PRDM6 promotes medulloblastoma by repressing chromatin accessibility and altering gene expression

Abstract

SNCAIP duplication may promote Group 4 medulloblastoma via induction of PRDM6, a poorly characterized member of the PRDF1 and RIZ1 homology domain-containing (PRDM) family of transcription factors. Here, we investigated the function of PRDM6 in human hindbrain neuroepithelial stem cells and tested PRDM6 as a driver of Group 4 medulloblastoma. We report that human PRDM6 localizes predominantly to the nucleus, where it causes widespread repression of chromatin accessibility and complex alterations of gene expression patterns. Genome-wide mapping of PRDM6 binding reveals that PRDM6 binds to chromatin regions marked by histone H3 lysine 27 trimethylation that are located within, or proximal to, genes. Moreover, we show that PRDM6 expression in neuroepithelial stem cells promotes medulloblastoma. Surprisingly, medulloblastomas derived from PRDM6-expressing neuroepithelial stem cells match human Group 3, but not Group 4, medulloblastoma. We conclude that PRDM6 expression has oncogenic potential but is insufficient to drive Group 4 medulloblastoma from neuroepithelial stem cells. We propose that both PRDM6 and additional factors, such as specific cell-of-origin features, are required for Group 4 medulloblastoma. Given the lack of PRDM6 expression in normal tissues and its oncogenic potential shown here, we suggest that PRDM6 inhibition may have therapeutic value in PRDM6-expressing medulloblastomas.

Figure 1

Expression and subcellular localization of PRDM6 in human neuroepithelial stem cells. (A) Schematic showing the generation of PRDM6-expressing NES cells from human iPS cells. (B) Wild-type (WT) NES cells do not express PRDM6. qRT-PCR analysis of PRDM6 expression in wild-type, empty vector (EV), and PRDM6-expressing (PRDM6) NES cells. Expression levels are fold changes relative to GAPDH. Error bars denote SEM. P-values were determined by an unpaired, two-tailed t-test. (C) PRDM6 protein levels in EV and PRDM6 NES cells. Equal amounts of whole cell extract from empty vector (EV) or PRDM6-V5-transduced (PRDM6) NES cells were probed with anti-V5 antibodies to assess PRDM6 expression and antibodies to β-Actin as a loading control. (D) Subcellular fractionation of PRMD6 NES cells. Equal amounts of whole cell extract (total), cytosol, and nuclear extract from NES cells expressing V5-epitope-tagged, human PRDM6 were probed with antibodies against V5 and the indicated proteins of known subcellular localization. (E) Representative confocal microscopy images of anti-V5-immunostained EV and PRDM6 NES cells. Nuclei were counterstained with DAPI. Scale bars, 10 µm.

Figure 2

PRDM6 represses chromatin accessibility in neuroepithelial stem cells. (A) Venn diagram showing the numbers and overlap of regions with accessible chromatin in EV (gray) and PRDM6 (green) NES cells. (B) Volcano plot of results from a quantitative analysis of differential chromatin accessibility in PRDM6 vs. EV NES cells (FDR < 0.01). Regions with a Log2 fold change ≥1 in chromatin accessibility are highlighted in yellow. Blue dots correspond to regions with significant (FDR < 0.01) but < 1 Log2 fold change. (C) Heatmaps showing ATAC-seq signal within 1.5-kb of all sites with differential chromatin accessibility (n = 35,441 sites, FDR < 0.01) across three replicate experiments (Rep1-3) in PRDM6 and EV NES cells. (D) left, Distribution of regions with differential accessibility (FDR < 0.01) across the indicated genomic annotations. Right, distribution of differential accessibility regions in PRDM6 NES relative to the expected genomic distribution.

Figure 3

PRDM6 expression causes widespread gene expression changes in neuroepithelial stem cells. (A) Volcano plot of gene expression changes in PRDM6 vs. EV NES cells (n = 3 biological replicates). Dots correspond to individual genes. Genes with significant (adjusted P-value < 0.01) up- or downregulation (Log2 fold change ≥1) are highlighted in yellow; genes with significant Log2 fold changes of  < 1 are in blue. Gray dots indicate genes without significant changes (adjusted P-value > 0.01). (B) Gene ontology (GO) analysis of upregulated (top) and downregulated (bottom) genes in PRDM6 NES cells. The bars correspond to significantly enriched GO terms and are colored by P-values in log base 10.

Figure 4

Impact of PRDM6 on transcription factor binding in open chromatin regions. (A) Volcano plot showing the quantification of differential transcription factor binding based on digital footprinting analysis of JASPAR transcription factor motifs (n = 841) in open chromatin regions of PRDM6 and EV NES cells. Transcription factor motifs with significantly differential binding scores (P < 0.0001) are indicated in blue; transcription factor motifs with high (≥0.075) changes in differential binding score (P < 0.0001) are highlighted in orange. Gray dots represent transcription factor motifs without significantly altered binding scores (P ≥ 0.0001). Footprinting was performed in n = 3 biological replicates per group. (B) Aggregate plots of footprinting signal across all SOX4, RFX1, and SOX3 motifs in EV (purple) and PRDM6 (blue) NES cells.

Figure 5

Elucidation of genome-wide PRDM6 binding in neuroepithelial stem cells. (A) Genomic distribution of PRDM6 binding sites determined by CUT&RUN. (B) Venn diagram showing the overlap of PRDM6-bound sites with regions showing altered chromatin accessibility (FDR < 0.01). (C) Gene ontology (GO) analysis of genes overlapping with or located within two kb of a PRDM6 binding site. Bars are colored by P-values in log base 10 and indicate significantly enriched GO terms. (D) Venn diagram showing the overlap of PRDM6-bound sites with H3K27me3-marked regions identified by CUT&RUN in PRDM6 NES cells. (E) Examples showing PRDM6 enrichment in relation to H3K27me3-marked regions in EV and PRDM6 NES cells, respectively. Bins per million (BPM)-normalized CUT&RUN and ATAC-seq signals are shown. SEACR peak calling results for PRDM6 CUT&RUN signal are indicated by gray bars. Genomic region coordinates and refGene annotations are shown.

Figure 6

PRDM6 expression in neuroepithelial stem cells induces medulloblastomas in mice. (A) Schematic of orthotopic human NES transplantation into the mouse cerebellum. (B) Outcomes of PRDM6 or EV NES cell transplantations in mice. d, days post-injection. (C) H&E staining of representative PRDM6 NES-derived tumor (left). Immunohistochemistry staining of PRDM6 via the V5 epitope tag (middle) and Ki67 (right) in PRDM6 NES-derived tumor. Dashed lines denote tumor margins. Main scale bars correspond to 100 µm; scale bars in insets correspond to 25 µm; black arrowheads in the middle panel inset denote anti-V5-reactive (PRDM6-expressing) nuclei; white arrowheads indicate PRDM6-negative nuclei; black arrowheads in the right panel inset denote Ki67-positive, proliferating cells. (D) Latency of tumor development after implantation of PRDM6-, MYCN- or MYCN-PRDM6 NES cells into the cerebellum of mice. Error bars represent SEM. P-values were determined by one-way ANOVA with Tukey’s post hoc correction. (E) Clustering of tumors derived from PRDM6-, MYCN-, or MYCN-PRDM6 NES cells with human medulloblastomas and pediatric, non-medulloblastoma (non-MB) CNS tumors.