Histone crotonylation regulates neural stem cell fate decisions by activating bivalent promoters

Abstract

Histone lysine crotonylation (Kcr), an evolutionarily conserved and widespread non‐acetyl short‐chain lysine acylation, plays important roles in transcriptional regulation and disease processes. However, the genome‐wide distribution, dynamic changes, and associations with gene expression of histone Kcr during developmental processes are largely unknown. In this study, we find that histone Kcr is mainly located in active promoter regions, acts as an epigenetic hallmark of highly expressed genes, and regulates genes participating in metabolism and proliferation. Moreover, elevated histone Kcr activates bivalent promoters to stimulate gene expression in neural stem/progenitor cells (NSPCs) by increasing chromatin openness and recruitment of RNA polymerase II (RNAP2). Functionally, these activated genes contribute to transcriptome remodeling and promote neuronal differentiation. Overall, histone Kcr marks active promoters with high gene expression and modifies the local chromatin environment to allow gene activation.

Histone lysine crotonylation (Kcr) is an active epigenetic mark with diverse functions during neural development. Kcr regulates neural stem cell fate decisions by activating developmentally primed bivalent promoters.

Introduction

Histone post‐translational modifications (PTMs) occur on histone amino‐terminal tails protruding from nucleosomes and globular histone cores forming scaffolds around which DNA is wrapped and regulate processes such as transcription, DNA replication, DNA damage response, and higher‐order chromatin structure (Li et al, ²⁰¹⁸). Histone lysine acetylation (Kac) is one of the most well‐studied histone PTMs. It facilitates nucleosome eviction by disturbing interactions between histones and DNA and promotes the binding of RNA polymerase II (RNAP2) to proximal promoters via recognition of reader proteins (Zentner & Henikoff, 2013; Lawrence et al, ²⁰¹⁶; Barnes et al, ²⁰¹⁹). Several non‐acetyl short‐chain lysine acylations including lysine crotonylation (Kcr), propionylation (Kpr), butyrylation (Kbu), 2‐hydroxyisobutyrylation (Khib), β‐hydroxybutyrylation (Kbhb), succinylation (Ksuc), malonylation (Kmal), glutarylation (Kglu), benzoylation (Kbz), and lactylation (Kla) have recently been found on both histones and non‐histone proteins (Haws et al, ²⁰²⁰). Among these, histone Kcr has attracted great attention because of its crucial role in transcriptional regulation (Tan et al, ²⁰¹¹; Sabari et al, ²⁰¹⁵; Sabari et al, ²⁰¹⁷).

Histone Kcr, specifically enriched at promoters and potential enhancers in the mammalian genome, has a stronger effect on transcription than histone Kac (Tan et al, ²⁰¹¹; Sabari et al, ²⁰¹⁵). Recent studies have implicated histone Kcr in development and disease processes, including protection from acute kidney injury (Ruiz‐Andres et al, ²⁰¹⁶), self‐renewal of mouse embryonic stem cells (mESCs) (Wei et al, ^2017a), spermatogenesis (Liu et al, ^2017b), latency of human immunodeficiency virus (HIV) (Jiang et al, ²⁰¹⁸), stress‐induced depression (Liu et al, ²⁰¹⁹), and mesendodermal commitment of human embryonic stem cells (hESCs) (Fang et al, ²⁰²¹). Histone Kcr and Kac are believed to have the same transferases and reader modules. More specifically, lysine acetyltransferases (KATs) such as KAT2A (GCN5), KAT3 (CBP and p300), and KAT8 (MOF) can catalyze Kcr on both histones and non‐histone proteins (Sabari et al, ²⁰¹⁵; Liu et al, ^2017c). Among the known reader modules of histone Kac, the YEATS domain of Taf14, YEATS2, and AF9, and the DPF domain of MOZ and DPF2 display selective binding affinity for histone Kcr (Andrews et al, ²⁰¹⁶; Li et al, ²⁰¹⁶; Xiong et al, ²⁰¹⁶; Zhao et al, ²⁰¹⁶). Class I histone deacetylases (HDAC I) and SIRT1 are major histone decrotonylases in mammals (Wei et al, ^2017a; Fellows et al, ²⁰¹⁸; Kelly et al, ²⁰¹⁸). Furthermore, the availability of intracellular crotonyl‐CoA regulated by crotonate (Sabari et al, ²⁰¹⁵), positive regulators (ACOX1, ACOX3, ACADS, and ACSS2) (Sabari et al, ²⁰¹⁵; Fang et al, ²⁰²¹), and negative regulators (CDYL and ECHS1) (Liu et al, ^2017b; Tang et al, ²⁰²¹) directly influence histone Kcr levels. However, little is known about the distribution and dynamic changes of histone Kcr at the genome level during development, which limits our knowledge of histone acylation in epigenetic and developmental studies.

As a transcription activating modification, histone Kac plays an important role in neurodevelopment and neurological diseases (Falkenberg & Johnstone, 2014; Tang et al, ²⁰¹⁹). Interestingly, a recent study reported that CDYL‐mediated histone Kcr plays a critical role in regulating stress‐induced depression (Liu et al, ²⁰¹⁹). Although histone Kcr is observed and exists in combination with other histone PTMs in the adult mouse brain (Tweedie‐Cullen et al, ²⁰¹²; Liu et al, ^2017b; Fellows et al, ²⁰¹⁸), its biological functions and regulatory mechanisms in neurogenesis remain unknown. Therefore, it is necessary to explore the existence patterns, regulation pathways, and biological functions of histone Kcr in neural systems to deepen our understanding of non‐acetyl histone acylation during neural development.

Here, we first investigated the genome‐wide distribution and associations with the gene expression of histone Kcr. Next, we explored dynamic changes under metabolic stimulation and regulatory mechanisms for the gene expression of histone Kcr in NSPCs. Additionally, we performed an integrated analysis with single‐cell RNA sequencing data from the embryonic cortex to illustrate the biological functions of histone Kcr. Finally, we studied the association of histone Kcr with neural differentiation in vivo. Collectively, our results indicate that histone Kcr marks active promoters and activates bivalent promoters by increasing chromatin openness, which stimulates gene expression and regulates the fate of NSPCs by remodeling the transcriptome.

Results

Genome‐wide distribution and associations with gene expression of histone Kcr

We first performed chromatin immunoprecipitation‐sequencing (ChIP‐seq) assays using an antibody specifically recognizing H3K9cr in the E13.5 forebrain (Fig EV1A and B) and found that H3K9cr was significantly enriched at active promoters co‐marked by H3K4me3 and H3K27ac or H3K9ac with low DNA methylation levels (Fig 1A). In addition, H3K9cr marked genes with high expression levels (Fig 1B). K‐means clustering divided annotated genes into distinct chromatin environments by the H3K9cr signal (Fig 1C). Genes with high levels of H3K9cr were usually accompanied by high levels of chromatin accessibility and gene expression and low DNA methylation levels in the E13.5 forebrain (Fig 1C). To further understand the general patterns of histone Kcr at the genome level, we analyzed publicly available ChIP‐seq data on H3K18cr in cells (Sabari et al, ²⁰¹⁵; Fellows et al, ²⁰¹⁸). Our analysis showed unambiguously that H3K18cr was mainly located at active promoters and marked genes with high expression levels (Appendix Fig S1). Together, these results provide strong evidence that histone Kcr marks active promoters with high gene expression.

Figure EV1. Existence patterns of histone Kcr in vivo .

