Article Figures & Data
Figures
- Figure 1. Regulation of Platr embryonic long noncoding RNAs during epithelial–mesenchymal transition (EMT).
(A) Schematic of the mouse tumor progression model. Non-transformed normal murine mammary gland (NMuMG) cells were silenced for hnRNP E1 and injected into the mammary fat pad of 6-wk old NOD/SCID mice. After ∼12 wk, cells from the primary tumor (M1P) and lung metastases (L1P) were isolated and harvested for culture in vitro. Each established cell line is presented on phase contrast micrographs. (B) Volcano plots of the RNA-Seq analysis of noncoding transcripts expression at the EMT (E1KD versus NMuMG) and metastatic (L1P versus M1P) steps of tumor progression. Only long noncoding transcripts are plotted. (C) Quantitative RT-PCR analysis of Platr18 transcript, mesenchymal markers vimentin (VIM), N-Cadherin (CDH2) and snail (SNAI1), epithelial markers occludin (OCLN) and E-cadherin (CDH1) in hnRNP E1–derived cells. Data are normalized to HPRT and expressed in fold changes compared with the control with mean ± SD. ANOVA P < 0.0001. (D) (Top) Heat map from RNA-Seq analysis of Platrs expression in the NMuMG-derived breast tumor progression model. Each cell line is in duplicate. (Bottom) Immunoblot analysis of PCBP1 expression in the tumor progression series. (E) Quantitative RT-PCR analysis of Platr18 transcript, mesenchymal markers vimentin (VIM), N-Cadherin (CDH2) and Snail (SNAI1), epithelial markers occludin (OCLN) and E-cadherin (CDH1) transcripts expression during TGFβ-mediated EMT in NMuMG cells. Data are normalized to HPRT and expressed in fold changes compared to the control with mean ± SD (n = 3–6). (F, G) Immunoblot analysis (top panel) of E-cadherin, ZO1 and occludin epithelial markers and vimentin and N-cadherin mesenchymal markers expression during TGFβ-induced EMT (F) or TGFβ-retrieval model of mesenchymal–epithelial transition after 10 d of continuous TGFβ exposure in NMuMG cells (G). Heat map (bottom panel) of Platr noncoding RNA genes expression during TGFβ-mediated cell EMT of NMuMG cells. RNA-Seq data extracted from GSE112797 with each time point in duplicate (n = 2). Data are mean ± SD; NS, nonsignificant, **P < 0.05, ***P < 0.01, ***P < 0.001, or P-values are from ANOVA. HSP90 serve as loading controls. Experiments have been repeated three times or as specified in the legend. Scale bars: 50 μm.
- Figure S1. Platr non-coding genes are reactivated during cancer cell plasticity and metastasis.
(A) Affymetrix transcriptomic analysis of genes expression in the tumor progression model. Heat map showing the differentially expressed genes across the series. Affymetrix array from GSE94637 with each sample in triplicate (left). Gene ontology term enrichment classification by cell component of the set of the most regulated genes (F score >5) (right). (B) Affymetrix expression profile of Platr18 in the NMuMG-derived breast tumor progression model. (C) Heat map of Platrs expression in the TGFβ-induced epithelial–mesenchymal transition (left) or TGFβ-retrieval model of mesenchymal–epithelial transition after 10 d of continuous TGFβ exposure (right) of the PyMT-1099 cell line derived from mouse mammary tumor virus-PyVT model of breast cancer. RNA-seq data extracted from (GSE114572) with each time point in duplicate. (D, E) Platr-14/-16/-20 expression in NMuMG-derived breast tumor progression and TGFβ-mediated epithelial–mesenchymal transition analysis by using (D) end point PCR and (E) quantitative PCR. (n = 6) Data are mean ± SD; NS, nonsignificant; ***P < 0.001 or P-value from ANOVA. GAPDH serves as a loading control. Experiments have been repeated three times or as specified in the legend.
- Figure 2. The TGFβ/E1KD/Platr18 axis controls semaphorin-4F expression in vitro and in vivo.
