3.36e⁸ –keep-dup all–broad–broad-cutoff 0.05'. Peaks from all samples of the same data type were merged using bedtools (v2.29.2) to generate the reference peaks (Quinlan and Hall 2010). Peaks were annotated to the citrus genome using the R package ChIPseeker (v1.24.0) (Yu et al. 2015). The whole genome was divided into three regions: promoter (from −3,000 bp to +500 bp of TSS), genic (from +500 bp of TSS to TES), and distal (other). Peaks located in promoter and genic regions were annotated to genes that overlapped with them, while peaks in distal regions were annotated to the nearest gene/TSS.

For quantification of ATAC-seq and CUT&Tag data, peaks from all samples were first merged to generate a unified reference peak set. Read counts under this reference set were then quantified and normalized to CPM (DBA_SCORE_TMM_READS_EFFECTIVE_CPM) using the R package DiffBind (v2.16.2) (Ross-Innes et al. 2012). The CPM values of the reference peak were used to perform PCA and hierarchical clustering analysis. The raw peak counts were used as input to identify differentially accessible peaks and differentially marked peaks by histone modification using the R package DESeq2 (v1.28.1) (Love et al. 2014). The heat maps centered on peaks were created using computeMatrix and plotHeatmap from deeptools (v3.5.0) (Ramírez et al. 2016).

To calculate correlations between transcriptome and epigenome data, Z-scaled TPM values of DEGs in 120 to 180 d and Z-scaled CPM values of peaks annotated to promoter and genic regions of these DEGs were calculated. Then, PCCs were calculated in R (v4.0.2) for each DEG.

Transcriptional regulatory network inference

TOBIAS was used for identifying TF footprints and establishing gene regulatory relationship inference (Bentsen et al. 2020). TF footprints with –motif-pvalue < 1e–4 were retained. Citrus TFs were predicted by PlantTFDB (http://planttfdb.gao-lab.org/prediction.php) and checked for best hits in Arabidopsis. The motif information was obtained from all Arabidopsis TFs in PlantTFdb (Tian et al. 2020). For the construction of the TRN, we focused on the set of 2,202 genes identified in Fig. 3e as being regulated by ATAC, which included both transcription factors and potential target genes. Among these, a total of 52 transcription factors were identified as regulators. In addition, to investigate the effect of histone modifications on fruit ripening, we paid special attention to the histone modification status of upstream TFs and target genes.

The TRN for different endosperm developmental stages was integrated and visualized using Cytoscape (v3.9.1) (Smoot et al. 2011). The connectivity for each gene, defined as the number of connections it has with other genes within the gene regulatory network, was calculated. Based on these connectivity scores, we identified the top-level genes with the highest connectivity, indicating their central role in the network.

DAP-seq data analysis

The DAP-seq libraries were sequenced on an Illumina NextSeq 500. The clean reads were aligned to the C. sinensis v3.0 genome (http://citrus.hzau.edu.cn/) using Bowtie2 (2.4.1) (Langmead and Salzberg 2012). Aligned reads were sorted, and duplicated reads were removed by samtools (v1.5) (Danecek et al. 2021). Peaks were called using MASC2 (v2.2.7.1) (Zhang et al. 2008). Putative genes associated with peaks were annotated using ChIPseeker (v1.24.0) (Yu et al. 2015).

Accession numbers

Sequence data can be downloaded from the NCBI SRA database under the accession number PRJNA1267850 or Citrus Genomics (http://citrus.hzau.edu.cn/) data libraries under accession numbers. CrbZIP5b (Cs_ont_1g023280), CrHSFA6B (Cs_ont_5g010840), CrACO1 (Cs_ont_1g028800), CrACO3 (Cs_ont_4g018980), CrSPSF1 (Cs_ont_4g002170), CrPSY (Cs_ont_6g006970), CrActin7 (Cs_ont_1g004160), CrPH4 (Cs_ont_2g035300), CrPH5 (Cs_ont_1g017980), CrPP2Ca (Cs_ont_9g027570), CrUGT71B6a (Cs_ont_6g011560), CrUGT71B6b (Cs_ont_6g000530), CrPYL8 (Cs_ont_9g027270), CrNCED3 (Cs_ont_5g033020), CrNCED1 (Cs_ont_7g028720), CrWDR5a-1 (Cs_ont_1g022800), CrWDR5a-2 (Cs_ont_9g026730), CrSRT1 (Cs_ont_4g020630), CrSRT2 (Cs_ont_1g006350), CrHDA2 (Cs_ont_4g005360), CrHDA5 (Cs_ont_5g038870), CrHDA6 (Cs_ont_5g045310), CrHDA8 (Cs_ont_8g009100), CrHDA9 (Cs_ont_8g007290), CrHDA14 (Cs_ont_1g005580), CrHDA15 (Cs_ont_5g037200), CrHDA19 (Cs_ont_8g029500), CrHD2B (Cs_ont_8g006930).

Author contributions

F.S., L-M.W., F-Q.T., C-L.L., and Y-C.J. conceived and designed the study; X.S., T-T.W., P.Z., T.L., C-M.S., Y.Z., and L-G.H conducted the experiments and data analysis, and Y-J.F.; X.S., T-T.W., and F.S. prepared the first draft; F-Q.T., A.B., and C-L.L. contributed to the final editing of the manuscript.

Supplementary material

Supplementary material is available at The Plant Cell online.

Funding

This work was supported by the National Key Research and Development Program of China (2023YFD2300600), the National Natural Science Foundation of China (32322073, U25A20687), Hubei Province Technology Innovation Plan Project (2024BBB085), Fundamental Research Funds for the Central Universities (2662025YLPY003), Hubei Province Supporting High Quality Development Fund Project for Seed Industry (HBZY2023B00501), Hubei Provincial Agricultural Science and Technology Innovation Fund (2024-620-000-001-023).

Data availability

Sequence data can be downloaded from the NCBl SRA database under the accession number PRJNA1267850.

Dive Curated Terms

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

• DAPI CHEBI: CHEBI:51231

• EMSA Gramene: Electrophoretic mobility shift assay

• EMSA Araport: Electrophoretic mobility shift assay

• PSY Gramene: AT5G17230

• PSY Araport: AT5G17230