Oral submucous fibrosis stimulates invasion and epithelial‐mesenchymal transition in oral squamous cell carcinoma by activating MMP‐2 and IGF‐IR

Abstract Oral submucous fibrosis (OSF) involves a high risk of malignant transformation and has been implicated in oral cancer. Limited studies have been conducted on the role of OSF in relation to the invasive capabilities and epithelial‐mesenchymal transition (EMT) in oral cancer. Herein, we investigated the effects of OSF on the microenvironment of human oral cancer cells. The results showed that the conditioned medium (CM) of fibrotic buccal mucosal fibroblasts (fBMFs) strongly induced the invasion of oral cancer cells and increased the activities of matrix metalloproteinase‐2. OSF significantly induced the EMT in oral cancer cells and downregulated epithelial markers, such as E‐cadherin, but significantly elevated vimentin, fibronectin, N‐cadherin, RhoA, Rac‐1 and FAK. Insulin‐like growth factor‐1 (IGF‐1) was elevated in OSF. The protein levels of the IGF‐1R were upregulated specifically in fBMF CM treatment for oral cancer cells, and the IGFR gene was confirmed by The Cancer Genome Atlas patient transcriptome data. The Kaplan‐Meier curve analysis revealed that patients with oral squamous cell carcinoma and high IGFR expression levels had poorer 5‐year survival than those with low IGFR expression (p = 0.004). The fBMF‐stimulated EMT cell model may recapture some of the molecular changes during EMT progression in clinical patients with oral cancer.

significantly increased in late-stage tumours and associated with poor prognosis in patient with OSCC. 5 Betel quid ingredients initiate deoxyribonucleic acid damage and induce reactive oxygen species production, which can lead to the promotion and progression of oral cancer. 6 Betel quid chewing is implicated in OSCC and oral submucous fibrosis (OSF), and the prevalence of OSF amongst betel quid chewers ranges from 0.9% to 4.7% in Mainland China. 7 Betel quid chewing causes the high-risk alleles and genotypes of transforming growth factor-β1 (TGF-β1), matrix metalloproteinases (MMPs) and LOX found in OSF patients may change the function and the expression of corresponding proteins and produce inflammation of fibroblasts and collagen deposition. 8 Compared with non-chewers group, betel quid dependency had the higher risk of oral potentially malignant disorders (adjusted odds ratios 8.0-51.3) than participants with non-dependent betel quid use (0-2 dependency domains; adjusted odds ratio 4.5-5.9) in six Asian populations. 9 Cancer-associated fibroblasts (CAFs) play a major regulatory role in matrix remodelling and influence the development of immunosuppression in the tumour microenvironment. 10 OSF is a chronic inflammatory disease that frequently occurs in buccal mucosa and is a progressive form of the fibrosis of lamina propria and deeper connective tissues. 11 This condition has been implicated in scar formation, tissue fibrosis and precancerous lesions due to its various mechanisms. Patients with OSF have a high correlation of oral cancer. 12 In oral cancer, OSF causes predisposition to its development and has tumour-promoting effects. 13 How bioactive molecules are secreted from OSF in the OSCC microenvironment remains unclear.
The majority of cancer-associated deaths and treatment failure amongst patients with cancer are not caused by the primary tumour but by distant metastasis, complex multistep processes including local invasion, proteolytic enzyme secretion, intravasation, extravasation and proliferation at distant sites. 14 The epithelialmesenchymal transition (EMT) is a cellular process that transforms an epithelial cell into a mesenchymal phenotype and is essential for tumour migration, cancer stem cell properties, chemotherapeutic resistance and metastatic potential. 15,16 This process in pathological states results in the complete loss of epithelial cell-cell junctions, such as α-catenin, occludin, claudins and E-cadherin, and the gain of mesenchymal phenotype markers, such as fibronectin, vimentin, Ncadherin, Snail and Slug. 17 Insulin-like growth factor-1 (IGF-1) is a circulating polypeptide and the main potential mitogen for normal and neoplastic cells with a central role in cancer development and progression. 18   shown to decrease programmed cell death and increase cancer cell differentiation. 19 Increased IGF-1 receptor (IGF-1R) expression is associated with malignancy and tumour growth in different cancer types, including paraganglioma and oral carcinoma. 20,21 IGF-1R expression is associated with poorer overall 5-year survival in patients with stage III/IV OCSCC. 22 IGF-1 functions as a tumour promoter by increasing drug resistance 23 and inducing EMT. 24 In our previous study, the mRNA and protein expression levels of IGF-1 were increased in human fibrotic buccal mucosal fibroblasts isolated from OSF specimens. 25 Although OSF enhances oral cancer progression, the molecular events responsible for the microenvironment of oral cancer remain elusive. In the present work, we aimed to discuss the characteristics of the OSCC microenvironment and explore the contribution of OSF to EMT and OSCC tumour infiltration.