(A) Single-molecule RNA-FISH labeling of NMuMG cells overexpressing Platr18 LincRNA or treated with TGFβ (bottom). Scale bar: 10 μm. (B) Volcano plot of the genome-wide transcriptome analysis of NMuMG overexpressing Platr18 (NMuMG-Platr18), silenced for hnRNP E1 (NMuMG-E1KD), treated with TGFβ for 4 d (NMuMG-TGFβ), and silenced for Platr18 in NMuMG-E1KD (NMuMG-E1KD-Platr18 KD). All genes are plotted. (C) (Left) Sema4F transcript expression in NMUMG cells overexpressing (+Platr18) or silenced for hnRNP E1(E1KD). (Right) Sema4F transcript expression after Platr18 silencing (Platr18 KD) in NMuMG or NMuMG silenced for hnRNP-E1 expression (NMuMG E1KD). Data are expressed in fold changes compared to the control with mean ± SD; *P < 0.05; ***P < 0.001 (n = 3). (D) Time course of Sema4F transcript expression under TGFβ exposure in (left) NMuMG cells or (right) NMuMG cells silenced for Platr18 expression (SCR versus Platr18 KD cells). Data are expressed in fold changes compared to the control with mean ± SD. TGFβ treatment: ANOVA P < 0.0001; ***P < 0.001; NS, not significant (n = 3). (E) Immunofluorescence of Sema4F protein expression and localization in NMUMG cells overexpressing Platr18 (NMUMG-Platr18), silenced for E1KD (NMUMG-E1KD), or treated with TGFβ (NMUMG + TGFβ) for 3 d. (F) Time course of IGSF11/VSIG3 transcript expression under TGFβ exposure, hnRNP E1 silencing (E1KD), or Platr18 expression. (n = 3) Scale bar: 20 μm. (G) Organoid cultures derived from mouse mammary tumor virus (MMTV)-Cre− PCBP1fl/fl or MMTV-Cre+ PCBP1fl/fl. Phase-contrast micrograph of organoid morphology and E-cadherin immunofluorescence (green) with nuclei counterstained by DAPI (blue). 60X magnification. Scale bar: 20 μm. (H) Immunofluorescence analysis of E-Cadherin and vimentin markers expression in MMTV-Cre− PCBP1fl/fl or MMTV-Cre+ PCBP1fl/fl hTert-immortalized cells in culture in vitro. Scale bar: 20 μm. (I) Quantitative PCR analysis of PCBP1 (hnRNP E1), Platr18, and Sema4F expression in 2D cultures derived from MMTV-Cre− PCBP1fl/fl and MMTV-Cre+ PCBP1fl/fl mice after exposure to TGFβ (n = 3). Data are expressed in fold changes compared to the control with mean ± SD; ***P < 0.001; ANOVA P < 0.0001. Experiments have been repeated three times or as specified in the legend.
- Figure S2. Platr18 silencing.
(A) Characterization of Platr18 silencing. Quantitative PCR analysis of Platr18 relative expression in NMuMG cells expressing scrambled (SCR) or Platr18 specific (Platr18 KD) shRNAs and treated with TGFβ (0–5 d; 5 ng.mL−1) (left) or silenced for hnRNP E1 (E1KD) (right). (n = 3) Data are mean ± SD. (B) Platr18 lincRNA silencing does not prevent TGFβ-induced epithelial–mesenchymal transition. Western-blot analysis of epithelial (E-cadherin; occludin) and mesenchymal (vimentin; N-cadherin) markers expression in NMuMG cells expressing scrambled (SCR) or Platr18 specific (Platr18 KD) shRNAs and treated with TGFβ for up to 5 d. (C) Immunofluorescence staining of vimentin/E-cadherin expression in NMuMG cells overexpressing Platr18 (Platr18), silenced for Platr18 (shPlatr18), and treated TGFβ (+beta 3 d) or after silencing of hnRNP E1 (E1KD). Experiments have been repeated three times or as specified in the legend. Scale bar: 20 μm.
