Vasohibin 2 Promotes Lymphangiogenesis of Lung Squamous Cell Carcinoma Through Snail- dependent VEGF-D Signaling Pathway

Pengpeng Liu Tianjin Medical University Cancer Institute and Hospital: Tianjin Tumor Hospital Rui Zhang Tianjin Medical University Cancer Institute and Hospital: Tianjin Tumor Hospital Lei Han Tianjin Medical University Cancer Institute and Hospital: Tianjin Tumor Hospital Xiao Zhang Tianjin Medical University Cancer Institute and Hospital: Tianjin Tumor Hospital Yingnan Ye Tianjin Medical University Cancer Institute and Hospital: Tianjin Tumor Hospital Wenwen Yu Tianjin Medical University Cancer Institute and Hospital: Tianjin Tumor Hospital Xiubao Ren Tianjin Medical University Cancer Institute and Hospital: Tianjin Tumor Hospital Weijia Zhang Icahn School of Medicine at Mount Sinai Jinpu Yu (  jyu@tmu.edu.cn ) Tianjin Medical University Cancer Institute and Hospital: Tianjin Tumor Hospital

Invasion and metastasis are a complex dynamic process involved multiple factors, which mainly occurs through blood vessels and lymphatic vessels. In LUSC, some studies have illuminated lymphangiogenesis could enhance the a nity of tumor cells to lymphatic vessels, further promoting invasion and metastasis through lymphatic vessels [5]. However, the mechanisms of lymphangiogenesis in LUSC development and progression are still unclear. Considering LUSC metastasis initiated from invading into lymph nodes through lymphatic vessels [6,7], speci c interdiction of lymphangiogenesis would provide more e cient inhibition on LUSC invasion compared to routine anti-angiogenesis treatment. Therefore, to discriminate the key biomarkers regulating the process of lymphangiogenesis in LUSC will throw a light on developing novel anti-cancer therapeutic strategies for LUSC [8][9][10].
In previous study, we analyzed mRNA expression pro les of 937 primary LUSC samples from GEO and TCGA databases and ltered out VASH2 as one of the clinical bene cial predictable biomarkers associated with LUSC poor prognosis. The gene for human VASH2 is located on chromosome 1q32.3.
The full-length human VASH2 is composed of 355 amino acid residues [11]. It was initially reported to be expressed in endothelial cells. VASH2 is mainly expressed in monocytes (MNCs), which are mobilized from bone marrow and in ltrated into the front of germination [12].
VASH2 has been demonstrated to be an angiogenic factor in a variety of tumors, including liver cancer, gastric cancer, breast cancer, and ovarian cancer [13][14][15], and plays an important role in tumor development [16]. However, none study on the expression and function of VASH2 in lung cancer, especially in LUSC has ever been reported. It has been reported that exogenous VASH2 signi cantly promoted tumor growth, increased microvessel density and hemoglobin concentration [17], while tumor angiogenesis was signi cantly inhibited in VASH2 knockout mice [18]. Furthermore, VASH2 was reported to trigger epithelial-mesenchymal transition (EMT), up-regulate chemotherapeutic resistance, and increase the proportion of cancer stem like cells [19,20]which can generate cellular traits associated with more aggressive malignancy with higher motility and invasiveness by imparting the trait of self-renewal to cancer cells. However, whether VASH2 promotes invasion and metastasis of lung cancer is related to angiogenesis or EMT has not been disclosed.
In this study, we explored the clinical signi cance of VASH2 in LUSC, and potential molecular mechanisms involved in VASH2 promoting the development and progression of LUSC. We demonstrated that high level of VASH2 was signi cantly associated with poor prognosis and promoted lymphatic metastasis in patients with LUSC. It was con rmed from cell experiments and animal models that overexpression of VASH2 promoted angiogenesis and lymphangiogenesis, and accelerated the invasion and metastasis of tumor cells by stimulating EMT. The expression of VASH2 can signi cantly increase the production of Snail and VEGF-D in LUSC cells. And blocking VASH2/VEGF-D signaling pathway could signi cantly inhibited the growth of tumor and the formation of lymphatics. In summary, we not only revealed VASH2 as a marker of LUSC, but also proposed new molecular mechanisms of lymphatic metastasis, which provided potential molecular diagnosis and targeted treatment strategies for patients with LUSC.
Mice BALB/c female mice between 6-8-week-old were purchased from Beijing SPF Biotechnology Co., Ltd. Mice were maintained at SPF animal laboratory of Tianjin Medical University Cancer Hospital and Institute. To establish the lung adenocarcinoma mouse models, the mouse lung adenocarcinoma cells H520 CTRL and H520 OV-VASH2 (1×10 6 cells in 100ul PBS) was injected into the mammary fat pads of BALB/c mice to induce tumor formation [1]. All procedures that involve mice were approved by the Ethics Committee for Animal Experiments at the Tianjin Medical University Cancer Hospital and Institute, and were performed in accordance with the Guide for the Care and Use of Laboratory Animals.

