The GTE4–EML chromatin reader complex concurrently recognizes histone acetylation and H3K4 trimethylation in Arabidopsis

Abstract

Histone acetylation and H3K4 trimethylation (H3K4me3) are associated with active transcription. However, how they cooperate to regulate transcription in plants remains largely unclear. Our study revealed that GLOBAL TRANSCRIPTION FACTOR GROUP E 4 (GTE4) binds to acetylated histones and forms a complex with the functionally redundant H3K4me3-binding EMSY-like proteins EML1 or EML2 (EML1/2) in Arabidopsis thaliana. The eml1 eml2 (eml1/2) double mutant exhibits a similar morphological phenotype to gte4, and most of the differentially expressed genes in gte4 were coregulated in eml1/2. Through chromatin immunoprecipitation followed by deep sequencing, we found that GTE4 and EML2 co-occupy protein-coding genes enriched with both histone acetylation and H3K4me3, exerting a synergistic effect on the association of the GTE4–EML complex with chromatin. The association of GTE4 with chromatin requires both its bromodomain and EML-interacting domain. This study identified a complex and uncovered how it concurrently recognizes histone acetylation and H3K4me3 to facilitate gene transcription at the whole-genome level in Arabidopsis.

The acetylated histone–binding protein GTE4 forms a complex with the H3K4me3-binding proteins EML1 and EML2 for the corecognition of histone modifications, facilitating transcription in Arabidopsis.

Introduction

Histone modifications play a crucial role in regulating the action of chromatin, including gene transcription, DNA replication, and DNA damage repair (Karlic et al. 2010; Bar-Ziv et al. 2016; Clouaire et al. 2018). One well-studied modification is histone acetylation at lysine sites, which is associated with increased chromatin accessibility and transcriptional activation (Struhl 1998; Bannister and Kouzarides 2011; Berr et al. 2011). Histone methylation at lysine sites is another histone modification type that is associated with transcription. The locations of methylated lysine sites are crucial for determining the effects of histone methylation on transcription. While the H3 lysines 4 and 36 methylation is associated with actively transcribed genes, the H3 lysines 9 and 27 methylation is associated with transcriptional repression (Suganuma and Workman 2011; Liu et al. 2018). Both histone acetylation and methylation are reversible modifications, which can be removed by histone deacetylases and demethylases, respectively (Mosammaparast and Shi 2010; Liu et al. 2014; Seto and Yoshida 2014; Hu and Du 2022). Gene expression is dynamically regulated by changes of histone modifications in plants (Brusslan et al. 2015; Song et al. 2019; Patrick et al. 2024). Moreover, various type of readers of histone modifications have been identified, and these readers are crucial for regulating chromatin status and gene expression (Marmorstein and Zhou 2014; Scheid et al. 2021). However, how different histone modification readers cooperate to regulate genome-wide transcription remains largely unclear.

The bromodomain is present in various chromatin-related proteins, including histone acetyltransferases, chromatin-remodeling complex components, and transcriptional regulators in eukaryotes (Pandey et al. 2002; Zeng and Zhou 2002; Filippakopoulos et al. 2012). In Arabidopsis (Arabidopsis thaliana), 29 bromodomain proteins have been identified (Pandey et al. 2002). However, only a small part of them has been characterized as histone acetylation readers (Zhang et al. 2016; Nie et al. 2019; Potok et al. 2019; Sijacic et al. 2019; Luo et al. 2020; Jaronczyk et al. 2021; Yu et al. 2021; Guo et al. 2022). Bromodomain and extra-terminal (BET) domain proteins constitute a significant subfamily of bromodomain proteins in eukaryotes (Pandey et al. 2002; Zeng and Zhou 2002). In fungi and metazoans, BET proteins have been shown to bind to acetylated histones, thereby regulating transcription (Ladurner et al. 2003; Wang et al. 2013), DNA damage signaling (Floyd et al. 2013), and cell cycling (LeRoy et al. 2008). In Arabidopsis, there are 12 BET proteins collectively known as GLOBAL TRANSCRIPTION FACTOR GROUP E 1 to 12 (GTE1 to GTE12). GTE1/IMBIBITION-INDUCIBLE 1 (IMB1) is involved in seed germination by facilitating the activation of genes related to cell wall metabolism and plastid-encoded genes (Duque and Chua 2003). GTE6 mediates the transcriptional activation of AS1, which encodes a MYB-domain protein involved in leaf development (Chua et al. 2005). GTE4 is essential for proper root and leaf development and plant immunity (Airoldi et al. 2010; Zhou et al. 2022). Additionally, GTE9 and GTE11, 2 closely related BET proteins, interact with BTB AND TAZ DOMAIN PROTEIN 2 (BT2) to mediate responses to abscisic acid (ABA) and sugar signaling (Misra et al. 2018). Despite the reported biological functions of these Arabidopsis BET proteins, the specific involvement of the bromodomain in their functions in plants has yet to be determined.

Yeast and metazoan BET proteins typically possess 2 to 5 tandem bromodomains that specifically bind to different combinations of acetylated lysine sites on histones, enabling the recognition of specific chromatin regions (Dyson et al. 2001; Pandey et al. 2002; Taniguchi 2016). The tandem arrangement of bromodomains enhances their binding affinity to acetylated histones synergistically (Moriniere et al. 2009; Filippakopoulos et al. 2012). In contrast, all Arabidopsis BET proteins have a single bromodomain (Pandey et al. 2002). Previous studies identified GTE1 as an acetylated histone–binding protein through in vitro assays, indicating that Arabidopsis BET family proteins likely possess a conserved ability with acetylated histones (Zhao et al. 2018). However, it remains unclear how these single-bromodomain BET proteins discriminate between chromatin regions with varying types or levels of histone acetylation. It is also unknown whether unique mechanisms for chromatin recognition have evolved in plants to compensate for the absence of tandem-bromodomain BET proteins. Furthermore, while several BET proteins in yeast and metazoans have been shown to interact with other chromatin regulators, such interactions have not yet been identified and characterized in plants (Floyd et al. 2013; Wang et al. 2013; Taniguchi 2016). Therefore, it is necessary to identify and study potential BET-interacting proteins in plants.

In this study, we identified 5 closely related Arabidopsis BET proteins (GTE2, GTE3, GTE4, GTE5, and GTE7) as acetylated histone–binding proteins and demonstrated that the binding of GTE4 to acetylated histones facilitates the association of GTE4 with chromatin. Moreover, we found that GTE4 interacts with 2 redundant EMSY-like (EML) proteins, EML1 and EML2 (EML1/2). EML1/2 contain an EMSY N-terminal (ENT) domain (Tsuchiya and Eulgem 2011), which is a characteristic of the BRCA2-interacting human EMSY protein (Hughes-Davies et al. 2003), and an Agenet/Tudor domain. The Tudor domain of EML1 was previously identified as a plant-specific reader of histone H3K4 trimethylation (H3K4me3) (Zhao et al. 2018). We demonstrated that GTE4 and EML1/2 co-occupy chromatin regions that are enriched with both histone acetylation and H3K4me3, and the occupancy of GTE4 and EML1/2 on chromatin is mutually dependent. Moreover, we have identified a conserved α-helix domain in GTE4 that is responsible for interacting with EML1 and EML2. These results provide insights into the cooperative recognition mechanism employed by single-bromodomain BET proteins and single-Tudor domain proteins and thereby shed light on the plant-specific combinatorial recognition of histone acetylation and H3K4me3.

Results

Identification and characterization of the GTE4 bromodomain

To identify acetylated histone–binding proteins in Arabidopsis, we performed a pull-down assay using the histone H4 peptide with acetylation at lysines 5, 8, 12, and 16 (H4K5/8/12/16ac) as a bait and then detected acetylated histone–interacting proteins by affinity purification followed by MS (AP-MS). This analysis revealed 5 BET proteins (GTE2, GTE3, GTE4, GTE5, and GTE7) as acetylated histone–interacting proteins (Fig. 1A; Supplementary Data Set 1). Although GTE1 was not found to be copurified with acetylated histones in our current AP-MS analysis, a previous in vitro pull-down assay demonstrated its strong binding affinity for the histone H4K5ac peptide (Zhao et al. 2018). Thus, it is likely that the acetylated histone–binding ability is conserved across all the Arabidopsis BET proteins.

Figure 1.

