The EZH1–SUZ12 complex positively regulates the transcription of NF-κB target genes through interaction with UXT

The epigenetic control of chromatin is crucial for transcription regulation and the proper response to extracellular signals in the cell. The polycomb repressive complexes 1 and 2 (PRC1 and PRC2) are the most studied protein complexes among all the Polycomb group (PcG) proteins and have long been considered as transcription repressors (Campos et al., 2014; Conway et al., 2015; Margueron and Reinberg, 2011). PRC1 inhibits transcription by regulating mono-ubiquitination on histone H2A, whereas PRC2 is responsible for the methylation on histone H3K27 (Margueron and Reinberg, 2011).

In Drosophila, three major subunits form PRC2 complex, including Enhancer of zeste [E(z)], extra sexcombs (ESC) and suppressor of zeste 12 (SUZ12), among which E(Z) is the enzyme for H3K27 methylation (Lanzuolo and Orlando, 2012). In mammals, two close homologs exist for E(Z), enhancer of zeste 1 polycomb repressive complex 2 subunit (EZH1) and enhancer of zeste 2 polycomb repressive complex 2 subunit (EZH2) (Shen et al., 2008). EZH2 mimics the function of E(Z) in Drosophila and is the main methyltransferase for H3K27 in mammalian cells (Shen et al., 2008). However, controversial reports exist about EZH1, which means its activities have been a puzzle for a long time. Compared with EZH2, EZH1 has lower enzymatic activity, suggesting it might have distinct functions (Margueron et al., 2008). Several groups reported that EZH1 methylates histone H3K27 and compensates for the functions of EZH2 upon its absence (Bae et al., 2015; Hidalgo et al., 2012; Shen et al., 2008). Recently, Mousavi et al. have reported that EZH1 behaves as a positive regulator for transcription and is required for proper recruitment of RNA polymerase II (Pol II) to target genes (Mousavi et al., 2012). Another study has demonstrated that EZH1 forms two different complexes with distinguished functions (Xu et al., 2015). One is similar to the classical PRC2 complex containing EZH1, embryonic ectoderm development (EED) and SUZ12, and the second complex is only with SUZ12 (Xu et al., 2015). The former complex inhibits transcription, while the latter one activates transcription (Xu et al., 2015). However, the detailed mechanisms of how the two EZH1 complexes regulate transcription remain elusive.

The nuclear factor κ-light-chain-enhancer of activated B cells (NF-κB) signaling pathway is a well-studied pathway involved in inflammation, immunity and anti-apoptosis (Iwai, 2012; Oeckinghaus et al., 2011). It is activated upon many extracellular signals, and tumor necrosis factor (TNFα) has been used as one of the typical molecules to activate the pathway (Hu et al., 2014; Wang et al., 2012; Zhang et al., 2014). Although the regulations of NF-κB pathway in the cytosol have been extensively studied, the events after NF-κB enters the nucleus are still not clear and have emerged as being crucial regulatory steps for pathway activation. Multiple histone methyltransferases are involved in pathway regulation. We have previously found that lysine (K)-specific methyltransferase 2A (KMT2A; also known as MLL1), a histone H3K4 methyltransferase, selectively regulates the activation of NF-κB target genes (Wang et al., 2012). EZH2 regulates the activation of NF-κB pathway by distinct mechanisms in different cell lines, which shows the complexity of epigenetic regulators (Lee et al., 2011). The NF-κB molecule itself is also regulated by protein methylation. SET-domain-containing (lysine methyltransferase) 7 (SET7; also known as SET9 and SETD7) methylates NF-κB at Lys314 and Lys315 and promotes NF-κB degradation (Yang et al., 2009). However, SET7 also methylates Lys37 on NF-κB and selectively activates target genes (Ea and Baltimore, 2009). Besides protein methylation, multiple transcription co-regulators have been found to be involved in the activation of NF-κB target genes. For example, the ubiquitously expressed prefoldin-like chaperone UXT (also known as STAP1 and ART-27), a small chaperone-like protein, has been reported to interact with the p65 subunit of NF-κB (also known as RELA) and functions as a transcription co-activator, as well as a cytoplasmic regulator for anti-viral pathway (Huang et al., 2011,, 2012; Sun et al., 2007). However, how UXT contributes to the activation of NF-κB target genes is still not clear.