1. Dot blot assay showing that anti‐H3K9cr antibody is specific to H3K9cr, with information for peptides used on the right.
2. Western blotting analysis showing that H3K9cr exists in the E13.5 forebrain (n = 2 embryonic mice).
3. Dot blot assay showing that anti‐Kcr antibody is specific to Kcr, with information for peptides used on the right.
4. Western blotting analysis showing the existence of Kcr in NSPCs, N: nuclear proteins, and C: cytoplasmic proteins (n = 3 independent biological replicates). H3 and β‐Actin were used as loading controls for the nuclear and cytoplasmic proteins, respectively.
5. Immunostaining of Kcr on the mouse forebrain sections at E12, E16, P0, and P50 (n = 1 mouse for each developmental stage). Scale bar, 50 μm; Scale bar in magnified box, 10 μm.

Source data are available online for this figure.

Figure 1. Multi‐omics profiling of H3K9cr in the embryonic forebrain.

1. Characterization of genome‐wide distribution of H3K9cr in the E13.5 forebrain. The panel on the left displays a heat map of the emission parameters, in which each row corresponds to a different state, and each column corresponds to a different histone mark. The darker blue color corresponds to a greater probability of observing the mark in the state. The heat map to the right of the emission parameters displays the overlap fold enrichment for various genomic regions (CpG island regions, ATAC: ATAC‐seq peak regions, H3K9ac: H3K9ac peak regions, and H3K9cr: H3K9cr peak regions). A darker tan or red color corresponds to greater fold enrichment for a column‐specific coloring scale. DNA methylation levels over genomic regions of each state are shown as a bar graph on the right. The heat map on the lower right shows the fold enrichment for each state within 2 kb around a set of transcription start sites (TSSs). The black blue color corresponds to greater fold enrichment.
2. Boxplot showing gene expression levels among different groups in the E13.5 forebrain. A total of 22,075 genes with transcripts per million (TPM) greater than zero were grouped into quartiles according to expression levels (Quartile 1–4, each quartile containing 25% of genes). AP: active promoter, co‐marked by H3K4me3 and H3K27ac. There were 11,451, 9,689, and 9,204 genes under the regulation of H3K9ac, H3K9cr, and the active promoter, respectively.
3. Density heat maps in the E13.5 forebrain for H3K9cr ChIP‐seq and ATAC‐seq at TSSs ± 5 kb of genes with TPM greater than zero (scaled by reads per genomic content [RPGC]), which were divided into three clusters and ranked from the highest to the lowest by H3K9cr signal, with the number of genes in each cluster labeled on the left. The heat map to the right of the density heat maps displays DNA methylation levels (scaled by % CpG methylation), expression levels (scaled by z‐score normalized log₂ [TPM]), and length (scaled by z‐score normalized log₂ [gene length in kb]) of genes ranked by decreasing H3K9cr signal.

Functional annotation of histone Kcr reveals its diverse functions

To investigate the biological functions of histone Kcr, we focused on H3K9cr peaks in the promoter regions (Fig 2A and Appendix Fig S2A). In our analysis, we found that there were too many genes regulated by H3K9cr to precisely uncover its predominately involved biological processes (Fig 1B). Therefore, we divided H3K9cr peaks located in promoter regions into two clusters based on their modification levels to explore the biological functions of H3K9cr (Appendix Fig S2B). De novo motif analysis of H3K9cr peak regions revealed several transcription factors, such as TCF4 and SOX9, which are important regulators of neural development (Scott et al, ²⁰¹⁰; Forrest et al, ²⁰¹⁴), thus suggesting the potential neural functions of histone Kcr (Fig 2B). Encouragingly, several genes were co‐regulated by H3K9cr and TCF4, which are associated with stemness maintenance and neural differentiation (Fig 2C). Furthermore, we found that histone Kcr, including H3K9cr and H3K18cr, primarily regulated genes involved in nucleic acid metabolism, protein quality control, and cell proliferation (Fig 2D and Appendix Fig S2C–J), which highlighted the functional importance of histone Kcr in regulating basic cellular processes.

Figure 2. Functional annotation of H3K9cr in the embryonic forebrain.

1. Pie chart showing the distribution of 12,492 H3K9cr peaks in annotated genomic regions in the E13.5 forebrain.
2. Sequence logos corresponding to enriched elements identified by de novo motif analysis of two clusters of H3K9cr peaks located in promoter regions in the E13.5 forebrain.
3. Top panel: Venn diagram of a combined comparison of genes under the regulation of H3K9cr or annotated in TCF4 peak regions. Bottom panel: Representative terms (P‐adjust < 0.05) for KEGG pathway enrichment analysis of 452 genes that were co‐marked by H3K9cr and TCF4.
4. KEGG pathway enrichment analysis of genes under regulation of H3K9cr in the E13.5 forebrain.

Genome‐wide alterations of histone Kcr under metabolic stimulation

Next, we used a specific anti‐Kcr antibody to profile the dynamics of Kcr (Fig EV1C) (Tan et al, ²⁰¹¹; Lu et al, ²⁰¹⁸). Kcr was detected in both cytoplasmic and nuclear proteins of NSPCs (Fig EV1D) and was observed in Nestin⁺ NSPCs of the ventricular zone (VZ) and subventricular zone (SVZ) of the developing mouse brain (Fig EV1E). Histone Kcr levels could be influenced by crotonate (Fig 3A), inhibitors of HDAC1‐3 (MS‐275) (Fig 3B), and CBP/p300 (C646) in NSPCs (Fig 3C). We also observed increased histone Kcr levels in the embryonic forebrain at E13.5 when treated with MS‐275 but not crotonate (Fig EV2A and B).

Figure 3. Dynamic changes of histone Kcr in NSPCs.

• A, B
  Core histones were prepared from NSPCs treated with various concentrations of crotonate (A) and 1 μM MS‐275 (B) for 24 h and subjected to Western blotting analysis using anti‐Kcr antibody, and the experiments were repeated three times.
• C
  NSPCs were treated with 20 μM C646, 10 mM acetate (Ac), 20 μM C646 and 10 mM acetate (Ac+C646), 10 mM crotonate (Cr), and 20 μM C646 and 10 mM crotonate (Cr+C646) for 24 h, and then, core histones were extracted and subjected to Western blotting analysis using anti‐Kcr antibody; the experiments were repeated three times.
• D
  ChIP‐seq density heat maps in NSPCs for histone Kcr at ± 1 kb from the center of peak regions with differential histone Kcr after crotonate (left panel) and MS‐275 (right panel) treatment. There were 3,308 and 836 peak regions with increased histone Kcr after crotonate and MS‐275 treatment, respectively, and 1,075 and 2,942 peak regions with reduced histone Kcr after crotonate and MS‐275 treatment, respectively.
• E
  Heat maps displaying overlap enrichment for various genomic regions (Up and Down: peak regions with increased and reduced histone Kcr in NSPCs after crotonate or MS‐275 treatment). A darker color in blue, red, or purple corresponds to greater fold enrichment for a column‐specific coloring scale.
• F
  GO analysis (biological process) of 2,442 genes with increased histone Kcr (left panel) and 726 genes with reduced histone Kcr (right panel) after crotonate treatment in NSPCs.

Source data are available online for this figure.

Figure EV2. Distribution and changes of histone Kcr in NSPCs.