GTE4 and its homologs as major readers of histone acetylation in Arabidopsis plants. A) Identification of GTEs as acetylated histone–binding proteins by AP-MS. Acetylated histone–binding proteins were identified by AP-MS using the histone H4 peptide with acetylation of lysines 5, 8, 12, and 16 (H4K5/8/12/16ac) as a bait. B) Determination of the binding affinity of the GTE4 bromodomain with histone peptides by ITC assays. The histone peptides used in ITC include H3 (1-19 aa), H3K4/9/14/18ac (1-26 aa), H4 (1-20 aa), and H4K5/8/12/16ac (1-20 aa). C) Analyses of the bromodomains in GTE proteins. The schematic diagram of conserved domains of GTE4 is shown in the upper panel, and the alignment of the bromodomains in indicated GTE and BROMODOMAIN (BRD) proteins is shown in the lower panel. The conserved residues mutated in this study are labeled on the top of aligned sequences. D) Determination of the binding of WT and mutated bromodomains of GTE4 to histone peptides with and without lysine acetylation. Biotin-labeled histone peptides were incubated with WT and mutated bromodomains of GTE4 and subjected to pull-down assays. H3ac and H4ac represent H3K4/9/14/18ac (1-26 aa) and H4K5/8/12/16ac (1-18 aa), respectively. E and F) Effect of the bromodomain mutations on the function of GTE4 in Arabidopsis plants. WT GTE4 and mutated GTE4-Y497A/N498A (YN/AA) were expressed in the gte4 mutant for complementation testing. The morphological phenotype of 21-day-old plants, including the 3rd, 4th, 5th, and 6th rosette leaves E), and the statistical analysis of 21-day-old seedling weight F) are shown. The values of plant weight of 21-day-old seedlings are shown as means ± Sd (n = 20). P-values were determined by Student t-test with 2 tails. Bar, 2 cm. WT, wild-type.

Among the BET proteins, GTE4 has been shown to influence root and leaf development (Airoldi et al. 2010). To explore how the binding of GTE4 to acetylated histones contributes to its function, we focused our investigation on this protein. We expressed and purified the bromodomain of GTE4 tagged by glutathione S-transferase (GST) from bacteria and then performed isothermal titration calorimetry (ITC) to determine its binding affinity for acetylated histones. The ITC results indicated that the GTE4 bromodomain has a strong affinity for the H4 peptide acetylated at lysines 5, 8, 12, and 16 (H4K5/8/12/16ac), while displaying a lower affinity for the H3 peptide acetylated at lysines 4, 9, 14, and 18 (H3K4/9/14/18ac), and no detectable binding to unmodified H3 or H4 peptides (Fig. 1B). A histone peptide pull-down assay corroborated these observations, confirming that the GTE4 bromodomain binds strongly to H4K5/8/12/16ac and to a lesser extent to H3K4/9/14/18ac (Supplementary Fig. S1). Furthermore, the pull-down assay indicated that the GTE4 bromodomain is capable of interacting with H4K5ac and H4K12ac and to a lesser extent with H4K8ac and H4K16ac (Supplementary Fig. S1). However, the binding affinity of the GTE4 bromodomain for these monacetylated H4 peptides was markedly lower than for H4K5/8/12/16ac (Supplementary Fig. S1). This aligns with previous findings that conserved bromodomains in humans preferentially bind to polyacetylated histones (Moriniere et al. 2009; Filippakopoulos et al. 2012). These results confirm that the GTE4 bromodomain has a strong binding affinity for histone H4 with acetylation at multiple lysine sites.

The bromodomain present in BRD1, BRD2, and BRD13—subunits of BRAHMA (BRM)-associated SWITCH/SUCROSE NON-FERMENTABLE (SWI/SNF) complexes—exhibits a conserved capacity to interact with acetylated histones (Zhao et al. 2018; Yu et al. 2021; Guo et al. 2022). Sequence alignment revealed that the GTE4 bromodomain possesses conserved residues L452, Y455, A494, Y497, and N498, which are also found in BET proteins as well as in BRD1, BRD2, and BRD13 (Fig. 1C). To assess the influence of these conserved residues on the binding of the GTE4 bromodomain to acetylated histones, we introduced point mutations (L452D, Y455A, A494D, and Y497A/N498A) into the GTE4 bromodomain and then performed a histone peptide pull-down assay. The results indicated that the GTE4 bromodomain has a strong binding affinity for H4K5/8/12/16ac but does not bind to unmodified H4 (Fig. 1D; Supplementary Fig. S1). Furthermore, the pull-down assay confirmed that all mutations introduced completely disrupted the binding of the GTE4 bromodomain to the acetylated H4 (Fig. 1D). These findings strongly suggest that the conserved residues within the GTE4 bromodomain are critical for its interaction with acetylated histones.

Additionally, we introduced the Y497A/N498A mutation into the GTE4 transgene to evaluate whether the conserved bromodomain of GTE4 is necessary for its biological function in Arabidopsis plants. In alignment with previous findings on the gte4 mutant (Airoldi et al. 2010), our analysis showed that the gte4 mutant (SALK_083697) exhibited a smaller rosette size and lower seedling weight when compared with the wild type (Fig. 1, E and F). Complementation tests indicated that the developmental defects in the gte4 mutant were restored by the wild-type GTE4 transgene but not by the GTE4 transgene with the Y497A/N498A mutation (Fig. 1, E and F; Supplementary Fig. S2). These findings suggest that the conserved bromodomain is vital for the biological function of GTE4 in Arabidopsis.

GTE4 interacts with the H3K4me3-binding homologs EML1 and EML2

To investigate how GTE4 regulates chromatin state, we generated transgenic Arabidopsis plants that express GTE4 fused with a FLAG epitope (GTE4-FLAG) and identified GTE4-interacting proteins via AP-MS using an anti-FLAG antibody. Our findings showed that 2 closely related EMSY-like proteins, EML1 and EML2 (EML1/2), were copurified with GTE4, suggesting a possible interaction between GTE4 and EML1/2 (Fig. 2A; Supplementary Data Set 2). To further examine this interaction, we produced transgenic Arabidopsis plants expressing FLAG-tagged EML1/2 and conducted AP-MS to isolate proteins that interact with EML1/2. We found that GTE4 was also copurified with EML1/2 (Fig. 2A; Supplementary Data Set 2). EML1/2 proteins contain an ENT domain and an Agenet/Tudor domain. A previous study indicated that the Agenet domain of EML1 binds to trimethylated H3K4 (H3K4me3) peptide (Zhao et al. 2018), and the interaction was confirmed using pull-down assays in our study (Supplementary Fig. S3). Given that both histone acetylation and H3K4me3 are associated with active transcription (Struhl 1998; Bannister and Kouzarides 2011), it is likely that the corecognition of histone acetylation and H3K4me3 by the GTE4–EML complex contributes to transcriptional regulation at genomic loci enriched with both modifications.

Figure 2.

Determination of the interaction between GTE4 and EML1 or EML2. A) Identification of EML1 and EML2 as GTE4-interacting proteins by AP-MS. The AP-MS experiment was performed using anti-FLAG antibody in GTE4-FLAG, EML1-FLAG, and EML2-FLAG transgenic plants. B) Determination of the interaction between GTE4 and EML1 or EML2 by Y2H assays. The yeast strains expressing GAL4-AD and GAL4-BD-fused proteins were grown on SD-WLH supplemented by 3 mm 3-AT for determining the protein–protein interaction. The same yeast strains were grown on SD-WL as controls. C) Determination of the interaction among GTE4, EML1, and EML2 by pull-down assays. The proteins fused with His or MBP tags were subjected to MBP pull-down assays, followed by immunoblotting. Target bands are marked by arrows. D) Schematic diagrams of full-length and truncated versions of GTE4 and EML1. The length and conserved domains of GTE4 and EML1 are indicated. Both the N- and C-terminal positions of each truncated version are labeled. The domains responsible for the GTE4–EML1 interaction as determined by Y2H were marked by boxes. E) Identification of the specific domains responsible for the GTE4–EML1 interaction by Y2H. WT and truncated versions of GTE4 (GTE4-1∼GTE4-8) and EML1 (EML1-1∼EML1-4) were used in the Y2H assay. F) Determination of the effect of EID deletion on the interaction of GTE4 with EML1 and EML2 by AP-MS in Arabidopsis plants. Transgenic plants expressing WT and EID-deleted GTE4-FLAG (ΔEID) were subjected to AP-MS using anti-FLAG antibody. G) Determination of the effect of EID deletion on the GTE4–EML2 interaction by co-IP in Arabidopsis plants. WT and EID-deleted GTE4-FLAG transgenic plants were separately crossed to EML2-Myc transgenic plants, and the F1 plants harboring both the FLAG- and Myc-tagged transgenes were subjected to co-IP. H and I) Effect of the EID deletion on the function of GTE4 in Arabidopsis plants. WT and EID-deleted GTE4-FLAG (ΔEID) were expressed in the gte4 mutant for complementation testing. The morphological phenotype of 21-day-old plants, including the 3rd, 4th, 5th, and 6th rosette leaves H), and the statistical analysis of 21-day-old seedling weight I) are shown. The values of plant weight of 21-day-old seedlings are shown as means ± Sd (n = 20). P-values were determined by Student t-test with 2 tails. Bar, 2 cm. WT, wild-type; VEC, empty GAL4-AD or GAL4-BD vectors.

