Information

H4. Recent Links and References - Biology

H4. Recent Links and References - Biology


We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

  1. Wasmer, C. Amyloid Fibrils of the HET-s(218-289) Prion Form a b Solenoid with a Triangular Hydrophobic Core. Science. 319, 1523 (2008)
  2. Balch, W. et al. Adapting Proteostasis for Disease Intervention. 319, 916 (2008)
  3. Krishnan and Linquist, Structural insights into a yeast prion illuminate nucleation and strain diversity. Nature, 435, 765 (2005)
  4. Ritter et al. Correlation of structural elements and infectivity of the HET-s prion. Nature, 435, 844 (2005)
  5. Nelson, R. Structure of the cross-b spine of amyloid-like fibrils. Nature 435, 773 (2005).
  6. Hardy and Selkoe. The Amyloid Hypothesis of Alzheimer's Disease: Progress and Problems on the Road to Therapeutics. Science 297, pg 353 (2002)
  7. Pepys et al. Targeted Pharmacological Depletion of Serum Amyloid P Component for Treatment of human amyloidosis. Nature. 254, 231 and 254 (2002)
  8. Bucciantini et al. Inherent toxicity of aggregates implies a common mechanism for protein misfolding diseases. 416, pgs. 483, 507 (2002)
  9. Walsh et al. Naturally secreted oligomers of amyloid b protein potently inhibit hippocampal long-term potentiation in vivo. 483, 535 (2002)
  10. Esler et al., A Portrait of Alzheimer Secretases - New Features and Familiar Faces. 293, pg 1449 (2001)
  11. Happarstrom et al. Trans-Suppression of Misfolding in an Amyloid Disease. Science 293 pg 2459 (2001)
  12. An Inflammatory Drug Prospect (for Alzheimers). 414, pg 199 (2001)
  13. Saborio et al. Sensitive Detection of Pathological prion protein by cyclic amplification of protein misfolding. 411, pg 810 (2001)
  14. Bence et al. Impairment of the Ubiquitin-Proteasome System by Protein Aggregation. 292, pg 1552 (2001)
  15. Cao et al. Elusive Protein Auditions for Several Roles (about APP in neural cells) Science. 293, pg 28, 115 (2001)
  16. Serio et al. Nucleated Conformational Conversion and the Replication of Conformational Information by a prion determinant. 289, pg 1317 (2000)
  17. Chien and Weissman. the Shape of a species barrier (in prion conformation transformation). 410, pg 161. 223 (2001)

Histamine H4 Receptor

The histamine H4 receptor shows generally restricted expression in immune cells and has been shown to mediate histamine induced-chemotaxis, suggesting that it may play a role in inflammation. This has led to the discovery and characterization of histamine H4 receptor antagonists from multiple chemical scaffolds. These antagonists have shown efficacy in numerous preclinical models of inflammation, further supporting the critical role of the histamine H4 receptor in inflammatory diseases, including rheumatoid arthritis, asthma, pruritis and inflammatory pain. The first histamine H4 receptor antagonists have entered into clinical trials and data are expected soon to establish the importance of targeting this receptor as a novel therapeutic in inflammation. This review focuses on the histamine H4 receptor antagonists that have been identified to date.


Histone H4 lysine 20 acetylation is associated with gene repression in human cells

Histone acetylation is generally associated with gene activation and chromatin decondensation. Recent mass spectrometry analysis has revealed that histone H4 lysine 20, a major methylation site, can also be acetylated. To understand the function of H4 lysine 20 acetylation (H4K20ac), we have developed a specific monoclonal antibody and performed ChIP-seq analysis using HeLa-S3 cells. H4K20ac was enriched around the transcription start sites (TSSs) of minimally expressed genes and in the gene body of expressed genes, in contrast to most histone acetylation being enriched around the TSSs of expressed genes. The distribution of H4K20ac showed little correlation with known histone modifications, including histone H3 methylations. A motif search in H4K20ac-enriched sequences, together with transcription factor binding profiles based on ENCODE ChIP-seq data, revealed that most transcription activators are excluded from H4K20ac-enriched genes and a transcription repressor NRSF/REST co-localized with H4K20ac. These results suggest that H4K20ac is a unique acetylation mark associated with gene repression.

Conflict of interest statement

N.N. is a founder of MAB Institute Inc.

Figures

Figure 1. Detection of H4K20ac by specific…

Figure 1. Detection of H4K20ac by specific antibody and mass spectroscopy.

Figure 2. Distribution of H4K20ac in HeLa-S3…

Figure 2. Distribution of H4K20ac in HeLa-S3 cells.

ChIP-seq analysis was performed using the H4K20ac-specific…

Figure 3. Correlation of H4K20ac with other…

Figure 3. Correlation of H4K20ac with other histone modifications.

( af ) Correlations…

Figure 4. Transcription factor binding sites in…

Figure 4. Transcription factor binding sites in H4K20ac-enriched sequences.


Results

Identification of a set of lncRNAs acting as lnc-eRNAs according to the epigenetic signatures of enhancers

Abnormal expression of lncRNAs has been reported in multiple cancer types [27] however, whether these lncRNAs act as lnc-eRNAs remains unclear. To identify lncRNAs that are transcribed from enhancers as lnc-eRNAs in MLL leukemia, we analyzed the lncRNA expression profile and enhancer epigenetic profile. Starting from a verified set of 128 upregulated lncRNAs from our previous microarray study [28], we first identified that the gene loci of 47 lncRNAs were marked by H3K27ac (Additional file 1: Table S1), a histone mark of active enhancers (as assessed by ChIP-seq data from the MLL leukemia cell line MV4-11), indicating that these lncRNAs might act as lnc-eRNAs. In addition, two other key histone marks (H3K4me1 and H3K4me3) [5] of enhancers were also measured by ChIP-qPCR. The results showed that 41 of 47 candidate lnc-eRNA loci showed high enrichment of H3K4me1 and relatively low enrichment of H3K4me3 (Fig. 1a and Additional file 2: Fig. S1), confirming that these lnc-eRNAs were transcribed from enhancer regions. These lnc-eRNAs were named GENE-Enhancer-LncRNAs (ELAs) according to their predicted target genes. The prediction of lnc-eRNAs target genes was based on previous report [6] and the genomic distance. Briefly, the nearest gene located within 100 kb of lnc-eRNA locus is considered as its target gene. We then selected 36 lnc-eRNAs according to the enrichment level of the epigenetic reader BRD4 of H3K27ac-enriched enhancer and obtained 29 lnc-eRNAs candidates using sensitivity to the BRD4 inhibitor i-BET151 treatment as a filter (Fig. 1a and Additional file 1: Table S1). Further qRT-PCR validation showed that the expression levels of 18 of these 29 lnc-eRNAs were decreased after i-BET151 treatment (Fig. 1b), indicating that a large proportion of upregulated lncRNAs could act as lnc-eRNAs and were transcribed through BRD4 regulation. We classified these 18 lnc-eRNAs as a group of “iBET-responsive lnc-eRNAs” and speculated that they might be involved in MLL leukemia progression.

A set of lncRNAs are identified as lnc-eRNAs and are activated through abnormal enhancer transcription. a Analysis of the gene loci of 128 upregulated lncRNAs with H3K27ac, H3K4me1, H3K4me3 and BRD4 enrichment. b Heatmaps of the qPCR assay results showed the expression levels of lnc-eRNAs in MV4-11 cells treated with i-BET151. Eighteen lnc-eRNAs were downregulated, 8 lnc-eRNAs were unchanged, and 3 lnc-eRNAs were not detected in three independent experiments. c The epigenetic environment of SEELA and its genomic relationship with SERINC2. SEELA is located 71.6 kb downstream of SERINC2, and its gene locus is marked by H3K27ac, BRD4 and POL II in two enhancer regions (as shown in schematic diagrams (gray)). The ChIP-qPCR primers were designed to target the gene body of SEELA (black). d DNA agarose gels showing that there were three and two variants of SEELA1 and SEELA2, respectively, according to the results of 5′ and 3′ RACE assays. *Nonspecific band. e The expression of SEELA was significantly upregulated in patients with MLL leukemia (n = 26) compared with MLL-wt (n = 75) patients (MLL-wt, leukemia patients not harboring MLL gene translocation). (Mann-Whitney test *P < 0.05 ***p < 0.001). The expression level was calculated using the 2 -ΔCT method and was normalized to that of GAPDH. f ChIP-qPCR detection of BRD4, POL II, and MLL fusion protein enrichment at the SEELA1 and SEELA2 gene loci in MV4-11 and RS411 cells after treatment with i-BET151. The error bars indicate the ± SEM values (*P < 0.05 **P < 0.01 ***P < 0.001 ns, no significant difference) in three independent experiments

Lnc-eRNA SEELA is activated through oncoprotein-driven transcription

To further explore the potential roles of lnc-eRNAs, we selected two lnc-eRNAs, SERINC2-ELA1 and SERINC2-ELA2 (named SEELA1 and SEELA2, respectively), that displayed the greatest decrease in expression after i-BET151 treatment (Fig. 1b) for further study. SEELA1 and SEELA2, annotated as LDC1P and LINC01226, respectively, in the human reference genome GRCh38/hg38, are located on chromosome 1p35.2. ChIP-seq data of the large-scale (including SEELA1/2, SERINC2, and two other nearby genes FABP3 and TINAGL1) and zoomed-in regions (only including SEELA1/2 and SERINC2) showed specific enrichment of H3K27ac and BRD4 at the SEELA1/2 locus and POL II enrichment at the SERINC2 and SEELA1/2 loci (Fig. 1c and Additional file 2: Fig. S2A). Furthermore, insertion of the H3K27ac-enriched DNA sequence of SEELA1/2 into the PGL-3 SERINC2 promoter plasmid significantly upregulated luciferase activity (Additional file 2: Fig. S2B). These results are consistent with the idea that SEELA1/2 is transcribed from the active enhancer of the SERINC2 gene [6]. Both SEELA1 and SEELA2 were 5′ m7G capped and 3′ polyadenylated (Additional file 2: Fig. S2C, D). RNA-seq tracks and RACE assays showed that SEELA1 had three variants (SEELA1-V1 to V3) and SEELA2 has two (SEELA2-V1 to V2) in MV4-11 cells, respectively (Fig. 1d and Additional file 2: Fig. S2E). Notably, SEELA2-V1 shared the same 5′ end with all variants of SEELA1, and there was no POL II or H3K27ac enrichment at the transcription start site of SEELA2-V2, suggesting that both SEELA1 and SEELA2 might be transcribed from the same enhancer (enhancer 1) and then processed into different isoforms. Thus, we called these lnc-eRNAs (including three SEELA1 variants and two SEELA2 variants) as the SEELA eRNA cluster (hereafter called SEELA). Notably, SEELA showed the same expression pattern across leukemia cell lines (Additional file 2: Fig. S2F) and we used the primers (Additional file 2: Fig. S2G) to examine the expression of all SEELA1 or SEELA2 variants in the following study. SEELA was upregulated in MLL leukemia patient samples (Fig. 1e). The MLL fusion oncoprotein has been reported to cooperate with the histone acetylation reader BRD4 to control the abnormal transcription of oncogenes [29] thus, we sought to determine whether activation of SEELA is associated with the MLL fusion protein-BRD4 complex. When cells were treated with i-BET151, the binding of the MLL fusion protein, BRD4, and POL II at the SEELA locus was dramatically reduced (Fig. 1f). The MYC and B2M genes were used as positive and negative controls, respectively, according to a previous report [23] (Additional file 2: Fig. S2H). These results indicated that abnormal transcription of SEELA was induced by the MLL fusion-BRD4 complex via recruitment of POL II to the enhancer locus.

