HIF-1α Promotes Hepatocellular Carcinoma Metastasis by Regulating Angiogenesis and Epithelial-Mesenchymal Transition

Background: Invasion and metastasis of hepatocellular carcinoma (HCC) still remain to be hard in medical society. However, little knowledge is known regarding the hypoxia impact in HCC with angiogenesis and epithelial-mesenchymal transition (EMT). The aims of this study were to explore the regulating roles of hypoxia-inducible factor-1α (HIF-1α) in angiogenesis and EMT of HCC. Method: The levels of HIF-1α, angiopoietin-2 (Ang-2), and vascular endothelial growth factor (VEGF) expression in a cohort of chronic liver diseases were detected by enzyme- linked immunosorbent assays, and their dynamic up-regulations were conrmed in model of rat hepatocyte malignant transformation. After HIF-1α gene transfected with specic miRNA, biological behaviors of HCC cells were analyzed by transwell or invasion assay; angiogenesis and EMT were analyzed at protein level by Western blot or at mRNA by quantitative real-time PCR. Results: The levels of circulating HIF-1α, VEGF, and Ang-2 in the HCC group (145.6 ± 32.6 μg/L, 458.9 ± 125.3 μg/L, and 42.9 ± 5.1μg/L) were signicantly higher (P < 0.001) than those in the LC (79.5 ± 8.4 μg/L, 206.8 ± 56.8 μg/L, and 26.2 ± 6.1 μg/L) or the CH (60.1 ± 18.8 μg/L, 178.1 ± 85.4 μg/L, and 21.8 ± 6.9 μg/L) group, respectively. Dynamic up-regulations of HIF-1α and angiogenic factors have been conrmed by rat model with hepatocyte malignant transformation. There were closely positive correlations (P < 0.001) between them of HIF-1α and VEGF or Ang-2. After HCC cells transfected with specic HIF-1α-miRNA, the levels


Background
Hepatocellular carcinoma (HCC) is one of the most commonly diagnosed and deadly malignant tumors worldwide [1]. HCC was still one of the most cancers in the inshore area of Yangtze River, China, due to hepatitis B virus (HBV) infection-related chronic hepatitis or liver cirrhosis [2]. There have been no more effective therapeutic options in the advanced HCC beyond sorafenib or a multi-targeted tyrosine kinase inhibitors, because hepatic hypoxia enhances cell proliferation, angiogenesis, and suppresses differentiation and apoptosis leads to resistance of transarterial chemoembolization (TACE) [3,4].
Hypoxia-inducible factor-1 (HIF-1) is a heterodimer consisting of two subunits (α tightly regulated by changes in oxygen regimes and β constitutively expressed), whereas HIF-1α is a master regulator of the transcriptional response of angiogenesis by multiple mechanisms such as some growth factors, tumor suppressor genes, oncogenes and formation of epithelial-mesenchymal transition (EMT), although it primarily involves protein ubiquitination [5,6]. Activation of HIF-1α might regulate a repertoire of key angiogenic genes including VEGF [7] and Ang-2 [8]. Given HIF-1α central role in angiogenesis, some angiogenic growth factors should become prime targets for therapeutic angiogenesis of HCC [9,10].
HIF-1α activation at early stages of rat hepatocyte malignant transformation has been con rmed, and HIF-1α up-regulation might be associated with angiogenesis of HCC progression [11,12]. Among the many regulatory factors of HCC, the dynamic change and relevance of HIF-1α and VEGF play vital roles and have been observed after TACE of HCC patients [13]. Anti-angiogenic therapy is bene cial to HCC patients following surgical resection of tumor. However, satisfactory results have not been achieved for HCC because of HIF-1α over-expression might in uence HCC biological behaviors and act in concert with signal pathways, and stimulate the required angiogenic growth factors endogenously [14 ~ 16]. However, HIF-1α how to mediate the molecular mechanisms of angiogenesis or EMT should be identi ed.
Therefore, the aims of the present study were to detect the serological levels of HIF-1α, angiogenic factors expressions in a cohort of cases with chronic liver diseases, con rm by the dynamic alterations in model of rat hepatocarcinogensis, and establish HIF-1α inactivation via speci c miRNA for analyzing on effects of HCC cell proliferation, angiogenesis and EMT formation.