1. MS‐275 affects histone Kcr levels in the embryonic forebrain. Top panel: Western blotting analysis indicating that histone Kcr levels are affected by MS‐275 abundance in the embryonic forebrain (n = 3 embryonic mice). In the experiment, 0.1% DMSO was intraperitoneally injected as a control, and 20 mg/kg MS‐275 was injected every 24 h for four continuous days on pregnant mice on day 9.5 of pregnancy. Bottom panel: Western blotting analysis indicating that crotonate cannot induce changes in histone Kcr levels in the embryonic forebrain (n = 3 embryonic mice). In the experiment, PBS was intraperitoneally injected as a control, and 500 μM/30 g crotonate was injected every 24 h for four continuous days on pregnant mice at day 9.5 of pregnancy.
2. Immunostaining results showing that MS‐275 enhances Kcr levels in VZ/SVZ at E13.5 stage (n = 1 embryonic mouse). Scale bar, 50 μm.
3. Boxplot showing gene expression levels among different groups in NSPCs under normal conditions. A total of 24,579 genes with TPM greater than zero were grouped into quartiles according to expression levels (Quartile 1–4, each quartile containing 25% of genes). AP: active promoter, co‐marked by H3K4me3 and H3K27ac. There were 13,506, 8,317, and 11,490 genes under the regulation of H3K9cr, histone Kcr, and active promoter, respectively.
4. The panel on the left displays a heat map of the emission parameters in which each row corresponds to a different state, and each column corresponds to a different mark in NSPCs under normal conditions. The darker blue color corresponds to a greater probability of observing the mark in the state. The heat map to the right of the emission parameters displays the overlap fold enrichment for various genomic regions (Up and Down: peak regions with increased and decreased H3K9cr after crotonate treatment, respectively). A darker tomato or blue4 color corresponds to greater fold enrichment for a column‐specific coloring scale.
5. NSPCs were treated with 10 mM crotonate (left panel) or 1 μM MS‐275 (right panel) for different time intervals, and then core histones were prepared and subjected to Western blotting analysis using anti‐Kcr and anti‐Kac antibodies. The time gradient experiments of the crotonate treatment were repeated twice, and the time gradient experiments of MS‐275 treatment were repeated once.
6. ChIP‐seq density heat maps in NSPCs for histone Kac at ± 1 kb from the center of peak regions with differential histone Kcr after crotonate (left panel) and MS‐275 (right panel) treatment, respectively.

Source data are available online for this figure.

To profile dynamic changes in histone Kcr at the genome level, we performed anti‐Kcr and anti‐H3K9cr ChIP‐seq assays in NSPCs and found that histone Kcr marked genes with high gene expression under both control (Fig EV2C) and crotonate‐treated conditions (Appendix Fig S3A). We then conducted quantitative comparisons of ChIP‐seq data to assess alterations in histone Kcr levels (Fig 3D). Interestingly, peaks with increased histone Kcr after crotonate or MS‐275 treatment were mainly located in bivalent promoter regions (Figs 3E and EV2D), the crucial cis‐regulatory elements involved in transcriptional regulation of developmentally primed and cell fate‐determining genes (Voigt et al, ²⁰¹³). Peaks with reduced histone Kcr were annotated in the proximal active promoter and enhancer regions (Fig 3E). Importantly, crotonate treatment was specific to changes in histone Kcr at both global and genomic levels compared with that of histone Kac (Fig EV2E and F).

Gene ontology (GO) enrichment analysis of genes annotated in the peak regions with differential histone Kcr indicated that they predominately regulated neuronal differentiation and cell proliferation (Fig 3F). Additionally, there was a high overlap of differentially expressed genes between crotonate‐treated and MS‐275‐treated groups (Appendix Fig S3B), and there were also several genes with increased histone Kcr both during neural development and after crotonate treatment in NSPCs, which were associated with neuronal differentiation (Appendix Fig S3C and D) (preprint: Dai et al, ²⁰²¹). We also found upregulation of genes encoding positive regulators of crotonyl‐CoA levels (Acss2, Acox1, and Acox3) and a sharp decline in the expression of genes encoding negative regulators of crotonyl‐CoA levels (Cdyl and Echs1) during neuronal differentiation processes (Appendix Fig S3E and F). These results suggest that the dynamic changes in histone Kcr and gene expression after crotonate treatment were physiologically relevant and tightly associated with NSPC fate decisions. Collectively, we propose that histone Kcr may remodel the local chromatin environment through its dynamic changes under metabolic stimulation.

Histone Kcr activates bivalent promoters to stimulate gene expression

To explore changes in bivalent mark status influenced by histone Kcr, we performed anti‐H3K4me3, anti‐H3K27me3, and anti‐H3K27ac ChIP‐seq assays. With increased histone Kcr, we observed significant loss of repressive H3K27me3 mark, slight increase in activating H3K4me3 mark, and less pronounced changes of H3K27ac in bivalent promoter regions (Fig 4A). Consistent results for changes in bivalent marks were obtained by CUT&Tag assays (Appendix Fig S4A and B), with high sensitivity and resolution compared with ChIP‐seq assays (Kaya‐Okur et al, ²⁰¹⁹). Importantly, genes under the regulation of these bivalent promoters were significantly upregulated and participated in neuronal fate decisions after crotonate treatment in NSPCs (Fig 4B and C). These results suggest that activation of bivalent promoters is accompanied by a robust increase in histone Kcr levels.

Figure 4. Histone Kcr activates bivalent promoters to stimulate gene expression.

1. ChIP‐seq density heat maps in NSPCs for histone Kcr, H3K27me3, H3K4me3, and H3K27ac at ± 1 kb (histone Kcr), ± 2.5 kb (H3K4me3 and H3K27ac), and ± 5 kb (H3K27me3) from the center of 1,447 peak regions located in bivalent promoters with increased histone Kcr after crotonate treatment.
2. Boxplot showing expression changes of 1,101 genes that were annotated in the 1,447 peak regions defined in (A). Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; ***P < 0.001.
3. GO analysis (biological process) of the 1,101 genes defined in (B).
4. BETA plot of combined computational analysis of histone Kcr ChIP‐seq and RNA‐seq data, peak regions located in bivalent promoters with increased histone Kcr as input.
5. Quantification analysis of changes in histone Kcr levels in NSPCs treated with DMSO (Control), 5 μM triptolide (FP), 5 μM flavopiridol and 10 mM crotonate (FP+CR), and 10 mM crotonate (CR) for 1 h. Data are presented as mean ± SEM of three independent biological replicates, n = 3. One‐way ANOVA with post hoc Tukey’s test was used to analyze statistical significance; NS: no significance, **P < 0.01, ***P < 0.001.
6. Co‐IP analysis indicating RNAP2 binding in H3K9cr enriched regions in NSPCs. H3 was used as a loading control for the input nuclear proteins.
7. Quantification analysis of changes in chromatin openness at Tgfb1 and Notum promoter regions by FAIRE‐qPCR among different groups of NSPCs. Data are presented as mean ± SEM of three independent biological replicates, n = 3. Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; *P < 0.05, **P < 0.01.

Source data are available online for this figure.

Next, we employed binding and expression target analysis (BETA) (Wang et al, ²⁰¹³), a method for integrated analysis of ChIP‐seq and RNA‐seq data, which models the influence of a binding site on the expression of a gene with a monotonically decreasing function that is based on the distance between the binding site and transcription start site instead of assigning one‐to‐one mapping between binding sites and genes, to detect associations between changes in histone Kcr and gene expression. We found that peaks with increased histone Kcr at both bivalent promoter and non‐bivalent promoter regions had significant effects as a gene activator (Fig 4D and Appendix Fig S4C), while peaks with reduced histone Kcr had significant effects as a gene repressor after crotonate treatment in NSPCs (Appendix Fig S4D). Importantly, there were positive correlations between changes in histone Kcr and gene expression at bivalent genes (Fig EV3A) and all genes (Fig EV3B) after crotonate treatment in NSPCs. In addition, histone Kcr was causal to changes in gene expression because transcription inhibition had no effect on the increase in histone Kcr (Figs 4E and EV3C and D). It is worth noting that RNAP2 binding was also observed in regions with H3K9cr enrichment (Fig 4F), suggesting the involvement of histone Kcr in the recruitment of transcription machinery.

Figure EV3. Crotonate stimulates gene expression via histone Kcr.

