Transcription factor 4 is a key mediator of oncogenesis in neuroblastoma by promoting MYC activity

Abstract

Super-enhancer-associated transcription factor networks define cell identity in neuroblastoma (NB). Dysregulation of these transcription factors contributes to the initiation and maintenance of NB by enforcing early developmental identity states. We report that the class I basic helix-loop-helix (bHLH) transcription factor 4 (TCF4; also known as E2-2) is a critical NB dependency gene that significantly contributes to these identity states through heterodimerization with cell-identity-specific bHLH transcription factors. Knockdown of TCF4 significantly induces apoptosis in vitro and inhibits tumorigenicity in vivo. We used genome-wide expression profiling, TCF4 chromatin immunoprecipitation sequencing (ChIP-seq) and TCF4 immunoprecipitation-mass spectrometry to determine the role of TCF4 in NB cells. Our results, along with recent findings in NB for the transcription factors T-box transcription factor TBX2, heart- and neural crest derivatives-expressed protein 2 (HAND2) and twist-related protein 1 (TWIST1), propose a role for TCF4 in regulating forkhead box protein M1 (FOXM1)/transcription factor E2F-driven gene regulatory networks that control cell cycle progression in cooperation with N-myc proto-oncogene protein (MYCN), TBX2, and the TCF4 dimerization partners HAND2 and TWIST1. Collectively, we showed that TCF4 promotes cell proliferation through direct transcriptional regulation of the c-MYC/MYCN oncogenic program that drives high-risk NB. Mechanistically, our data suggest the novel finding that TCF4 acts to support MYC activity by recruiting multiple factors known to regulate MYC function to sites of colocalization between critical NB transcription factors, TCF4 and MYC oncoproteins. Many of the TCF4-recruited factors are druggable, giving insight into potential therapies for high-risk NB. This study identifies a new function for class I bHLH transcription factors (e.g., TCF3, TCF4, and TCF12) that are important in cancer and development.

Keywords: MYCN; TCF4; core regulatory circuit; neuroblastoma; super‐enhancer; transcription factor.

Fig. 1

TCF4 is highly expressed in neuroblastoma and JQ1 suppresses the expression of TCF4 in multiple neuroblastoma lineage cells. (A) Dose–response of cells after 4‐day treatment with half‐log dilutions of the BET inhibitor JQ1. Cell lines include neural crest cells (NCCs), NCCs overexpressing N‐Myc (N‐Myc NCC), NCCs pooled from p53−/− and p53+/− embryos (NCC p53 mixed), cell line from N‐Myc tumor (N‐Myc Tu), cell lines from N‐Myc p53 mixed tumors (N‐Myc; p53 mixed Tu) and control NIH3T3. Results were normalized to control + SE. n = 3 independent experiments. (B) A relationship plot generated from the Super‐enhancer database (SEdb 2.0) using the SE‐based transcription factor (TF) TF‐GENE Analysis program to comprehensively analyze TFs gene pairs mediated by super‐enhancers (SEs) in neuroblastoma (NB) cell lines. (C) TCF4 expression (log₂) in NB tumors and NB cell lines as compared to other tumors or cell lines. Boxplots show the 1st quartile up to the 3rd quartile of the data values and median as a line within the box. Number of samples for each cancer are shown in parenthesis. (D) TCF4 mRNA levels were determined by quantitative real‐time PCR after NB cell lines and primary NCCs were treated with DMSO, 1 μm JQ1 or SJ018 for 3 h. Expression values are shown relative to the DMSO condition for each cell line. n = 3 biological replicates. Data are presented as the mean ± SE two tailed unpaired t‐test (***P < 0.001 vs. control). (E) Immunoprecipitation (IP) of TCF4 using Kelly and SK‐N‐AS whole lysate. n = 3 independent experiments, and the clearest mages from these replicates were selected for presentation. Western blot is probed with the indicated antibodies. Control IP by rabbit IgG and 10% input are also shown.

Fig. 2

TCF4 is essential for neuroblastoma viability. (A) Quantitative RT‐PCR analysis showing TCF4 expression in Kelly and SK‐N‐AS stable cells treated with or without 1 μg·mL⁻¹ doxycycline for 3 days. Data are presented as the means ± SE from three independent experiments. Two tailed unpaired t‐test (**P < 0.01, ***P < 0.001 vs. control). (B) Whole‐cell protein lysates were analyzed by western blotting using TCF4 antibody 5 days after 1 μg·mL⁻¹ doxycycline treatment. N = 3 independent experiments, and the clearest mages from these replicates were selected for presentation. (C) CyQuant proliferation assay performed using SK‐N‐AS TCF4 sh #1, #2, #3 stable cell lines compared to empty vector control (NTC) cell line 7 days after doxycycline treatment. n = 3 biological replicates. Two tailed unpaired t‐test (*P < 0.05, **P < 0.01, ***P < 0.001 vs. control). (D) Colony formation assays were performed following TCF4 knockdown in SK‐N‐AS. Cells were cultured for 10 days in the presence or absence of 1 μg·mL⁻¹ of doxycycline. (E) % of cells in each phase of the cell cycle 5 days following TCF4 knockdown in the SK‐N‐AS cell line. Cell cycle was assayed by flow cytometry. n = 3 biological replicates. Data are presented as the mean ± SE two tailed unpaired t‐test (**P < 0.01 vs. control). (F) Quantitative analysis of the percentage of apoptotic cells (Annexin V+/FITC+) in SK‐N‐AS TCF4 stable cell lines treated with or without 1 μg·mL⁻¹ doxycycline for 7 days. Two tailed unpaired t‐test (*P < 0.05, **P < 0.01, vs. control). (G) TCF4 overexpression was analyzed with western blot analysis in the stable SK‐N‐AS clones. (H) Annexin V/FITC staining of parental SK‐N‐AS cells and SK‐N‐AS overexpressing TCF4 cultured in the presence of increasing concentrations of the BET inhibitor JQ1 for 4 days. Data are presented as the means ± SE from three independent experiments. Two tailed unpaired t‐test (*P < 0.05 vs. control).