In this study, we discovered that UXT physically interacts with EZH1 and SUZ12, but not with EED or EZH2. EZH1 and UXT are required for p65 recruitment to NF-κB target genes and the subsequent induction of expression. EZH1 and UXT are associated with Pol II, and regulate its loading to chromatin. Our study indicates that UXT, EZH1 and SUZ12 together help to link p65 with Pol II and regulate the activation of NF-κB target genes.

To investigate the role of EZH1 in regulating transcription, a yeast two-hybrid screen was performed with full-length EZH1 as the bait. Multiple clones were identified encoding the open reading frame of UXT, which has been shown to be a transcription co-factor and to interact with multiple transcription factors (Carter et al., 2014; Li et al., 2014; McGilvray et al., 2007; Sun et al., 2007). We speculated that UXT might regulate transcription by bridging the master transcription factors and epigenetic regulators and continued with the following studies.

We cloned the full-length cDNA of EZH1 and UXT into mammalian expression vectors, and performed immunoprecipitation with anti-FLAG or anti-HA antibodies. HA–EZH1 successfully pulled down FLAG–UXT and vice versa (Fig. 1A). To examine the interaction between endogenous proteins, we generated antibodies against the two proteins and performed immunoprecipitation. The results showed that endogenous EZH1 and UXT are associated with each other (Fig. 1B). To further characterize whether EZH1 directly interacts with UTX, both proteins were expressed in bacteria and purified with GST or His affinity resins. Then His–UXT was bound to Ni resins, which successfully pulled down GST–EZH1. His–SPOP was used as a negative control (Fig. 1C). These results indicate that EZH1 and UXT directly interact with each other. Given that UXT is a small protein and contains only one prefoldin α domain, we just mapped the interacting domain in EZH1. A series of EZH1 truncations were generated (Fig. 1D) and co-immunoprecipitation assays indicated that the fragment from 430 to 480 residues in EZH1, which represents the SANT domain, is crucial for the interaction (Fig. 1E–G).

One recent paper has demonstrated that two different EZH1 complexes exist in the cell, one with EED and SUZ12, and another with SUZ12 only (Xu et al., 2015). To study which complex is associated with UXT, we performed immunoprecipitation with the anti-UXT antibody. The result indicated that EZH1 and SUZ12 are associated with UXT, but not EED, suggesting UXT is associated only with the EZH1–SUZ12 complex (Fig. 2A). To further confirm this result, exogenously expressed or endogenous SUZ12 was pulled down with the corresponding antibody and UXT was found to be associated with SUZ12 (Fig. 2B,C). By contrast, immunoprecipitation of EED did not pull down UXT, nor did the reciprocal UXT immunoprecipitation (Fig. 2D). The assay was repeated more than three times in both directions to confirm that endogenous EED and UXT are not associated with each other. These results suggest that UXT specifically interacts with the EZH1–SUZ12 complex, but not with EED.

Previously, UXT has been reported to interact with p65 as a co-activator to transcriptionally regulate the expression of NF-κB target genes (Sun et al., 2007). Given that EZH1 interacts with UXT, we studied whether EZH1 also regulates the NF-κB pathway. UXT was knocked down by small interfering RNA (siRNA) in the HCT116 cell, and the TNFα-induced expression of NFKBIA (also known as IκBα), CXCL8 and TNFAIP3 (also known as A20), three typical NF-κB target genes, were significantly downregulated, as reported previously (Fig. 3A; Sun et al., 2007). Then, we knocked down EZH1 in HCT116 with two different siRNAs. Similar to upon UXT knockdown, the induction of the above three genes was greatly impaired, indicating that EZH1 positively regulates the activation of NF-κB target genes (Fig. 3B). We also knocked down EZH1 in RKO, HEK293 and U2OS cell lines, and observed the same results (Fig. S1A–C). To investigate whether EZH1 regulates the activation of NF-κB target genes during the anti-viral response, Sendai virus (SeV) was used to treat HCT116, and we found that the expression of NF-κB downstream genes was significantly impaired after EZH1 knockdown (Fig. S1D). To further confirm the role of EZH1 in NF-κB signaling, we profiled the gene expression pattern after EZH1 knockdown with next-generation sequencing. EZH1 deficiency impaired the expression of almost all the TNFα-induced genes, similar to the effect of UXT (Fig. 3C), whereas EZH1 only regulated the basal expression of a small portion of genes (Table S1). This is different from some of the other epigenetic regulators, such as KMT2A, which selectively regulates the activation of NF-κB target genes by modifying H3K4 methylation (Wang et al., 2012), suggesting that EZH1 might regulate NF-κB signaling through a distinctive mechanism.

