(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



MYCN drives oncogenesis by cooperating with the histone methyltransferase G9a and the WDR5 adaptor to orchestrate global gene transcription [1]

['Zhihui Liu', 'Pediatric Oncology Branch', 'National Cancer Institute', 'Bethesda', 'Maryland', 'United States Of America', 'Xiyuan Zhang', 'Man Xu', 'Jason J. Hong', 'Amanda Ciardiello']

Date: 2024-04

MYCN activates canonical MYC targets involved in ribosome biogenesis, protein synthesis, and represses neuronal differentiation genes to drive oncogenesis in neuroblastoma (NB). How MYCN orchestrates global gene expression remains incompletely understood. Our study finds that MYCN binds promoters to up-regulate canonical MYC targets but binds to both enhancers and promoters to repress differentiation genes. MYCN binding also increases H3K4me3 and H3K27ac on canonical MYC target promoters and decreases H3K27ac on neuronal differentiation gene enhancers and promoters. WDR5 facilitates MYCN promoter binding to activate canonical MYC target genes, whereas MYCN recruits G9a to enhancers to repress neuronal differentiation genes. Targeting both MYCN’s active and repressive transcriptional activities using both WDR5 and G9a inhibitors synergistically suppresses NB growth. We demonstrate that MYCN cooperates with WDR5 and G9a to orchestrate global gene transcription. The targeting of both these cofactors is a novel therapeutic strategy to indirectly target the oncogenic activity of MYCN.

Data Availability: All the home generated RNA-seq and ChIP-seq can be found in the Gene Expression Omnibus (GEO) database. GEO accession number for data generated in this study is GSE208424. RNA-seq of MYCN silencing for 72 h in IMR32 and LAN5 can be found under GEO accession number GSE183641. GEO accession number for publicly available ChIP-seq data derived from BE2C cells is GSE94822. GEO accession number for publicly available ChIP-seq data of MYCN, and histone marks derived from rhabdomyosarcoma cell line RH4 is GSE83728.

This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

To investigate how MYCN globally regulates gene transcription, we combined a protein interactome assay and genome-wide approaches including RNA sequencing (RNA-seq) and chromatin immunoprecipitation followed by DNA sequencing (ChIP-seq). Our study demonstrates that WDR5 assists MYCN to bind promoters to activate canonical MYC targets, whereas MYCN recruits G9a to enhancers to repress neuronal differentiation genes in NB. The simultaneous targeting of both WDR5 and G9a-regulated transcriptional activities is a more effective approach to indirectly target MYCN and its oncogenic program.

Most eukaryotic TFs act by recruiting coactivators, or corepressors, which include chromatin remodeling complexes and covalent histone-modifying complexes [ 12 ]. Only a handful of coactivators and corepressors of MYCN have been experimentally demonstrated to mediate its transcriptional activity, and in these studies only a few target genes have been assessed [ 5 , 13 – 18 ]. However, the cooperation between MYCN and its cofactors has not been systematically investigated on a genome-wide level. MYCN is a bona fide oncogenic driver in NB [ 5 , 19 – 21 ] and it is known that the silencing of MYCN results in a decrease in cell proliferation and induction of cell differentiation in NB cells [ 22 , 23 ]. The therapeutic aim of directly targeting MYCN remains challenging due to its structural flexibility. However, one can target MYCN indirectly by identifying the enzymatically active cofactors that mediate MYCN regulated oncogenic transcriptional programs. This requires a genome-wide understanding of the critical cofactors by which MYCN regulates global gene expression.

The deregulation of MYC family oncogenes including c-MYC, MYCN, and MYCL occurs in most cancers and frequently marks those associated with poor prognosis [ 1 – 5 ]. MYCN is implicated in many pediatric embryonal tumors such as neuroblastoma (NB), rhabdomyosarcoma, medulloblastoma, and more recently in therapy-resistant adult cancers including subtypes of breast cancer and prostate cancers [ 2 , 5 ]. MYCN encodes a basic helix-loop-helix-leucine zipper transcription factor (TF) named N-Myc or MYCN and exhibits high-structural homology with c-MYC [ 2 ]. The c-MYC TF directly regulates gene transcription controlling cell growth, cell cycle progression, ribosome biogenesis, protein synthesis, genomic stability, glucose and nucleotide metabolism, apoptosis, etc. [ 1 , 3 ]. Additionally, c-MYC inhibits differentiation in normal hematopoietic, mesenchymal, adipocytic, neuronal, and muscle cells, as well as in cancer cells such as pheochromocytoma and erythroleukemia [ 6 – 8 ]. Some studies showed that c-MYC represses cell differentiation by blocking or repressing the expression of differentiation-inducing genes [ 6 ]. Early studies indicated that MYCN overexpression in NB cells leads to a transcriptome enriched in canonical MYC target genes including genes involved in ribosome biogenesis and protein synthesis [ 9 , 10 ]. Later, the identification of a functional MYCN signature gene set in one NB cell line indicated that MYCN suppresses genes associated with neuronal differentiation [ 11 ]. However, the molecular mechanisms by which MYCN orchestrates these global gene expression changes at a genome-wide level remain unclear.

