Limits...
CELF1 is a central node in post-transcriptional regulatory programmes underlying EMT

View Article: PubMed Central - PubMed

ABSTRACT

The importance of translational regulation in tumour biology is increasingly appreciated. Here, we leverage polyribosomal profiling to prospectively define translational regulatory programs underlying epithelial-to-mesenchymal transition (EMT) in breast epithelial cells. We identify a group of ten translationally regulated drivers of EMT sharing a common GU-rich cis-element within the 3′-untranslated region (3′-UTR) of their mRNA. These cis-elements, necessary for the regulatory activity imparted by these 3′-UTRs, are directly bound by the CELF1 protein, which itself is regulated post-translationally during the EMT program. CELF1 is necessary and sufficient for both mesenchymal transition and metastatic colonization, and CELF1 protein, but not mRNA, is significantly overexpressed in human breast cancer tissues. Our data present an 11-component genetic pathway, invisible to transcriptional profiling approaches, in which the CELF1 protein functions as a central node controlling translational activation of genes driving EMT and ultimately tumour progression.

No MeSH data available.


Related in: MedlinePlus

Enhanced CELF1 expression in human breast cancer correlates with increased metastasis.(a) Heat map depicting log2-fold difference in mRNA expression of indicated genes between tumour-normal matched pairs for 111 TCGA patients. Topmost coloured bar indicates PAM50 assignments. Columns are ordered by PAM50 subtype. Top: heat map shows exemplary set of genes correlated with subtype, and bottom: heat map depicts genes translationally regulated by CELF1 during EMT and CELF1. For each heat map, rows are ordered by decreasing variance from top to bottom. Columns are ordered by PAM50 assignment and within PAM50 assignment by increasing mean log2-fold change of genes from left to right. (b) qRT-PCR-based determination of change in steady-state and polyribosomal-bound CELF1 mRNA using total (T) and polyribosomal (P) mRNA obtained from untreated and TGF-β-treated MCF10A cells. Data are represented as mean±s.e.m. All panels are representative of a minimum of three experimental replicates. (c) CELF1 is hyperphosphorylated in mesenchymal MCF10A cells. Immunoprecipitates obtained using anti-CELF1 antibody or mouse IgG from MG-132 treated (8 h) MCF10A±TGF-β (72 h) cells were probed with indicated antibodies. (d) Inhibition of proteasome-mediated degradation results in increased CELF1 protein levels in epithelial MCF10A cells. Cells were incubated with MG-132 for the indicated times and immunoblotted for the indicated proteins. HSP90 serves as a loading control. (e) Representative image of CELF1 protein expression in mammary carcinoma and tumour adjacent normal mammary gland tissue, as analysed by IHC staining. Image was obtained with × 20 objective. Brown staining indicates antigen, while blue staining represents counterstain. Scale bar, 100 μm. (f) Relative CELF1 protein expression as assessed by the per cent score in the 29 matched breast tumour and adjacent normal tissue (P<0.0001; Wilcoxon's signed-rank test). (g) Intracellular CELF1 staining was scored in breast tumour samples from patients with tumour size <20 mm and no lymph node involvement (T1,N0) (n=10), and with lymph node involvement along with tumour size between 20–50 mm (T2,N>0) (n=53) and >50 mm (T3,N>0) (n=13), and the P-value of the trend was determined using the Jonckheere–Terpstra test. Box plots represent the 25th to 75th quartiles with the bold horizontal line representing the median value. (h) Trend between CELF1 expression and tumour grade I (n=12), grade II–III (n=97) and grade III (n=7). The P-value of the trend was determined using the Jonckheere–Terpstra test. Box plots represent the 25th to 75th quartiles with the bold horizontal line representing the median value. See also Supplementary Fig. 6. Full scans of blots are shown in Supplementary Fig. 12.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5121338&req=5

