Comprehensive multi-omics analysis uncovers a group of TGF-β-regulated genes among lncRNA EPR direct transcriptional targets

Abstract

Long non-coding RNAs (lncRNAs) can affect multiple layers of gene expression to control crucial cellular functions. We have previously demonstrated that the lncRNA EPR, by controlling gene expression at different levels, affects cell proliferation and migration in cultured mammary gland cells and impairs breast tumor formation in an orthotopic transplant model in mice. Here, we used ChIRP-Seq to identify EPR binding sites on chromatin of NMuMG mammary gland cells overexpressing EPR and identified its trans binding sites in the genome. Then, with the purpose of relating EPR/chromatin interactions to the reshaping of the epitranscriptome landscape, we profiled histone activation marks at promoter/enhancer regions by ChIP-Seq. Finally, we integrated data derived from ChIRP-Seq, ChIP-Seq as well as RNA-Seq in a comprehensive analysis and we selected a group of bona fide direct transcriptional targets of EPR. Among them, we identified a subset of EPR targets whose expression is controlled by TGF-β with one of them-Arrdc3-being able to modulate Epithelial to Mesenchymal Transition. This experimental framework allowed us to correlate lncRNA/chromatin interactions with the real outcome of gene expression and to start defining the gene network regulated by EPR as a component of the TGF-β pathway.

Figure 1.

(A) qRT-PCR analysis of nascent mRNAs in NMuMG cells stably transfected with either the empty vector (mock, black bars) or EPR (NMuMG-EPR, green bars). Please note the logarithmic scale of the graph. The values of qRT-PCR experiments shown are averages (±SEM) of three independent experiments performed in triplicate. Statistical significance: **P < 0.001 (Student's t test). (B) Schematic representation of the experimental framework adopted in this study.

Figure 2.

ChIRP-Seq analysis of chromatin targets of EPR. (A) Representative snapshots of ChIRP-Seq experiments—centered on the indicated genes—showing two proximal (Il6ra, Mettl7a1) and two distal (Hdac11, Ocln) targets of EPR. The genomic coordinates of each target gene are represented on the top of each panel. The position of the relevant ChIRP peaks (overlapping in both EVEN and ODD samples) is marked by a green arrowhead while blue arrows indicate the transcription direction. (B) qPCR analysis of the EPR genomic targets. Both input DNA and DNA purified using either ODD (red bars) or EVEN (blue bars) tiling oligonucleotides were analyzed by qPCR to amplify a region in Rpl32 and B2m genes (negative controls) or the indicated target genes. Values, represented as percentage of input, are averages (±SEM) of three independent experiments performed in triplicate.

Figure 3.

Analyses of EPR genomic targets. (A) HOMER de novo transcription factor binding motifs enriched in either proximal (left) or distal (right) binding sites of EPR to chromatin. P-values for motif enrichment are shown. (B) Chromatin prepared from NMuMG-EPR was immunoprecipitated using either normal mouse IgG (cIgG) or mouse monoclonal anti-SMAD3 antibody. The association of SMAD3 with a select group of EPR binding sites on chromatin was verified by qPCR using specific primers. The values of qPCR experiments shown are averages (±SEM) of three independent experiments performed in triplicate. Statistical significance: **P < 0.001 (Student's t test). (C) Triplex Domain Finder (TDF) analysis of the interaction between EPR and its proximal (left) or distal (right) targets. Upper panels, the number of triplexes is shown in blue while regions highlighted in grey indicate significant DNA binding domains (DBDs) (y-axis). The position of triplexes and DBDs is presented with respect to the EPR sequence (x-axis). Pink bars mark the position of EPR regions able to undergo autobinding. Lower panels, TDF analysis reveals a high propensity (higher z-score) of domains I, V and VI of EPR to form triple helices when compared to other domains. (D) Pie charts showing the percentage of EPR target sequences that can associate with EPR DNA binding domains.

Figure 4.

Profiling of activation histone marks. (A and B). Volcano plots showing genes with differential occupancy by either H3K27ac (panel A) or H3K4me3 (panel B) marks in NMuMG-EPR versus mock. (C and D) Pie graphs showing the percentage of genes that, among those induced in NMuMG-EPR when compared to mock, display enhanced occupancy by either H3K27ac (panel C) or H3K4me3 (panel D) marks in proximal and distal regions as indicated.

Figure 5.

Identification of bona fide direct transcriptional targets of EPR. (A) Pie graph showing the percentage of genes that, among those directly bound by EPR, display enhanced H3K4me3 deposition at proximal or distal target regions upon EPR overexpression. (B) Pie graph showing the percentage of genes that, among those directly bound by EPR, display enhanced H3K27ac deposition at proximal or distal target regions upon EPR overexpression. (C) Pie graph showing the percentage of genes that, among those induced EPR overexpression in NMuMG cells, display EPR binding to either proximal or distal target regions. (D) UpSet plots showing the integration of the distinct genomic analyses performed in this study. The matrix shows the number of genes with the indicated combinations of enhanced gene expression (RNA-Seq), enhanced occupancy by either H3K27ac or H3K4me3, and presence of EPR binding (ChIRP-Seq). The size of datasets is represented by the horizontal bars displayed on the far right. (E) qRT-PCR analysis of the indicated transcripts in either mock or NMuMG-EPR cells. (F) qRT-PCR analysis of the indicated transcripts in either NMuMG-EPR (upper) or wild-type NMuMG cells (lower) transiently transfected with either control siRNA (siC) or siRNA designed to silence EPR expression (siEPR). The values of qRT-PCR experiments shown are averages (±SEM) of three independent experiments performed in triplicate. Statistical significance: *P < 0.01, **P < 0.001 (Student's t test).

Figure 6.

A group of direct EPR targets is regulated by TGF-β and silencing of the Arrdc3 gene upregulates the expression of EMT factors. (A) qRT-PCR analysis of the indicated targets of EPR in NMuMG cells serum-starved (2% FBS, 16 h) and either treated with TGF-β (5 ng/ml) for 24 h or untreated. (B) qRT-PCR analysis of the indicated EPR target transcripts in either mock or EPR-overexpressing (EPR) NMuMG cells serum-starved and either treated with TGF-β (+) for 24 h or left untreated (−). (C) qRT-PCR analysis of the indicated transcripts in NMuMG-EPR transiently transfected with either control esiRNA (siC) or esiRNA designed to silence Arrdc3 expression (siArrdc3). (D) qRT-PCR analysis of the indicated transcripts in NMuMG-EPR transiently transfected with either control esiRNA (siC) or esiRNA designed to silence Arrdc3 expression (siArrdc3). 24 hours after transfection, cells were serum starved and then either treated with TGF-β (+) for 24 h or left untreated (−). The values of qRT-PCR experiments shown are averages (±SEM) of three independent experiments performed in triplicate. Statistical significance: *P < 0.01, **P < 0.001 (Student's t test).