- Figure S3. TGFβ/hnRNPE1/Platr18 axis controls neurogenesis-related genes expression.
(A, B, C) RNA-seq experiment interpretation by gene ontology term enrichment classification (biological process) of the transcripts up-regulated in NMuMG cells treated with TGFβ (A) up-regulated in NMuMG cells silenced for hnRNP E1 (B) or down-regulated in E1KD cells silenced for Platr18 expression (C). (D) Sema4F is highly regulated gene during TGFβ-mediated epithelial–mesenchymal transition of PyMT-derived cells. RNA-Seq analysis of the transcriptomes of PyMT-1099 cells treated with TGFβ for 96 h (PyMT-1099 + TGFβ versus PyMT-1099). Data extracted from (GSE114572). All genes are plotted. (E) Gene ontology term enrichment classification of up-regulated transcripts in PyMT-1099 cells treated with TGFβ. (F) Phase contrast micrograph of 2D cultures of mouse mammary tumor virus-Cre− PCBP1fl/fl and mouse mammary tumor virus-Cre+PCBP1 fl/fl ± TGFβ treatment for 7 d.
- Figure S4. Supplementary figure 4.
(A) Gene ontology term enrichment classification of transcripts up-regulated in human liver carcinoma HePG2 cells silenced for hnRNP E1. RNA-Seq data extracted from the ENCODE project ENCSR603TCV & ENCSR635FRH. (B) Gene ontology term enrichment classification of transcripts up-regulated in 4T1 cells treated with TGFβ for 24 h. RNA-seq data extracted from GSE110912. (C) Validation of Sema4F silencing. Immunofluorescence staining of E1KD cells expressing scrambled (E1KD SCR) or Sema4F-specific shRNAs (E1KD S4F KD). Nuclei are counterstained with DAPI. Scale bar: 20 μm. (D) Sema4F silencing does not impair TGFβ-mediated changes in cell morphology. Phase contrast micrographs of NMuMG cells expressing scrambled (NMuMG SCR) or Sema4F specific (NMuMG S4F KD) shRNAs and treated with TGFβ for up to 2 d (0–2 d; 5 ng.mL−1). Scale bar: 100 μm. (E) PC12 cells do not differentiate after TGFβ exposure. Bright-field micrographs of PC12 cultures exposed to TGFβ (1–50 ng.mL−1) for 5 d. Nerve growth factor (NGF) serves as a positive control. Scale bar: 50 μm. (F) NMuMG-derived models do not produce significant amount of NGF. Western blot analysis of NGF production from whole cell lysates of NMuMG cells (WT), NMuMG cells overexpressing Platr18 (OE Platr18), silenced for hnRNP E1 (E1KD), or treated with TGFβ for 3 d. Mouse whole brain (Whole Brain) and N2A cell line (Neuro2A) lysates are used as positive controls. HSP90 is used as a loading control. Experiments have been repeated three times or as specified in the legend.
- Figure 3. The TGFβ/E1KD/Platr18 axis controls axonogenesis in vitro.
