Hypoxia-induced vasculogenic mimicry formation in human colorectal cancer cells: Involvement of HIF-1a, Claudin-4, and E-cadherin and Vimentin

Vasculogenic mimicry (VM) plays an important role in colorectal cancer (CRC) metastasis, and both hypoxia and the epithelial-mesenchymal transition (EMT) are necessary for VM. In this study, HIF-1α expression was upregulated in the VM-positive CRC cell line HCT-116 and thereby affected the expression of the EMT-related markers Claudin-4, E-cadherin (E-cd) and Vimentin(VIM). SB431542 and U0126EtOH, which can inhibit of EMT were used to treat HCT-116 and HCT-8 in these experiments. Both of the inhibitors had significant effect on EMT markers and the formations of VM in CRC cells. In addition, knockdown of HIF-1α in the HCT-116 cells inhibited their capacity for VM. Our study reveals a regulatory role for HIF-1α in VM and suggests that targeting either HIF-1α or EMT may be a valuable strategy for the elimination of CRC metastasis.

Scientific RepoRts | 6:37534 | DOI: 10.1038/srep37534 promotes vasculogenic mimicry by inducing EMT in ovarian carcinoma 11 . However, it is still unclear whether EMT is associated with VM or if EMT directly causes VM in CRC.
For these reasons, we used in vivo and in vitro assays to explore the mechanisms by which the HIF-1α /EMT pathway regulates VM in CRC. The data may offer a theoretical basis for the development of targeted CRC therapies.

Materials & Methods
Experimental animals. We confirmed that the Ethical Committee of the Shanghai University of Traditional Chinese Medicine approved all the experiments described in this paper. All experiments were performed in accordance with the official recommendations of the Chinese Community Guidelines. A total of 48 BABL/c nude male mice (18 ± 2 g) raised under specific pathogen-free conditions were purchased from Shanghai Slac Laboratory Animal Co. Ltd (Shanghai, China; animal license number SCXK (Shanghai) 2008-0016); they were fed in the experimental animal room of Putuo Hospital affiliated to Shanghai University of Traditional Chinese Medicine (Shanghai, China, animal laboratory license number, SCHK (Shanghai) 2007-0005). Sixteen animals were randomly allocated into 2 groups prior to the experiment. Human HCT-116 and HT-29 colorectal cancer cells were used to establish xenografts. Cells were re-suspended at a density of 1 × 10 7 cells/ml, and the resulting suspensions (0.1 ml/10 g body weight) were injected subcutaneously. After 12 days, tumor nodules were palpable. On day 21, the mice were killed by cervical dislocation, and the tumors were removed, fixed in 4% paraformaldehyde (pH 7.2) and embedded in paraffin. Tumor sections were then stained with CD31 in order to observe tumor angiogenesis, and the number of vessels was enumerated in each of ten high-power fields. In addition, PAS staining was also performed to visualize channels formed by VM. The density of PAS-positive vessel networks was estimated using a photomicroscope (Leica, Germany), and for each tumor, 5 microscopic fields per slide in each of 5 slides were analyzed. Finally, slides were also stained for laminin, a component of the basement membrane of VM channels.