We assessed the interaction between GTE4 and EML1/2 using yeast 2-hybrid (Y2H) assays. The results revealed interactions not only between GTE4 and EML1/2 but also between EML1 and EML2 (Fig. 2B). These interactions were corroborated by pull-down assays (Fig. 2C). To identify the domains responsible for the GTE4–EML1 interaction, we conducted Y2H assays with truncated forms of GTE4, which led us to discover a domain (amino acids 251 to 316) critical for the interaction with EML1 (Fig. 2, D and E; Supplementary Fig. S4). This domain was predicted to form a conserved α-helix by AlphaFold (Supplementary Fig. S5). Importantly, deleting this domain from GTE4 disrupted its interaction with EML1 (Fig. 2E). Therefore, we designated this domain as the EML-interacting domain (EID). Additionally, Y2H assays indicated that the N-terminal portion of EML1 (amino acids 1 to 293), which includes the ENT and Agenet domains is crucial for interacting with GTE4 (Fig. 2, D and E).

To assess the effect of EID deletion on the interaction between GTE4 and EML1/2 in Arabidopsis, we generated transgenic plants expressing both wild-type GTE4 and a GTE4 variant lacking the EID domain (GTE4-ΔEID) (Supplementary Fig. S2). We then performed AP-MS to check whether EML1/2 could be copurified with either wild-type GTE4 or GTE4-ΔEID. The results showed that EML1/2 were copurified with wild-type GTE4, but not with GTE4-ΔEID (Fig. 2F; Supplementary Data Set 2). The effect of EID deletion on the GTE4–EML2 interaction was further verified through co-immunoprecipitation (co-IP) assays (Fig. 2G). These results confirm that the EID domain is essential for the GTE4–EML1/2 interaction in Arabidopsis. To further demonstrate the functional importance of this interaction, we conducted complementation testing by introducing the wild-type GTE4 transgene and the GTE4-ΔEID transgene into the gte4 mutant to evaluate their ability to restore the observed developmental defects. Notably, the wild-type GTE4 transgene effectively rescued the developmental defects, whereas the GTE4-ΔEID transgene did not (Fig. 2, H and I; Supplementary Figs. S2 and S6). These results suggest that the GTE4–EML1/2 interaction is essential for the functionality of GTE4 in Arabidopsis plants.

GTE4 and EML1/2 coregulate development and gene expression

To investigate the functional relationship between EML1/2 and GTE4 in Arabidopsis, we examined the phenotypes of the eml1 and eml2 single mutants. We found that the eml1 (SALK_077088) and eml2 (SALK_116222) single mutants did not display any noticeable developmental defects compared with the wild type. We therefore created an eml1 eml2 (eml1/2) double mutant through genetic crossing and assessed its developmental phenotypes alongside those of the gte4 mutant. We found that the eml1/2 double mutant and the gte4 mutant exhibited similar developmental defects, such as reduced rosette size and lower seedling weight (Fig. 3, A to C), indicating that EML1 and EML2 function redundantly in the GTE4–EML complex. To determine the genetic relationship between gte4 and eml1/2, we generated a gte4 eml1/2 (gte4/eml1/2) triple mutant via genetic crossing and compared the phenotypes of gte4, eml1/2, and gte4 eml1/2 mutants. The gte4/eml1/2 mutant showed more pronounced developmental defects than either gte4 or eml1/2 mutants alone (Fig. 3, A to C). These findings support the inference that GTE4 and EML1/2 collaboratively influence plant development.

Figure 3.

GTE4 and EML1/2 coregulate plant development and gene expression. A to C) Effect of gte4, eml1/2, and gte4/eml1/2 mutations on plant development. The morphological phenotype, including the 3rd, 4th, 5th, and 6th rosette leaves A), rosette width B), and seedling weight C) are indicated in 21-day-old plants of WT (Col-0), gte4, eml1/2, and gte4/eml1/2 mutants. The values of rosette width and plant weight of 21-day-old seedlings are shown as means ± Sd (n > 20). P-values were determined by Student t-test with 2 tails. Bar, 2 cm. D) Heatmaps showing the effect of gte4, eml1/2, and gte4/eml1/2 on the expression of the combined set of DEGs identified in gte4, eml1/2, and gte4/eml1/2. The log₂ (fold change of reads between mutant and WT) values are represented by color bars. E) The overlaps of DEGs identified in gte4, eml1/2, and gte4/eml1/2, relative to the WT (Col-0). The overlaps of downregulated (upper panel) and upregulated DEGs (lower panel) are independently shown. F) Scatter plots showing the correlation between the effects of gte4, eml1/2, and gte4/eml1/2 on gene expression. The DEGs identified in the gte4 mutant were subjected to the analysis. WT, Col-0. G) Relative expression levels of representative genes as determined by RNA-seq. The gene expression levels in the mutants were normalized to those in the WT (Col-0). The values are shown as means ± Sd (n = 3).

In order to understand how GTE4 and EML1/2 regulate development, we performed RNA deep sequencing (RNA-seq) for 3 biological replicates to analyze the effect of gte4, eml1/2, and gte4/eml1/2 on gene expression. Heat maps demonstrated that the differentially expressed genes (DEGs) identified in the replicates exhibited a strong correlation (Supplementary Fig. S7). By integrating the 3 replicates, we found that most DEGs (log₂FC > 0.5 or <−0.5, False Discovery Rate (FDR) < 0.05) identified in the gte4, eml1/2, and gte4/eml1/2 mutants were coregulated (Fig. 3D; Supplementary Data Set 3). A significant overlap of both upregulated and downregulated genes was observed among the 3 mutants (Fig. 3E). Scatter plot analysis showed a high positive correlation in the expression levels of DEGs between gte4 and eml1/2 mutants, as well as between gte4 and gte4/eml1/2 mutants (Fig. 3F). Notably, many DEGs in the gte4/eml1/2 mutant were not present in the gte4 or eml1/2 mutants (Fig. 3E), suggesting a synergistic effect between gte4 and eml1/2 in the regulation of gene expression. These results suggest that while GTE4 and EML1/2 can coregulate gene expression in the GTE4–EML complex, they also function independently of this complex.

In our analysis of coregulated genes in gte4, eml1/2, and gte4/eml1/2 mutants, we noted a marked decrease in the expression of CYC3, a member of the cyclin gene family (Wang et al. 2004), as well as PILS4, which is associated with auxin transport (Barbez et al. 2012) (Fig. 3G). This finding may account for the reduced size observed in these mutants. Furthermore, we detected a notable decline in the expression of genes related to disease resistance, including DIR9, DIR21, LECRK-V.3, and LECRK-V.4 (Bouwmeester and Govers 2009; Paniagua et al. 2017), in all 3 mutants (Fig. 3G). This aligns with earlier research highlighting the roles of EML1/2 in plant defense mechanisms against pathogens (Tsuchiya and Eulgem 2011; Coursey et al. 2018). Although the number of DEGs in the gte4/eml1/2 mutant was significantly greater than that in the gte4 and eml1/2 mutants (Fig. 3E), the effect of gte4/eml1/2 on the expression of specific target genes was comparable with that of the gte4 and eml1/2 mutations (Fig. 3G). This supports the notion that GTE4 and EML1/2 form a complex to coregulate the expression of these genes through the same pathway.

GTE4 and EML1/2 co-occupy genes enriched with histone acetylation and H3K4me3

To determine whether the GTE4–EML complex regulates gene expression by directly suppressing transcription, we investigated the occupancy of this complex on its target genes using chromatin immunoprecipitation followed by deep sequencing (ChIP-seq). Due to the functional redundancy between EML1 and EML2, we focus on EML2 as a representative in the ChIP-seq experiment. Using GTE4-FLAG and EML2-FLAG transgenic plants, our ChIP-seq analysis demonstrated that the results from 2 independent replicates were highly correlated (Supplementary Fig. S8), identifying 13,775 and 14,231 peaks for GTE4 and EML2, respectively (Supplementary Data Sets 4 and 5). Most peaks for GTE4 and EML2 were located within intragenic regions close to the transcription start site of protein-coding genes (Fig. 4, A to C). By assigning these peaks to genes, we identified 13,371 genes bound by GTE4 and 13,797 genes bound by EML2. A significant overlap was observed between the genes bound by GTE4 and EML2 (Fig. 4D), suggesting that GTE4 and EML1/2 co-occupy genes across the entire genome.

Figure 4.

