EZH2 is a mediator of EWS/FLI1 driven tumor growth and metastasis blocking endothelial and neuro-ectodermal differentiation

Abstract

Ewing tumors (ET) are highly malignant, localized in bone or soft tissue, and are molecularly defined by ews/ets translocations. DNA microarray analysis revealed a relationship of ET to both endothelium and fetal neural crest. We identified expression of histone methyltransferase enhancer of Zeste, Drosophila, Homolog 2 (EZH2) to be increased in ET. Suppressive activity of EZH2 maintains stemness in normal and malignant cells. Here, we found EWS/FLI1 bound to the EZH2 promoter in vivo, and induced EZH2 expression in ET and mesenchymal stem cells. Down-regulation of EZH2 by RNA interference in ET suppressed oncogenic transformation by inhibiting clonogenicity in vitro. Similarly, tumor development and metastasis was suppressed in immunodeficient Rag2(-/-)gamma(C)(-/-) mice. EZH2-mediated gene silencing was shown to be dependent on histone deacetylase (HDAC) activity. Subsequent microarray analysis of EZH2 knock down, HDAC-inhibitor treatment and confirmation in independent assays revealed an undifferentiated phenotype maintained by EZH2 in ET. EZH2 regulated stemness genes such as nerve growth factor receptor (NGFR), as well as genes involved in neuroectodermal and endothelial differentiation (EMP1, EPHB2, GFAP, and GAP43). These data suggest that EZH2 might have a central role in ET pathology by shaping the oncogenicity and stem cell phenotype of this tumor.

Fig. 1.

EZH2 expression and regulation by EWS/FLI1. (A) Transient transfection of EWS/FLI1 specific siRNA down-regulates EZH2 expression. EWS/FLI1 #I and EWS/FLI1 #II represent different siRNAs (see Materials and Methods; control siRNA: non silencing siRNA). Results of real-time RT-PCR after 85 h of transfection are shown. NTC, no template control. (B) Retroviral gene transfer of EWS/FLI1 cDNA (see Materials and Methods) into MSC lines L87 and V54.2 results in a strong EWS/FLI1 expression as shown by Western blot analysis with FLI1 antibody. Numbers for different obtained lines are given; vc, vector control. HPRT was used as loading control. (C) EZH2 real-time RT-PCR of selected stem cell clones as identified by Western blot analysis in B reveals an EWS/FLI1-dependent EZH2 induction. (D) EWS/FLI1 binds to the EZH2 promoter in vivo. ChIPs were performed in A673 cells with the indicated antibodies, and analyzed by PCR and real-time PCR for binding to different regions of the EZH2 promoter. The gel picture indicates specific enrichment of anti-Fli-1 ChIPs to a region starting at −1081 bp upstream of the transcriptional start site (TSS). Different regions were also analyzed by quantitative real-time PCR, and normalized for IgG and nonspecific binding to an unrelated genomic region. The bars indicate mean and SE of 4 independent ChIPs. FLI1 is enriched at an evolutionary conserved consensus site at −1081 bp upstream of the TSS. The −4,400-bp region is devoid of ETS recognition sequences, and served as another negative control.

Fig. 2.

Effects of EZH2 down-regulation on in vitro and in vivo tumor growth and metastasis. (A) Stable EZH2 shRNA infectants, generated by retroviral gene transfer of pSIREN-shEZH2 constructs (see Materials and Methods; pSIREN^n.siRNA: non silencing siRNA) into A673 cells were analyzed for their siRNA mediated down-regulation of EZH2 expression by real-time RT-PCR. Error bars represent SD. Inset: Western blot analysis of the same cells as analyzed in real-time RT-PCR. Anti-Histone H3 (H3) was used as loading control. (B) Stable A673-EZH2shRNA infectants pSIREN^EZH2-1 and pSIREN^EZH2-2 lost their ability for colony formation in methylcellulose assay. Results are shown in duplicate. (C) Immunodeficient Rag2^−/−γ_C^−/− mice were injected s.c. intrainguinal with parental A673 cells and shRNA infectants (pSIREN^n.siRNA, pSIREN^EZH2-1, and pSIREN^EZH2-2). Mice with an average tumor size >10 mm in diameter were considered as positive and killed. Combined Kaplan–Meyer plot of 2 individual experiments with altogether 9–10 mice per group are shown. (D) To analyze metastatic growth, A673 cells and EZH2 siRNA infectants were injected into the tail vein of immunodeficient mice. Four weeks later, mice were killed and metastatic spread was analyzed. (Top) Affected lungs are shown. (Middle and Bottom) Histology of affected lungs. (Middle) Low magnification of lung sections stained with Hematoxylin and Eosin (magnification 12×). Note the difference in the number of lung metastases. A673pSIREN^EZH2-1 cells induced no lung metastases, whereas A673pSIREN^EZH2-2 cells caused less and smaller metastases in the lung when compared with the A673pSIREN^n.siRNA control cells. The metastases observed in the animals treated with A673pSIREN^n.siRNA replace almost the entire parenchyma of the lung. (Bottom) Higher magnification of the tumor cells (magnification 400×). The metastases are composed of cells with abundant cytoplasm, irregular nuclei, and prominent nucleoli. Foci of necrosis and mitosis are frequently observed. Lungs of control mice are of the same age as those injected with EZH2 shRNA infectants.

Fig. 3.

EZH2 blockade in ET induces a number of genes important for epithelial and neuroectodermal differentiation. (A) Microarray data of selected genes after SAM analysis with their normalized fluorescent signal intensities (see Materials and Methods). Combined results of 2 independent experiments with RNA derived from cells after DMSO/TSA treatment or transient transfection with different siRNAs are shown. Cells were treated with DMSO/TSA for 24 h, collected, and then analyzed; siRNA transfectants were harvested after 48 h. Each column represents 1 individual array. The first 100 most significant genes of the analysis are shown. (B) Real-time RT-PCR of different neuroectodermal and endothelial genes 60 h after transient siRNA transfection of A673 and MHHES1 cells, respectively. Down-regulation of EZH2, EED, and SUZ12, (“siRNA target expression” graph) similarly regulate the expression of identified genes EPHB2, EMP1, GAP43, and NGFR alluding to their common regulation by PRC2 complex. Similar results were obtained for ALCAM and GFAP (Table S3). Potential contribution of noncoding RNAs to an AGO1-mediated PRC2 regulation of target genes was only observed for NGFR.

Fig. 4.

Blockade of genes of PRC2 complex in ET induces a number of genes important for and enables epithelial and neuroectodermal differentiation. (A) Neurogenic differentiation of stable A673-infectants pSIREN^EZH2-1 and pSIREN^n.siRNA treated for 5 days with 0.1 mM BHA was analyzed. Phase contrast microscopy (magnification 20×), showing neuron-like morphology after neurogenic induction. Immunofluorescent staining of GFAP shown at 20× magnification. (B) To analyze endothelial differentiation potential of ET cells, transiently siRNA transfected A673, MHHES1, and constitutively infected A673 cells were analyzed for tube formation potential by using matrigel assay. Cells were grown in 3D culture on collagen type I matrix. First 2 rows (A673 transfectants) and right column (MHHES1 transfectants) cells after calcein staining, images at 4× magnification. The bottom 3 images from left, phase contrast microscopy at 10× magnification; neg.siRNA, non silencing siRNA. Tube formation was not only enabled by EZH2 down-regulation, but also by inhibition of other components of PRC2 complex such as EED and SUZ12.