Adrenomedullin promotes angiogenesis in epithelial ovarian cancer through upregulating hypoxia-inducible factor-1α and vascular endothelial growth factor

Adrenomedullin (ADM) is a multi-functional peptide related to many kinds of tumors. This study was aimed to investigate the role of ADM on angiogenesis in epithelial ovarian cancer (EOC) and its possible mechanism. The expressions of ADM, vascular endothelial growth factor (VEGF), hypoxia-inducible factor-1α (HIF-1α) and CD34 were examined by immunohistochemistry staining. The relationship among ADM, HIF-1α, VEGF and micro-vessel density (MVD) was assessed in 56 EOC tissues. CAOV3 cells were stably transfected with pcDNA-ADM (plasmid overexpressing ADM gene) or pRNA-shADM (small interfering RNA for ADM gene). Real-time PCR and western blot analysis were performed to detect the expressions of HIF-1α and VEGF. The MTT, transwell migration assay and in vitro tube formation analysis were used to evaluate the proliferation, migration, and tube formation ability of human umbilical vein endothelial cells (HUVECs) which were pretreated with ADM or ADM receptor antagonist ADM22-52. Our findings showed that ADM expression was positively correlated with the expressions of HIF-1α, VEGF or MVD in EOC. ADM upregulated expression of HIF-1α and VEGF in CAOV3 cells. ADM promoted HUVECs proliferation, migration and tube formation. In conclusion, ADM was an upstream molecule of HIF-1α/VEGF and it promoted angiogenesis through upregulating HIF-1α/VEGF in EOC.


Positive Correlation between the Expressions of ADM and MVD in EOC.
To detect the correlation between ADM expression and tumor angiogenesis, we assessed the micro-vessel density (MVD), the marker of angiogenesis, in the 56 EOC tissues. MVD was defined as the mean number of CD34 + vessels per section. A significantly higher MVD in EOC was found compared to normal ovarian tissues (p = 0.02). MVD in EOC was associated with degree of differentiation (p = 0.014), but not with age and Federation International of Gynecology and Obstetrics (FIGO) stage (Table 2). EOC tissues with higher integrated optical density (IOD) of ADM expression had a significantly higher MVD value. Likewise, EOC tissues with higher IOD of VEGF or HIF-1α expression showed similar phenomena. The correlations of ADM, HIF-1α or VEGF expression with MVD were positive ( Table 3).

Expressions of HIF-1a and VEGF was regulated by ADM in CAOV3 cells. To determine whether
ADM regulates HIF-1α and VEGF transcriptional activation in EOC, CAOV3 cells were transfected with pcDNA-ADM (plasmid overexpressing ADM) or pRNA-shADM (small interfering RNA for ADM) to upregulate or knockdown the ADM gene, then the expression levels of HIF-1α and VEGF in CAOV3 cells were examined using real-time PCR and western blot. As shown in Fig. 2A, when ADM gene was upregulated in CAOV3 cells with plasmid pcDNA-ADM, HIF-1α and VEGF mRNA expressions were enhanced greatly (p = 0.001 and 0.006, respectively), especially VEGF more than 3 times higher compared with control group. On the contrary, when ADM gene was silenced with shRNA ( Fig. 2B), VEGF and HIF-1α mRNA expressions were inhibited significantly (p = 0.000 and 0.046, respectively). The protein expression of HIF-1α showed the similar change

ADM Promotes HUVECs Proliferation, Migration and Tube Formation in Vitro.
