Decreased miR-200b-3p in cancer cells leads to angiogenesis in HCC by enhancing endothelial ERG expression

Transcription factor ERG (erythroblast transformation-specific (ETS)-related gene) is essential in endothelial differentiation and angiogenesis, in which microRNA (miR)-200b-3p targeting site is expected by miRNA target prediction database. miR-200b is known decreased in hepatocellular carcinoma (HCC), however, the functional relation between ERG and miR-200b-3p, originating from pre-miR-200b, in HCC angiogenesis remains unclear. We investigated whether hepatocyte-derived miR-200b-3p governs angiogenesis in HCC by targeting endothelial ERG. Levels of miR-200b-3p in HCC tissues were significantly lower than those in adjacent non-HCC tissues. Poorly differentiated HCC cell line expressed lower level of miR-200b-3p compared to well-differentiated HCC cell lines. The numbers of ERG-positive endothelial cells were higher in HCC tissues than in adjacent non-HCC tissues. There was a negative correlation between the number of ERG-positive cells and miR-200b-3p expression in HCC tissues. Culture supernatants of HCC cell lines with miR-200b-3p-overexpression reduced cell migration, proliferation and tube forming capacity in endothelial cells relative to the control, while those with miR-200b-3p-inhibition augmented the responses. Exosomes isolated from HCC culture supernatants with miR-200b-3p overexpression suppressed endothelial ERG expression. These results suggest that exosomal miR-200b-3p from hepatocytes suppresses endothelial ERG expression, and decreased miR-200b-3p in cancer cells promotes angiogenesis in HCC tissues by enhancing endothelial ERG expression.

www.nature.com/scientificreports/ target prediction database. Recently, miRNAs are known to target ERG (erythroblast transformation-specific (ETS)-related gene) in prostate cancer and colorectal cancer 12,13 . ERG plays an essential role in endothelial homeostasis, differentiation, and angiogenesis 14,15 , in which miR-200b-3p targeting site is expected by miRNA target prediction database. We hypothesized that down-regulated miR-200b-3p in HCC may cause enhanced endothelial ERG expression, leading to increased angiogenesis in the cancer microenvironment.
In this study, we, for the first time, demonstrate that ERG is a target of miR200b-3p and hepatocyte-derived miR-200b-3p reduces the endothelial cell migration, proliferation and tube forming capacity in endothelial cells. Additionally, we demonstrate that HCC tissues exhibit reduced miR-200b-3p expression, which causes augmented endothelial ERG expression, promoting angiogenesis in the cancer microenvironment. Furthermore, miR-200b-3p appears to be transferred by exosomes released from hepatocytes. Thus, miR-200b-3p can be a novel therapeutic target for the regulation of cancer angiogenesis.

Results
Expression of miR-200b-3p in HCC tissues and cell lines. The expression levels of miR-200b-3p were analyzed in forty pairs of clinical HCC and adjacent non-cancer tissues by qRT-PCR. The cases for the enrolled 40 patients with HCC are shown in Table 1. As shown in Fig. 1a, the miR-200b-3p expression levels in the HCC tissues were significantly lower than those in adjacent non-cancer tissues. The miR-200b-3p expression levels tended to be lower with decreasing grade of cancer differentiation although it was not statistically significant (Fig. 1b). Further detailed clinical data from enrolled patients with HCC were shown in Table 2. Among HCC with trabecular pattern, miR-200b-3p expression levels in moderately plus poorly differentiated HCC tended to be lower than those in well differentiated HCC (Table 2). Next, we examined the expression levels of miR-200b-3p in three HCC cell lines. The expression levels of miR-200b-3p in poorly differentiated cell line (HLE cell) were significantly lower than those in two well-differentiated HCC cell lines (Hep3B and HepG2 cells) (Fig. 1c). In contrast, the expression levels of miR-200b-3p in HUVECs was extremely low when compared to those in three HCC cell lines (Fig. 1c). These data indicate that non-cancer tissues, likely non-cancer hepatocytes, exhibit high levels of miR-200b-3p expression, whereas HCC tissues, likely cancer cells, exhibit decreased expression depending on the degree of cancer differentiation.

