The breast cancer oncogene EMSY represses transcription of antimetastatic microRNA miR-31

Abstract

Amplification of the EMSY gene in sporadic breast and ovarian cancers is a poor prognostic indicator. Although EMSY has been linked to transcriptional silencing, its mechanism of action is unknown. Here, we report that EMSY acts as an oncogene, causing the transformation of cells in vitro and potentiating tumor formation and metastatic features in vivo. We identify an inverse correlation between EMSY amplification and miR-31 expression, an antimetastatic microRNA, in the METABRIC cohort of human breast samples. Re-expression of miR-31 profoundly reduced cell migration, invasion, and colony-formation abilities of cells overexpressing EMSY or haboring EMSY amplification. We show that EMSY is recruited to the miR-31 promoter by the DNA binding factor ETS-1, and it represses miR-31 transcription by delivering the H3K4me3 demethylase JARID1b/PLU-1/KDM5B. Altogether, these results suggest a pathway underlying the role of EMSY in breast cancer and uncover potential diagnostic and therapeutic targets in sporadic breast cancer.

Graphical abstract

Figure 1

EMSY Overexpression Results in Oncogenic Transformation (A) EMSY increases anchorage-independent growth in soft-agar assays. NIH 3T3 cells stably transfected with human EMSY gene or an empty vector were grown in soft agar for 14 days. Error bars represent SD of the mean. Top, representative pictures of the colonies. Right, representative western blot analysis of EMSY expression. Each value is the average of three independent experiments, and error bars display the SD of the mean. (B) Soft-agar assay for MCF-7 cells stably overexpressing EMSY or the empty vector. Results are presented as in (A). Each value is the average of three independent experiments, and error bars display the SD of the mean. (C) MCF-7-EMSY cells were injected into the mammary fat pad of 5-week-old female athymic nu/nu mice (n = 10). Mice were sacrificed 34 days after injection, and then tumors were excised and weighed. Quantifications of tumor volume (left) and weight (middle) are shown. Right, representative photographs of the tumors. Error bars display the SD of the mean. (D) MCF-7-EMSY cells induce micrometastasis in the lungs after injection into the tail vein of 5-week-old female athymic nu/nu mice (n = 10). Mice were sacrificed 62 days after injection. Left, metastasis was analyzed macroscopically and by H&E tissue staining. Representative photographs of the micrometastasis are shown, and arrows indicate the micrometastasis. Right, quantifications of the micrometastasis obtained. Error bars display the SD of the mean. See also Figure S1.

Figure 2

Expression of EMSY Anticorrelates with miR-31 Levels In Vitro and In Vivo (A) Box plots illustrating the distribution of EMSY copy-number-segmented means across the integrative subgroups (left) and PAM50 intrinsic subtypes (right) for 1,283 tumors from the METABRIC cohort (see also Table S1). (B) Box plot of miR-31 expression for cases with EMSY amplification (AMP; n = 45, red boxes) versus neutral copy numbers (NEUT; n = 809, blue boxes). The p value was determined by a Wilcoxon rank-sum test. (C) EMSY represses miR-31 transcription. qRT-PCR analysis of mature forms of the indicated miRNAs in MCF-7 cells stably overexpressing EMSY or control. Error bars represent SD of the mean. (D) Left, relative expression levels of miR-31, miR-198, and miR-181-a2 determined by qRT-PCR in MCF-7 cells transfected with siRNAs targeting human EMSY. Error bars represent SD of the mean. Right, representative western blot analysis of EMSY expression in MCF7 cells transfected with siRNAs targeting human EMSY. (E) Relative expression of the miR-31-regulated prometastatic genes upon EMSY overexpression in MCF-7 cells. Transcripts levels were measured by qRT-PCR. Error bars represent SD of the mean. See also Figure S2.