GTE4 and EML2 co-occupy target genes enriched with both H4K5ac and H3K4me3. A) Distribution of GTE4 and EML2 ChIP-seq peaks at different genomic regions. Proximal and distal promoter regions represent 1 to 400 and 401 to 1,000 bp regions upstream to the TSS, respectively. Sample size: GTE4 peaks (n = 13,775); EML2 peaks (n = 14,231). B and C) Distribution of GTE4 and EML2 ChIP-seq signals over the GTE4- and EML2-enriched genes. Metaplots B) and heatmaps C) are independently shown. In heatmaps, the GTE4- and EML2-enriched genes were sorted by the enrichment levels of ChIP-seq signals. D) Venn diagram showing the overlap between GTE4- and EML2-bound genes. P-value determined by 1-tailed hypergeometric test indicates the significance of overlap. E) The H4K5ac and H3K4me3 ChIP-seq levels of genes co-occupied by GTE4 and EML2 as well as random genes in the wild-type plants. In box plots, the center line and box edges are the median and the interquartile range, respectively. Whiskers extend within 1.5 times the interquartile range. Sample size: genes co-occupied by GTE4 and EML2 (n = 12,535) and random genes (n = 10,000). P-values were determined by 2-tailed, Mann–Whitney U test (unpaired). F) The overlaps among genes co-occupied by GTE4 and EML2, H4K5ac-enriched genes, and H3K4me3-enriched genes. The significance of overlapping was indicated by P-values as determined by 1-tailed hypergeometric test. G) Genome viewers showing the ChIP-seq signals for GTE4, EML2, H4K5ac, and H3K4me3 at representative genomic region. The scales of RPKM are shown for ChIP-seq signals. H) The ratio of DEGs co-occupied by GTE4 and EML2 to total DEGs identified in gte4, eml1/2, and gte4/eml1/2 mutants. The upregulated and downregulated DEGs were independently analysed. TSS, transcription start site; TES, transcription end site.

Given that GTE4 binds to acetylated histones while EML1/2 bind to H3K4me3, we performed immunoblotting to determine the impact of gte4 and eml1/2 mutations on histone acetylation and H3K4me3. The results indicated that the H4K5ac, H4K12ac, and H3K4me3 levels remained unchanged in the gte4, eml1/2, or gte4/eml1/2 mutants compared with wild type (Supplementary Fig. S9). Additionally, GTE4 and EML1/2 did not interact with any histone acetylation or methylation modifying enzymes (Supplementary Data Set 2). These findings suggest that the GTE4–EML complex functions downstream of histone acetylation and H3K4me3 modifications.

Subsequently, we conducted a combined analysis of the ChIP-seq data for GTE4 and EML2 alongside the H4K5ac and H3K4me3 ChIP-seq data to determine whether GTE4 and EML2 preferentially bind to genes that are enriched with H4K5ac and H3K4me3. Since histone acetylation at different lysine sites are correlated (Earley et al. 2007), H4K5ac can be recognized as a representative of histone acetylation. By examining previously published H4K5ac and H3K4me3 ChIP-seq data (Shang et al. 2021; Zheng et al. 2023), we discovered that genes co-occupied by GTE4 and EML2 had significantly higher levels of H4K5ac and H3K4me3 compared to randomly selected genes (Fig. 4E; Supplementary Data Set 6). The enrichment of GTE4 and EML2 increased in gene deciles with rising levels of H4K5ac and H3K4me3, except for the last decile (Supplementary Fig. S10, A and B). While only a small subset of genes enriched for either H4K5ac or H3K4me3 showed co-occupied by GTE4 and EML2, the vast majority of genes that were enriched for both H4K5ac and H3K4me3 were co-occupied by GTE4 and EML2 (Fig. 4, F and G; Supplementary Fig. S10C). These findings suggest that GTE4 and EML1/2 preferentially associate with genes that are enriched in both histone acetylation and H3K4me3.

By conducting a combined analysis of RNA-seq and ChIP-seq data, we found that the genes co-occupied by GTE4 and EML2 were enriched in downregulated DEGs, as opposed to upregulated ones, in the gte4, eml1/2, and gte4/eml1/2 mutants (Fig. 4H). This suggests that the GTE4–EML complex associates with chromatin to promote the transcription of its target genes.

The binding of GTE4 and EML1/2 to chromatin is partially interdependent

To investigate the impact of EML1/2 on the chromatin binding of GTE4, we conducted ChIP-seq analysis using GTE4-FLAG transgenic plants in both the eml1/2 mutant and wild-type backgrounds (Supplementary Fig. S11A). Two replicates of the ChIP-seq data showed a strong correlation (Supplementary Fig. S8), suggesting a high quality of the data. The ChIP-seq results revealed that the numbers of genes bound by GTE4 in the eml1/2 mutant and the wild type were comparable and exhibited a significant overlap (Fig. 5A; Supplementary Data Set 4). Additionally, we performed ChIP-seq using EML2-FLAG transgenic plants in both the gte4 mutant and wild type (Supplementary Fig. S11B) and obtained high-quality ChIP-seq data (Supplementary Fig. S8). We found that the genes bound by EML2 in the gte4 mutant exhibited comparable abundance to those in the wild type, with a high degree of overlap (Fig. 5A; Supplementary Data Set 5).

Figure 5.

GTE4 and EML1/2 exhibit a partial interdependent binding to their target genes. A) Venn diagrams showing the overlap of GTE4-bounds genes between WT (Col-0) and eml1/2 mutant (upper panel) and the overlap of EML2-bound genes between WT (Col-0) and gte4 mutant (lower panel). P-values were determined by 1-tailed hypergeometric test. B) Determination of the effect of eml1/2 on GTE4 ChIP-seq signals and the effect of gte4 on EML2 ChIP-seq signals. Meta plots show the average GTE4 and EML2 ChIP-seq signals in the WT (Col-0), gte4, or eml1/2 mutants, across genes co-occupied by GTE4 and EML2, GTE4-specific genes, and EML2-specific genes. TSS, transcription start site; TES, transcription end site. C) Box plots showing the effect of eml1/2 on the enrichment of GTE4 and the effect of gte4 on the enrichment of EML2. The enrichment levels of GTE4 and EML2 were determined at peaks co-occupied by GTE4 and EML2 (n = 12,761 for GTE4 peaks and n = 12,790 for EML2 peaks), GTE4-specific peaks (n = 839), and EML2-specific peaks (n = 1,284). In box plots, the center line and box edges are the median and the interquartile range, respectively. Whiskers extend within 1.5 times the interquartile range. P-values were determined by 2-tailed, paired Mann–Whitney U test. D) Genome viewers showing the effect of eml1/2 on GTE4 enrichment and the effect of gte4 on EML2 enrichment at the representative genomic region. The scale of RPKM is shown. E) Bar charts showing the quantities of GTE4-bound genes with eml1/2-affected enrichment and the quantities of EML2-bound genes with gte4-affected enrichment. GTE4-bound genes identified in either the WT or eml1/2 mutant were combined for determining the effect of eml1/2 on the enrichment of GTE4; EML2-bound genes identified in either the WT or gte4 mutant were combined for determining the effect of gte4 on the enrichment of EML2. F and G) The numbers of GTE4-bound genes for which GTE4 enrichment are decreased by eml1/2, alongside EML2-bound genes where EML2 enrichment is reduced by gte4, within deciles that show increasing levels of H4K5ac F) and H3K4me3 G). WT (wild type) represents Col-0.

Furthermore, we examined the ChIP-seq levels of GTE4 and EML2 at genes co-occupied by both factors, in addition to those occupied by either of them. Our analysis indicated that GTE4 enrichment was substantially higher at genes co-occupied by GTE4 and EML2 compared with those unique to GTE4 (Fig. 5, B and C; Supplementary Data Set 7). In the eml1/2 mutant, GTE4 enrichment at genes co-occupied by GTE4 and EML2 was notably decreased, while there was only a slight reduction at genes exclusive to GTE4 (Fig. 5, B and C). Likewise, EML2 enrichment was considerably greater at genes shared with GTE4 than at genes exclusive to EML2 (Fig. 5, B and C; Supplementary Data Set 7). In the gte4 mutant, EML2 enrichment at co-occupied genes was significantly reduced but remained unchanged for genes specific to EML2 (Fig. 5, B and C). These observations support the notion that GTE4 and EML2 mutually enhance their binding to common target genes.

Despite the reduced chromatin binding of GTE4 and EML2 in the eml1/2 and gte4 mutants, respectively, their binding was not entirely eliminated (Fig. 5, B to D). This suggests that the chromatin binding of GTE4 and EML1/2 is partially interdependent. Furthermore, we observed that most target genes of GTE4 and EML2 identified in the wild type exhibited reduced enrichment levels in the eml1/2 and gte4 mutants, respectively, whereas the other target genes showed no significant change in GTE4 and EML2 enrichment in the mutants (Fig. 5E). These findings confirm the idea that GTE4 and EML1/2 binding to chromatin is partially interdependent.

