Chromatin accessibility and histone landscapes define immune signal-specific transcriptional reprogramming in Arabidopsis

Abstract

Plant immunity relies on dynamic reprogramming of the transcriptome, but the underlying regulatory mechanisms that modulate chromatin accessibility and thus transcriptional activity remain poorly defined. In this study, we explore how salicylic acid (SA) and the pathogen-associated molecular pattern (PAMP) peptide flg22 reprogram immune gene expression through epigenetic modifications in Arabidopsis thaliana. We integrated RNA-seq, ChIP-seq, and ATAC-seq data to show that exogenous SA and flg22 induce distinct, widespread patterns of chromatin remodeling and differential gene expression. SA predominantly promotes a coordinated increase in H3K4me3 and H3 acetylation (H3Ac), two histone marks that are strongly associated with enhanced chromatin accessibility, at defense-related loci. By contrast, flg22 mainly activates defense gene expression via H3Ac, with minimal changes in H3K4me3 levels, suggesting distinct epigenetic strategies. Greater chromatin accessibility was closely linked to the activation of immune-responsive genes and correlated positively with H3K4me3 and H3Ac levels. Network analysis highlighted central roles for NPR1, ICS1, PR1, and proteasome-related genes in SA-triggered immunity. Collectively, our results reveal that SA and flg22 modulate immune gene expression through shared and unique chromatin-based pathways, offering insight into the epigenetic regulation of plant defense.

1. Introduction

Plants are challenged by numerous potential pathogens and diseases that pose a significant threat to food security. In response, plants have evolved complex immune systems that protect them against diverse microbial pathogens. Plant immune systems consist of two primary layers: pattern-triggered immunity (PTI) and effector-triggered immunity (ETI). ETI involves intracellular nucleotide-binding leucine-rich repeat receptors (NLRs) that detect effectors derived or secreted from the pathogen, while PTI is activated upon recognition of pathogen-associated molecular patterns (PAMPs) by pattern recognition receptors (PRRs) at the cell surface [[1], [2], [3]]. Well-characterized PAMPs include flg22 (a 22-amino acid [aa] peptide derived from bacterial flagellin), elf18 (an 18-aa peptide derived from the bacterial elongation factor EF-Tu), Pep1, necrosis- and ethylene-inducing-like proteins (NLPs), chitooctaose, 3-OH-FA (a bacterial hydroxylated fatty acid), and oligogalacturonides, each recognized by specific PRRs [[4], [5], [6], [7], [8], [9]]. In Arabidopsis (Arabidopsis thaliana), flg22 is recognized by its cognate receptor kinase FLAGELLIN-SENSITIVE 2 (FLS2). The binding of flg22 to FLS2 induces the latter's interaction with its co-receptor BRI1 ASSOCIATED RECEPTOR KINASE (BAK1), forming a heterodimer that activates plant immunity [6,10,11]. The immune response triggered by flg22 serves as one of the most common model systems for elucidating the mechanisms of plant-microbe interactions.

Plant defense responses such as stomatal closure, the reactive oxygen species (ROS) burst, callose deposition, and activation of defense-related phytohormones including salicylic acid (SA), jasmonic acid (JA), and ethylene [9,12]. SA is a major immune signal for basal and systemic acquired resistance (SAR) [13]. In Arabidopsis, approximately 20% of SA-induced changes in gene expression are mediated by the downstream signaling component NONEXPRESSER OF PR GENES 1 (NPR1). NPR1 and its homologs NPR1-LIKE PROTEIN 3 (NPR3) and NPR4 function as SA receptors with varying binding affinities [[14], [15], [16], [17]]. SA is a key phytohormone involved in resistance to biotrophic and hemibiotrophic pathogens. SA biosynthesis primarily occurs via the phenylalanine ammonia-lyase (PAL) and isochorismate synthase (ICS) pathways [18]. In rice, PAL4 and PAL6 are essential for pathogen-induced SA accumulation [19]. The transcription factors SAR DEFICIENT 1 (SARD1) and CALMODULIN-BINDING PROTEIN 60-LIKE g (CBP60g) regulate SA biosynthesis and SAR; their loss of function impairs systemic immunity [20]. SA primarily induces defense responses against biotrophic and hemibiotrophic pathogens [21]. NPR1 forms a complex with TGA MOTIF-BINDING PROTEINs (TGAs) and the histone acetyltransferase HAC, driving transcriptional activation of pathogenesis-related (PR) genes via histone acetylation [[22], [23], [24]]. Conversely, HDA6 represses SA biosynthesis by deacetylating chromatin at the CBP60g and SARD1 loci [25].

Epigenetics encompasses heritable changes in gene expression that do not involve changes in DNA sequence. Epigenetic regulatory mechanisms include DNA methylation, histone modifications, non-coding RNAs, chromatin remodeling, and RNA modifications [26,27]. In plants, epigenetic regulation governs a wide range of developmental programs such as flowering time, gametogenesis, morphological differentiation, and responses to environmental stimuli. Epigenetic modifications intersect with plant immune signaling. For example, flg22 suppresses the expression of genes involved in RNA-directed DNA methylation (RdDM), facilitating SA-mediated defense gene activation [28]. Histone methyltransferases such as Arabidopsis TRITHORAX 1 (ATX1), ATX1-RELATED 1 (ATXR1), and SET DOMAIN GROUP 8 (SDG8) contribute to PTI, ETI, and SAR by promoting immunity-related gene expression [29,30]. Genome-wide hypomethylation in response to flg22 has been reported in both rice (Oryza sativa) and tomato (Solanum lycopersicum), predominantly in the CHH context [31].

Genomic architecture and chromatin accessibility are key determinants of the functional state of a cell. Mounting evidence suggests that chromatin structure plays an important role in regulating plant defense responses through physical association with components or regulators of plant hormone signaling pathways [[32], [33], [34]]. In Arabidopsis, the major chromatin-remodeling factor families include SWITCH/SUCROSE NONFERMENTING (SWI/SNF), IMITATION SWITCH (ISWI), INOSITOL REQUIRING 80 (INO80), Swi2/Snf2-related 1 (SWR1), and chromodomain helicase DNA-binding (CHD) [35,36]. Epigenetic modulators such as the SWR1 complex, the SWI/SNF-related ATPase BRAHMA (BRM), and the histone demethylase LYSINE-SPECIFIC HISTONE DEMETHYLASE-LIKE 3 (LDL3) are also involved in SA-responsive regulation of gene expression [[37], [38], [39]]. The Arabidopsis SWI/SNF chromatin-remodeling ATPase SPLAYED (SYD) transcriptionally activates a subset of genes downstream of the JA and ethylene signaling pathways and confers resistance to the necrotrophic pathogen Botrytis cinerea. The chromatin-remodeling protein SWP73A suppresses expression of NLR genes through both direct transcriptional repression and modulation of alternative splicing. This suppression is relieved upon bacterial infection when induced small RNAs silence SWP73A, thereby activating NLR gene expression [40]. In soybean (Glycine max), poised chromatin states balance plant growth and innate immunity by maintaining PRR and NLR genes in a transcriptionally primed and low-expression state. Furthermore, despite displaying distinct epigenetic features, clustered PRR and NLR genes within the same topologically associating domains exhibit coordinated chromatin states and expression dynamics to ensure rapid defense activation upon stress [41].