Interestingly, when analyzing the signaling pathway after UXT knockdown, the differentially expressed genes were found to be specifically enriched in the gene ontology terms ubiquitin-mediated proteolysis, cell cycle, p53 signaling etc, suggesting that UXT might play roles in cancer-related signaling pathways (Fig. S2).

To further explore the roles of other PRC2 subunits in NF-κB signaling, we knocked down SUZ12 or EED and analyzed the expression of NF-κB target genes. Interestingly, SUZ12 knockdown, but not EED knockdown, impaired the activation of NFKBIA and CXCL8 (Fig. 3D). The above results suggest that the EZH1–SUZ12 complex positively regulates the activation of NF-κB target genes, probably through interaction with UXT, whereas EED is not involved.

EZH2 is the major H3K27 methyltransferase in the mammalian cell and EZH1 has been proposed to compensate for its function under certain circumstances (Margueron et al., 2008; Shen et al., 2008). We investigated the impact of UXT on global H3K27 methylation by western blotting. UXT was knocked down by two different siRNAs and the global H3K27me1, H3K27me2 and H3K27me3 levels were measured with corresponding antibodies. Multiple experiments were performed and all of them indicated that the global levels of the three methylation statuses were not significantly affected (Fig. 4A). The similar results were observed with two different EZH1 siRNAs (Fig. 4B), whereas depletion of EZH2 successfully decreased the global H3K27me3 levels (Fig. S3A). This is consistent with the previous report that EZH2, but not EZH1, is the major enzyme involved in H3K27 methylation (Margueron et al., 2008).

Although EZH1 and UXT do not regulate the global H3K27 methylation level, it was still possible that they regulated the methylation on NF-κB target genes. We investigated the H3K27 methylation status on NFKBIA and CXCL8 by performing a chromatin immunoprecipitation (ChIP) assay. Surprisingly, repeated experiments supported that the levels of H3K27me1, H3K27me2 and H3K27me3 on these genes were not significantly reduced after EZH1 or UXT knockdown, although their expression decreased (Fig. S3B). Here, GAPDH and MYOD1 were used as controls to ensure each experiment was performed correctly (Fig. S3B). To further confirm this result, we performed H3K27me1, H3K27me2 and H3K27me3 ChIP-Seq analysis. Consistent with the above result, although EZH1 or UXT deficiency altered H3K27 methylation on a few genes, the average levels on NF-κB target genes were not changed significantly (Fig. 4C). Actually, the average signals of all three forms of H3K27 methylation were quite low on NF-κB target genes, in comparison with other genes (Fig. 4C), which hints that H3K27 methylations might not be crucial in regulating the expression of NF-κB target genes here.

The above data demonstrate that the regulation of NF-κB signaling by UXT and EZH1 is not dependent on H3K37 methylation. A previous study has reported that UXT deficiency impairs the translocation of p65 from the cytosol to the nucleus (Sun et al., 2007), we started to investigate whether EZH1 regulates NF-κB pathway through a similar mechanism. We firstly confirmed the role of UXT by western blotting after fractionation. After UXT knockdown, the p65 level in the nucleus was modestly reduced compared with the control (Fig. 5A). Then the effect of EZH1 knockdown was investigated. Interestingly, EZH1 knockdown did not affect the p65 level in the nucleus (Fig. 5B).

Then we examined the subcellular localization of EZH1. UXT was localized both in the cytosol and the nucleus (Fig. 5C), as reported previously (Sun et al., 2007). However, the results of western blotting after fractionation and fluorescent staining both indicated that EZH1 was mostly localized in the nucleus (Fig. 5C,D). Hence, it is highly possible that EZH1 interacts with UXT and regulates the expression of NF-κB target genes only in the nucleus.