SEELA binds histone components to activate the target gene SERINC2 in cis

Since SEELA was transcribed from the enhancer region of SERINC2, we investigated the relationship between SEELA and SERINC2. A positive correlation between SERINC2 and SEELA expression was observed in clinical samples (Fig. 2a, b) and cell lines (Additional file 2: Fig. S3A). This correlation was further validated by modulating the expression of the SEELA (Fig. 2c, d Additional file 2: Fig. S3B, C). Interestingly, knockdown SEELA1 or SEELA2 by siRNA, respectively, could reduce the expression of each other (Additional file 2: Fig. S3D), further supporting that this eRNA cluster is transcribed coordinately. Furthermore, silencing SEELA had little effect on the expression of two other nearby genes (FABP3 and TINAGL1) that were located near the SEELA locus (Additional file 2: Fig. S3E), indicating that SEELA mainly regulated the expression of SERINC2.

Lnc-eRNA SEELA binds histone components to activate target genes in cis. a The expression level of SERINC2 was significantly upregulated in patients with MLL leukemia (n = 26) compared with MLL-wt (n = 75) patients (Mann-Whitney test, **, P < 0.01). The expression level was calculated using 2 -ΔCT method and was normalized to GAPDH. b Positive correlation of SERINC2 with SEELA was observed in patient samples (n = 101). Spearman analysis was used, and r = 0.44, P < 0.001 (upper) and r = 0.36, P < 0.001 (bottom). c qPCR showed that the mRNA levels of SERINC2 were decreased after knocking down SEELA. Error bars reflect ±SD (**P < 0.01 ***P < 0.001) in three independent experiments. d A schematic diagram shows the SEELA sites within their loci targeted by sgRNAs of the CRISPR-VPR activation system (upper). qPCR showed that the mRNA levels of SERINC2 and SEELA were significantly upregulated (bottom). Error bars reflect ±SD (**P < 0.01 ***P < 0.001) in three independent experiments. e The SEELA was enriched in the chromatin fraction of MV4-11 cells. The relative concentration was adjusted according to the chromatin concentration of positive control NEAT1. MALAT1 was another positive control and GAPDH was negative control. Error bars reflect ± SD in three independent experiments. f ChIP-qPCR showed the decreased POL II binding to the SERINC2 promoter after knocking down SEELA. Error bars reflect ±SEM (*P < 0.05) in three independent experiments. An IgG antibody was used as a negative control. g RIP-qPCR detection was used to assess the association of histone components with SEELA in MV4-11 and RS411 cells. Error bars reflect ± SEM in three independent experiments. An IgG antibody acted as a negative control. h, i Immunoblot detection of H3, and H4 retrieved by in vitro-transcribed tRSA-tagged SEELA sections from MV4-11 cell lysates. SEELA1 (SEELA1-V1/2/3) and SEELA2 (SEELA2-V1/2) presented significant enrichment. TRSA and beta-tubulin were used as negative controls. j A schematic diagram shows that upregulated lnc-eRNA SEELA acted in cis to activate SERINC2 transcription via binding with histone components

The majority of eRNAs are located preferentially on chromatin and function locally to regulate the transcription of neighboring genes [13], although the mechanism remains unclear. We firstly investigated whether SEELA functions at the chromatin level. SEELA was retained in the nuclear fraction (Additional file 2: Fig. S3F) and enriched in the chromatin fraction (Fig. 2e). Binding of POL II at the transcription start site of SERINC2 was reduced after knockdown of these lnc-eRNAs (Fig. 2f), indicating that SEELA transcriptionally activated SERINC2 expression in cis. We next sought to determine the mechanism by which the lnc-eRNAs are enriched in the chromatin fraction. Given that histones are the main component of chromatin, we detected whether lnc-eRNAs bind chromatin by interacting with histone components. RIP analysis using anti-histone H3 antibodies showed that SEELA was highly enriched in the histone component immunoprecipitation samples (Fig. 2g Additional file 2: Fig. S3G, H), whereas two other highly expressed lncRNAs, HOTAIRM1 and ENST00000413525, used as negative RNA controls, were not associated with histone components (Additional file 2: Fig. S3I). A tRSA RNA pull-down assay [30] using the full-length sequences of all variants of the two lnc-eRNAs further confirmed that SEELA interacted with histones H3 and H4 (Fig. 2h, i). Collectively, these results suggested that the upregulated SEELA acted in cis by binding with histone components, as shown in Fig. 2j.

SEELA directly binds K31 amino acid of histone H4

We next sought to determine which histone component binds SEELA. As SEELA is transcribed from loci enriched in histone marks H3K27ac and H3K4me1, we speculated that SEELA might bind these two modifications. Unexpectedly, no preferential binding to H3K27ac or H3K4me1 was observed (Additional file 2: Fig. S4A). Additionally, we used SEELA1-variant 1 (V1) to further test its ability to bind to H3K4me1 and H3K27ac in RNA, and no difference was observed in the SEELA1 pull-down assay when we mutated K4, K27, and a control site (K36) from K residues to M residues (Additional file 2: Fig. S4B), suggesting that enhancer modification might not be the essential histone characteristic required for SEELA binding.

We then sought to determine whether lnc-eRNAs interact with specific proteins of histone components. By scanning the available public RNA-binding proteomes [31,32,33,34], we found that four independent studies identified histone H4 (Fig. 3a) as an RNA-binding protein. RIP assays with HA-tagged histone H4 or H3 fragments as bait showed that only the H4 fragment with amino acids 1-50 (H4, 1-50 aa) coprecipitated with SEELA1 (Fig. 3b and Additional file 2: Fig. S4C), implying that SEELA1 binds to histone H4 (1-50 aa). However, a discrepancy was found between the binding of SEELA to truncated 1-50 aa and full-length H4. This difference could be the reason that SEELA is associated with chromatin and that truncated H4 might not incorporate into nucleosomes as efficiently as full-length and, hence, might bind less SEELA than full-length H4. But we could not fully exclude the potential contribution of the C terminus of histone H4 in binding SEELA.

SEELA binds K31 aa of histone H4 to promote the enhancer recognition of histone modification reader. a The reported RNA-binding proteomes identified histone H4 (P62805) as a RNA-binding protein candidate. b A schematic diagram shows the truncated fragments of H3 and H4 (left). RIP-qPCR for SEELA1-V1 pull-down by HA-tagged H3 and H4 sections in 293 T cells (right). The anti-HA antibody was used in RIP-qPCR assays. ce The in vitro tRSA RNA pull-down assay with the purified different mutants of H4 proteins showed that the K31 aa of H4 mainly interacted with SEELA1-V1. f RIP-qPCR detection was used to assess the association of BRD4 with SEELA in MV4-11 and RS411 cells. Error bars reflect ± SEM in three independent experiments. An IgG antibody acted as a negative control. g Immunoblot detection of Flag-tagged BRD4 truncated fragments retrieved by in vitro-transcribed tRSA-tagged SEELA1-V1 from 293 T cell lysates. BRD4 (aa 1-1362, aa 1-722, and aa 466-1362) presented significantly higher enrichment of SEELA1-V1. h ChIP-qPCR assays after knocking down SEELA showed that the enrichments of BRD4 at the SEELA locus were downregulated in MV4-11. Error bars reflect ±SEM (*P < 0.05 **P < 0.01) in three independent experiments. An IgG antibody was used as a negative control in the ChIP assay. i A schematic diagram (left) shows stable overexpressed HA-tagged H4-full length or H4K31A into MV4-11 cells. The binding of SEELA is disrupted in H4K31A mutants, as a result, the missing of the SEELA-induced SERINC2 expression. The expression of SEELA and SERINC2 (right) were downregulated in H4K31A mutant expression cells versus that in the H4-full length expression cells. Error bars reflect ± SD (***P < 0.001) in three independent experiments

In vitro RNA pull-down assays showed that SEELA1, rather than tRSA or SEELA1 antisense RNA, coprecipitated with full-length H4 (Additional file 2: Fig. S4D). Additionally, neither recombinant GST protein nor GST-tagged full-length H3 protein interacted with SEELA1 (Additional file 2: Fig. S4E). Furthermore, the 1-50 aa region of H4 showed high affinity for SEELA1 (Fig. 3c), indicating that SEELA1 directly binds the 1-50 aa of H4. Using the histone H4 truncation mutants (H4-Δ1-24, H4-Δ25-36), we found that the 25-36 aa region of H4 directly interacted with SEELA1 (Fig. 3d). The amino acid sequence of this region is “DNIQGITKPAIR,” and arginine (R) and lysine (K) are amino acids that preferentially bind RNA [35]. Substitution of K31 of H4 with alanine decreased the interaction of H4 with SEELA1, but mutation of either R35 or the control amino acids (R24 and R36) did not affect the binding of H4 to SEELA1 (Fig. 3e). Consistent with this result, the RIP assay showed that mutation of K31 significantly reduced the association between H4 and SEELA1 (Additional file 2: Fig. S4F), suggesting the importance of H4K31 in mediating RNA binding. Among all histone components, the histone H4 protein is highly conserved, and H4K31 is conserved among almost all types of eukaryotes, including nematodes, drosophila, plants, and human (Additional file 2: Fig. S4G). This conservation suggests that the RNA interaction domain of H4 may play a pivotal role in eukaryotic histone function. The nucleosome is an octamer containing two copies of each of the core histone proteins (H2A, H2B, H3, and H4) [36]. H4K31 lies at the N-terminus of the histone H4α1 helix and is exposed on the outer surface of the nucleosome (as shown in Additional file 2: Fig. S4H). The location of H4K31 may provide the space for the SEELA-H4 interaction, which could be a possible mechanism for the specificity of H4K31 in lnc-eRNA binding.

SEELA acts as a modular scaffold to promote enhancer recognition of histone modifiers

The observation above showed that lnc-eRNAs bind histone H4 to achieve specific chromatin localization, which is the premise for their function in activating nearby genes in cis. We next sought to determine how lnc-eRNAs affect enhancer activity to regulate transcription. Enhancer activity is driven by binding of histone modifiers. Notably, the majority of these modifiers and readers, including MLL3/4, CBP/P300, TIP60, CHD7, and BRD4, were identified in previous studies to have RNA binding ability [31,32,33,34, 37, 38]. Thus, we speculated that H4-binding lnc-eRNAs may act as the key mediators to strengthen histone modifier recruitment to affect enhancer activity and oncogene activation. To address this question, we selected BRD4 to verify its interaction with SEELA. BRD4 is the transcriptional regulator of SEELA, making it more spatially accessible to SEELA. We found that BRD4 directly bound to SEELA in the RIP and tRSA pull-down assays (Fig. 3f and Additional file 2: Fig. S5A-D). Specifically, pull-down of SEELA1 with a series of truncated BRD4 fragments showed that SEELA1 might bind the 466-722 aa region of BRD4 (Fig. 3g and Additional file 2: Fig. S5E). Importantly, BRD4 enrichment at enhancer loci was significantly downregulated after knockdown of SEELA1 or SEELA2 (Fig. 3h), suggesting that SEELA bound histone H4 to influence histone modification reading via BRD4 recruitment. Furthermore, we constructed stable MV4-11 cells overexpressing HA-tagged H4-FL or H4K31A. Interestingly, the expression of SEELA and SERINC2, but not FABP3 and TINAGL1, were reduced in H4K31A-expressing cells, indicating that binding of SEELA to H4K31 sustained the expression of SEELA-SERINC2 axis components (Fig. 3i and Additional file 2: Fig. S5F,G). Notably, the SEELA binding site was close to the histone H4 tail, and the acetylated histone H4 tail was found to be a BRD4 binding site, which supported our finding that SEELA acted as a modular scaffold by interacting with the 466-722 aa of BRD4 and K31 aa of histone H4 (Additional file 2: Fig. S5H), implying an important role for lnc-eRNAs. Together, these data suggested that SEELA bound to K31 aa of histone H4 and subsequently induced the occupancy of histone modification reader on chromatin to promote enhancer activity.