Cell Lines
Human lung adenocarcinoma cell line H520 was obtained from Chinese Academy of Medical Sciences tumor cell libraries.

Cell line culture
The human LUSC cell line H520 was cultured in RPMI 1640 complete medium containing 10% fetal bovine serum in at 37℃, 5% CO2. By transfecting wild-type H520 cells (H520 CTRL ) with VASH2-speci c short hairpin RNA (shRNA) lentiviral vectors, H520 OV-VASH2 cells were established as indicated in our previous work. Stably transfected cells were cloned by limiting dilution and validated using RT-PCR and Western blot as our previous work described.

Clinical samples
This study recruited 64 cases of LUSC patients treated with partial lung resection surgery at the Department of Hepatobiliary Oncology of the Tianjin Medical University Cancer Institute and Hospital from November 2004 to November 2006. These patients included 49 males and 19 females with a median age of 62.7 years. No prior treatments, including chemotherapy or radiotherapy, were conducted before lung resection surgery was performed. Postoperative follow-up time was 40-65 months. This project was approved by the Ethics Committee of Tianjin Medical University. All experiments were performed in accordance with the principles of the Declaration of Helsinki. Written consents were obtained from the patients. Tissue samples were ash frozen and stored at 80℃.

Data Processing for Public Datasets
Public datasets were downloaded from the GEO and TCGA database. 10 studies of LUSC expression pro ling with a minimum of 24 LUSC patients were collected from GEO database for the meta-analysis study, including 41 dataset from GSE50081, 78 dataset from GSE41271, 55 dataset from GSE30219, 24 dataset from GSE19188, 72 dataset from GSE8894, 43 dataset from GSE42127, 53 dataset from GSE3141, 52 dataset from GSE14814, 66 dataset from GSE37745, 38 dataset from Tianjin database and 415 dataset from TCGA. Detailed information for these datasets is listed in Figure S1.
The work ow for the meta-analysis of the discovery set of 1 Tianjin and 10 public LUSC datasets was depicted in Figure S1. Meta genes associated with short survival (<5 years: 5 years was the median survival of 11 cohorts) and long survival ( >5 years) were identi ed by a meta p-value of less than 0.05 by combining p values from 11 datasets and individual p value <0.05 in at least 4 out of 11 datasets from the 11 datasets, . The meta genes were subjected to Gene Ontology (GO) and canonical pathway enrichment with multiple pathways databases (KEGG, Reactome, PID, Wiki Pathways, Biocarta and Panther).