CD31/PAS double staining.
For CD31/PAS double staining, CD31(Cell Signaling, USA, dilution 1:2000) immunohistochemical staining was performed first. Specifically, after application of the diaminobenzidine chromagen, slides were rinsed with distilled water, treated with 0.5% periodic acid solution for 10 min, and rinsed again with distilled water for 2-3 min. Next, the slides were incubated with Schiff solution for 15-30 mins in the dark, rinsed with distilled water, and finally counterstained with hematoxylin.
Three-dimensional cultures. The formation of vascular channels was assessed using in vitro 3-D cultures composed of rat tail collagen type I. First, the 6-well plates were pretreated with a certain percentage of collagen for 30 minutes. Then, 4 × 10 5 tumor cells were plated onto the surface of the collagen and incubated at 37 °C. Additionally, for in-gel methods, the tumor cells were mixed with Matrigel, which was then allowed to polymerize. Filtered CoCl 2 was added to the medium in the hypoxic group 48 h after polymerization.
Transfections were performed using Lipofectamine ™ 2000 reagent (Invitrogen, USA) according to the manufacturer's protocol, and the cells were incubated at 37 °C for 48 hours before analysis. Then, the expression of HIF-1α mRNA was detected by quantitative PCR, and protein expression was analyzed by western blot.
Western blot analysis. Cells were lysed in 10 volumes (w/v) of lysis buffer. After centrifugation, the supernatant was collected, and the protein concentrations of each sample were quantified. Equal amounts of total protein were separated by 12% SDS-PAGE and then transferred to a PVDF membrane. The membrane was then Scientific RepoRts | 6:37534 | DOI: 10.1038/srep37534 incubated overnight at 4 °C with primary rabbit anti-human monoclonal antibodies to HIF-1α (Abcam, UK; dilution 1:500), E-cd (Cell Signaling, USA, dilution 1:1000), Claudin-4 (Santa Cruz, USA, dilution 1:1000), and vimentin (Afinity, USA; dilution 1:1000), as well as rabbit anti-human polyclonal antibodies to β -actin (Bioworld, USA; 1:500). After incubation with an HRP-conjugated secondary antibody at room temperature for 2 h, protein bands were visualized using enhanced chemiluminescence (ECL) and detected using the Bio-Imaging System. Immunohistochemistry (IHC). Tumor tissues were fixed in 4% paraformaldehyde for 48 hours, embedded in paraffin and sectioned into 4 μ m slides. Tissue sections were then processed by de-paraffinization, rehydration through an alcohol gradient, peroxidase clearing, antigen retrieval and blocking, antibody binding, DAB staining, washing with distilled water, hematoxylin staining, niacin alcohol differentiation, dilute ammonia bluing, incremental graded alcohol dehydration, rinsing in xylenes, and mounting with a conventional resin. Slides were then incubated overnight at 4 °C with the following primary antibodies: rabbit-anti-human HIF-1α monoclonal antibody (Abcam, Cambridge, MA, USA; dilution 1:50), rabbit-anti-human E-cadherin monoclonal antibody (Cell Signaling, USA, dilution 1:100), Claudin-4 (Santa Cruz, USA, dilution 1:100), and vimentin (Afinity, USA; dilution 1:100). Secondary biotin-conjugated antibodies were used at 1:200. Positive tan staining for CD31 and E-cadherin was visualized with a light yellow or tan substrate and cell nuclei were counterstained  In vitro migration and invasion assays. HCT-116 and HT-29 cell invasion was assayed using 24-well Transwells with 8-μ m-pore polycarbonate membrane inserts (Corning, NY, USA) and Matrigel (BD Biosciences) according to the manufacturer's instructions. Briefly, 20 μ l of Matrigel (1:5 dilution) was added to each insert. Then, 100 μ l of a cell suspension containing 3 × 10 5 cells was transferred to the upper chamber and allowed to incubate for 24 h. The filters were then stained with hematoxylin, and cells that appeared on the lower surface of the filter were counted in five random high-magnification microscope fields. Each experiment was performed in triplicate.
Wound healing assay. Cell motility was assessed using a scratch assay. The width of the wound was measured after 12 and 24 hours, and the speed of wound closure was determined at each time point by calculating the wound width relative to that at 0 hours. Each experiment was performed in triplicate.

Statistical analysis.
Results were expressed as the means ± SD, and SPSS 18.0 was used for all statistical analyses. qRT-PCR data were analyzed using the paired Student's t-test, whereas statistical analysis of other results was determined using one-way ANOVA. P < 0.05 was considered statistically significant.

In vitro VM capacity of colorectal cancer cells is associated with invasiveness, migration and
motility. The formation of vascular channels indicative of VM was tested using a 3-D culture system composed of rat tail collagen type I 4 . Two human CRC cell lines with different metastatic potentials were used, including a highly-metastatic (HCT-116) and a poorly-metastatic cell line (HT-29) 14 . HCT-116 cell VM was first observed at 24 h after seeding, and channel formation was completed by 6 days. However, HT-29 cells did not form vascular channels by VM (Fig. 1A). Next, cell invasion was assayed using Transwell chambers and serum-free media without the addition of a chemoattractant. After 24 h, a significant number of HCT-116, but not HT-29, cells had invaded the Matrigel, (Fig. 1B). Furthermore, as shown in Fig. 1C, quantitative analysis of a wound healing assay revealed significant differences between the wound healing speeds of HCT-116 and HT-29 cells.
Taken together, these results indicate that VM-positive HCT-116 cells were highly invasive, exhibiting both high cell motility and migration, while VM-negative HT-29 cells were less invasive. These results show a correlation between in vitro VM capacity and metastatic potential in CRC cells.  cells were incubated for 0-8 h in either normoxic or hypoxic conditions. To determine the expression levels of HIF-1α , we analyzed the levels of HIF-1α protein in two CRC cell lines using western blotting. Both cell lines showed a significant increase in HIF-1α protein expression under hypoxic conditions ( Fig. 2A). We then detected the mRNA expression levels of HIF-1α and several EMT markers in a panel of CRC cell lines using qRT-PCR. As shown in Fig. 2B, HCT-116 and HCT8 cells exhibited high expression levels of HIF-1α , Fn1 and vimentin (Vim), whereas HT-29 and LS174T cells exhibited low expression levels of these markers. However, the reverse was true for the markers TGF-β 1, Claudin-4 and E-cd. Specifically, we found that HCT-116 and HCT8 cells exhibited low expression levels of these three markers, whereas HT-29 and LS174T, exhibited high levels. Moreover, Vim expression levels in HCT-116 and HCT8 cells were higher than those in HT-29 and LS174T cells. The trends for TGF-β 1, Claudin-4, Vim and E-cd protein expression in HCT-116, HCT8, HT-29 and LS174T cells agreed with their respective levels of mRNA expression (Fig. 2C).