SEELA promotes leukemia progression by affecting the transcription of the oncogene SERINC2 and, in turn, regulating oncometabolites

The observations above showed that SEELA modulated enhancer activity to transcriptionally activate SERINC2 expression (Figs. 2 and 3). The SERINC family has been suggested to be involved in the synthesis of serine-derived lipids [39], including sphingolipids, which are frequently reported as oncometabolites in cancer progression [40]. Thus, we hypothesized that SEELA may exert its biological effects by activating SERINC2 expression. To address this question, we performed in vitro and in vivo functional studies. As shown in Fig. 4a–d and Additional file 2: Fig. S6A-C, suppressing SEELA or SERINC2 reduced leukemic cell proliferation and inhibited cell cycle progression. To further demonstrate the cis effect of SEELA on SERINC2 expression and leukemia progress, rescue experiments were performed. We overexpressed SERINC2 in the SEELA knockdown cells (si-SEELA2-1) via doxycycline (Dox) induction and lentiviral systems and found that the proliferation inhibition and cell cycle arrest caused by transfection of SEELA siRNAs were reversed (Fig. 4e, f and Additional file 2: Fig. S6D). Additionally, we introduced an eGFP-tagged SERINC2 protein into si-SEELA cells to distinguish its expression from the endogenous protein and found that it could restore the effect of si-SEELA (Additional file 2: Fig. S6E). These results suggested that SERINC2 is an important downstream target of SEELA. To further investigate whether SEELA exerts its oncogenic effects mainly through modulating SERINC2 expression, we performed RNA-seq in MV4-11 cells transfected with si-NC or si-SEELA2-1. Comparing with si-NC, 57 genes were dysregulated in the si-SEELA samples with absolute value of fold change > 1.5 (|FC| > 1.5), FDR < 0.05. In the si-SEELA samples, SERINC2 was significantly downregulated (log2 fold change = − 1.3, FDR = 8.64 × 10 −10 ), while the other two neighbor genes FABP3 or TINAGL1 did not show obvious variation (Additional file 2: Fig. S6F), consistent with our previous qPCR results. We have also performed RNA-seq in the cells transfected with si-SERINC2-1 in order to see how many dysregulated genes of si-SEELA group are altered in si-SERINC2 group. We found that 42 of 57 dysregulated genes (73.7%) in si-SEELA group were also found to be dysregulated in si-SERINC2 group (|FC| > 1.5, FDR < 0.05) (Fig. 4g and Additional file 3: Table S2), indicating that SEELA and SERINC2 could affect a large number of same downstream gene sets. Based on the RNA-seq data, it can be inferred that SEELA regulates its downstream genes mainly through modulating SERINC2.

SEELA-SERINC2 axis promotes leukemia progression in vitro and in vivo through sphingolipid metabolism. a-d Inhibition of SEELA2 (a, b) and SERINC2 (c, d) disrupted the cell proliferation and induced the accumulation of G0/G1 phase cells via CCK-8 assay and cell cycle assay by flow cytometry in MV4–11 and RS411 cells. Error bars reflect ± SEM (*P < 0.05 **P < 0.01 ***P < 0.001) in three independent experiments. e, f Schematic outline of SERINC2 rescue strategy (e left upper) in MV4-11 (e) and RS411 (f) cells. Western blot showed the protein level of SERINC2 in indicated cells. CCK8 and cell cycle assay showed the cell proliferation and cell cycle arrest of indicated treatment. The error bars indicate the ± SEM values (***P < 0.001) in three independent experiments. g Total 57 genes were dysregulated in the si-SEELA group (|FC| > 1.5, FDR < 0.05), and 42 of 57 genes (73.7%) were dysregulated si-SERINC2 group (|FC| > 1.5, FDR < 0.05). h A schematic diagram shows the xenotransplantation model. i Representative flow cytometry graphs show the decreased percentages of human leukemic GFP+ cells in the bone marrow from mice treated with sh-SEELA, and sh-SERINC2 treated MV4-11 cells relative to the levels observed in control mice (left). A scatter plot shows the statistical values (right). Error bars reflect ± SEM (*P < 0.05). j Kaplan-Meier survival curves show that the mice injected with sh-SEELA and sh-SERINC2 survived longer than those of the control sh-NC group (n = 4). The P values were calculated using a log-rank (Mantel-Cox) test. (*P < 0.05 **P < 0.01). k, l Heatmaps show the expression of sphingolipids from the mass spectrometry (MS) assay. Most of the sphingolipids were upregulated after knocking down SEELA (si-SEELA1-1 or si-SEELA2-1) (k) or SERINC2 (si-SERINC2-1) (l) in MV4-11 cells. Ceramides, Cer sphingomyelins, SM glucosylceramides, GluCer lactosylceramides, LacCer ganglioside, GM2 monosialo-dihexosyl gangliosides, GM3 trihexosylceramide, Gb3 sphingosine, SPH and sphingosine-1-phosphate, S1P. m The survival rate of MV4-11 cells was measured using a CCK-8 kit at 24 h. The sh-NC, sh-SEELA established cells were treated with FTY720 at concentrations from 1 to 9 μM

We further used a NOD-SCID mouse model to explore the functions of SEELA and SERINC2 (Fig. 4h) in MV4-11 cells transfected with different shRNAs (the effect of SERINC2 shRNA is shown in Additional file 2: Fig. S6G). The percentages of human leukemic GFP+ cells were decreased in the bone marrow (Fig. 4i) of mice implanted with sh-SEELA and sh-SERINC2 MV4-11 cells compared with mice in the sh-NC cell group. In addition, mice in the sh-SEELA1 (P < 0.05), sh-SEELA2 (P < 0.05), and sh-SERINC2 (P < 0.01) groups survived longer than those in the control sh-NC group (Fig. 4j), showing that SEELA and SERINC2 affect leukemia progression in vitro and in vivo.

We then investigated whether SEELA-SERINC2 axis could influence sphingolipid synthesis to drive leukemia progress. Unbiased sphingolipid profiling by mass spectrometry (MS) was performed in MV4-11 cells with suppression of SEELA and SERINC2 expression. The altered profile of sphingolipid metabolism in SEELA knockdown cells strongly resembled that in SERINC2 knockdown cells (Fig. 4k, l and Additional file 2: Fig. S7A, B), further indicating that SERINC2 is the essential downstream effector of SEELA that influences sphingolipid synthesis and leukemia progression. Importantly, ceramides (Cer), tumor-suppressor sphingolipids [41], were found to be upregulated when either SEELA or SERINC2 was knocked down (Additional file 2: Fig. S8A, B). Suppression of SEELA expression also resulted in the sensitivity of MV4-11 cells to FTY720 (Fig. 4m and Additional file 2: Fig. S8C-E), an anticancer drug that induces cell death through ceramide accumulation [42, 43] indicating that the SEELA-SERINC2 axis affects leukemia progression by mediating ceramide accumulation.

The enhancer activity of SEELA is programmed via distal binding of HOXA9/10

Lnc-eRNAs have been shown to be transcribed from established active enhancer loci. We finally sought to determine the mechanism by which the enhancer activity at lnc-eRNAs loci is programmed in MLL leukemia. Given that HOXA9, the most important downstream gene in MLL leukemia, was reported to have enhancer-binding preference [44] and to rewire leukemia-specific enhancers [25], we speculated that HOXA genes might mediate the programming of the SEELA enhancer. H3K27ac ChIP-seq identified that three regions located upstream of SEELA1 were marked by H3K27ac. We named these regions SEELA-enhancer1/2/3 (ELA-enhancer1/2/3). ChIP-qPCR showed that HOXA9/10 preferentially bound ELA-enhancer1 (Fig. 5a) rather than ELA-enhancer2/3 or the SEELA locus (Additional file 2: Fig. S9A). Additionally, a 150 bp DNA region within ELA-enhancer1 was enriched with HOXA9, as identified in the ChIP-seq data. Notably, a 14 bp binding motif of HOXA9/10 was found in this region using the JASPAR prediction tool [45] (Additional file 2: Fig. S9B, C). The region contained two HOXA9/10 binding motifs on the two strands due to the opposing position of the DNA strands (Fig. 5b). EMSA was performed to validate the binding of HOXA9 to this motif (Fig. 5c). The levels of H3K27ac and H3K4me1 histone modifications at the SEELA locus were reduced after knockdown of HOXA9 and HOXA10 (Fig. 5d), which was followed by a significant reduction in the expression of both SEELA and SERINC2 but not the other two nearby genes FABP3 and TINAGL1 (Fig. 5e, f and Additional file 2: Fig. S9D). Collectively, these results indicated that HOXA9/10 controls SEELA transcription by programming enhancer activity. To further explore whether HOXA9/10 initiates enhancer-driven activation of global lnc-eRNA loci, we reanalyzed ChIP-seq data of HOXA9 [46] in MV4-11 cells and found that 16 of 18 i-BET-responding lnc-eRNA (Fig. 1b) loci exhibited HOXA9 enrichment (Fig. 5g and Additional file 4: Table S3), implying that most of these lnc-eRNAs could be activated via this HOXA initiation-BRD4 transcriptional activation mechanism and that modulating the expression of these lnc-eRNAs may be an alternative strategy to target the HOXA cluster [47] in MLL leukemia.

HOXA9/HOXA10 binds to the upstream enhancer of SEELA to reprogram the enhancer-related histone mark. a A schematic diagram shows the genomic location and the H3K27ac enrichment of ELA-enhancer 1/2/3 (located at the region 20.5 kb, 38.2 kb, and 59.3 kb upstream of SEELA1, respectively). ChIP-qPCR assays of HOXA9 (middle) and HOXA10 (bottom) showed that HOXA9/HOXA10 bound to ELA-enhancer 1 instead of ELA-enhancer 2/3 Error bars reflect ± SD (*P < 0.05 ***P < 0.001 ns, no significant) in three independent experiments. An IgG antibody was used as negative controls in the ChIP assay. b, c A schematic diagram shows the normal and mutant DNA sequences that are predicted in HOXA9 located (b). EMSA experiments validated the binding motif of HOXA9/10, which was a 14 bp motif (GGTAATTAATTACG) (c). d ChIP-qPCR assay results of H3K27ac and H3K4me1 after knocking down HOXA9 and HOXA10, the enrichment of enhancer histone marks in the SEELA locus were decreased. Error bars reflect ± SEM (*P < 0.05, **P < 0.01, and ***P < 0.001) in three independent experiments. An IgG antibody was used as a negative control in the ChIP assay. e, f qPCR and western blotting showed that the mRNA and protein levels of SEELA and SERINC2 were reduced after knocking down HOXA9 (e) or HOXA10 (f). Error bars reflect ± SD (*P < 0.05, **P < 0.01, and ***P < 0.001) in three independent experiments. g Analysis of the chip-seq data showed that 16 of 18 i-BET responding lnc-eRNAs loci were marked by HOXA9

Collectively, our data suggest a model of the activation and function of lnc-eRNAs in leukemia progression. As shown in Fig. 6, lnc-eRNAs are activated via two steps: the HOXA cluster induces enhancer activity by modulating epigenomic signatures, and BRD4 binds histone acetylation marks to promote lnc-eRNA transcription. The upregulated lnc-eRNA SEELA enhances the chromatin occupancy of histone modifiers by directly interacting with K31 aa of histone H4, thereby promoting enhancer activity and SERINC2 expression to affect sphingolipid metabolism during leukemia progression.

A working model shows the proposed activation and function of lnc-eRNAs


Discussion

Our results suggest that the amino acid sequence similarity among histone H4 genes is maintained primarily by strong purifying selection rather than by concerted evolution. At the nucleotide level, the numbers of synonymous differences between member genes from the same species are generally very large and often near the saturation level. This high level of synonymous differences suggests that H4 genes are subject to birth-and-death evolution at the DNA level and that many genes have persisted in the genome for a long time. This is quite interesting, considering the fact that H4 proteins from distantly related species (e.g., human, trout, and chicken) are identical ( fig. 1 ). This long-term conservation of protein sequences can only be explained by strong purifying selection.