Results

MYCN governs a malignant NB cell identity by activating canonical MYC target genes and suppressing neuronal differentiation genes We systematically investigated MYCN biological functions and transcriptional activity in several NB cell lines through both loss and gain of function studies. As previously reported [22,23], the knockdown of MYCN in IMR32 cell line using 2 different siRNAs (siMYCN_2 and siMYCN_4) reproducibly resulted in a decrease in cell proliferation and an increase in neurite extension (Figs 1A–1D and S1A). These consistent outcomes establish that the observed effects of MYCN loss of function are not due to off-target effects of the siRNAs. Thus, in subsequent studies, we utilized one of these 2 siRNAs (siMYCN_2) in KCNR, LAN5, and BE(2)C cell lines to show that the silencing of MYCN resulted in a decrease in cell proliferation and an increase in neurite extension (S1B–S1J Fig). For gain of MYCN function studies, the non-tumorigenic SHEP NB cell line that does not express MYCN (MYCN non-amplified) was used. Overexpression of MYCN in SHEP for 2 days resulted in a cell morphology change with a flatter, more round cell bodies compared to control cells (S1K and S1L Fig). Consistent with a previous report [24], an anchorage-independent cell proliferation assay showed that overexpression of MYCN in SHEP cells increased soft agar colony formation (S1M and S1N Fig). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. MYCN governs a malignant NB cell identity by directly activating canonical MYC target genes and suppressing neuronal differentiation genes. (A) MYCN knockdown using 2 different siRNAs in IMR32 cells for 72 h decreases MYCN protein levels as detected by western blot assay. (B) MYCN knockdown using 2 different siRNAs in IMR32 cells decreases cell number as detected by IncuCyte cell confluence assay. (C) and (D) MYCN knockdown in IMR32 cells increases neurite length as assessed by the IncuCyte neurite analysis assays and phase-contrast images. (E) k-Means clustering of MYCN and histone marks ChIP-seq around MYCN binding sites of NB cell line IMR32 (±3 kb) shows that MYCN binds to proximal regulatory elements containing active promoters marked by H3K27ac and H3K4me3 signals, and binds distal regulatory elements containing enhancers marked by H3K4me1 and H3K27ac signals. (F) GREAT GO analysis indicates that MYCN-bound promoter-associated genes are enriched in RNA processing and MYCN-bound enhancer-associated genes are enriched in nervous system development. (G) HOMER motif analysis shows the enrichment of canonical E-box in the promoters and the enrichment of non-canonical E-box in the enhancers. (H) The combination of RNA-seq and ChIP-seq data analysis in IMR32 cells after the silencing of MYCN shows that more MYCN-bound promoter-associated genes are down-regulated (521 vs. 434) while more MYCN-bound enhancer-associated genes are up-regulated (368 vs. 189) after the silencing of MYCN. (I) GREAT GO analysis indicates that MYCN-bound promoter-associated genes down-regulated after MYCN knockdown in IMR32 cells are enriched in ribosome biogenesis and RNA processing (left panel), and MYCN-bound enhancer-associated genes up-regulated after MYCN knockdown are enriched in neuronal differentiation (right panel). (J) Signal tracks show the MYCN, H3K27ac, and H3K4me3 ChIP-seq signals at the promoter of NAT10 gene (cyan box). (K) Signal tracks show the MYCN, H3K27ac, and H3K4me3 ChIP-seq signals at the promoter (cyan box) and enhancers (pink box) of GAP43 gene. (L) The silencing of MYCN in IMR32 cells results in a significant down-regulation of genes involved in ribosome formation based on the RNA-seq results, with MYCN binding to the promoters of these genes. The p-value is calculated using one-way ANOVA. (I) The silencing of MYCN in IMR32 cells results in a significant up-regulation of genes involved in neuron projection morphogenesis based on the RNA-seq results, with MYCN binding to the enhancers of these genes. The indicated p-value is calculated using a one-way ANOVA. The data underlying the graphs in the figure are shown in S1 Data. CPM, counts per million; GO, Gene Ontology; GREAT, genomic regions enrichment of annotation tool; NB, neuroblastoma. https://doi.org/10.1371/journal.pbio.3002240.g001 To identify MYCN-regulated genes, we conducted RNA-seq experiments after silencing MYCN expression. Additionally, we explored published RNA-seq datasets following MYCN silencing (GSE183641) [25] (S1 Table). Gene set enrichment analysis (GSEA) [26,27] showed that the silencing of MYCN using 2 different siRNAs (siMYCN_2 or siMYCN_4) in IMR32 resulted in a similar negative enrichment of MYC targets and ribosome biogenesis genes, whereas neuron markers and genes that positively regulate synaptic transmission were positively enriched (S1O and S1P Fig). Similar results were observed when MYCN was knocked down in other NB cell lines by using siMYCN_2 (S1Q and S1R Fig), or when MYCN was overexpressed in SHEP cells (S1S Fig). These loss and gain of MYCN function studies in multiple NB cell lines indicate that MYCN activates canonical MYC target genes and represses neuronal differentiation genes to govern a malignant NB cell identity.

