SWI/SNF chromatin remodeling factor BRAHMA promotes de novo shoot regeneration by epigenetic priming via H3K27me3 removal

SUMMARY

Plants have a remarkable capacity for regeneration. Recent studies have identified associations between plant regeneration and epigenetic regulators, thereby supporting the hypothesis that dynamic gene expression changes occur during the regeneration process. Notably, the association with chromatin remodeling factors remains to be elucidated. In this study, we demonstrated that BRAHMA (BRM), a core ATPase of the BRM‐associated SWI/SNF (BAS) chromatin remodeling complex, plays a crucial role in the shoot regeneration process via root‐derived callus formation. Phenotypic and transcriptomic analyses demonstrated that BRM exerts a substantial effect on the transition of gene expression from callus formation to shoot regeneration. Furthermore, epigenomic analysis revealed that BRM contributes to the removal of the silencing mark H3K27me3 indirectly from shoot regeneration‐related genes during callus formation, suggesting cooperative functions with plant‐specific H3K27me3 demethylases. The transcriptional activation of shoot regeneration‐related genes from which H3K27me3 was removed during callus formation did not occur until shoot induction. This suggests that BRM is involved in epigenetic priming, which puts shoot regeneration‐related genes in a primed state that allows gene expression immediately after shoot induction. We identified 24 BRM‐mediated epigenetic priming targets, which are not expressed during callus formation but are rapidly transcribed after shoot induction. Furthermore, out of these targets, the transcription factor NGATHA 3 (NGA3) and the glycine‐rich protein DEFECTIVELY ORGANIZED TRIBUTARIES 1 (DOT1), are involved in the shoot regeneration process through epigenetic priming.

Significance Statement

The chromatin remodeler BRAHMA (BRM) promotes shoot regeneration through epigenetic priming. In callus formation, H3K27me3 at BRM‐primed genes is reduced without transcriptional change, creating a poised state. REF6 and ELF6 demethylases may contribute to this process. Upon shoot induction, these genes for shoot regeneration are upregulated while retaining H3K27me3.

BRM facilitates de novo shoot regeneration by mainly involving gene expression changes from callus formation to shoot regeneration, and epigenetic priming via H3K27me3 removal. Understanding the mechanism of plant regeneration has the potential to promote the improvement of the breeding of useful plants through tissue culture techniques and genetic modification, as a solution to current food and environmental problems.

INTRODUCTION

It has long been recognized that plants possess a remarkable regenerative capacity, which had already attracted scientific interest by the end of the 19th century (Fehér, 2019). Plants frequently face physical damage caused by biotic or abiotic environmental factors (Fehér, 2019). For this reason, they are equipped with dedicated developmental pathways to repair damaged tissues and regenerate lost organs (Fehér, 2019). This regenerative capacity has given rise to practical techniques such as cutting and grafting in horticulture and agriculture, which are still widely used today (Ikeuchi et al., 2019; Melnyk & Meyerowitz, 2015). Tissue culture techniques have further extended this principle, enabling the genetic improvement of valuable plants and offering potential solutions to current food security and environmental challenges. Consequently, understanding the plant regenerative capacity is crucial for advancing in biotechnology and agriculture.

One established approach to study the molecular mechanism of regeneration is the use of in vitro tissue culture systems that induce pluripotent cell masses known as calli. In this system, small pieces of plant tissue (explant) are cultured on callus inducing medium (CIM) rich in auxin to form a callus. The callus is then cultured on a shoot inducing medium (SIM) rich in cytokinin to regenerate a shoot (Skoog & Miller, 1957; Sugimoto et al., 2019; Valvekens et al., 1988). This regeneration process requires dynamic trans‐differentiation. In fact, callus, the preliminary stage of organogenesis, has been found to exhibit traits similar to those of the root apical meristem (RAM), regardless of the tissue of origin (Sugimoto et al., 2010). It has also been suggested that the removal of RAM tissue memory and the increased expression of shoot apical meristem (SAM)‐related genes are important for shoot regeneration from callus (Ishihara et al., 2019). Furthermore, it has been revealed that the acquisition of regenerative capacity occurs during callus formation (Che et al., 2007) and is essential for shoot regeneration (Zhai & Xu, 2021). Trans‐differentiation alters the fate of a large number of cells, resulting in widespread changes in gene expression (Che et al., 2007; Sugimoto et al., 2010; Tanurdzic et al., 2008). Recent studies have revealed that these dynamic, genome‐wide gene expression changes are influenced by epigenetic regulation.

Epigenetic regulation, involving factors that influence DNA methylation and histone modifications, has been strongly correlated with plant organ regeneration. Notable enzymes involved in this process include the following: DNA methyltransferase METHYLTRANSFERASE 1 (MET1) (Li et al., 2011); histone deacetylase (HDAC) HISTONE DEACETYLASE 19 (HDA19) (Temman et al., 2023); histone demethylase LYSINE‐SPECIFIC DEMETHYLASE 1‐LIKE 3 (LDL3) (Ishihara et al., 2019); histone demethylase JUMONJI C DOMAIN‐CONTAINING PROTEIN 30 (JMJ30) and histone methyltransferase ARABIDOPSIS TRITHORAX‐RELATED 2 (ATXR2) (Lee et al., 2018); and histone acetyltransferase (HAT) GENERAL CONTROL NONDEREPRESSIBLE 5 (GCN5) (Kim et al., 2018). However, the relationship between chromatin remodeling complexes and plant regeneration remains unclear. Chromatin remodeling complexes are evolutionarily conserved protein complexes that regulate chromatin structure by altering nucleosome composition and interactions (Narlikar et al., 2013). Eukaryotic chromatin remodeling complexes include the SWItch/Sucrose Non‐Fermenting (SWI/SNF) subfamily remodelers, the Imitation SWItch (ISWI) subfamily remodelers, the Chromodomain Helicase DNA binding (CHD) subfamily remodelers, and the Inositol Requiring 80 (INO80) subfamily remodelers (Ojolo et al., 2018). Among these remodelers, the SWI/SNF subfamily remodelers are among the most comprehensively studied chromatin remodeling complexes (Ojolo et al., 2018). SWI/SNF chromatin remodelers utilize the energy of ATP hydrolysis and play a crucial role in modifying chromatin structure (Guo & He, 2024). In recent years, remarkable progress has been made in elucidating the subunit composition (Huang et al., 2025). In Arabidopsis thaliana, SWI/SNF ATPases include BRAHMA (BRM), SPLAYED (SYD), and MINUSCULE 1/2 (MINU1/2) (Guo & He, 2024). Recent studies have revealed that three distinct SWI/SNF complexes exist in A. thaliana, termed BRM‐, SYD‐, and MINU1/2‐associated SWI/SNF complexes (BAS, SAS, and MAS), based on the ATPase subunits they contain (Fu et al., 2023; Guo et al., 2022). Until these two studies were conducted, the full composition of the plant SWI/SNF remodeler had long remained unclear (Huang et al., 2025). They are involved in diverse processes such as transcription, DNA replication, and maintenance of genome stability (Guo & He, 2024). In plants, they participate in key processes such as growth and development including meristem formation and maintenance, cell differentiation, and organogenesis (Ojolo et al., 2018). Thus, chromatin remodeling factors play an essential role in transcription through dynamic epigenetic regulation by altering chromatin structure, which may be relevant to plant regeneration processes that require trans‐differentiation. Therefore, this study focuses on the chromatin remodeling complexes and aims to elucidate their role during the shoot regeneration process.

In recent years, epigenetic analyses have also provided insights into the preparatory mechanisms underlying future growth, differentiation, and regeneration (Handa & Matsunaga, 2024). During this preparatory phase, neither phenotypic changes nor detectable alterations in gene expression occur (Handa & Matsunaga, 2024). This unique process, termed ‘epigenetic priming’, sets the stage for rapid transcription in response to environmental stimuli or differentiation induction (Bonifer & Cockerill, 2017; Dillon, 2012). Epigenetic priming has been reported in humans in contexts such as embryonic stem cell differentiation (Bryan et al., 2025), cancer development (Vicente‐Dueñas et al., 2018), blood cell differentiation (Walter et al., 2008), and skin repair (Handa & Matsunaga, 2024). In plant shoot regeneration, the elimination of H3K4me2 by LDL3 during callus formation represents an epigenetic priming process that facilitates the rapid activation of SAM‐related genes following shoot induction (Ishihara et al., 2019). However, beyond this example of H3K4me2 removal, it remains unclear to what extent epigenetic priming in plant organ regeneration extends to other histone modifications.