Patients
A cohort of 478 patients with chronic liver diseases including HCC (n = 298), liver cirrhosis (LC, n = 92), and chronic hepatitis (CH, n = 88) from the Hospital of Nantong University, China were investigated with written or verbal consent. Total 74 healthy people as normal controls (NC) were obtained from the Nantong Central Blood Bank, with negative viral markers (HBsAg, HBV-DNA, and anti-HCV antibody), normal ALT activity and liver B ultrasonic examination. Data of patients with chronic liver diseases & normal control are shown in Table 1. Total 5 mL of blood from each patient or d healthy control were collected in the morning, and serum α-fetoprotein (AFP) and biochemistries were detected at once.
Diagnostic criteria of liver cancer and hepatitis were based on the Chinese National Collaborative Cancer Research Group and Viral Hepatitis Meeting, respectively. This study was approved by the Ethics Committee of Nantong University Hospital (TDFY2013008) of China and performed in line with medical ethics of the Helsinki Declaration.

Hepatocarcinogenesis model
Rat models were approved by the guidelines of Animal Care and Use Committee of Nantong University, China. Total 48 Sprague-Dawley rats (SD) with 4 ~ 6-wk-old that obtained from the Experimental Animal Center of Nantong University were made for hepatocarcinogenesis model in clean environment, 12-h light/dark cycle, and 55% humidity, and schematic representation of the model and rat grouping are shown in Fig. 1. Control rats (n = 12) were fed normal diet, model rats (n = 36) was fed with containing 2acetylamino uorene (2-AAF, 0.05%, Sigma) diet, and checked every day. After rats sacri ced at different time, livers and blood were collected for analysis. Livers were used for pathology and total protein extraction. Liver pathological examination with Hematoxylin & Eosin (H.&E.) staining was diagnosed by independent pathologists, and histological grouping. Dynamic alterations of whole gene expression pro ling were detected by Affymetrix GeneChip® Rat Genome 230 2.0 Array (total 28,000 gene, YESLAB., Shanghai, China) and levels of VEGF, Ang-2 and HIF-1α in liver tissue supernatant and sera of rats were quantitatively detected by enzyme-linked immuno-sorbent assays.

In vitro transfection
Human HepG2 or Hep3B cells (5 × 10 3 ) were cultured on six-well plates for overnight incubation. They were divided into blank control (Con), negative miRNA (Neg) and miRNA (MiR) groups. The MiR or Neg group was transfected with HIF-1α miRNA or negative miRNA according to the instructions of relatedreagent kit (Roche, Germany).

Cell migration or invasion assay
Quantitative and qualitative analysis of HepG2 or Hep3B cells migration were assessed by in vitro Transwell assay with modi ed Boyden Chambers and Transwell-coated Matrigel membrane lter (BD Biosciences, Bedford, MA, USA). Cells (5 × 10 3 ) from Con, Neg, and MiR groups (n = 3/per group) were plated onto the upper compartment in without FBS or 10% FBS in the lower chamber as a chemoattractant. Fluorescent images of nuclear Hoechst staining (10 µg/mL) were captured at 24 h of incubation in a 5% CO 2 humidi ed at 37 ℃. Percentages of migrated cells in each group were counted from 10 random microscope elds for each sample in 3 independent experiments. For cell migration analysis, the modi ed Boyden Chambers without the Transwell-precoated Matrigel membrane lter in above method was performed.
Quantitative real-time PCR