Genome-wide mapping of MYCN binding To identify genes directly regulated by MYCN and the chromatin status associated with these genes, we performed ChIP-seq experiment using an MYCN antibody and antibodies that recognize different histone marks in IMR32 cells. High-confidence ChIP-seq peaks were called by MACS2 and peaks were normalized to reads per kilobase per million reads normalized read numbers (RPKM, see Materials and methods for details). Peaks from ChIP-seq of MYCN and histone marks were selected based on p-value (all p < 10−7). The MYCN ChIP-seq heatmap represented genome-wide stringent sets of MYCN peaks (MYCN binding sites, total 19,707 peaks) within the whole genome (Fig 1E). These MYCN peaks were segmented based on their colocalizations with the indicated histone marks (Fig 1E) through k-means clustering, revealing a total of 2 distinct clusters of MYCN peaks that colocalize with differing histone marks. In general, H3K4me1 marks both active and poised enhancers, H3K27ac marks both active enhancers and promoters, H3K4me3 marks active promoters, while H3K27me3 marks repressed chromatin. Based on the peak distribution and the signal intensity of the histone marks, cluster 1 of the heatmap represented active promoters or proximal regulatory regions, cluster 2 represented both active enhancers and weak or poised enhancers based on H3K4me1 and H3K27ac signal intensities (Fig 1E). The heatmap showed that MYCN overlapped with active histone marks but not the repressive histone mark H3K27me3 (Fig 1E). Among all the 19,707 MYCN peaks, 47.2% (9,299 peaks) are within promoters and 52.8% (10,408 peaks) are within enhancers. Genomic regions enrichment of annotation tool (GREAT) [28] was used to analyze the peak distribution of each of these clusters. Consistent with being located in promoter and enhancer regions, the results showed that the majority of peaks for cluster 1 were within 5 kb of the transcription start sites (TSSs), whereas the majority of peaks for cluster 2 were over 5 kb from the TSS (S1T Fig). GREAT Gene Ontology (GO) analysis of MYCN binding sites associated genes showed that MYCN-bound promoters (cluster 1) associated genes are enriched in RNA processing, ribosome assembly, metabolic process, protein synthesis, and other processes (Fig 1F, left panel, S2 Table), while MYCN-bound enhancers (cluster 2) associated genes are enriched in development such as top-ranked nervous system development (Fig 1F, right panel). Additionally, GREAT analysis of MYCN-bound active enhancers in IMR32, marked by H3K27ac peaks that do not overlap with H3K4me3 peaks, indicates that these enhancers’ associated genes are involved in regulating nervous system development (top-ranked). Notably, genes related to other developmental processes, such as limb bud formation and arterial endothelial cell differentiation, as well as pathways that are unrelated to differentiation, are also enriched (S2 Table). Furthermore, GREAT analysis indicates that genes associated with all active enhancers, or genes associated with active enhancers with or without MYCN binding are enriched in similar pathways (S2 Table). HOMER motif analysis showed that in IMR32 cells, E-boxes were enriched in MYCN-bound promoters and enhancers (Fig 1G). We further dissected MYCN binding sites and their associated genes using a publicly available MYCN ChIP-seq dataset generated in another MYCN-amplified cell line BE(2)C (GSE94822). Here, we simply separated MYCN binding sites into 2 groups, which include promoter regions (−1 kb–+100 bp from TSS) and distal regulatory regions (the regions outside of the promoter) as annotated by the HOMER tool. GREAT GO analysis of MYCN binding sites associated genes showed that MYCN-bound promoter-associated genes were enriched in canonical MYC target genes that regulate RNA processing and ribosome biogenesis, whereas MYCN-bound distal regulatory regions associated genes were enriched in nervous system development (S1U Fig). Altogether, the ChIP-seq analyses in both IMR32 and BE(2)C cells indicate that MYCN-bound promoters are associated with canonical MYC target genes and MYCN-bound distal regulatory regions are associated with neuronal genes in NB.