Expression of ERG in HCC tissues.
Expression of ERG in HCC tissues was next analyzed. As shown in Fig. 2a, nuclei of endothelial cells in tissue samples were stained positive for ERG. The numbers of ERG-positive endothelial cells in cancer tissue were more than those in adjacent non-cancer tissue (Fig. 2b). As shown in Fig. 2c, the statistical analysis indicated a negative correlation between the number of ERG-positive endothelial cells and the expression of miR-200b-3p in the HCC tissues (r = − 0.4836, p < 0.0001), suggesting that decreased miR-200b-3p expression in cancer tissue causes vascular hyperplasia in HCC tissues. miR-200b-3p negatively regulates endothelial ERG expression. ERG is a transcription factor, and plays a crucial rule in vascular development, angiogenesis and vascular stability [14][15][16] . We assumed that miR-200b-3p may affect endothelial ERG expression. To investigate this assumption, miR-200b-3p was overexpressed in HUVECs, after which the cells were cultured with complete medium containing VEGF, and the ERG expression was examined by western blotting. As shown in Fig. 3a, levels of ERG in HUVECs overexpressing miR-200b-3p were significantly lower than those in controls. Although HUVECs exhibited much lower miR-200b-3p expression levels compared to HCC cell lines (Fig. 1c), inhibition of miR-200b-3p in HUVECs increased the ERG expression when compared to controls (Fig. 3b). These data show that miR-200b-3p partly downregulates endothelial ERG expression.
ERG is a direct target of miR-200b-3p. Analyzing the microRNA target prediction database and tools, such as TargetScanHuman (https ://www.targe tscan .org/vert_72/), microRNA.org (https ://www.micro rna.org/ micro rna/home.do) and miRDB (https ://mirdb .org) revealed that the seed region of miR-200b-3p and 3′-UTR of ERG mRNA (position 624-631) are complementary (Fig. 4a). We generated an oligonucleotide of 23 base pairs containing miR-200b-3p binding sequence of ERG 3′-UTR and cloned into pmirGLO Dual-Luciferase miRNA Target Expression Vector and analyzed if ERG is a direct target of miR-200b-3p by using the dual-luciferase reporter assay. A similar construct lacking the target sequence of 3′-UTR of ERG was used as mutant construct. As shown in Fig. 4b, co-expression of precursor miRNA-200b-3p and wild type ERG 3′-UTR construct in HEK239T cells significantly reduced luciferase activity. There was no significant difference in luciferase activities when the ERG 3′-UTR sequence was mutated (p = 0.1686) (Fig. 4b). These data indicate that miR-200b-3p can directly target the 3′-UTR of the ERG gene and suppress the ERG expression.
Hepatic miR-200b-3p suppresses endothelial ERG expression. To examine the relation between hepatic miR-200b-3p and endothelial ERG, HUVECs were cultured with two types of HCC culture supernatants. For gain-of-function analysis, we overexpressed miR-200b-3p in HLE cells (miR-200b-3p expression; low, Fig. 1c) and harvested culture supernatants at 48 h (Fig. 5a). For loss-of-function experiments, we treated Hep3B cells (miR-200b-3p expression; high, Fig. 1c) with a specific inhibitor or inhibitor control and harvested culture supernatants (Fig. 5b). The culture supernatants were mixed with complete medium at a ratio of 1:1, and the HUVECs were cultured. As shown in Fig. 5c, the protein expression levels of ERG in HUVECs cultured with miR-200b-3p overexpression supernatants were lower than those with the control. By contrast, the protein expression levels of ERG in HUVECs treated with culture supernatants of miR-200b-3p-inhibition were higher than those with control (Fig. 5d). This suggests that hepatocyte-derived miR-200b-3p downregulates ERG expression in HUVECs. www.nature.com/scientificreports/ Hepatic miR-200b-3p inhibits angiogenesis. Angiogenesis requires not only endothelial cell proliferation but also their migration and formation of tubes. Scratch wound healing assay was performed using HUVECs to examine the role of miR-200b-3p in endothelial cell migration. Culture supernatants with miR-200b-3p overexpression inhibited the closure of HUVECs when compared to control (Fig. 6a, upper). In contrast, culture supernatants with miR-200b-3p-inhibition accelerated the wound closure (Fig. 6a, lower). miR-200b-3p-overexpressing culture supernatants decreased the proliferation of HUVECs (Fig. 6b, upper), which was increased by culture supernatants with miR-200b-3p-inhibition (Fig. 6b, lower). Next, tube formation assay was performed to evaluate the ability of HUVECs to form capillary-like structures. HUVECs cultured with miR-200b-3p-overexpressing supernatants exhibited reduced capacity to form capillary-like structures (Fig. 6c, left), whereas miR-200b-3p-inhibited supernatants increased the capacity when compared to control (Fig. 6c, right). These data suggest that hepatic miR-200b-3p downregulates angiogenic capability of HUVECs.  17,18 . Hence, we examined if exosomes released from cancer cells could transfer miR-200b-3p to endothelial cells. Exosomes were isolated from HCC cell lines and expression levels of miR-200b-3p in exosomes were measured. The expression levels of miR-200b-3p in the exosomes were high in Hep3B cells and low in HLE cells ( Fig. 7a), which were proportional to those detected in cells (Fig. 1c). Next, HUVECs were cultured with exosomes isolated from HLE cell culture supernatants and expression levels of ERG in HUVECs were measured. As shown in Fig. 7b, addition of exosomes suppressed the expression of ERG in HUVECs as compared with control (no exosomes). To confirm the role of exosome-associated miR-200b-3p in inhibition of ERG expression in HUVECs, we overexpressed miR-200b-3p in HLE cells, and exosomes were isolated from miR-200b-3p-overexpressing HLE cell culture supernatants (Fig. 7c). Exosomes isolated from miR-200b-3p-overexpressing HLE cell culture medium significantly reduced ERG expression in HUVECs when compared to control (Fig. 7d). These data suggest that miR-200b-3p is transferred via exosomes from hepatocytes to endothelial cells, resulting in suppression of endothelial ERG expression.