| Preparation of conditioned medium and treatment cells with CM
FBMFs (passages 3-10) were cultured in DMEM with 10% v/v FBS and 1% streptomycin/penicillin at a density of 2 × 10 5 cells/10 cm dish until confluence for 5 days at 37°C incubation with 5% CO 2 , and then the CM was collected and centrifuged at 200 g for 5 min and frozen at −20°C until needed (not exceeding 2 weeks). For the treatment group, FaDu and OECM-1 cells were grown in RPMI and CM at a ratio of 2:1. The medium was changed every 3 days. To exclude that the observed effects on OSCC cell invasion and migration were due to the amounts of serum in CM, FaDu and OECM-1 cells were cultured in RPMI and DMEM standard culture media with 10% v/v FBS and 1% streptomycin/penicillin at a ratio of 2:1 in the control group.

| Assessment of invasion by a Transwell
Transwell® polycarbonate membrane cell culture inserts (Corning, MA, USA) were coated with Matrigel® matrix (BD Biosciences) for the assessment of their invasive potential in vitro. After being pretreated with or without CM from fBMFs for 30 days, the FaDu and OECM-1 cells were washed with phosphate buffer saline (PBS) and harvested and added to Transwell® inserts in serum-free medium and then incubated for an additional 24 h at 37°C incubation with 5% CO 2 .
The 8µm pore polycarbonate membrane inserts were coated with Corning® Matrigel® matrix (standard formulation with growth factor, product number: 354234, Corning Incorporated, Life Sciences), and the standard culture medium (RPMI with 10% v/v FBS) was added into the lower compartment of the apparatus. The invaded cells were fixed with 100% methanol for 20 min at room temperature and stained with Giemsa for 2 h. The cell numbers were photographed and counted under a light microscope. The data were expressed as the average cell numbers of five random fields for each treatment condition.

| Wound healing migration assay
After being pretreated with or without CM from fBMFs for 30 days, the FaDu and OECM-1 cells were plated onto six-well plates until confluence for another 24 h. The FaDu cells were treated with or without 50 ng/ml IGF-1 for 30 days, and the medium was changed every 3 days. NVP-AEW541 (IGF-IR inhibitor) was used for treatment or without for another 24 h, and then the cells were plated onto six-well plates. The cells were wounded using culture inserts (Ibidi GmbH) to create a clear line and then incubated with RPMI containing 0.5% FBS for 12 (for FaDu) or 24 h (for OECM-1). The cells that migrated into the wound area were photographed using a phase-contrast microscope (×100). 26

| Determination of matrix metalloproteinase-2 activity by gelatine zymography
The FaDu cells were treated with the CM of fBMF or 50 ng/ml IGF-1 for 30 days, and the cells were washed with PBS and cultured in serum-free RPMI medium for 24 or 48 h. The condition media (for 24 or 48 h) was collected and spun at 12,500 g for 5 min to remove any cell debris, and then the total number of viable FuDu cells was calculated using a Trypan Blue Exclusion Assay (Sigma Chemical Co.).
The equal amounts of collected conditioned media (20 μl) were added with 5× non-reducing sample buffer and loaded into 0.1% gelatin-8% sodium dodecyl sulphate (SDS) polyacrylamide gel electrophoresis (PAGE) to examine the activity of MMP-2. After electrophoresis, the gels were washed twice with wash buffer (2.5% Triton X-100) and incubated in reaction buffer (40 mM Tris-HCl, pH 8.0, containing 0.02% NaN 3 and 10 mM CaCl 2 ) at 37°C for 12 h. The gel was then stained with Coomassie brilliant blue R-250 for 30 min at room temperature and incubated with a destaining solution (10% methanol, 5% acetic acid in dH 2 O) until bands were visible. 27 The stained gel was photographed and bands corresponding to the activity of MMP-2 were quantified using ImageJ software (NIH). The quantification of the MMP-2 activity was normalized by the total number of viable FuDu cells, calculated using a Trypan Blue Exclusion Assay.