MYCN binds promoters to activate canonical MYC targets but binds to both enhancers and promoters to repress neuronal differentiation genes in NB To investigate how DNA bound MYCN affected gene transcription, we performed an integrative analysis of the MYCN ChIP-seq and RNA-seq data in IMR32 cells. A total of 1,823 genes, showing an expression change of >1.5-fold or <−1.5-fold with an adjusted p-value of <0.05 detected by RNA-seq upon silencing of MYCN, are considered MYCN-regulated genes (S1 Table). Genes associated with MYCN-bound promoters or enhancers (Fig 1E) are determined by using the HOMER peak annotation tool. We compared the list of MYCN-regulated genes identified by RNA-seq with the list of genes associated with MYCN-bound promoters or enhancers. This analysis revealed that among the 1,823 MYCN up- and down-regulated genes, 1,512 were bound by MYCN within the promoters, enhancers, or both. This analysis suggests that these genes are direct targets of MYCN. We found that for MYCN-bound promoter-associated genes, 521 genes (54.6%) were down-regulated and 434 genes (45.4%) were up-regulated after the silencing of MYCN. However, for MYCN-bound enhancer-associated genes, a greater number of genes were up-regulated (368 genes, 66.1%) compared to down-regulated genes (189 genes, 33.9%) (Fig 1H). MYCN-bound promoter-associated MYCN-activated genes (whose expression decreased after MYCN silencing) were significantly enriched in ribosome biogenesis and RNA processing (Fig 1I, left panel). In contrast, MYCN-bound promoter-associated, MYCN-repressed genes (whose expression increased after MYCN silencing) were enriched in pons development and response to axon injury (S1V Fig). The MYCN-bound enhancer-associated MYCN-repressed genes were significantly enriched in neuronal differentiation (Fig 1I, right panel), with MYCN-bound enhancer-associated MYCN-activated genes enriched in chordate embryonic development (S1W Fig). RPKM normalized signal tracks showed an MYCN ChIP-seq peak at the promoter of NAT10 (N-Acetyltransferase 10), a gene required for ribosome biogenesis (Fig 1J). For the neuronal differentiation gene, GAP43 (growth associated protein 43), a single MYCN ChIP-seq peak was observed at the promoter while multiple peaks were observed at its enhancers (Fig 1K). We found that in addition to binding to the enhancers of neuronal genes in NB cells, MYCN also binds to the promoters of these genes. The expression changes of representative MYCN-bound promoter-associated ribosome biogenesis genes after MYCN depletion were shown in Fig 1L, and those for representative MYCN-bound enhancer-associated neuronal genes were shown in Fig 1M. These results indicate that MYCN activates canonical MYC target genes mainly through binding to promoters in NB, whereas MYCN binds to both enhancers and promoters to suppress neuronal differentiation genes. Bona fide MYC-target genes encompass positive regulators of cell cycle progression, such as CDK4 and cyclin A, genes involved in ribosome biogenesis and protein synthesis, as well as those involved in metabolism. Therefore, a cell exhibiting high MYC levels would be primed for active proliferation due to increased cycling activity, larger cell mass, and enhanced competence in energy production [6,29]. In addition to genes related to ribosome biogenesis and protein synthesis, we specifically examined canonical MYC-target genes associated with cell proliferation and metabolism. Our findings indicate that MYCN binds to the promoters, but not enhancers, of these genes in most cases. For example, signal tracks revealed that MYCN colocalizes with H3K27ac and H3K4me3 on the promoters, but not enhancers, of cell cycle genes CDK4 and CCNA2, as well as metabolic genes ODC1, LDHA, GLS, PRIM1, and PKM (S1X Fig). Consistently, RNA-seq data analysis demonstrated that MYCN knockdown results in a negative enrichment of genes involved in cell cycling and metabolism (S1Y and S1Z Fig).