Discussion
Angiogenesis is necessary for cancer growth and metastasis and regulated by various molecules. In this study, we investigated the role of miR-200b-3p in regulating cancer angiogenesis in HCC. We used 40 paired human HCC tissues and adjacent non-cancer liver tissues and examined the expression of miR-200b-3p. We used Taqman advanced miRNA cDNA synthesis kit and a pair of advanced miRNA control to avoid amplification bias and reflect true expression level of mature miRNA. miR-200b is known decreased in hepatocellular carcinoma (HCC) 19 23 and are involved in promoting cancer growth and invasion. ERG is an essential regulator of www.nature.com/scientificreports/ www.nature.com/scientificreports/   www.nature.com/scientificreports/ endothelial homeostasis and tumor angiogenesis 24 , but there is no study about ERG as miRNA-200b-3p target.
In this study, we demonstrated that ERG is a miR-200b-3p target. The numbers of ERG positive endothelial cells in highly vascular cancer tissues were upregulated when compared to those in non-cancer tissues. Our in vitro experiments manipulating miR-200b-3p expression revealed that the decreased levels of miR-200b-3p in HCC tissues drive ERG expression in HUVECs. We know from the literature that angiogenesis in cancer tissues is regulated by various molecules. Among them, our data suggest that decreased levels of miR-200b-3p in cancer hepatocytes partly contribute to angiogenesis in HCC tissues. Cancer cells are known to release large numbers of exosomes, which deliver molecules that have a pathogenic role in cancer pathology 25,26 . Exosomes are heterogeneous membrane-enclosed structures released by cells. Exosomes mediate both autocrine and paracrine signal transduction by transferring proteins, RNA, and miRNAs 27,28 . Recent evidence indicates that exosomes promote HCC cell proliferation, growth, invasion, and metastasis, as well as the development of drug resistance 29 . In this study, we demonstrated evidence that miR-200b-3p is released from hepatocytes inside exosomes and transferred to endothelial cells. Although we did not quantify levels of exosomes from non-tumor and tumor hepatocytes, the expression levels of miR-200b-3p in cancer area were quite lower than those in non-cancer area. It is likely that lower expression levels of exosomal miR-200b-3p is due to decreased expression of miR-200b-3p in HCC cells. www.nature.com/scientificreports/ Our data indicate that hepatocyte-derived miR-200b-3p inhibits ERG expression in HUVECs. However, its effect was partial as assessed by overexpression or inhibition of miR-200b-3p. Although ERG is a target of miR-200b-3p in endothelial cells, a single transcript may be regulated by multiple miRNAs, as each miRNA is known to regulates hundreds of genes. For instance, miR-196a and miR-196b induced ERG downregulation in leukemia 30 . ERG is a target of miR-145 in colorectal cancer and prostate cancer 12,31 . Alternatively, angiogenic growth factors, including vascular endothelial growth factor (VEGF), platelet-derived growth factor (PDGF), and fibroblast growth factor (FGF), released from HCC cells 32 may affect ERG expression in HUVECs. Thus, ERG expression in endothelial cells is regulated by multiple mechanisms and a single inhibition of miR-200b-3p has a limited effect. Regulation of ERG by miRNA-200b-3p in other cancers is a subject for future investigation.
In this study, we did not evaluate the mechanisms underlying the downregulation of miR-200b-3p in HCC cells. Recent studies have revealed an importance of epigenetic control in the expression of distinct miRNAs. The expression of the miR-200 family is reported to be regulated by DNA methylation and histone modifications 33 . The promoter methylation status of miR-200b is reported to determine tumor outcome in gastric cancer 34   The results of our study are illustrated in Fig. 8. In non-tumor tissues, exosomal miR-200b-3p released from hepatocytes downregulates vascular proliferation by inhibiting ERG expression in endothelial cells. By contrast, miR-200b-3p expression in HCC cells is downregulated, which drives vascular proliferation by enhancing ERG expression in endothelial cells. Hence, miR-200b-3p may be a novel molecular target for the treatment of HCC.