If H4 genes evolve according to the model of birth-and-death evolution under strong purifying selection, pseudogenes may be generated ( Nei and Hughes 1991 Nei, Gu, and Sitnikova 1997 ). Indeed, H4 pseudogenes have been found in X. laevis ( Turner et al. 1983 ), mice ( Liu, Liu, and Marzluff 1987 DeBry 1998 ), humans ( Kardalinou et al. 1993 Albig and Doenecke 1997 ), and Arabidopsis ( Tacchini and Walbot 1995 ). Our analysis of C. elegans genome has suggested that there is at least one H4 pseudogene. Some (i.e., Arabidopsis pseudogene) of these pseudogenes appear to have emerged quite recently, whereas others (e.g., human and C. elegans pseudogenes) seem to be quite old, as shown by the level of sequence divergence from other genes ( table 4 ).

We have shown that members of the histone H4 gene family do not evolve in a concerted manner in long-term evolution. Similar findings have also been reported in the highly conserved histone H3 family ( Rooney, Piontkivska, and Nei 2002 ) and the ubiquitin gene family ( Nei, Rogozin, and Piontkivska 2000 ). Furthermore, the model of birth-and-death evolution applies to many immune system gene families, such as the MHC ( Nei and Hughes 1992 Gu and Nei 1999 ), immunoglobulin ( Ota and Nei 1994 ), TCR ( Su and Nei 2001 ), and ribonuclease genes ( Zhang, Dyer, and Rosenberg 2000 ), as well as other multigene families ( Duda and Palumbi 2000 Robertson 2000 ). It appears that birth-and-death evolution is the major mode of evolution of multigene families in eukaryotes.


Results

Rapid induction of H4 deacetylation by DSBs

Increased and decreased H4 acetylation have both been suggested to occur at various times following DNA damage in different experimental systems ( Bird et al., 2002 Jazayeri et al., 2004 Gupta et al., 2005 Tamburini and Tyler, 2005 Murr et al., 2006 Alao et al., 2009 Miller et al., 2010 Xu et al., 2010). We chose to focus on the first hour of the response to DSBs to minimize the possibility of detecting changes in H4 acetylation or H4K20 methylation due to checkpoint activation or other alterations in cell cycle progression. Focal accumulation of 53BP1 and reduced H4K16ac immunostaining were observed in multiple human cell types 1 h after adding the DSB inducer bleocin (Figure 1A). Decreased H4K16ac was detectable by immunoblotting 30 min after bleocin addition but recovered to pretreatment levels by ∼8 h in the continued presence of bleocin and persistent H2A.X-Ser139 phosphorylation (γH2A.X) (Supplementary Figure S1). The number of 53BP1 foci formed in individual cells was negatively correlated with their residual H4K16ac immunostaining intensity, suggesting that K16 acetylation antagonizes 53BP1 foci formation (Figure 1B). This was also suggested by the finding that bleocin induced less γH2A.X and fewer 53BP1 foci in U2OS cells, a line with higher pretreatment levels of H4K16ac, compared with HeLa cells (Figure 1C and D). Notably, the global levels of H4K20me1/2/3 were not altered by bleocin treatment. Immunoblots with additional antisera suggest that DSBs induce multisite H4 deacetylation (Supplementary Figure S2A). To further investigate the possibility that K16 acetylation antagonizes 53BP1 foci formation, we compared 53BP1 foci formation in HeLa cells expressing wild-type FLAG-H4 or FLAG-H4 with single glutamine substitutions to mimic constitutive acetylation at K5, K8, K12, or K16. 53BP1 foci formation was significantly attenuated in cells expressing the K16Q and K12Q mutants compared with cells expressing either wild type, K5Q, or K8Q FLAG-H4 (Supplementary Figure S2B and C).

Rapid induction of H4K16 deacetylation by DNA DSBs. (A) U2OS and MCF-7 cells were treated with bleocin (10 μg/ml, 1 h) prior to staining with antisera for 53BP1 and H4K16ac. DNA was stained with TO-PRO-3. (B) Bleocin-treated U2OS cells were subdivided into groups exhibiting low, medium, and high nuclear H4K16ac staining intensity after normalization for DNA content. The average number of 53BP1 foci observed in three independent experiments is plotted for each group. (C) HeLa and U2OS cells were treated with bleocin (10 μg/ml, 1 h) and harvested for immunoblotting with antisera to γH2A.X, H4K16ac, H4K20me1, H4K20me2, H4K20me3, and total histone H3. The average relative levels of γH2A.X and H4K16ac observed in three independent experiments are shown in the bar charts. The relative levels of K20me1, 2, and 3 did not differ significantly between bleocin-treated and untreated cells (bar charts not shown). (D) HeLa and U2OS cells were treated with bleocin (10 μg/ml, 1 h) and then stained with antisera to 53BP1. The mean number of 53BP1 foci (±SEM) determined by analyzing >50 cells for each sample are shown in the bar chart. Similar values in the bar charts are indicated by identical letters whereas an asterisk or different letters indicate those which are significantly different (P < 0.05). Scale bars in micrographs = 10 μm. A.U. = arbitrary units.

Rapid induction of H4K16 deacetylation by DNA DSBs. (A) U2OS and MCF-7 cells were treated with bleocin (10 μg/ml, 1 h) prior to staining with antisera for 53BP1 and H4K16ac. DNA was stained with TO-PRO-3. (B) Bleocin-treated U2OS cells were subdivided into groups exhibiting low, medium, and high nuclear H4K16ac staining intensity after normalization for DNA content. The average number of 53BP1 foci observed in three independent experiments is plotted for each group. (C) HeLa and U2OS cells were treated with bleocin (10 μg/ml, 1 h) and harvested for immunoblotting with antisera to γH2A.X, H4K16ac, H4K20me1, H4K20me2, H4K20me3, and total histone H3. The average relative levels of γH2A.X and H4K16ac observed in three independent experiments are shown in the bar charts. The relative levels of K20me1, 2, and 3 did not differ significantly between bleocin-treated and untreated cells (bar charts not shown). (D) HeLa and U2OS cells were treated with bleocin (10 μg/ml, 1 h) and then stained with antisera to 53BP1. The mean number of 53BP1 foci (±SEM) determined by analyzing >50 cells for each sample are shown in the bar chart. Similar values in the bar charts are indicated by identical letters whereas an asterisk or different letters indicate those which are significantly different (P < 0.05). Scale bars in micrographs = 10 μm. A.U. = arbitrary units.

The significance of H4 modifications at DSBs for 53BP1 foci formation

The multiplicity of DSBs generated by bleocin facilitated the detection of H4K16 deacetylation by immunostaining and immunoblotting (Figure 1), but made it difficult to discern whether deacetylation is localized to DSBs or occurs more globally. To distinguish between these alternatives, we used chromatin immunoprecipitation (ChIP) to monitor 53BP1, H4K16ac, and H4K20me2 dynamics at the single DSB generated by I-SceI nuclease in DR-GFP U2OS cells ( Nakanishi et al., 2005). Similar experiments have provided evidence that MMSET facilitates focal accumulation of 53BP1 by locally enhancing the levels of H4K20me1/2/3 at DSBs ( Pei et al., 2011). DR-GFP U2OS cells expressing the glucocorticoid receptor ligand-binding domain fused to wild-type I-SceI or an inactive point mutant of I-SceI were analyzed 1 h after the addition of dexamethasone to induce nuclear translocation of the respective fusion proteins ( Soutoglou et al., 2007). Consistent with the findings of Pei et al. (2011), 53BP1 was enriched at the I-SceI site in cells expressing wild-type I-SceI-GR compared with cells expressing inactive 44N-I-SceI-GR (Figure 2B). We also found that the level of H4K16ac at this site was reduced in cells expressing active I-SceI-GR, as expected from our data shown in Figure 1. These changes appeared to be localized to chromatin flanking the DSB created by I-SceI because they were not observed at a distant site (Sat2 in Figure 2B). In contrast to the findings reported by Pei et al. (2011), we did not detect a significant change in the level of H4K20me2 at either of the sites assayed (Figure 2B). These results suggest that transient, localized H4K16 deacetylation at DSBs facilitates 53BP1 foci formation in the absence of changes in the level of H4K20me2. These findings are further supported by evidence that the level of H4K20me2 within 53BP1 foci resembles that of bulk chromatin, and that 53BP1 foci tend to be excluded from chromatin which is enriched in H4K16ac (Supplementary Figures S3 and S4).

H4K16 acetylation is reduced near a single DSB generated by I-SceI. DR-GFP U2OS cells were transfected with plasmids encoding an inactive mutant (44N) or wild-type (WT) I-SceI-GR fusion for 48 h, followed by dexamethasone treatment (100 nM, 1 h) to induce nuclear translocation. (A) The diagram shows the relative locations of the primers used to assay ChIP DNA (gray arrows) and I-SceI digestion efficiency (black arrows). (B) The relative levels of sceGFP DNA (gray arrows in A) associated with 53BP1, K16ac, and K20me2 as determined by ChIP-qPCR. ChIP-qPCR performed in parallel using Sat2 primers reported on a site distant from the I-SceI cutting site. (C) The percentage of the intact I-SceI site remaining in the ChIP input DNA after transfection and dexamethasone treatment of cells were assayed by PCR using primers that span the I-SceI cutting site (black arrows in A). All values represent mean ± SEM. Asterisks indicate values significantly different from the respective control (P < 0.05). (D) Nonimmune rabbit and mouse immunoglobulins (r-Ig and m-Ig) were used as negative controls in ChIP (solid bar for rabbit open bar for mouse). A single asterisk indicates a significant difference compared to r-Ig, a double asterisk indicates a significant difference compared to m-Ig.

H4K16 acetylation is reduced near a single DSB generated by I-SceI. DR-GFP U2OS cells were transfected with plasmids encoding an inactive mutant (44N) or wild-type (WT) I-SceI-GR fusion for 48 h, followed by dexamethasone treatment (100 nM, 1 h) to induce nuclear translocation. (A) The diagram shows the relative locations of the primers used to assay ChIP DNA (gray arrows) and I-SceI digestion efficiency (black arrows). (B) The relative levels of sceGFP DNA (gray arrows in A) associated with 53BP1, K16ac, and K20me2 as determined by ChIP-qPCR. ChIP-qPCR performed in parallel using Sat2 primers reported on a site distant from the I-SceI cutting site. (C) The percentage of the intact I-SceI site remaining in the ChIP input DNA after transfection and dexamethasone treatment of cells were assayed by PCR using primers that span the I-SceI cutting site (black arrows in A). All values represent mean ± SEM. Asterisks indicate values significantly different from the respective control (P < 0.05). (D) Nonimmune rabbit and mouse immunoglobulins (r-Ig and m-Ig) were used as negative controls in ChIP (solid bar for rabbit open bar for mouse). A single asterisk indicates a significant difference compared to r-Ig, a double asterisk indicates a significant difference compared to m-Ig.

To further investigate the role of H4K20me2 in 53BP1 recruitment, we compared 53BP1 foci formation in HeLa cells depleted of either MMSET alone or SETD8 + SUV420H1 + SUV420H2 in combination (three-in-one) (Supplementary Figure S5). Combined depletion of SETD8 + SUV420H1 + SUV420H2 decreased H4K20me2 levels in immunostaining and immunoblotting, 53BP1 foci formation, and significantly reduced end joining (EJ) activity. In contrast, depletion of MMSET had no effect on these characteristics. The changes in the SETD8 + SUV420H1 + SUV420H2 knockdown cells appear to be attributable to depletion of H4K20me2 rather than nonspecific effects because depletion of these enzymes is not associated with gross alterations in chromatin structure, cell cycle progression, or viability in HeLa or Drosophila S2 cells ( Pesavento et al., 2008 Yang et al., 2008), and we did not observe changes in viability or nuclear morphology in these experiments (Figure 3A and unpublished data). Although it has been reported that de novo formation of H4K20me1 by SETD8 is necessary for 53BP1 foci formation at sites of laser-induced damage ( Oda et al., 2010), we found that accumulation of high levels of H4K20me1 in SUV420H1 + SUV420H2-depleted cells (2-in-1) (Supplementary Figure S7B) ( Schotta et al., 2008 Yang et al., 2008) did not rescue 53BP1 foci formation, H4K20me2 immunostaining, or EJ activity (Supplementary Figures S6 and S7). Together, the data shown in Figure 3 and Supplementary Figures S5–S7 suggest that H4K20me2 established by the SETD8/SUV420H1/SUV420H2 pathway prior to DSB induction by bleocin is more significant for 53BP1 foci formation and NHEJ than H4K20me1 formed by SETD8 or H4K20 methylation purportedly mediated by MMSET.

