Cross-Cohort Analysis Identifies a TEAD4-MYCN Positive Feedback Loop as the Core Regulatory Element of High-Risk Neuroblastoma

Abstract

High-risk neuroblastomas show a paucity of recurrent somatic mutations at diagnosis. As a result, the molecular basis for this aggressive phenotype remains elusive. Recent progress in regulatory network analysis helped us elucidate disease-driving mechanisms downstream of genomic alterations, including recurrent chromosomal alterations. Our analysis identified three molecular subtypes of high-risk neuroblastomas, consistent with chromosomal alterations, and identified subtype-specific master regulator proteins that were conserved across independent cohorts. A 10-protein transcriptional module-centered around a TEAD4-MYCN positive feedback loop-emerged as the regulatory driver of the high-risk subtype associated with MYCN amplification. Silencing of either gene collapsed MYCN-amplified (MYCN^Amp) neuroblastoma transcriptional hallmarks and abrogated viability in vitro and in vivo Consistently, TEAD4 emerged as a robust prognostic marker of poor survival, with activity independent of the canonical Hippo pathway transcriptional coactivators YAP and TAZ. These results suggest novel therapeutic strategies for the large subset of MYCN-deregulated neuroblastomas.Significance: Despite progress in understanding of neuroblastoma genetics, little progress has been made toward personalized treatment. Here, we present a framework to determine the downstream effectors of the genetic alterations sustaining neuroblastoma subtypes, which can be easily extended to other tumor types. We show the critical effect of disrupting a 10-protein module centered around a YAP/TAZ-independent TEAD4-MYCN positive feedback loop in MYCN^Amp neuroblastomas, nominating TEAD4 as a novel candidate for therapeutic intervention. Cancer Discov; 8(5); 582-99. ©2018 AACR.This article is highlighted in the In This Issue feature, p. 517.

Figure 1

High-risk neuroblastoma molecular subtypes classification and inference of master regulators. (A) Unsupervised consensus clustering of high-risk neuroblastoma GEPs was performed to establish molecular subtypes. Three subgroups were identified according to robustness of clustering and consistency between two cohorts, TARGET and NRC. (B) The overlap of top 500 up-regulated (red) and down-regulated (blue) genes for each subtype in TARGET and NRC dataset using Stage 1 GEPs as reference, with its corresponding odds ratios. (C) Copy number frequency per genomic location of individual molecular subtypes showing segregated pattern of 11q, 3p and 1p loss. Gains are considered when the log₂ ratio between tumor and blood > 1.1, while losses are considered for log₂ ratios < 0.9 (D) Top activated MRs (red) of high-risk subtypes are represented using VIPER inference of TF activity using stage 1 samples as a control group. (E, F and G) REACTOME pathway enrichment analysis of MYCNA, 11q-LOH and MES subtype gene expression signatures. Axis represents –log₁₀ of the p-value while retaining the directionality of the enrichment score. Also see Figure S1 and supplementary experimental procedures.

Figure 2

RNAi screening identifies MYCNA subtype specific MRs. (A) Master regulator analysis of MYCNA versus Stage 1 tumors performed independently on both TARGET and NRC datasets show strong reciprocate reproducibility of both activated (red) and deactivated (blue) top 50 MRs (Supplementary Table S4). (B) The top 25 integrated MRs of MYCNA subtype selected for validation. The map shows distribution of positively (red) and negatively (blue) regulated targets of each MR ranked by differential expression between MYCNA subtype versus stage I patient samples. (C) In-vivo pooled shRNA screening in MYCNA (BE2) versus control cells (SKNAS) and (D) In-vitro pooled shRNA screening in MYCNA (BE2, IMR-5) versus control cells (NLF, SK-NAS), depicting average effect of putative MR silencing in MYCNA cells compared to control cells. For both (C, D), tumor-enriched shRNAs were amplified, sequenced and counted to identify enrichment and dropouts. shRNA abundance for a gene was integrated into a score and calculated as a ratio of T_final to T_initial. The MRs were first screened to include only the ones with p <0.05 in MYCNA group (red) and average fold change between MYCNA cells versus control cells was calculated. The grey dashed line shows the cutoff for −2.0 fold change. (E) Scatter plot of average relative cell viability of MYCNA cells (BE2, IMR5, IMR32, NB1, LAN1) versus control cells (SY5Y, SKNAS, SKNFI) upon transduction with 2 shRNAs per MR, normalized to control shRNA, measured 72 to 96hrs post transduction. (F) Scatter plot of average cell viability of MYCNA cells (BE2, IMR5, SKNDZ) versus control cells (SY5Y, SKNAS) upon transfection with ON-Target smartpool siRNA against each MR normalized to control siRNA, measured 96hrs post transfection. For both (E, F), the red dashed lines show the cutoff of a < 0.8 (cell viability reduction) for MYCNA cells and b - a > 0.2 (cell type specificity). Experiments were run in triplicate. Representative experiments are shown. (G) Venn diagram depicting potential MYCNA subtype specific MRs from (C, D) MRs common to both in-vitro and in-vivo negative selection pooled shRNA screening (E) individual shRNA screening (F) and siRNA screening. Additional data in Supplementary Figure S5 and Table S5.