Immunohistochemistry
All samples were heated for 0.5 h at 56°C, depara nized in xylene and rehydrated through graded alcohol. Antigens were retrieved by heating in citrate buffer (pH 6.0) for 20 min. Endogenous peroxidase activity was quenched in a bath of methanol and hydrogen peroxide for 30 min. The samples were then incubated overnight at 4°C with mouse antibody, and a biotinylated secondary goat anti-mouse or rabbit IgG antibody labeled with streptavidin-horseradish peroxidase (HRP) using a DAB staining kit according to the manufacturer's instructions. Five representative high-power elds (400× magni cation) were selected for each tissue section for histological evaluation. For each protein, two parameters, namely, positive rate (PR) and staining intensity (SI), were used to describe the expression on the basis of the extensity and intensity of the positively stained cells in the samples. PR denotes the percentage of positively stained cells in cancer tissues: ≤15%, negative (scored as 0); 16%-50%, positive at low frequency (scored as 1); 51%-80%, positive at medium frequency (scored as 2); and ≥80%, positive at high frequency (scored as 3). SI refers to the ranked staining intensity of positively stained cells in LUSC samples. This value ranged from 0 to 3, which corresponded to negative, weakly positive, moderately positive, and strongly positive. The sum represented the nal scores of each protein marker in the samples because protein expression was comprehensively evaluated on the basis of both parameters. A nal score of <4 was de ned as low/negative expression and a nal score of ≥4 was de ned as high expression.

Lentivirus construction
The vector pHBLV-CMV-MCS-3FLAG-EF1-ZsGreen-T2A-PURO was digested at 37℃ and the gel was recovered. Then the fragments were recovered by PCR. The target fragment was connected with the vector and the reaction system(20ul). 100ulcompetent cell suspension was taken from -70℃ refrigerator and thawed at room temperature. Add plasmid DNA solution, shake gently, place on ice for 30 minutes, heat shock in 42℃ water bath for 90 seconds and quickly place on ice to cool for 3-5 minutes after heat shock. 1 ml LB liquid medium (without antibiotics) was added into the tube, and then it was stirred at 37℃ for 1 hour in a 220 RPM shaker. The bacteria were restored to normal growth state and expressed antibiotic resistance gene encoded by plasmid. After shaking the above bacterial solution, centrifugation was performed to remove 900ul supernatant. After the remaining culture medium was sucked and mixed, 100ul was applied to the screening plate containing antibiotics. The plate was placed upside down for half an hour. After the bacterial solution was completely absorbed by the medium, the plate was inverted and cultured at 37℃ for 16-24 hours. After the transformation, the plates were picked up and shaken for 14 hours at 230 RPM / min at 37℃. The positive clones were sent to sequencing company for sequencing for construction of VAHS2 and Snail overexpression lentivirus. The synthesized VASH2 shRNA and Snail shRNA were inserted into plvx-U6-CMV-RFP-P2A-BSD and pHBLV-U6-MCS-CMV-ZsGreen-PGK-PURO lentiviral vectors respectively. The lentivirus was generated as described above.
Tube formation H520 CTRL and H520 OV-VASH2 cells were cultured as described above. When the cells reached 80% con uence, the culture medium was changed to DMEM without fetal bovine serum. After an additional 48h culture, the supernatant was collected as CM and stored at 20℃. After thawed at 4℃ overnight, the Matrigel was coated in 96-well plate then incubated at room temperature for at least 30 min to gel.
HUVECs or HLECs were suspended at a density of 2×10 5 cells/ml in the different CMs. The cell suspensions (100 ml) were added to each Matrigel coated well. DMEM was substituted for CM for the negative control. DMEM and VASH2 factor were substituted for CM for the positive control. After 18h, the formed networks were photographed and analyzed to calculate the the formation rate of tubes was calculated according to the following equation: Formation rate (%) = Cell number Tube quantity /Cell number Total × 100%.