To explore the association of H4K5ac and H3K4me3 with GTE4 and EML2 target genes at the whole-genome level, we investigated the levels of H4K5ac and H3K4me3 in 2 groups of GTE4 target genes, which were classified based on the impact of eml1/2 on GTE4 enrichment. The box plots demonstrated that the eml1/2 mutation typically decreases GTE4 enrichment at target genes exhibiting relatively high levels of H4K5ac and H3K4me3 (Supplementary Fig. S12A). Similarly, we evaluated the H4K5ac and H3K4me3 levels in 2 categories of EML2 target genes, classified based on the effect of the gte4 mutation on EML2 enrichment. Interestingly, we found that the gte4 mutation tends to decrease EML2 enrichment at target genes with relatively low levels of H4K5ac and H3K4me3 (Supplementary Fig. S12B). Furthermore, we quantified the GTE4-bound genes with reduced GTE4 enrichment in the eml1/2 mutant and the EML2-bound genes with decreased EML2 enrichment in the gte4 mutant, across gene deciles sorted by H4K5ac levels. Our analysis showed that in the deciles with low H4K5ac levels, the number of GTE4-bound genes affected by the eml1/2 mutation was comparable with that of EML2-bound genes influenced by the gte4 mutation (Fig. 5F). Conversely, in the high H4K5ac deciles, the count of GTE4-bound genes impacted by the eml1/2 mutation was markedly greater than that of EML2-bound genes affected by the gte4 mutation (Fig. 5F). We conducted the same analysis in the H3K4me3 deciles and found that both the eml1/2 and gte4 mutations have a similar effect to that observed in the H4K5ac deciles (Fig. 5, F and G). These findings suggest that EML1/2 play a more important role than GTE4 in mediating the association of the GTE4–EML complex with genes exhibiting high levels of H4K5ac and H3K4me3, while they have a similar impact on the association of this complex with genes showing lower levels of these modifications.

The bromodomain and EID are crucial for the association of GTE4 with genes enriched for histone acetylation and H3K4me3

Although the binding of the GTE4 bromodomain to acetylated histones has been demonstrated (Fig. 1, B to D), it remains unclear whether this binding is necessary for the association of GTE4 with chromatin in vivo. To investigate this, we performed ChIP-seq using transgenic plants expressing wild-type GTE4 and the mutated GTE4-Y497A/N498A in which the mutations in the bromodomain disrupted its ability to bind acetylated histones (Fig. 1D; Supplementary Fig. S2). The ChIP-seq results showed that the genes bound by the GTE4-Y497A/N498A exhibited a significant overlap with those bound by wild-type GTE4 (Fig. 6A; Supplementary Data Set 8). However, in genomic regions bound by wild-type GTE4, the overall enrichment of GTE4-Y497A/N498A was significantly lower than that of wild-type GTE4 (Fig. 6, B to E; Supplementary Data Set 9), suggesting that the conserved bromodomain is involved in the association of GTE4 with chromatin. The Y497A/N498A mutation resulted in a notable decrease in GTE4 enrichment at the majority of genes it binds (Fig. 6, F and G), reinforcing the notion that the bromodomain binding to acetylated histones is crucial for the association of GTE4 with chromatin.

Figure 6.

Both the bromodomain and EID are involved in the binding of GTE4 to chromatin. A) The overlaps of genes bound by WT GTE4, GTE4-Y497A/N498A (YN/AA), and GTE4-ΔEID. P-values were determined by 1-tailed hypergeometric test. B) The average ChIP-seq signals of WT GTE4, GTE4-Y497A/N498A (YN/AA), and GTE4-ΔEID (ΔEID) over the WT GTE4-bound genes. C) The enrichments of WT GTE4, GTE4-Y497A/N498A (YN/AA), and GTE4-ΔEID (ΔEID) at the WT GTE4 ChIP-seq peaks (n = 13,775). In box plots, the center line and box edges are the median and the interquartile range, respectively. Whiskers extend within 1.5 times the interquartile range. P-values shown in the box-plot chart were determined by 2-tailed, paired Mann–Whitney U test. D) Heat maps showing the distribution of ChIP-seq signals for WT GTE4, GTE4-Y497A/N498A (YN/AA), and GTE4-ΔEID (ΔEID) over GTE4-bound genes. The genes were sorted by the degree of GTE4 enrichment at its target genes. E) Genome viewers showing the effects of Y497A/N498A (YN/AA) mutation and EID deletion (ΔEID) on GTE4 enrichment at the representative genomic region. The scale of RPKM is shown. F) Scatter plots showing the comparison between the enrichment of WT GTE4 and the enrichments of GTE4-Y497A/N498A (YN/AA) or GTE4-ΔEID (ΔEID) at WT GTE4-bound peaks. G) Bar charts showing the numbers of genes with upregulated and downregulated levels of GTE4 enrichment caused by the Y497A/N498A mutation (YN/AA) and the EID deletion (ΔEID). H and I) The H4K5ac and H3K4me3 levels divided into 2 sets of GTE4-bound genes according to the effect of the Y497A/N498A mutation (YN/AA) H) or the EID deletion (ΔEID) I) on GTE4 enrichment. In box plots, the center line and box edges are the median and the interquartile range, respectively. Whiskers extend within 1.5 times the interquartile range. Sample size: YN/AA vs WT (down: n = 10,754; unchanged: n = 2,299); ΔEID vs WT (down: n = 10,899; unchanged: n = 2,170). P-values were determined by 2-tailed, Mann–Whitney U test (unpaired). TSS, transcription start site; TES, transcription end site; WT, wild-type GTE4.

Given that EML1 has been found to interact with H3K4me3 (Zhao et al. 2018), we sought to investigate whether disrupting the GTE4–EML interaction through EID domain deletion affects the association of GTE4 with chromatin. We performed ChIP-seq for EID-deleted GTE4 (GTE4-ΔEID) using transgenic plants expressing GTE4-ΔEID (Supplementary Fig. S2). Our analysis revealed that the genes bound by GTE4-ΔEID significantly overlapped with those bound by wild-type GTE4 (Fig. 6A; Supplementary Data Set 8). However, at the genomic loci bound by GTE4, the enrichment of GTE4-ΔEID was notably reduced compared to that of wild-type GTE4 (Fig. 6, B to E; Supplementary Data Set 9), indicating that the EID domain is necessary for the association of GTE4 with chromatin. Furthermore, we detected a significant decrease in GTE4 enrichment at most of its target loci when the EID domain was deleted (Fig. 6, F and G). Notably, the effect of EID deletion on GTE4 enrichment was more pronounced than that observed in the eml1/2 mutant (Figs. 5, B to D and 6, B to E), suggesting that the EID domain can also participate in the association of GTE4 with chromatin in an EML1/2-independent manner. Considering the presence of 2 additional EML homologs (EML3 and EML4) in Arabidopsis, it is plausible that EML3 and/or EML4 may function redundantly with EML1/2 to enhance the association of GTE4 with chromatin.

Additionally, we examined the H4K5ac and H3K4me3 levels in GTE4 target genes where GTE4 enrichment was diminished due to the Y497A/N498A mutation and EID deletion. We classified GTE4 target genes into 2 groups based on whether the Y497A/N498A mutation or the EID deletion affect GTE4 enrichment and assessed their histone modifications. The box plots indicated that GTE4 target genes with reduced enrichment due to the Y497A/N498A mutation and those affected by the EID deletion exhibited relatively high levels of H4K5ac and H3K4me3 when compared with those not affected by the Y497A/N498A mutation or the EID deletion (Fig. 6, H and I). These findings support the idea that the bromodomain and the EML1/2-interacting domain in GTE4 are both involved in its association with target genes that are enriched in H4K5ac and H3K4me3.

Discussion

The tandem PLANT HOMEODOMAIN (PHD)-bromodomain module has been found in numerous chromatin-related proteins in yeast and metazoans (Ito et al. 1999; Schultz et al. 2001; Ruthenburg et al. 2007; Plotnikov et al. 2014; Tallant et al. 2015; Bardhan et al. 2023). BROMODOMAIN PHD FINGER TRANSCRIPTION FACTOR (BPTF), a core subunit of the human NUCLEOSOME REMODELING FACTOR (NURF) chromatin-remodeling complex, contains a PHD-bromodomain module that facilitates coordinated recognition of histone acetylation and H3K4me3 (Ruthenburg et al. 2011). Although a PHD-bromodomain module was reported in MBD9, a component of SWR1 complex responsible for H2A.Z deposition, in Arabidopsis (Nie et al. 2019; Potok et al. 2019; Sijacic et al. 2019; Luo et al. 2020), the disruption of PHD or bromodomain did not affect the biological function of MBD9 (Luo et al. 2020). It remains unclear whether and how the coordination between histone acetylation and H3K4me3 play a crucial role in plants. In this study, we found that the acetylated histone–binding protein GTE4 forms a complex with the H3K4me3-binding EML1/2, revealing coordination between the 2 histone modifications. Considering the well-established roles of histone acetylation and H3K4me3 in transcriptional activation (Struhl 1998; Bannister and Kouzarides 2011), the combinatorial recognition of the 2 histone modifications likely synergistically contributes to the activation of transcription. GTE4 was previously thought to have both activating and repressive roles in transcriptional regulation (Zhou et al. 2022). However, through the analysis of RNA-seq and ChIP-seq data, we discovered that the genes bound by both GTE4 and EML are predominantly enriched in downregulated DEGs, rather than upregulated ones, in the gte4 and eml1/2 mutants, supporting the notion that GTE4 primarily functions in transcriptional activation.

