Two primary sources for the ESCC data were obtained from Oncomine database (https://www.oncomine.org) (37,38) to determine the mRNA expression levels of SREBP1. The two groups had sample sizes of 56 and 17 cases, respectively. Median age, sex and age range data were not available in the Oncomine database.

Tissue samples were collected during surgical resection after informed written consent had been provided by patients with ESCC (Table I). The patients were diagnosed by two pathologists in accordance with the World Health Organization classification without chemotherapy or radiotherapy before resection, as described previously (39). The 77 paired ESCC tumor and adjacent normal tissue samples that were 2 cm away from the tumors' edge, were collected between February 2018 and July 2019 at the Affiliated Hospital of Jiangsu University (Zhenjiang, China). The IHC staining procedure used to determine SREBP1 expression levels was performed as previously described (7,39). The sections were incubated with antibody against SREBP1 (1:100; catalog no. 14088-1-AP; Proteintech Group, Inc.) overnight at 4°C. Briefly, slides were evaluated at a magnification of ×200 and 5 representative fields of each section were assessed independently by 2 trained observers who were blinded to patient information. Identical settings were used for all of the images. The integrated absorbance and area of the images were counted by Image-Pro Plus version 6.0 software (Media Cybernetics, Inc.), and the sections were classified as either high or low expression. The intensity of staining was scored as follows: 0, no staining; 1, weak staining; 2, moderate staining; and 3, strong staining. The proportion of stained cells was scored as follows: 0, no cells stained; 1, <50% of cells stained positive; 2, 50 to 75% of cells stained positive; and 3, >75% of cells stained positive. The final score was determined by combining the 2 scores. A score of ≤6 was considered low expression, and a score >6 was considered high expression. The present study was approved by The Medical Ethical Committee of Affiliated Hospital of Jiangsu University (Zhenjiang, China).

ESCC cell lines TE-1, ECA-109 and KYSE-150 were purchased from Shanghai GeneChem Co., Ltd., which were authenticated using Short Tandem Repeat profiling. All experiments using these cell lines were performed within 6 months of receipt or thawing after −80°C cryopreservation. ESCCs were cultured in RPMI-1640 (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (FBS; Biological Industries) and 1% penicillin-streptomycin (Gibco; Thermo Fisher Scientific, Inc.) at 37°C in a thermostatic incubator containing 5% CO₂. Immortalized human normal esophageal epithelial cells, Het-1A (BeNa Culture Collection), were maintained in DMEM (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% FBS and containing 1% penicillin-streptomycin, as described previously (39). Wnt/β-catenin inhibitor ICG-001 was obtained from Selleck Chemicals and dissolved in dimethyl sulfoxide as previously described (7,40). Fatostatin (inhibitor of SREBP) and Mevastatin [an inhibitor of HMG-CoA reductase (HMGCR)] were purchased from MedChemExpress and dissolved in dimethyl sulfoxide. ESCC cells received treatment with ICG-001, Fatostatin and Mevastatin at the recommended concentration of 10 µM at 37°C for 48 h (41,42).

Since ECA-109 and KYSE-150 are classic cell lines of human esophageal squamous cell carcinoma and these 2 cell lines are stable after lentivirus transfection, which is convenient for cell sub-culture (43). ECA-109 and KYSE-150 cells were chosen for transfection experiments in the present study. Lentiviral vectors with small hairpin (sh)RNA-SREBP1 (shSREBP1 Bank Id: NM_004176, NM_001321096 and NM_001005291), SREBP1-expressing vector (Bank Id NM_001321096), empty vector control and scrambled control shRNA as negative control (shNC) were purchased from Shanghai GeneChem Co., Ltd. The cell concentration was adjusted to 5×10⁴ cells/ml and inoculated in 6-well plate with 2 ml/well of culture medium. After 24 h of cell adherence, culture medium was replaced by serum-free medium (1 ml) and then 20 µl lentiviral vectors, shRNA and controls (1×10⁸ TU/ml) and 4 µl of HitransG A (Shanghai GeneChem Co., Ltd.) were added. The efficiency of transfection was verified via western blotting 72 h following transfection and subsequent experiments were performed. The concentration of shRNA, vector and controls transfected are according to the manufacturer's protocol. The transfection reagent used was HitransG A (Shanghai GeneChem Co., Ltd.). The small interfering (si)RNA against NC, SREBP1 and SCD1 were designed and synthesized by Shanghai GenePharma Co., Ltd. In order to assess the effects of short- and long-term silencing of SREBP1, shSREBP1 was used to silence SREBP1 for a long time and siSREBP1was used to silence SREBP1 for a short time. The siRNA sequences were as follows: NC, 5′-GCGACGAUCUGCCUAAGAU-3′; SREBP1, 5′-GCUCCUCACUUGAAGGCUUTT-3′; and SCD1, 5′-TGAAAGAAGATATTCACGA-3′. siRNA (40 nM) was transfected using Lipofectamine^® 2000 (Thermo Fisher Scientific, Inc.) in serum-free conditions for 7 h prior to replacement with complete medium. The efficiency of silencing or overexpression were confirmed by western blotting.