RT-qPCR analysis for gene expression
Total RNA was exacted using the Trizol kit according to the manufacturer's protocol. Complementary DNA was synthesized using equivalent amounts of total RNA (1 μg) in a 20 μl reverse transcriptase reaction mixture and detected using SYBR Premix Ex Taq™ in 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). The primers were synthesized from Santa Cruz (Dallas, TX, USA). The sequences of primers are shown in Table 2. All experiments were performed in triplicates and were calculated for △CT. △CT= CT target gene -CT reference gene . And relative expression quantity of mRNA is 2-△CT. Most of the PCR primer sequences used in this study were referenced from literatures (27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38). But for the primer sequences of 8 EMT-related genes (Zeb1, Zeb2, Twist1, Snail, Slug, β-catenin, N-cadherin and E-cadherin) and VEGF family genes (VEGF-A, VEGF-C and VEGF-D), they were designed by ourselves and validated in our previous work.
Enzyme-linked immunosorbent assay (ELISA) Cells (4 × 10 5 ) were seeded into 6-well plates and the culture supernatants of H520 CTRL andH520 OV-VASH2 cells were collected after 48 h to determine the VASH2 and VEGF-D levels using the VASH2/VEGF-D ELISA kit.

Western blot analysis
The cells were harvested and washed twice with PBS (pH 7.4, 0.15 M). Total protein was extracted by RIPA buffer (Beyotime, Shanghai, China). Approximately 30 μg of total protein was subjected to SDS-PAGE and transferred to PVDF membranes. The membranes were blocked with 5% skim milk in TBST and incubated with the primary antibody (1:1000) in TBST containing 5% BSA overnight at 4 °C. The membranes were washed twice with TBST and incubated with HRP-conjugated secondary antibody (1:4000; Santa Cruz, Dallas, TX, USA) at room temperature for 2 h. The membrane was exposed using an

Cell Apoptosis Analysis
The apoptotic in H520 CTRL andH520 OV-VASH2 cells was measured using Annexin-V-FITC. The Annexin-V-FITC Apoptosis Detection kit was used to measure apoptosis. After 12 hours starvation treatment, cells were washed with PBS and resuspended in 1×binding buffer at a concentration of 1×10 6 cells/ml.
Subsequently, 5µl of Annexin-V and 10µl of PI were added to 100µl of the cell suspension, and the mixture was incubated for 15min in the dark. The analyses were performed using a ow cytometry. Because the growth factors were solved in DMSO for cell culture experiment, DMSO was added into medium and diluted at the nal concentration of 5μg/ml, serving as a negative control in order to eliminate the disturbance of solvent. The experiments were repeated at least 3 times.

Wound Healing Assay
Cells were seeded into 6-well plates for 24h to reach 80~90% con uence. The cell monolayer was wounded in a straight line with a 10μl pipette tip to create a ''wound''. Cell debris was removed by PBS and the wounded cells were cultured with fresh medium. Images were taken at 0, 24h and 48h later to measure the distance of the wound. The cell migration rate = (the distance of the wound at 0h -the distance of the wound at 24h or 48h) / the distance of the wound at 0h × 100%. The experiments were repeated at least 3 times.

Trans-well invasion assay
The Matrigel and Trans-well were used to carry out lung cancer cell invasion assays to build invading chambers for separating high invasive and low invasive cells. Cells were seeded at a density of 1×10 5 cells in the Matrigel and 100μl serum-free RPMI-1640 to the 24-well plate Trans-well system with an 8-μm pore size polycarbonate lter membrane (Corning, NY, USA). The lower chambers contained 10% FBS containing medium. The cells were incubated for 48h. The cells on the lower surface of the membrane were xed with methanol and stained with 1% toluidine blue. The stained membranes were photographed through the microscope and invading cells were counted. The experiments were repeated at least 3 times.
In Vivo Tumorigenicity Study H520 CTRL andH520 OV-VASH2 cells were collected and an equal number of 1×10 6 cells in 100μl of PBS were mixed with Matrigel respectively. The mixed cells were injected subcutaneously into both anks of BALB/c mouse. The tumor sizes were monitored every 3 days. Each group contains 5 mice. And the experiments were repeated at least 3 times. The tumor volume (V) was calculated by the formula: V=3.14×L×W×H/6 (L: length, W: width, H: height). The animal protocol used in this study was approved by the Ethics Committee for Animal Experiments of the Tianjin Medical University Cancer Hospital and Institute, and was performed in accordance with the Guide for the Care and Use of Laboratory Animals.
Once palpable tumors were formed (in ~7-10 days) treatment cycles was initiated. The mice were anesthetized by intraperitoneal injection of 100 mg/kg anti-VEGFD and 10 mg/kg anti-VASH2. Treatment involved the mice being randomly divided into three groups, control, anti-VEGFD, and anti-VASH2-with each group containing 5 mice. The control group of mice were subjected to 1 × PBS injections also for 5 consecutive days. After the completion of the treatment cycle the mice were sacri ced and the tumors were excised. Tumor volumes for the excised tumors were measured using a caliper and volumes were calculated as previously mentioned. Tumors were then xed and processed for immunohistochemistry.