Despite recent advances, the broader epigenetic dynamics regulating immune gene expression remain insufficiently characterized. Most studies focus on single genes or a single modification, and little is known about how different immune signaling pathways globally affect the chromatin landscape. In particular, how plants achieve specificity in targeting immune genes through general epigenetic mechanisms remains unclear. The PAMP flg22 is a well-established elicitor that triggers rapid and early PTI responses (such as early kinase cascades and ROS burst). By contrast, SA is the primary phytohormone acting downstream of early PAMP perception to mediate sustained defense responses and SAR. Previous studies have mostly focused on either early immune responses or late-stage disease resistance signaling, with little research on comparing and connecting these two pathways.

To distinguish between components that are purely responsive to primary early immune signaling (flg22-triggered) and those dependent on secondary downstream phytohormone accumulation (SA-triggered), we investigated chromatin epigenetic homeostasis in Arabidopsis plants treated with these two classic immune elicitors individually. We integrated results from transcriptome deep sequencing (RNA-seq), chromatin immunoprecipitation sequencing (ChIP-seq), and assays for transposase-accessible chromatin sequencing (ATAC-seq) to profile global changes in chromatin accessibility and histone modifications. We demonstrate that SA promotes a coordinated rise in H3K4me3 and H3 acetylation (H3Ac), while flg22 primarily modulates H3Ac independently of H3K4me3. Chromatin accessibility correlates strongly with activation of gene expression and enrichment of active histone marks. We identified key immune regulators such as NPR1, PR1, and ICS1, as central targets. These observations lay a theoretical foundation for understanding the interplay between plant immunity and chromatin dynamics, revealing that SA and flg22 employ distinct epigenetic mechanisms to precisely regulate plant immune responses.

2. Results

2.1. Transcriptome profiles of plants treated with salicylic acid or flg22

To elucidate the role of epigenetic mechanisms in the regulation of immune genes, we profiled the transcriptome of wild-type Arabidopsis accession Col-0 to SA or to the PAMP peptide elicitor flg22, representing phytohormone-mediated and PTI-triggered immune signals, respectively. We treated two-week-old seedlings with water (as mock), 1 μM SA for 10 h, or 1 μM flg22 for 1 h. We performed RNA-seq using two independent biological replicates (Fig. 1A). Principal component analysis (PCA) indicated strong reproducibility between replicates (Fig. S1A and B).

Fig. 1.

Transcriptional landscape induced by SA and flg22 immune signals. A Schematic overview of the experimental workflow for SA and flg22 treatment of Arabidopsis thaliana. B Volcano plots showing differentially expressed genes (DEGs) upon SA or flg22 treatment compared to the mock controls. Red dots indicate significantly upregulated genes, blue dots indicate downregulated genes, gray denotes non-significant changes, and purple marks genes significantly regulated by both treatments. C Venn diagram illustrating the overlap of DEGs between SA and flg22 treatments. D Heatmaps showing the expression patterns of common DEGs (left), SA-specific DEGs (middle), and flg22-specific DEGs (right). M-S: SA mock; SA: SA treatment; M-F: flg22 mock; flg22: flg22 treatment.

We defined differentially expressed genes (DEGs) based on a comparison to mock-treated seedlings, using the criteria of |log₂ (fold-change)| > 1 and P_adj < 0.05. We obtained 2496 and 1712 DEGs in the SA-treated and flg22-treated samples, respectively, with 429 shared genes (Fig. 1B–D, Table S1). A gene ontology (GO) term enrichment analysis showed that these shared DEGs, such as CBP60g, SARD1, PHYTOALEXIN DEFICIENT 4 (PAD4), and CHITIN ELICITOR RECEPTOR KINASE 1 (CERK1), are enriched in functional terms related to defense responses (Fig. S1C and D).

SA treatment led to 1414 upregulated DEGs and 1082 downregulated DEGs, whereas flg22 treatment predominantly induced gene expression, with 1476 upregulated DEGs and only 236 downregulated DEGs, suggesting that flg22 treatment significantly activates the PTI signaling pathway (Fig. 1B, Table S1). Among SA-specific DEGs, 1053 DEGs were upregulated and 1014 DEGs were downregulated, including the WRKY family transcription factor genes WRKY38 and WRKY54 (Fig. S1E). flg22 treatment specifically induced the significant upregulation of 1094 genes, including the PRR receptor gene FLS2, BOTRYTIS-INDUCED KINASE 1 (BIK1) encoding a receptor-like cytoplasmic kinase, and BAK1 encoding the co-receptor of FLS2 (Fig. S1F), whereas only 189 genes were significantly downregulated. These results confirm the robust and distinct transcriptomic changes initiated by SA and flg22.

2.2. Genome-wide histone modification dynamics induced by SA or flg22

To assess how immune signals affect chromatin states, we performed ChIP-seq (two independent biological replicates) for different histone modifications on the same plant samples. We profiled seven key histone marks: histone H3 (H3) acetylation (H3Ac), H3K4me1, H3K4me2, H3K4me3, H3K9me2, H3K27me3, and H3K36me3. PCA confirmed high reproducibility among replicates (Fig. S2). We looked for differentially enriched histone peaks mapping to DEGs to link chromatin dynamics with changes in gene expression (Tables S2 and S3).

To investigate whether the above SA-responsive and flg22-responsive changes in histone modifications influence the transcription of immunity-related genes, we examined the changes in histone marks at DEGs in response to SA or flg22. Among all examined histone modifications, H3Ac displayed the most notable dynamics while H3K4me3 showed the smallest changes. The H3Ac mark was significantly enriched at genes upregulated by both SA and flg22 treatments, and notably present at lower levels along downregulated DEGs (Fig. 2A and B). By contrast, H3K4me2 levels were moderately lower at upregulated DEGs but slightly higher at downregulated DEGs, consistent with the proposed role of this histone modification as a repressive epigenetic mark in plants [42]. H3K4me2 is a debated modification, and its function requires further exploration [42,43]. Moreover, there was no significant difference in the level of any of the other histone modifications along either upregulated or downregulated DEGs. These results highlight H3Ac and H3K4me3 as key histone modifications in transcriptomic regulation induced by SA or flg22 immune signaling, whereas H3K4me2 may play a secondary or context-dependent role.

Fig. 2.

SA and flg22 treatments induce dynamic histone modifications linked to the activation of immune-responsive genes. A Metaplots showing the average distribution of various histone modifications around the transcription start site (TSS), transcription end site (TES), and ±1.5 kb flanking regions of up- and downregulated DEGs under SA or flg22 treatment. B Heatmaps depicting the enrichment of histone marks across DEGs upon SA (B) and flg22 (C) treatment compared to the mock control.

