SOX7 blocks vasculogenic mimicry in oral squamous cell carcinoma

Background : Vasculogenic mimicry (VM) is the formation of an alternative circulatory system by aggressive tumor cells. The characteristics of VM and its underlying mechanism in oral squamous cell carcinoma (OSCC) remain unclear. This study aims to determine the relationship between VM channels in OSCC tissues and clinical outcomes and to investigate the biological role of SOX7 in VM in OSCC cells. Methods : CD31/PAS double staining was performed to evaluate VM status in OSCC tissue. The relationships between VM and clinicopathological variables, and VM and SOX7 levels were analyzed. VM channel formation was assay performed to observe VM channels in the OSCC cell lines. To investigate the role of SOX7 in VM channel formation, SOX7 was transiently over-expressed in SCC-9 cells. VM-modulating genes were identied by Western blotting. Results : We conrmed the presence of VM channels in OSCC tissue and several cell lines and found a positive correlation between VM and lymph node metastasis and patient survival in OSCC ( P = 0.003). We also found that the presence of VM channels in OSCC tissue was inversely associated with SOX7 expression ( P = 0.020). We observed that overexpression of SOX7 impaired VM channel formation by reducing the expression of VE-cadherin, thereby inhibiting cell migration and invasion. Conclusion : These results suggest that SOX7 plays an important role in the regulation of VM channel formation and may inhibit OSCC metastasis. followed and Endothelial cells were used as and the was omitted as the negative control. node marker in also provide that the of channels in OSCC tissue is inversely associated with SOX7 expression and that SOX7 restrains in vitro VM channel formation by inhibiting VE-cadherin expression. To the of our knowledge, this study is the rst to demonstrate that SOX7 plays an important role in VM and consequently contributes to OSCC suppression. Therefore, further investigation of SOX7 as a potential therapeutic strategy for inhibiting VM channel formation in OSCC is necessary in the future. Y-box VM: Vasculogenic Oral squamous cell PAS: acid–Schiff; VE-cadherin: Vascular endothelial cadherin; EphA2: Erythropoietin-producing hepatocellular receptor A2; PTK: Protein tyrosine kinases; EMT: Epithelial–mesenchymal transition; MMPs: Matrix metalloproteinases.


Abstract
Background : Vasculogenic mimicry (VM) is the formation of an alternative circulatory system by aggressive tumor cells. The characteristics of VM and its underlying mechanism in oral squamous cell carcinoma (OSCC) remain unclear. This study aims to determine the relationship between VM channels in OSCC tissues and clinical outcomes and to investigate the biological role of SOX7 in VM in OSCC cells.
Methods : CD31/PAS double staining was performed to evaluate VM status in OSCC tissue. The relationships between VM and clinicopathological variables, and VM and SOX7 levels were analyzed. VM channel formation was assay performed to observe VM channels in the OSCC cell lines. To investigate the role of SOX7 in VM channel formation, SOX7 was transiently over-expressed in SCC-9 cells. VMmodulating genes were identi ed by Western blotting. Results : We con rmed the presence of VM channels in OSCC tissue and several cell lines and found a positive correlation between VM and lymph node metastasis and patient survival in OSCC ( P = 0.003). We also found that the presence of VM channels in OSCC tissue was inversely associated with SOX7 expression ( P = 0.020). We observed that overexpression of SOX7 impaired VM channel formation by reducing the expression of VE-cadherin, thereby inhibiting cell migration and invasion. Conclusion : These results suggest that SOX7 plays an important role in the regulation of VM channel formation and may inhibit OSCC metastasis.

