3D genome mapping identifies subgroup-specific chromosome conformations and tumor-dependency genes in ependymoma

Abstract

Ependymoma is a tumor of the brain or spinal cord. The two most common and aggressive molecular groups of ependymoma are the supratentorial ZFTA-fusion associated and the posterior fossa ependymoma group A. In both groups, tumors occur mainly in young children and frequently recur after treatment. Although molecular mechanisms underlying these diseases have recently been uncovered, they remain difficult to target and innovative therapeutic approaches are urgently needed. Here, we use genome-wide chromosome conformation capture (Hi-C), complemented with CTCF and H3K27ac ChIP-seq, as well as gene expression and DNA methylation analysis in primary and relapsed ependymoma tumors, to identify chromosomal conformations and regulatory mechanisms associated with aberrant gene expression. In particular, we observe the formation of new topologically associating domains ('neo-TADs') caused by structural variants, group-specific 3D chromatin loops, and the replacement of CTCF insulators by DNA hyper-methylation. Through inhibition experiments, we validate that genes implicated by these 3D genome conformations are essential for the survival of patient-derived ependymoma models in a group-specific manner. Thus, this study extends our ability to reveal tumor-dependency genes by 3D genome conformations even in tumors that lack targetable genetic alterations.

Fig. 1. 3D tumor genome profiling identifies PFA and ZFTA ependymoma specific chromatin conformations and enhancer associated genes.

a Overview of the major results obtained by the application of genome-wide chromosome conformation capture (Hi-C) in ependymoma brain tumors. b Characteristics of ependymoma samples analyzed by Hi-C. One group of PFA ependymoma samples has no apparent copy-number variants, while the other group of PFA samples exhibits chromosome 1q gains associated with an unfavorable outcome. In addition, the copy number status of selected chromosome arms that are most commonly affected in ependymoma is reported. c Unsupervised hierarchical clustering of PFA and RELA ependymoma tumors based on DNA interactions (Hi-C) stratifies the samples into the expected molecular groups. d Integrative analysis of enhancers (H3K27ac ChIP-seq), chromosome conformation (Hi-C) and gene expression (RNA-seq) shows that genes are more strongly expressed when their promoters physically interact with other promoters or with enhancers. Shown are tumors (n = 6 independent samples) for which sample-matched H3K27ac ChIP-seq, RNA-seq, and Hi-C data are available. P-values from the bootstrap t-test are included. The box plot center line, box limits and whiskers indicate the median, upper/lower quartiles and 1.5× interquartile range respectively. e) The Hi-C data reliably detect the structural variants that lead to the ZFTA-RELA fusion gene in supratentorial ZFTA-fusion associated tumors (top row), while no such signals were found in PFA tumors (bottom row). Green boxes highlight SVs predicted by the computational methods applied.

Fig. 2. Transcriptional activation of RCOR2 by neo-TADs in RELA ependymoma.

a Chromatin contacts in a reconstructed ZFTA tumor genome (sample 4EP53) including the tandem duplication that leads to the ZFTA-RELA fusion (chr1:63532174-65429788, green boxes). Solid black boxes show TADs identified by applying TopDom to the Hi-C data mapped on to the reconstructed tumor genome, including a neo-TAD that spans the DNA breakpoint. b, c Reconstructed genomic locus containing the ZFTA-RELA fusion gene in the ZFTA ependymoma sample 4EP53 (b) and 11EP22 (c). The black boxes/ triangles indicate TADs reported by TopDom when applied to the reconstructed tumor genome. A neo-TAD is identified that spans the DNA breakpoint and places RCOR2 into a new regulatory environment. d Boxplot of RCOR2 gene expression across ependymoma groups using Affymetrix gene expression data (n = 393). RCOR2 is significantly upregulated in ZFTA tumors (ZFTA vs all other tumor classes limma p-val.: 7.62e−27). The center line, box limits, whiskers, and points indicate the median, upper/lower quartiles, 1.5× interquartile range and outliers, respectively. e Correlation between RCOR2 and ZFTA in ZFTA (left side, n = 76, cor = 0.663, p-val = 6.93e−11) and PFA ependymoma samples (right side, n = 200, cor=0.336 p-val = 1.13e−06). f–i shRNA time-course knock-down experiments in ZFTA (EP1NS) and PFA (EPD210FH) ependymoma cell lines using a scrambled control and two shRNA constructs each targeting either RCOR2 in EP1NS (f), RCOR2 in EPD210FH (g), LSD1 in EP1NS (h) and LSD1 in EPD210FH (i). All constructs are GFP tagged and GFP positive cells are sorted by FACS. For panel (f), error bars represent mean ± SD for n = 3 independent experiments (two-tailed paired t test p-val = 0.0018 and 0,0046; shRCOR#1 and shRCOR2#2, respectively). For panels (g–i), normalized data represent mean from n = 2 independent experiments per cell line. j, k Dose–response curves of single-compound treatment with ORY-1001 (j) or Entinostat (k) of ZFTA (EP1NS, BT165 and ST-1) and PFA (EPD210FH, BT214) ependymoma spheroids over a 72-h time-course using Celltiter-Glo cell viability assays. For each sample the results are presented as percentage of the Luminescence signal from control condition (i.e. water for ORY-1001 and DMSO for Entinostat). Error bars represent mean ± SD for n = 3 independent experiments (one-way ANOVA test p-val < 0,0001).