Figure 3

miR-31 Reverts EMSY Oncogenicity In Vitro (A) Invasion and migration assay of MCF-7 cells stably expressing EMSY. Reintroduction of miR-31 reduces EMSY-mediated effects. Error bars represent SD of the mean. (B) Invasion and migration assay of MCF-7 cells transiently depleted from miR-31 using oligonucleotides. Bottom, relative expression levels of miR-31 measured by qRT-PCR. (C) Stable depletion of miR-31 reduces anchorage-independent growth of MCF-7 cells. Error bars represent SD of the mean. (D) Soft-agar assay of MCF-7 cells stably overexpressing EMSY after the reintroduction of miR-31. Results are presented as in (C). (E) The expression levels of the indicated miRNAs were measured by qRT-PCR in MDA-MB-175, harouring EMSY amplification, and stably depleted from EMSY. Error bars represent SD of the mean. (F) Soft-agar assay of MDA-MB-175 cells stably depleted from EMSY. (G) Exogenous expression of miR-31 reduces anchorage-independent growth of of MDA-MB-175. Results are presented as in (C). (H) Invasion and migration assay of MDA-MB-175 cells either stably depleted from EMSY or stably overexpressing miR-31. Error bars represent SD of the mean. ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. See also Figure S3.

Figure 4

EMSY Represses miR-31 Transcription and Associates to miR-31 Promoter (A) pri-miRNAs level determined by qRT-PCR in MCF-7 cells transiently depleted from EMSY. Error bars display SD of the mean. (B) EMSY associates to miR-31 promoter. ChIP in MCF-7 cells treated with control siRNA or EMSY siRNA using anti-EMSY or anti-GFP (control). The data have been normalized to “input DNA” amplification levels for each amplicon. Each amplicon was analyzed in duplicate. Experiments were repeated at least two times with similar results. A representative example of a ChIP analysis is shown. Error bars display the SD of the mean. (C) EMSY depletion leads to increased levels of RNA polymerase II at miR-31 promoter. ChIP with anti-RNA polymerase II in MCF-7 cells treated with control siRNA or EMSY siRNA. Results are presented as in (B). (D) EMSY and RNA polymerase II occupancy at the miR-31 promoter were assessed in MCF-7, MDA-MB-175, and MDA-MB-415 cells. Results are presented as in (B). See also Figure S4.

Figure 5

EMSY Is Recruited to the miR-31 Promoter via ETS-1 Transcription Factor and Interacts with H3K4me3 Histone Demethylase KDM5B (A) Schematic view of the indicated miRNA promoter region (1,000 bp upstream from the putative transcription start site; arrow). The locations of predicted putative TF binding sites are shown for ETS-1, ETV4 (PEA3), and GATA1. Numbering is relative to the transcription start site. (B) Expression of the miR-31, miR-198, and miR-181a-2 genes were studied by qRT-PCR after cells were treated with control, ETS-1, ETV4 (PEA3), or GATA1 siRNAs. Error bars represent SD of the mean. (C) MCF-7 cells were transiently transfected with the cis-regulatory element of miR-31 promoter fused to the luciferase reporter vector pGL4, as indicated in the scheme (left). The luciferase activity of empty pGL4 was used as the control. Luciferase activities, expressed as relative luciferase units, were normalized to the level of Renilla luciferase activity and then to control RLU (empty vector). Site-directed mutagenesis was used to change the sequence of each or both of the ETS-1 binding sites. Error bars represent SD of the mean. (D) Chromatin from MCF-7 cells depleted from ETS-1 was immunoprecipitated with the indicated antibodies. Results are presented as in Figure 4B. (E) Endogenous coimmunoprecipitation assays for detecting the interaction between EMSY, ETS-1, and KDM5B in MCF-7 cells. Whole-cell extracts were immunoprecipitated with anti-ETS-1, anti-KDM5B, anti-SUZ12, and anti-EMSY or preimmune GFP used as control IgG. The presence of EMSY, ETS-1, and KDM5B in immunoprecipitates was detected by western blotting. (F) KDM5B depletion affects H3K4me3 levels on miR-31 promoter. ChIP in control and KDM5B-siRNA-treated MCF-7 cells. (G) The expression levels of mature forms of the indicated miRNAs were measured by qRT-PCR analysis in control and KDM5B-siRNA-treated MCF-7 cells. (H) Migration assay of MCF-7 cells overexpressing EMSY or an empty vector after cells were treated with siRNA against ETS-1, KDM5B, or a combination of both. Error bars represent SD of the mean. ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. See also Figure S5.