• A, B
  Pearson correlation for changes in histone Kcr and gene expression of 1,101 genes under regulation of bivalent promoters with increased histone Kcr (left panel) and all annotated genes (right panel).
• C
  Blots showing changes in RNAP2, RNAP2 serine 5 CTD phosphorylation (S5p), and RNAP2 serine 2 CTD phosphorylation (S2p) levels in NSPCs treated with DMSO, 5 μM triptolide (TP), 5 μM flavopiridol (FP), and 25 μg/mL actinomycin D (AD) for 1 h. In this experiment, triptolide (S3604, Selleck Chemicals, CAS: 38748‐32‐2), flavopiridol HCl (S2679, Selleck Chemicals, CAS: 131740‐09‐5), and actinomycin D (S8964, Selleck Chemicals, CAS: 50‐76‐0) are widely used compounds that inhibit transcription inhibition, transcription elongation, and whole transcription processes, respectively. Flavopiridol was selected for all studies because it exhibited the strongest inhibitory effect on transcription, which was indicated by the loss of S5p (a mark of transcription initiation) and S2p (a mark of transcription elongation).
• D
  Blots showing changes in histone Kcr levels among different groups of NSPCs. FP: 5 μM flavopiridol, FP+CR: 5 μM flavopiridol and 10 mM crotonate, and CR: 10 mM crotonate.
• E
  Genome‐browser view at Tgfb1 and Notum genes indicating that they are under the regulation of bivalent promoters, with annotation of chromatin state on the top. Changes of histone Kcr at Tgfb1 and Notum promoter regions are 2.12 and 2.70 folds, respectively, and changes of gene expression for Tgfb1 and Notum are 7.75 and 7.83 folds after crotonate treatment in NSPCs, respectively.
• F, G
  Quantification analysis of changes in histone Kcr at Tgfb1 (F) and Notum (G) promoter regions by ChIP‐qPCR among different groups of NSPCs. Data are presented as mean ± SEM of three independent biological replicates, n = 3. Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; *P < 0.05, ***P < 0.001.
• H, I
  Quantification analysis of expression changes of Tgfb1 (H) and Notum (I) by qRT‐PCR among different groups of NSPCs. Data are presented as mean ± SEM of three independent biological replicates, n = 3. Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; **P < 0.01.
• J, K
  Quantification analysis of changes in histone Kac at Tgfb1 (J) and Notum (K) promoter regions by ChIP‐qPCR among different groups of NSPCs. Data are presented as mean ± SEM of three independent biological replicates, n = 3. Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; NS: no significance, *P < 0.05, **P < 0.01.
• L, M
  Quantification analysis of changes in histone Kcr at Tgfb1 (L) and Notum (M) promoter regions by ChIP‐qPCR among different groups of NSPCs. Data are presented as mean ± SEM of three independent biological replicates, n = 3. Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; NS: no significance, **P < 0.01.

Source data are available online for this figure.

Tgfb1 and Notum were genes under the regulation of bivalent promoters in NSPCs, and levels of both histone Kcr and expression of these two genes were significantly increased after crotonate treatment (Fig EV3E). Therefore, we selected these two genes to validate the sequencing data and elucidate the mechanisms underlying histone Kcr‐mediated activation of bivalent promoters. ChIP‐qPCR and formaldehyde‐assisted isolation of regulatory elements followed by qPCR analysis (FAIRE‐qPCR) data indicated that crotonate significantly increased histone Kcr and chromatin openness in their promoter regions (Figs 4G and EV3F and G), accompanied by significant upregulation of gene expression (Fig EV3H and I). In contrast, minor changes in histone Kac were observed in the promoter regions of Tgfb1 and Notum after crotonate treatment compared with that of MS‐275 treatment (Fig EV3J and K). Interestingly, the ability of histone Kcr to stimulate gene expression was comparable to that of histone Kac after the combined comparison of different treatments (Fig EV3L and M, and Appendix Fig S4E and F). Moreover, there was a quicker and more robust response to crotonate for histone Kcr than to H3K4me3, H3K27me3, and histone Kac (Fig EV2E and Appendix Fig S4G), which suggested that histone Kcr might be causal to changes in bivalent marks. Overall, these results provide evidence that histone Kcr activates bivalent promoters to stimulate gene expression by modulating chromatin openness and recruitment of the transcriptional machinery.

Histone Kcr regulates NSPCs fate via remodeling transcriptome

To further explore the cellular functions of histone Kcr, we examined the effects of crotonate on cell proliferation and differentiation. By using bromodeoxyuridine (BrdU) pulse‐labeling, we found that less BrdU was incorporated in crotonate‐treated NSPCs than in the control group (Fig 5A and B). After three days of differentiation, crotonate‐treated NSPCs differentiated into more TUJ1⁺ neurons than those in the control group (Fig 5C and D). Consistent with phenotypic changes, upregulated and downregulated genes were associated with neuronal differentiation and cell proliferation after crotonate treatment in NSPCs, respectively (Fig EV4A and B).

Figure 5. Crotonate alters NSPCs fate via transcriptome remodeling.

• A, B
  Representative images (A) and quantification analysis (B) of changes in cell proliferation ability measured by BrdU incorporation rate among different groups of NSPCs. Scale bar, 50 μm. Data are presented as mean ± SEM of three independent biological replicates, n = 3. Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; **P < 0.01.
• C, D
  Representative images (C) and quantification analysis (D) of changes in neuronal differentiation potential defined by TUJ1 positivity among different groups of NSPCs. Scale bar, 20 μm. Data are presented as mean ± SEM of three independent biological replicates, n = 4. Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; ***P < 0.001.
• E, F
  Odds‐ratio analysis of overlapping genes displaying different expression (DEx) versus representing changes in cell proliferation ability (E) and neuronal differentiation potential (F) in NSPCs. Np: Non‐proliferative cells, and P: proliferative cells. Up and Down: upregulated and downregulated genes after crotonate treatment in NSPCs, respectively. Insert numbers indicate respective P‐values for associations, with the number of genes overlapping in parentheses, and two‐sided Fisher's exact test was used to analyze statistical significance.
• G
  Genome‐browser view at the Mir‐203 gene of different sequencing datasets, with annotation of chromatin state at the top.
• H
  Quantification analysis of changes in histone Kcr at Mir‐203 proximal regions by ChIP‐qPCR among different groups of NSPCs. The amplified genomic regions of two pairs of primers (P1 and P2) are shown in (G). Data are presented as mean ± SEM of three independent biological replicates, n = 3. Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; *P < 0.05, **P < 0.01.
• I, J
  Quantification analysis of expression changes of Mir‐203 (I) and Bmi1 (J) by qRT‐PCR among different groups of NSPCs. Data are presented as mean ± SEM of three independent biological replicates, n = 3. Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; *P < 0.05, **P < 0.01.
• K
  Representative blots (top panel) and quantification analysis (bottom panel) of changes in BMI1 protein levels among different groups of NSPCs. Data are presented as mean ± SEM of three independent biological replicates, n = 3. Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; *P < 0.05.
• L
  Venn diagram of a combined comparison of different gene sets.
• M, N
  GO analysis (biological processes) of overlapping genes in (L) between red circle with blue circle (M), and red circle with purplish red circle (N).

Source data are available online for this figure.

Figure EV4. Transcriptome remodeling in crotonate‐treated NSPCs.