Fig. 3. Long-range DNA loops reveal a complex chromatin complex in PFA ependymomas.

a Hi-C DNA interaction matrices wherein a ~5 million base pair segment of chromosome 2 is aligned along the diagonals shown for PFA (9EP1, left) and ZFTA (4EP53, middle) tumors and normal cerebellum astrocytes (CAs, right). b Hi-C DNA interactions of a PFA tumor (sample BT214) wherein the same ~5 million base pair segment of chromosome 2 shown in panel (a) is aligned horizontally. Circles and dashed lines highlight long-range DNA interactions. c Genome browser view of the PFA-specific chromatin cluster shown in panels (a) and (b). The included data tracks show DNA interactions in PFA and ZFTA tumors via loops. Tracks for RNA-seq, H3K27ac and Hi-C derived DNA loop were obtained from merging PFA (9EP1,9EP9, 7EP18) or ZFTA (11EP22, 4EP53, 7EP41) samples, respectively. Differential specificity of the PFA loop is confirmed from statistical comparison (adjusted p-val 0.0089) via DiffLoop tool. d–f Genetic (CRISPR-Cas9) time-course inhibition of MAP3K20 in one PFA EPD210FH (d) and ZFTA EP1NS (e) and VBT372 (f) cell lines. Changes in the percentage of GFP positive cells are presented after normalization. GFP percentage was normalized to day 4 post infection and presented as day 0. Normalized data represent mean from n = 2 independent experiments in cell lines EPD210FH, EP1NS and mean ± SD for n = 3 independent experiments for VBT373. g EPD210FH and EP1NS cells are treated with the MAP3K20 inhibitor (M443) for 6 days and cell viability is measured by CellTiterGlo assay and IC50 value is calculated by GraphPAD as respectively, 16, 7 and 37, 5 uM. Error bars represent mean ± SD, n = 3 biological replicates are used for all experiments. h, i Genetic (CRISPR-Cas9) time-course inhibition of ITGA6 in PFA EPD210FH (h) and RELA EP1NS (i) cells using a control sgRNA and two individual sgRNA constructs. All constructs are GFP tagged and GFP positive cells are sorted by FACS. Normalized data represent mean from n = 2 independent experiments in each cell line.

Fig. 4. Hypermethylation replaces CTCF binding sites in PFA ependymoma.

a Proposed mechanism of epigenetic oncogene activation in PFA ependymoma tumors. Top: The CTCF insulator is replaced by DNA methylation allowing enhancer to activate oncogene. Below: The oncogene is separated from an enhancer by topological barrier. Created with BioRender.com. b The volcano plot of differential CTCF binding sites between PFA and ZFTA ependymoma tumors (min p-value: 0.1). CTCF binding sites significantly hypermethylated in PFA are marked in orange (min q-value: 0.05). c Comparison of CTCF binding strength (CTCF ChIP-seq, x-axis, min p-value 0.1, min fold change: 0.5) and DNA methylation (WGBS, y-axis, min q-value: 0.05, min difference: 0.1) at differential CTCF binding sites between PFA and ZFTA ependymoma tumors. d Heatmap of WGBS-derived DNA methylation at the 300 most significant differential CTCF binding sites in three PFA (left) vs. three ZFTA (right) ependymoma tumors. e Genome browser visualization of PFA ependymoma-specific DNA loops that associate two PFA enhancers (E1 and E2) with the ARL4C gene. Tracks for RNA-seq, H3K27ac, CTCF and Hi-C derived DNA loops are obtained by merging PFA (9EP1,9EP9, 7EP18) or ZFTA (11EP22, 4EP53, 7EP41), respectively. f WGBS-derived DNA methylation and CTCF ChIP-seq data from PFA and ZFTA tumors at PFA-specific hypermethylated CTCF loci. g ARL4C expression is positively correlated with activity of enhancer E1 (chr2:237763494 − 237764993) in ependymoma tumors (n = 24). h Genetic (CRISPR-Cas9) time-course inhibition of ARL4C in PFA cells (EPD210FH) using a control sgRNA and two individual sgRNA constructs. All constructs are GFP tagged and GFP positive cells are sorted by FACS. Normalized data represent mean from n = 2 independent experiments. i Expanded view of the CTCF motif targeted by CRISPR-Cas9: two sgRNAs and protospacer adjacent motif (PAM) direct Cas9 nuclease to the motif. Sequencing of target site demonstrates the formation of indels (insertion or deletions). j qPCR reveals increased ARL4C expression up on targeting CTCF by CRISPR-Cas9 in ZFTA cells. Results are normalized to control gRNA and data represent mean from n = 2 independent experiments. k Images depict ARL4C expression in ZFTA cells (EP1NS) after targeting the CTCF binding site by either gControl/Cas9 or gCTCF#1/Cas9 at 10 days post-infection.