We next assessed whether EZH1 regulates the recruitment of p65 to chromatin. First, we studied whether EZH1 interacts with p65. As reported previously, UXT interacts with p65 (Fig. 5E; Sun et al., 2007). We did not detect an interaction between EZH1 and p65 at the endogenous level; however, upon co-expressing EZH1 and p65 in the cell, we did observe an interaction between the exogenous proteins (Fig. 5F). This suggests that the interaction between EZH1 and p65 is quite weak and perhaps not direct. Then we studied whether EZH1 is bound to NF-κB target genes on chromatin. Because we do not have a ChIP grade anti-EZH1 antibody, we generated a stable HCT116 cell line expressing HA-tagged EZH1 (Fig. S4A) and performed a ChIP assay with anti-HA antibody (Fig. 5G). The results demonstrated that EZH1 binds to the promoters of NFKBIA and CXCL8 on chromatin even without TNFα treatment, and when EZH1 was knocked down, the signal significantly decreased (Fig. 5G).

Next, a p65 ChIP assay was performed in HCT116 cells. When UXT was knocked down, the recruitment of p65 to NFKBIA and CXCL8 promoters was greatly impaired, consistent with the previous report (Fig. 5H; Sun et al., 2007). When EZH1 was knocked down, the same result was observed (Fig. 5H). The result was then confirmed in 293FRT cells (Fig. S4B), as well as after virus treatment (Fig. S4C). These results demonstrate that EZH1 regulates the induced transcription of NF-κB target genes not through modifying histones, but by modulating the recruitment of NF-κB to target genes.

We further investigated whether p65 also regulates EZH1 binding to target genes. The RELA gene was knocked down in the HA–EZH1 stable cell line and a ChIP assay was performed with anti-HA antibody. The data indicated that TNFα treatment increases EZH1 binding to target genes and p65 deficiency impairs the binding (Fig. 5I). All the above data suggest that p65 and EZH1 probably work synergistically to regulate the activation of their target genes.

Pol II is the sole enzyme responsible for mRNA transcription and in a previous study it has been reported EZH1 is associated with Pol II (Mousavi et al., 2012). To investigate whether SUZ12 and UXT are associated with Pol II, we performed immunoprecipitation with a widely used commercial monoclonal antibody 8WG16 in HCT116, and found that EZH1, UXT and SUZ12 were all associated with Pol II, but that EED was not (Fig. 6A,B). The recruitment of Pol II to NFKBIA and CXCL8 was also greatly reduced in the absence of EZH1, UXT or SUZ12, but not EED (Fig. 6C,D). The ChIP data was confirmed in 293FRT cells (Fig. S4D). These results are consistent with our previous data, suggesting that EZH1, SUZ12 and UXT function synergistically in regulating NF-κB signaling, but that EED does not.

TNFα treatment not only activates NF-κB signaling, but also the apoptosis pathway. The activation of NF-κB pathway strongly inhibits apoptosis, so the inhibition of NF-κB activation often enhances TNFα-induced apoptosis. To further confirm the role of EZH1 in TNFα signaling, we studied its effect on the apoptosis induced by TNFα. EZH1 or UXT knockdown alone slightly increased the percentage of sub-G1 cells, which represents dead cells (Fig. 7A). With TNFα treatment, EZH1 or UXT knockdown significantly increased the percentage of apoptotic cells (Fig. 7A). To further confirm this result, we used annexin V and propidium iodide to double label the cell, and flow cytometry analysis confirmed that EZH1 and UXT deficiency increased the cell death after TNFα treatment (Fig. 7B). NF-κB signaling inhibits apoptosis through the activation of downstream anti-apoptosis genes. We surveyed our RNA-seq data and confirmed by RT-PCR that TNFα activates the expression of BIRC2 (encoding cIAP1) and BIRC3 (encoding cIAP2), two well-known anti-apoptotic genes, in the studied HCT116 cell line (Fig. S4E). Further study showed that EZH1 and UXT both regulate the expression of BIRC2 and BIRC3 induced by TNFα (Fig. 7C). These results suggest that EZH1 and UXT increase the sensitivity of the cells to apoptosis, probably through regulating the expression of BIRC2 and BIRC3.