In this study, we focus on BRAHMA (BRM), a core ATPase of the BAS chromatin remodeling complex (Fu et al., 2023; Guo et al., 2022). BRM plays an important role in plants by regulating transcription (Li et al., 2015, 2022; Yang et al., 2022), phytohormone signaling pathways (Efroni et al., 2013; Nishioka et al., 2020; Wu et al., 2015; Yang et al., 2015), and chromatin loosening, in association with histone acetylation to reduce susceptibility to DNA double‐strand breaks (Sakamoto et al., 2018). We demonstrate that BRM contributes to the removal of H3K27me3 from shoot regeneration‐related genes during callus formation, potentially in cooperation with the plant‐specific H3K27me3 demethylases, RELATIVE OF EARLY FLOWERING 6 (REF6) and EARLY FLOWERING 6 (ELF6). We further identify the B3‐type transcription factor NGATHA 3 (NGA3) and the glycine‐rich protein DEFECTIVELY ORGANIZED TRIBUTARIES 1 (DOT1) as the primary targets of this process. In particular, mutants of each gene showed suppressed shoot regeneration phenotypes, consistent with the epigenetic priming effect mediated by BRM. Here, we propose that the elimination of H3K27me3 via the chromatin remodeling factor BRM, is a crucial step in epigenetic priming for the shoot regeneration process.

RESULTS

The chromatin remodeling factor BRM promotes shoot regeneration from root‐derived callus

To investigate the role of BRM on shoot regeneration from root explants, we used a two‐step organ regeneration system via callus formation (Sugimoto et al., 2010) (Figure 1a). Roots were excised 2 cm from the root tip at 7 days after germination (DAG), transferred to CIM for 7 days, and subsequently transferred to SIM for 14 days. The shoot regeneration rate was calculated as the percentage of explants that regenerated shoots relative to the total number of explants. The results showed that the shoot regeneration rate was significantly reduced in both T‐DNA insertion mutants brm‐3 (Farrona et al., 2007) and brm‐20 (Zhang et al., 2017) (Figure 1b) compared to Col‐0 (wild type; WT) (Figure 1c,d). They are hypo‐morphic alleles of brm, predicted to be truncated in the bromodomain and C‐terminal short sequences, respectively, by T‐DNA insertion (Figure 1b; Li et al., 2022). Taken together, these results indicate that BRM promotes shoot regeneration from root‐derived callus.

Figure 1.

BRAHMA (BRM) facilitates de novo shoot regeneration from root‐derived callus.

(a) Schematic diagram of de novo shoot regeneration via root‐derived callus formation. DAG, days after germination.

(b) Structure and sites of T‐DNA insertion in the BRM genes. T‐DNA insertion sites of each mutant are indicated under the structure. Boxes represent exons and lines between the boxes indicate introns. Each blue box represents a protein domain. The figure was drawn with reference to Farrona et al. (2007), Kwon et al. (2006), Yu et al. (2021), and Zhang et al. (2017).

(c) Phenotypes of wild type (WT), brm‐3 and brm‐20 callus. C7, CIM incubation for 7 days; C7S14, SIM incubation for 14 days after CIM incubation for 7 days; Bars, 5 mm; Arrows, regenerated shoot.

(d) Shoot regeneration rate for WT, brm‐3, and brm‐20. The percentage of shoot regenerating explants to the total number of explants is shown as a percentage (22 ≤ n ≤ 50). Data are means + SD (three biological replicates). Asterisks indicate significant differences based on Dunnett's test (*P < 0.05).

The effect of BRM on the shoot regeneration process is independent of callus growth

Compared to WT, the shoot regeneration rate was reduced in brm mutants (Figure 1c,d). To determine whether this phenotype was dependent on callus growth, morphological observations were conducted after 7 days of callus formation. Propidium iodide (PI) staining was performed to help distinguish normal cells and callus cells (Figure S1a,b). For measurement and quantification, the area of callus cells was quantified by imaging the region around the center of the explant (1 cm from the cutting site and root tip) (Figure S1a). We found no significant difference in callus cell area between WT and brm‐3 (Figure S1b,c). Thus, BRM does not affect callus growth, suggesting that its role in the shoot regeneration process is independent of callus formation.

BRM contributes mainly to the transition of gene expression from callus formation to shoot regeneration

To examine changes in gene expression during the shoot regeneration process in brm‐3 mutant, RNA‐seq was performed at three time points: just before callus formation (C0), callus formation (C7), and shoot regeneration (C7S1) where there was no phenotypic difference between WT and brm‐3 (Figure S2). Analysis of differentially expressed genes (DEGs) (q <0.05, fold change (FC) >2 or <0.5) at each time point (Table S1) revealed that more genes were differentially expressed at C7 and C7S1 than at C0 (Figure 2a). First, we focused primarily on C7S1, as there was no significant difference in area between WT and brm‐3 at C7 (Figure S1b,c).

Figure 2.

BRAHMA (BRM) plays an important role in gene expression at C7S1.

(a) Differentially expressed genes between wild type (WT) and brm‐3 at C0, C7, and C7S1. Red and blue dots indicate significantly upregulated genes (q <0.05, fold change (FC) >2) and significantly downregulated genes (q <0.05, FC <0.5), respectively.

(b) Gene ontology (GO) terms of genes upregulated (q <0.05, FC >2) in brm‐3 compared to WT at C7S1.

(c) GO terms of genes downregulated (q <0.05, FC <0.5) in brm‐3 compared to WT at C7S1. See also Figures S1 and S2.

Gene ontology (GO) analysis at C7S1 revealed that upregulated genes in brm‐3 compared to WT (q <0.05, FC >2) were associated with defense responses and systemic acquisition resistance (SAR), a type of immune response (Figure 2b). Meanwhile, downregulated genes (q <0.05, FC <0.5) were associated with gibberellin (GA) responses, cell division, and meristem growth (Figure 2c). Recently, it has been reported that the accumulation of salicylic acid (SA) suppresses shoot regeneration in callus (Koo et al., 2024) and that GAs positively regulate almost all aspects of plant growth and development (Robil et al., 2024). Taken together, these findings suggest that BRM may contribute to shoot regeneration through the regulation of phytohormone pathways.

Next, to investigate the role of BRM in trans‐differentiation, which is essential for the shoot regeneration process, we analyzed gene expression before and after shoot induction (C7 and C7S1) in WT to identify genes upregulated (q <0.05, FC >2) or downregulated (q <0.05, FC <0.5) in response to shoot induction (Table S2). We then examined the expression changes of these genes in brm‐3. Among the genes upregulated in response to shoot induction in WT, we found both downregulated and further upregulated genes in brm‐3 on SIM (Figure 3a). GO analysis revealed that the downregulated genes in brm‐3 included those involved in root patterning, while the further upregulated genes were enriched for functions related to defense responses and SA (Figure S3a–d). In contrast, the expression levels of the genes downregulated in response to shoot induction in WT were not substantially different between WT and brm‐3 (Figure 3b). These results indicate that BRM contributes to the proper regulation of genes that are normally upregulated during shoot induction in WT.

Figure 3.

BRAHMA (BRM) contributes mainly to gene expression after C7 during the shoot regeneration process.

(a) Changes in the expression of upregulated genes (q <0.05, FC >2) from C7 to C7S1 in wild type (WT), and their expression patterns in brm‐3. aveRPM, average read counts per million mapped reads.

(b) Changes in the expression of downregulated genes (q <0.05, FC <0.5) from C7 to C7S1 in WT, and their expression patterns in brm‐3. aveRPM, average read counts per million mapped reads.

(c) Gene expression changes in key genes for the shoot regeneration process based on Transcriptional regulation in table 1 of Ikeuchi et al. (2019). Genes marked root, callus, or shoot were selected and described. Symbols ^r, ^c, and ^s indicate genes marked root, callus, and shoot, respectively. RAM‐related genes (Ishihara et al., 2019) are shown in bold. aveRPM, average read counts per million mapped reads. See also Figures S3 and S4.