mRNA expression levels in the clinical specimens of paired ESCC and normal control tissues were compared. In total, 20 pairs of frozen ESCC tumors and matched normal tissues were randomly selected, immediately snap-frozen in liquid nitrogen and stored at −80°C. RNA was extracted from these tissues using RNAiso Plus (Takara Biotechnology Co., Ltd.), in accordance with the manufacturer's instructions, and prepared for RT into cDNA using a reverse transcription reagent kit (cat. no. R123-01; Vazyme Biotech Co., Ltd). qPCR was performed using a PCR kit (cat. no. Q311-02/03; Vazyme Biotech Co., Ltd.), and the primers used were synthesized by Invitrogen; Thermo Fisher Scientific, Inc. The qPCR thermocycling conditions were as follows: 1 cycle at 95°C for 30 sec; 40 cycles at 95°C for 10 sec, 60°C for 30 sec and 3 cycles at 95°C for 15 sec, 60°C for 60 sec, 95°C for 15 sec following the protocols of ChamQ™ SYBR qPCR Master Mix (Vazyme Biotech Co. Ltd.). The sequences were as follows: SREBP1 forward, 5′-CGGCGCTGCTGACCGACATC-3′ and reverse, 5′-CCCTGCCCCACTCCCAGCAT-3′; Snail1 forward, 5′-CTTCTCCTCTACTTCAGTCTCTTCC-3′ and reverse, 5′-TGAGGTATTCCTTGTTGCAGTATTT-3′; Slug forward, 5′-AACAGAGCATTTGCAGACAGGTC-3′ and reverse, 5′-GCTACACAGCAGCCAGATTCC-3′; MMP2 forward, 5′-CTTCCAAGTCTGGAGCGATGT-3′ and reverse, 5′-TACCGTCAAAGGGGTATCCAT-3′; MMP9 forward, 5′-GGGACGCAGACATCGTCATC-3′ and reverse, 5′-TCGTCATCGTCGAAATGGGC-3′; VEGF-A forward, 5′-AGGAGGAGGGCAGAATCATCA-3′ and reverse, 5′-CTCATTGGATGGCAGTAGCT-3′; VEGF-C, forward, 5′-TGTGTGTCCGTCTACAGATGTG-3′ and reverse, 5′-TCGGCAGGAAGTGTGATTGG-3′; and GAPDH forward, 5′-AGCCACATCGCTCAGACAC-3′ and reverse, 5′-GCCCAATACGACCAAATCC-3′. Relative expression levels were calculated using the 2^−ΔΔCq method and GAPDH was used as an internal reference gene (44).