Background
The formation of new blood vessels in solid tumors is essential for their growth and metastasis. Without the oxygen and nutrients provided by additional blood vessels, tumors undergo necrotic cell death due to ischemia and hypoxia [1,2]. Conventional anti-angiogenic therapy, which targets vascular endothelial cells to inhibit tumor blood vessel formation, has proven unsatisfactory [2,3]. The failure of such therapy suggests that tumors have an alternative strategy beyond angiogenesis for attaining a blood supply adequate to sustain growth.
A new interpretation of previous ndings was reported by Maniotis et al., who observed highly patterned vessel-like channels in human melanomas in which red blood cells were detected, independent of endothelium-dependent angiogenesis [4]. Formed by the process of vasculogenic mimicry (VM), these structures consist of periodic acid-Schiff (PAS) + and CD31-negative basement membranes lined by tumor cells rather than endothelial cells [5]. VM has been vividly described in numerous types of malignant tumors and is closely associated with a poor clinical outcome [6][7][8]. Experimental evidence has con rmed a relationship between VM and tumor metastasis in a variety of cancer cell lines and has shown that tumor cells can form VM channels in Matrigel culture [9][10][11]. The tumor cells lining VM networks express vascular endothelial cadherin (VE-cadherin) [12], erythropoietin-producing hepatocellular receptor A2 (EphA2) [13], and matrix metalloproteinases [14], the expression of which correlates with tumor metastasis and invasion [15]. Although many studies have contributed new insights into VM and its underlying molecular pathways [3,16], the features of VM and its underlying mechanism in oral squamous cell carcinoma (OSCC) are still unclear.
We recently reported that the expression of SOX7, a member of the sex-determining region Y-box (SOX) family, is associated with advanced TMN stage, lymph node metastasis, and poor overall survival in patients with OSCC [17]. Additionally, several studies have reported a tumor-inhibitory function of SOX7 associated with cancer regulatory pathways [18,19]. The effect of SOX7 on VM channel formation and the regulation of VM-related genes has not been studied in any cancer.
The present study investigates whether VM channels are present in human OSCC tissues and cell lines.
The relationship between VM channels and clinical outcomes in OSCC tissues is examined, and the biological effect of SOX7 on VM channel formation in OSCC cells is explored.

Methods
Patients and tissue specimens Specimens were retrieved from 46 patients with OSCC who were surgically treated at the Department of Oral and Maxillofacial Surgery at Seoul National University Dental Hospital between 1999 and 2005.
Clinical data including age, sex, recurrence, and survival were collected from patient medical records.
Tumors were graded based on the World Health Organization (WHO) Classi cation of Tumors [20] and staged according to the TNM system recommended by the American Joint Committee on Cancer [21].
The clinicopathological features of the OSCC patients are summarized in Table 1. This study was approved by the Institutional Review Board (IRB) of Seoul National University Dental Hospital (IRB number: ERI16008).

CD31/ Periodic acid-Schiff (PAS) double staining
Formalin-xed, para n-embedded specimens were sectioned to a thickness of 4 μm. Tissue sections were depara nized in xylene and rehydrated in graded alcohol solutions. Endogenous peroxidase was inactivated by incubation in 3% H 2 O 2 solution for 5 min. Heat-induced epitope retrieval was performed in citrate buffer (pH 6.0) for 10 min using a microwave oven. The sections were incubated overnight at 4°C with mouse monoclonal anti-CD31 (clone JC/70A; 1:100; Thermo Fisher Scienti c, Rockford, IL, USA).
The immunohistochemical reaction was performed using REAL EnVision/HRP (Dako, Glostrup, Denmark) and visualized by diaminobenzidine. Sections were incubated for 5 min in 0.5% periodic acid solution followed by 15 min in Schiff reagent. Slides were washed for 5 min in lukewarm water, counterstained with Mayer hematoxylin, dehydrated, and mounted. Endothelial cells were used as an internal positive control for CD31, and the primary antibody was omitted as the negative control.
Identi cation of vasculogenic mimicry CD31/PAS double staining was evaluated blindly by two oral pathologists (KYO and SDH). Based on recent reviews of VM [22,23], VM + was de ned as the presence of CD31 − /PAS + vessel-like structures surrounded by tumor cells on the non-luminal side. Although red blood cells may be observed in the lumen, the absence of endothelial cells was con rmed by the absence of CD31 expression on the luminal side. Microscopic elds with prominent in ammatory in ltrates, necrosis, and hemorrhage were not included for VM evaluation. According to these criteria, all 46 OSCC cases were classi ed as VM + or VM − .