• A, B
  KEGG pathway enrichment analysis of 2872 upregulated genes (A) and 585 downregulated genes (B) after crotonate treatment in NSPCs.
• C, D
  Venn diagram of a combined comparison of different gene sets. In (C), Np vs P_up and Np vs P_down: upregulated and downregulated genes when differential expression analysis was performed between Np (non‐proliferative cells) and P (proliferative cells), respectively. In (D), Neuron vs NSPC_up and Neuron vs NSPC_down: upregulated and downregulated genes when differential expression analysis was performed between Neuron and NSPC, respectively. RNA‐up and RNA‐down: upregulated and downregulated genes after crotonate treatment in NSPCs, respectively.
• E, F
  Boxplot showing changes in gene expression (E) and histone Kcr (F) among different groups of genes. After crotonate treatment in NSPCs, 2,872 upregulated genes were grouped into quartiles by the degree of changes in gene expression (Quartile 1–4, each quartile containing 25% of genes). Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; *P < 0.05, ***P < 0.001.

To link transcriptome changes with cell fate decisions, we integrated our bulk RNA sequencing data with published single‐cell RNA sequencing data for embryonic neurogenesis (Loo et al, ²⁰¹⁹). A total of 22 cell clusters were identified in the E14.5 cortex (Appendix Fig S5A and B). Specific cell markers were used to distinguish radial glial cells, progenitors, neurons, proliferative cells, and non‐proliferative cells (Appendix Fig S5C and D). We found that downregulated genes were predominately associated with the category that regulated NSPC proliferation and stemness (Figs 5E and F, and EV4C). On the contrary, upregulated genes were mainly related to the category that promoted the differentiation of NSPCs into neurons (Figs 5F and EV4D).

Interestingly, miR‐203, a microRNA that was previously found to inhibit proliferation via Bmi1 repression in NSPCs (Liu et al, ^2017a), was under the regulation of bivalent promoters and displayed increased histone Kcr levels at its proximal promoter regions (Fig 5G and H), and upregulation of expression after crotonate treatment in NSPCs (Fig 5I, Appendix Fig S6A and B). Consistent with our previous findings, reduction in both mRNA and protein levels of Bmi1 was observed concordant with upregulation of miR‐203 expression (Fig 5J and K). In contrast, we found sharp declines in histone Kcr in promoter regions of mitosis‐related genes and histone genes that are important for the S phase of the cell cycle (Appendix Fig S6C and D).

Upregulated genes with higher degrees of expression changes were also accompanied by higher levels of changes in histone Kcr (Fig EV4E and F), and promoters of upregulated genes were significantly enriched in bivalent promoter regions, which was consistent with changes in histone Kcr (Fig 3E and Appendix Fig S6E). Furthermore, GO enrichment analysis of the overlap between genes annotated in bivalent promoter regions with increased histone Kcr and genes related to neural cell fate transitions or encoding transcription factors and cofactors (Fig 5L), revealed that increased histone Kcr‐activated neuronal differentiation‐related regulatory networks after crotonate treatment in NSPCs (Fig 5M and N). Taken together, our molecular and functional characterization of histone Kcr suggests that activation of bivalent promoters via histone Kcr may be key regulatory mechanisms for NSPC fate decisions.

Elevated histone Kcr is accompanied by neuronal differentiation in vivo

We next employed brain explant culture to explore tissue‐level changes of histone Kcr and transcriptome. E13.5 brain explants were separated and cultured for 3 h under a series of concentration gradients of crotonate. We found that crotonate dramatically increased H3K9cr and histone Kcr levels (Fig 6A–D). Consistent with the results of NSPCs, transcriptome profiling indicated cell fate changes in crotonated‐treated explants, in that upregulated and downregulated genes were associated with neuronal differentiation and cell proliferation, respectively (Fig 6E and F).

Figure 6. Regulation and dynamic changes of histone Kcr in vivo .

• A, B
  Core histones were prepared from E13.5 forebrain explants treated with various concentrations of crotonate for 3 h and subjected to Western blotting analysis using anti‐H3K9cr (A) and anti‐Kcr (B) antibodies. Naïve: isolated forebrains without subsequent culture and treatment.
• C, D
  Quantification analysis of changes in H3K9cr (C) and H3‐Kcr (D) levels by Western blotting among different groups of explants. Data are presented as mean ± SEM of three independent biological replicates, n = 3. Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; NS: no significance, **P < 0.01, ***P < 0.001.
• E, F
  GO analysis (biological process) of genes with increased (E) and decreased (F) expression levels after crotonate treatment in E13.5 forebrain explants.
• G, H
  Quantitative analysis of changes in H3K9cr (G, left panel of H) and H3K14la (right panel of H) levels by Western blotting among different groups of the embryonic forebrain. Data are presented as mean ± SEM of three independent biological replicates, n = 3. Two‐tailed unpaired Student’s t‐test was used to analyze statistical significance; NS: no significance, *P < 0.05, ***P < 0.001.
• I
  Quantitative analysis of changes in H3K9cr levels in the developing forebrain. Data are presented as mean ± SEM of three independent biological replicates, n = 3. One‐way ANOVA with post hoc Tukey’s test was used to analyze statistical significance; NS: no significance, *P < 0.05, **P < 0.01.

Source data are available online for this figure.

We also examined the regulation of histone Kcr in vivo. MS‐275 treatment increased both H3K9cr and H3K18cr levels in the E13.5 forebrain, while ACSS2 inhibition was not able to reduce histone Kcr levels in vivo (Figs 6G and EV5A). However, interfering with the tricarboxylic acid cycle by inhibiting mitochondrial pyruvate transport 1 (MPC1) had no obvious effect on histone Kcr levels, but enhanced histone Kla levels (H3K14la), which was recently reported to be sensitive to cytoplasmic pyruvate levels (Figs 6H and EV5B). Interestingly, there were increasing trends of H3K9cr and H3K18cr levels during neurogenesis stages (Figs 6I and EV5C and D), highlighting their potential roles in neural differentiation. Together, these results demonstrate a strong association between histone Kcr and neuronal differentiation in vivo.

Figure EV5. Associations of histone Kcr with neuronal differentiation in vivo .

• A, B
  Western blotting analysis showing changes in H3K9cr, H3K18cr, and H3K14la levels in vivo. In the experiment, 10 mg/kg MS‐275 (A), 5 mg/kg ACSS2 inhibitor (S8588, Selleck Chemicals, CAS: 508186‐14‐9) (A), and 5 mg/kg UK 5099 (S5317, Selleck Chemicals, CAS: 56396‐35‐1) (B) were injected every 12 h for continuous two days (total four times) on pregnant mice at day 11.5 of pregnancy. The core histones of the E13.5 forebrain were extracted and subjected to Western blotting analysis.
• C
  Western blotting analysis showing changes in TUJ1, H3K9cr, and H3K18cr levels among different development stages of the forebrain.
• D
  Quantitative analysis of changes in H3K18cr levels in the developing forebrain. Data are presented as mean ± SEM of three independent biological replicates, n = 3. One‐way ANOVA with post hoc Tukey’s test was used to analyze statistical significance, and NS: no significance.
• E
  Dynamic expression changes of Hdac1‐3 and Sirt1 in the developing forebrain.

Source data are available online for this figure.

Discussion

Although histone Kcr has been previously shown to localize at promoters and potential enhancers of active genes (Tan et al, ²⁰¹¹), detailed distribution patterns and dynamic changes of histone Kcr during neurogenesis at the epigenetic level remain largely unknown. Here, we provide comprehensive investigations of genome‐wide distribution and dynamic changes in histone Kcr, as well as its associations with gene expression, and explored potential roles of histone Kcr as key regulatory mechanisms underlying gene activation and disturbed NSPC fate. Under normal conditions, histone Kcr marks active promoters with high levels of H3K4me3, H3K27ac, and chromatin openness. The highly expressed genes under the regulation of histone Kcr are involved in nucleic acid, protein metabolism, and cell proliferation (Fig 7). However, crotonate‐derived crotonyl‐CoA may break the balance of histone crotonylation and de‐crotonylation in the local chromatin environment, especially bivalent promoters, to stimulate gene expression by facilitating CBP/p300‐mediated deposition of histone Kcr in NSPCs. It is also accompanied by enhanced chromatin openness and RNAP2 binding. Contributing to transcriptome remodeling, these activated genes promote the transition of NSPCs from the stemness state to a state that tends to favor neuronal differentiation (Fig 7). For the first time, we uncovered the metabolic‐epigenetic roles of histone Kcr in NSPCs.