The marked association of H3Ac and H3K4me3 marks with DEGs prompted us to further investigate their relationship with gene expression. We identified differentially enriched peaks (DE peaks) for each mark (false discovery rate [FDR] < 0.05). In SA-treated seedlings, we identified 1048 H3Ac-enriched peaks and 348 H3K4me3-enriched peaks (Fig. 3A and B). There were far more upregulated peaks than downregulated peaks, suggesting that SA treatment mainly leads to significantly higher levels for these two markers. Among the 343 SA-responsive H3K4me3 DEGs, 250 (∼73%) overlapped with SA-responsive H3Ac-enriched genes (Fig. 3B). This highly asymmetric overlap indicates that while H3K4me3 modifications are strongly accompanied by H3Ac during the SA response, H3Ac acts as a broader regulatory mark that frequently operates independently of H3K4me3. GO term enrichment analysis of these 250 overlapping genes indicated strong enrichment in defense-related biological processes (Fig. S3A). For instance, the SA receptor gene NPR4 was more highly expressed following SA treatment, alongside significant gains in H3Ac and H3K4me3 (Fig. 3C), while no such changes were observed in response to flg22. Similarly, WALL-ASSOCIATED KINASE 1 (WAK1), encoding an oligogalacturonide receptor involved in oxidative burst signaling [44], also showed strong SA-mediated induction of expression, together with elevated H3Ac and H3K4me3 levels (Fig. 3C). Notably, in addition to the synergistic regulation of some genes by H3Ac and H3K4me3, some genes exhibited selective enrichment of a single mark. For example, the JA-associated transcription factor gene WRKY50 in the JA defense pathway [45] showed a prominent increase in H3Ac peaks without a corresponding change in H3K4me3 (Fig. 3C), illustrating that the specificity of regulatory mechanisms is mainly associated with a robust increase of H3Ac.

Fig. 3.

H3K4me3 and H3Ac modifications contribute to SA- and flg22-induced activation of immune-responsive genes in a synergistic or independent manner. A Volcano plots showing the differential enrichment of H3Ac and H3K4me3 under SA (upper) and flg22 (lower) treatment relative to the mock controls. Red: significantly increased; blue: significantly decreased; gray: unchanged. B Venn diagram illustrating the overlap between differentially H3Ac- and H3K4me3-modified genes under SA or flg22 treatment. C Genome browser snapshots of representative genes showing ChIP-seq (H3Ac, H3K4me3) and RNA-seq signal tracks under mock, SA, and flg22 treatment. D Metaplot of H3K4me3 density centered on SA-induced differential H3Ac peaks. M, Mock; SA, SA treatment. E Boxplots comparing (left) H3K4me3 signal intensity and (right) gene expression (FPKM) at SA-induced differential H3Ac peaks. Student's t-test: ∗∗∗, P < 0.001. In all box plots, the central horizontal line indicates the median. The lower and upper borders of the box represent the 25th and 75th percentiles (first and third quartiles), respectively. The whiskers extend to the minimum and maximum values within 1.5 times the interquartile range (IQR) from the box borders. Outliers are shown as individual points. F Metaplot of H3Ac density centered on SA-induced differential H3K4me3 peaks. G Boxplots comparing (left) H3Ac signals and (right) gene expression at SA-induced differential H3K4me3 peaks. ∗∗∗, P < 0.001. H Metaplot of H3K4me3 density centered on flg22-induced differential H3Ac peaks. M, Mock; flg22, flg22 treatment. I Boxplots comparing (left) H3K4me3 signals and (right) gene expression at flg22-induced differential H3Ac peaks. Statistical significance was assessed by Student's t-test; NS: not significant, ∗, P < 0.05. J Metaplot of H3Ac density centered on flg22-induced differential H3K4me3 peaks. (K) Boxplots comparing (left) H3Ac signals and (right) gene expression at flg22-induced differential H3K4me3 peaks.

In contrast to SA, flg22 treatment resulted in 1054 DE-H3Ac genes and 590 DE-H3K4me3 genes (Fig. 3A, lower panels), but with minimal overlap between the two gene sets (Fig. 2B). This observation suggests that flg22-responsive immune signaling mobilizes H3Ac and H3K4me3 in largely independent fashions. For example, ETHYLENE-RESPONSE FACTOR 19 (ERF19), encoding a dehydration-responsive element (DREB) subfamily A-5 member of the APETALA 2 (AP2)/ERF transcription factor family known to suppress PAMP-triggered immunity [46], was significantly upregulated upon flg22 treatment and exhibited elevated H3K4me3 levels, with no correspondingly higher H3Ac levels (Fig. 3C). We observed a similar pattern at L-TYPE LECTIN RECEPTOR KINASE IX.2 (LECRK-IX.2), a gene encoding a receptor-like kinase involved in PTI and SA signaling [47,48]. Conversely, the calcium channel gene CYCLIC NUCLEOTIDE-GATED CHANNEL 20 (CNGC20), which contributes to calcium homeostasis and calcium-mediated autoimmunity [49], exhibited increased H3Ac deposition but unchanged H3K4me3 levels, highlighting gene-specific regulation.

Notably, only four genes exhibited co-enrichment of both H3Ac and H3K4me3 in response to treatment with SA or flg22 (Fig. 3B and C), including BAK-TO-LIFE 2 (BTL2). BTL2 was reported to mediate immune signaling via sensing BAK1 and SERK4 integrity and activating CNGC20 expression through a kinase-dependent mechanism [50]. These observations underscore the stimulus-specific nature of histone dynamics in plant immunity and suggest that SA and flg22 elicit chromatin remodeling through both shared and distinct epigenetic pathways.

Collectively, these data reveal that immune signals such as SA and flg22 induce significant, target-specific changes in histone modifications. While SA promotes coordinated increases in H3Ac and H3K4me3 deposition at immune-responsive genes, flg22 triggers more divergent, gene-specific patterns. This differential histone remodeling likely contributes to the specificity and breadth of immune gene activation in response to diverse pathogenic cues.

2.3. H3K4me3 and H3Ac mediate SA- and flg22-induced transcriptome changes in both synergistic and independent manners

To explore the roles of H3K4me3 and H3Ac in the regulation of gene expression induced by SA and flg22 in more detail, we assessed the extent of correlation in reciprocal scatterplots between H3K4me3 signal density at DE-H3Ac peaks under each treatment. Upon SA treatment, the densities of H3K4me3 and H3Ac signals were significantly elevated at DE-H3Ac and DE-H3K4me3 loci, respectively, compared to mock-treated controls (Fig. 3D–G). Furthermore, the genes associated with these SA-induced histone modifications (i.e., DE-H3K4me3 peaks and DE-H3Ac peaks) were significantly more highly expressed (Fig. 3E and G, right panels), suggesting a coordinated epigenetic regulation by H3K4me3 and H3Ac during SA-triggered immune responses.

By contrast, under flg22 treatment, there were no significant changes in H3K4me3 or H3Ac densities at the corresponding flg22-induced DE-H3Ac genes and DE-H3K4me3 peaks when compared to mock controls (Fig. 3H–K). Nonetheless, the expression of genes associated with these flg22-specific histone modifications was significantly upregulated (Fig. 3I and K, right panels), indicating that flg22 can independently modulate H3K4me3 and H3Ac levels to activate immune gene expression.