We found that the bromodomain of GTE4 exhibits a strong binding affinity for H4 with multiple acetylated lysine sites. Previous studies have investigated the binding specificity of bromodomains to different acetylated lysine sites on histones in various BET proteins of other eukaryotes (Owen et al. 2000; Zeng and Zhou 2002). Structural analyses have revealed how a mono bromodomain can simultaneously bind to polyacetylated histone lysine sites (Moriniere et al. 2009). In our current study, we observed that the single bromodomain of GTE4 has significantly higher binding affinity for polyacetylated histone peptides compared with their monoacetylated counterparts. This suggests that the recognition of polyacetylated histone lysine sites by a mono bromodomain is conserved in plants. The strong binding affinity of GTE4 for polyacetylated H4 likely contributes to its association with chromatin exhibiting high levels of histone acetylation. Previously, EML1 was identified as an H3K4me3-binding protein, and the Agenet/Tudor domain of EML1 was found to act as a plant-specific H3K4me3-binding module (Zhao et al. 2018). Our study suggests that the GTE4–EML complex corecognizes histone acetylation and H3K4me3 on their shared target genes, thereby establishing a connection between the 2 histone modifications in plants.

Although our study indicates that GTE4 and EML1/2 form a complex in Arabidopsis, we cannot exclude the possibility that either the GTE4 homologs or the EML1/2 homologs are also involved in the formation of redundant GTE–EML complexes. In addition to GTE4, several GTE4 homologs, including GTE3, GTE7, GTE8, GTE9, GTE10, and GTE11, were copurified with EML2, as determined by AP-MS (Supplementary Data Set 2). Similarly, in addition to EML1/2, the EML1/2 homolog EML3 was copurified with GTE4 (Supplementary Data Set 2). These results support the inference that the GTE4 and EML1/2 homologs form redundant GTE–EML complexes. The functional redundancy of the GTE4 and EML1/2 homologs provides a plausible explanation for the finding that only a small part of GTE4- and EML1/2-occupied genes are differentially expressed in the gte4 and eml1/2 mutants. It is also possible that the DEGs in the gte4 and eml1/2 mutants can only be identified during specific developmental stages or under certain adverse environmental conditions but not under the conditions tested in the current study.

Previous studies indicated that GTE4 and EML proteins are involved in the regulation of plant development (Milutinovic et al. 2019) and disease resistance (Tsuchiya and Eulgem 2011; Coursey et al. 2018; Zhou et al. 2022), suggesting the GTE4–EML complex has multiple functions in Arabidopsis plants. We predict that the GTE–EML complexes are responsible for establishing a priming state for transcriptional activation rather than for directly mediating transcriptional activation, and certain uncharacterized transcription regulators are specifically responsible for the activation of different sets of GTE4- and EML1/2-occupied genes. Further studies are required to identify which transcription factors are involved in the activation of different sets of GTE4- and EML1/2-occupied genes and to determine how the transcription factors are coordinated with the GTE4–EML complex to regulate the activation of these genes.

Materials and methods

Plant materials and constructs

Arabidopsis plants used in this study are in the Columbia (Col-0) background. The T-DNA insertion mutants gte4 (SALK_083697), eml1 (SALK_077088), and eml2 (SALK_116222) were obtained from Arabidopsis Biological Resource Center. The eml1/2 double mutant and the gte4/eml1/2 triple mutant were generated by crossing. Arabidopsis seedlings were grown on MS medium plates with a 16-h light and 8-h dark photoperiod at 22 °C. Twelve-day-old plants were transferred to soil and grown in the culture room under the same photoperiod and temperature conditions.

The full-length GTE4, EML1, and EML2 genomic DNA sequences driven by their own promoters were fused to 3×FLAG tag or 3×Myc tags at their 3′-terminal regions and cloned into the modified pCAMBIA1305 vector. The full-length and EID-deleted versions of GTE4 coding sequences driven by the native GTE4 promoter was also fused to 3×FLAG tag at their 3′-terminal regions and cloned into the modified pCAMBIA1305 vector. The primers used for generating the constructs are shown in Supplementary Data Set 10. The constructs were transformed into the Agrobacterium strain GV3101 and introduced into Arabidopsis plants via the floral-dip method. The T1 transgenic seedlings were selected on MS medium supplemented by 30 mg/L hygromycin and 50 mg/L ampicillin. The transgenic plants were subjected to AP-MS. The EID-deleted GTE4 fragment was amplified from the pGTE4::GTE4-FLAG construct and cloned into the pCAMBIA1305 vector, thereby obtaining the pGTE4::GTE4-ΔEID-FLAG construct. The Y497A/N498A mutations were introduced into GTE4 using the Mut Express MultiS Fast Mutagenesis Kit (Vazyme, C215-01).

Yeast 2-hybrid assay

The full-length and a series of truncated sequences of GTE4, EML1, and EML2 cDNA were cloned into pGADT7 and pGBKT7 vectors. The pGADT7 and pGBKT7 constructs were transformed into the yeast strains AH109 and Y187, respectively. The strain AH109 grew on synthetic dropout medium lacking Leu (SD-L); the strain Y187 grew on synthetic dropout medium lacking Trp (SD-W). Then, the positive AH109 and Y187 strains were subjected to mating in yeast peptone dextrose adenine liquid medium for 16 to 20 h. After mating, the culture was spread on a synthetic dropout medium plate lacking both Trp and Leu (SD-WL). The positive colonies grown on SD-WL medium were suspended and spotted on synthetic dropout medium lacking Trp, Leu, and His (SD-WLH) supplemented by 3 mm 3-amino-1,2,4-triazole (3-AT) and on SD-WL.

Affinity purification, MS, and co-IP

Three grams of plant material grown on MS medium was harvested and ground into fine powder in liquid nitrogen. The powder was suspended in 10 mL lysis buffer (50 mm Tris-HCl [pH 7.5], 150 mm NaCl, 5 mm MgCl₂, 10% glycerol, 0.1% NP-40, 0.5 mm DTT, 1 mm PMSF, and 1% Roche protease inhibitor cocktail) and rotated for 30 min at 4 °C, followed by centrifugation at 12,000 × g for 10 min. The supernatant containing total plant proteins was passed through Miracloth (Millipore, 475855-1R) and incubated with 100 μL of anti-FLAG M1 Agarose affinity gel (Sigma, A4596) for 2 h. The beads were harvested by centrifugation at 150 × g for 3 min at 4 °C and then washed 4 times with lysis buffer. The proteins bound to the beads were eluted with 3×FLAG peptide (Sigma, F4799) for 30 min at 4 °C. The eluate was run on a 10% SDS-PAGE gel, followed by silver staining with the ProteoSilver Silver Stain Kit (Sigma, PROT-SIL1).

Acetylated histone–binding proteins were identified from Arabidopsis plants as previously described (Qian et al. 2021). In brief, total proteins were extracted from 2 g of wild-type plants and then incubated with 2 μg of biotin-labeled H4K5/8/12/16ac (1 to 20 aa) peptides (Scilight-Peptide, Beijing, China) for 3 h at 4 °C with rotation. Forty microliters of of Streptavidin MagneSphere Paramagnetic Particles (Promega, Z5481) were added to the mixture and incubated for 1 h at 4 °C with rotation. After the particles were washed 3 times with wash buffer 1 (320 mm NaCl, 50 mm Tris-HCl [pH 8.0], and 0.5% NP-40) and 3 times with wash buffer 2 (150 mm NaCl, and 50 mm Tris-HCl [pH 8.0]), they were suspended and boiled in 1 × SDS sample buffer and run on a 10% SDS-PAGE gel, followed by silver staining.

The MS analysis was performed as previously described (Tan et al. 2020). In brief, the silver-stained protein bands were excised, de-stained, and digested with trypsin at 37 °C overnight. The peptides were eluted on a capillary column and sprayed into a linear trap quadrupole mass spectrometer equipped with a nano-ESI ion source (Thermo Fisher Scientific). The obtained spectra were searched on the Mascot server (Matrix Science Ltd., London, UK) against the International Protein Index Arabidopsis protein database.

For co-IP, the total proteins extracted from Arabidopsis plants were subjected to immunoprecipitation using anti-FLAG M1 Agarose affinity gel (Sigma, A4596), followed by immunoblotting. pEML2::EML2-Myc transgenic plants were crossed to pGTE4::GTE4-FLAG and pGTE4::GTE4-ΔEID-FLAG transgenic plants, and the F1 plants harboring both the Myc- and FLAG-tagged transgenes were used for co-IP.