Depletion of MMSET does not affect the levels of H4K20me2, 53BP1 foci formation, or EJ activity after DSB induction. (A) HeLa cells transfected with siRNA against luciferase (Luc), MMSET (MM), or a combination of siRNA against SETD8 + SUV420H1 + SUV420H2 (3-in-1) for 8 days were treated with bleocin (10 μg/ml, 1 h) and then stained with antisera to 53BP1 and K20me2. (B) The mean number of 53BP1 foci per cell, and the relative intensity of K20me2 staining after normalization for DNA content are shown. (C) HeLa cells were treated with siRNA and bleocin as in A and then assayed for EJ activity (see Supplementary Figure S9 for details). All values represent mean ± SEM. Lettering in the bar charts designates similar values (identical letters) versus those which are significantly different (P < 0.05) (different letters). Scale bars in micrographs = 10 μm. A.U. = arbitrary units.

Depletion of MMSET does not affect the levels of H4K20me2, 53BP1 foci formation, or EJ activity after DSB induction. (A) HeLa cells transfected with siRNA against luciferase (Luc), MMSET (MM), or a combination of siRNA against SETD8 + SUV420H1 + SUV420H2 (3-in-1) for 8 days were treated with bleocin (10 μg/ml, 1 h) and then stained with antisera to 53BP1 and K20me2. (B) The mean number of 53BP1 foci per cell, and the relative intensity of K20me2 staining after normalization for DNA content are shown. (C) HeLa cells were treated with siRNA and bleocin as in A and then assayed for EJ activity (see Supplementary Figure S9 for details). All values represent mean ± SEM. Lettering in the bar charts designates similar values (identical letters) versus those which are significantly different (P < 0.05) (different letters). Scale bars in micrographs = 10 μm. A.U. = arbitrary units.

Concurrent H4K16ac suppresses 53BP1 binding to H4K20me2/1

As most K16 acetylation occurs in conjunction with K20 dimethylation ( Pesavento et al., 2008), we hypothesized that K16 acetylation antagonizes the binding of the 53BP1 tandem Tudor domain (53BP1-TT) to K20me2 on the same molecule of H4. We tested this hypothesis by comparing the interaction of 53BP1-TT with H4 N-terminal peptides that differed in K16 acetylation and K20 methylation (Supplementary Table S4). Consistent with earlier findings ( Botuyan et al., 2006), we found that wild-type 53BP1-TT preferentially bound the K20me2 H4 N-terminal peptide, interacted more weakly with the K20me1 peptide, and displayed negligible binding for the unmodified and K20me3 peptides in an ELISA assay (Figure 4A) and a pull-down assay (Figure 4C). Strikingly, the interaction of 53BP1-TT with the K20me2 and K20me1 peptides was suppressed by concurrent acetylation at K16 (Figure 4A and C), with binding diminished to the background levels observed for the unmodified and K20me3 peptides and a point mutant of 53BP1-TT that abrogates methyl-lysine recognition (Figure 4B) ( Botuyan et al., 2006).

Concurrent acetylation at K16 antagonizes recognition of H4K20me2 by the tandem Tudor domain of 53BP1. The binding of wild-type (A) or W1495A (B) recombinant 6× His-tagged 53BP1 tandem Tudor domain (53BP1-TT) to synthetic H4 peptides (Supplementary Table S3 100 pmol/well) was compared using antisera to the 6× His-tag in an ELISA assay. (C) H4 peptides coupled to iodoalkyl agarose beads were incubated with wild-type 53BP1-TT in a pull-down assay. After washing to remove unbound proteins, bead-associated proteins were eluted, resolved using SDS-PAGE, and 53BP1-TT detected on immunoblots probed with antisera to the 6× His-tag. Single and double asterisks indicate values significantly different (P < 0.05) from the (unmodified) control peptide and from each other, respectively. A.U. = arbitrary units.

Concurrent acetylation at K16 antagonizes recognition of H4K20me2 by the tandem Tudor domain of 53BP1. The binding of wild-type (A) or W1495A (B) recombinant 6× His-tagged 53BP1 tandem Tudor domain (53BP1-TT) to synthetic H4 peptides (Supplementary Table S3 100 pmol/well) was compared using antisera to the 6× His-tag in an ELISA assay. (C) H4 peptides coupled to iodoalkyl agarose beads were incubated with wild-type 53BP1-TT in a pull-down assay. After washing to remove unbound proteins, bead-associated proteins were eluted, resolved using SDS-PAGE, and 53BP1-TT detected on immunoblots probed with antisera to the 6× His-tag. Single and double asterisks indicate values significantly different (P < 0.05) from the (unmodified) control peptide and from each other, respectively. A.U. = arbitrary units.

Additional evidence further substantiates our hypothesis that H4K16 acetylation regulates 53BP1 foci formation. Increasing the levels of H4K16ac by pretreating U2OS cells with media supplemented with 50 or 100-mM sodium acetate ( Wilhelm and McCarty, 1970) reduced 53BP1 foci formation (Figure 5). Expression of FLAG-H4(K16Q) led to a marked reduction in 53BP1 foci formation compared with cells expressing wild-type FLAG-H4 or FLAG-H4(K16R) (Figure 6A and Supplementary Figure S8). Finally, we found that butyrate suppressed 53BP1 foci formation more in cells expressing wild-type FLAG-H4 than in cells expressing FLAG-H4(K16R) (Figure 6B). Taken together, the data shown in Figures 1–6 suggest that localized H4K16 deacetylation facilitates focal accumulation of 53BP1 on pre-existing H4K20me2, and possibly H4K20me1, at DSBs.

Hyperacetylation at H4K16 suppresses 53BP1 foci formation. (A) U2OS cells were maintained in media supplemented with sodium acetate (NaOAc) for 48 h prior to immunoblotting using antisera to K16ac and histone H3 (loading control). The osmotic pressure of all cultures was adjusted to equivalence using sodium chloride (NaCl). (B) Acetate-treated U2OS cells were exposed to bleocin (10 μg/ml, 1 h) and then stained with antisera to K16ac and 53BP1. (C) The mean number of 53BP1 foci per cell (upper panel) and the average K16ac staining intensity (lower panel) are plotted. Asterisks indicate values significantly different from the control cultures (0 mM NaOAc) (P < 0.05). Scale bars in micrographs = 10 μm. A.U. = arbitrary units.

Hyperacetylation at H4K16 suppresses 53BP1 foci formation. (A) U2OS cells were maintained in media supplemented with sodium acetate (NaOAc) for 48 h prior to immunoblotting using antisera to K16ac and histone H3 (loading control). The osmotic pressure of all cultures was adjusted to equivalence using sodium chloride (NaCl). (B) Acetate-treated U2OS cells were exposed to bleocin (10 μg/ml, 1 h) and then stained with antisera to K16ac and 53BP1. (C) The mean number of 53BP1 foci per cell (upper panel) and the average K16ac staining intensity (lower panel) are plotted. Asterisks indicate values significantly different from the control cultures (0 mM NaOAc) (P < 0.05). Scale bars in micrographs = 10 μm. A.U. = arbitrary units.

Loss of H4K16ac dynamics affects 53BP1 recruitment and function. (A) HeLa cells transfected for 48 h with wild-type FLAG-H4 (WT) or the K16Q or K16R point mutants (16Q, 16R) were treated with bleocin (10 μg/ml, 1 h) and then stained with antisera to 53BP1 and the FLAG-tag. The mean number of 53BP1 foci per cell (from >30 cells per group) is plotted in the bar chart. (B) HeLa cells were transfected for 48 h with wild-type FLAG-H4 (WT) or the K16R point mutant and butyrate (10 mM) was applied for the last 16 h. Bleocin was then added (10 μg/ml, 1 h) and cells stained with antisera to 53BP1 and the FLAG-tag. The mean number of 53BP1 foci per cell is plotted in the bar chart. (C) HeLa cells transfected for 48 h with the FLAG-H4 constructs shown were then transfected with linearized pCMV-HA-Venus to assay EJ activity (see Supplementary Figure S9 for assay details). All values represent mean ± SEM. Identical letters in the bar charts signify similar values, and differing letters and asterisks indicate values that differ significantly from the respective controls (P < 0.05). Only cells expressing similar levels of FLAG-H4 were included in the analysis. Scale bars in micrographs = 10 μm. An immunoblot comparing the expression attained for each of the FLAG-H4 constructs is shown in Supplementary Figure S8.

Loss of H4K16ac dynamics affects 53BP1 recruitment and function. (A) HeLa cells transfected for 48 h with wild-type FLAG-H4 (WT) or the K16Q or K16R point mutants (16Q, 16R) were treated with bleocin (10 μg/ml, 1 h) and then stained with antisera to 53BP1 and the FLAG-tag. The mean number of 53BP1 foci per cell (from >30 cells per group) is plotted in the bar chart. (B) HeLa cells were transfected for 48 h with wild-type FLAG-H4 (WT) or the K16R point mutant and butyrate (10 mM) was applied for the last 16 h. Bleocin was then added (10 μg/ml, 1 h) and cells stained with antisera to 53BP1 and the FLAG-tag. The mean number of 53BP1 foci per cell is plotted in the bar chart. (C) HeLa cells transfected for 48 h with the FLAG-H4 constructs shown were then transfected with linearized pCMV-HA-Venus to assay EJ activity (see Supplementary Figure S9 for assay details). All values represent mean ± SEM. Identical letters in the bar charts signify similar values, and differing letters and asterisks indicate values that differ significantly from the respective controls (P < 0.05). Only cells expressing similar levels of FLAG-H4 were included in the analysis. Scale bars in micrographs = 10 μm. An immunoblot comparing the expression attained for each of the FLAG-H4 constructs is shown in Supplementary Figure S8.

DSB-induced H4K16 deacetylation impacts fundamental processes

The suppression of 53BP1:H4K20me2 interactions by K16 acetylation is likely to affect classical NHEJ and other types of EJ ( Dimitrova et al., 2008 FitzGerald et al., 2009). Consistent with this, we found that cells expressing FLAG-H4(K16Q) displayed less EJ activity than cells expressing FLAG-H4(WT) (Figure 6C, Supplementary Figures S8 and S9). EJ was also suppressed in cells expressing FLAG-H4(K16A) or FLAG-H4(K16R), suggesting that normal H4K16 acetylation dynamics may be necessary to attain full EJ activity. This possibility was further supported by the reduced EJ activity observed for U2OS cells exposed to butyrate before and during DSB induction by bleocin (Supplementary Figure S11A). As K16 acetylation has been implicated in regulating gene transcription ( Akhtar and Becker, 2000 Shogren-Knaak et al., 2006), we used BrUTP to label nascent RNA in control and bleocin-treated cells and discovered that bleocin treatment repressed transcription (Supplementary Figure S11B). Transcription by both RNAPI and RNAPII and appear to be affected since BrUTP incorporation was reduced in both nucleolar and extranucleolar chromatin. These effects appear to be mediated, at least in part, by K16 deacetylation because less repression was observed in cells expressing FLAG-H4(K16Q) compared with those expressing either wild-type FLAG-H4 or FLAG-H4(K16R) (Supplementary Figure S11C).