The epigenetic regulation of signaling pathways has emerged to be one of the crucial steps in the cell for responding to the extra- and intra-cellular signals. Polycomb group proteins are key factors in determining cell status and transcriptional programs. However, the functions and mechanisms of EZH1 have remained controversial. In the current study, we demonstrate that EZH1 regulates the transcription of NF-κB target genes by interacting with UXT, a small protein functioning as a transcription co-activator. EED is required for PRC2 activity (Montgomery et al., 2005,, 2007), but because UXT here is just associated with EZH1–SUZ12, and not EED, it is reasonable that H3K27 methylation is not involved in the transcription activation regulated by UXT and EZH1. Instead, both EZH1 and UXT are required for the expression of NF-κB target genes, and proper p65 recruitment and Pol II loading to their target genes. Here, UXT, EZH1 and SUZ12 seem to act like a bridge to link NF-κB and Pol II together.

The functional difference between EZH2 and EZH1 on transcription regulation is also puzzling. Here, we show that EZH1 positively regulates transcription in an EED- and H3K27-methylation-independent manner. Our data are consistent with the recent report by Xu et al. (2015). In fact, EZH2 sometimes also positively regulates transcription independently of its enzyme activity in a context-dependent manner (Lee et al., 2011; Xu et al., 2012). Our study is helpful to understand the detailed mechanisms of how EZH1 and EZH2 regulate gene expression by different mechanisms.

UXT has been reported to regulate multiple signaling pathways, making it an important regulator for inducible transcription (Carter et al., 2014; Li et al., 2014; McGilvray et al., 2007; Sun et al., 2007). Meanwhile, EZH1 is widely associated with positive transcription genome. Hence, it is highly possible that UXT and EZH1 use the same mechanisms to regulate other signaling pathways. It will be interesting to study whether EZH1 regulates other UXT-dependent pathways.

The 293FRT cell line was purchased from Invitrogen, and HEK293 and HCT116 cell lines were purchased from Cell Bank of Chinese Academy of Sciences. Cells were grown in Dulbecco's modified Eagle's medium (DMEM) (4.5 g/l D-glucose) with L-glutamine, 25 mM HEPES and 10% fetal bovine serum (FBS, HyClone) and a penicillin–streptomycin supplement. The stable cell line containing HA-tagged EZH1 was constructed with a lentivirus system (pHAGE, psPAX2 and pMG2.G) in HCT116 cells. The antibodies against FLAG (1:1000 for immunoblotting, 1:500 for immunoprecipitation; M2, Sigma), HA (1:1000 for immunoblotting, 1:500 for immunoprecipitation; clone CB051, Origene), GST (1:1000 for immunoblotting; Abmart, M20007), His (1:1000 for immunoblotting; Abmart, M20001), SUZ12 (1:1000 for immunoblotting, 1:500 for immunoprecipitation; clone D39F6, CST 3737), EED (1:1000 for immunoblotting, 1:500 for immunoprecipitation; Proteintech, 16818-1-AP), H3K27me1 (1:1000 for immunoblotting, 1:250 for ChIP; Millipore 07-448), H3K27me2 (1:1000 for immunoblotting, 1:250 for ChIP; ACTIVE MOTIF 39919), H3K27me3 (1:1000 for immunoblotting, 1:250 for ChIP; ABclonal A2363, Millipore 07-449), H3 (1:1000 for immunoblotting; ABclonal A2348), NFKBIA (1:1000 for immunoblotting; Epitomics, 1130-1), 8WG16 (1:2000 for immunoblotting, 1:500 for ChIP; Covance) and p65 (1:1000 for immunoblotting, 1:250 for ChIP; Abcam, ab7970) were purchased from indicated companies. The antibodies against EZH1 and UXT were raised against full-length EZH1 or UXT expressed in bacteria using the pET30-C plasmid. The siRNA information is in Table S2.

Cells were harvested and spun down in cold PBS. 10 volumes of buffer A (10 mM Tris-HCl pH 7.4, 5 mM MgCl₂, 10 mM NaCl, 1 mM DTT, proteinase inhibitors) was added to the cells, which were then incubated on ice for 10 min. Next, 0.5 volumes of buffer B (10 mM Tris-HCl pH 7.4, 5 mM MgCl₂, 10 mM NaCl, 1 mM DTT, proteinase inhibitors, 10% NP-40) was added to the cells, which were incubated on ice for 1 min. The cell suspension was vortexed for 5 s and centrifuged at 800 g for 5 min at 4°C. The supernatant was collected as cytoplasm fraction. The above steps were repeated once more and the supernatant was discarded. The sediment was suspended in 10 volumes of PBS as the nuclear fraction. SDS loading buffer was added to the cell fractions for western blotting.