RNA transcript analysis

Twenty-day-old wild-type, gte4, eml1/2, and gte4/eml1/2 plants grown in soil were used for RNA-seq analysis. One gram of plant material was subjected to total RNA extraction using Trizol reagent (Invitrogen) strictly adhering to the manufacturer's instructions. Subsequently, the extracted total RNA was sent to Berry Genomics Co., Ltd for reverse transcription and library construction. The constructed libraries were then sequenced using an Illumina NovaSeq6000 instrument with a paired-end scheme (2 × 150 bp). After adapters and poor-quality reads were removed, the clean reads were mapped to the Arabidopsis genome (TAIR 10) with HISAT2 (v2.2.0) (Kim et al. 2019). The DEGs were identified with a filter (FDR < 0.05, |log₂FC| > 0.5) with the R (v4.0.3) package edgeR (v3.32.1) (Robinson et al. 2010). The RNA-seq results shown in this study were from 3 independent biological replicates. The heatmap of DEGs was draw by using the R package gplots (v3.1.1). Venn diagrams were drawn by using the R package ggvenn (v0.1.9). Scatter plots were drawn using the R package ggplot2 (v3.3.5).

Chromatin immunoprecipitation

Transgenic Arabidopsis plants used for ChIP-seq comprise: wild-type GTE4-FLAG, GET4-Y497A/N498A-FLAG, and GTE4-ΔEID-FLAG transgenic plants in the gte4 background, GTE4-FLAG transgenic plants in both wild-type and eml1/2 backgrounds, and EML2-FLAG transgenic plants in both wild-type and gte4 backgrounds. Four grams of 12-day-old seedlings were harvested and fixed with 1% formaldehyde (Sigma, F8775). Samples were ground in liquid nitrogen and homogenized in 40 mL of NEB1 buffer (10 mm Tris-HCl [pH 8.0], 0.4 m sucrose, 10 mm MgCl₂, 5 mm DTT, 0.1 mm PMSF, and 1% Roche protease inhibitor cocktail tablet). The homogenate was filtrated through 2 layers of Miracloth and then centrifuged at 2,000 × g for 20 min at 4 °C. After washing 4 times with NEB2 buffer (10 mm Tris-HCl [pH 8.0], 0.25 m sucrose, 10 mm MgCl₂, 1% Triton X-100, and 0.1 mm PMSF), the nuclei were resuspended in 1.2 mL of NEB3 buffer (10 mm Tris-HCl [pH 8.0], 1.7 m sucrose, 2 mm MgCl₂, 0.15% Triton X-100, 0.1 mm PMSF, and 1% Roche protease inhibitor cocktail tablet). Subsequently, 600 μL of NEB3 was added to a new tube and overlaid with the sample solution before centrifugation at 14,000 × g for 1 h at 4 °C. The pellet was resuspended in 400 μL of nuclear lysis buffer (50 mm Tris-HCl [pH 8.0], 10 mm EDTA, 1% SDS, 0.1 mm PMSF, and 1% Roche protease inhibitor cocktail tablet) and diluted by adding 800 μL of dilution buffer (16.7 mm Tris-HCl [pH 8.0], 1.2 mm EDTA, 167 mm NaCl, 1.1% Triton X-100, 0.1 mm PMSF, and 1% Roche protease inhibitor cocktail tablet). The sample was then sonicated using Bioruptor (Diagenode). Following centrifugation at 16,000 × g for 15 min at 4 ˚C, the supernatant was diluted by adding 2.8 mL of dilution buffer. After preclearing using protein G beads (Thermo Fisher Scientific), the chromatin was immunoprecipitated using anti-FLAG antibody (SIGMA, F1804) and then incubated with protein G beads. The slurry was washed twice with low salt buffer (20 mm Tris-HCl [pH 8.0], 2 mm EDTA, 150 mm NaCl, 0.1% SDS, 1% Triton X-100, and 0.1 mm PMSF), high salt buffer (20 mm Tris-HCl [pH 8.0], 2 mm EDTA, 500 mm NaCl, 0.1% SDS, 1% Triton X-100, and 0.1 mm PMSF), LiCl wash buffer (10 mm Tris-HCl [pH 8.0], 1 mm EDTA, 0.25 m LiCl, 1% sodium deoxycholate, and 1% NP-40, 0.1 mm PMSF), and TE buffer (10 mm Tris-HCl [pH 8.0], 1 mm EDTA, and 0.1 mm PMSF). The protein–DNA complex was eluted with elution buffer (0.1% SDS and 0.1 m NaHCO₃), reverse cross-linked at 65 °C, and treated with proteinase K and RNase. The DNA was purified and used to prepare DNA libraries with the NEXTflex Rapid DNA-Seq Kit (Bioo Scientific, Austin, TX, USA).

The libraries were sent to Berry Genomics Co., Ltd and sequenced on an Illumina NovaSeq6000 instrument using a paired-end scheme (2 × 150 bp). After adapters and low-quality reads were removed, the clean reads were mapped to the Arabidopsis genome by Bowtie2 (v2.3.4) allowing 1 mismatch (Langmead and Salzberg 2012). Probable PCR duplicates were removed by using Picard Tools (v2.23.0) with MarkDuplicates. For FLAG ChIP-seq, enriched peaks were identified by MACS2 (v2.2.7.1) using the input reads as a negative control (Zhang et al. 2008). The read counts were normalized to reads per kilobase per million mapped reads (RPKM) by the number of clean reads mapped to the genome in each library. The ChIP-seq results shown in this study are from 2 independent biological replicates. The boxplots and scatterplots were drawn by using the R package ggplot2 (v3.3.5). The profile plot and heatmap of ChIP-seq were drawn by using DeepTools (v3.5.1). Venn diagrams were drawn by using the R package ggvenn (v0.1.9).

Protein purification and pull-down assay

For protein purification, the full-length or truncated coding sequences of GTE4, EML1/2 were cloned into the pET28a+, pMAL-c5x, and pGEX6p-1 vectors in fusion with His, MBP, and GST tags, respectively. The primers utilized for cloning are detailed in Supplementary Data Set 10. Site-directed mutagenesis was performed using Mut Express II fast mutagenesis Kit V2 (Vazyme Biotech) to express the mutated bromodomain versions of GTE4, in which the mutations include L452D, Y455A, A494D, and Y497A/N498A. The resultant constructs were verified by Sanger sequencing and subsequently transformed into the Escherichia coli strain BL21 (Rosetta DE3) for protein expression.

To induce protein expression, the cultures were grown in LB medium supplemented with 0.1 mm isopropyl β-D-thiogalactopyranoside at 16 °C for 12 h. Following this, the bacteria were harvested and resuspended in His-tag lysis buffer (20 mm Tris-HCl [pH 7.5], 500 mm NaCl, 20 mm imidazole, 1 mm DTT, and 1 mm PMSF), MBP-tag lysis buffer (20 mm Tris-HCl [pH 7.5], 200 mm NaCl, 1 mm EDTA, 1 mm DTT, and 1 mm PMSF), or GST-lysis buffer (20 mm Tris-HCl [pH 7.5], and 150 mm NaCl). The resuspended samples were sonicated and then centrifuged for 1 h at 4 °C. The supernatant was incubated with His-beads (70666-4; Millipore), Dextrin-beads (SA077025; Smart-lifesciences), or Glutathione-Sepharose 4B (GE Healthcare, 71024800-EG) at 4 °C for 1 h. After incubation, the protein-bound beads were washed 5 times with their respective lysis buffers and eluted using His-tag elution buffer (His-tag lysis buffer containing 250 mm imidazole), MBP-tag elution buffer (MBP-tag lysis buffer with 10 mm maltose), or GST elution buffer (50 mm Tris-HCl [pH 7.5], 150 mm NaCl, and 20 mm glutathione).

For pull-down assays, proteins tagged with MBP and His were mixed in MBP-tag lysis buffer and incubated at 4 °C for 1 h. Prior to bead incubation, 10% of the mixture was removed as an input control. The remaining mixture was then incubated with Dextrin-beads, with gentle rotation at 4 °C for another hour. Following incubation, the beads were washed 5 times, each for 5 min, using MBP-tag lysis buffer. The proteins were subsequently eluted using MBP-tag elution buffer containing 10 mm maltose. Both the elution and input samples were boiled in 1× SDS loading buffer at 100 °C for 10 min and analysed by SDS-PAGE. Immunoblotting was performed using MBP antibody (AE016; ABclonal; 1:5000 dilution) and His antibody (TA150088; ORIGENE; 1:5000 dilution) to detect the presence of the respective proteins.

