Aberrant super-enhancer landscape reveals core transcriptional regulatory circuitry in lung adenocarcinoma

Abstract

Lung adenocarcinoma (LUAD) relies on dysregulated gene expression to sustain its infinite growth and progression. Emerging evidence indicates that aberrant transcriptional program results from core transcriptional regulatory circuitry (CRC) which is driven by super-enhancers (SEs). In this study, by integrating profiles of H3K27Ac chromatin immunoprecipitation sequencing (ChIP-seq) from normal adult lung and LUAD cell lines, we revealed that widespread alterations of the super-enhancer were presence during lung carcinogenesis. With SE-based modeling of regulatory circuits and assessments of transcription factor (TF) dependencies, we reconstructed an interconnected transcriptional regulation network formed by three master TFs, including ELF3, EHF, and TGIF1, all of which promoted each other's expression that confirmed by ChIP-qPCR and western blot. Loss-of function assay revealed that each of them is essential for LUAD cells survival, invasion and metastasis. Meanwhile, the rescue assay also illustrated the transacting transcriptional regulatory circuitry. In addition, the mRNA levels of ELF3, EHF, and TGIF1 were differentially expressed in LUAD tumors and peritumoral tissue. IHC of serial sections revealed that high expressions of CRC (ELF3/EHF/TGIF1-High) were closely associated with high proliferative activity in tumor tissue and poor prognosis on patients with LUAD. Finally, we used small molecular inhibitors to perturb the transcriptional circuitry, also exhibited a prominent anti-cancer effect in vitro. Our findings reveal the mechanism of the transcriptional dysregulation and addiction of LUAD.

Fig. 1. The super-enhancer landscape of LUAD cells differed significantly from that of the normal adult lung.

A Enhancer regions of the normal adult lung, A549, PC-9 cells were plotted in increasing order based on their H3K27Ac ChIP-Seq signal. Enhancers above the inflection point of the curve were defined as SEs. Examples of SE-genes are listed along with their respective rank. B SE-associated gene in A549 and PC-9 cells were systematically compared with SE-associated genes in normal lung tissue. Totally, 1278 and 848 SE-associated genes were acquired in A549 and PC-9 cells, respectively. A total of 1792 SE-associated genes were acquired in A549 and PC-9 cells, in which 375 SE-associated genes were commonly acquired in both cell lines. C The gene ontology enrichment analysis suggested that LUAD-SE genes were significantly associated with biological processes essential to cancer sustainability. D The KEGG pathway enrichment analysis revealed that LUAD-SE genes were significantly enriched in multiple cancer-relating signaling pathways. E Volcano plot illustrating the biological and statistical significance of LUAD-SE genes are highlighted in red, and non-SE-genes are highlighted in gray. F The expression profile of the common LUAD-SE gene was clearly segregated between LUAD samples and NT samples in the unsupervised clustering analysis.

Fig. 2. Master TFs in CRC, identified from LUAD-SE genes are frequently overexpressed in human LUAD and are correlated with poor prognosis in LUAD patients.

A Experimental validation for LUAD cell circuitry. Core regulatory circuit containing ELF3, EHF, and TGIF1 for LUAD cell lines, A549 and PC-9. B Intersections of TFs between two individual LUAD cell lines, A549 and PC-9, identified by CRC models. C The expression levels of selected TFs in tumor or normal tissues of patients with LUAD based on the TCGA illumina HiSeq RNA-Seq data. The x-axis represents the different tissues of LUAD patients and the y-axis represents expression read counts of TCGA. D Kaplan–Meier survival analysis with TCGA datasets indicates that higher ELF3 (p = 0.011), EHF (p = 0.00024), and TGIF1 (p = 0.00028) expression is associated with a worse overall survival in patients with LUAD. E ChIP-seq occupancy profiles of H3K27Ac at the normal adult lung, A549 cell line and PC-9 cell line. H3K27Ac ChIP-seq signal was highly enriched in the ELF3, EHF and TGIF1 genomic loci of A549 and PC-9 cells but not the normal liver. TCGA The Cancer Genome Atlas.

Fig. 3. The regulatory interaction network in CRC model of LUAD-SE-associated TFs is interdependent.

A–C, J Expressions of all master TFs in the knockdown of any master TF and validating the efficiency of siRNA targeting to master TFs by both real-time PCR and western blot. D–I Western blot and real-time PCR demonstrating that siRNA co-transfected with pcDNA3.1 could decrease the master TF expression significantly in both protein and mRNA level. This phenomenon could be partly reversed by either of another two master TFs, also could be reversed almost entirely in the combination of another two master TFs. K–M Data are shown as fold enrichments of master TFs promotor sub-regions in each antibody immunoprecipitate, ELF3 antibody (K), EHF antibody (L), TGIF1 antibody (M), over control IgG immunoprecipitate. GAPDH and β-actin were used as internal control, the lower panels showed the gray scale ratio of protein (ELF3) to GAPDH and protein (EHF and TGIF1) to β-actin. *p < 0.05; **p < 0.01; ***p < 0.001.