Finally, to examine the expression patterns of genes previously reported to be expressed during the shoot regeneration process, we analyzed our RNA‐Seq data with reference to data from Ikeuchi et al. (2019). During the shoot regeneration process, removal of RAM tissue memory from callus and increased expression of SAM‐related genes are important (Ishihara et al., 2019). As a result, we did not observe a marked difference in the expression of genes important for callus formation between WT and brm‐3 during callus formation from C0 to C7 (Figure 3c). However, during shoot induction from C7 to C7S1, we found that RAM‐related genes (Ishihara et al., 2019), which were highly expressed in callus but suppressed on SIM in WT, remained relatively highly expressed in brm‐3 (Figure 3c). Therefore, we examined in detail the genes whose expression increased from C0 to C7 (q <0.05, FC >2) and decreased from C7 to C7S1 (q <0.05, FC <0.5). Two such genes, LATERAL ORGAN BOUNDARIES‐DOMAIN 16 (LBD16) and WUSCHEL RELATED HOMEOBOX 5 (WOX5), were significantly upregulated at C7S1 in brm‐3 than in WT (Figure S4), and WOX5, was also significantly upregulated in brm‐3 at C7 (Figure S4). In summary, these results suggest that BRM primarily contributes to the regulation of gene expression after C7 during the shoot regeneration process.

BRM contributes to H3K27me3 demethylation from shoot regeneration‐related genes during callus formation

Previous studies have shown that the levels of H3K4me3 and H3K27me3 are altered in brm mutants (Li et al., 2015; Yang et al., 2015; Zhao et al., 2015), suggesting that the suppressed shoot regeneration phenotype observed in brm‐3 may be due to changes in these histone modifications. Therefore, to investigate the effects of BRM, which dynamically alters chromatin structure, and histone modifications involved in epigenetic regulation during the shoot regeneration process, we analyzed H3K4 methylation (H3K4me1, H3K4me2, and H3K4me3) and H3K27me3 at three time points (C0, C7, and C7S1) in WT and brm‐3 using chromatin immunoprecipitation followed by sequencing (ChIP‐seq).

First, we compared methylation levels between WT and brm‐3 and found that H3K27me3 levels were increased in brm‐3 at each time point, while H3K4me1/2/3 levels showed no distinct differences (Figure 4a). We then identified genes with increased H3K27me3 levels at each time point (P < 0.05, FC >1) (Figure 4b; Table S3). GO analysis of these gene sets revealed enrichment of terms related to shoot regeneration, such as ‘plant organ formation’, ‘shoot system development’, and ‘cell differentiation’ at each time point (Figure S5a–c).

Figure 4.

BRAHMA (BRM) is involved in the removal of H3K27me3 from genes related to shoot regeneration during callus formation.

(a) Input, H3K4 methylation, and H3K27me3 levels in brm‐3 compared to wild type (WT) at C0, C7, and C7S1. r, Pearson correlation coefficient; RPM, read counts per million mapped reads.

(b) Genes with increased H3K27me3 in brm‐3 compared to WT at C0, C7, and C7S1 (P <0.05, FC >1).

(c) Gene Ontology terms of genes with increased H3K27me3 in brm‐3 compared to WT specifically at C7 (P <0.05, FC >1; 802 genes).

(d) Expression patterns of the BRM translational reporter (pBRM::BRM‐GFP, shown in green) at C0, C3, and C7. Cell outlines were visualized using propidium iodide staining (shown in magenta). Bar: 100 μm. See also Figures S5–S7.

Next, we focused on genes with increased H3K27me3 levels specific to each time point. As a result of GO analysis, only one GO term was enriched for genes specific to C0 (Figure S6a), and no GO terms were detected for genes specific to C7S1. In contrast, C7‐specific genes with increased H3K27me3 levels were significantly enriched for GO terms associated with shoot regeneration, such as ‘shoot system morphogenesis’, ‘cell differentiation’ and ‘shoot system development’ (Figure 4c). H3K27me3 levels of these genes in WT were also found to be significantly decreased throughout callus formation (C7) and shoot regeneration (C7S1) compared to root tips (C0) while the H3K27me3 levels remained higher in brm‐3 (Figure S6b). Thus, our results suggest that BRM is involved in the demethylation of H3K27me3 from genes related to shoot regeneration during callus formation. To determine whether BRM is functional during callus formation, we observed its localization using a transgenic line expressing pBRM::BRM‐GFP. Fluorescence intensity increased during callus formation from C0 to C7 and spread throughout the callus cells (Figure 4d), suggesting that BRM plays a role in callus formation.

We also examined the relationship between the repression of RAM‐related genes, which were identified as BRM‐regulated genes by RNA‐Seq, and H3K27me3. However, no significant changes in the H3K27me3 levels were observed for RAM‐related genes throughout the shoot regeneration process (Figure S7), indicating that the repression of these genes is not substantially affected by H3K27me3.

BRM regulates 24 genes as targets for epigenetic priming during the shoot regeneration process

BRM was implicated in the demethylation of H3K27me3 from the genes involved in shoot regeneration during callus formation (Figure 4b,c). Based on this result, we considered the potential involvement of BRM in epigenetic priming. Epigenetic priming is a potential molecular mechanism that creates a primed state without changes in gene expression through epigenetic changes such as histone modifications (Ishihara et al., 2019). Recently, epigenetic priming by the histone demethylase LDL3 has been reported to be involved in plant regeneration. LDL3 removes H3K4me2 from shoot regeneration‐related genes during callus formation, establishing a primed state that facilitates their gene activation during shoot induction (Ishihara et al., 2019). Therefore, this study suggests that BRM and H3K27me3 may similarly contribute to epigenetic priming in plant regeneration.

We first performed an integrated analysis of RNA‐seq and ChIP‐seq data to find primed genes by BRM. In this study, we defined them as genes present in the intersection of the following three criteria: genes upregulated (q <0.05, FC >2) in response to shoot induction (from C7 to C7S1) in WT, genes downregulated (q <0.05, FC <0.5) in brm‐3 compared to WT at C7S1, and genes with increased H3K27me3 levels in brm‐3 compared to WT at C7 (P < 0.05, FC >1). This analysis identified 24 candidate primed genes of BRM‐mediated epigenetic priming (Figure 5a,b; Table S4). In WT, these primed genes were upregulated during shoot induction (Figure 5b; Figure S8a), while no significant changes in gene expression were observed during callus formation (Figure S8a). They also exhibited significantly higher levels of H3K27me3 in brm‐3 compared to WT (Figure S8b), resembling the pattern observed in genes with increased H3K27me3 specifically at C7 (Figure S6b). In WT, H3K27me3 levels of these primed genes decreased markedly during callus formation (from C0 to C7) but showed relatively little change after shoot induction (from C7 to C7S1) (Figure S8b). These results suggest that BRM‐mediated primed genes undergo no major changes in gene expression while H3K27me3 was removed during callus formation from C0 to C7, and subsequently upregulated while maintaining constant levels of H3K27me3 during shoot induction from C7 to C7S1. Notably, this behavior contrasts with the genome‐wide trend, in which gene expression typically decreases as H3K27me3 levels increase at both C7 and C7S1 (Figure S8c,d).

Figure 5.

BRAHMA (BRM) is involved in epigenetic priming via H3K27me3 removal in the shoot regeneration process.

(a) Identification of BRM target genes associated with epigenetic priming (24 genes).

(b) Gene expression of the genes explained in (a) in wild type (WT) and brm‐3 before and after shoot induction (from C7 to C7S1). aveRPM, average read counts per million mapped reads.

(c) Gene expression of NGA3 and DOT1 in WT and brm‐3 before and after shoot induction (from C7 to C7S1). Data are means ± SD (three biological replicates). aveRPM, average read counts per million mapped reads. Asterisks indicate significant differences based on RNA‐seq (*q <0.05).

(d) Occupancy of H3K27me3 at NGA3 and DOT1 in WT and brm‐3 before and after shoot induction (from C7 to C7S1). Data range: 0–50.

(e) Phenotypes of WT, nga3‐1 #3, and nga3‐1 #6 callus. C7, CIM incubation for 7 days; C7S14, SIM incubation for 14 days after CIM incubation for 7 days; Bars, 5 mm; Arrows, regenerated shoot.