Figure 7.

Working models for histone Kcr to regulate neural stem cell fate decisions.

How histone Kcr regulates transcription remains to be determined. In most previous studies, histone Kcr has been considered as a stimulator for gene expression, which is based on the consistency between changes in histone Kcr and gene expression under certain circumstances, such as HDAC1‐VRPP overexpression, CDYL deletion, and AF9 mutation (Li et al, ²⁰¹⁶; Wei et al, ^2017a; Liu et al, ^2017b). However, two recent studies have revealed the role of histone Kcr in transcriptional repression. Notably, accompanied by an increase in histone Kcr, downregulated genes display simultaneous reduction in histone Kac during the metabolic cycle of yeast and in NEAT‐knockdown cells (Gowans et al, ²⁰¹⁹; Wang et al, ²⁰¹⁹). Therefore, the balance between opposite changes in histone Kcr and Kac makes it difficult and ambiguous to determine whether histone Kcr directly represses transcription. In general, increased histone Kcr stimulates gene expression via loss of local H3K27me3 and enhanced recognition by reader modules, such as AF9, which can recruit the transcriptional machinery (Li et al, ²⁰¹⁶; Liu et al, ²⁰¹⁹). Here, we provide evidence supporting the important role of histone Kcr in activating silenced bivalent promoters by enhancing chromatin openness and binding RNAP2 in NSPCs. Importantly, we demonstrated that histone Kcr is not influenced by transcriptional inhibition. These findings lead us to propose that histone Kcr may play a causative, not just correlative, role for activation of bivalent promoters and gene expression, which is consistent with the chemical properties of histone Kcr such as altering strong electrostatic interactions between oppositely charged nucleosome DNA (Onufriev & Schiessel, 2019).

Metabolism plays an important role in cell fate decisions through epigenetic regulation, especially during neurogenesis (Xie & Sheppard, 2018). Acetate, a precursor of acetyl‐CoA, delays the differentiation of ESCs and blocks early histone deacetylation in a dose‐dependent manner (Moussaieff et al, ²⁰¹⁵). Crotonate is converted to crotonyl‐CoA through ACSS2, which acts as a donor for crotonylation of histones and non‐histone proteins, and increases histone Kcr levels by promoting the recruitment of histone crotonyltransferases (Sabari et al, ²⁰¹⁵; Wei et al, ^2017b; Kollenstart et al, ²⁰¹⁹). By integrating published single‐cell sequencing data of embryonic neurogenesis, we found that crotonate promoted transcriptome remodeling, which favors a switch from a stemness state to a neuron‐like state via histone Kcr‐mediated activation of bivalent promoters (Fig 7). Although crotonate has been reported to inhibit the proliferation of cancer cells and influence histone Kcr levels in the cell cycle, the mechanisms by which histone Kcr regulates cell proliferation remain unclear (Wei et al, ^2017b; Fellows et al, ²⁰¹⁸). We propose that the histone Kcr‐miR‐203‐Bmi1 regulatory axis may play critical roles in regulating the proliferation of NSPCs (Fig 5G–K). In addition, the sharp decline of histone Kcr at coding regions of histone genes and genes involved in the progression of the M phase (Appendix Fig S6C and D) provides alternative ways for histone Kcr to regulate proliferation. In this scenario, decrease in histone Kcr may act as “brakes” for expression of genes that are important for cell cycle progression in response to signals that disfavor proliferation. Crotonyl‐CoA metabolism may be tightly associated with NSPC fate during neural development, because our results regarding the functions of histone Kcr in crotonate‐treated NSPCs are highly consistent with our most recent study, which indicates that histone Kcr is positively correlated with neuronal differentiation in vivo (preprint: Dai et al, ²⁰²¹).

Hyperacetylation of histones has been reported to promote neuronal differentiation. TSA, a pan‐HDAC inhibitor, increases the expression of marker genes in neurons in E15.5 brain explants (Večeřa et al, ²⁰¹⁸). Similarly, NSPCs in the embryonic forebrain with deletion of HDAC1, 2, or 3 are prone to differentiate into neurons (Li et al, ^2019a; Tang et al, ²⁰¹⁹), while CBP knockdown in cultured NSPCs inhibits their neuronal differentiation (Wang et al, ²⁰¹⁰). However, inhibition or deletion of HDACs and knockdown of CBP influence both histone Kac and Kcr simultaneously (Fig EV2E) (Sabari et al, ²⁰¹⁵; Wei et al, ^2017a). There were opposite trends between changes in H3K9cr levels and HDAC expression in the developing forebrain (Figs 6I and EV5E) (Večeřa et al, ²⁰¹⁸; Tang et al, ²⁰¹⁹; preprint: Dai et al, ²⁰²¹). Considering our results that histone hypercrotonylation promoted neuronal differentiation of NSPCs by activating bivalent promoters (Fig 7), we propose that HDACs limit both histone Kac and Kcr at neuronal differentiation‐related genes during neural development. Therefore, any interference breaking the balance of histone acylation and de‐acylation at these genes would re‐write their epigenetic state and pre‐activate their expression (Wang et al, ²⁰⁰⁹), which finally influences NSPC fate decisions and may cause severe neurodevelopmental diseases.

Additionally, we employed anti‐Kcr antibody to profile dynamic changes in histone Kcr by ChIP‐seq, as in other studies (Tan et al, ²⁰¹¹; Liu et al, ^{2017b, 2019}; Lu et al, ²⁰¹⁸; Fang et al, ²⁰²¹). Although we also performed anti‐H3K9cr ChIP‐seq assays to validate the results obtained with the anti‐Kcr antibody (Fig EV2C and D, Appendix Fig S3A), the potential recognition of crotonylated non‐histone proteins cannot be excluded for this antibody. Overall, we provided genome‐wide dissection of histone Kcr in the neural system and identify unique roles of this histone mark in the activation of bivalent promoters and NSPC fate decisions. Our findings provide a deeper understanding of epigenetic regulation mediated by histone Kcr and offer new perspectives on the underlying mechanisms of neural development and neurological diseases.

Materials and Methods

Animals

C57BL/6 and ICR mice were purchased from SiPeiFu Biotechnology Co., Ltd. (Beijing, China). All mouse experiments were approved by the Animal Committee of the Institute of Zoology, Chinese Academy of Sciences, Beijing, China.

Cell and brain explants culture

Neural stem/progenitor cells isolated from the embryonic forebrain at E13.5 were cultured in proliferation medium (DMEM/F12 medium (Gibco) supplemented with 20 ng/ml epidermal growth factor and fibroblast growth factor (EGF/FGF, PeproTech), 0.5 × B27 (Gibco), 0.5 × N2 (Gibco), and 1% penicillin‐streptomycin (Gibco)). For passaging, neurospheres were dissociated into single cells and sub‐cultured. Explanted E13.5 mouse brains were cultured in DMEM/F12 medium supplemented with 10% FBS (Gibco) and 1% penicillin‐streptomycin, and treated with 0, 10, 20, or 50 mM crotonate for 3 h. Isolated forebrains were subjected to Western blotting and RNA‐seq analysis.