For histone peptide pull-down assays, 2 μg of GST-fused protein was incubated with 1 μg of biotin-labeled histone peptides in 600 μL of binding buffer (50 mm Tris-HCl [pH 7.5], 150 mm NaCl, and 0.05% NP-40) for 4 h at 4 °C with rotation. Twenty-five microliters of Streptavidin MagneSphere Paramagnetic Particles (Promega, Z5481) were added to each tube and incubated for 2 h at 4 °C to capture histone peptides. A tube without addition of histone peptides served as a negative control. After the particles were washed with 1 mL binding buffer for 4 times, the proteins bound to the particles were dissolved in 1 × SDS sample buffer by boiling and then run on SDS-PAGE gel for immunoblotting. The GST-fused proteins without addition of histone peptides were used as negative controls. The histone peptides, including H3 (1 to 19 aa), H3K4/9/14/18ac (1 to 26 aa), H3K14ac (1 to 19 aa), H4K5/8/12/16ac (1 to 20 aa), H4K5ac (1 to 20 aa), H4K8ac (1 to 20 aa), H4K12ac (1 to 20 aa), and H4K16ac (1 to 20 aa) were commercially produced (Scilight-Peptide). Additional histone peptides used in this study include H3 (1 to 21 aa) (Millipore, 12-403), H4 (2 to 24 aa) (Millipore, 12-372), H4K5/8/12/16ac (1 to 18 aa) (Millipore, 12-379), H3K36ac (EpiCypher, 12-0021), and H3K4me3 (Millipore, 12-564).

ITC assay

The binding affinity of the bromodomain of GTE4 with histone peptides was assessed by a standard ITC assay. ITC binding curves were measured using a Microcal PEAQ-ITC instrument (Malvern). Purified proteins were dialyzed against ITC buffer containing 100 mm NaCl, 20 mm HEPES (pH 7.0), and 2 mm β-mercaptoethanol. The peptides were dissolved in the same buffer. Proteins were diluted to 0.04 mm with the ITC buffer, and 320 μL of the sample was added to the ITC instrument. Peptides were diluted to 0.6 mM, and 60 μL of the peptides was added to the ITC instrument. The histone peptides used in ITC including H3 (1 to 19 aa), H3K4/9/14/18ac (1 to 26 aa), H4 (1 to 20 aa), and H4K5/8/12/16ac (1 to 20 aa) were commercially produced (Scilight-Peptide). Titrations were performed at 20 °C. Data were analyzed using Origin 7.0.

Nuclear protein extraction and immunoblotting

A 0.6 g sample of 12-day-old seedlings grown on MS medium plates was collected and ground to a powder in liquid nitrogen. The sample was immediately resuspended in nucleus extraction buffer (20 mm Tris-HCl [pH 7.5], 2.5 mm MgCl₂, 20 mm KCl, 2 mm EDTA [pH 8.0], 250 mm sucrose, 25% glycerol, 1 mm DTT, 0.1 mm PMSF, and 1% cocktail, Roche) and gently rotated at 4 °C for 10 min. The homogenate was passed through 2 layers of Miracloth (Millipore, 475855), and the filtrate was centrifuged at 1500 × g for 10 min. The pellet was washed with Nuclear Resuspension Buffer with 0.2% Triton X-100 (NRBT) (20 mm Tris-HCl [pH 7.5], 2.5 mm MgCl₂, 25% glycerol, and 0.2% Triton X-100) until the pellet was no longer green, which typically took 4 washes. The pellet was then resuspended in nuclear extraction buffer. The sample was incubated in SDS loading buffer at 100 °C for 10 min, followed by immunoblotting for the detection of H4K5ac, H4K12ac, and H3K4me3. The H3 antibody (Abcam, ab1791, 1:5,000 dilution), H4K5ac antibody (Abcam, ab51997, 1:5,000 dilution), H4K12ac antibody (Abcam, ab46983, 1:5,000 dilution), and H3K4me3 antibody (Millipore, 07-473, 1:5,000 dilution) were used for immunoblotting.

Statistical analysis

The statistical analysis of phenotypic data was conducted using Student's t-test (2 tailed). The statistical analysis of 2nd-generation sequencing data was performed utilizing Mann–Whitney U test (2 tailed). Both tests were performed using R package. Detailed description of all statistical tests is provided in the figure legends, and the complete statistical data are available in Supplementary Data Set 11.

Accession numbers

Accession numbers of the genes used in this study are as follows: GTE4 (AT1G06230), GTE2 (AT5G10550), GTE3 (AT1G73150), GTE5 (AT1G17790), GTE7 (AT5G65630), GTE8 (AT3G27260), GTE9 (AT5G14270), GTE10 (AT5G63320), GTE11 (AT3G01770), EML1 (AT3G12140), EML2 (AT5G06780), and EML3 (AT5G13020). Raw RNA-seq and ChIP-seq data have been deposited in the Gene Expression Omnibus (GEO) database with the accession code GSE245273.

Author contributions

X.-J.H. conceived and supervised the project. F.Q., Q.-Q.Z., J.-X.Z., and X.-J.H. designed experiments, analysed the data, and wrote the manuscript. F.Q., Q.-Q.Z., J.-X.Z., and Z.-Z.L. performed experiments. D.-Y.Y. and Y.-N.S. performed bioinformatics analysis. L.L. and S.C. performed MS analysis.

Supplementary data

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

Supplementary Figure S1. Determining the binding ability of the GTE4 bromodomain with acetylated histones by histone peptide pull-down assays.

Supplementary Figure S2. The expression levels of wild-type and mutated versions of FLAG-tagged GTE4 transgenes.

Supplementary Figure S3. Validating the ability of EML1 to bind to the H3K4me3 peptide by a pull-down assay.

Supplementary Figure S4. Determining the interaction between the truncated versions of GTE4 and EML1 by yeast 2-hybrid assays.

Supplementary Figure S5. Structural prediction of GTE family proteins by AlphaFold.

Supplementary Figure S6. Determining the impact of the EID deletion on GTE4 function by complementation testing.

Supplementary Figure S7. Correlation among biological replicates of RNA-seq data for gte4, eml1/2, and gte4/eml1/2.

Supplementary Figure S8. Evaluation of ChIP-seq data replicability using principal component analysis.

Supplementary Figure S9. Determining the effect of gte4 and eml1/2 on histone modifications by immunoblotting.

Supplementary Figure S10. The effect of H4K5ac and H3K4me3 on the distribution of GTE4-bound genes at the whole-genome level.

Supplementary Figure S11. Determining the expression levels of the GTE4-FLAG and EML2-FLAG transgenes by immunoblotting.

Supplementary Figure S12. Analyzing the H4K5ac and H3K4me3 levels of GTE4 and EML2 target genes in the wild type.

Supplementary Data Set 1. Identification of acetylated histone–interacting proteins by AP-MS using histone H4K5/8/12/16ac peptides.

Supplementary Data Set 2. Identification of GTE4-, GTE4-ΔEID-, EML1-, and EML2-interacting proteins by AP-MS in Arabidopsis plants.

Supplementary Data Set 3. DEGs identified by RNA-seq in gte4, eml1/2, and gte4/eml1/2 relative to the wild type.

Supplementary Data Set 4. List of GTE4-enriched peaks and genes in the wild-type and eml1/2 backgrounds.

Supplementary Data Set 5. List of EML2-enriched peaks and genes in the wild-type and gte4 backgrounds.

Supplementary Data Set 6. H4K5ac and H3K4me3 levels of GTE4- and EML2-enriched genes.

Supplementary Data Set 7. The effect of eml1/2 on GTE4 enrichment at GTE4-bound peaks and the effect of gte4 on EML2 enrichment at EML2-bound peaks.

Supplementary Data Set 8. List of peaks and genes enriched with wild-type GTE4, GTE4-Y497A/N498A, and GTE4-ΔEID.

Supplementary Data Set 9. The effect of ΔEID and Y497A/N498A on GTE4 enrichment at GTE4-enriched peaks.

Supplementary Data Set 10. DNA oligos used in this study.

Supplementary Data Set 11. Results of statistical analysis.

Funding

This work was supported by the National Natural Science Foundation of China (32025003 and 32470628 to X.-J.H.).

Data availability

The raw RNA-seq and ChIP-seq data generated in this study are available in the Gene Expression Omnibus (GEO) database under the accession code GSE245273.

Dive Curated Terms

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

• GTE6 Gramene: AT3G52280

• GTE6 Araport: AT3G52280

• GTE1 Gramene: AT2G34900

• GTE1 Araport: AT2G34900

• GTE4 Gramene: AT1G06230

• GTE4 Araport: AT1G06230

• EML2 Gramene: AT5G06780

• EML2 Araport: AT5G06780

• EML3 Gramene: AT5G13020

• EML3 Araport: AT5G13020

• EML1 Gramene: AT3G12140

• EML1 Araport: AT3G12140

• GTE2 Gramene: AT5G10550

• GTE2 Araport: AT5G10550

• GTE3 Gramene: AT1G73150

• GTE3 Araport: AT1G73150

• GTE5 Gramene: AT1G17790

• GTE5 Araport: AT1G17790

• GTE7 Gramene: AT5G65630

• GTE7 Araport: AT5G65630

• GTE8 Gramene: AT3G27260

• GTE8 Araport: AT3G27260

• GTE9 Gramene: AT5G14270

• GTE9 Araport: AT5G14270

• GTE10 Gramene: AT5G63320

• GTE10 Araport: AT5G63320

• GTE11 Gramene: AT3G01770

• GTE11 Araport: AT3G01770