Method
Human tissue samples. In this study, we employed 40 pairs of tumor and adjacent non-tumor tissue derived from patients with HCC, who underwent surgical resection between January 2015 and December 2016 at Okayama University Hospital. The patients who underwent chemotherapy or radiotherapy before the resection were not included in this study. All the hematoxylin and eosin-stained tumor glass slides used for diagnosis were reviewed and the degree of differentiation of cancer was recorded. One representative tumor slide was   www.nature.com/scientificreports/ relative expression levels validated by miR-191-5p and miR-26a-5p were constantly similar (not shown). Therefore, we have shown the expression data relative to miR-191-5p. miR-200b-3p expression in the exosomes was normalized using 1 µL of 1 pM spike-in control, ath-miR-159a (478411-mir). All primers were purchased from Applied Biosystems. All experiments were performed in triplicates and the relative expression was calculated using the ∆∆CT method.
Immunohistochemistry. Immunostaining was performed using the Histofine Simple Stain MAX-PO (Nichirei Biosciences Inc, Tokyo, Japan), following the manufacturer's instructions. Briefly, the sections (4 µm slices) were deparaffinized, rehydrated, and subjected to heat-induced epitope retrieval in 0.5 M EDTA buffer (pH 8.0). Next, the sections were treated with 0.3% H 2 O 2 in methanol and incubated with rabbit monoclonal anti-ERG antibody (1:1) (Histofine, Tokyo, Japan) for 90 min at room temperature. The sections were rinsed and incubated with peroxidase-conjugated secondary antibodies at room temperature for 45 min. Diaminobenzidine (DAKO, Carpinteria, CA, USA) was used as a chromogen. The images of four random consecutive areas (each area, 371 µm 3 ) were captured under 20 × magnification of a light microscope. The number of ERG-positive nuclei in the endothelial cells was counted using the Image J software.  Western blotting. The cells were lysed in RIPA buffer containing protease inhibitor. The total protein concentrations were measured using the BCA Protein Assay Kit (TaKaRa). Equal amounts of lysates were loaded per lane on 4-12% polyacrylamide gels (Thermo Fisher Scientific). The proteins were resolved by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. The resolved proteins were transferred onto nitrocellulose membranes (pore size, 0.45 µm). The membrane was incubated with anti-ERG antibody (1:10) or anti-GAPDH antibody (1:5,000) (Cell Signaling, MA, USA) overnight. Next, the membrane was incubated with the horseradish peroxidase-conjugated rabbit IgG antibody (1:1,000) (Cell Signaling, MA, USA). The proteins were visualized using the enhanced chemiluminescence detection reagents (ImmunoStar LD; FUJIFILM Wako, Osaka, Japan). The protein band intensity was quantified using Image Studio Lite software. All experiments were repeated in triplicates.

Dual
Scratch wound healing assay. The HUVECs were seeded in a 6-well plate to obtain the confluent monolayer. The wounds were generated by scratching the monolayer with a sterile 200-µL tip. The wounded cell monolayers were cultured for 24 h. The wound distance at different time points was measured by Image J software and presented as the percentage of wound closure at time 0. The experiments were performed in triplicates.
Proliferation assay. HUVECs were cultured in 96-well plate for 24 h. 50 µL of XTT labeling mixture (final XTT concentration 0.3 mg/mL, Cell proliferation Kit II, Sigma-Aldrich) was added into each well. After incubation at 37 °C for 1 h, the absorbance of each well was measured at 480-650 nm by using ELISA reader. Each sample was tested in duplicates, and the experiments were performed in triplicates.
Tube formation assay. The HUVECs were starved for 6 h in Medium 200 Phenol Red Free medium (Gibco). The cells (4 × 10 4 /well) were seeded onto the 24-well plate precoated with Matrigel Basement Membrane Matrix (Corning, NY, USA) and incubated with culture supernatants containing 1% FBS for 6 h. Next, the cells were incubated with 2 μg/mL Calcein (Invitrogen) for 30 min. The tube formation was visualized and photographed using the BZ-X700 fluorescence microscope (Keyence Corp, Osaka, Japan). The total tube length was measured by cellSens standard under 10X magnification. Each treatment was performed in triplicates, and 4 visions were counted to obtain an average.