Analysis of NSPC proliferation and differentiation

Crotonate (113018, Sigma‐Aldrich, CAS: 107‐93‐7) dissolved in distilled water was adjusted to pH 7.4 using sodium hydroxide and MS‐275 (S1053, Selleck Chemicals, CAS: 209783‐80‐2) was dissolved in dimethyl sulfoxide (DMSO, Sigma‐Aldrich). 10 mM crotonate or 1 μM MS‐275 was added to the culture medium. To study cell proliferation, NSPCs were dissociated into single cells and plated on poly‐L‐ornithine/laminin (PLL)‐coated coverslips at a density of 3 × 10⁴ cells/well in proliferation medium, and BrdU (Sigma‐Aldrich) was added to the culture medium at a final concentration of 10 μM for 6 h after 18‐h crotonate or MS‐275 treatment. NSPCs were seeded at a density of 5 × 10⁴ cells/well to study cell differentiation. After 24‐h treatment with 10 mM crotonate or 1 μM MS‐275 in proliferation medium, NSPC differentiation was induced by changing to Neurobasal medium (Gibco) supplemented with 1% penicillin‐streptomycin, 1 × B27, and 1 × GlutaMAX Supplement (Gibco), 5 μM forskolin (FSK, Sigma‐Aldrich), and 1 μM retinoic acid (RA, Sigma‐Aldrich) for 3 days.

Immunocytochemistry and immunostaining

DAPI (Sigma‐Aldrich) was used to label cell nuclei. For immunocytochemistry analysis, cells on coverslips were fixed in 4% formaldehyde for 15 min at room temperature and then washed three times with phosphate‐buffered saline (PBS). To detect BrdU incorporation, fixed cells were pre‐treated with 1 M HCl for 30 min at 37°C, and then washed sequentially twice with 0.1 M borate buffer (pH 8.5) and three times with PBS. Next, cells were incubated with blocking solution (2% bovine serum albumin, 0.3% Triton X‐100, and 0.1% sodium azide) at room temperature for 1 h. Coverslips were then incubated with primary antibodies diluted in blocking solution at 4°C overnight and labeled using appropriate secondary antibodies at room temperature for 2 h. For immunostaining, embryonic and adult mouse brains were cut into 10 and 40 μm thick cryosections, respectively. Staining procedures for brain slices were similar to those used for immunocytochemistry analysis. Detailed information on the primary and secondary antibodies used for immunocytochemistry and immunostaining analysis is summarized in Table EV1. All images were acquired using confocal microscopy (Zeiss, LSM710).

Western blotting and co‐immunoprecipitation

Cytoplasmic and nuclear proteins were sequentially acquired using cytoplasmic lysis buffer (5 mM PIPES, pH 8.0, 85 mM KCl, 0.5% NP40, and 1 × protease inhibitor cocktail [Bimake]) and nucleus lysis buffer (50 mM Tris–HCl, pH 8.1, 10 mM EDTA, 1% SDS, and 1 × protease inhibitor cocktail), and protein concentrations were measured using a BCA protein assay kit (Beyotime Biotechnology, Shanghai, China). Histones were isolated from cells and tissues following standard acid extraction protocols (Shechter et al, ²⁰⁰⁷), and their concentrations were determined using the Bradford protein assay kit (Beyotime). Co‐immunoprecipitation was performed following the instructions of Protein A/G PLUS‐Agarose (sc‐2003, Santa Cruz), and 200 µg of nuclear proteins were used for each co‐immunoprecipitation reaction.

Proteins were separated using 6–15% SDS–PAGE and transferred onto polyvinylidene fluoride membranes (Millipore). Blotted membranes were blocked in 5% skim milk (BD Biosciences) in Tris‐buffered saline with 0.1% Tween 20 (TBS‐T) and incubated with primary antibodies at 4 °C overnight. Membranes were then washed three times with TBS‐T and incubated with secondary antibodies at room temperature for 1–2 h. The immunoreactive products were detected using SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Scientific) and a Tanon‐5200 Chemiluminescent Imaging System (Tanon, Shanghai, China). ImageJ was used to perform quantitative analysis of Western blotting results as the ratio of target histone marks or protein gray value to the gray value of reference proteins (H3 or β‐Actin). For dot blot assays, peptide samples were spotted onto PVDF membranes, which were first blocked at room temperature for 1 h and then washed in TBS‐T. Primary and secondary antibodies were incubated at room temperature for 2 h and 45 min, respectively, followed by washing three times with TBS‐T and signal detection. Detailed information on the primary and secondary antibodies used for Western blotting analysis is summarized in Table EV1.

ChIP‐seq, ChIP‐qPCR and FAIRE‐qPCR

ChIP was performed as previously described (Liu et al, ²⁰¹⁰). Briefly, cells (digested forebrain cells and NSPCs) were cross‐linked with 1% formaldehyde (Sigma‐Aldrich) for 10 min and stopped with 125 mM glycine for 5 min, and then washed three times with ice‐cold PBS. Next, chromatin fractions were isolated and sonicated for an average size of approximately 200–600 bp. Sheared chromatin was precleared with salmon sperm DNA/protein A‐agarose beads (16‐157, Millipore) and then incubated with antibodies at 4°C overnight in IP dilution buffer (0.01% SDS, 1.1% Triton X‐100, 1.2 mM EDTA, 16.7 mM Tris–HCl, pH 8.1, 167 mM NaCl, and 1 × protease inhibitor cocktail). Antibodies‐chromatin complexes were collected using 60 μl beads by incubation with rotation for 2 h at 4°C, and the beads were sequentially washed twice with IP dilution buffer, twice with TSE‐500 buffer (0.1% SDS, 1% Triton X‐100, 2 mM EDTA, 20 mM Tris–HCl, pH 8.1, and 500 mM NaCl), twice with LiCl wash buffer (100 mM Tris–HCl, pH 8.1, 300 mM LiCl, 1% NP40, and 1% deoxycholic acid), and twice with TE buffer (100 mM Tris–HCl, pH 8.1 and 1 mM EDTA) (10 min at 4°C with rotation for each wash). Immune complexes were eluted from the beads twice for 15 min at room temperature by vortexing using 250 μl elution buffer (1% SDS and 100 mM NaHCO₃). Formaldehyde‐induced protein‐DNA crosslinking was heat‐reversed by incubation at 65°C overnight with 200 mM NaCl. Then, IPed DNA was purified using the phenol‐chloroform‐isoamyl alcohol (25:24:1) and precipitated with two volumes of 100% ethanol and 10 µg linear acrylamide (Invitrogen) at −20°C for 2 h. Library construction and sequencing of ChIP samples were performed by Beijing Genomics Institute Co., Ltd. (Shenzhen, China) on the BGISEQ‐500 platform and Annoroad Gene Technology Co., Ltd. (Beijing, China) on Illumina platforms. ChIP samples were subjected to qRT‐PCR on a Roche LightCycler 480 II, and the results were expressed as percent of input DNA. FAIRE‐qPCR was performed according to previously published methods (Li et al, ^2019b). Detailed information for primers used in ChIP‐qPCR and FAIRE‐qPCR is provided in Table EV2.

CUT&Tag and WGBS

We used the Hyperactive In‐Situ ChIP Library Prep Kit for Illumina (pG‐Tn5) (TD901, Vazyme, Nanjing, China) to construct CUT&Tag libraries according to the manufacturer’s instructions. The WGBS library was constructed by Annoroad Gene Technology Co., Ltd. and sequenced on an Illumina platform with 150 bp per read length.