(f) Shoot regeneration rate for WT, nga3‐1 #3, and nga3‐1 #6. The percentage of shoot regenerating explants out of the total number of explants is shown as a percentage (20 ≤ n ≤ 22) on the left. Data are means + SD (three biological replicates). The number of shoots per explant is shown on the right. Data are means + SD (60 ≤ n ≤ 65). Asterisks indicate significant differences based on Dunnett's test (*P < 0.05; **P < 0.01; n.s., no significant [P > 0.05]).

(g) Phenotypes of WT, dot1‐2 #8, and dot1‐2 #10 callus. C7, CIM incubation for 7 days; C7S14, SIM incubation for 14 days after CIM incubation for 7 days; Bars, 5 mm; Arrows, regenerated shoot.

(h) Shoot regeneration rate for WT, dot1‐2 #8, and dot1‐2 #10. The percentage of shoot regenerating explants out of the total number of explants is shown as a percentage (10 ≤ n ≤ 23) on the left. Data are means + SD (three biological replicates). The number of shoots per explant is shown on the right. Data are means + SD (35 ≤ n ≤ 64). Asterisks indicate significant differences based on Dunnett's test (*P < 0.05; ***P < 0.001). See also Figures S8–S12.

We next analyzed genes with H3K27me3 dynamics similar to those of BRM‐mediated primed genes, defined as those showing significant decreases in H3K27me3 in WT during callus formation (from C0 to C7, P < 0.05 and FC <1) but no significant changes during shoot induction (from C7 to C7S1, P ≥ 0.05) (Figure S9a). Among the 3806 identified genes, H3K27me3 accumulated to higher levels in brm‐3 than in WT (Figure S9b), indicating that BRM is also involved in removing H3K27me3 from these genes beyond BRM‐mediated primed genes. Expression analysis revealed that these genes tended to be induced from C7 to C7S1 in WT, although the fold changes were generally lower than those of BRM‐mediated primed genes (Figures S8a and S9c). This suggests that BRM‐dependent H3K27me3 removal also contributes to the induction of these genes. Importantly, both these genes and BRM‐mediated primed genes showed no expression changes from C0 to C7 in WT despite reductions in H3K27me3 (Figure S9b,c). Taken together, these results suggest that epigenetic priming, in which gene expression is induced after the removal of repressive histone marks, may regulate many genes during callus formation and shoot induction.

To explore potential mechanisms underlying the activation of BRM‐mediated primed genes after shoot induction, we performed a cis‐element enrichment analysis of the promoter regions of these genes. The results showed that the promoter regions of BRM‐mediated primed genes contained commonly enriched motifs that overlapped with motifs enriched in the promoter regions of SIM‐inducible genes in WT (C7S1 versus C7), but not with the top 30 motifs enriched in the promoters of CIM‐inducible genes in WT (C7 versus C0) (Figure S10a–c). These results suggest that BRM‐mediated primed genes are specifically induced only after shoot induction (Figure S8a), likely due to the requirement for SIM‐specific transcription factors. The absence of cis‐elements related to CIM‐inducible genes in the promoter regions of BRM‐mediated primed genes is also consistent with the lack of expression changes during callus formation (Figure S8a).

BRM‐mediated primed genes including NGA3 and DOT1 may collectively contribute to gene expression after shoot induction

Among BRM‐mediated primed genes, we focused on NGATHA3 (NGA3) and DEFECTIVELY ORGANIZED TRIBUTARIES 1 (DOT1), which have been previously reported to be abnormal in leaf morphology and differentiation in mutants. NGA3 encodes a B3‐type transcription factor, and loss of the gene delays the differentiation of leaf cells (Lee et al., 2015). In addition, a reduced shoot regeneration rate has been recently reported in the nga2 mutant, which belongs to the same gene family (Wu et al., 2022). DOT1 encodes a glycine‐rich protein, and mutants show abnormal phenotypes such as fused and cup‐shaped leaves and reduced trichomes (Petricka et al., 2008). Both genes were significantly upregulated upon shoot induction in WT, but expression during shoot regeneration was suppressed in brm‐3 compared to WT (Figure 5c). H3K27me3 levels at these loci were accumulated in brm‐3 during both callus formation and shoot regeneration (Figure 5d). These expression and H3K27me3 patterns were consistent with those observed for BRM‐mediated primed genes overall (Figure S8a,b) and were further validated by quantitative reverse transcriptase‐polymerase chain reaction (qRT‐PCR) and ChIP‐qPCR (Figure S11a,b).

To assess the functional relevance of these genes, we performed shoot regeneration assays using two different mutant lines of nga3 (nga3‐1; Sato et al., 2018) and dot1 (dot1‐2; Petricka et al., 2008). The results showed no significant difference between WT and nga3‐1 in the percentage of explants with regenerated shoots out of all explants, but a significantly decreased regeneration rate in nga3‐1 in the number of regenerated shoots per explant (Figure 5e,f). In dot1‐2, the shoot regeneration rate was significantly reduced in both parameters (Figure 5g,h). Thus, these results indicate that NGA3 and DOT1, like BRM, promote shoot regeneration.

Next, to explore the potential contribution of the transcription factors primed by BRM through H3K27me3 (Figure 5b), we performed a cis‐element enrichment analysis using the promoter regions of upregulated and downregulated genes in brm‐3 at C7S1 to transcriptomic changes in brm‐3. The results revealed significant enrichment of motifs associated with NGA‐related (CCCTG/TACCTG/CCGTG/CACTTG) (Li et al., 2020; Sato et al., 2018), DOF‐related (AAAG) (Zou & Sun, 2023), GATA‐related (GATA) (Schwechheimer et al., 2022), MYB‐related ([C/T]NGTT[G/A]) (Millard et al., 2019), NAC‐related (CACG) (Nakashima et al., 2012), and G box core (ACGT) (Ezer et al., 2017) transcription factors among the promoter regions of upregulated and downregulated genes in brm‐3 (Figure S12a,b). This result suggests that each transcription factor primed by BRM, including NGA3, may collectively contribute to the broad transcriptomic changes on SIM. This further implies that BRM may exert indirect effects on genes upregulated during shoot induction when BRM‐mediated primed genes, including NGA3, activate downstream genes (Figure 3a; Figure S3a).

BRM, REF6, and ELF6 may cooperatively regulate BRM‐mediated epigenetic priming

BRM is known to directly interact with the plant‐specific H3K27me3 demethylase RELATIVE OF EARLY FLOWERING 6 (REF6) (Li et al., 2016), which has a close homolog, EARLY FLOWERING 6 (ELF6) (Yu et al., 2008). Based on this, we also investigated the involvement of BRM, REF6, and ELF6 in the shoot regeneration process. ref6‐1, ref6‐3, and elf6‐3 are T‐DNA inserted mutants (Lu et al., 2011; Yan et al., 2018), while ref6 ^c and ref6 ^c elf6‐3 were created by the RNA‐guided Cas9 method and conventional crossing (Yan et al., 2018). In shoot regeneration assays using these five mutants, all showed decreased shoot regeneration rates compared to WT (Figure 6a,b), indicating that REF6 and ELF6 promote shoot regeneration from root‐derived callus, similar to BRM.

Figure 6.

REF6 and ELF6 may facilitate de novo shoot regeneration with BRAHMA (BRM) cooperatively.

(a) Phenotypes of wild type (WT), ref6‐1, ref6‐3, ref6 ^c , ref6 ^c elf6‐3 and elf6‐3 callus. C7, CIM incubation for 7 days; C7S14, SIM incubation for 14 days after CIM incubation for 7 days; Bars, 5 mm; Arrows, regenerated shoot.

(b) Shoot regeneration rate for WT, ref6‐1, ref6‐3, ref6 ^c , ref6 ^c elf6‐3, and elf6‐3. The percentage of shoot regenerating explants to the total number of explants is shown as a percentage (10 ≤ n ≤ 42). Data are means + SD (three biological replicates). Asterisks indicate significant differences from WT based on Dunnett's test (*P < 0.05; ***P < 0.001).

(c) Venn diagram showing genes upregulated during shoot induction in WT and genes downregulated in ref6 ^c elf6‐3 at C7S1, and BRM‐mediated primed genes. Significant overlap was shown by Fisher's exact test (P = 2.888e‐11). See also Figures S13–S17.