Flow Cytometry
The concentration of antibodies used for staining and the time of co-incubation were followed the manufacturer's instruction. PE anti-human Podoplanin antibody was used at the concentration of 12.5 μg/ml. 1×10 5 cells were resuspended in 100μl PBS, incubated for 30 min with different antibodies before washing and being xed using 4% paraformaldehyde. Since all protein markers are membrane proteins, none cells were permeabilized. The samples were analyzed by BD FACS Aria ow cytometer (BD Biosciences).

Data Analysis and Statistical Methods
Data were analyzed using SPSS 17.0. Quantitative data were presented as mean ± standard deviation.
We performed Student's t-test when analyzing the statistical comparisons between two groups. Meanwhile, for more than 2 groups included in this study, we used one-factor ANOVA assay to evaluate the difference among all groups, followed by LSD method to compare the difference between each 2 groups. Statistical signi cance was set at p<0.05.

Results
High expression of VASH2 correlated with worse prognosis and more lymphatic metastasis in LUSC patients We collected and analyzed the global gene expression pro ling of 937 primary LUSC tissue samples from 11 database based on the disparity in clinical outcomes. We de ned the patients with a survival time more than 40 months as good prognosis group, and those with a survival time less than 40 months as poor prognosis group. And we identi ed 12 highly differentially expressed genes which signi cantly correlated with patients' poor prognosis ( Figure S1). We collected LUSC tissue samples from Tianjin  Figure 1A). Among them, Overexpression ofVASH2 was signi cantly correlated with the poor prognosis in 4 of 11 databases, indicating that VASH2 is a novel prognostic factor of LUSC whose high expression was signi cantly correlated with comparably shorter OS in different databases ( Figure 1B). The nding of TCGA data implied that VASH2, previously reported to be associated with angiogenesis, might be a potential prognosis predictive biomarker for LUSC which was signi cantly up-regulated in LUSC cases suffering from shorter OS.
In order to validate the correlation between VASH2 and clinical outcome in LUSC, we collected another independent cohort composing of 64 LUSC samples from TJMUCH. The mRNA level of VASH2 in LUSC samples was detected using real-time quantitative PCR (qPCR) assay. We observed that low VASH2 expression was mainly found in patients survived more than 40 months, while level of VASH2 mRNA of good prognosis group is much lower than that of poor prognosis group (10.68±1.13 vs. 10.14±0.96, p=0.01 Figure 1C). In addition, we compared the level of VASH2protein in the corresponding LUSC para n specimens using IHC analysis, and found that VASH2 protein was predominately higher in patients with short OS compared to that in patients with long OS (63.82±5.18% vs. 20.88±3.74%, p=0.01, Figure 1D). Above results indicated that both the transcription and