Western Blot Analysis
Cell lysates were prepared by homogenization with RIPA buffer (EMD Millipore, Billerica, CA, USA) according to the protocol supplied. Protein concentrations was measured using the DC Protein Assay Kit (Bio-Rad Laboratories, Hercules, CA, USA). After normalization, equal amounts of protein were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to immunoblot polyvinylidene di uoride membranes (Pall Corporation, Port Washington, NY, USA). The membrane was blocked with 5% skim milk at room temperature for 2 h, incubated with speci c primary antibodies, and probed with the corresponding horseradish peroxidase-conjugated secondary antibodies (GTX213110 for anti-Rabbit and GTX213111 for anti-mouse). Antibodies against SOX7 (SC-20093), VE-cadherin (SC-9989), and α-tubulin (SC-5286) were obtained from Santa Cruz Biotechnology. Rabbit anti-human monoclonal antibody against EphA2 (#6697) was obtained from Cell Signaling Technology, Inc (Danvers, MA, USA). The immunoreactive proteins were detected by SuperSignal West Pico Chemiluminescent Substrate (SC-2048; Santa Cruz Biotechnology), and immunoreactive bands were visualized using ImageQuant LAS 500 (GE Healthcare Life Sciences, Pittsburgh, PA, USA). Densitometric analysis of western blots was performed using ImageJ software (Version 1.51k, NIH, Bethesda, MD, USA).
Three-dimensional culturing for VM channel formation assay A 24 well plate was coated with 150 µL of growth factor reduced Matrigel (BD Bioscience, Bedford, MA, USA), and the cell suspension was seeded onto the solid Matrigel at a concentration of 1.5 × 10 5 cells with serum-free medium. After incubation for 24 h, cells were stained with PAS (Thermo Fisher Scienti c) to label the basement membrane of tubular structures. Five random elds of view for channels and intersections were photographed at 100× magni cation using an inverted microscope (Nikon Ti, Nikon Instruments Inc., Melville, USA) and counted using Image J software. Network Intersection was de ned as a junction between 3 neighbors [24].

Statistical analysis
Pearson chi-square test and Fisher's exact test were performed to evaluate the association between VM status and clinicopathological features or SOX7 expression. Cumulative survival curves were generated using the Kaplan-Meier method, and the difference between the curves was analyzed using the log-rank test. All in vitro data are expressed as the mean ± standard deviation (SD). The statistical comparisons between groups were evaluated using Student's t-test for two groups. P < 0.05 was considered statistically signi cant. All statistical analyses were performed using IBM SPSS Statistics version 25 (IBM, Armonk, NY, USA).

Results
VM was associated with poor prognosis in OSCC patients VM was identi ed in 23.9% (11/46) of the OSCC specimens studied. PAS + tubular structures were lined by endothelial-like tumor cells with CD31 − staining ( Figures 1A and B). Vascular-like structures were observed in 3 OSCC cell lines (HN22, OSC20, and YD10B), and PAS + staining indicating VM channel formation was observed ( Figure 1C). VM was signi cantly associated with lymph node metastasis (P = 0.003) ( Table 2). The 5-year survival rate was 47.8% (22/46) in patients with OSCC, and the VM + group tended to have lower overall survival (P = 0.073) ( Figure 1D).

VM correlated with absence of SOX7 expression in OSCC
Of the 46 OSCC specimens, 14 (30.4%) were classi ed as SOX7 + , and all were VM − (Figure 2). All VM + exhibited no nuclear staining for SOX7, indicating a strong association between the positive VM status and the absence of SOX7 expression in OSCC (P = 0.020) ( Table 3).

SOX7 impaired VM channel formation in vitro
To assess the in uence of SOX7 expression on VM channel formation in vitro, SCC-9 cells were transiently transfected with a SOX7 overexpression vector (Figures 3A and 3B). SOX7 and control (transfected with vehicle vector) cells were then cultured on Matrigel for 24 h. SCC-9 cells transfected with SOX7 had fewer channels than did control cells and did not form extensive channel networks ( Figure 3C).
These results were con rmed by PAS staining of VM networks ( Figure 3C) and quanti cation of channel formation and the number of channel intersections indicated a decrease in VM channel formation in SOX7-overexpressing cells ( Figure 3D). VE-cadherin and EphA2 have been identi ed as major proteins involved in VM channel formation [12]. While the expression of EphA2 did not change signi cantly, we found that only VE-cadherin expression was reduced in SOX7-overexpressing cells (Figures 3E and 3F).

SOX7 inhibited the migration and invasion of SCC-9 cells
VM channel formation is thought to be related to tumor cell migration and invasion [5]. To investigate the function of SOX7 expression in the migration and invasion ability of SCC-9 cells, we performed transwell migration and invasion assays in SCC-9 cells. As expected, SOX7 overexpression strongly inhibited the migration and invasion capacity of SCC-9 cells (Figure 4).