Western blot analysis
Total proteins from HCC cells were lysed in RIPA buffer with protease and phosphatase inhibitors (Roche) and the concentrations were quanti ed with the Bicinchoninic acid Protein Assay Kit (Beyotime Institute of Biotechnology, Shanghai, China), and an equal amount of 50 µg protein was separated by 10% sodium dodecyl sulfate-polyacrylamide gels electrophoresis (SDS-PAGE) and then transferred onto polyvinylidine di uoride (PVDF) membranes (Millipore, Billerica, MA, USA), blocked with 5% bovine serum albumin (BSA) in blocking buffer (Solarbio, China) for 2 h at room temperature, and incubated with speci c primary rabbit anti-human antibodies overnight at 4 ℃, and β-actin (CST, USA) was used as protein loading control. The HIF-1α primary antibody was obtained from Santa Cruz (Univ-Bio., Shanghai, China) and the antibodies against Ang-2, Vimentin, E-Cadherin, Twist, and Snail were purchased from Abcam (Cambridge, MA, USA). Then the membranes were incubated with horseradish peroxidase-conjugated secondary goat anti-rabbit antibody (Abbkine, China). Detection was performed by enhanced chemiluminescence kit (Beyotime Institute of Biotech., Shanghai, China). All images were taken by the Quantity One software (Bio-Rad, Laboratories, Inc., USA).

Enzyme-linked immunosorbent assay (ELISA)
Concentrations of VEGF, Ang-2, and HIF-1α in sera or supernatant were quantitatively detected according to the manufacturer's instructions with ELISA kits. Their levels were calculated using a standard curve generated with speci c standards provided by the manufacturer with inter and intra-assay variances under 10%. ELISA kits for VEGF, Ang-2 and HIF-1α detections were purchased from the R&D systems, Abingdon, UK; ADL Biotech Dev Co., USA; and Abcam Co., Shanghai, China, respectively.

Statistical analysis
Patients were divided into HCC, CH, and LC groups, with healthy persons as a NC group. Rat livers by pathological examination (H&E staining) were divided into four groups of rHCC, Pre-C, Deg, and NC. Data were expressed as mean ± standard deviation (± SD), and analyzed by SPSS19.0. Pearson χ 2 test, ANOVA and q test were performed to analyze the difference between different groups. A P value < 0.05 was considered to be statistically signi cant.

Results
Serum VEGF, HIF-1α and Ang-2 in HCC patients Comparative analysis of serum HIF-1α, VEGF and Ang-2 levels in a cohort of patients with benign or malignant chronic liver diseases are shown in Table 2. The levels of VEGF, HIF-1α and Ang-2 were signi cantly higher (P < 0.001) in the HCC group than those in the NC, CH or LC group. If the cutoff values were set at over 300 µg/L for VEGF, 100 µg/L for HIF-1α or 35 µg/L for Ang-2, the positive rates in the HCC group (83.6%, 88.9% or 88.6%) were signi cantly higher (P < 0.001) than those in the LC group (16.3%, 27.2% or 8.7%) or in the CH group (13.6%, 2.3% or 4.6%), and none in the NC group, respectively. The circulating levels of HIF-1α, VEGF and Ang-2 were synchronous increasing from NC, benign liver diseases to HCC progression.
HIF-1α, VEGF and Ang-2 in rat hepatocarcinogenesis Schematic representation of rat hepatocarcinogenesis model with grouping according to the liver pathological changes (H.&E. staining) is shown in Fig. 1. During hepato-carcinogensis, lots of relatedgenes were found with alterations (Sup- Fig. 1 and Sup- Table 1) and HIF-1α with increasing expressions in rat livers (Sup- Fig. 2) at different stages of HCC formation. The levels of HIF-1α, VEGF and Ang-2 expressions in livers (speci c concentration, ng/per mg wet livers) or sera (µg/L) of rats were quantitatively investigated from normal control (NC) liver to granulose degeneration (Deg) at early stage, precancerous lesions (Pre-C) at middle stage, and rat HCC (rHCC) formation at later stage. The dynamic alterations of hepatic or circulating HIF-1α, VEGF and Ang-2 expressions in the different groups are shown in Table 3. No matter livers or blood, the levels of HIF-1α, VEGF and Ang-2 were dynamically upregulated during rat hepatocyte malignant transformation to accelerate the formation of new blood vessels to meet the need for oxygen, especially in the Pre-C or rHCC group.