Fig. 4. ELF3, EHF, and TGIF1 enhances the malignant phenotypes of LUAD cells.

A, B Inhibition of invasive and migrate activity by knockdown master TFs, ELF3, EHF, and TGIF1. PC-9 cells were transfected with indicated siRNA-pool and subjected to invasion and migration assays (see “Materials and methods”). Invaded and migrated cells were stained with crystal violet and counted (see Fig. S3). Histograms represent the number of invasion or migration. Representative photographs were shown. C Colony assay performed on PC-9 cells transfected with either scramble or siRNA-pool targeting ELF3, EHF, and TGIF1. D Silencing of ELF3, EHF, and TGIF1 increased clonal cell growth in a 3D spheroid culture system. PC-9 cells transfected with scramble or siRNA-pool were cultured as spheroid in DMEM medium and the spheroids was observed after 48 h. E Inhibition of cell migration by knockdown master TFs, ELF3, EHF, and TGIF1. PC-9 cells were transfected with indicated siRNA-pool and subjected to scratch assay (see “Materials and methods”). Representative photographs were taken and migration areas were measured. Statistical analysis was performed by using t tests; data are shown as bar graphs of the mean ± SD. of three independent experiments. Scale bars (white) indicate 250 μm. F The effects of pcDNA3.1-ELF3 and/or pcDNA3.1-EHF co-transfected with TGIF1 siRNA on cell migration and invasiveness in LUAD cells were examined by Transwell assays using a Boyden chamber in the presence or absence of Matrigel, respectively. Histograms represent the number of invasion or migration. Data were representative of two to three separate experiments performed in triplicate. *p < 0.05; **p < 0.01; ***p < 0.001.

Fig. 5. Inhibition of the SE-associated key targets attenuated LUAD malignant progression via the suppression of master TFs in CRC model.

A–C Perturbation of SE associated key targets, including BRD4, EP300, and CDK7, by small-molecule inhibitors reduced the expression of master TFs significantly at protein levels. D Expressions of master TFs in LUAD cell line PC-9 treated with SE associated targets inhibitors in mRNA level were analyzed by real-time PCR. E Inhibition of invasive and migrate activity by perturbation of SE associated key targets via small molecular inhibitions. PC-9 cells were cultured with indicated small molecular inhibitions and subjected to invasion and migration assays (see “Materials and methods”). Invaded and migrated cells were stained with crystal violet and counted. Representative photographs were shown. F, G Histograms represent the number of invasion or migration cells. H In PC-9 cell lines, alterations of ELF3, EHF, and TGIF1 at protein levels while JQ1 treated for 100 nM 48 h together with or without transfected with overexpression plasmids. GAPDH and β-actin were used as internal control, the lower panels showed the gray scale ratio of protein (ELF3) to GAPDH and protein (EHF and TGIF1) to β-actin. *p < 0.05; **p < 0.01; ***p < 0.001.

Fig. 6. High expression of CRC (ELF3/EHF/TGIF1-High) were associated with poor prognosis on patients with LUAD.

A qRT-PCR analysis of ELF3, EHF, and TGIF1 among 20 pairs of matched lung adenocarcinoma and adjacent normal tissue. B Correlation analyses conducted between the mRNA expression of ELF3, EHF, and TGIF1 from (A). C ELF3, EHF, and TGIF1 were detected in peritumoral tissue and representative LUAD samples by immunohistochemistry staining. Scale bar, 100 μm. D Kaplan–Meier analysis of overall survival of LUAD patients (TCGA-LUAD) with low (low expression of ELF3, EHF, and TGIF1, n = 78) and high (high expression of ELF3, EHF, and TGIF1, n = 52) ELF3/EHF/TGIF1 expression (log-rank test, two-sided). E Left panel: a schematic diagram showing the key targets and small-molecule inhibitors of SE. JQ1 and OTX015 targeting BRD4, THZ1 targeting CDK7; CBPC30 targeting EP300. Right panel: CRCs are assembled as fully interconnected loops of auto-regulated TFs. Schematic graph of the model of interconnected circuitry, with hexagons representing enhancer elements and proteins, respectively. Data in (A) were mean ± SD; n = 3 independent experiments, two-tailed paired Student’s t test, ns not significant, *p < 0.05; **p < 0.01; ***p < 0.001 (B) R, Pearson correlation coefficients (r) and p values.