| Microculture tetrazolium proliferation assay
The FaDu and OECM-1 cells were treated with or without CM (from fBMFs) for 30 days and then seeded in 24-well plates at a density of 2 × 10 4 cells/well for 24 h or 48 h. The cells were washed with PBS and incubated with 0.5 mg/ml 3-(4,5-dimethylthiazole-2-y1)-2,5-diphenylte trazolium bromide (Sigma) in culture medium for another 4 h. The blue formazan crystals of viable cells were dissolved with 2-propanol and measured by a Hitachi U-1900 spectrophotometer (Hitachi) at 570 nm. 28

| Western blot analysis
The total cell lysates were collected using a cold mammalian protein extraction buffer kit (GE Healthcare Bio-Sciences Corp.) containing protease inhibitor cocktails (Merck Millipore). The cell lysates were subjected to a centrifugation of 10,000 g for 30 min at 4°C, after a vortex at 0°C for 10 min. The concentration of the total protein was determined using the Bradford protein assay kit (Thermo Fisher Scientific). Equal

| RNA preparation and TaqMan quantitative real-time polymerase chain reaction
After the cells were pretreated with or without CM from fBMF cells for 30 days, the total cellular RNA was isolated from FaDu and OECM-1 cells using Trizol reagent (Thermo Scientific, Waltham,

MA, USA) in accordance with the manufacturer's instructions.
Quantitative real-time PCR analysis was conducted using TaqMan one-step PCR Master Mix (Applied Biosystems). We added 100 ng of total cDNA per 25 μl reaction with MMP-2 or GAPDH primers and TaqMan probes designed using commercial software. Quantitative real-time PCR assays were performed in triplicate on a StepOnePlus sequence detection system. The threshold was set above the nontemplate control background and within the linear phase of target gene amplification to calculate the cycle number at which the transcript was detected.

| Immunofluorescence staining
The FaDu cells were treated with or without CM of fBMFs for 30 days or with or without IGF-1 or NVP-AEW541 (IGF-1R inhibitor) for 30 days. The cells were then cultured on sterile glass coverslips in six-well plates, washed in PBS for 20 min, fixed in 4% paraformaldehyde for 20 min, and permeabilized with 0.5% Triton X-100 in PBS.

| Measurement of MMP-2 promoter activity
The FaDu and OECM-1 cells were treated with or without CM from fBMFs for 30 days. A 460 bp (−218 to +243) segment from the 5ʹ-promoter region of the MMP-2 gene was cloned. The pGL3-MMP-2 plasmids were transfected into FaDu or OECM-1 cells using TurboFect transfection reagent (Thermo Scientific) in accordance with the manufacturer's instructions for 48 h. After transfection, the MMP-2 promoter plasmids and cells lysates were prepared and examined by a luciferase assay system (Promega). Firefly luciferase activities were standardized for the β-galactosidase activity. 30

| Cell-matrix adhesion assay
The FaDu and OECM-1 cells were treated with or without CM of fBMFs for 30 days. The cells were seeded into 24-well dishes coated with collagen type I for 40 min. Non-adherent cells were removed by washing the dishes with PBS. Adherent cells were fixed with 100% methanol for 20 min and stained with 0.1% crystal violet for 30 min. After staining, the fixed cells were lysed in 30% acetic acid in water, and the absorbance was measured at 550 nm with a Hitachi U-1900 spectrophotometer (Hitachi). 27

| Human-activated receptor tyrosine kinase array
The FaDu and OECM-1 cells were treated with or without CM of fBMF cells for 30 days and then rinsed with PBS and solubilized cells at 1 × 10 7 cells/ml in lysis buffer for 30 min. In brief, 300 μg of protein of whole-cell lysates were incubated with human phospho-RTK arrays (ARY001B) (R&D Systems), in which the arrays were incubated with the total cell lysates overnight at 4°C with shaking, washed with 1× washing buffer and incubated with 1X streptavidin-horseradish peroxidase for 2 h at room temperature on a rocking platform shaker before being incubated with a Chemi Reagent Mix and exposure of the membranes to X-ray film. The arrays were scanned with the Bio-Rad Molecular Imager Gel Doc XR system. 27  mouse was intraperitoneally injected with 150 mg D-luciferin/kg bodyweight and was anaesthetized with inhaled 2% isoflurane. Tumour growth was performed using an IVIS 50 animal imaging system (Xenogen Corporation), and the signal intensity was analysed using live imaging software. 27 After 45 days, the animals were euthanized with 30-70% displacement rate of chamber air with CO 2 gas/min.