To determine the direct angiogenic effect of ADM, various angiogenic properties were studied using human umbilical vein endothelial cells (HUVECs). MTT assay was performed to evaluate the ADM effect on HUVECs proliferation. Figure 3A indicated that ADM promoted HUVECs proliferation directly in a time-dependent manner (24 h, 48 h, 72 h, p = 0.000, 0.049, and 0.025, respectively) and ADM22-52 inhibited HUVECs proliferation also in a time-dependent manner (24 h, 48 h, 72 h, p = 0.047, 0.002, and 0.021, respectively). ADM resulted in a promotion of 133.26% in 72 h treatment. Early at the 24 h treatment, the inhibition of ADM22-52 could be observed. So, in the following studies we chose 24 h as a representative time point of ADM22-52. Interference of ADM expression also regulated the protein levels of VE-cadherin and MMP-9, which plays the important role in tumor angiogenesis ( Fig. 3B,C). Therefore, a transwell chamber system was employed to determine the effects of ADM on the migration of HUVECs. As shown in Fig. 4A, ADM (100 nM for 24 h) increased HUVECs migration across the transwell membrane (p = 0.000). Contrarily, ADM22-52 (1 nM for 24 h) decreased the invasive potential of HUVECs (p = 0.000) co-culture with CAOV3 cells, indicating ADM secreted from CAOV3 cells is crucial for HUVECs migration (Fig. 4B). We also evaluated the effect of ADM on the formation of capillary-like tube structures by plating HUVECs on Matrigel. The result showed that ADM promoted tube like structure formation of cultured HUVECs, and ADM22-52 inhibited this effect (p = 0.000, Fig. 5A,B).

Discussion
The present study demonstrated that ADM promoted tumor angiogenesis through upregulating HIF-1α /VEGF in EOC in vitro, ADM/HIF-1α /VEGF might be a new signaling pathway playing a role in this process. Strong evidence suggested that ADM involved in physiological and pathological angiogenesis in some tissues and cell lines. ADM induces the growth of human endometrial microvascular endothelial cells and regulates the angiogenesis in female reproductive tract 12 . In uterine leiomyomas and renal tumors, the expression of ADM mRNA is correlated with vascular density 13,14 . In xenografted tumor models utilizing human endometrial, breast, pancreatic tumor cell lines or human glioblastoma cells, ADM overexpressing transfectants increase the growth of blood vessels or vascular density [15][16][17][18] . VEGF and HIF-1α are potent inducer of angiogenesis and tumor growth 19,20 . Preclinical studies suggest that VEGF-mediated angiogenesis is important in initiating and mediating the growth of ovarian cancers 21 . HIF-1α is often upregulated in human cancers to regulate VEGF expression by binding to the hypoxia responsive element of VEGF promoter 22 .
In the present study, we investigated the relationship between ADM expression and MVD of 56 EOC tissues with immunohistochemical analysis, and also the relationship among ADM, VEGF and HIF-1α in EOC for the first time. We found the direct clues that ADM, VEGF and HIF-1α expressions were all positively related to MVD in EOC in vivo, which agreed with the results shown in renal tumors 14 and uterine leiomyomas 13 . We also observed a positive correlation among ADM, VEGF and HIF-1α expressions in EOC in vitro, which is consistent with our previous report 23 . Higher level of ADM was closely correlated with expressions of HIF-1α and VEGF. These data indicated that ADM might induce HIF-1α /VEGF expression and contribute to angiogenesis in clinical EOC. The effect of ADM on angiogenesis may be bound up with VEGF and HIF-1α . Besides, we found that MVD in EOC was only associated with degree of differentiation, which was similar to ADM expression in EOC in our previous study 24 . We therefore considered that MVD could reflect the biological aggressiveness in EOC as ADM.     In order to further investigate the role of ADM/HIF-1α /VEGF on angiogenesis in EOC, we constructed plasmid pcDNA-ADM and pRNA-shADM to endogenously increase or decrease the ADM gene expression in EOC cell line CAOV3. Then we provided exogenous ADM and ADM22-52, the ADM antagonist, to CAOV3 cells. HIF-1α /VEGF expressions were examined by real time PCR and western blot. We found that ADM upregulated HIF-1α and VEGF expressions, and ADM22-52 downregulated HIF-1α and VEGF expressions in CAOV3 cells. These results suggested that ADM was an upstream molecule of HIF-1α /VEGF. HIF-1α and VEGF were important regulators in tumor angiogenesis, which were required for tumorigenesis and tumor development 25,26 . ADM may promote angiogenesis via upregulating HIF-1α /VEGF.
In addition, we treated HUVECs with ADM and ADM22-52 to determine the direct effect of ADM on angiogenesis. We observed that ADM promoted HUVECs proliferation, migration and tube formation in vitro while ADM22-52 inhibited these effects. These observations were consistent with the previous results 27 .