(A) Schematic of the PC12 axonogenesis assay in vitro (top panel). PC12 cells derived from pheochromocytoma of the rat adrenal medulla are exposed to culture supernatants of tumor cells (lower panels). Micrographs of PC12 cells complemented with cultured media from NMuMG cells (NMuMG), overexpressing Platr18 (-Platr18) silenced for Sema4F expression (Sema4F-KD), silenced for hnRNP E1 (-E1KD), silenced for Platr18 (Platr18 KD) and treated with TGFβ (+TGFβ). (B) Schematic of the Neuro2A axonogenesis co-culture assay in vitro (top panel). Mouse Neuro2A neuroblastoma cells are transduced with a lentivirus carrying the eGFP gene, FACS-sorted, and then cultured in the presence of tumor cells. The neurite outgrowth is tracked by fluorescence microscopy that allows distinguishing the neuronal component of the co-culture through the GFP expression (bottom panels). DAPI+/eGFP+ = Neuro2A; DAPI+/eGFP− = tumor cells. Neuro2A confocal microscopy pictures of the Neuro-2A cells cultivated alone (N2a Alone) or in the presence of NMuMG cells overexpressing Platr18 (Platr18), silenced for hnRNP E1 (E1KD), silenced for Platr18 (Platr18 KD), or pretreated with TGFβ prior (not during the co-culture) their addition to the Neuro2AGFP+ culture. (C) Scatter plot of the neurite outgrowth absolute quantification of the PC12 axonogenesis assay in vitro. For each culture condition (bottom table) the number of dendrites per cell is reported. (n=) number of PC12 cells per condition. The histogram displays the median ± interquartile range. *P < 0.05; **P < 0.01; ***P < 0.001. Experiments have been repeated three times or as specified in the legend. Scale bars: 50 μm.
- Figure 4. E1KD/Platr18/Sema4F axis controls primary tumor axonogenesis in vivo.
(A) hNRNP E1/PCBP1 genetic deletion in mouse mammary tumor virus (MMTV)/PyMT-driven breast cancer progression model in vivo. MMTV-PyMT, MMTV-Cre− (PyMT-WT), MMTV-PyMT, MMTV-Cre+ PCBP1fl/wt (PyMT-E1KO-Het) and MMTV-PyMT, MMTV-Cre+ PCBP1fl/fl (PyMT-E1KO), mouse genetic models were generated. Histograms represent the number of tumors observed per mouse for each group. Data expressed as mean ± SEM. P-value from ANOVA (n = 4). (B) Quantitative RT-PCR analysis of hnRNP E1 (PCBP1), Platr18, fibronectin (FN1), and N-cadherin (CDH2) expression in the primary tumors. Data are normalized to HPRT and expressed in fold changes compared with the control with mean ± SEM. P-values from ANOVA. (n = 5–11 as indicated). (C) Tubulin β3 (Tubβ3) IHC staining on primary tumors. 20× magnification. (D) (Left) Western blot analysis of neuron-specific tubulin β3 (Tubβ3) marker and hnRNP E1/PCBP1 expression in primary tumors of PyMT-WT and PyMT-E1KO primary tumors and (Right) quantification of the tubulin β3 signal compared with HSP90. Mean ± SD *P < 0.05. (E) Characterization of the PyMT-E1KO primary tumors innervation. Co-immunofluorescence analysis of neuron-specific marker tubulin β3 (Tubβ3–green) with sympathetic-specific markers (red) tyrosine hydroxylase (TH), sensory-specific marker capsaicin receptor (TRPV1), or parasympathetic marker choline acetyltransferase (ChAT). For each marker, a representative nerve twigs or a more organized fiber is displayed. 120× magnification. (F) In vivo analysis of primary tumor-related axonogenesis and lung metastasis in female NOD/SCID mice injected into the mammary fat-pad with NMuMG-E1KD control cells (SCR), silenced for Platr18 (Platr18 KD), or Sema4F (Sema4F KD). Tubulin β3 (Tubβ3) and Ki67 markers are used to IHC stain tumor innervation and lung metastasis, respectively. 20× magnification. (G) Characterization of the primary tumors innervation in the orthotopic model. Co-immunofluorescence analysis of neuron-specific marker tubulin β3 (Tubβ3 – green) with sympathetic-specific marker tyrosine hydroxylase (TH), sensory-specific marker capsaicin receptor (TRPV1), or parasympathetic marker choline acetyltransferase (ChAT). For each marker, a representative nerve twig or a more organized fiber is displayed. 120× magnification. (H) Western-blot analysis of neuron-specific tubulin β3 (Tubβ3) and sympathetic-specific tyrosine hydroxylase (TH) markers expression in primary tumors. (I) Quantification of tumor weight (g), lung metastases (number of nodules), and tumor axonogenesis (quantification of Tubβ3 marker expression in primary tumors) of mice injected with NMuMG-E1KD control cells (SCR) or silenced for Platr18 (Platr18 KD) or Sema4F (Sema4F KD). (n = 5 mice per group) Data are mean ± SD; **P < 0.01; ***P < 0.001, or P-value from ANOVA. HSP90 serves as a loading control. Experiments have been repeated three times or as specified in the legend. Scale bars: 50 μm.