f7: Enhanced CELF1 expression in human breast cancer correlates with increased metastasis.(a) Heat map depicting log2-fold difference in mRNA expression of indicated genes between tumour-normal matched pairs for 111 TCGA patients. Topmost coloured bar indicates PAM50 assignments. Columns are ordered by PAM50 subtype. Top: heat map shows exemplary set of genes correlated with subtype, and bottom: heat map depicts genes translationally regulated by CELF1 during EMT and CELF1. For each heat map, rows are ordered by decreasing variance from top to bottom. Columns are ordered by PAM50 assignment and within PAM50 assignment by increasing mean log2-fold change of genes from left to right. (b) qRT-PCR-based determination of change in steady-state and polyribosomal-bound CELF1 mRNA using total (T) and polyribosomal (P) mRNA obtained from untreated and TGF-β-treated MCF10A cells. Data are represented as mean±s.e.m. All panels are representative of a minimum of three experimental replicates. (c) CELF1 is hyperphosphorylated in mesenchymal MCF10A cells. Immunoprecipitates obtained using anti-CELF1 antibody or mouse IgG from MG-132 treated (8 h) MCF10A±TGF-β (72 h) cells were probed with indicated antibodies. (d) Inhibition of proteasome-mediated degradation results in increased CELF1 protein levels in epithelial MCF10A cells. Cells were incubated with MG-132 for the indicated times and immunoblotted for the indicated proteins. HSP90 serves as a loading control. (e) Representative image of CELF1 protein expression in mammary carcinoma and tumour adjacent normal mammary gland tissue, as analysed by IHC staining. Image was obtained with × 20 objective. Brown staining indicates antigen, while blue staining represents counterstain. Scale bar, 100 μm. (f) Relative CELF1 protein expression as assessed by the per cent score in the 29 matched breast tumour and adjacent normal tissue (P<0.0001; Wilcoxon's signed-rank test). (g) Intracellular CELF1 staining was scored in breast tumour samples from patients with tumour size <20 mm and no lymph node involvement (T1,N0) (n=10), and with lymph node involvement along with tumour size between 20–50 mm (T2,N>0) (n=53) and >50 mm (T3,N>0) (n=13), and the P-value of the trend was determined using the Jonckheere–Terpstra test. Box plots represent the 25th to 75th quartiles with the bold horizontal line representing the median value. (h) Trend between CELF1 expression and tumour grade I (n=12), grade II–III (n=97) and grade III (n=7). The P-value of the trend was determined using the Jonckheere–Terpstra test. Box plots represent the 25th to 75th quartiles with the bold horizontal line representing the median value. See also Supplementary Fig. 6. Full scans of blots are shown in Supplementary Fig. 12.

Mentions: Our analysis revealed similar or decreased levels of GRE-containing mRNA transcripts in several defined molecular subtypes of human breast cancer as compared with normal controls (Fig. 7a). Strikingly, there were essentially no changes in the expression of CELF1 in this comparison. We thus returned to our primary MCF10A model, finding that the observed increase in relative CELF1 protein expression associated with the mesenchymal state (Fig. 4a) occurred independently of significant changes in total CELF1 mRNA expression or ribosomal occupancy (Fig. 7b). However, blocking proteasomal degradation via treatment with MG-132 in epithelial MCF10A cells resulted in marked increases of CELF1 protein in these cells (Fig. 7c). In other systems, the stability and activity of CELF1 protein has been shown to be impacted by phosphorylation26. Consistent with these observations, CELF1 protein was characterized by markedly increased phosphorylation of serine and threonine (but not tyrosine) residues in mesenchymal MCF10A cells (Fig. 7d). Under the rationale that CELF1 gene expression might be similarly regulated in the context of human breast cancer, we next asked whether dysregulation of CELF1 protein expression in breast cancer might be visualized via immunohistochemistry.


CELF1 is a central node in post-transcriptional regulatory programmes underlying EMT
Enhanced CELF1 expression in human breast cancer correlates with increased metastasis.(a) Heat map depicting log2-fold difference in mRNA expression of indicated genes between tumour-normal matched pairs for 111 TCGA patients. Topmost coloured bar indicates PAM50 assignments. Columns are ordered by PAM50 subtype. Top: heat map shows exemplary set of genes correlated with subtype, and bottom: heat map depicts genes translationally regulated by CELF1 during EMT and CELF1. For each heat map, rows are ordered by decreasing variance from top to bottom. Columns are ordered by PAM50 assignment and within PAM50 assignment by increasing mean log2-fold change of genes from left to right. (b) qRT-PCR-based determination of change in steady-state and polyribosomal-bound CELF1 mRNA using total (T) and polyribosomal (P) mRNA obtained from untreated and TGF-β-treated MCF10A cells. Data are represented as mean±s.e.m. All panels are representative of a minimum of three experimental replicates. (c) CELF1 is hyperphosphorylated in mesenchymal MCF10A cells. Immunoprecipitates obtained using anti-CELF1 antibody or mouse IgG from MG-132 treated (8 h) MCF10A±TGF-β (72 h) cells were probed with indicated antibodies. (d) Inhibition of proteasome-mediated degradation results in increased CELF1 protein levels in epithelial MCF10A cells. Cells were incubated with MG-132 for the indicated times and immunoblotted for the indicated proteins. HSP90 serves as a loading control. (e) Representative image of CELF1 protein expression in mammary carcinoma and tumour adjacent normal mammary gland tissue, as analysed by IHC staining. Image was obtained with × 20 objective. Brown staining indicates antigen, while blue staining represents counterstain. Scale bar, 100 μm. (f) Relative CELF1 protein expression as assessed by the per cent score in the 29 matched breast tumour and adjacent normal tissue (P<0.0001; Wilcoxon's signed-rank test). (g) Intracellular CELF1 staining was scored in breast tumour samples from patients with tumour size <20 mm and no lymph node involvement (T1,N0) (n=10), and with lymph node involvement along with tumour size between 20–50 mm (T2,N>0) (n=53) and >50 mm (T3,N>0) (n=13), and the P-value of the trend was determined using the Jonckheere–Terpstra test. Box plots represent the 25th to 75th quartiles with the bold horizontal line representing the median value. (h) Trend between CELF1 expression and tumour grade I (n=12), grade II–III (n=97) and grade III (n=7). The P-value of the trend was determined using the Jonckheere–Terpstra test. Box plots represent the 25th to 75th quartiles with the bold horizontal line representing the median value. See also Supplementary Fig. 6. Full scans of blots are shown in Supplementary Fig. 12.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5121338&req=5