| Bio-Plex Pro-human cytokine assay
FBMFs were cultured in DMEM with 10% FBS and 1% streptomycin/penicillin at a density of 2 × 10 5 cells/10 cm dish until confluence for 5 days at 37°C incubation with 5% CO 2 , and then the CM was collected and centrifuged at 1000 rpm for 5 min and frozen at were rinsed three times with wash buffer before the detection antibody was added to each well. After another washing step, 50 μl of streptavidin-phycoerythrin solution was added to each well, and they were incubated for 10 min. After the last incubation step, the beads were resuspended in 125 μl of assay buffer with shaking at 1100 rpm for 30 s. The cytokine data were analysed using the Bio-Plex 200 suspension array system (Bio-Rad Laboratories).
The cytokine protein expression of fBMF was compared with the standard DMEM with 10% FBS (control group).

| Colony formation assay
The FaDu and OECM-1 cells were treated with or without CM from fBMFs for 30 days, and then five thousand OECM-1 or FaDu cells were plated as single cells in a six-well plate for 8 days. The medium was changed every 3 days. After fixing with methanol, the cells were stained with crystal violet (Millipore, Sigma) for 30 min.

| IGF-1R expression analysis of The Cancer Genome Atlas
The Cancer Genome Atlas, a landmark cancer genomics programme,

| Statistical analysis
The statistical analysis was conducted using one-way analysis of variance (ANOVA) with a post hoc Dunnett's test (GraphPad Software).
The p-values less than 0.05 were considered as statistically significant differences.

| OSF induces the invasion and migration of OSCC
As OSF is a chronic fibrosis with a long-term effect in the tumour  Figure 1A).
The wound healing migration assay showed that fBMFs significantly increased the migration capability of FaDu and OECM-1 cells ( Figure 1B). These data indicated that the invasion and migration activity of FaDu and OECM-1 treated with fBMFs were increased for a long time treatment.

| OSF exerts an increased effect on MMP-2 of OSCC
The FaDu cells were treated with the CM of fBMFs for 30 days, and The data indicated that fBMFs significantly elevated the activity of MMP-2 in FaDu cells (Figure 2A). The MMP-2 protein level was also examined using a western blot assay, and the expression of MMP-2 of the FaDu and OECM-1 cells was increased with fBMF treatment ( Figure 2B). The mRNA expression ( Figure 2C) and MMP-2 transcription activity ( Figure 2D) of FaDu and OECM-1 cells increased when treated with fBMFs. fBMF treatment also elevated the cell-matrix interactions of FaDu and OECM-1 cells ( Figure 2E).

| OSF reduces the cell proliferation and tumour growth of OSCC
The FaDu and OECM-1 cells were cultured with or without CM from fBMFs for 30 days, and the cell proliferation of oral cancer cells was investigated using an MTT assay. OSF inhibited the cell growth of FaDu and OECM-1 cells compared with the 48 h control group ( Figure 3A). We performed a colony formation assay to examine the long-term effect of fBMF-reduced cell growth. The colonies number of OECM-1 and FaDu cells significantly reduced after fBMFs CM treatment ( Figure 3B). Luciferase-expressing OECM-1 cells were treated with or without CM from fBMFs for 30 days and then subcutaneously inoculated into the right and left flank of immunodeficient nude mice, respectively, to determine the in vivo effects of OSF. The tumour growth of the fBMF-treated groups was remarkably inhibited compared with that of the control groups ( Figure 3C).

| OSF improved the EMT of OSCC
OECM-1 and FaDu cells were pretreated with or without the CM of fBMFs for 30 days to elucidate whether the OSF could affect cell morphology changes. The cells exhibited fibroblast-like morphology and suppressed cell-to-cell contact ( Figure 4A). The CM of fBMFs induced cell scattering and fibroblast-like morphology as investigated by immunofluorescence staining with Texas-568 phalloidin to visualize the actin cytoskeleton ( Figure 4B). The effect of OSF on the major regulators and markers of EMT in OSCC was also analysed.