MYCN depletion alters histone modifications on its target genes We next asked whether the activation of canonical MYC target genes and repression of neuronal differentiation genes by MYCN are associated with changes of histone modifications after MYCN depletion in IMR32 cells. To compare the ChIP-seq signal intensity in control and MYCN silenced samples, the ChIP-seq peaks were RPKM normalized. While changes in MYCN levels led to decreases in the average MYCN ChIP-seq signals, there were no changes in the average ChIP-seq signals for histone marks globally (S3A Fig). Analysis of MYCN binding focused on whole genome-wide TSS showed similar results (S3B Fig). Furthermore, we focused on MYCN-bound genes whose expression was modulated after changes in MYCN expression. After silencing MYCN, the integrative analysis of the MYCN ChIP-seq and RNA-seq merged data identified 601 genes whose expression decreased and thus was directly activated by MYCN (bound by 936 MYCN peaks), and 625 genes whose expression increased and thus was directly suppressed by MYCN (bound by 1,420 MYCN peaks) whose expression increased with MYCN silencing. Ingenuity Pathway Analysis (IPA) showed that MYCN-bound genes whose expression decreased after MYCN silencing were enriched in RNA posttranscriptional modification and protein synthesis (Fig 3A), whereas the MYCN-bound genes that were up-regulated after MYCN silencing were positively enriched in neuronal differentiation (Fig 3B). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. MYCN depletion alters histone modifications on its target genes. (A) The integrative analysis of the MYCN ChIP-seq and RNA-seq merged data by IPA shows that the MYCN-bound, down-regulated genes after MYCN silencing in IMR32 cells are enriched in RNA posttranscriptional modification and protein synthesis. (B) IPA of the RNA-seq data shows that the MYCN-bound up-regulated genes after MYCN silencing are positively enriched in neuronal differentiation. (C) When focusing on MYCN-bound promoters associated, down-regulated genes, metagene plots show that MYCN silencing results in a decrease in the average ChIP-seq signals of H3K4me3 and H3K27ac at the MYCN peak center. (D) When focusing on MYCN-bound promoters associated, up-regulated genes, metagene plots show that MYCN silencing results in an increase in the average ChIP-seq signal of H3K27ac at the MYCN peak center. (E) When focusing on MYCN-bound enhancers associated, up-regulated genes, metagene plots show that MYCN silencing results in an increase in the average ChIP-seq signal of H3K27ac at the MYCN peak center. (F) When focusing on MYCN-bound enhancers associated, down-regulated genes, metagene plots show that MYCN silencing results in a decrease in the average ChIP-seq signal of H3K27ac at the MYCN peak center. (G) Signal tracks show decreases in MYCN, H3K27ac, and H3K4me3 ChIP-seq signals at the promoter of RPL8 after the depletion of MYCN (cyan box). (H) Signal tracks show increases in H3K27ac ChIP-seq signals at the promoter (cyan box) and enhancers (pink boxes) of GAP43 after the depletion of MYCN. The data underlying the graphs in the figure are shown in S1 Data. IPA, Ingenuity Pathway Analysis. https://doi.org/10.1371/journal.pbio.3002240.g003 Since we discovered that MYCN binds to the promoters of MYCN-activated canonical MYC target genes and binds to the enhancers of MYCN-repressed neuronal genes (Fig 1I), we next focused on how the silencing of MYCN affects the epigenetic modifications at MYCN binding sites within these promoters and enhancers. By focusing on the promoters of MYCN-bound down-regulated genes due to MYCN silencing, we found a decrease in the average ChIP-seq signals of active promoter marks H3K27ac at the MYCN peak center, while the decrease in the average H3K4me3 signals was not significant (Fig 3C). When focused on the promoters of MYCN-bound, up-regulated genes after MYCN silencing, we found an increase of the average ChIP-seq signal of H3K27ac at the MYCN peak center (Fig 3D). By focusing on the enhancers of MYCN-bound genes after MYCN silencing, there was an increase in the average H3K27ac ChIP-seq signal at the MYCN peak center for up-regulated genes (Fig 3E) but a decrease in the H3K27ac signal for down-regulated genes (Fig 3F). For example, signal tracks showed decreases of MYCN, H3K27ac, and H3K4me3 ChIP-seq signals at the promoter of RPL8 (ribosomal protein L8) after the depletion of MYCN (Fig 3G), whereas signal tracks for the neuronal differentiation gene, GAP43, showed decreases of MYCN ChIP-seq signals and increases of H3K27ac ChIP-seq signals at both the promoter and enhancers after the depletion of MYCN (Fig 3H). These results suggest that MYCN directly activates canonical MYC target genes through increasing promoter activity, whereas MYCN directly represses neuronal differentiation genes through repressing enhancer activity. Subsequently, we investigated the distribution of MYCN peaks after MYCN knockdown. ChIP-seq results revealed that silencing MYCN reduced the number of MYCN peaks from 19,707 to 6,012. Analyzing the distribution of MYCN peaks among the remaining 6,012 peaks following MYCN silencing in IMR32 cells through k-means clustering (S3C Fig) showed a significant decrease in the percentage of MYCN peaks within enhancers and a notable increase in the percentage of MYCN peaks within promoters, compared to the MYCN peak distribution in control cells (S3D Fig). This finding aligns with the observation that elevated levels of MYC oncoproteins tend to bind to distal regulatory regions [31–33].

