Capsaicin Inhibits Migration and Invasion via Inhibiting Epithelial-Mesenchymal Transition in Esophageal Squamous Cell Carcinoma By UpRegulation of Cldn3 Expression

Purpose To investigate the effects of capsaicin on migration and invasion of esophageal squamous cell carcinoma and the roles of claudin-3 and epithelial-mesenchymal transition (EMT). Methods Cldn3 expression was detected in paired cancerous and adjacent normal esophageal tissues by IHC. The association of Cldn3 with clinical data was analyzed. The migration and invasion of ESCC cells were investigated by exposure of capsaicin. The migration and invasion of ESCC cells with Cldn3 silencing or Cldn3 overexpression were also evaluated. The effects of capsaicin on lung metastasis of ESCC with or without Cldn3 silencing were evaluated. Results: The positive expression of Cldn3 was associated with better prognosis of ESCC patients. Capsaicin attenuated migration and invasion of ESCC and inhibited EMT in a dose-dependent manner. Cldn3 silencing enhanced migration and invasion and EMT in ESCC cells. Cldn3 overexpression inhibited migration and invasion and EMT, similar to exposure of capsaicin. Cldn3 silencing also attenuated inhibition of migration and invasion and EMT by exposure of capsaicin. Cldn3 silencing attenuated inhibition of lung metastasis of ESCC by exposure of capsaicin in vivo. Conclusion The positive expression of Cldn3 showed a better prognosis of ESCC, which was a potentially prognostic indicator. Capsaicin inhibited migration and invasion of ESCC through up-regulation of Cldn3 and inhibition of EMT. The effects of capsaicin on Cldn3 expression and invasion in ESCC were In this study, the association between Cldn3 expression and prognosis in ESCC was evaluated. The effects of capsaicin on migration and invasion of via and EMT dependent mechanism were investigated. ESCC cell lines including KYSE150 and were used to address this overexpression plus capsaicin by wound healing assay; B) Comparison of invaded ESCC cells treated with capsaicin or Cldn3 overexpression plus capsaicin by Transwell assay; C) Quantication of mRNA expression of Cldn3 and EMT markers in ESCC cells treated with Cldn3 overexpression or Cldn3 overexpression plus capsaicin; D) Comparison of protein expression of Cldn3 and EMT markers in ESCC cells treated with Cldn3 overexpression or Cldn3 overexpression plus capsaicin. All data were as mean SD of three independent experiments. Statistical analysis was performed using an unpaired


Introduction
Esophageal cancer was the sixth leading cause of cancer-related death worldwide and the fourth leading cause of cancer-related death in China 1,2 . The mortality of esophageal cancer in China approximated 14.71/100,000 in 2013 based on National Central Cancer Registry of China 3 . Despite advances in multimodality therapy, the prognosis of ESCC remained poor with an average ve-year survival less than 30% [4][5][6][7][8] . The poor prognosis was partially led by locoregional invasion or distant metastasis at initial diagnosis. Locoregional recurrence and distant metastasis remained the primary cause of cancer-related death in ESCC patients after curative surgery 9 . Therefore, a better understanding of metastasis and identi cation of molecular targets for treatment were urgent for ESCC intervention.
Tight junction TJ was the most important intercellular junction of epithelial tissues, known as functioning in intercellular adhesion, cellular polarity and permeability of ion and macro molecular across epithelial barrier 10,11 . Tight junction proteins, physiologically existing among normal differentiated epithelial cells, were often lost in tumor tissues. Tumor cells were frequently featured by dysregulation of TJ function, less differentiation and aberrant cellular polarity 12, 13. Claudins, which belonged to the tetraspan transmembrane protein family, were the major constituents of TJ. Cldn3 was important for TJ formation, paracellular permeability and epithelial barrier function. Cldn3 was overexpressed in a variety of tumor tissues 14 . However, lower expression of Cldn3 achieved poor prognosis in certain types of tumors including ESCC [15][16][17] . The methylation of CLDN3 tended to increase in squamous dysplasia and ESCC comparing with normal esophageal epithelium, contributing to lower expression of Cldn3 in preneoplastic and ESCC 18, 19 . Loss of Cldn3 expression was associated with distant metastasis and the lower apoptotic index of ESCC 17 .
Epithelial-mesenchymal transition (EMT), which was the principal driver of metastasis and invasion of tumor cells, was observed in a variety of tumors including esophageal squamous cell carcinoma (ESCC). EMT prompted epithelial phenotype trans-differentiating into mesenchymal phenotype, which enable tumors cells to penetrate basement membrane and move to locoregional lymph nodes or distant organs 20 . Loss of E-cadherin was a hallmark of EMT in epithelial tumors, which was more prone to metastasis 21,22 . EMT markers played an important role in invasion and metastasis, which was closely related to prognosis 23 . Therefore, EMT markers were potential markers for diagnosis and therapy. It was reported Cldn3 could maintain the homeostasis of E-cadherin in ovarian cancer cells 24 .
Capsaicin was a naturally occurring alkaloid responsible for spices, food additives and drugs 25 , and had extensive pharmacological effects, e.g., pain killing, anti-in ammation and anti-oxidation. Recently, several studies indicated capsaicin also had anti-tumor effects through inhibiting proliferation and angiogenesis of tumor tissues 26, 27 . Capsaicin was known to regulate cytoskeleton and tight junctions of epithelial and endothelial cells. Capsaicin was able to irreversibly open tight junction of microvascular endothelial cells, which potentially increased the permeability of macromolecular across blood-brain barrier 28 . Capsaicin also induced reversible opening of TJ through change of cytoskeleton proteins (e.g., Co lin, F-actin) and tight junction proteins (e.g., Occludin) in epithelial cells [29][30][31][32] .
The effects of capsaicin on Cldn3 expression and invasion in ESCC were unclear. In this study, the association between Cldn3 expression and prognosis in ESCC was evaluated. The effects of capsaicin on migration and invasion of ESCC via Cldn3 and EMT dependent mechanism were investigated. ESCC cell lines including TE-1, TE-10, KYSE150 and ECa109 were used to address this issue.