RNA‐seq and qRT‐PCR

RNA was extracted using TRIzol reagent (Invitrogen), and mRNA was purified from total RNA using poly T oligo‐attached magnetic beads. Libraries were then constructed using the NEBNext Ultra RNA Library Prep Kit for Illumina (NEB), following the manufacturer’s instructions, and sequenced on Illumina platforms by Annoroad Gene Technology Co., Ltd. RNA samples were subjected to qRT‐PCR using a Roche LightCycler 480 II. In brief, total RNA was transcribed into cDNA using the TransScript One‐Step gDNA Removal and cDNA Synthesis Kit (TransGen Biotech, Beijing, China). For qRT‐PCR analysis, cDNA was quantified by using Hieff qPCR SYBR Green Master Mix (YEASEN, Shanghai, China) in a 20‐μl reaction system according to instructions, PCR steps were performed 5‐min initial pre‐denaturation at 95°C, followed by 45 cycles of each 10 s at 94°C, 30 s at 60°C, 30 s at 72°C and samples were run in triplicate. The relative expression was calculated using the 2^−ΔΔCt method. Detailed information about the primers used in qRT‐PCR is provided in Table EV2.

ChIP‐seq, CUT&Tag, and WGBS data analysis

ChIP‐seq libraries were sequenced, generating 50‐bp single reads or 150‐bp paired ends. Raw read data were filtered using Trimmomatic (v.0.36) and quality‐controlled using FastQC (v. 0.11.7) (Bolger et al, ²⁰¹⁴). High‐quality reads were aligned using Bowtie 2 (v2.4.1) to the mouse reference genome using default parameters (Langmead & Salzberg, 2012). Samtools (v.1.1.0) was then used to convert files into bam format and filter reads with parameters “‐F 1804 ‐q 30” for single‐end sequencing data and “‐F 1804 ‐f 2 ‐q 30” for paired‐end sequencing data (Li et al, ²⁰⁰⁹). PCR duplicates were removed using the Mark Duplicates function in Picard (v.2.22.0), and MACS (v.2.2.7.1) was used to perform peak calling (‐q 0.05) for the IP samples relative to the control samples (IgG or input) (Feng et al, ²⁰¹²). CUT&Tag analysis was identical to ChIP‐seq except aligning with parameters “‐‐mm ‐‐local ‐‐very‐sensitive‐local ‐‐no‐unal ‐‐no‐mixed ‐‐no‐discordant ‐‐phred33 ‐I 10 ‐X 700” by Bowtie 2 (Kaya‐Okur et al, ²⁰¹⁹).

MAnorm (v.1.2.0) was used for quantitative comparison of ChIP‐Seq data (Shao et al, ²⁰¹²). Peaks with differential histone Kcr or Kac were defined by P‐values < 0.05, and absolute M‐values greater than 1.5. Peak annotation was performed using ChIPseeker (v.1.22.1) at the gene level, and promoter regions were defined as ± 1 kb of TSS (Yu et al, ²⁰¹⁵). Genes were considered under regulation of histone marks with at least one peak annotated in their promoter regions. The BEDTools (v2.29.0) “coverage” function was used for read counting within ± 1 kb of TSS or peak regions (Quinlan & Hall, 2010). Reads per kilobase per million reads (RPKM) values for ChIP‐seq and CUT&Tag analysis were calculated as follows: [(read counts) / (region length in kb)] / (total mapped and filtered reads in Mb).

ChromHMM (v.1.20) was employed to identify chromatin states and perform genome annotation (Ernst & Kellis, 2012). “BinarizeBam” (‐f 5) and “LearnModel” functions in ChromHMM were used to convert a set of bam files of aligned reads into binarized data files and learn chromatin state models, respectively. DeepTools (v. 3.4.0) “computeMatrix,” “plotHeatmap” and “plotProfile” functions were used to generate heat maps and profile plots (Ramírez et al, ²⁰¹⁶). For HOMER (v. 4.11.1) de novo motif analysis, “findMotifsGenome.pl (‐size 500 ‐S 5 ‐p 30)” function was used for motif finding of histone Kcr peaks at promoters (Heinz et al, ²⁰¹⁰). For genome browser representation, data in bigwig files generated by deepTools were visualized using IGV (v. 2.4.10) (Thorvaldsdóttir et al, ²⁰¹³).

WGBS data were aligned to the bisulfite‐converted GRCm38 reference genome using Bismark (v0.22.3) (Krueger & Andrews, 2011). We used the deduplicate_bismark script to remove duplicates and the bismark_methylation_extractor script to extract methylation status. Only CpGs with at least five reads covering them were used for downstream analysis. Quality control of ChIP‐seq, CUT&Tag, and WGBS data are displayed in Table EV3.

RNA‐seq and Single‐cell RNA‐seq data analysis

High‐quality reads of RNA‐seq data were quantified using Salmon (v.1.1.0) with parameters “‐i ‐g ‐‐gcBias ‐‐validateMappings” (Patro et al, ²⁰¹⁷). Differential gene expression analysis was conducted using DESeq2 (v1.26.0), and differentially expressed genes were defined by P‐adjust < 0.05, and absolute fold change > 1.5 (Love et al, ²⁰¹⁴). For RNA‐seq data of brain explants, we employed pseudo‐replicates of FASTQ files (three biological replicates were combined first and then randomly divided into three equal files to perform downstream analysis) to uncover precise gene expression changes; differentially expressed genes were defined by P‐adjust < 0.05, and absolute fold change > 0. The quality control of RNA‐seq data is shown in Table EV3.

Seurat (v3.1.1) was used to analyze the single‐cell RNA‐seq data (Butler et al, ²⁰¹⁸). Only genes that were present in at least 10 cells were considered. Cells with fewer than 500 detectable genes or whose mitochondrial contribution exceeded 10% of transcripts or with fewer than 60 transcripts were removed. The data were normalized, the batch effect was removed, and 3000 variable genes were identified using the “SCTransform” function. Next, principal component analysis (PCA) was carried out, and the top 15 principal components (PCs) were used to find clusters with a clustering resolution set to 1.6, using the “FindClusters” function. “FindMarkers” function was adopted to search differentially expressed genes across different cell types with the following parameters: ident.1 = (cell type A), ident.2 = (cell type B), assay = “SCT”, slot = “data”, test.use = “MAST”, min.pct = 0, logfc.threshold = 0, min.cells.feature = 0, min.cells.group = 0, pseudocount.use = 1, genes with p_val_adj < 0.05 and avg_logFC > 0 were defined as cell type A‐specific genes.

Gene enrichment analysis was performed using clusterProfiler (v3.14.3) (Yu et al, ²⁰¹²). BETA “basic” function (‐‐df 0.05 ‐‐da 1 ‐c 0.001) was used for activating and repressive function prediction of peak regions with differential histone Kcr (Wang et al, ²⁰¹³). The mouse reference genome sequence (vM24) and gene annotation (vM24) were downloaded from GENCODE (https://www.gencodegenes.org/). The public and published sequencing data analyzed in this study are listed in Table EV4.

Statistics

Two‐tailed unpaired Student’s t‐test, one‐way ANOVA with post hoc Tukey’s test, and two‐sided Fisher’s exact test were used to analyze statistical significance. Differences were regarded as significant by P value; NS: no significance, *P < 0.05, **P < 0.01, and ***P < 0.001. All statistical analyses and diagrams were performed using GraphPad Prism 8.0 and R (v3.6.3).

Author contributions

SKD and CML, conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; PPL, HZD, XL, YJX, CL, YYW, and ZQT, collection and assembly of data.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

Table EV1

Table EV2

Table EV3

Table EV4

Source Data for Expanded View and Appendix

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

Source Data for Figure 6

EMBO reports (2021) 22: e52023.

Data availability

The ChIP‐seq, CUT&Tag, WGBS, and RNA‐seq datasets generated and analyzed during the current study have been deposited in the NCBI Gene Expression Omnibus (GEO) and are accessible through the series accession numbers GSE124540, GSE172310, and GSE178097.