Genome-wide colocalization of MYCN and its cofactors To investigate the genome-wide interactions of MYCN and the coactivator WDR5 or corepressor G9a, we performed ChIP-seq analysis of MYCN, WDR5, and G9a in IMR32 cells. MYCN peaks were segmented based on their colocalizations with specific histone marks, WDR5, and G9a through k-means clustering. We found that WDR5 bound to both promoters and enhancers, although with a stronger signal intensity at the promoter regions (Fig 4B). In contrast, G9a predominantly bound to the enhancers (Fig 4B). ChIPPeakAnno analysis showed that 73% WDR5 and 52% G9a binding sites overlapped with MYCN binding sites (Fig 4C). Consistent with the heatmap shown in Fig 4B, ChIPPeakAnno analysis indicated that among the 11,844 MYCN and WDR5 overlapped peaks, 6,723 (56.8%) are within promoters, 5,250 (44.3%) are within enhancers, and 361 (3.0%) are within super-enhancers. For the 2,352 peaks where MYCN and G9a overlap, 105 (4.5%) are within promoters, 2,250 (95.6%) are within enhancers, and 158 (6.7%) are within super-enhancers (Fig 4C). Notably, a small number of peaks were assigned to both promoters and enhancers, or both enhancers and super-enhancers. HOMER motif scan showed that the top 2 MYCN binding motifs are canonical and non-canonical E-boxes, while the non-canonical E-box was found to be enriched in WDR5 and G9a binding sites (Fig 4D–4F). Consistent with the k-means clustering analysis, GREAT peak distribution analysis showed that few G9a binding sites were within 5 kb of TSS (<5%), while 55% of WDR5 binding sites were within 5 kb of TSS (Fig 4G). The genome-wide colocalization of MYCN with either WDR5 or G9a suggested that each of these cofactors cooperates with MYCN to regulate a subset of MYCN target genes. GREAT GO analyses showed that MYCN WDR5 ChIP-seq peak-associated genes were enriched in RNA processing and ribosome biogenesis (Fig 4H, left panel), suggesting its potential role as an MYCN coactivator. On the other hand, G9a binding site-associated genes were enriched in nervous system development (Fig 4H, right panel), suggesting its potential role as an MYCN corepressor. These results suggest that MYCN cooperates with WDR5 to regulate canonical MYC target genes while MYCN cooperates with G9a to regulate neuronal differentiation genes.