Cells were cultured on coverslips and fixed with freezing methanol after washing twice in PBS. The coverslips were then washed three times by PBS and blocked in PBS with 1% BSA for 10 min. The coverslips were hybridized with primary and secondary antibodies for 1 hour each. Then the coverslips were mounted with prolong anti-fade kit (Invitrogen) and observed with fluorescent microscopy.

The cells were harvested and lysed in NP40 lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.5% NP40) or high-salt lysis buffer (20 mM HEPES pH 7.4, 10% glycerol, 0.35 M NaCl, 1 mM MgCl₂, 0.5% Triton X-100, 1 mM DTT) with proteinase inhibitors. The supernatant was then incubated with protein G beads (GE Healthcare) and the desired antibody at 4°C for 4 h. The beads were spun down and washed three times with lysis buffer. The final drop of wash buffer was vacuumed out and SDS loading buffer was added to the beads, followed by western blotting.

ChIP assays were performed as previously described (Wu et al., 2008). Briefly, ∼10⁷ cells were fixed with 1% formaldehyde and quenched by glycine. The cells were washed three times with PBS and then harvested in ChIP lysis buffer (50 mM Tris-HCl, pH 8.0, 1% SDS and 5 mM EDTA). DNA was sonicated to 400–600 bp before extensive centrifugation. Four volumes of ChIP dilution buffer (20 mM Tris-HCl pH 8.0, 150 mM NaCl, 2 mM EDTA and 1% Triton X-100) was added to the supernatant. The resulting lysate was then incubated with protein G beads and antibodies at 4°C overnight. The beads were washed five times and DNA was eluted in ChIP elution buffer (0.1 M NaHCO₃, 1% SDS and 30 μg/ml proteinase K). The elution was incubated at 65°C overnight and DNA was extracted with a DNA purification kit (Tiangen). The purified DNA was assayed by quantitative PCR with Biorad MyIQ. Assays were repeated at least three times. Data shown are mean±s.d. of representative experiments. The primer information is in Table S2. At least three biological replicates were analyzed in each experiment. A t-test was used for statistical analysis.

Cells were scraped down and collected by centrifugation. Total RNA was extracted with an RNA extraction kit (Yuanpinghao) according to the manufacturer's manual. Approximately 1 µg of total RNA was used for reverse transcription with a first-strand cDNA synthesis kit (Toyobo). The amount of mRNA was assayed by quantitative PCR. β-actin was used to normalize the amount of each sample. Assays were repeated at least three times. Data shown are mean±s.d. of one representative experiment. The primer information is in Table S2. At least three biological replicates were analyzed in each experiment. A t-test was used for statistical analysis.

The extracted mRNA from three biological replicates was subjected to high-throughput sequencing. The mRNA-seq library was constructed by using the Illumina TruSeq library construction kit. 5 μg total RNA was used for initiation, prepared according to the manufacturer's instructions. mRNA-seq libraries were sequenced using HiSeq2000 for 100-bp paired-end sequencing. Quality control of mRNA-seq data was performed using Fatsqc, and low-quality bases were trimmed. After quality control, data were mapped to hg19 genome reference by Tophat2 allowing a maximum of two mismatches. Cufflinks was used to assess the differentially expressed genes. Gene ontology analysis was performed using DAVID (http://david.abcc.ncifcrf.gov).

ChIP was performed by using antibody against H3K27me1, H3K27me2 and H3K27me3. After ChIP, a sequence library was prepared by using a KAPA Hyper Prep Kit. We used 10 ng of ChIP DNA as initiation, prepared according to the manufacturer's instructions. The ChIPed DNA from three biological replicates were mixed together and subjected to high-throughput sequencing. ChIP DNA and matched input DNA were prepared for end repair and ‘A’-tailing, adaptor ligation and library amplification. ChIP-Seq sequencing was performed by using an Illumina HiSeq2500 platform for 100-bp paired-end sequencing. Quality control of ChIP-seq data was performed using Fatsqc, and low-quality bases and library adaptors were trimmed. After quality control, data were mapped to hg19 genome reference by bowtie Tophat2 allowing a maximum of two mismatches. For histone modification analysis, SICER software was used to call peaks and identify the differential modification region.

We thank Dr Hong-bing Shu of Wuhan University and Dr Yan-Yi Wang of the Institute of Viology for sharing reagents and plasmids.