Discussion
Avascular tumors (>1-2 mm 3 ) cannot be su ciently supplied with oxygen and nutrients and cannot rely entirely on host endothelial cells for angiogenesis [25]. Therefore, tumors require a system independent of angiogenesis, such as VM. A three-stage phenomenon underlies VM: 1) highly malignant tumor cell plasticity, 2) remodeling of the extracellular matrix (ECM), and 3) connection of VM channels and host blood vessels to obtain a blood supply from the host tissue [26]. Since the rst description of VM in uveal melanoma, a CD31 − /PAS + pro le has been recognized as the hallmark of VM [10,24,27]. In this study, we observed small vessel-like structures with a lumen lined with tumor cells in OSCC tissue (Figures 1A and  1B). VM channels were also detected in several OSCC cell lines using Matrigel culture system ( Figure 1C). In addition, the occurrence of VM was positively associated with lymph node metastasis ( Table 2), and Kaplan-Meier survival analysis showed that VM + OSCC patients had a worse overall survival rate than did VM − patients ( Figure 1D). Our present ndings are consistent with a recent meta-analysis reporting that VM is associated with lymph node metastasis and overall survival in patients with squamous cell carcinoma of the head and neck [28]. These observations strongly suggest that VM may be used as an independent predictor of prognosis in OSCC.
We previously demonstrated that SOX7 expression was lower in OSCC tissue samples than in corresponding normal oral mucosal tissues and that the absence of SOX7 expression was signi cantly associated with advanced TNM stage, incidence of lymph node metastasis, and poor prognosis [17]. SOX7 expression was found to be frequently down-regulated in a variety of tumors; its forced expression inhibited cellular proliferation, migration, and invasion, suggesting that SOX7 functions as a tumor suppressor [29][30][31]. Because both VM and SOX7 are closely associated with OSCC prognosis, we hypothesized that SOX7 might play an essential role in VM channel formation in OSCC. In the present study, we observed VM channels in 34.4% of OSCC in the absence of SOX7 expression; SOX7 expression correlated negatively with VM status (Figure 2 and Table 3). In vitro experiments also showed that overexpression of SOX7 severely disrupted the formation of VM channels in SCC-9 cells in a Matrigel culture system ( Figure 3C and 3D). These results further support the hypothesis that SOX7 regulates VM channel formation in OSCC. To the best of our knowledge, this is the rst report elucidating the role of SOX7 in VM in OSCC.

Given that tumor cells can be transformed into endothelial-like cells (mesenchymal cells), many studies
have proposed that the epithelial-mesenchymal transition (EMT) is vital for VM [32][33][34][35]. A recent study reports that the EMT-related transcription factors Twist1 and Snail were up-regulated in VM + hepatocellular carcinoma samples and that their expression correlated with VM channel formation [33,36]. In contrast, our results show that VM status was not associated with the expression of Snail or Twist in OSCC (data not shown), indicating that other factors are involved in VM in OSCC. VM-associated molecules and signaling pathways have been investigated in numerous types of highly aggressive malignant tumors and have served as a strategic roadmap for drug development [15]. VE-cadherin, a transmembrane protein in the cadherin family, is expressed exclusively by endothelial cells [37]. Expressed in highly aggressive melanoma cells, its down-regulation was observed to abolish VM, indicating its involvement in VM structure formation [38]. VE-cadherin also mediates the activities of EphA2, a member of the ephrin-receptor family of PTKs, which is crucial for angiogenesis [39]. Mitra et al., recently found that phosphorylation of EphA2 was signi cantly associated with the occurrence of VM in invasive breast carcinoma [40]. Co-localization of VE-cadherin with EphA2 is reported to induce the activation of MMPs, resulting in matrix remodeling and VM channel formation [12,15,41]. These ndings suggest that VE-cadherin, EphA2, and MMPs are important VM-modulating genes. In this study, we observed that SOX7 overexpression signi cantly attenuated the protein expression of VE-cadherin in SCC-9 cells but did not affect VM-modulating genes such as EphaA2 and MMP-2 and -9 ( Figure 2E; Supplementary Figure 1). These results suggest that the molecule responsible for the observed regulatory effects of SOX7 on VM channel formation in OSCC is VE-cadherin.

Page 8/19
Our ndings show that the occurrence of VM correlated positively with lymph node metastasis and may be a promising marker of patient prognosis in OSCC. We also provide evidence that the presence of VM channels in OSCC tissue is inversely associated with SOX7 expression and that SOX7 restrains in vitro VM channel formation by inhibiting VE-cadherin expression. To the best of our knowledge, this study is the rst to demonstrate that SOX7 plays an important role in VM and consequently contributes to OSCC suppression. Therefore, further investigation of SOX7 as a potential therapeutic strategy for inhibiting VM channel formation in OSCC is necessary in the future.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Competing interests
The authors declare that they have no competing interests.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. AdditionalFile1.docx