H4. Recent Links and References - Biology

All articles published by MDPI are made immediately available worldwide under an open access license. No special permission is required to reuse all or part of the article published by MDPI, including figures and tables. For articles published under an open access Creative Common CC BY license, any part of the article may be reused without permission provided that the original article is clearly cited.

Feature Papers represent the most advanced research with significant potential for high impact in the field. Feature Papers are submitted upon individual invitation or recommendation by the scientific editors and undergo peer review prior to publication.

The Feature Paper can be either an original research article, a substantial novel research study that often involves several techniques or approaches, or a comprehensive review paper with concise and precise updates on the latest progress in the field that systematically reviews the most exciting advances in scientific literature. This type of paper provides an outlook on future directions of research or possible applications.

Editor’s Choice articles are based on recommendations by the scientific editors of MDPI journals from around the world. Editors select a small number of articles recently published in the journal that they believe will be particularly interesting to authors, or important in this field. The aim is to provide a snapshot of some of the most exciting work published in the various research areas of the journal.


The chemical brothers: nucleosomes and transcription

The positioning of nucleosomes along the DNA with regard to various genetic elements is thought to affect transcription by competing with transcription factors for binding DNA. Using a high-resolution chemical approach to map nucleosomes, Voong et al. provide new insights into the interplay between nucleosomes, transcription and splicing.

Looking to improve on the accuracy of the common nucleosome-mapping method MNase-seq, the authors developed a genome-wide nucleosome-mapping approach that determines nucleosome centre (dyad) positions at nucleotide resolution based on chemical cleavage of the DNA. The method requires substituting Ser47 of histone H4, which flanks the nucleosome centre with a Cys residue (H4 S47C ), which can be covalently bound by a copper-chelating compound. The copper ions direct cleavage of nucleosome DNA near the dyad by hydroxyl radicals, and the resulting DNA fragments are subjected to deep sequencing.

The authors substituted most of the endogenous H4 proteins in mouse embryonic stem (ES) cells with H4 S47C . MNase-seq nucleosome maps generated in H4 S47C and wild-type ES cells, as well as previously generated maps from different organisms, were generally in agreement. They showed the presence of nucleosome-depleted regions (NDRs) upstream of transcription start sites (TSSs) and at transcription termination sites (TTSs) in actively transcribed genes. By contrast, the chemical map revealed generally high nucleosome occupancy spanning the TSS, coding sequence and TTS of actively transcribed genes.

To investigate how nucleosome positioning correlates with RNA polymerase II (Pol II) elongation kinetics, the authors used an available global run-on and sequencing (GRO-seq) data set, from which sites of Pol II accumulation can be inferred. Alignment of Pol II accumulation at promoter-proximal sites with the chemically determined nucleosome positions revealed that occupancy of the +1 position relative to the TSS was positively associated with Pol II pausing.

“the chemical map revealed that nucleosomes are enriched at exon boundaries”

The rate of transcription can influence co-transcriptional splicing. In contrast to recent studies, which found that nucleosomes have higher occupancy around exon centres, the chemical map revealed that nucleosomes are enriched at exon boundaries. Importantly, at all of the expressed genes (regardless of expression levels), Pol II accumulation correlated with nucleosome occupancy at exon boundaries, which is indicative of Pol II stalling at exon–intron junctions close to the nucleosome centre, where the strongest DNA–histone interactions occur.

It is still unclear whether pluripotency transcription factors can bind to their target sites when the DNA is bound by nucleosomes. The chemical map showed that the pluripotency factors OCT4, SOX2, Nanog and Krüppel-like factor 4 bind to their target sites within nucleosomes and modulate nucleosomes in the flanking regions. This suggests that they function as pioneer factors, which can induce chromatin opening and the formation of NDRs.

This study supports a dynamic function of nucleosome in gene regulation. At both promoter-proximal regions and exon–intron junctions, nucleosomes could function as transient barriers for Pol II progression, thereby regulating the kinetics of transcription elongation and splicing.


Results

AS characterizes hESC differentiation

The role of AS in the regulation of ES cell fates adds another notable regulatory layer to the known mechanisms that govern stemness and differentiation [55]. To screen the AS events associated with ES cell fate decision, we investigated a panel of RNA-seq data during hESC (H1 cell line) differentiation [15]. We considered four cell types directly differentiated from H1 cells, including trophoblast-like cells (TBL), mesendoderm (ME), neural progenitor cells (NPC), and mesenchymal stem cells (MSC). We also considered IMR90, a cell line for primary human fetal lung fibroblast, as an example of terminally differentiated cells. These cells represent five cell lineages of different developmental levels (Additional file 1: Figure S1A). We identified thousands of AS events of all types with their changes of “per spliced in” (ΔPSIs) are > 0.1 (inclusion-loss) or < − 0.1 (inclusion-gain), and with the false discovery rates (FDRs) are < 0.05 based on the measurement used by rMATS [56] (Additional file 1: Figure S1B and Table S1, see “Methods”). We implemented further analyses only on the most common AS events, including 3513 MXEs and 3678 SEs, which are referred to as hESC differentiation-associated AS exons (Additional file 1: Figure S1C and Additional file 2: Table S2).

These hESC differentiation-related AS exons possess typical properties, as previously described [57, 58], as follows: (1) most of their hosting genes are not differentially expressed between hESCs and differentiated cells (Additional file 1: Figure S1D) (2) they tend to be shorter with much longer flanking introns compared to the average length of all exons and introns (RefSeq annotation), respectively (Additional file 1: Figure S2A, B) (3) the arrangement of shorter AS exons surrounded by longer introns is consistent across cell lineages and AS types (Additional file 1: Figure S2C, D) and (4) the lengths of AS exons are more often divisible by three to preserve the reading frame (Additional file 1: Figure S2E).

During hESC differentiation, about 20% of expressed genes undergo AS (2257 genes for SE and 2489 genes for MXE), including previously known ESC-specific AS genes, such as the pluripotency factor FOXP1 [13] (Fig. 1a) and the Wnt/β-catenin signalling component CTNND1 [14] (Fig. 1b). These hESC differentiation-related AS genes include many TFs, transcriptional co-factors, chromatin remodelling factors, housekeeping genes, and bivalent domain genes implicated in ESC pluripotency and development [39] (Fig. 1c and Additional file 1: Figure S1C). Enrichment analysis based on a stemness gene set [59] also shows that hESC differentiation-related AS genes are enriched in the regulators or markers that are most significantly associated with stemness signatures of ESCs (Additional file 1: Figure S3A, see “Methods”).

AS characterizes the hESC differentiation. a, b Sashimi plots show two AS events of previously known ESC-specific AS events, FOXP1 (a) and CTNND1 (b). Inset histograms show the PSIs (Ψ) of the AS exons in all cell types based on the MISO estimation. c The bar graph shows that the number of total AS events and lineage-specific AS events increase coordinately with the developmental levels. Higher developmental level induces more (lineage-specific) AS events. MXE.sp. and SE.sp. indicate the percentage of lineage-specific AS events. d Heat maps show the differential “percent splice in” (ΔPSIs) of SE (left) and MXE (right) AS events (rows) for each cell lineage (columns). For MXE event, the ΔPSIs are of the upstream exons. e, f The hosting genes of MXE (e) and SE (f) AS events characterize cell lineages. Black and white bars refer to the common AS genes shared by all cell lineages, while the colour bars indicate the lineage-specific AS genes. The length of the colour bars is proportional to the percentage of lineage-specific genes. Dark fills indicate the inclusion-gain events, while light fills indicate the inclusion-loss events. The numbers in the bars are the proportion of corresponding parts the numbers in the parentheses are the numbers of common AS genes or lineage-specific AS genes of each lineage. Gain or loss for MXE events refers to the upstream exons. Also see Additional file 1: Figures S1–S3

Clustering on AS events across cell lineages show lineage-dependent splicing patterns (Fig. 1d). Upon hESC differentiation, the SE exons tend to lose their inclusion levels (inclusion-loss), while the upstream exons of MXE events are likely to gain their inclusion levels (inclusion-gain) (Fisher’s exact test, p = 3.83E-107). The numbers of AS events increase accordingly with the developmental level following hESC differentiation (Fig. 1c). For example, the differentiation to ME involves the fewest AS events and ME presents the most stem-cell-like AS profiles, while the IMR90 has the most AS events and exhibits the most similar AS profiles to adult cells (Fig. 1c, d). Inter-lineage comparisons show, on average, that 42.0% of SE and 56.4% of MXE events (Fig. 1c, d and Additional file 1: Figure S3B, C), involved in 29.6% and 38.6% of AS hosting genes (Fig. 1e, f and Additional file 1: Figure S3D, E), are lineage-specific. In contrast, only 0.65% of SE and 0.14% of MEX events (Additional file 1: Figure S3B, C), involved in 0.49% and 1.52% of AS hosting genes, are shared by all lineages (Fig. 1e, f and Additional file 1: Figure S3D, E). Similar trends are observed from pairwise comparisons (Additional file 1: Figure S3F). Furthermore, one-third of AS genes (n = 881) have both MXE and SE events (Additional file 1: Figure S3G). Only four genes are common across all cell lineages and AS types, of which the AS events of Ctnnd1 and Mbd1 have been reported to regulate mESC differentiation [14]. Together, these results demonstrate that AS depicts lineage-dependent and developmental level-dependent characterizations of hESC differentiation.

Dynamic changes of HMs predominantly occur in AS exons

In ESCs, epigenetic mechanisms contribute mainly to maintaining the expression of pluripotency genes and the repression of lineage-specific genes in order to avoid exiting from stemness. Upon differentiation, epigenetic mechanisms orchestrate the expression of developmental programs spatiotemporally to ensure the heritability of existing or newly acquired phenotypic states. Though epigenetic signatures are mainly found to be enriched in promoters and enhancers, it has become increasingly clear that they are also present in gene bodies, especially in exon regions, implying a potential link of epigenetic regulation to pre-mRNA splicing [60, 61]. Consistent with previous reports [36, 37, 62], we also observed that few involved SFs are differentially expressed during H1 cells differentiation (Additional file 1: Figure S3H, see “Methods”), which confirms the existence of an additional layer of epigenetic regulations on AS. However, the extents to which the AS is epigenetically regulated and how these AS genes contribute to the cell fate decision are poorly understood. We focused on 16 HMs, including nine histone acetylation and seven histone methylation that have available data in all six cell types (see “Methods”) and aimed to reveal their associations with AS genes during hESC differentiation.

To investigate whether the dynamic changes of these HMs upon cell differentiation prefer the AS exons consistently (Fig. 2a, b), we profiled the differential HM patterns of around the hESC differentiation-associated AS exons and the same number of randomly selected constitutive splicing (CS) exons of the same AS genes for each differentiation lineage. We compared the changes of ChIP-seq reads count (normalized Δ reads count, see “Methods”) in ± 150-bp regions around the splice sites upon hESC differentiation (Fig. 2c and Additional file 1: Figure S4, see “Methods”). Except for a small part of cases (with black dots or boxes in Fig. 2d), most HMs changed more significantly around AS exons than around constitutive exons upon hESC differentiation (Mann–Whitney–Wilcoxon test, p ≤ 0.05, Fig. 2d and Additional file 1: Figure S4). Nevertheless, some HMs displayed strong links to AS, such as H3K79me1 and H3K36me3, while others only had weak link strengths, such as H3K27me3 and H3K9me3 (Fig. 2d). This result is consistent with the fact that the former are involved in active expression and AS regulation [38, 44, 63], while the latter are the epigenetic marks of repressed regions and heterochromatin [64]. The link strengths are presented as the -log10 p values to test whether the HM changes consistently prefer the AS exons across different cell lineages and AS types (Fig. 2d sidebar graph, see “Methods”). Taken together, these results, from a global view, revealed a potential regulatory link from HMs to RNA splicing, of which some are strong while the others are weak.