Patients and samples
Para n-embedded sections of ESCC tissues and adjacent normal tissues were prepared from 93 ESCC patients (71 males and 22 females, average age 64.1 ± 6.2-year-old) under radical esophagectomy from 2013 to 2014. None of the patients received neoadjuvant therapy before surgery. The tissue samples from primary tumor and regional lymph nodes were histologically examined with hematoxylin-eosin staining. The staging was con rmed according to International Union Against Cancer TNM classi cation 53 . The differentiation was histologically subdivided into three levels: well-, moderate-and poor differentiation. The patients dying of surgery or causes other than esophageal cancer was excluded from this research. Demographic data including age, gender, stage, differentiation and histopathological factors were recorded and analyzed. All tissue samples were xed in 10% formalin and embedded in para n wax. 4-µm serial sections were sliced and examined by immunohistochemistry (IHC
Final score of 0-4 was de ned as negative expression and score of 5-12 as positive expression 20 .
Cell culture and protein extraction Human ESCC cell lines including TE-1, TE-10, KYSE150 and Eca109 were cultured in RPMI 1640 medium supplemented with 10% fetal calf serum (FCS). At the indicated time point, total protein was extracted using cell lysis buffer (Beyotime Biotechnology, China).

RNA Interference
Three target siRNA sequences were selected from different loci of human CLDN3. First, the most effective siRNA against human CLDN3 were screened. The optimal small hairpin RNA (shRNA) cassette targeting human CLDN3 (CCGGCGACCGCAAGGACTACGTCTACTC_GAGTAGACGTAGTCCTTGCGGTCGTTTTTG) and a scrambled sequence (RiboBio Co. Ltd, Guangzhou, China) was transfected into TE-1 and Eca109 cell lines in 6-well plates using jetPRIME transfection reagent (Polyplus-Transfection, Illkirch-Graffenstaden, France) according to the manufacturer's instructions. Transfected cells were treated with 200µM capsaicin and control medium. To con rm the e cacy of siRNA CLDN3, Cldn3 expression in transfected cells were measured by RT-PCR and Western Blot.

Lentiviral-mediated overexpression of Cldn3
The cDNA coding full-length human CLDN3 was ampli ed by PCR and cloned into the lentiviral vector pGag/Pol (Gene Pharma Co., CHN) (pLV-CLDN3). To produce a negative control virus (LV-NC), a short noncoding sequence was cloned into the lentiviral pGag/Pol and processed in parallel with the target gene. TE-1 and Eca109 cells (3×10 5 ) were transduced with pLV-CLDN3 and LV-NC using polybrene (5µg /ml). Transduced cells were treated with 200µM capsaicin or control medium. To con rm the e cacy of pLV-CLDN3, Cldn3 expression in transduced cells were measured by RT-PCR and Western Blot.