expression of VASH2 gene in LUSC are signi cantly correlated with poor survival outcomes consistently. Furthermore, we analyzed the association between VASH2 and other clinicopathology and found that high expression of VASH2 positively corresponds with more lymph node metastasis, lager tumor size, and worse clinical stage ( Figure 1E, Table 1). These results indicated that VASH2 was a valuable biomarker to predict poor prognosis and lymphatic metastasis in LUSC.
VASH2 accelerated the proliferation and invasion potential of LUSC cells, and promoted tube-formation capacity of HUVEC and HLEC cells Then, the supernatants from cultures of H520 CTRL and H520 OV-VASH2 were collected. In order to investigate the underlying biological roles of VASH2 in carcinogenesis and lymphangiogenesis, VASH2 was up-regulated in LUSC cell line NCI-H520 cells by transfecting recombinant lentivirus ( Figure S2A).
High level of VASH2 was con rmed in the supernatants of H520 OV-VASH2 compared to that inH520 CTRL .
Next, HUVECs and HLECs were co-cultured with the supernatants from either H520 OV-VASH2 or H520 CTRL cells. The HUVECs and HLECs cultured in DMEM medium were used as negative controls.  Figure 2H-I). Above results implied that VASH2 not only effectively promoted proliferation and invasion potential of LUSC cells, but also signi cantly enhanced lymphangiogenesis in vitro.
VASH2 facilitated lymphangiogenesis via up-regulating VEGF-D production in LUSC cells In order to de ne the molecular events regulating VASH2-related lymphangiogenesis, the transcription and expression of lymphangiogenesis-related VEGF family members inH520 OV-VASH2 cells, including VEGF-A, VEGF-C and VEGF-D, were measured using qPCR assay. No signi cant difference at the mRNA levels of VEGF-A and VEGF-C was detected between H520 CTRL and H520 OV-VASH2 cells. But mRNA and protein level of VEGF-D in H520 OV-VASH2 cells was signi cantly higher than that in H520 CTRL cells (p=0.008, p=0.046, Figure 3A-B). The level of VEGF-D protein in the supernatants of H520 OV-VASH2 and H520 CTRL was compared by ELISA assay. We found H520 OV-VASH2 secreted more VEGF-D than H520 CTRL (497.37±67.51pg/ml vs. 218.54±45.36pg/ml, p=0.023, Figure 3C). Furthermore, speci c VEGF-D blocking antibody could e ciently inhibited the tube formation of HLECs co-cultured with H520 OV-VASH2 (29.67±4.67 vs. 64.48±8.12, p=0.003, Figure 3D). These data implied that VASH2 over-expression promoted the production and secretion of VEGF-D in LUSC cells, which might directly facilitate VASH2indiced lymphangiogenesis in vitro. VEGF-D antibody was added to the two groups of cells respectively. The effect of VEGF-D on VASH2 production and release was detected by ELISA and PCR. The level of VASH2 protein in the supernatants has no difference in H520 OV-VASH2+VEGF-DAB compared with H520 OV-VASH2 (563.54±89.21pg/ml vs 592.72±60.34pg/ml, p=0.24, Figure 3E). There is still no difference in mRNA expression (6.83±1.64 vs 8.61±1.51, p=0.24, Figure 3F).