f7: Enhanced CELF1 expression in human breast cancer correlates with increased metastasis.(a) Heat map depicting log2-fold difference in mRNA expression of indicated genes between tumour-normal matched pairs for 111 TCGA patients. Topmost coloured bar indicates PAM50 assignments. Columns are ordered by PAM50 subtype. Top: heat map shows exemplary set of genes correlated with subtype, and bottom: heat map depicts genes translationally regulated by CELF1 during EMT and CELF1. For each heat map, rows are ordered by decreasing variance from top to bottom. Columns are ordered by PAM50 assignment and within PAM50 assignment by increasing mean log2-fold change of genes from left to right. (b) qRT-PCR-based determination of change in steady-state and polyribosomal-bound CELF1 mRNA using total (T) and polyribosomal (P) mRNA obtained from untreated and TGF-β-treated MCF10A cells. Data are represented as mean±s.e.m. All panels are representative of a minimum of three experimental replicates. (c) CELF1 is hyperphosphorylated in mesenchymal MCF10A cells. Immunoprecipitates obtained using anti-CELF1 antibody or mouse IgG from MG-132 treated (8 h) MCF10A±TGF-β (72 h) cells were probed with indicated antibodies. (d) Inhibition of proteasome-mediated degradation results in increased CELF1 protein levels in epithelial MCF10A cells. Cells were incubated with MG-132 for the indicated times and immunoblotted for the indicated proteins. HSP90 serves as a loading control. (e) Representative image of CELF1 protein expression in mammary carcinoma and tumour adjacent normal mammary gland tissue, as analysed by IHC staining. Image was obtained with × 20 objective. Brown staining indicates antigen, while blue staining represents counterstain. Scale bar, 100 μm. (f) Relative CELF1 protein expression as assessed by the per cent score in the 29 matched breast tumour and adjacent normal tissue (P<0.0001; Wilcoxon's signed-rank test). (g) Intracellular CELF1 staining was scored in breast tumour samples from patients with tumour size <20 mm and no lymph node involvement (T1,N0) (n=10), and with lymph node involvement along with tumour size between 20–50 mm (T2,N>0) (n=53) and >50 mm (T3,N>0) (n=13), and the P-value of the trend was determined using the Jonckheere–Terpstra test. Box plots represent the 25th to 75th quartiles with the bold horizontal line representing the median value. (h) Trend between CELF1 expression and tumour grade I (n=12), grade II–III (n=97) and grade III (n=7). The P-value of the trend was determined using the Jonckheere–Terpstra test. Box plots represent the 25th to 75th quartiles with the bold horizontal line representing the median value. See also Supplementary Fig. 6. Full scans of blots are shown in Supplementary Fig. 12.
Mentions: Our analysis revealed similar or decreased levels of GRE-containing mRNA transcripts in several defined molecular subtypes of human breast cancer as compared with normal controls (Fig. 7a). Strikingly, there were essentially no changes in the expression of CELF1 in this comparison. We thus returned to our primary MCF10A model, finding that the observed increase in relative CELF1 protein expression associated with the mesenchymal state (Fig. 4a) occurred independently of significant changes in total CELF1 mRNA expression or ribosomal occupancy (Fig. 7b). However, blocking proteasomal degradation via treatment with MG-132 in epithelial MCF10A cells resulted in marked increases of CELF1 protein in these cells (Fig. 7c). In other systems, the stability and activity of CELF1 protein has been shown to be impacted by phosphorylation26. Consistent with these observations, CELF1 protein was characterized by markedly increased phosphorylation of serine and threonine (but not tyrosine) residues in mesenchymal MCF10A cells (Fig. 7d). Under the rationale that CELF1 gene expression might be similarly regulated in the context of human breast cancer, we next asked whether dysregulation of CELF1 protein expression in breast cancer might be visualized via immunohistochemistry.

View Article: PubMed Central - PubMed

ABSTRACT

The importance of translational regulation in tumour biology is increasingly appreciated. Here, we leverage polyribosomal profiling to prospectively define translational regulatory programs underlying epithelial-to-mesenchymal transition (EMT) in breast epithelial cells. We identify a group of ten translationally regulated drivers of EMT sharing a common GU-rich cis-element within the 3&prime;-untranslated region (3&prime;-UTR) of their mRNA. These cis-elements, necessary for the regulatory activity imparted by these 3&prime;-UTRs, are directly bound by the CELF1 protein, which itself is regulated post-translationally during the EMT program. CELF1 is necessary and sufficient for both mesenchymal transition and metastatic colonization, and CELF1 protein, but not mRNA, is significantly overexpressed in human breast cancer tissues. Our data present an 11-component genetic pathway, invisible to transcriptional profiling approaches, in which the CELF1 protein functions as a central node controlling translational activation of genes driving EMT and ultimately tumour progression.

No MeSH data available.


Related in: MedlinePlus