Collectively, these results support the idea that SA and flg22 initiate the transcriptional regulation of immunity-related genes through distinct or overlapping mechanisms involving H3K4me3 and H3Ac, either synergistically or independently.

2.4. SA treatment induces globally more accessible chromatin

Given the dynamic changes in histone modifications observed upon immune stimulation, we investigated whether chromatin accessibility at defense-related loci is similarly modulated by SA and/or flg22. To this end, we performed ATAC-seq on seedlings treated with SA, flg22, or mock to assess chromatin accessibility across the genome (Fig. S2A, PCA). We defined differentially accessible peaks (DE-ATACs) using an FDR <0.05 and |log₂ fold-change| > 0. In SA-treated samples, we obtained a total of 2798 DE-ATAC peaks (Table S2), the vast majority of which (2746 peaks) were more accessible relative to the mock samples (Fig. 4A and B). This observation suggests that SA treatment promotes a widespread chromatin transition from a closed state to a more open state. GO enrichment analysis of the 2593 genes associated with these 2798 peaks revealed strong enrichment in biological processes related to defense and plant immunity (Fig. S4A).

Fig. 4.

SA-induced chromatin accessibility promotes the transcriptional regulation of immune-responsive genes. A Circos plot showing RNA-seq and ATAC-seq profiles of Arabidopsis under SA and mock control treatment. From the outer to inner ring: chromatin accessibility under SA treatment, chromatin accessibility under mock treatment, transcriptome data under SA treatment, and transcriptome data under mock treatment. B MA plot showing genes associated with significantly increased (n = 2746) or decreased (n = 52) chromatin accessibility upon SA treatment. Blue dots: not significant. C Metaplots (upper) and heatmaps (lower) showing chromatin accessibility across gene bodies (TSS to TES) and ±1.5 kb flanking regions for all genes and DEGs under SA treatment. D Protein–protein interaction network of the top 50 genes ranked by MCC from SA-induced differentially accessible genes. E Heatmap showing fold changes in chromatin accessibility and expression levels for the top 50 MCC genes. F Genome browser tracks showing the transcription and chromatin accessibility of representative immune-responsive genes under SA vs. mock treatment. G Venn diagram showing the overlap between SA-induced differentially accessible genes and DEGs. H Boxplot showing the expression levels of the 436 overlapping genes in (G). Statistical significance was assessed using Student's t-test (∗∗, P < 0.01).

Importantly, while chromatin accessibility showed only a modest rise globally, the chromatin of DEGs following SA treatment was markedly more open (Fig. 4C). These results indicate that SA signaling selectively enhances chromatin accessibility at immunity-related genes, thereby facilitating their transcriptional activation.

2.5. SA-induced chromatin remodeling facilitates transcriptional regulation of immune genes

To identify central regulators involved in SA-mediated transcriptional regulation, we analyzed the protein–protein interaction (PPI) networks of SA-responsive differentially accessible genes (DAGs) using the STRING database. We used the cytoHubba plugin in Cytoscape to calculate maximal clique centrality (MCC) scores; we constructed a network from the 50 genes with the highest MCC values (Fig. 4D).

Most of these 50 genes were associated with plant immune responses. Notably, key components of the SA signaling pathway, including the SA receptor gene NPR1, the SA biosynthetic gene ICS1, and the SA-responsive marker gene PR1, ranked highly in centrality (Fig. 4E and F). Multiple genes encoding subunits of the 20 S proteasome complex, such as PAF1, PBF1, PAE1, PAD2, RPT6A, PTP4B, and PBB2, also exhibited high and significant centrality, suggesting a potential role for proteasome-mediated protein degradation in maintaining the homeostasis of immune regulators during SA signaling. In addition, key transcription factor genes involved in regulation of immune gene expression were also prominent within the network, including members of the WRKY family (WRKY33, WRKY70) and the TGA family (TGA3).

Among the 2593 SA-induced DAGs, 436 genes were also differentially expressed compared to mock (Fig. 4G), and were generally upregulated following SA treatment (Fig. 4H). Taken together, these results demonstrate that greater SA-induced chromatin accessibility plays a pivotal role in the transcriptional regulation of immune genes, and suggest the central gene networks that may coordinate this response.

2.6. Greater flg22-induced chromatin accessibility enhances transcription of immune genes

To investigate the effect of flg22 treatment on chromatin architecture, we similarly performed ATAC-seq following flg22 treatment and identified 1550 genomic regions with higher accessibility and 982 regions with lower accessibility relative to the mock-treated control (Fig. 5A and B, Table S3). These results indicate that activation of PTI signaling significantly modulates chromatin accessibility. Consistent with observations under SA treatment, chromatin accessibility at DEGs was markedly elevated in flg22-treated samples, compared to the overall gene set (Fig. 5C). This result suggests that both SA and flg22 signaling pathways enhance chromatin accessibility specifically at immune-responsive loci, thereby facilitating their transcriptional reprogramming.

Fig. 5.

Flg22-induced chromatin accessibility contributes to the transcriptional regulation of immune-responsive genes. A Circos plot showing RNA-seq and ATAC-seq profiles of Arabidopsis under flg22 and mock control treatment. Rings (outer to inner): chromatin accessibility under flg22 treatment, chromatin accessibility under mock treatment, transcriptome data under flg22 treatment, and transcriptome data under mock treatment. B MA plot showing genes associated with significantly increased (n = 1550) or decreased (n = 982) chromatin accessibility upon flg22 treatment. Blue dots: not significant. C Metaplots (upper) and heatmaps (lower) showing chromatin accessibility across gene bodies and ±1.5 kb flanking regions in all genes and DEGs under flg22 treatment. D Protein–protein interaction network of the top 50 MCC-ranked genes from flg22-induced differentially accessible genes. E Heatmap showing fold changes in chromatin accessibility and gene expression for the top 50 MCC genes. F Genome browser views showing the transcriptional and chromatin accessibility profiles of representative genes under flg22 vs. mock treatment. G Venn diagram illustrating the overlap between flg22-induced differentially accessible genes and DEGs. H Boxplot showing the expression of the 437 overlapping genes in (G). Statistical significance was assessed using Student's t-test (∗∗∗, P < 0.001).

MCC analysis revealed that the top 50 genes with highest centrality in the flg22-induced DAG network were predominantly ribosome-related genes, including RPS7 and various RPL family members (Fig. 5D–F). Among the 2285 DAGs identified under flg22 treatment, 437 genes were also differentially expressed, and their expression levels were significantly elevated following treatment (Fig. 5G and H). These results indicate that flg22-responsive chromatin remodeling contributes directly to the transcriptional activation of immune genes, including those involved in the translation machinery and protein biosynthesis.

2.7. Coordination between SA-induced chromatin accessibility and histone modifications

Given the substantial changes in chromatin accessibility and histone dynamics following immune stimulation, we investigated the interplay between histone modifications and chromatin accessibility in regulating transcriptional reprogramming. Accordingly, we quantified the levels of various histone marks at SA-responsive or flg22-responsive DAGs before and after treatment. SA treatment resulted in significantly higher H3K4me3 and H3Ac levels, with a modest elevation in H3K4me2 deposition. By contrast, the levels of other histone modifications showed minimal or no significant changes (Fig. 6A and B). These results suggest a positive correlation between chromatin accessibility and the levels of H3K4me3 and H3Ac, two marks typically associated with transcriptional activation.