To further investigate the roles of REF6 and ELF6, we performed RNA‐seq on the ref6 ^c elf6‐3 due to redundancy between REF6 and ELF6 (Yan et al., 2018). As with BRM, we focused on C7S1 and performed GO analysis on DEGs compared to WT (q <0.05, FC >2 or <0.5) (Figure S13a,b; Table S5). At C7S1, upregulated genes in ref6 ^c elf6‐3 compared to WT (q <0.05, FC >2) were enriched in categories related to SAR and defense responses (Figure S13a), similar to brm‐3 (Figure 2b). Conversely, downregulated genes (q <0.05, FC <0.5) were enriched in terms related to tropism (Figure S13b; Table S5), which involves directed growth and deformation in response to stimuli (Moulton et al., 2020), implying impaired directional cell regulation during shoot regeneration in ref6 ^c elf6‐3.

We next compared RNA‐Seq data from brm‐3 and ref6 ^c elf6‐3 to determine whether BRM, REF6, and ELF6 function cooperatively during shoot regeneration. A significant overlap was observed in both upregulated and downregulated genes at C7S1 (Figure S14a,b). GO analysis revealed that overlapping upregulated genes were associated with ‘defense response’ and ‘immune response’, whereas overlapping downregulated genes were associated with ‘regulation of asymmetric cell division’ and ‘regulation of meristem growth’ (Figure 2b,c; Figure S14c,d). These results suggest that BRM, REF6, and ELF6 cooperatively regulate gene expression during shoot regeneration. This is further supported by the observation that the expression of genes downregulated in both brm‐3 and ref6 ^c elf6‐3 was higher than that of genes downregulated only in brm‐3 (Figure S15), indicating that REF6 and ELF6 contribute to the enhanced expression of BRM downstream genes. To further confirm whether BRM downstream genes can also be targeted by REF6 or ELF6, we examined the distribution of REF6‐binding motifs (CTCTGYTY (Y = T or C)) (Cui et al., 2016; Li et al., 2016) within the genomic regions of downregulated genes at C7S1. The proportion of genes with REF6‐binding sites was significantly higher in the genes commonly downregulated in both brm‐3 and ref6 ^c elf6‐3 than in those downregulated only in brm‐3 (Figure S16a). Similarly, the number of REF6‐binding sites within these genomic regions was also significantly higher in the overlapped genes relative to all A. thaliana genes and downregulated genes only in brm‐3 (Figure S16b). These results further support the cooperative role of BRM, REF6, and ELF6 in regulating downstream genes during shoot regeneration.

Finally, we investigated whether REF6 and ELF6 are involved in regulating BRM‐mediated primed genes. A significant overlap was found between BRM‐mediated primed genes and genes downregulated in ref6 ^c elf6‐3 compared to WT, which show gene activation during shoot induction in WT (Figure 6c; Table S6). This suggests that REF6 and ELF6 may also regulate BRM‐mediated primed genes cooperatively. As REF6 and ELF6 are H3K27me3 demethylases, and BRM‐mediated primed genes are demethylated during callus formation and exhibit dynamic changes in H3K27me3 levels throughout callus formation in the shoot regeneration process (Figure S8b), we suggest that REF6 and ELF6 may participate in the demethylation of H3K27me3 at these loci. This is further supported by ChIP‐qPCR of BRM‐mediated primed genes cooperatively regulated by REF6 and ELF6 (Figure 6c; Table S6), which showed accumulation of H3K27me3 at NGA3 and FASCICLIN‐LIKE ARABINOOGALACTAN 9 (FLA9) in ref6 ^c elf6‐3 compared to WT (Figure S17). FLA9 is involved in normal embryo development (Cagnola et al., 2018). The accumulation of H3K27me3 at these two genes suggests that REF6 and ELF6 contribute to BRM‐mediated epigenetic priming through the removal of H3K27me3.

In conclusion, BRM contributes mainly to gene expression after callus formation in the shoot regeneration process from root‐derived callus and is involved in epigenetic priming mediated by H3K27me3 removal during callus formation. This demethylation process is suggested to be mediated by the plant‐specific H3K27me3 demethylases REF6 and ELF6, which promote shoot regeneration by targeting BRM‐mediated primed genes.

DISCUSSION

In this study, we found that the ATPase BRM, a component of the BAS chromatin remodeling complex, plays an essential role in the shoot regeneration process (Figure 1c,d). Specifically, we identified BRM as a player in epigenetic priming, showing that it facilitates H3K27me3 demethylation from shoot regeneration‐related genes named BRM‐mediated primed genes during callus formation. Our findings further suggested that the plant‐specific H3K27me3 demethylases REF6 and ELF6 promote shoot regeneration and may cooperatively regulate BRM‐mediated primed genes (Figure 6a–c). In addition, we revealed that NGA3 and DOT1 positively contribute to shoot regeneration downstream of BRM (Figure 5e–h). In particular, we propose that chromatin remodeling should be recognized as a critical layer of regulation in addition to previously reported epigenetic mechanisms through the analysis of BRM, which plays a central role in chromatin remodeling itself and transcription in the chromatin remodeling complexes.

We previously reported that epigenetic priming through the elimination of H3K4me2 by the histone demethylase LDL3 contributes to plant regeneration (Ishihara et al., 2019). In this study, we demonstrate that epigenetic priming involving the removal of H3K27me3 also plays a key role in shoot regeneration. H3K27me3 alteration in the shoot regeneration process via callus formation (Figure 4a,b) suggests that this change is mediated by the callus‐specific function of BRM. Supporting this suggestion, the expression of BRM was observed in callus cells during callus formation (Figure 4d). Our GO analysis revealed that genes with increased H3K27me3 specifically during callus formation were enriched in terms related to shoot regeneration, such as ‘shoot system morphogenesis’ and ‘cell differentiation’ (Figure 4b,c). Callus observation showed that the callus area of WT and brm‐3 had no significant difference (Figure S1b,c), suggesting that although callus can grow properly, the H3K27me3 levels of shoot regeneration‐related genes are increased in those cells during callus formation and their gene expression is suppressed after shoot induction in brm‐3. Thus, histone modification of shoot regeneration‐related genes is altered during callus formation as part of epigenetic priming. Transcriptome analysis indicated that BRM contributes mainly to the transition of gene expression from callus formation to shoot regeneration (Figure 3a–c). These findings are consistent with previous work (Ishihara et al., 2019), considering that by epigenetic priming, changes in histone modifications that have a major impact on shoot regeneration occur mainly during callus formation, and changes in gene expression occur mainly during shoot regeneration. This is further supported by results showing that H3K27me3 levels in BRM‐mediated primed genes decreased from C0 to C7 during callus formation without significant changes in their expression. However, these genes were upregulated from C7 to C7S1 during shoot induction, despite sustained H3K27me3 levels (Figure S8a,b). This relationship between the repressive histone mark H3K27me3 and gene expression deviates from a simple inverse correlation (Figure S8c,d), indicating that epigenetic priming is a distinct regulatory mechanism, as previously described (Bryan et al., 2025; Handa & Matsunaga, 2024; Vicente‐Dueñas et al., 2018; Walter et al., 2008). Beyond BRM‐mediated primed genes, we also identified additional genes that exhibited similar H3K27me3 patterns but showed no changes in expression during callus formation from C0 to C7, despite a marked reduction in H3K27me3 in WT (Figure S9b,c). Interestingly, from C7 to C7S1, these genes maintained stable H3K27me3 levels, yet their expression was upregulated (Figure S9b,c). These results suggest that epigenetic priming, in which epigenetic changes create a primed state for gene expression, may be a general mechanism by which BRM regulates many genes in the shoot regeneration process. Our cis‐element enrichment analyses also revealed that the activation of BRM‐mediated primed genes during shoot induction should likely require SIM‐specific transcription factors (Figure S10a–c), implying that cytokinin signaling pathways may be involved in the activation process. Furthermore, the unchanged expression of BRM‐mediated primed genes during callus formation could be explained by the absence of cis‐elements that are enriched with CIM‐inducible genes in the promoter regions of these genes at this stage. Both NGA3 and DOT1, identified as target genes of BRM‐mediated epigenetic priming, are known to be involved in leaf differentiation and morphogenesis (Lee et al., 2015; Petricka et al., 2008), but their involvement in shoot regeneration has not been reported. The expression patterns during callus formation and shoot regeneration, as well as the changes in H3K27me3, confirm epigenetic priming of altered expression levels without changes in H3K27me3 (Figure 5c,d). In particular, shoot regeneration assays using nga3 mutant showed no significant difference from WT in the percentage of explants with regenerated shoots, but a significant difference in the number of regenerated shoots (Figure 5e,f). dot1 mutant also showed a decreased shoot regeneration rate (Figure 5g,h). These findings suggest that NGA3 and DOT1 play a functionally relevant role in shoot regeneration downstream of BRM. Given that NGA3 and DOT1 regulate leaf cell differentiation and morphogenesis (Lee et al., 2015; Petricka et al., 2008), it is likely that they also contribute to shoot meristem formation and tissue patterning during the regeneration process. Furthermore, in addition to NGA3, other transcription factors among BRM‐mediated primed genes may also contribute moderately to transcriptional regulation during shoot regeneration, as suggested by the cis‐element enrichment analyses (Figure S12a,b). These findings suggest that BRM promotes shoot regeneration by regulating various transcription factors through epigenetic priming. Finally, we suggest that REF6 and ELF6 cooperatively regulate BRM‐mediated primed genes (Figure 6a–c), implying that these two enzymes may be involved in the demethylation of H3K27me3 as part of BRM‐mediated epigenetic priming. Previous studies have reported that chromatin remodeling factors cooperate with histone modifying enzymes (Wu & Roberts, 2013). It is also known that BRM and REF6 coordinately regulate gene expression (Li et al., 2016), and our results extend this cooperation to the shoot regeneration process (Figure 6c; Figures S14a–d and S15). Consistently, the analysis of REF6‐binding sites suggests a functional connection among BRM, REF6, and ELF6 (Figure S16a,b), raising the possibility that REF6 and ELF6 act together with BRM. Furthermore, H3K27me3 levels at NGA3 and FLA9, which are BRM‐mediated primed genes, were increased in ref6 ^c elf6‐3 compared to WT (Figure S17), supporting the role of REF6 and ELF6 in H3K27me3 demethylation at these loci. However, the precise contribution of REF6 and ELF6 to H3K27me3 removal during the shoot regeneration process remains unclear. Additionally, it is important to note that the SWI/SNF chromatin remodeling complexes cannot bind to DNA directly but are instead recruited to promoter regions via interactions with DNA binding proteins (Nishioka et al., 2020). In a previous study, REF6 was shown to be required for BRM to bind to chromatin (Li et al., 2016), while other reports have shown that the DNA‐binding domains of BRM can bind to DNA (Farrona et al., 2007), and that another component of the SWI/SNF complexes, B‐cell lymphoma/leukemia protein 7 A/B (BCL7A/B) homologous subunits, require BRM to bind target genes in plants (Lei et al., 2024). Therefore, further studies to identify the common demethylation target genes of BRM, REF6, and ELF6 would provide new insights into the mechanisms underlying BRM‐mediated epigenetic priming. Moreover, it will be crucial to investigate how BRM, REF6, and ELF6 interact with the genome during the shoot regeneration process.