Relationship between HIF-1α and VEGF or Ang-2
Down-regulating levels of VEGF and Ang-2 expression after intervening HIF-1αmRNA transcription in HepG2 or Hep3B cells are shown in Table 4. To further estimate the effect of HIF-1α on VEGF or Ang-2, an endogenous gene product of HIF-1α miRNA transcription activity in the culture media of HepG2 or Hep3B cells were analyzed at 24 h, 48 h, and 72 h after miRNA transfection. The levels of VEGF and Ang-2 were signi cantly decreased (P < 0.01) between the MiR group and the Con group.

inhibition of HCC biological behaviors
The interfering HIF-1α mRNA expression on effects of the biological behavior of HCC cells are shown in Fig. 3. After the HepG2 or Hep3B cells transfected with miRNA interfering plasmid for 48 h or 72 h, the cell proliferation rates were signi cantly lower than that of the negative or control group (Fig. 3a), with the longer time and the higher inhibition rate (Fig. 3b), with lowest the clone formation number of HCC cells (Fig. 3c1). After the HCC cells transfected with miRNA interfering plasmid for 72 h, the migration capacity of HCC cells in the miRNA group were signi cantly lower (P < 0.001, Fig. 3d) than those of the control or negative group. Also, the ability of HCC cells invasiveness in the miRNA group were signi cantly lower (P < 0.001, Fig. 3e) than those of the control or negative group, indicated that interfering with HIF-1 gene transcription could signi cantly affect the biological behaviors of HCC cells, especially in migration and invasiveness.

HIF-1α activation promotes EMT formation
Interfering HIF-1α gene transcription of HCC cells on effects of EMT-related protein expressions are shown in Fig. 4. Alterations of EMT-related epithelial marker (E-cadherin), mesenchymal indicator (Vimentin), transcriptional factor (Snail & Twist) levels in the culture medium of HCC cells were analyzed by Western blot analysis. There were no signi cant differences of those proteins between the Neg group and the Con group. when HepG2 cells (a ~ d) or Hep3B cells (e ~ f) with stable silencing HIF-1α transcription in the MiR group, the levels of the E-cadherin were signi cantly (P < 0.001) increasing than that in the NC group (Fig. 4a, a1& Fig. 4e, e1); otherwise the levels of Vimentin (P < 0.001, Fig. 4b, b1& Fig. 4f, f1), and Snail (P < 0.001, Fig. 4c, c1 & Fig. 4g, g1) and Twist (P < 0.001, Fig. 4d, d1 & Fig. 4h, h1) were signi cantly lower less than those in the NC group.
The expressions of EMT-related proteins in HCC cells were signi cantly inhibited by interfering with HIF-1α mRNA except of E-cadherin, and it should be a novel regulating mechanism (Fig. 5) insight into the invasion and metastasis of HCC is by decreasing E-cadherin or increasing Vimentin, Twist and Snail signaling for molecular-targeted therapy. HIF-1α over-expression could promote the proliferation of cancer cells, EMT, and angiogenic factor expressions for HCC metastasis.