The angiogenic process is regulated by several "classic" factors, such as VEGF, HIF-1α and fibroblast growth factor-2 (FGF-2). These factors together with their receptors are currently the main targets against angiogenesis. Our results confirmed that ADM is an upstream molecule of HIF-1α /VEGF, so we suppose ADM may be involved in the regulation of other angiogenic "classic" factors to promote tumor angiogenesis. It was reported that HIF-1α induced the expression of ADM mRNA under normoxic and hypoxia conditions in the human ovarian carcinoma cell line OVCAR-3 10 , which was opposite to what we observed. Furthermore, recent observation 24,28 proved that focal adhesion kinase (FAK) and ADM may play a cooperative role in EOC. Additional studies are needed to further demonstrate the role of ADM on EOC angiogenesis and the relationship between ADM and HIF-1α .
In summary, our present study showed that ADM was an upstream molecule of HIF-1α /VEGF, and promoted angiogenesis through upregulating HIF-1α /VEGF in EOC. Thus, ADM/HIF-1α /VEGF signaling pathway may be a possible therapeutic target in EOC in the future.
Human umbilical vein endothelial cells (HUVECs) were freshly isolated as described previously 29 , from human umbilical veins of newborn obtained from a parturient at Shenyang 242 Hospital who gave written informed consent. HUVECs were routinely grown in DMEM (Gibco, USA) supplemented with 10% fetal bovine serum (Gibco, USA) and endothelial cell growth supplement (BD Biosciences, USA) at 37 °C and 5% CO 2 . In our experiments, only the first three passages of each HUVECs primary cultured were used.  Immunohistochemical Staining. Unstained 4 μ m paraffin sections from tissue sample were deparaffinized and rehydrated. The sections had been hematoxylin-and-eosin (HE) stained to confirm histological diagnosis by two pathologists according to the World Health Organization (WHO) classifications. All the sections were subjected to antigen retrieval by heating in Tris-EDTA buffer at pH 8.0 in an autoclave sterilizer for 2 min. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide (H 2 O 2 ) in methanol and non-specific binding sites were blocked with protein blocking solution (5% normal horse and 1% normal goat serum). Primary antibodies against ADM (1: 100 dilution, R&D, USA)/VEGF (1:200 dilution, Santa Cruz, CA)/HIF-1α (1:50 dilution, Santa Cruz, CA)/CD34 (1:100 dilution, DAKO Cytomation, Glostrup, Denmark) were added and sections were incubated over night at 4 °C. Then the sections were treated with secondary antibody and incubated with streptavidin-peroxidase (SP) complex (Maixin Biotechnology, Fujian, China) for 40 min at room temperature. Binding sites were visualized with 3, 3-diainobenzidine (DAB) (Maixin Biotechnology, Fujian, China) after 1 min incubation. Finally, sections were counterstained with hematoxylin, dehydrated with ethanol, fixed with xylene and mounted. Phosphate-buffered saline (PBS) was used instead of the primary antibodies for the negative control. For quantitative analysis of immunostaining intensity, integrated optical density (IOD) was employed to compute the relative value of each section. Digitally fixed images were analyzed at x200 magnification using an AxioImager A1 (Zeiss, Germany) light microscope equipped with an image analyzer (Image Pro Plus, Italy). IOD was calculated for arbitrary areas (20 arbitrary areas/samples, 1000 μ m × 1500 μ m) and each section analyzed with the same size.
Microvessel density (MVD) per section was measured using immunostaining with a CD34-monoclonal antibody. MVD was assessed according to the international consensus 30 . The entire section was scanned systematically at low magnification (× 100) in order to identify the most intense areas of neovascularization (hotspots) within the tumor. After five hotspots areas with the highest number of capillaries and small venules were identified, microvessels were counted at high power magnification (× 400), and the average in five fields was calculated.
Stably Transfected CAOV3 Cells. CAOV3 cells (5 × 10 4 ) were seeded in 6-well tissue culture plates (BD Falcon, USA) and allowed to attach overnight to achieve 70% confluence at the time of transfection. 2 μ g of purified plasmid DNA, encoding either vector alone, pcDNA-ADM or pRNA-shADM was transfected into cells using Lipofectamine TM 2000 (Carlsbad, CA) per manufacturer's protocol. The cells were incubated for 24 h, and then 1 ml medium containing 20% serum was added to each well. The cells were cultured and screened in medium containing 10% serum and 400 μ g/ml G418 for at least 3 weeks. Then, stable transfectants were formed. pcDNA3.1 or pRNAU6.1/neo vector was used as a control. The targeting sequences were validated by real-time PCR and western blot analysis.