Dynamic changes of HMs predominantly occur in AS exons. a, b Genome browser views of representative H3K36me3 changes in MXE (exemplified as FGFR2) and SE (exemplified as CDC27) events, respectively, showing that the changes of H3K36me3 around the AS exons (blue shading) are more significant than around the flanking constitutive exons (gray shading) in 4 H1-derived cell types and IMR90. The tracks of H1 are duplicated as yellow shadings overlapping with other tracks of the derived cells (green) for a better comparison. c Representative profiles of HM changes (normalized Δ reads number) around the AS exons and randomly selected constitutive splicing (CS) exons upon hESC differentiation, shown as the average of all cell lineages pooled together. The ± 150-bp regions (exons and flanking introns) of the splice sites were considered and 15 bp-binned to produce the curves. It shows that the changes of HMs are more significant around AS exons than around constitutive exons, especially in exonic regions (gray shading). The p values, Mann–Whitney–Wilcoxon test. d The statistic significances for changes of all 16 HMs in all cell lineages and pooling them together (pooled), represented as the -log10 p values based on Mann–Whitney–Wilcoxon test. The detailed profiles are provided in Additional file 1: Figure S4. Black boxes indicate the cases that HMs around constitutive exons change more significantly than around AS exons, corresponding to the red-shaded panels in Additional file 1: Figure S4. Sidebars represent the significances whether the changes of HMs are consistently enriched in AS exons across cell lineages, showing the link strength between AS and HMs and represented as the -log10 p value based on Fisher’s exact test. The yellow vertical line indicates the significance cutoff of 0.05. Also see Additional file 1: Figure S4

Three HMs are significantly associated with AS upon hESC differentiation

To quantitatively associate the HMs with AS, all ChIP-seq data were processed for narrow peak calling using MACS2 [65]. For each AS exon of each differentiation lineage, we then quantified the differential inclusion levels, i.e. the changes of “percent splice in” (ΔPSIs, Additional file 1: Figure S1B), and the differential HMs signals, i.e. the changes of normalized narrow peak height of ChIP-seq (ΔHMs, Additional file 1: Figure S5A, see “Methods”) between H1 and differentiated cells. We observed significant differences in all HM profiles (except H3K27me3, Additional file 1: Figure S5B) between the inclusion-gain and inclusion-loss exons across cell lineages and AS types (Mann–Whitney–Wilcoxon test, p ≤ 0.05) (Fig. 3a and Additional file 1: Figure S5B). However, three independent correlation tests showed only weak global quantitative associations between the ΔPSIs and ΔHMs for some HMs (Fig. 3c and Additional file 1: Figure S5C), including eight HMs for MXE AS exons and eight HMs for SE AS exons. The weak associations may indicate that only subsets of AS exons are strongly associated with HMs and vice versa, which is consistent with a recent report [66].

A subset of HMs and AS are strongly associated upon hESC differentiation. a Representative profiles of HM (H3K36me3) changes (normalized Δ reads number) around the inclusion-gain (red lines) and inclusion-loss (blue lines) AS exons, as well as randomly selected constitutive splicing (CS) exons (black lines) for both MXE (left) and SE (right) AS events. It shows that HM changes are significantly different between inclusion-gain and inclusion-loss AS exons (p values, Mann–Whitney–Wilcoxon test). Additional file 1: Figure S5B provides the whole significances of all HMs across AS types and cell lineages. b Pearson correlation test between differential HM signals (ΔHMs) and differential inclusion levels (ΔPSIs), taking H3k36me3 as an example. Additional file 1: Figure S5C provides the correlation test results of other HMs based on two more tests. c A representative k-means cluster shows a subset of SE AS events having a negative correlation between the ΔPSIs and the ΔHMs of H3K36me3. Additional file 1: Figures S5D and S6 provide all the clustering results. d Scatter plot shows that HM-associated AS events display significant correlations between the ΔPSIs and the ΔHMs upon hESC differentiation, taking H3K27ac–associated (positively) MXE events as an example. Also see Additional file 1: Figures S5, S6

To explore the subsets of highly associated AS exons and corresponding HMs, we performed k-means clustering on the sets of inclusion-gain and inclusion-loss exons of SE and MXE events, separately, taking the ΔHMs of eight identified HMs as epigenetic features (Fig. 3c and Additional file 1: Figures S5D and S6, see “Methods”). We obtained three subsets of HM-associated SE exons and three subsets of HM-associated MXE exons (Additional file 3: Table S3). The three HM-associated SE subsets include 180, 664, and 1062 exons and are negatively associated with H4K8ac (Additional file 1: Figure S6), negatively associated with H3K36me3 (Fig. 3c), and positively associated with H3K36me3 (Additional file 1: Figure S6), respectively. The three HM-associated MXE subsets include 99, 821, and 971 exons and are positively associated with H3K27ac (Fig. 3d), negatively associated with H3K36me3 (Additional file 1: Figure S6), and positively associated with H3K36me3 (Additional file 1: Figure S6), respectively. The exons of each subset show significant correlations between their ΔPSIs and ΔHMs upon hESC differentiation (Fig. 3d). These HM-associated AS exons account for an average of 52.8% of hESC differentiation-related AS events, on average (Additional file 1: Figure S5E).

Of the three AS-associated HMs, H3K36me3 has both positive and negative correlations with AS exons. This is consistent with the fact that H3K36me3 has dual functions in regulating AS through two different chromatin-adapter systems, PSIP1/SRSF1 [45] and MRG15/PTBP1 [44]. The former increases the inclusion levels of targeting AS exons, whereas the latter decreases the inclusion levels [38]. As expected, 139 and 11 of our identified H3K36me3-associated AS genes have been reported to be regulated by SRSF1 [67, 68] (Additional file 1: Figure S5F) and PTBP1 [69] (Additional file 1: Figure S5G), respectively. Taken together, our analysis showed that more than half (52.8%) of hESC differentiation-associated AS events are significantly associated with three of 16 HMs during hESC differentiation, including H3K36me3, H3K27ac, and H4K8ac.

HM-associated AS genes predominantly function in G2/M phases to facilitate hESC differentiation

Epigenetic mechanisms have been proposed to be dynamic and play crucial roles in human ESC differentiation [15, 16]. Given the aforementioned associations between HMs and AS, and the well-established links between AS and hESC differentiation, we hypothesized that the three HMs (H3K36me3, H3K27ac, and H4K8ac) may contribute to stem cell differentiation through their associated AS events. To test our hypothesis and gain more insights into the differences between the HM-associated and HM-unassociated AS events, we performed comparative function analyses between their hosting genes, revealing that HMs are involved in alternatively splicing the core components of cell-cycle machinery and related pathways to regulate stem cell pluripotency and differentiation.

We found that HMs prefer to be associated with even shorter AS exons (Additional file 1: Figure S7A, p < 0.001, Student’s t-test), though AS exons are shorter than the average length of all exons (Additional file 1: Figure S2A). HM-associated genes (n = 2125) show more lineage specificity, i.e. more genes (49.76% vs 29.6% of MXE or 38.6% of SE genes) are lineage-specific (Additional file 1: Figures S7B and S3D, E), regardless of whether IMR90 is included or not (Additional file 1: Figure S7C). Only a few HM-associated genes are shared by different cell lineages, even in pairwise comparisons (Additional file 1: Figure S7D) the most common shared genes are lineage-independent housekeeping genes (Additional file 1: Figure S7E). These suggest that HM-associated AS genes contribute more to lineage specificity. In addition, the HM-associated AS genes (966 of 2125) are more enriched in stemness signatures than unassociated AS genes (429 of 1057) (Fig. 4a). TF binding enrichment analysis shows that HM-associated AS genes are likely to be regulated by TFs involved in cell differentiation, whereas HM-unassociated AS genes are likely to be regulated by TFs involved in cell proliferation and growth (Fig. 4b). All these results suggest that HM-associated and HM-unassociated AS genes function differently during hESC differentiation.

HM-associated AS genes predominantly function in G2/M cell-cycle phases contributing to hESC differentiation. a HM-associated AS genes are enriched more significantly in stemness signatures than HM-unassociated AS genes. b TF binding enrichment shows that HM-associated AS genes prefer to be regulated by TFs involved in cell differentiation, while the HM-unassociated AS genes are prone to be regulated by TFs involved in cell proliferation and growth. c GO enrichment analysis shows that HM-associated AS genes are enriched more significantly in cell-cycle progression than HM-unassociated AS genes, shown as the -log10 p values after FDR (≤ 0.05) adjustment. d The significant enrichment of HM-associated AS genes in the cell cycle are consistent across cell lineages, with the MSC as an exception that no significant enrichment was observed. e The top 20 enriched functions show that HM-associated AS genes involved in cell-cycle progression prefer to function in G2/M phases and DNA damage response. f The canonical pathway enrichment shows that AMT/ATR-mediated DNA damage response is the top enriched pathway of HM-associated AS genes. The vertical lines (yellow) indicate the significance cutoff of 0.05. Also see Additional file 1: Figures S7, S8

Gene Ontology (GO) enrichment analysis shows that more than half of the HM-associated AS genes (1120 of 2125) function in cell-cycle progression and exhibit more significant enrichment than do HM-unassociated AS genes (376 of 1057, Fig. 4c, d and Additional file 1: Figure S8A). The significance of the top enriched GO term (GO:0007049, cell cycle) is consistent across cell lineages, although HM-associated AS genes exhibit more lineage specificity and few of them are shared among lineages (Additional file 1: Figures S7B–D and S8B). These results suggest the involvement of HMs in AS regulation of the cell-cycle machinery that has been reported to be exploited by stem cells to control their cell fate decision [20].

Further study of the top enriched cell-cycle AS genes (Fig. 4d and Additional file 1: Figure S8A) shows that HM-associated (n = 282) and HM-unassociated AS genes (n = 150) play roles in different cell-cycle phases and related pathways. The former is prone to function in G2/M phases and DNA damage response (Fig. 4e, f). This indicates that HMs contribute to cell differentiation, at least partially, via AS regulations in these phases, which is consistent with the fact that inheritance of HMs in daughter cells occurs during the G2 phases [20]. The latter play roles in G1 phase, cell-cycle arrest, and Wnt/β-catenin signalling (Additional file 1: Figure S8C, D). Since cell fate choices seem to occur or at least be initiated during G1/S transition [53], while cell fate commitment is achieved in G2/M [54], it could be rational for stem cells to change their identity during the G2 phase when HMs are reprogrammed [20].

Intriguingly, the top enriched pathway of HM-associated AS genes is “ATM/ATR-mediated DNA damage response,” which is activated in S/G2 phases and has been recently reported as a gatekeeper of the pluripotency state dissolution (PSD) that participates in allowing hESC differentiation [54]. Together with our previous results [19], it suggests the presence of a combinational mechanism involving HMs and AS, wherein HMs facilitate the PSD and cell fate commitment by alternatively splicing the key components of the ATM/ATR pathway. Additionally, many cell-cycle TF genes are involved in the top enriched HM-associated AS gene set. The pre-B-cell leukaemia transcription factor 1 (PBX1) is one of these genes that contribute to cell-cycle progression and is discussed later in next section. Taken together, we suggest that three of 16 HMs function in positive or negative ways affect the AS of subsets of genes and further contribute to hESC differentiation in a cell-cycle phase-dependent manner. The results suggest a potential mechanistic model connecting the HMs, AS regulations, and cell-cycle progression with the cell fate decision.