Based on callus observations (Figure S1b,c), we focused mainly on the transcriptome during shoot regeneration and before and after shoot induction and found that upregulated and downregulated genes during shoot regeneration were enriched in GO terms related to SA (Figure 2b) and GA (Figure 2c), respectively. SA is a well‐established signaling molecule for SAR and plays crucial roles in basal defense and the amplification of local immune responses (Peng et al., 2021). Notably, it has recently been reported that SA accumulation suppresses shoot regeneration (Koo et al., 2024). GAs also regulate almost all aspects of plant growth and development (Robil et al., 2024). Organ regeneration often involves the activation of mitosis, and cell proliferation is necessary for successful organ regeneration (Ikeuchi et al., 2019). Based on these findings, we hypothesized that the alteration of these genes influences the suppressed shoot regeneration phenotypes in brm‐3. The expression changes before and after shoot induction (Figure 3a,b) suggest that BRM contributes mainly to the upregulated genes in response to shoot induction, supporting its role in epigenetic priming. The cis‐element enrichment analysis using the promoter regions of downregulated genes in brm‐3 at C7S1 is also consistent with these results, suggesting that several transcription factors in BRM‐mediated primed genes may activate target genes on SIM (Figure S12b). Among these, the downregulated and more highly upregulated genes in brm‐3 included those involved in root patterning, SAR, and SA (Figure S3a–d). As for the genes involved in root patterning, which were downregulated at C7S1, it has been previously reported that RAM and callus show similar traits (Sugimoto et al., 2010). In a recent review (Lee et al., 2024), a previous study showed that callus is not a disorganized cell but spatially organized (Atta et al., 2009; Sugimoto et al., 2010, 2011; Varapparambath et al., 2022; Zhai & Xu, 2021). This suggests that disorganization of callus cell organization by the suppression of root patterning‐related genes has a negative effect on shoot regeneration in brm‐3. As for the genes related to SAR and SA, whose expression was upregulated at C7S1, it is suggested that the response to SA (Peng et al., 2021), a well‐known signaling molecule for SAR, is activated and suppresses shoot regeneration (Koo et al., 2024). Consistent with the evidence that BRM is highly involved in gene expression before and after shoot induction in the expression of known regeneration‐related genes, the expression of some genes including RAM‐related genes (Ishihara et al., 2019) remained elevated in brm‐3 at C7S1 compared to WT (Figure 3c; Figure S4). LBD16, which was significantly upregulated at C7S1, and WOX5, which was significantly upregulated in both C7 and C7S1, are key genes in the acquisition of callus pluripotency (Shin et al., 2020). These findings suggest that callus is produced normally in brm‐3 (Figure S1b,c), but that trans‐differentiation to regenerate de novo shoots cannot be successful. Finally, since H3K27me3 played a minor role in the regulation of RAM‐related genes (Figure S7), it is considered that these genes may be influenced by other transcriptional regulatory mechanisms.

In summary, we show that the ATPase BRM, a component of the BAS chromatin remodeling complex, promotes shoot regeneration via epigenetic priming through the removal of H3K27me3 during the shoot regeneration process. We also show that REF6 and ELF6 are likely to cooperatively regulate BRM‐mediated primed genes, suggesting that they are involved in the demethylation of H3K27me3 in the shoot regeneration process. Furthermore, BRM does not affect callus growth, but epigenetic priming already occurs in callus cells, resulting in the activation of shoot regeneration‐related gene expression upon shoot induction. In addition, our data suggest that the expression of genes important for callus characteristics is regulated by BRM, thereby contributing to trans‐differentiation. However, since histone modifications were examined as indirect indicators of chromatin remodeling, dynamic changes in chromatin itself need to be investigated in detail in the future. It also remains to be clarified whether the reduced shoot regeneration in brm‐3 results from altered H3K27me3 at C7 or from the loss of BRM function at C7S1. Clustered regularly interspaced short palindromic repeats/nuclease‐dead Cas (CRISPR/dCas)‐mediated epigenome editing (Cheng et al., 2024) expressed at specific developmental stages will be helpful to distinguish between these possibilities.

MATERIALS AND METHODS

Plant materials and growth conditions

All A. thaliana plants used in this study were of the Columbia background. Col‐0 was used as WT. The T‐DNA insertion mutant lines are brm‐3 (SALK_088462; Farrona et al., 2007), brm‐20 (SALK_002500; Zhang et al., 2017), nga3‐1 (SAIL_232_E10; Sato et al., 2018), dot1‐2 (SALK_001950; Petricka et al., 2008), ref6‐1 (SALK_001018), ref6‐3 (SAIL_747_A07), and elf6‐3 (SALK_074694). All T‐DNA insertion mutant lines of ref6 and elf6 have been described previously (Lu et al., 2011). ref6 ^c and ref6 ^c elf6‐3 are created by the RNA‐guided Cas9 method and conventional crossing (Yan et al., 2018). The transgenic line expressing pBRM::BRM‐GFP with BASTA resistance (Smaczniak et al., 2012; Zhong et al., 2008) is brm‐1 (SALK_030046; Hurtado et al., 2006) background. These seeds were sterilized with 50% (v/v) bleach for up to 5 min. Sterilized seeds were rinsed with autoclaved water three times before sowing onto media containing MGRL solution (Fujiwara et al., 1992), 1% (w/v) sucrose (Wako, Osaka, Japan), and 0.8% (w/v) gellan gum (Wako) for other than the transgenic line or half‐strength Murashige and Skoog media (Murashige & Skoog, 1962) containing 1% (w/v) sucrose (Wako), 0.4% (w/v) gellan gum (Wako), and 10 μgml⁻¹ BASTA for the transgenic line, or soil. For MGRL media and half‐strength Murashige and Skoog media, after 2 days of stratification at 4°C under darkness, seeds were stimulated for germination at 22°C. For soil, sterilized seeds were sown on rockwool (Grodan, Roermond, the Netherlands) in pots filled with vermiculite (Asahi industries, Okayama, Japan). Plants were grown under a long‐day (16 h light [100 μmolm⁻² sec⁻¹]/8 h dark) photoperiod. Plants homozygous for the mutation were selected for genotyping. The fluorescence of transgenic plants selected by BASTA was observed (see later). Primers used for genotyping are shown in Table S7.