Real-time PCR. Total RNA was isolated from treated confluent CAOV3 cells using Trizol reagent according to the manufacturer's instructions, resuspended in RNase-free water and stored at − 80 °C. RNA concentration was measured by absorbance reading at 260 nm. Total RNA (2 μ g) was reverse transcribed into cDNA using reverse transcription system (Promega, USA). The final cDNA product was stored at − 20 °C. Real time PCR was carried out using SYBR Premix Ex Taq Kit (Takara, Tokyo, Japan) on Applied Biosystems 7500 (Foster City, CA, USA). The PCR reaction mixture of 20 μ l contained 10 μ l SYBR Premix Ex Taq (2 ×), 0.7 μ l forword primer, 0.7 μ l reverse primer and 2.0 μ l cDNA. The forward and reverse PCR primers were summarized below (Supplementary Table 1). The reaction conditions were 50 °C for 2 min, 95 °C for 10 min, 95 °C for 30 sec and 60 °C for 30 sec for 50 cycles. The mRNA levels were normalized with respect to the levels of GAPDH in each sample.
Western Blot. Total protein lysates were isolated from treated confluent CAOV3 cells using lysis buffer (0.01 mmol/l Tris-HCl, pH 7.6, 0.1 mmol/l NaCl, 0.001 mol/l EDTA, pH 8.0, 1 μ g/ml Aprotinin, 100 μ g/ml PMSF). Protein concentrations were determined by BCA Protein Assay Kit (PIERCE, Rockford, IL). Proteins were separated by 10% SDS-PAGE and then transferred to polyvinylidene fluoride membranes, which were then blocked for 2 h in 5% defatted milk in Tris-buffered saline containing Tween-20 (TBST, 10 mM Tris-HCl, 150 mM NaCl, 0.1% Tween-20). Membranes were then incubated for 2 h at room temperature with the following primary anti was added to each well and incubated for 4 h at 37 °C, which was followed by adding 150 μ l dimethyl sulfoxide (DMSO) and incubating at 37 °C for an additional 10 min. Absorbance was read at 560 nm on a microplate reader.
Migration Assay. For HUVECs migration assay, Transwell inserts (8 μ m pore, Corning Costar Corp, Cambridge, MA, USA) were used as described 32 . At the top chambers, HUVECs were counted and resuspended in serum-free DMEM containing 0.1% BSA at a final concentration of 5 × 10 4 cells/100 μ l/well. The bottom chambers (600 μ l) were filled with DMEM with 10% FBS. The HUVECs were co-incubated with PBS (as control), ADM (100 nM) or ADM 22-52 (1 nM) and allowed to invade for 24 h at 37 °C with 5% CO 2 . After incubation, the noninvasive cells that remained on the upper surface of the filter were removed by a cotton swab. The invaded cells were fixed and stained with 0.1% crystal violet, and then counted using a light microscope in 5 random fields per well.
In Vitro Tube Formation Assays. The effect of ADM on angiogenesis in vitro was examined by tube formation assay. The wells of a 96-well plate were coated with 50 μ l of ice-cold Growth Factor Reduced Matrigel (BD Bioscience, San Jose, CA) at 37 °C for 1 h. HUVECs were seeded at a density of 5 × 10 4 cells per well in 200 μ l complete culture medium containing ADM (100 nM) or ADM 22-52 (1 nM) or PBS (equal volume as control). After incubation for 24 h at 37 °C with 5% CO 2 , the cultures were photographed and the tube-like structures were evaluated.
Statistical Analysis. Data were presented as the mean ± standard deviation (SD), and analyzed with one-way ANOVA or Student's t test where applicable, by using SPSS (version 16.0, Chicago, IL, USA). Spearman rank correlation analysis was performed to examine the correlations among different variables. All the experiments were repeated at least three times. A p < 0.05 was defined as statistically significant.