RT-PCR
Total RNA was isolated and cDNA was reversal transcribed according to the manufacturer's protocol. The mRNAs of target genes were measured by RT-PCR and analyzed by Rotor-Gene 3000 system (Corbett Research, Australia). Negative control was applied in each assay to ensure no contamination. All samples were detected in triplicate. The primer sequences were listed in Table 1.
Cell migration by wound healing assay ESCC cell lines including KYSE150, TE-1, TE-10 and Eca109 were grown to con uence on 6-well plates and wounded by removing about 400µm-wide strips of cells across the well with a standard 200µl-pipette tip. After washing with HBSS, fresh culture medium was added and cells were incubated at 37℃ in a humid environment with 5% CO 2 . Wound closure was photographed every 15 mins by Nikon Eclipse Ti-S/L 100 microscope system. The experiments were performed twice and assayed in triplicate.

Cell invasion by Transwell assay
Cell invasion was evaluated using a Transwell chamber assay (Costar, USA). ESCC cell lines including TE-1, TE-10, Eca109, KYSE150 were serum starved for 24 h. Transwell inserts (BD Biosciences, USA) of 8mmpore size were coated with 150 ml of 1:60 diluted Matrigel (BD Biosciences, USA). Cells (1×10 5 ) were plated onto the top of each coated lters in 150 ml serum-free medium. 300 ml of the same medium containing 20% FCS was placed in the lower chamber to act as a chemoattractant. After 24h incubation at 37°C, cells which did not move across the pores of Transwell inserts were manually removed with cotton swabs. The inserts were xed in cold methanol for 10 mins followed by staining with 0.01% crystal violet in 20% ethanol. Cell counting was performed by Coomassie blue staining followed by visualizing under microscope (Leica, Germany). The experiments were performed twice and assayed in triplicate.

Tail vein metastatic assay
The tail vein metastatic assay was performed as previously described 54 . TE-1 cells (1 × 10 6 ) treated with capsaicin 200 μM or siRNA Cldn3 plus capsaicin 200 μM were injected intravenously into nude mice through tail vein. The mice were sacri ced at 8 weeks after treated TE-1 cells injection. The metastatic lesions of lungs were harvested and examined with histology. The number of metastatic foci from lungs were quanti ed. Eight mice were included in each group.

Statistical Analysis
For the data from histologic analysis, comparisons were assessed using the chi-squared test. Survival curve was performed by Kaplan-Meier method. The discrepancies in survival distributions were evaluated by log-rank test. Cox's proportional hazards modeling was performed to con rm factors in uencing on survival. For the data of RT-PCR and Western blotting, comparisons between groups were assessed by one-way ANOVA followed by Bonferroni's test. The data were presented as the means ±SD.
Statistical signi cance was estimated using SPSS version 19.0 software (SPSS, Inc., Chicago, IL, USA). P<0.05 was considered to a signi cant difference.

Results
The association of Cldn3 expression with clinical data in ESCC patients Cldn3 was primarily located at cellular membrane and cytoplasm of ESCC cells ( Figure 1A-C), while negative or mild staining of Cldn3 was observed in adjacent normal tissues of ESCC. The positive expression of Cldn3 accounted for 43% of ESCC tissues (40/93). It was indicated that negative expression of Cldn3 was associated with locoregional lymph node metastasis and poor survival (P<0.05; Table2). There was no signi cant association among Cldn3 expression and gender, age, location, differentiation, tumor size and T status (P > 0.05; Table 2). The median follow-up was 25.6 months ranging from 1 to 52 months. Kaplan-Meier analysis revealed the median survival time for ESCC patients with positive expression of Cldn3 was 38.0 months, whereas 24.1 months for patients with negative expression of Cldn3 (P<0.05; Figure 1E). It was indicated ESCC patients with negative expression of Cldn3 achieved poor prognosis (P<0.05; Table 2).

Effects of capsaicin on migration and invasion of ESCC cells
Four ESCC cell lines including TE-1, TE-10, Eca109 and KYSE150 were treated with different doses (0µM, 100μM and 200μM) of Capsaicin for 48h respectively. The spreading speed of TE-1, TE-10, Eca109 and KYSE150 cells with Capsaicin administering was signi cantly slower than that in control cells by wound healing assay (p<0.05; Figure 2A-B). The average number of invading cells treated with capsaicin was signi cantly reduced as compared with that of control cells by Transwell assay (p<0.05; Figure 2C-D).

Page 8/23
ESCC cells including TE-1, TE-10, Eca109 and KYSE150 were exposed to 0µM, 100µM, 200µM capsaicin for 12h respectively. The expression of Cldn3 mRNA and protein was detected. The expression of Cldn3 mRNA and protein was up-regulated by exposure of capsaicin in a dose-dependent manner (P<0.05; Figure 3A-C). ESCC cells exposed by capsaicin showed up-regulation of E-cadherin and down-regulation of N-cadherin and vimentin in a dose-dependent manner (P<0.05; Figure3A-C).