Fig. 6.

Integration of chromatin accessibility, histone modification, and gene expression data with SA-induced transcriptional regulation. A Metaplot showing the distribution of histone marks across gene bodies (TSS to TES) and ±1.5 kb flanking regions at genes with significantly increased chromatin accessibility under SA treatment. B Boxplot comparing histone modification levels between SA-treated and mock-treated samples. ∗, P < 0.05, ∗∗, P < 0.01, NS: not significant (Student's t-test). C Metaplots showing chromatin accessibility, H3Ac, and H3K4me3 levels across upregulated (left) and downregulated (right) DEGs. D Correlation analysis of chromatin accessibility and histone modifications in up- (left) and downregulated (right) DEGs. Red/yellow: positive correlation; blue: negative correlation. ‘+’ indicates SA treatment, ‘-’ indicates mock treatment. E Venn diagrams showing the overlap among SA-induced deH3Ac, deH3K4me3, DEGs, and increased or decreased chromatin accessibility. F GO enrichment of overlapping genes from (E), highlighting biological processes (BP) and molecular functions (MF) (n = 89). G Heatmap showing fold changes in chromatin accessibility, gene expression, H3Ac, and H3K4me3 levels for overlapping genes (n = 89). H Genome browser snapshots illustrating chromatin, histone modification, and expression signals at selected target genes under SA and mock conditions.

We examined the ChIP-seq and ATAC-seq signals at SA-responsive DEGs. Upon SA treatment, H3Ac and H3K4me3 signals were markedly enriched at upregulated genes (Fig. 6C), while the abundance of these marks was lower at downregulated genes. Surprisingly, chromatin accessibility was elevated for both upregulated and downregulated genes, suggesting that chromatin access alone is not always sufficient for transcriptional activation and must be interpreted in the context of histone modification states.

We performed a Spearman correlation analysis to assess and quantify the potential relationships among histone modifications, chromatin accessibility, and gene expression. For both upregulated and downregulated SA-responsive DEGs, levels of H3Ac, H3K4me3, and chromatin accessibility exhibited strong pairwise positive correlations, whereas H3K4me2 and H3K36me3 displayed weaker positive associations (Fig. 6D). Conversely, levels of H3K9me2, H3K27me3, and H3K4me1, which are commonly associated with repression of gene expression [51], showed significant negative correlations with H3Ac, H3K4me3, and chromatin accessibility. Consistent with these trends, a substantial number of SA-responsive DEGs demonstrated concurrent elevation in chromatin accessibility, H3Ac, and H3K4me3 levels (Fig. 6E–G). Functional annotation revealed that most of these genes are involved in SA signaling and plant defense responses (Fig. 6F), including key regulators such as SARD1 and WRKY70 (Fig. 6H).

Together, these results strongly support a model in which SA-responsive immune signaling enhances chromatin accessibility and modulates active histone marks (H3K4me3 and H3Ac) at immune-responsive loci. This epigenetic reprogramming facilitates transcriptional activation, thereby contributing to robust immune gene expression during plant defense responses.

2.8. Interplay between flg22-responsive chromatin accessibility and histone modification dynamics

To investigate the epigenetic mechanisms underlying flg22-responsive transcriptional regulation, we examined the relationship between chromatin accessibility and histone modification dynamics. Analysis of genes exhibiting greater chromatin accessibility upon flg22 treatment revealed a significant elevation in their H3Ac levels compared to mock controls (Fig. 7A and B), consistent with trends observed under SA treatment (Fig. 6A and B). Notably, the levels of other histone modifications, including H3K4me3, remained largely unchanged following flg22 treatment (Fig. 7A and B). This observation contrasts with the SA-responsive response, in which both H3Ac and H3K4me3 levels rose significantly (Fig. 6A and B), suggesting distinct epigenetic regulatory mechanisms between PTI (flg22) and SA signaling.

Fig. 7.

Integration of chromatin accessibility, histone modification, and gene expression data with flg22-induced transcriptional regulation. A Metaplot showing histone mark distributions across gene bodies and ±1.5 kb flanking regions at genes with significantly altered chromatin accessibility under flg22 treatment. B Boxplot comparing histone modification levels between flg22-and mock-treated samples (∗∗∗, P < 0.001, NS: not significant; Student's t-test). C Metaplots showing chromatin accessibility, H3Ac, and H3K4me3 levels at up- (left) and downregulated (right) DEGs under flg22 treatment. D Correlation analysis of chromatin accessibility and histone modifications in flg22-induced up- (left) and downregulated (right) DEGs. ‘+’ indicates flg22 treatment, ‘-’ indicates mock treatment. E Venn diagrams showing the overlap among flg22-induced deH3Ac, deH3K4me3, DEGs, and increased or decreased chromatin accessibility. F GO enrichment of overlapping upregulated genes (n = 23) focusing on the BP category. G Heatmap showing fold changes in chromatin accessibility, gene expression, H3Ac, and H3K4me3 levels for overlapping genes (n = 23). H Genome browser snapshots illustrating transcription, chromatin, and histone modification signals at representative flg22-responsive loci under flg22 and mock conditions.

We further explored the distribution of H3K4me3, H3Ac, and ATAC-seq signals across flg22-responsive DEGs before and after treatment. Consistent with the observed SA responses, chromatin accessibility was significantly higher at both upregulated and downregulated genes following flg22 treatment (Fig. 7C). For upregulated genes, H3Ac and H3K4me3 levels were both moderately elevated, with H3Ac showing a more prominent change (Fig. 7C). By contrast, downregulated genes exhibited substantially lower H3Ac levels, while H3K4me3 levels remained largely unchanged, indicating that flg22-responsive chromatin opening at downregulated genes may occur independently of H3K4me3 remodeling.

To assess the relationship between chromatin accessibility and histone marks, we performed another Spearman correlation analysis. Consistent with the SA response, chromatin accessibility exhibited a strong positive correlation with both H3Ac and H3K4me3 levels in flg22-responsive upregulated and downregulated DEGs (Fig. 7D). Notably, H3K4me1 levels displayed a strong negative correlation with chromatin accessibility in both upregulated and downregulated DEGs, whereas H3K9me2 and H3K27me3 levels showed weaker but nevertheless negative associations, particularly in the upregulated gene set (Fig. 7D). A subset of 23 upregulated genes displayed concurrent increases in H3Ac, H3K4me3, and chromatin accessibility (Fig. 7E–G). These genes included several key regulators of immune responses, such as the SA biosynthesis gene CBP60g, the SA-responsive gene CYSTEINE-RICH RLK 9 (CRK9) [52], the H₂O₂-responsive gene ZINC FINGER OF ARABIDOPSIS THALIANA 11 (ZAT11) involved in cell death [53], and the IMPAIRED OOMYCETE SUSCEPTIBILITY 1 (IOS1) gene encoding a receptor-like kinase functioning in PTI through association with FLS2 and EF-TU RECEPTOR (EFR) (Fig. 7H).