Fig. 3

TCF4 knockdown suppresses tumor growth in vivo. (A, B) Soft‐agar colony formation assay (Clonogenic Assay) of the Kelly and SK‐N‐AS TCF4 stable cell lines at 21 days after doxycycline treatment. Experiments were performed in triplicates. Representative images are shown, scale bar, 500 μm. Growth curves for subcutaneous xenograft transduced with (C) Kelly TCF4 sh2, (D) SK‐NA‐S sh2 injected in the flank region of nude mice. One week after the injections, mice were assigned to either (−Dox) or (+Dox) feed (Con group = 4 mice, Dox group = 5 mice). Data is presented as the means ± SE two tailed unpaired t‐test (*P < 0.05, **P < 0.01). (E) TCF4 knockdown after doxycycline treatment was confirmed by immunoblot in tumors formed from Kelly and SK‐N‐AS neuroblastoma cells, control or transduced cells with an shRNA targeting TCF4. Experiments were performed in triplicates.

Fig. 4

Knockdown of TCF4 deregulates gene expression of MYC target genes as well as genes involved in cell cycle. (A, B) Enrichment plots acquired from the gene set enrichment analysis (GSEA). Four significant pathways among the top enriched ones in vehicle‐treated cells compared to doxycycline‐treated cells upon TCF4 knockdown in Kelly and SK‐N‐AS. FDR < 0.01 was defined as statistically significant. (C, D) Enrichr pathway analysis of downregulated differentially expressed genes (DEG) following TCF4 knockdown in Kelly and SK‐N‐AS cells using the ARCHS4 TF Coexp. The lists of genes were analyzed based on the combined score ranking. P‐value < 0.05 was used as the significance threshold.

Fig. 5

TCF4‐dependent regulatory network in neuroblastoma. (A) Heatmap image represents genes down‐ or up‐regulated in both Kelly and SK‐N‐AS cells after TCF4 knockdown using two different shRNAs (#2, #3) n = 3 biological replicates, containing a TCF4 ChIP‐Seq peak within the promoter (1000 bp from TSS based on gene orientation) (2 biological replicates), the gene body or capture Hi‐C data. Highlighted in red are DREAM complex components. (B) Heatmap indicating the binding intensity of TCF4 at promoters or enhancers (Homer annotation) within 5 kb of ChIP‐seq peaks in the Kelly and SK‐N‐AS cell lines. The color scale shows the intensity of the distribution signal. n = 2 biological replicates. (C) Enriched DNA‐binding motifs identified by HOMER corresponding to known transcription factors. (D) Aggregated ChIP‐seq signals for TCF4, H3K27ac, ATAC, MYCN, and the CRC members HAND2, PHOX2B, GATA3, ISL1, ASCL1 peaks in the Kelly cell line for the regions (−5000 to +5000 bp) from the TCF4 peak summits of all TCF4 peaks. Two replicates in Kelly are depicted.

Fig. 6

TCF4 interactome in neuroblastoma. (A) Proteins interacting with TCF4 in both Kelly and SK‐N‐AS neuroblastoma cells, identified by immunoprecipitation coupled to mass spectrometry (IP‐MS). Normal rabbit IgG was used as a negative control. Identified proteins are high‐confidence proteins identified in at least two independent (IP‐MS) reactions per cell line, found in both Kelly and SK‐N‐AS cells. n = 3 independent experiments. TCF4 interaction partners shared between Kelly and SK‐N‐AS cells are denoted in gray, TCF4 interactors identified in Kelly cells only are in blue, and SK‐N‐AS only are in orange. Bold circles represent transcription factors identified as TCF4 interactors in the (IP‐MS) analysis and have enriched DNA‐binding motifs identified in the TCF4 ChIP‐seq in both cell lines. (B) Volcano plot of HAND2 interacting proteins identified by IP‐MS with fold‐change versus significance of the change due to TCF4 silencing. Proteins that are significantly lost (blue or purple) or gained (red) after TCF4 silencing. n = 3 independent experiments. Purple points indicate proteins that complex with HDAC proteins. Dash lines demarcate proteins that are significantly changed (P < 0.05) and have a fold‐change greater than ±2‐fold.