Effects of capsaicin on migration and invasion of ESCC cells via Cldn3-and EMT-dependent mechanisms
TE-1 and Eca109 cells were selected for Cldn3 RNA interference based on higher endogenous expression of Cldn3 protein ( Figure 3C). CLDN3 gene silencing contributed to aggravated invasion and migration of both ESCC cells (P<0.05; Figure 4A Figure 6A-B). It was indicated capsaicin inhibited lung metastasis of ESCC by overexpression of Cldn3.

Discussion
It was reported capsaicin possessed the properties of anti-tumor and anti-proliferation; however, the exact mechanism was unknown yet. In this study, it was suggested capsaicin inhibited invasion and migration of human ESCC cells via Cldn3-and EMT-dependent mechanism. It was provided a theoretical basis for capsaicin as a potential anti-tumor medicine to ESCC.
Intercellular junction proteins played a key role in the process of progress and metastasis of tumors. Normal epithelial cells were linked by tight junctions (TJ), adheren junctions, gap junctions, desmosomes, and etc. Cldn3 was an important structural protein to participate in TJ formation, which maintained homeostasis of epithelial and endothelial barrier function [33][34][35] . The expression of Cldn3 was also associated with invasion and prognosis of epithelial tumors. Low expression of Cldn3 was reported to be signi cantly associated with poor prognosis of hepatocellular carcinoma and lung squamous cell carcinoma, which was caused by promoting cell motility, invasion and metastasis 36,37 . High expression of Cldn3 in lung adenocarcinoma indicated poor survival and cancer progression 38 . The discrepancy of Cldn3 expression might be explained by histological heterogeneity of tumors. Cldn3 was also considered as a potential blood-based biomarker in prostate cancer 39 . In this study, low expression of Cldn3 was associated with a poor prognosis of ESCC ( Figure 1D).
Capsaicin was a kind of pungent ingredient extracted from red hot pepper. Capsaicin possessed properties of anti-tumor and pro-apoptosis in some tumors. Capsaicin e ciently induced apoptosis in hepatocellular carcinoma via acceleration of intracellular Ca 2+ production, activation of reactive oxygen species (ROS) and regulation of Bcl-2 and caspase-3 40,41 . Capsaicin also induced programmed cell death of colon cancer cells via activation of ROS and regulation of PPARγ, Bcl-2 and caspases 42,43 . However, the effect of capsaicin on migration and invasion of human ESCC cells was still unclear. In this study, capsaicin signi cantly inhibited migration and invasion of ESCC cells via regulation of Cldn3 and EMT, which was a potential preventive and therapeutic agent for ESCC treatment.
EMT was the process by which polarized epithelial cells that were connected via adhesion lose their characteristics and acquired migratory and invasive properties characteristic of a mesenchymal cells 44,45 . Cells under EMT displayed loss of epithelial cell markers (e.g., E-cadherin) and gain of mesenchymal cell markers (e.g., N-cadherin and Vimentin). EMT promoted migration and invasion of epithelial cells, accelerated metastasis of cancer cells and achieved acquisition of stem cell-like properties 46-50 . In this study, capsaicin inhibited migration and invasion of ESCC cells via attenuating EMT.
Capsaicin was a known anti-tumor agent through interfering with signal pathways, e.g., NFκB and Wnt pathway 51,52 . However, there was little known about the relationship between capsaicin exposure and Cldn3 expression. In this study, capsaicin triggered activation of Cldn3 expression. CLDN3 gene silencing led to down-regulation of E-cadherin and up-regulation of N-cadherin and Vimentin. In contrast, Overexpression of Cldn3 contributed to up-regulation of E-cadherin and down-regulation of N-cadherin and Vimentin. It was suggested capsaicin inhibited migration and invasion of ESCC via Cldn3-and EMTdependent manner.
In conclusion, low expression of Cldn3 was associated with poor prognosis of ESCC patients. Cldn3 was a potential prognostic indicator of ESCC. Capsaicin inhibited invasion and metastasis of ESCC via Cldn3 overexpression and inhibition of EMT. Capsaicin was a potential anti-tumor agent for ESCC treatment.     Quanti cation of invaded ESCC cells treated with capsaicin of different doses by Transwell assay. All data were presented as mean ± SD of three independent experiments. Statistical analysis was performed using an unpaired Student's t test. *, P<0.05. Scale bar equals to 50μm.  cells treated with CLDN3 gene silencing or CLDN3 gene silencing plus capsaicin. All data were presented as mean ± SD of three independent experiments. Statistical analysis was performed using an unpaired Student's t test. *, P<0.05; **, P<0.01. Scale bar equals to 50μm.