These results suggest that flg22-triggered transcriptional activation of specific immune genes is facilitated by chromatin remodeling and selective enrichment of active histone marks, particularly H3Ac, and to a lesser extent H3K4me3. Overall, our results support a model in which PTI signaling via flg22 promotes transcriptional regulation through targeted epigenetic modulation, which is distinct from, but also partially overlapping with, SA-mediated immune regulation.

2.9. Histone-modifying enzymes orchestrate chromatin remodeling to regulate transcriptional reprogramming of immune genes

Previous data showed that H3K4me3 and H3Ac deposition undergoes significant changes upon treatment with SA or flg22. To evaluate the changes in chromatin modifications of immune-related genes under different treatments and genetic backgrounds, we performed ChIP-qPCR to examine the enrichment of H3Ac and H3K4me3 at the loci BTL2, WAK1, NPR4, ERF19, and LECRK-IX.2 in the wild type (Col-0) and mutants (hda6, atx1 atx2 jmj28) following mock treatment with water (Mock), SA, or flg22. In Col-0, SA treatment promoted the significant enrichment of H3Ac and H3K4me3 at BTL2, WAK1, and NPR4 (Fig. 8A–C). Furthermore, flg22 treatment significantly induced H3Ac and H3K4me3 deposition at BTL2, ERF19, and LECRK-IX.2 (Fig. 8A, D, E). Under mock conditions, the basal H3Ac enrichment levels for all tested genes in the hda6 mutant were significantly higher than those in the Col-0 Mock-treated group. Upon treatment with SA or flg22, H3Ac levels in the hda6 mutant rose compared to the hda6 Mock-treated group. However, with the exception of LECRK-IX.2 under flg22 treatment (where enrichment reached levels similar to those in flg22-treated wild type), the absolute induced H3Ac levels in the mutant remained lower than those in the wild-type-treated groups (Col-0-SA or Col-0-flg22).

Fig. 8.

Mutations in HDA6, ATX1, ATX2, and JMJ28 alter H3Ac/H3K4me3 deposition and chromatin accessibility at immune-responsive genes upon SA or flg22 perception. A-E ChIP-qPCR of H3Ac and K4me3 at immune-responsive genes in the wild type, hda6, and atx1 atx2 jmj28 under SA and flg22 treatment. Different letters above the bars indicate significant differences based on one-way ANOVA followed by Tukey's honestly significant difference (HSD) test, P < 0.05. F qPCR of chromatin accessibility in the wild type and atx1 atx2 jmj28 mutant upon SA and flg22 treatment. G qPCR of chromatin accessibility in the wild type and hda6 mutant upon SA and flg22 treatment.

Under mock conditions, the basal H3K4me3 levels at BTL2, NPR4, ERF19, and LECRK-IX.2 in the atx1 atx2 jmj28 mutant with loss of three H3K4 methyltransferases were significantly lower than those in Col-0. Following SA or flg22 treatment, the H3K4me3 levels in the triple mutant rose relative to its corresponding mock-treated group, although they remained significantly lower than those in the treated Col-0 groups. Notably, H3K4me3 levels at WAK1 under mock treatment were comparable between Col-0 and the atx1 atx2 jmj28 mutant. However, upon SA treatment, H3K4me3 enrichment was higher in the mutant than in Col-0 (Fig. 8B).

To further investigate the direct consequences of changes in H3K4me3 and H3Ac levels on the physical architecture of target gene promoters, we employed ATAC-qPCR to assess the chromatin accessibility of key immunity-related genes in Col-0, atx1 atx2 jmj28, and hda6 following flg22 or SA treatment. In Col-0, SA or flg22 treatment significantly enhanced chromatin accessibility at the promoter regions of ERF19, BTL2, CBP60g, and NPR4, indicating a shift towards a more open chromatin state (Fig. 8F and G). However, in the atx1 atx2 jmj28 mutant, chromatin opening induced by SA or flg22 treatment was severely impaired (Fig. 8F). This structural occlusion is highly consistent with the defective H3K4me3 enrichment observed in this mutant via ChIP-qPCR. The hda6 mutant led to elevated basal chromatin accessibility and abnormal induction, consistent with the previously mentioned elevated H3Ac levels (Fig. 8G).

3. Discussion

Plants rely on a sophisticated immune system to recognize and respond to a wide variety of pathogens. A variety of immune signals, including the PTI-activating PAMP flg22 and SA, are known to regulate the expression of defense genes. However, the chromatin-level mechanisms that translate external immune signals into changes in gene expression remain underexplored. In this study, we provide a comprehensive and integrated analysis of chromatin accessibility, histone modification dynamics, and the transcriptome of Arabidopsis seedlings during immune responses induced by SA or flg22. Our results demonstrate that SA and flg22 trigger distinct but also partially overlapping epigenetic changes that converge on chromatin accessibility as a key regulatory hub for transcriptional activation of immune genes.

A major contribution of this work is the delineation of stimulus-specific chromatin-remodeling programs. Through a combined analysis of RNA-seq, ChIP-seq, and ATAC-seq data, we showed that SA treatment leads to widespread increases in chromatin accessibility and enrichment of active histone marks, particularly H3K4me3 and H3Ac, at immune-related genes. These results support a model in which SA-driven immunity involves coordinated modification of both nucleosome positioning and histone modifications to achieve robust transcriptional regulation.

The involvement of histone acetylation and H3K4me deposition in plant immune regulation has been widely reported. Histone acetyltransferases from the CBP/p300 family form a coactivator complex with NPR1 and are recruited to PR genes through TGAs upon reception of the SA signal, resulting in the activation of PR genes via histone acetylation–mediated epigenetic regulation [54]. In turn, the histone deacetylases HDA6 and HDA19 repress the expression of PR genes by removing histone acetylation from their promoters [55,56]. HDA19 can also interact with WRKY38 and WRK62, two transcription factors required for transcriptional activation and SA-mediated suppression of JA signaling [57,58], to repress their transcriptional activation activity. In addition, HDA6 can also target the CBP60g and SARD1 genes for transcriptional repression through histone deacetylation [25]. For H3K4me3, a group of methyltransferases has been implicated in immune regulation. For example, ATX1 is required for the indirect activation of PR1 expression [59], and ATXR7 participates in the transcriptional activation of many PTI, ETI, and SAR genes [30]. Removal of H3K4me3 by histone demethylases also exerts an important influence on plant immunity. In Arabidopsis, the H3K4 demethylases JUMONJI 14 (JMJ14), JMJ8, and LDL4 have been shown to regulate the activation of immune genes from different pathways, including SUPPRESSOR OF NPR1-1, INDUCIBLE 1 (SNI1) and WRKYs [[60], [61], [62]]. These studies indicate that H3K4me3 and H3Ac are the master epigenetic regulators of plant immune gene transcriptional regulation, which is consistent with the conclusions of this study.