Analysis of HIF-1α and EMT expression in VM
EMT regulated in VM formation. In order to demonstrate the EMT regulated VM formation in CRC, SB431542 and U0126EtOH, which can inhibit of EMT were used to treat HCT-116 and HCT-8 in these experiments. Protein expression levels of EMT markers E-cd, Claudin-4 and Vim in CRC cells were detected by western blot. As shown in Fig. 3A, both of the inhibitor had significant effect on EMT markers. What's more, the formations of VM in CRC cells were distinctly weakened after treated with EMT inhibitor (Fig. 3B). That is to say, EMT regulated VM formation in CRC. HIF-1α induced changes in EMT will be illustrated in the follow experiments.

HIF-1α induced changes in EMT gene expression levels and altered VM in HCT-116 cells.
To further investigate the role of HIF-1α in the formation of vascular channels under hypoxic conditions, an shRNA construct (herein referred to as shRNA-HIF-1α ) was used to knock down HIF-1α in HCT-116 cells. As shown in Fig. 4A, this construct significantly reduced HIF-1α expression. We also observed significant changes in the mRNA and protein levels of the EMT markers E-cd, Claudin-4, Vim and Fn1 following HIF-1α knock down (Fig. 4B). Furthermore, as expected, shRNA-HIF-1α significantly reduced the VM capacity of HCT-116 cells under hypoxic conditions relative to control cells (Fig. 4C). Taken together, these results indicate that hypoxia-induced VM is dependent upon HIF-1α expression.

VM is associated with HIF-1α and EMT markers in CRC xenograft tumors.
In vivo data showed that HCT-116, but not HT-29 cells, formed vascular channels (Fig. 5A). In addition, the formation of these channels was associated with increased expression of HIF-1α and vimentin, as well as reduced expression of E-cadherin and Claudin-4 in CRC xenograft tumors. Moreover, knock down of HIF-1α in HCT-116 tumor xenografts prevented VM and reduced gene expression changes implicated in the process of EMT ( Fig. 5C and D).

Discussion
Vasculogenic mimicry (VM) is the phenomenon where cancer cells mimic endothelial cells by forming blood vessels 15 . Tumor blood vessel formation is caused not only by angiogenesis dependent on host endothelial cells but also by VM, which plays an equally substantial role in this process [16][17][18] . VM, which refers to a process whereby tumor cells directly line up to form blood vessels, was first reported in melanoma by Hendrix and colleagues in 1999 4 . It was described as the unique ability of highly aggressive tumor cells to differentiate into multiple cellular phenotypes, obtain endothelial-like characteristics, and form vessel-like structures to obtain blood supply either actively or passively. Hence, it is suggested that the presence of tumor cell-lined channels may provide a potential new route by which metastases may leave the tumor to reach distant sites, using the exchange flow between the VM channels and normal vascular system. These findings explain why a class of drugs once heralded as a game-changer in cancer treatment was less effective than hoped 19,20 . This study was not the first in which researchers suggested that cancer cells could make their own blood vessels.
The induction of EMT in human mammary epithelial cells results in the acquisition of mesenchymal traits and the expression of stem cell markers 21 . Previous data show that cancer cell lines underwent morphological EMT-like changes (more fibroblastoid morphology and loss of cellular cohesiveness) under hypoxic conditions. Hypoxia led to an increase in migration, invasion and is critical for cell plasticity and VM formation 22 . Hypoxia (or induction of HIF-1α ) mediates the behavior of important factors within the tumor microenvironment that induce EMT and promote a stem-like phenotype 23 . EMT is the main process that promotes VM in cancer, and VM has been associated with metastasis in CRC and reduced survival 24 . In this study, we provide evidence that VM is associated with HIF-1α expression, alterations in the levels of EMT markers, and metastasis in CRC. In addition, the use of shRNA-HIF-1α showed that the down-regulation of HIF-1α reduced VM in HCT-116 cells and the expression of the EMT markers E-cadherin, Claudin-4, Vim and Fn1. Concomitantly, our in vivo data support that VM capacity is significantly impacted by HIF-1α expression and EMT. Taken together, our data imply that hypoxia may directly cause VM by inducing EMT.
To our knowledge, this is the first report discussing the relationship between EMT, HIF-1α and VM using VM-positive CRC cells as a model. HIF-1α plays an important role in the development of VM, but further studies are needed to support this result. Most of the research conducted regarding EMT focuses on the invasion and metastasis of cancer cells. Taken together, our results indicate that the combined use of angiogenesis-and VM-targeting therapies may be an effective cancer treatment. Identification of individual EMT molecules that regulate VM in CRC is the focus of our ongoing work.