Discussion
HCC is one of the most common malignant cancers worldwide and a high malignancy solid tumor with hypoxia microenvironment, high-vascular, and traditional-therapy resistance [16]. HIF-1α is a key transcription regulator for many genes and promotes tumorigenicity by up-regulating its target genes, which were involved in apoptosis, proliferation, angiogenesis, invasion and metastasis of HCC such as VEGF, Ang-2, Wnt3a, IGF-II, and so on [17]. This study has investigated the relationship between HIF-1α and HCC-related angiogenesis or EMT, and demonstrated that abnormalities of HIF-1α, Ang-2 and VEGF expressions in HCC, the up-regulating expression in rat hepatocarcinogenesis, and the acting mechanism of HIF-1α were con rmed on effects of the proliferation, angiogenesis and EMT of HCC cells with speci c miRNA transfection.
Maintenance appropriate oxygen and angiogenic factors are essential for HCC growth. HIF-1α has been implicated in regulating hepatic in ammatory progression, and associated with multiple in ammationrelated cancers or signaling pathway activation of HBV-related HCC [18]. The presence of hypoxia in tissues restricts HCC overgrowth because of HIF-1α regulating cellular metabolism, immune escape, angiogenesis, metastasis, extracellular matrix remodeling, cancer stem cells (CSC) [19] or MDR via regulating PI3K/AKT/HIF-1α/MDR-1 pathway [20]. HIF-1α-de cient HCC cells displayed signi cantly reduced anchorage-independent growth and enhanced sensitivity toward chemotherapy [4]. Angiogenesis is thought to depend on a perfectly coordinated balance between endogenous-positive and negative regulatory factors. The investigation of HIF-1α, VEGF and Ang-2 levels in sera of patients with benign or malignant chronic liver diseases suggested that their synchronous increasing expression from normal control, chronic hepatitis, liver cirrhosis to HCC [21].
Abnormalities of HIF-1α and angiogenic factors were associated with rat hepatocarcinogenesis [22]. HIF-1α binds to an evolutionary conserved HRE located in the rst introns of Ang-2. All of the HIF-1α, VEGF and Ang-2 expressions demonstrated a tendency to increase with the histopathological changes: HCC more than precancerosis more than degeneration more than controls. Their levels were markedly higher in the rHCC and precancerous groups than those in the degeneration or control group. Blood levels of HIF-1α, VEGF and Ang-2 expressions in the HCC or precancerous group were higher than in the degeneration or control group. There was an apparent positive correlation between in blood and livers. HIF-1α overexpression was con rmed during hepatocytes malignant transformation with HCC progress, because it regulates the transcription of downstream numerous genes including angiogenesis or proliferation to set basis for HCC growth and metastasis [23]. However, the speci c miRNA decreased HIF-1α expression and suppressed angiogenesis in the HCC cell lines by down-regulating VEGF and Ang-2, indicated that HIF-1α regulate the angiogenesis of HCC.
EMT is a cellular programmed that is known to be crucial for malignant progression that increases the HCC invasive and metastasis potential with increasing expression of mesenchymal indicators or transcription factors and down-regulating epithelial marker levels [24]. In the context of liver neoplasias, EMT confers on cancer cells promoting tumor-initiating and easily metastatic potential, and more resistance to elimination by therapeutic regimens for HCC. Vimentin might increase the migration and invasiveness of HCC and relate to reduce E-cadherin and up-regulate N-cadherin, while increased vimentin was associated with poor prognosis of HCC. In this study, the relationship between the EMT-related biomarkers (E-cadherin, Vimentin, Snail, and Twist) and HIF-1α were investigated at cell level. After the HepG2 or Hep3B cells with stable silencing HIF-1α in the MiR group compared with the CN or Neg group, the E-cadherin at protein level was signi cantly un-regulation, but the levels of Vimentin, Snail, and Twist were markedly down-regulating expression, suggesting that HIF-1α promote HCC metastasis by increasing EMT ability and as a potential molecular target for HCC therapy.

Conclusions
Our data provide novel evidence that HIF-1α activation could promote metastasis and invasion of HCC via regulating angiogenesis and EMT-related proteins. It should provide a regulating mechanism insight into HCC metastasis. From silencing HIF-1α gene transcription could inhibit the metastasis with angiogenesis or EMT formation of HCC, and had got some associations between HIF-1α and metastasis.
However, the exact mechanisms were needed to be con rmed with more studies in vitro and in vivo. Further work should be done how to application of HIF-1α miRNA plus multi-targeting strategies for HCC effective therapy.   Silencing HIF-1α with EMT of HCC cells Epithelial or mesenchymal biomarker and transcriptional factors were involved in HCC metastasis with EMT. a ~ d, HepG2 cells; e ~ f, Hep3B cells; After silencing HIF-1α gene, the related-proteins were analyzed by the Western blotting with β-actin as loading control. a & e, Ecadherin, and a1 & e1, the relative ratio (n = 6); b & f, vimentin, and b1 & f1, the relative ratio (n = 6); c & g, snail, and c1 & g1, the relative ratio (n = 6); d & h, twist, and d1 & h1, the relative ratio (n = 6). Con, the Con group; Neg, the Neg group; MiR, the MiR group. *P < 0.05 or **P < 0.001, compared with the Con group.