The epigenetic patterns associated with flg22-induced PTI responses were different from those affected by SA treatment. Although chromatin became more accessible at immune-responsive genes upon flg22 treatment, the effect on H3K4me3 levels was minimal, with H3Ac being the primary histone mark enriched. Compared to previous studies [63], these results provide a fuller picture of the underlying chromatin dynamics of PTI responses. Our data suggest that flg22-responsive transcriptional activation may rely more heavily on rapid chromatin opening and acetylation, bypassing the need for stable trimethylation of H3K4, a mark often associated with long-term transcriptional memory or poised states. These results suggest that rapid stress responses in plants may be driven by dynamic histone acetylation and nucleosome remodeling rather than changes in histone methylation.

Notably, following treatment with either SA or flg22, we observed significant chromatin opening at both upregulated and downregulated genes (Fig. 6, Fig. 7C). This observation underscores the complexity of chromatin-based regulation, suggesting that chromatin opening alone is not always sufficient for transcriptional activation, but instead may prime loci for context-specific regulation by transcription factors or additional chromatin modifiers. The positive correlations between chromatin accessibility, H3Ac, and H3K4me3 observed in our Spearman analyses support the notion of coordinated epigenetic remodeling at activated genes (Fig. 6, Fig. 7D). Conversely, negative correlations with repressive marks such as H3K9me2, H3K27me3, and H3K4me1 emphasize the antagonistic relationship between active and repressive chromatin states in the immune response.

In addition, our study integrates chromatin accessibility data with protein–protein interaction networks, thereby enabling the identification of central regulatory nodes within the SA-responsive and/or flg22-responsive gene sets. In SA-treated samples, the top 50 genes ranked by MCC included key immune regulator genes such as NPR1, ICS1, and PR1, as well as several genes encoding 20 S proteasome subunits (Fig. 4D). The prominence of proteasome components in the SA network suggests that protein degradation may play an important role in maintaining immune homeostasis, potentially by regulating the turnover of transcription factors and signaling proteins. This idea is supported by earlier work showing that NPR1 stability is modulated by the proteasome and is crucial for its transcriptional coactivator function [63]. By contrast, the top MCC genes in the flg22-induced accessibility network were primarily associated with ribosomal function and translation. While ribosome biogenesis is not traditionally associated with immune signaling, recent studies have begun to uncover a role for translational control in PTI [64,65]. Our results suggest that flg22-triggered chromatin remodeling may extend beyond transcriptional regulation to influence the protein synthesis machinery, providing a potential mechanism for rapid immune protein production.

We also identified a subset of immune genes, including CBP60g, WRKY70, ZAT11, and IOS1, that displayed coordinated increases in chromatin accessibility, as well as H3K4me3 and H3Ac deposition following SA and flg22 treatments. These genes appear to represent epigenetically primed loci capable of responding robustly to diverse immune stimuli. Similar patterns have been observed in studies of SAR and stress memory, where early exposure to pathogens or abiotic stress induces a primed state characterized by modified chromatin structure and faster reactivation upon subsequent challenge [66,67]. Our genome-wide data expand on these models by demonstrating that epigenetic priming can occur through multiple chromatin-related mechanisms and is pathway-specific.

A major contribution of our study is its elucidation of the distinct epigenetic strategies that govern chromatin accessibility during different phases of plant immunity. We demonstrate that PAMP-triggered immunity (via flg22) relies primarily on a rapid H3Ac-driven mechanism to induce local chromatin opening. In contrast, SA-mediated defense requires a more complex, coordinated deposition of both H3K4me3 and H3Ac. The existence of these two distinct epigenetic responses serves as a global regulatory rheostat: the single-mark (H3Ac) approach allows for rapid, transient gene activation during early basal defense, whereas the dual-mark (H3K4me3 and H3Ac) synergy drives robust, sustained, and amplified transcriptional reprogramming essential for SA-triggered immunity. Thus, the precise interplay between specific histone-modifying complexes (ATX1, ATX2, JMJ28, and HDA6) and chromatin architecture ensures tailored and energetically efficient defense responses.

Together, our results support a model in which chromatin accessibility acts as a central integrator of immune signaling and transcriptional control. SA and flg22 both transcriptionally regulate immune genes but do so through distinct chromatin signatures: SA predominantly relies on H3K4me3 and H3Ac co-enrichment, while flg22 activates genes primarily through H3Ac. This distinction reflects a broader theme in plant biology, namely, that epigenetic regulation is highly flexible and context-dependent, capable of tailoring gene expression programs to the specific needs of plant immunity (Fig. 9).

Fig. 9.

The roles of salicylic acid and flg22 in the epigenetic regulation of plant immunity in Arabidopsis. The model illustrates the divergent chromatin modification strategies employed during immune responses induced by flg22 (left panel) and SA (right panel), as evaluated by RNA-seq, ChIP-seq, and ATAC-seq. A Perception of the flg22 peptide at the plasma membrane activates a signaling pathway that relies primarily on increased histone H3 acetylation, with minimal changes observed in H3K4me3 levels. This H3Ac-dependent process leads to increased chromatin accessibility and the subsequent activation of defense genes. B By contrast, SA treatment is associated with a coordinated increase in both H3K4me3 and H3Ac modifications. This coordinated epigenetic response is closely associated with enhanced chromatin accessibility, resulting in the strong activation of immune-responsive genes, including those with central roles in SA-triggered immunity, such as NPR1, ICS1, PR1, and proteasome-related genes.

4. Materials and methods

4.1. Plant growth conditions and treatments

Wild-type Arabidopsis Col-0 plants were grown in a growth chamber at 22 °C under a 16-h light/8-h dark cycle. Fourteen-day-old seedlings were subjected to four treatments: two groups were treated with 1 mM SA and water, respectively, for 10 h, while the other two groups were treated with 1 μM flg22 and water, respectively, for 1 h. A portion of each seedling sample was blotted dry and frozen in liquid nitrogen for RNA extraction, another portion of fresh tissue was used for ChIP-seq library construction, and the rest of the sample was used to construct an ATAC-seq library.

4.2. RNA-seq and ChIP-seq

For RNA-seq, total RNA was extracted from frozen tissue using TRIzol reagent. Library construction and deep sequencing were performed using the Illumina HiSeq 4000 platform according to the manufacturer's protocols (Majorbio, Shanghai, China). Two or three biological replicates were performed.

ChIP assays were performed as previously described [68]. In brief, 3 g of seedling tissue was harvested and cross-linked using 1% formaldehyde. Nuclei were extracted from the sample using nuclei isolation buffer (0.25 M sucrose, 15 mM PIPES pH 6.8, 5 mM MgCl₂, 60 mM KCl, 15 mM NaCl, 1 mM CaCl₂, 0.9% Triton X-100, and protease inhibitor cocktail). The nuclei were lysed with 1% SDS buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM PMSF, 1% SDS, 0.1% Na deoxycholate, 1% Triton X-100, and protease inhibitor cocktail) to release the chromatin, which was sonicated into 200–600 bp fragments using a Bioruptor (Diagenode). The sample was diluted 10-fold, and the chromatin was precipitated overnight using the appropriate antibodies (antiH3, 1:200 dilution; anti-H3K4me1, 1:200 dilution; anti-H3K4me2, 1:200 dilution; anti-H3K4me3, 1:200 dilution; anti-H3K27me3, 1:200 dilution; anti-H3K9me2, 1:200 dilution; anti-H3K36me3, 1:200 dilution; anti-H3Ac, 1:200 dilution) and incubated with Protein G Dynabeads at 4 °C for 2 h. After two sequential washes with low salt, high salt, LiCl, and TE buffer, the DNA-protein complex was eluted at room temperature using elution buffer (0.5% SDS and 0.1 M NaHCO₃). After reversing cross-linking and performing proteinase K/RNase A treatment, the DNA was recovered using phenol/chloroform/isoamyl alcohol. For ChIP-seq, DNA libraries were generated using a NEBNext Ultra II DNA Library Prep Kit for Illumina and sequenced on the HiSeq 2500 platform (Illumina).