Shoot regeneration assays

Root explants (0–2 cm from the root tip) were excised from seedlings at 7 DAG and cultured on CIM containing Gamborg's B‐5 medium (Wako) with 20 g L⁻¹ glucose (Wako), 0.5 g L⁻¹ MES (Nacalai tesque, Kyoto, Japan), 1 × Gamborg's vitamin solution (Duchefa Biochemie, Haarlem, The Netherlands), 500 μgL⁻¹ 2,4‐D (Sigma‐Aldrich, St. Louis, MO, USA), 50 μgL⁻¹ kinetin (Wako or Sigma‐Aldrich), and 0.8% gellan gum (Wako); the pH was adjusted to 5.7 using 1.0 and 0.1 N KOH (Nacalai tesque). Callus induction was conducted at 25°C under continuous light (130 μmolm⁻² sec⁻¹).

After culturing for 7 days on CIM, the explants were transferred to SIM containing Gamborg's B‐5 medium, 10 g L⁻¹ glucose, 0.5 g L⁻¹ MES, 1 × Gamborg's vitamin solution, 2 μgml⁻¹ trans‐zeatin (Wako or Tokyo Chemical Industry, Tokyo, Japan), 0.4 μgml⁻¹ indole‐3‐butyric acid (Nacalai tesque), 1 μgml⁻¹ d‐biotin (Nacalai tesque), and 0.8% gellan gum; the pH was adjusted to 5.7 using 1.0 and 0.1 N KOH. Shoot induction was conducted at 25°C under continuous light (130 μmolm⁻² sec⁻¹).

After incubation on SIM for 14 days, either the number of explants that regenerated shoots on all explants or the number of regenerated shoots per explant was evaluated. All phenotypic assays and microscopic observations using the stereomicroscope (SMZ18, Nikon, Tokyo, Japan) and NIS‐Elements BR software (Nikon) were performed with at least three biological replicates. Images were taken by the microscope camera (DS‐Ri2, Nikon). Fiji software (Schindelin et al., 2012) and Stitching plugin (Preibisch et al., 2009) were used to process the images. Bar plots were drawn by ggplot2 package v3.4.2 (Wickham, 2016) in R.

RNA‐Seq

Root explants derived from WT, brm‐3, and ref6 ^c elf6‐3 seedlings were collected on Day 0 of CIM (C0), Day 7 of CIM (C7), and Day 1 of SIM (C7S1) incubation. Total RNA was isolated from the explants using the Monarch^® Total RNA Miniprep Kit (New England Biolabs, Ipswich, MA, USA). The integrity of the purified RNA was assessed using a spectrophotometer (IMPLEN, Munich, Germany) and the 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). The extracted RNA (500 ng) was used to construct a transcriptome library with NEBNext Poly(A) mRNA Magnetic Isolation Module (New England Biolabs) and NEB Next Ultra II RNA Library Prep Kit for Illumina (New England Biolabs). Libraries were quantified using the Qubit dsDNA High Sensitivity Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) and the 2100 Bioanalyzer (Agilent Technologies). Quantified libraries were pooled, and RNA‐seq was performed under the following conditions: 150 bp paired‐end reads of WT and brm‐3 at C0 were obtained using the Hiseq X Ten sequencer (Illumina, San Diego, CA, USA), whereas 100 bp single‐end reads of the other samples (C0 of ref6 ^c elf6‐3, C7, and C7S1) were obtained using the Novaseq 6000 sequencer (Illumina). Three independent biological replicates were analyzed for each genotype. For the paired‐end sample, only one read of each pair was used for downstream analysis to maintain consistency with the single‐end data.

RNA‐Seq data analysis

Raw fastq files were quality‐filtered by fastp v0.20.0 (Chen, 2023) with ‐q 20 ‐l 50 parameters, then ribosomal RNA was removed by bowtie2 v2.2.4 (Langmead & Salzberg, 2012). Quality‐filtered reads were mapped onto the cDNA sequences of annotated genes and other transcripts of TAIR10 using STAR 2.6.0a (Dobin et al., 2013) with the following parameters: ‐‐outSAMtype BAM SortedByCoordinate ‐‐outFilterMultimapNmax 1 ‐‐outFilterMismatchNmax 2 ‐‐outSAMunmapped Within ‐‐outFilterType BySJout ‐‐outSAMstrandField intronMotif ‐‐quantMode GeneCounts. The resulting sorted BAM files were converted to BED files using BEDTools v2.26.0 (Quinlan & Hall, 2010). The ‘slop’ function of BEDTools was used to extend the 5′ end of RNA‐seq reads toward the 3′ direction to fit the average insertion size (300 bp) of the sequenced libraries. Then, the ‘coverage’ function of BEDTools was used to calculate the number of reads that overlapped with each annotation unit. DEGs were identified using the edgeR package v3.34.1 (Chen et al., 2025) in R version 4.1.1 treating biological triplicates as paired samples. Genes with q <0.05 and FC >2 or <0.5 in each comparison were identified as DEGs other than low‐expressed genes defined as count per million mapped reads (CPM) ≤0.2 (Tables S1, S2, and S5). In addition, volcano plots and bar plots were drawn by the ggplot2 package v3.4.2 (Wickham, 2016) in R. GO analysis was conducted using Shiny GO ver 0.76.3 (Ge et al., 2020). Heatmaps were described using the genefilter package v1.74.1 and the ‘heatmap.2’ function of gplots package v3.1.3 in R from calculated read per million mapped reads (RPM). Venn diagrams were also drawn using BioVenn (Hulsen et al., 2008). See also Methods S2.