- Figure 5. Epithelial–mesenchymal transition (EMT) controls tumor axonogenesis and metastasis in vivo.
(A) Characterization of dominant negative TGFβ type II receptor strategy in NMuMG model. Micrographs of NMuMG cell cultures expressing WT (NMuMG WT) or dominant negative TGFβ type II receptor (NMuMG DN RII) and treated with TGFβ to induce EMT. (B) Quantitative PCR analysis of Sema4F, fibronectin (FN1; mesenchymal marker), and E-cadherin (CDH1; epithelial marker) transcripts induction by TGFβ in the dominant negative TGFβ type II receptor model. Data are mean ± SD (n = 3). (C) Schematic of the dominant negative TGFβ type II receptor (RII) expression strategy to block TGFβ signaling. (D) Western blot analysis of Smad2 phosphorylation 30 m after TGFβ exposure (0.5 ng.mL−1) in the mouse 4T1 breast cancer cells expressing either the wild-type (WT) or dominant negative TGFβ type II receptor (DN RII) in vitro. (E) (Top) Lung metastases quantification are reported in the table. LM+ = number of mice positive for lung metastases; five mice per group. (Bottom) Representative picture of mice lungs 3 wk after mammary fat pad injection of 4T1 cells expressing WT (WT RII) or truncated TGFβ type II receptor (DN RII) & obtained tumors weight (g). Data are mean ± SD (n = 5 mice per group). (F) Tumor-related axonogenesis of tumors generated by mammary fat pad injection of NOD/SCID mice with 4T1 cells enabled (WT) or blocked (DN RII) for TGFβ signaling. Tubulin β3 immunostaining visualizes intra-tumoral innervation. (G) Western blot analysis of axonogenesis in tumors abolished for TGFβ signaling. Tubulin β3 and TH neuronal markers were used to quantify axonogenesis. Densitometry quantification of Tubulin β3 is provided in scattered dot plots on the right of the blots. Data are mean ± SD; ***P < 0.001. HSP90 is used as a loading control. (n = 5 mice per group). (H) Characterization of primary tumors innervation in the 4T1 orthotopic model of tumor progression. Co-immunofluorescence analysis of neuron-specific marker Tubulin β3 (Tubβ3 – green) with sympathetic-specific marker tyrosine hydroxylase (TH), sensory-specific marker capsaicin receptor (TRPV1), or parasympathetic marker choline acetyltransferase (ChAT). For each marker, a representative nerve twig or a more organized fiber is displayed. 120× magnification. (I) Cross-correlation between EMT and tumor innervation in human breast cancer samples. (Top) Analysis of human breast cancer primary tumor axonogenesis (TUBB3 gene expression) and EMT-related mesenchymal (TGFβ1, FN1, CDH2, and SNAI1) and epithelial (TJP1 and CTNNB1) genes expression in 1,218 human breast tumor samples from the Cancer Genome Atlas BRCA RNA-sequencing database. Tumor innervation is ranked through tubulin-β3 expression analysis. Red color indicates tumors with the highest innervation. (J) Correlation analysis between primary tumor axonogenesis and EMT related factors. Genes expression is expressed in Log2 (normalized count + 1) and Pearson’s rho (r=) and P-value (p=) is provided for each correlation. HSP90 serves as a loading control. (K) Kaplan–Meier survival curve comparing breast cancer innervation to overall survival. Data obtained from the KM plotter breast dataset in overall breast cancer samples (n = 4,929). HSP90 serves as a loading control. Experiments have been repeated three times or as specified in the legend. Scale bars: 50 μm.