MYCN silencing alters the genomic DNA binding of its cofactors To identify whether MYCN recruits its cofactors to its binding sites, we performed ChIP-seq experiments using MYCN, WDR5, and G9a antibodies in siCtrl and siMYCN transfected IMR32 cells. MYCN silencing did not alter the steady-state protein levels of G9a but did cause a 50% decrease in WDR5 based on densitometric analysis (Fig 5A). ChIP-seq results showed that the knockdown of MYCN in IMR32 cells caused a 15% decrease in WDR5 ChIP-seq peak numbers (18,263 to 15,612 peaks) (S5 Table), which was possibly caused by the decrease of WDR5 protein. By focusing on MYCN and WDR5 overlapped peaks, metagene plots showed a 50% decrease in the average MYCN ChIP-seq signals and a 5% decrease in the average WDR5 ChIP-seq signals at the summit of MYCN peak centers (Fig 5B) after silencing of MYCN. For example, signal tracks for the ribosome gene RPL8 promoter and the cell adhesion molecule coding gene PVR promoter showed decreases in WDR5 binding after MYCN silencing (S5A Fig). However, it is worth noting that silencing MYCN resulted in a 50% decrease of WDR5 at the protein level but was accompanied by only a slight decrease in WDR5 ChIP-seq peak numbers and average ChIP-seq signal intensity, suggesting that the interaction between MYCN and DNA is not required for WDR5 to bind to DNA. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Silencing of MYCN selectively alters genomic DNA binding of its cofactors. (A) Western blot and densitometric analyses show the effect of 72 h MYCN knockdown in IMR32 cells on the expression of its cofactors at the protein levels. (B) Metagene plots show that the knockdown of MYCN results in a decrease of average MYCN and WDR5 ChIP-seq signal at the MYCN peak center of the MYCN and WDR5 overlapped binding sites. (C) Metagene plots show that the knockdown of MYCN results in a decrease in average MYCN and G9a signal at the MYCN peak center of MYCN and G9a overlapped binding sites. (D) ChIP-seq heatmaps (left) and profiles (right) of MYCN and G9a in control and MYCN knockdown IMR32 cells around MYCN binding sites (±3 kb) of MYCN and G9a overlapped peaks. (E) IPA of the G9a ChIP-seq data shows that the G9a binding sites with decreased ChIP-seq signal (>1.2-fold decrease of G9a signal after the silencing of MYCN) associated genes are enriched in nervous system development. The data underlying the graphs in the figure are shown in S1 Data. IPA, Ingenuity Pathway Analysis. https://doi.org/10.1371/journal.pbio.3002240.g005 To investigate the WDR5-independent MYCN target genes, we compared the list of genes regulated by MYCN (silencing MYCN for 48 h) and the list of genes regulated by WDR5 (silencing WDR5 for 48 h) by analyzing the RNA-seq data (threshold, >1.5-fold and <−1.5-fold, q < 0.05) in IMR32 cells. We found that among the 1,908 MYCN target genes, 83% (1,598/1,908 genes) were regulated by MYCN but not by WDR5. Among the WDR5 target genes, 66% (610/920 genes) were regulated by WDR5 but not by MYCN (S6 Table). IPA showed that genes commonly regulated by MYCN and WDR5, or uniquely regulated by WDR5, are involved in regulating protein translation, while genes only regulated by MYCN are enriched in regulating nervous system development and other pathways (S7 Table). These results indicate that MYCN regulates broad transcriptional programs, which overlap with WDR5 to regulate transcriptional programs associated with protein translation. When focused on G9a, we found that MYCN silencing resulted in a 64% decrease in G9a ChIP-seq peak numbers (6,723 to 2,428 peaks) (S8 Table). Metagene plots of MYCN and G9a overlapping peaks show a 5% decrease in the average G9a ChIP-seq signals at the summit of MYCN peak centers after silencing MYCN (Fig 5C). For example, signal tracks for the KCNK3 gene showed that MYCN knockdown decreased MYCN and G9a binding signals within the KCNK3 intron (S5B Fig). MYCN silencing resulted in a subtle decrease in the average G9a ChIP-seq signals (Fig 5C) but dramatically reduced G9a ChIP-seq peak numbers (S8 Table). Thus, we further analyzed the MYCN and G9a overlapping peaks by dissecting them into 3 clusters. The first cluster was classified as down-regulated (down) peaks, which included G9a peaks with at least >1.2-fold decrease of G9a ChIP-seq signal after MYCN silencing (Fig 5D); the second cluster was classified as not-altered (stable) peaks, which included G9a peaks with <1.1-fold changes of G9a ChIP-seq signal after the silencing of MYCN (Fig 5D); whereas the third cluster was classified as increased (up) peaks, which included G9a peaks with >1.2-fold increase of G9a ChIP-seq signal after the silencing of MYCN (Fig 5D). More G9a binding peaks showed decreases in their ChIP-seq signal than the ones that were stable or had increased signals in MYCN silenced cells (Fig 5D). This is consistent with the observation of a reduced number of G9a ChIP-seq peaks after the silencing of MYCN (S8 Table). IPA of the G9a peaks associated genes with a decreased or stable ChIP-seq signal revealed that these genes were enriched in nervous system development (Figs 5E and S5C). Furthermore, the genes associated with G9a binding sites whose ChIP-seq signals increased were enriched in the development of other tissues such as hair and skin development (S5D Fig). Our results indicated that MYCN selectively recruits G9a to MYCN binding sites that are associated with neuronal genes.