| OSF promoted EMT and invasion by upregulating IGF-1R expression in human oral cancer
Considering that OSF significantly elevates the invasion and migration of cancer cells, we subsequently examined whether OSF influ- A gelatine zymography assay revealed that IGF-1 elevated the MMP-2 activity of FaDu cells ( Figure 5F). After treatment with IGF-

1, the migration and invasion of the FaDu cells were increased, and
NVP-AEW541 attenuated the IGF-1-induced migration ( Figure 5G) and invasion ( Figure 5H). Thus, IGF-1 was elevated in OSF and, consequently, increased the expression of IGF-IR and promoted the EMT and invasion by upregulating IGF-1R in human oral cancer.

| IGF-1R is increased in patients with OSCC
The Cancer Genome Atlas (TCGA) OSCC data set (n = 328) was further used to verify our findings. The mRNA expression of IGF-1R in OSCC tissues was higher than that in normal tissues (p < 0.0001; Figure 6A). Higher IGF-1R levels were detected in OSCC tissues (n = 33) compared with their corresponding adjacent non-cancerous tissues (p < 0.0001; Figure 6B). A positive correlation was found between the MMP-2 and IGF-1R expression in OSCC tumours (p < 0.0001; Figure 6C). The results from the TCGA data set of patients with OSCC indicated that high IGF-1R expression was associated with a poor survival rate (p = 0.004; Figure 6D).  Figure 6E). A positive correlation was also found between Cadherin-2 (CDH2, also known as N-cadherin) and IGF-1R expression in OSCC tumours (p < 0.01; Figure 6F). These data suggest that IGF-1R expression may promote the mortality of patients with OSCC.
Taken together, we provided molecular evidence of the tumourpromoting activity of OSF-derived fibroblast on OSCC cells by showing marked induction of the IGF-1 production. These findings suggested that OSF-derived fibroblast is able to promote migration and invasion and induce an epithelial-mesenchymal-like state via the upregulation of IGF-1R in oral cancer cells (Figure 7).