4.3. ATAC-seq

ATAC-seq libraries were prepared as previously described with some modifications [69]. In brief, 1 g of seedling tissue was harvested, mixed with 500 μL of pre-chilled lysis buffer (15 mM Tris-HCl pH 7.5, 20 mM NaCl, 80 mM KCl, 0.5 mM spermine, 5 mM 2-mercaptoethanol, and 0.2% Triton X-100) on ice, and quickly chopped with a blade. The resulting solution was filtered through a 40 μm filter, and the filtrate was collected into a Falcon 5 mL Round Bottom Polystyrene Test Tube (Falcon, Cat No:352054) and resuspended in 2.0 mL lysis buffer. The sample was stained with DAPI (Sigma, Cat: MBD0015), and nuclear sorting was performed using a flow cytometer (FACSAria III, BD Biosciences). The nuclei were sorted into 500 mL lysis buffer (50,000 nuclei per reaction), centrifuged at 1000 g for 10 min at 4 °C, and washed once with Tris-Mg buffer (10 mM Tris-HCl pH 8.0, 5 mM MgCl₂). The purified nuclei were incubated in Tn5 transposase and tagmentation buffer (Vazyme, Cat: TD501) at 37 °C for 30 min for fragmentation. The sample was purified using VAHTS DNA Clean Beads (Vazyme, Cat: N411-01), followed by PCR amplification for 12–15 cycles using high-fidelity polymerase. DNA fragments (300–500 bp) were selected through two rounds of selection using different ratios of VAHTS DNA Clean Beads. Two or three biological replicates were performed. Following library construction, the library was sequenced by the company (GENEWIZ, Suzhou) using the Illumina NovaSeq 4000 platform.

4.4. RNA-seq data analysis

For RNA-seq data processing, the paired-end raw reads were cleaned and adapter sequences removed using TrimGalore (Cutadapt) [70]. After trimming, quality control was performed using FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and MultiQC to obtain clean FQ files. Clean reads were separately aligned to the reference genome using HISAT2 [71]. The number of paired mapping reads that overlapped with each annotated gene (Araport11) was counted using featureCounts [72] with the parameter "-p". The count files were then used as input for differential gene expression analysis using DESeq2 [73]. The threshold of differential gene expression was “p.adj <0.05 and abs (log₂FoldChange) > 1". Additionally, ggplot2 (https://ggplot2.tidyverse.org/), tidyr (https://tidyr.tidyverse.org/), dplyr (https://dplyr.tidyverse.org), and other R packages were used for data visualization and plotting.

4.5. ChIP-seq data analysis

For ChIP data processing, the paired-end raw reads were cleaned and adapter sequences removed using TrimGalore. After trimming, quality control was performed using FastQC and MultiQC to obtain clean FQ files. Bowtie2 (https://bowtie-bio.sourceforge.net/bowtie2/index.html) was used to align the clean reads to the Arabidopsis thaliana reference genome TAIR10. Mapped reads with low mapping quality (MAPQ <30) or multiple positions in the genome were identified and removed by SAMtools [74]. Biological replicates were merged using SAMtools. Sorted BAM files were used for peak calling with MACS2 [75]. For the H3K9me2 and H3K27me3 ChIP-seq datasets, the “broad” parameter was used. To standardize and visualize individual and merged replicate datasets, BAM files were converted to BigWig format using bamCoverage from deepTools [76]. A bin size of 10 bp was applied, and normalization was performed using Reads Per Kilobase per Million mapped reads (RPKM). The peaks called using MACS2 and differential peaks identified by DiffBind (https://hbctraining.github.io/Intro-to-ChIPseq/lessons/08_diffbind_differential_peaks.html) were annotated based on “Arabidopsis_thaliana.TAIR10.51.gff3” and the ChIPseeker package [77] using the “annotatePeak” function. Peaks/genes with FDR <0.05 and log₂FoldChange >0 were considered to be peaks/genes with increased chromatin accessibility, FDR <0.05 and log₂FoldChange <0 as peaks/genes with decreased chromatin accessibility, and FDR >0.05 as peaks/genes with no significant changes in chromatin accessibility. Metaplot diagrams and heatmaps of chromatin modification intensity were generated using the ComputeMatrix, plotProfile, and plotHeatmap programs from deepTools. GO analysis was performed using clusterProfiler [78], with reference pathway data from the ‘org.At.tair.db’ (v3.19.1) (https://www.bioconductor.org). Integrative Genomics Viewer was used to visualize the signals.

4.6. ATAC-seq data analysis

ATAC-seq data analysis was performed as previously described with some modifications [69]. Raw data was trimmed using fastp [79] with default parameters (for ATAC-seq, the adapter sequence was set to "-a CTGTCTCTTATACACATCT”). After trimming, Bowtie2 was used for alignment with the parameter --very-sensitive. SAMtools was used to sort and filter out low-quality data, followed by duplicate removal using Picard (https://broadinstitute.github.io/picard/). To standardize and visualize individual and merged replicate datasets, BAM files were converted to BigWig format using bamCoverage from deepTools. A bin size of 10 bp was applied, and normalization was performed using Reads Per Kilobase per Million mapped reads (RPKM). The peaks called by MACS2 and differential peaks identified by DiffBind were annotated based on “Arabidopsis_thaliana.TAIR10.51.gff3” using the ChIPseeker package with the “annotatePeak” function. GO analysis was then performed using ClusterProfiler. Genes with different accessibility levels were used to construct the regulatory network. The cytoHubba [80] plugin in Cytoscape [81] was used to compute the network maximal clique centrality (MCC) values, allowing transcription factors to be sorted in descending order based on MCC values. Genes with the Top 50 MCC values were used to construct a heatmap. Integrative Genomics Viewer was used to visualize the signals.

CRediT authorship contribution statement

Ding-Tian Yu: Writing – original draft, Project administration, Methodology, Investigation, Data curation. Si-Si Xie: Project administration, Investigation, Data curation. Cheng-Guo Duan: Writing – review & editing, Writing – original draft, Project administration, Conceptualization.

Declaration of competing interest

The authors declare that they have no competing interests. Author Cheng-Guo Duan was not involved in the journal's review of this manuscript.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Cheng-Guo Duan (duanchg@mail.sysu.edu.cn). All unique/stable reagents generated in this study are available from the lead contact. The ATAC-seq, ChIP-seq and RNA-seq data (BioProject PRJCA038790) were deposited in National Genomics Data Center (https://ngdc.cncb.ac.cn).