The depletion of MYCN cofactors antagonizes MYCN-mediated gene expression changes Our study indicated that WDR5 and G9a interact with MYCN and that these proteins colocalize to the promoters or enhancers of genes they regulate (Fig 4B). To investigate whether the presence of WDR5 or G9a is necessary to regulate MYCN target genes, we silenced WDR5, G9a, or MYCN using siRNAs for 48 h in IMR32 cells and performed RNA-seq analysis. Western blot results showed that the silencing of WDR5 or G9a had no effect on MYCN expression at this time point (S7A Fig). After MYCN silencing for 48 h in IMR32 cells, GSEA results showed significant negative enrichment of canonical MYC target genes involved in ribosome biogenesis, RNA processing, ribosome formation, and cytoplasmic translation (S7B Fig), with significant positive enrichment of neuronal genes involved in axon development, neuron differentiation, glutamatergic synapse, and neuron projection guidance (S7C Fig). To compare genes regulated by MYCN, WDR5, and G9a in IMR32 cells after 48 h MYCN silencing, we generated an “MYCN-activated canonical MYC targets” gene sets by defining genes that were significantly down-regulated after the silencing of MYCN in the gene sets of ribosome biogenesis, RNA processing, ribosome formation, and cytoplasmic translation (S11 Table). In parallel, we generated an “MYCN-repressed neuronal genes” gene sets by defining those genes that were significantly up-regulated after the silencing of MYCN in the gene sets of axon development, neuron differentiation, glutamatergic synapse, positive regulation of synaptic transmission, and neuron projection guidance (S11 Table). The silencing of WDR5 resulted in a significant negative enrichment of “MYCN-activated canonical MYC targets” gene set such as genes involved in ribosome formation and cytoplasmic translation (Fig 7A). GSEA results did not exhibit significant positive enrichment of neuronal genes after the silencing of WDR5. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. The depletion of MYCN cofactors antagonizes MYCN-mediated gene transcription regulation, which makes them potential therapeutic targets. (A) GSEA shows that the silencing of WDR5 results in a significant negative enrichment of genes involved in ribosome formation and protein synthesis that are activated by MYCN. (B) GSEA shows that the silencing of G9a results in a significant positive enrichment of genes involved in axon development and neuron differentiation that are repressed by MYCN. (C) The silencing of either MYCN or WDR5 but not G9a results in a significant down-regulation of genes involved in ribosome formation and protein translation based on the RNA-seq results. The p-value indicated is calculated in one-way ANOVA. (D) The silencing of either MYCN or G9a but not WDR5 results in a significant up-regulation of genes involved in axon development and neuron differentiation based on the RNA-seq results. The p-value indicated is calculated in one-way ANOVA. (E) CellTiter-Glo assay shows the drug effect of the OICR-9429 (OICR, 40 μm) + UNC0642 (UNC, 8 μm) treatment on IMR32 cell viability at 72 h. (F) Realtime PCR shows that the inhibition of both WDR5 and G9a results in a significant down-regulation of ribosomal genes and up-regulation of neuronal genes. (G) Heatmaps show the percentage of cell viability after different doses of WDR5 inhibitor OICR-9429 and G9a inhibitor UNC0642 treatment in MYCN-amplified NB cell lines. Cells are treated with the drugs for 72 h and cell viability is measured by CellTiter-Glo Cell Viability Assay. (H) SynergyFinder online tool is used for bliss synergistic analysis to evaluate the synergistic effect of the combination treatment in MYCN-amplified NB cell lines shown in (G). The data underlying the graphs in the figure are shown in S1 Data. CPM, counts per million; GSEA, gene set enrichment analysis; NB, neuroblastoma. https://doi.org/10.1371/journal.pbio.3002240.g007 G9a predominantly binds to the enhancers (Fig 4B) and G9a-bound peak-associated genes were enriched in nervous system development (Fig 4H). GSEA results showed that the silencing of G9a resulted in a significant positive enrichment of the “MYCN-repressed neuronal genes” gene set such as genes involved in axon development, neuron differentiation, glutamatergic synapse, and neuron projection guidance (Figs 7B and S7D), while no significant negative enrichment of canonical MYC target genes was observed after the silencing of G9a. Representative genes commonly down-regulated after the silencing of MYCN or WDR5 that belong to the gene sets of “MYCN-activated canonical MYC targets” were shown in Fig 7C. The same group of genes was not down-regulated after G9a silencing (Fig 7C). Representative genes commonly up-regulated after the silencing of MYCN or G9a that belong to “MYCN-repressed neuronal genes” gene set were shown in Fig 7D. The same group of genes was not up-regulated after the silencing of WDR5 (Fig 7D). These results indicated that WDR5 silencing antagonizes MYCN-mediated activation of canonical MYC target genes, while G9a silencing antagonizes MYCN-mediated repression of neuronal genes.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002240

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.