ChIP‐Seq

Root explants derived from WT and brm‐3 seedlings were collected on Day 0 of CIM (C0), Day 7 of CIM (C7), and Day 1 of SIM (C7S1) incubation; 0.1 g of explants were frozen in liquid nitrogen, ground into a fine powder, cross‐linked, and nuclear‐extracted in the nuclei isolation buffer (10 mM HEPES [pH 7.6], 1 m sucrose, 5 mM KCl, 5 mM MgCl₂, 5 mM EDTA [pH 8.0]) with 0.6% formaldehyde (Thermo Fisher Scientific or PolySciences, W. Touhy Avenue Niles, IL, USA), 0.6% Triton X‐100, and 14.4 mM 2‐mercaptoethanol (Wako), 1 mM Pefabloc^® SC (Roche, Basel, Switzerland) and cOmplete™ protease inhibitor cocktail (Roche). Samples were sonicated using BIORUPTOR^® UCD‐250HSA (COSMO BIO, Tokyo, Japan) and BIORUPTOR^® tube (COSMO BIO). The conditions were set to Power: H, Time: 30 sec ON/30 sec OFF for nine times; then the tube was quickly ejected, and ice was added to the machine, and again Time: 30 sec ON/30 sec OFF for eight times. The sonicated samples were incubated overnight at 4°C with antibodies other than Input: rabbit anti‐H3K4me1 (ab8895; Abcam, Cambridge, UK), rabbit anti‐H3K4me2 (ab32356; Abcam), rabbit anti‐H3K4me3 (ab8580; Abcam), and mouse anti‐H3K27me3 (ab6002; Abcam). Dynabeads™ Protein G (Thermo Fisher Scientific) and Dynabeads™ M280 Sheep anti‐mouse IgG (Thermo Fisher Scientific) were used for immunoprecipitation of H3K4me1/2/3 and H3K27me3, respectively. Beads were washed twice each with PBS buffer, low‐salt RIPA buffer (50 mM Tris–HCl [pH 7.8], 150 mM NaCl, 1 mM EDTA [pH 8.0], 1% Triton X‐100, 0.1% SDS, 0.1% sodium deoxycholate) with 1% cOmplete™ protease inhibitor (Roche), high‐salt RIPA buffer (50 mM Tris–HCl [pH 7.8], 600 mM NaCl, 1 mM EDTA [pH 8.0], 1% Triton X‐100, 0.1% SDS, 0.1% sodium deoxycholate) with 1% cOmplete™ protease inhibitor (Roche), LNDET buffer (250 mM LiCl, 1% IGEPAL, 1% sodium deoxycholate, 1 mM EDTA [pH 8.0], and 10 mM Tris–HCl [pH 7.8]), and with TE buffer. After adding the elution buffer (10 mM Tris–HCl [pH 7.8], 0.3 m NaCl, 5 mM EDTA [pH 8.0], and 0.5% SDS), all beads were incubated at 65°C overnight. Lysates were treated with 10 mgml⁻¹ RNaseA at 37°C for 30 min and then treated with 20 mgml⁻¹ proteinase K and 20 mgml⁻¹ glycogen at 37°C for 2 h. After phenol‐chloroform extraction and ethanol precipitation, the pellet was suspended in Resuspension Buffer (Illumina). The collected DNA was quantified using the Qubit dsDNA High Sensitivity Assay Kit (Thermo Fisher Scientific), and 1 ng for H3K27me3 or 2.5 ng for others of DNA was used to construct a sequencing library with the KAPA Hyper Prep Kit for Illumina (Kapa Biosystems, Wilmington, MA, USA). Dual‐size selection was performed using Agencourt AMPure XP (Beckman Coulter, Brea, CA, USA) to enrich 300–500 bp fragments. Libraries were quantified using the Qubit dsDNA High Sensitivity Assay Kit (Thermo Fisher Scientific) and the 2100 Bioanalyzer (Agilent Technologies). Quantified libraries were pooled, and 100 bp single‐read sequences were obtained using the NovaSeq 6000 sequencer (Illumina). Two independent biological replicates were analyzed for each genotype.

ChIP‐Seq data analysis

Raw fastq files were quality‐filtered by fastp v0.20.0 (Chen, 2023) with ‐q 20 ‐l 30 parameters, then quality‐filtered reads were mapped onto the cDNA sequences of annotated genes and other transcripts of TAIR10 using bowtie2 v2.2.4 (Langmead & Salzberg, 2012). The resulting SAM files were converted to sorted BAM files using SAMtools 1.9 (Danecek et al., 2021) and then converted to BED files using BEDTools v2.26.0 (Quinlan & Hall, 2010). The ‘slop’ function of BEDTools was used to extend the 5′ end of RNA‐seq reads toward the 3′ direction to fit the average insertion size (300 bp) of the sequenced libraries. Then, the ‘coverage’ function of BEDTools was used to calculate the number of reads that overlapped with each annotation unit. Scatter plots were described using RPM. For H3K27me3, differentially hyper‐methylated genes were identified using the edgeR package v3.34.1 (Chen et al., 2025) in R version 4.1.1 treating two biological replicates as paired samples. Genes with P < 0.05 and FC >1 in each comparison were identified as differentially hyper‐methylated genes, other than low‐methylated genes defined as CPM ≤0.2 (Table S3). For visualization of H3K27me3 binding sites, TDF files were created using igvtools from BAM files and visualized using the Integrative Genome Viewer (Robinson et al., 2023). Waveforms of ChIP‐seq results are shown for only one of the biological replicates. In addition, GO analysis was conducted using Shiny GO ver 0.76.3 (Ge et al., 2020). Venn diagrams were also drawn using BioVenn (Hulsen et al., 2008). See also Methods S3.

Microscopic imaging

For observation of GFP, PI was applied to samples prior to imaging for the counterstaining of cell outlines. The staining solution was prepared by mixing 1 mgml⁻¹ PI solution (Sigma–Aldrich) with sterilized water to a final concentration of 0.1 mgml⁻¹. The solution was applied for 5 min before imaging. Explants were observed using an Olympus FV1200 confocal microscope with a UPLSAPO20X (NA = 0.75, WD = 0.6 mm) objective lens (Olympus, Tokyo, Japan) and FV10‐ASW software ver. 4.2a (Olympus). To detect the GFP signal, a 473 nm laser line was used for excitation and a 490–540 nm band pass filter was used for signal collection. To detect PI staining, a 559 nm laser line was used for excitation and a 575–675 nm band pass filter was used to collect the signal. The images were processed by Fiji software (Schindelin et al., 2012).

AUTHOR CONTRIBUTIONS

MD initiated this project. HS, TS, MD, and SM designed the experiments. AH, YI, and MD conducted experiments. AH and HS analyzed data. YS supported RNA‐Seq and ChIP‐Seq analysis technically. AH, HS, TS, MD, and SM wrote the manuscript. All authors contributed through discussions and reviewed the manuscript.

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

Supporting information

Figure S1. BRM does not appear to affect callus growth.

Figure S2. Phenotypes of wild type (WT) and brm‐3 at C7S1.

Figure S3. Gene ontology (GO) terms of genes downregulated or more upregulated specifically in brm‐3 upon shoot induction.

Figure S4. Expression in reported genes compared to WT and brm‐3 in the shoot regeneration process.

Figure S5. GO terms of genes with increased H3K27me3 compared to WT and brm‐3.

Figure S6. Profiles of genes with increased H3K27me3 specifically at each timepoint compared to WT and brm‐3.

Figure S7. H3K27me3 levels of RAM‐related genes.

Figure S8. Profiles of expression and H3K27me3 for BRM‐mediated primed genes and genome‐wide genes in WT and brm‐3 at each time point.

Figure S9. Profiles of expression and H3K27me3 for genes with H3K27me3 patterns similar to BRM‐mediated primed genes at each time point in WT and brm‐3.

Figure S10. Cis‐element enrichment analysis using the promoter regions in these BRM‐mediated primed genes.

Figure S11. Gene expression and H3K27me3 levels of NGA3 and DOT1 based on qRT‐PCR and ChIP‐qPCR.

Figure S12. Cis‐element enrichment analysis using the promoter regions of upregulated and downregulated genes in brm‐3 at C7S1.

Figure S13. GO terms of differentially expressed genes (DEGs) compared to WT and ref6 ^c elf‐3.

Figure S14. GO terms of genes upregulated or downregulated in brm‐3 and ref6 ^c elf6‐3 at C7S1.

Figure S15. Gene expression of genes downregulated in both brm‐3 and ref6c elf6‐3 and those downregulated only in brm‐3 at C7S1.

Figure S16. Analysis of REF6‐binding sites within genomic regions of downregulated genes in brm‐3 and ref6 ^c elf6‐3 at C7S1.

Figure S17. H3K27me3 levels of NGA3 and FLA9 based on ChIP‐qPCR.

Methods S1. Morphological observation and image quantification.

Methods S2. Additional methods for RNA‐Seq analysis.

Methods S3. Additional methods for ChIP‐Seq analysis.

Methods S4. Methods for cis‐element enrichment analysis.

Methods S5. Methods for qRT‐PCR.

Methods S6. Methods for ChIP‐qPCR.

Methods S7. Methods for motif analysis.

Table S1. Lists of differentially expressed genes (DEGs) compared to WT and brm‐3 (q <0.05, fold change (FC) >2 or <0.5).

Table S2. Lists of DEGs compared to C7 and C7S1 in WT (q <0.05, FC >2 or <0.5).

Table S3. Lists of differentially hyper‐methylated genes compared to WT and brm‐3 for H3K27me3 (P < 0.05, FC >1).

Table S4. The candidate target genes of BRM‐mediated epigenetic priming.

Table S5. Lists of DEGs compared to WT and ref6 ^c elf6‐3 (q <0.05, FC >2 or <0.5).

Table S6. The common target genes of BRM, REF6, and ELF6 for BRM‐mediated epigenetic priming.

Table S7. All primers used in this study.

DATA AVAILABILITY STATEMENT

Raw RNA‐Seq and ChIP‐Seq data have been deposited with links to BioProject accession numbers PRJDB12618 and PRJDB12627 in the DDBJ BioProject database.