| DISCUSS ION
EMT activation plays an essential role in the initial oral cancer metastasis and generates cancer stem cells in OSCC. 16 EMT is induced by several signalling pathways, including Hedgehog, Wnt/β-catenin, hypoxia and TGF-β1. TGF-β1 is a multifunctional growth factor and plays a dual role in pro-and antitumoral effects. The expression of TGF-β1 was correlated with the induction of the angiogenic pathway in OSCC. 39 TGF-β1 is also a potent inhibitor of epithelial cell growth in the early stages of carcinogenesis in cancer, and this property contributes to its role as a tumour-suppressor. 40 Earlier reports indicated that high expression EMT markers in malignant mesenchymal tumours will reduce tumour growth. Researchers suggested that turning off EMT-inducing transcription factor Twist1 expression to allow reversion of the EMT is essential for disseminated tumour cells to proliferate and promotes colonization in distant sites. 41 The data presented here demonstrate that OSF induced the invasion and EMT of FaDu and OECM-1 cells but reduced the cell proliferation and tumour growth in vivo. Rho guanine nucleotide exchange factor. 42 RhoA mediates the cell motility of many diverse types of cancer as indicated by its regulation of cytoskeletal organization and gene expression. 46 RhoA is also implicated in mediating the activation of FAK in response to the development of aggressive non-small cell lung cancer. 47 Our results showed that OSF activated IGF-1R and increased RhoA expression and FAK activation in human OSCC. This phenomenon indicates that IGF-1 from OSF has a potential role in increasing cancer progression in the OSCC microenvironment.
IGF-1 is a 70-amino acid polypeptide hormone and a major determinant in cancer development and pathogenesis. 42 The insulin/IGF system was implicated in the development of drug resistance to epidermal growth factor receptor-targeted agents and chemotherapeutic drugs in colorectal cancer. 48 Increased IGF-1R signalling is associated F I G U R E 5 Effects of IGF-1 on the morphology, MMP-2, and invasion in oral cancer cells. (A) FaDu and OECM-1 cells were treated with the CM of fBMF for 30 days, and cell lysates were used to detect 49 different receptor tyrosine kinases by human phosphor-RTK array (from R&D system). (B) FaDu cells were treated with the CM of fBMF for 30 days, and the cell lysates were subjected to western blot analysis to detect IGF-1 receptor (IGF-1R) and p-IGF-1R with β-actin as an internal control. (C) CM was collected from fBMFs and analysed for IGF-1 with an ELISA kit. (D) FaDu cells were treated with the CM of fBMF and 50 ng IGF-1 with or without 2 μM NVP-AEW541. The cells were imaged with a microscope (100×). (E) Immunofluorescence staining of FaDu cells (treated with or without IGF-1 or NVP-AEW541 for 30 days) with Texas-568 phalloidin was used to visualize the actin cytoskeleton (400×). (F) FaDu cells were treated with 50 ng/ml IGF-1 for 30 days, and then the condition media (for 24 h) was collected for the analysis of MMP-2 by gelatine zymography. (G) FaDu cells were treated with IGF-1 with or without NVP-AEW541 by a wound healing assay. (H) FaDu cells were treated with CM of fBMF and 50 ng/ml IGF-1 with or without 2 μM NVP-AEW541 by a cell invasion assay (40×). (I) Cytokines from fBMFs were analysed using a Bio-Plex Pro-human cytokine 27-plex assay. CM was collected from fBMFs and was analysed for 27 different cytokines. The cytokine expression of fBMFs was compared with standard DMEM with 10% FBS as a control group. Results representing three separate experiments are shown for B-H, and the statistical significance was analysed through one-way ANOVA with a post hoc Dunnett's test (***, p < 0.001) with a poor response to anti-EGFR treatment in head and neck cancer cells. 49 IGF-1 promotes the motility of muller glial cells by activating MMP-2 and the phosphoinositide 3-kinase signalling pathway. 50 High IGF-1R expression in tumour cells is associated with poor outcomes in patients with resected pancreatic ductal adenocarcinoma. 51 In this work, patients with OSCC and high IGFR expression levels had poorer 5-year survival than those with low IGFR expres- in oral squamous cell carcinoma. A significant correlation was found between FN1 and IGF-1R (Spearman rank correlation coefficient r = 0.3405, p < 0.0001). (F) Correlations between the mRNA levels of Cadherin-2 (CDH2) and IGF-1R in oral squamous cell carcinoma. A significant correlation was found between CDH2 and IGF-1R (Spearman rank correlation coefficient r = 0.1499, p < 0.01). The p values were determined using a log-rank test F I G U R E 7 Tumour-promoting activity of OSF on human OSCC cells is attributed to the activation of IGF-1R, invasion and EMT in human OSCC cells produce inflammatory cytokines, chemokines and numerous growth factors, and these secretory products facilitate communication between CAFs and cancer cells, resulting in cancer progression. 52 In the current study, five cytokines, including IL-6, eotaxin, G-CSF, IP-10 and MCP-1 were increased by fBMFs in the Bio-Plex Pro-human cytokine assay. These five cytokines may require penetration to induce cell invasion and migration. Further research is required to verify the role of cytokines and factor expression by fBMFs in the progression of OSCC.
A positive correlation was also found between EMT markers (FN1 and CDH2) and the IGF-1R expression in OSCC tumours. OSF was reported to be associated with better prognosis amongst OSCC patients. 53 As the TCGA data did not provide the cases that may have fibrosis in OSCC patients, there was a limitation in linking oral fibrosis and IGF-1R expression.
The tumour microenvironment is considered as an attractive target for new anti-tumour therapies for patients with OSCC. OSF and cancer-associated fibroblasts have been shown to have tumourpromoting effects in oral cancer. This study discusses the characteristics of the microenvironment of OSCC and investigates the contribution of OSF to the epithelial-mesenchymal transition and tumour infiltration of OSCC. The present study showed that the tumour-promoting activity of OSF on human OSCC cells was attributed to the activation of IGF-1R, invasion, and EMT in human OSCC cells. Additionally, the high IGF-1R expression was associated with a poor survival rate in patients with OSCC. These data suggest that IGF-1R may promote the mortality of patients with OSCC. IGF-1, which is secreted in OSF, promoted the migration ability in oral cancer cells.

CO N FLI C T S O F I NTE R E S T
The authors declare that there is no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used to support the findings of the present study are available from the corresponding author upon request.