Splicing of PBX1 links H3K36me3 to hESC fate decision

The past few years have identified key factors required for maintaining the pluripotent state of hESCs [70, 71], including NANOG, OCT4 (POU5F1), SOX2, KLF4, and c-MYC, the combination of which was called Yamanaka factors and sufficient to reprogram somatic cells into induced pluripotent stem cells (iPSCs) [72]. These factors appear to activate a transcriptional network that endows cells with pluripotency [73]. The above integrative analyses showed strong links between three HMs and RNA splicing, revealing a group of epigenetic regulated AS genes involved in cell-cycle machinery. PBX1 was one of the genes that their ASs are positively associated with H3K36me3 (Fig. 5a, b). Its protein is a member of the TALE (three-amino acid loop extension) family homeodomain transcription factors [74, 75] and well-known for its functions in lymphoblastic leukaemia [76,77,78,79] and several cancers [80,81,82,83,84,85,86,87,88,89]. PBX1 also plays roles in regulating developmental gene expression [90], maintaining stemness and self-renewal [80, 91, 92], and promoting the cell-cycle transition to the S phase [93]. Additionally, multiple lines of evidence obtained from in vivo and in vitro highlighted its functions as a pioneer factor [86, 94]. However, few studies have distinguished the distinct functions of its different isoforms.

Isoform switch from PBX1a and PBX1b during hESC differentiation. a Genome browser view shows the AS event and H3K36me3 signals of PBX1 upon hESC differentiation. The green horizontal bars below the ChIP-seq tracks indicate the narrow peaks called by MACS2. b The inclusion level for exon 7 of PBX1 is significantly correlated to the H3K36me3 signals over this exon across cell lineages. c The sequence difference of three protein isoforms of PBX1 and the main functional domains. d The relative expressions of PBX1a and PBX1b in 56 cells/tissues, representing the differential expressions of two isoforms in three groups based on their developmental states. e The expression levels of NANOG and OCT4 genes are negatively correlated with the expression of PBX1b. f The expression levels of PSIP1 and SRSF1 show significant positive correlations with the expression level of PBX1a. Also see Additional file 1: Figures S9, S10

PBX1 has three isoforms [95], including PBX1a, PBX1b, and PBX1c (Fig. 5c and Additional file 1: Figure S9A). PBX1a and PBX1b are produced by the AS of exon 7 (Fig. 5a) and attract most of the research attention of PBX1. PBX1b retains the same DNA-binding homeodomain as PBX1a, but changes 14 amino acids (from 334 to 347) and truncates the remaining 83 amino acids at the C-terminus of PBX1a (Fig. 5c and Additional file 1: Figure S9A). This C-terminal alteration of PBX1a has been reported to affect its cooperative interactions with HOX partners [96], which may impart different functions to these two isoforms. We here revealed its H3K36me3-regulated isoform switch between PBX1a and PBX1b, which functions at the upstream of pluripotency transcriptional network to link H3K36me3 with ESC fate decision.

We first observed differential transcript expressions of these two isoforms between the hESCs and differentiated cells, wherein PBX1a was predominantly transcribed in hESCs, while PBX1b was predominantly induced in differentiated cells (Fig. 5a and Additional file 1: Figure S9B). The same trend was also observed in an extended dataset of 56 human cell lines/tissues (Fig. 5d) from the Roadmap [97] and ENCODE [98] projects (Additional file 4: Table S4). Additionally, we did not observe significantly different expression of the total PBX1 and three other PBX family members across cell types (Additional file 1: Figure S9C, fold change < 2), indicating that the isoform switch of PBX1, rather than the differential expression of its family members, plays more important roles during hESC differentiation. To further test the possible mechanism by which PBX1b contributes to stem cell differentiation, we investigated the transcription levels of Yamanaka factors. Of these TFs, the NANOG is activated by direct promoter binding of PBX1 and KLF4, which is essential for stemness maintenance [91, 99]. Consistently, all these core factors are repressed in differentiated cells where PBX1b is highly expressed (Additional file 1: Figure S9D–G), even though the PBX1a is expressed. Based on the 56 human cell lines/tissues, we also observed significant negative correlations between expression of most important pluripotent factors (NANOG and OCT4) and PBX1b (Fig. 5e), as well as positive correlations between these two factors and PBX1a (or inclusion level of exon 7, Additional file 1: Figure S10A, B). Consistent with previous reports showing that the PBX1a and PBX1b differ in their ability to activate or repress the expression of reporter genes [100, 101], we hypothesize that PBX1a promotes the activity of the pluripotent regulatory network by promoting the expression of NANOG, whereas PBX1b may attenuate this activity by competitively binding and regulating the same target gene, since PBX1b retains the same DNA-binding domain as PBX1a. These observations are strongly suggestive that the switch from PBX1a to PBX1b is a mechanism by which PBX1 contributes to hESC differentiation via regulating the pluripotency regulatory network.

Exon 7 of PBX1 shows significantly positive correlations between its inclusion levels (PSIs) and the surrounding epigenetic signals of H3K36me3 in hESCs and differentiated cells (Fig. 5b). It suggests a potential role of H3K36me3 in regulating the isoform switch between PBX1a and PBX1b. To investigate the regulatory role of H3K36me3, we focused on two previously proved chromatin-adaptor complexes, MRG15/PTBP1 [44] and PSIP1/SRSF1 [45], which endow dual functions to H3K36me3 in AS regulation [38]. Based on the 56 cell lines/tissues from the Roadmap/ENCODE projects, we first found significant positive correlations between the expressions of PBX1a (or inclusion level of exon 7) and PSIP1/SRSF1 (Fig. 5f), but not with MRG15/PTBP1 (Additional file 1: Figure S10C, D). This result suggests that the AS of PBX1 is epigenetic regulated by H3K36me3 through the PSIP1/SRSF1 adaptor system, which was strongly supported by a recent report using the HeLa cell lines [67]. The overexpression of SRSF1 in Hela cells introduces a PSI increase of 0.18 for exon 7 of PBX1 (chr1: 164789308–164,789,421 based on NCBI37/hg19 genome assembly) based on the RNA-seq (Table S1 of [67]). Additionally, this exon was one of the 104 AS exons that were further validated using radioactive reverse transcription polymerase chain reaction (RT-PCR) (Table S2 of [67]). Their results showed that exon 7 of PBX1 is indeed a splicing target of SRSF1, supporting our conclusions.

We then validated the above hypotheses on MSCs and IM90 cells, since these two cells types show the most significant difference from H1 cells regarding our hypotheses (Fig. 5b). We cultured H1 cells, IMR90 cells, and induced H1 cells to differentiate into MSCs (H1-MSCs, see “Methods” for details). Additionally, we also included other two sources of MSCs, including one derived from human bone marrow (hBM-MSCs) and the other derived from adipose tissue (hAT-MSCs) (see “Methods” for details). Consistent with the results from RNA-seq, the same expression patterns of Yamanaka factors in H1, MSCs, and IMR90 cells were observed using quantitative RT-PCR (qRT-PCR) and western blot (Fig. 6a), which confirmed the pluripotent state of H1 cells and the differentiated states of other cell types. We then detected the isoform switch from PBX1a to PBX1b in our cultured cells, which are consistent both in mRNA and protein levels (Fig. 6b and Additional file 1: Figure S10E) and further confirmed by the western blot using PBX1b-specific antibody (anti-PBX1b) (Fig. 6b bottom and Additional file 1: Figure S10E iii). These results have verified that the PBX1b was significantly induced in differentiated cells, where the PBX1a was significantly reduced.

Isoform switch of PBX1 links H3K36me3 to hESC fate decision. a qRT-PCR and western blot show the expression levels of Yamanaka factors in H1, MSC, and IMR90 cells. Whiskers denote the standard deviations of three replicates. b RT-PCR and western blot show the isoform switches between PBX1a and PBX1b from H1 cells to differentiated cells. c i. ChIP-PCR shows the differential binding of PBX1b to NANOG promoter in H1 cells and differentiated cells ii. ChIP-PCR shows the reduced H3K36me3 signal in differentiated cells iii. ChIP-PCR shows the differential recruitment of PSIP1 to exon 7 of PBX1. d RIP-PCR show the differential recruitment of SRSF1 around exon 7 of PBX1. e Co-IP shows the overall physical interaction between PSIP1 and SRSF1 in all studied cell types. f The mechanism by which H3K36me3 is linked to cell fate decision by regulating the isoform switch of PBX1, which functions upstream of the pluripotency regulatory network. Also see Additional file 1: Figures S9, S10

We also validated the mechanism by which the splicing of PBX1 links H3K36me3 to stem cell fate decision. We first confirmed that PBX1b also binds to the promoter of NANOG at the same region where PBX1a binds to and the binding signals (ChIP-PCR) were high in the differentiated cells but very low in H1 stem cells (Fig. 6c i and Additional file 1: Figure S10F i). Consistent with the results from ChIP-seq, we also observed reduced H3K36 tri-methylation around exon 7 of PBX1 based on ChIP-PCR assay (Fig. 6c ii and Additional file 1: Figure S10F ii). Furthermore, the chromatin factor PSIP1 only showed high binding signal in H1 stem cells (Fig. 6c iii and Additional file 1: Figure S10F iii), which recruit the SF SRSF1 to the PBX1 exclusively in H1 stem cells (Fig. 6d and Additional file 1: Figure S10G) even though the physical binding between these two factors were universally detected in all cell types (Fig. 6e). All these experimental results suggested that, upon differentiation, stem cells reduced the H3K36 tri-methylation and may attenuate the recruitment of PSIP1/SRSF1 adaptor around exon 7 of PBX1, leading to the exclusion of exon 7 and highly expressed PBX1b in differentiated cells. High expression of PBX1b may attenuate the activity of PBX1a in promoting the pluripotency regulatory network.

Taken together, we suggested that H3K36me3 regulates the AS of PBX1 via the PSIP1/SRSF1 adaptor system, leading the isoform switch from PBX1a to PBX1b during hESC differentiation. Subsequently, PBX1b competitively binds to NANOG and abolishes the bindings of PBX1a. This competitive binding attenuates the pluripotency regulatory network to repress self-renewal and consequently facilitate differentiation (Fig. 6f). These findings revealed how the PBX1 contributes to cell fate decision and exemplify the mechanism by which AS links HMs to stem cell fate decision.


Histone H4 directly stimulates neutrophil activation through membrane permeabilization

Kevan L. Hartshorn MD, Department of Medicine, Section of Hematology Oncology, Boston University School of Medicine, Boston, MA, USA.

Department of Medicine, Section of Hematology Oncology, Boston University School of Medicine, Boston, Massachusetts, USA

Department of Medicine, Section of Hematology Oncology, Boston University School of Medicine, Boston, Massachusetts, USA

Department of Medicine, Section of Hematology Oncology, Boston University School of Medicine, Boston, Massachusetts, USA

Department of Medicine, Section of Hematology Oncology, Boston University School of Medicine, Boston, Massachusetts, USA

Kevan L. Hartshorn MD, Department of Medicine, Section of Hematology Oncology, Boston University School of Medicine, Boston, MA, USA.

Abstract

Extracellular histones have been implicated as a cause of tissue inflammatory injury in a variety of disorders including sepsis, lung, and liver diseases. However, little is known about their interactions with neutrophils and how this might contribute to injury. Here, it is shown that histone H4 acts as neutrophil activator by inducing hydrogen peroxide production, degranulation, cell adhesion, and IL-8 generation. Histone H4 caused permeabilization of the neutrophil membrane (a phenomenon described in other cell types) leading to accelerated cell death. H4 caused sustained rise in neutrophil intracellular calcium that is necessary for respiratory burst activation and degranulation. Convincing evidence was not found for TLRs or ATP receptors in H4 mediated activation. However, pertussis toxin and wortmannin (inhibitors of G protein and PI3K) inhibited H4-induced hydrogen peroxide production and degranulation. These studies suggest that release of histone H4 in sites of infection or inflammation may potentiate neutrophil activation and promote additional inflammatory responses. These studies may provide a better basis for developing novel therapeutic strategies to block neutrophil extracellular trap (NET) and H4-related pathology in sepsis and various forms of lung injury including that induced by viruses like influenza or SAR-CoV2.


Watch the video: Connecting links in Evolution #shorts 1 minute Biology. biologyexams4u (October 2022).