VASH2 enhanced tumor proliferation and invasion via up-regulating Snail expression in LUSC cells
The transcription of multiple EMT-related biomarkers was detected using qPCR assay, and we found that the level of Snail mRNA increased in H520 OV-VASH2 compared to H520 CTRL but that of E-cadherin decreased which indicated VASH2 promoting EMT in LUSC cells. However, there was no signi cant difference of the mRNA levels of the mesenchymal cell biomarkers β-catenin and N-cadherin, and other EMT-related TFs, such as Zeb family (Zeb1 and Zeb2), Slug, and Twist1 ( Figure 4A). These results implied that VASH2 promoted LUSC cell EMT by stimulating Snail transcription.
In order to determine if Snail regulated VASH2-stimulated VEGF-D production, H520 CTRL , H520 OV-VASH2 , H520 OV-VASH2+sh-Snail and H520 OV-Snail cells were collected to compare the mRNA levels of VEGF-A, VEGF-C and VEGF-D at by qPCR assay. Compared to H520 OV-VASH2 , a signi cant decrease of the mRNA level of VEGF-Din H520 OV-VASH2+sh-Snail , but no difference was detected in the mRNA levels of VEGF-A and VEGF-C ( Figure 5E). Consistently, compared to H520 CTRL , over-expression of Snail induced a signi cant increase of VEGF-D at 6.47 folds rather than VEGF-A and VEGF-C (p=0.008, Figure 5F). Above results indicated that Snail regulated VASH2-related lymphangiogenesis rather than angiogenesis via stimulating VEGF-D production.
VASH2 was signi cantly correlated with either Snail or VEGF-D in primary LUSC tissues The protein levels of Snail and VEGF-D in 64 cases of primary LUSC tissues were assessed using IHC staining assay ( Figure 6A). We used qPCR to detect the mRNA levels of VASH2, Snail and VEGF-D in LUSC. Snail and VEGF-D were highly expressed in VASH2 overexpression samples, and VEGF-D was also highly expressed in snail overexpression samples ( Figure 6B). In addition, the results of IHC were consistent with this result. The positive rate and intensity of Snail inVASH2-positiveLUSC tissues were higher than that in VASH2-negativeLUSC tissues (73.92±2.43% vs. 20.16±2.72%, p=0.019, Figure 6C). The similar results were observed in the average positive rate of VEGF-D (67.06±3.85% vs. 25.08±3.36%, p=0.035, Figure 6C). And among the Snail-positive LUSC patients, the average positive rate of VEGF-D was much higher than that in Snail-negative LUSC patients (61.74±4.19% vs. 20.98±2.87%, p=0.033, Figure 6C).
To further compare the correlation amongVAHS2/Snail/VEGF-D and prognosis of patients with LUSC, we divided them into double positive group. We compared the OS between VASH2 + Snail + LUSC patients and non-VASH2 + Snail + patients, as well as between VASH2 + VEGF-D + patients and non-VASH2 + VEGF-  Figure 6D).
In order to con rm if targeted therapy against either VASH2 or VEGF-D was e cient to inhibit tumor metastasis and local lymphangiogenesis in vivo, mice were treated with anti-VASH2 or anti-VEGF-D antibodies at a dose of 2.5mg/kg bodyweight by i.p. injection three times per week. We found the tumor sizes were signi cantly reduced after either treatment (H520 OV-VASH2 : 1805.62±171.56mm 3 ; H520 OV-
Then, we extracted tumor RNA and compared the mRNA levels of Snail and VEGF-D between different groups using qPCR assay. We noticed that the mRNA levels of above genes were signi cantly upregulated in H520 OV-VASH2 compared to H520 CTRL , which was consistent with the observation in vitro.  Figure   7D-E). The same conclusion could be reached by detecting protein level by western blot. The protein levels of Snail and VEGF-D in H520 OV-VASH2 was higher than that in H520 CTRL Snail: 1.84±0.49 vs 1.13±0.095, p=0.044; VEGF-D: 1.49±0.64 vs 0.92±0.12, p=0.026, Figure 7D-E). The protein levels of Snail and VEGF-D were signi cantly decreased by adding VASH2 antibody (Snail: 0.57±0.094 vs 1.84±0.49, p=0.012; VEGF-D: 0.89±0.084 vs 1.49±0.64, p=0.017, Figure 7D-E). When VEGF-D antibody was added, the protein level of Snail had no difference (Snail: 1.52±0. 28 Figure 7D-E).