Figure 3

TEAD4 is the master regulator of MYCNA subtype. (A) Heatmap representing gene expression changes of MYCNA subtype specific MRs from Figure 2G, upon transduction of BE2 cells with control or respective shRNAs against each MR, measured by qRT-PCR, 48hrs post-transduction. The genes showing >1.5 fold downregulation of transcript upon treatment with the shMR was considered to be a target and are displayed in the map. Samples were run in triplicate and representative experiments are shown. (B) The inter-regulatory network derived from the results in (A). *MYCN binds to the promoter of the genes by ChIP assay (Figure S5F); *TEAD4 binds to proximal region of these genes by ChIP-seq experiment (Supplementary Table S6) (C) TEAD4 (x-axis) and MYCN (y-axis) knockdowns signatures compared with MYCNA versus stage1 signature (red-blue heat colors). (D) Venn-diagram showing proportion of MYCNA subtype signature up-regulated (red) and down-regulated (blue) genes by MYCN, TEAD4 or both knockdown signatures. (E) REACTOME and (F) Gene Ontology pathway enrichment analysis performed on TEAD4 (x-axis) and MYCN (y-axis) knockdown signatures. Red circle represent positive TEAD4 targets genes (down-regulated upon knockdown) while blue circle represents negative targets (up-regulated after knockdown). Fisher’s exact test was used to calculate the statistical significance of both overlaps using a background list of 18,179 genes included in the RNA-seq signature. (G) Overlap between differentially expressed genes after TEAD4 knockdown, peak targeted genes from TEAD4-Ab ChIP-seq, and MYCNA subtype signature up-regulated genes, with the corresponding KEGG pathway enrichment analysis on the overlapping genes. See also Supplementary Figure S6 and Supplementary Tables S6 and S7.

Figure 4

TEAD4 promotes MYCN protein stabilization. (A) Immunoblot of TEAD4, MYCN, CDK1 and AURKA proteins in BE2 cells transduced with control or two different TEAD4 shRNAs in a time course experiment (B) qPCR analysis showing transcript levels of the corresponding genes, 2 days post-transduction. (C) Regulatory hierarchical model showing a TEAD4 ↔ MYCN positive feedback loop controlling the master regulatory module (D) Immunoblot of TEAD4 and MYCN proteins in BE2 cells transduced with control and TEAD4 shRNA, and treated with CHX for indicated times (E) Quantification of MYCN protein stability from results shown in (D) where MYCN levels were normalized to GAPDH (F) Immunoblot of TEAD4 and MYCN 72hrs post-transduction from cells treated with DMSO or MG-132, 4hrs before harvesting (G) Densitometry analysis of MYCN proteins from results shown in (F), where MYCN levels were normalized to GAPDH. Representative experiments are shown.

Figure 5

TEAD4 is required for cell cycle progression and cell growth of MYCNA cell lines. (A) Scatter plot representing MYCN and MYC expression in MYCNA and non-MYCNA samples from TARGET cohort. Single sample activity of TEAD4, is represented as normalized enrichment score (NES). NRC cohort results provided in Supplementary Figure S7H. (B) The effect of TEAD4 on anchorage-independent growth in MYCNA and control cell lines was evaluated by soft agar colony assays, 21 days post transduction (C) Immunoblot analysis confirming silencing of TEAD4 in the corresponding cell lines (D) GSEA plot evaluating enrichment for KEGG cell cycle gene set in shTEAD4 signature (upper) and leading edge cell cycle genes (lower) colored by their signature t-score; yellow and red asterisk indicate genes with assigned anti-TEAD4 ChIP-seq peaks by proximity and overlap criterion respectively (supplementary Experimental Procedures). (E) Cell cycle profile and (F) cellular proliferation, assessed upon treatment of BE2 cells with control or TEAD4 shRNA, 48hrs post transduction by flow cytometry. Representative experiments are shown. See also Figure S8.

Figure 6

TEAD4 is overexpressed in high-risk neuroblastoma tumors. (A) Expression of TEAD4 across high risk and stage 1 subtypes (B) VIPER transcriptional activity of TEAD4 across high risk and Stage 1 subtypes. (C) Histogram of primary neuroblastoma samples stained for TEAD4 protein by immunohistochemistry on a tissue microarray, segregated by risk level and (D) MYCN-amplification status in high-risk neuroblastoma, showing differential pattern of TEAD4 protein staining intensity (where 0=no staining; 1=low staining; 2=moderate staining; 3=high staining). See also Supplementary Table S8.