Discussion
Although many studies have con rmed the predominant role of microvascular development, especially angiogenesis in tumor growth and metastasis, the anti-angiogenesis therapy has not achieved su cient clinical e cacy as expected until now [21]. Actually, recent studies showed that the initial metastasis of most malignant tumors is not through blood vessels, but rstly through lymphatic vessels to local lymph nodes [22,23]. Lymphangiogenesis, the development of new lymph vessels, has been reported to promote lymphatic invasion and lymph node metastasis. Therefore, anti-lymphangiogenesis therapy has become a relatively hot topic of clinical investigation and might be more promising strategy to inhibit the metastasis and spread of tumors in early stage.
In this study, based on 937 LUSC gene expression pro ling data from 11 public datasets in GEO and TCGA dataset, as well as 64 LUSC primary tissues from TJMUCH independent cohort, we demonstrated that VASH2 was an independent prognostic biomarker for LUSC. We validated the positive correlation between VASH2 and poor prognosis of LUSC patients both at mRNA and protein levels. The IHC results of an independent Chinese LUSC cohort showed that there was no correlation between the expression of VASH2 and CD31 and other vascular endothelial markers, while the pathological analysis showed that the expression of VASH2 was signi cantly related to the number of lymph node metastasis in LUSC patients. Furthermore, we found that VASH2 expression was signi cantly positively correlated with the number of lymph node metastasis rather than distant metastasis. We indicated that VASH2 promoted ampli cation and tube-formation of HUVEC and HLEC cells, and accelerated proliferation and invasion of LUSC cells. Further studies demonstrated that VASH2 facilitated lymphangiogenesis via up-regulating VEGF-D production in LUSC cells and enhanced tumor proliferation and invasion via up-regulating Snail expression in LUSC cells.
VASH2 was reported as an angiogenic factor promoting tumor development and progression [13,24]. VASH2 regulates the expression of VEGF through autocrine and paracrine mechanisms, and promotes angiogenesis [25]. VEGF family was identi ed as an endothelial cell-speci c mitogen that has the capacity to induce physiological and pathological angiogenesis and lymphangiogenesis. Among the VEGF family members, VEGF-C and VEGF-D have similar structures, but bind and activate VEGF receptor VEGFR-2 and VEGFR-3, respectively. VEGF-C and VEGF-D promoted angiogenesis and lymphangiogenesis within and at the periphery of solid tumors, as well as enhancing solid tumor growth and metastatic spread to lymph nodes and distant organs [26,27]. Actually, the understanding of the molecular mechanisms of VEGF-C in lymphatic development and lymphangiogenesis has been greatly appreciated in multiple cancers, however, the role of VEGF-D is less well understood. Previous studies showed that VEGF-D could compensate for the loss of VEGF-C for the sprouting of lymphatics, indicating VEGF-D might play a subtler role in regulating lymphatic vascular development than VEGF-C [28][29][30]. In this study, we found VEGF-D was signi cantly up-regulated in H520OV-VASH2 cells, but the expression of VEGF-C did not change signi cantly, suggesting that VASH2 promoted lymphangiogenesis mainly through upregulation of VEGF-D, but not VEGF-C signaling.
Snail has been reported to promote lymph node metastasis [31]. Snail is a transcription factor involved in EMT, but plays an important role in lymphatic metastasis, and increased lymph node metastasis. Consistently, we found that Snail positively correlated with the level of VASH2 and the status of lymph node metastasis. The relationship between Snail and lymphangiogenesis, as well as Snail and VEGF-D, are not fully investigated yet [32].
In this study, we found that Snail participated in VASH2-mediated lymphangiogenesis via up-regulation of VEGF-D in LUSC cells. Inhibition VEGF-D via anti-VEGF-D antibody could reduce tumor growth in vivo, not affecting Snail expression, which implied a "VASH2-Snail-VEGF-D" axis in promoting lymphangiogenesis and tumor progression in LUSC. Furthermore, we observed that inhibition of VASH2 showed better therapy effect than inhibition of VEGF-D in the mouse model, which is most likely due to the function of VASH2 monoclonal antibodies in interfering proliferation of cancer cells as well as lymphangiogenesis in tumor tissues simultaneously.

Conclusion
In conclusion, we found that the high expression of VASH2 was signi cantly related to the poor prognosis and lymph node metastasis of LUSC patients. VASH2 could increase the expression of VEGF-D, upregulate the expression of Snail and down-regulate the expression of E-cadherin in LUSC cells, and promote the formation of lymphatics. Therefore, further investigations for a comprehensive understanding of VASH2 in tumor development will probably provide us more effective promising therapeutic strategies for treating metastatic disease.

Consent for publication
My material is original research, has not been previously published and has not been submitted for publication elsewhere while under consideration.