The total cellular or tissue protein was extracted using RIPA lysis buffer (Beijing Solarbio Science & Technology Co., Ltd.) at 4°C. Cytoplasmic protein and nuclear proteins were extracted using the Cytoplasmic Protein Extraction kit (cat. no. BC3740; Beijing Solarbio Science & Technology Co., Ltd.) and Nuclear Protein Extraction kit (cat. no. R0050; Beijing Solarbio Science & Technology Co., Ltd.) according to the manufaturer's protocols. Then, protein concentration was calculated using a BCA protein assay kit (CoWin Biosciences). Following that, equivalent amounts of total protein (40 µg/lane) were added into each lane of 8–15% gel, resolved using SDS-PAGE and transferred to PVDF membranes. Membranes were blocked using 5% bovine serum albumin (BSA) or non-fat milk diluted in TBST for 1 h at room temperature. Thereafter, the blocked membranes were incubated with the following primary antibodies at 4°C overnight: SREBP1 (1:1,000; catalog no. sc-13551; Santa Cruz Biotechnology, Inc., or 1:1,000; catalog no. 14088–1-AP; Proteintech Group, Inc. Two different antibodies were used to ensure the specificity of antibodies and results), SREBP2 (1:1,000; cat. no. sc-13552; Santa Cruz Biotechnology, Inc.), SCD1 (1:1,000; catalog no. ab236868); GSK-3β (1:1,000; catalog no. ab68476), phosphorylated (p)-GSK-3β (1:1,000; catalog no. ab32391), VEGF-A (1:1,000; catalog no. ab1316), MMP2 (1:1,000; catalog no. ab92536), MMP9 (1:1,000; catalog no. ab76003; Abcam), E-cadherin (1:1,000; catalog no. 14472), Vimentin (1:1,000; catalog no. 5741), Snail (1:1,000; catalog no. 3879), β-catenin (1:1,000; catalog no. 8480); Met (1:1,000; catalog no. 8198), CD44 (1:1,000; catalog no. 3570), T cell factor 1 (TCF1; 1:1,000; catalog no. 2203), lymphoid enhancer binding factor 1 (LEF1; 1:1,000; catalog no. 2230), cyclin D1 (1:1,000; catalog no. 2978), MMP7 (1:1,000; catalog no. 3801), c-Myc (1:1,000; catalog no. 5605), β-actin (1:1,000; catalog no. 4970; Cell Signaling Technology, Inc.), N-cadherin (1:500; catalog no. WL01047; Wanleibio Co., Ltd.), Ki-67 (1:500; catalog no. WL01384a; Wanleibio Co., Ltd.), VEGF-C (1:400; catalog no. BA0548; Wuhan Boster Biological Technology, Ltd.) and Lamin B1 (1:1,000; catalog no. 66095-1-Ig; Proteintech Group, Inc.). Subsequent to washing the membranes with TBST, the membranes were incubated with the anti-rabbit and anti-mouse secondary antibodies conjugated to horseradish peroxidase (1:1,000; catalog nos. 7076 and 7074; Cell Signaling Technology, Inc.) for 1 h at room temperature. After washing the membranes with TBST, the membranes were ameliorated using an ECL kit (Vazyme Biotech Co., Ltd.), followed by obtaining images of the emitted signals using a ChemisScope-4300 imager (Clinx Science Instruments Co., Ltd.).

Cell growth rate was assessed by the CCK-8 Cell Counting kit (cat. no. A311-01/01; Vazyme Biotech Co., Ltd.) according to the manufacturer's instructions. After 48 h of transfection, the untreated control, negative control vector, scramble shRNA-transfected-SREBP1 and SREBP1-expression vector-transfected cells were seeded in a 96-well plate and subsequently followed by 24, 48 and 72 h incubation at 37°C. At every time point, 10 µl CCK-8 solution was added to each well of the plates and the cells were incubated for another 1 h at 37°C in darkness. The absorbance value at 450 nm was detected to represent cell viability using a microplate reader (BioTek Instruments, Inc.). Assays were repeated at least three times.

For the wound healing assay, ECA-109-NC, ECA-109-shSREBP1, KYSE-150-Vector and KYSE-150-SREBP1 cells were seeded onto 24-well plates to form a cell monolayer (80% confluence). The cell layer was mechanically disrupted using a sterile 200-µl pipette tip to generate a linear wound, washed three times with PBS to remove floating cells or debris and then cultured with serum-free medium (RPMI-1640; Thermo Fisher Scientific, Inc.). Images were captured 24 h after scraping using a light microscope (Olympus Corporation) (magnification, ×200). The wound area for each condition was quantified using Image-Pro Plus version 6.0 software (Media Cybernetics, Inc.) and the migration rate was calculated using GraphPad Prism 5.0 software (GraphPad Software, Inc.) at each time point in three independent samples.

Cell invasion was analyzed using 24-well Transwell plates with a pore size of 8 µm (Corning Life Sciences). For invasion assay, the Transwell inserts were pre-coated with 25 µl Matrigel (BD Biosciences) at 37°C and the plates were incubated for 24 h at 37°C and then ECA-109-NC, ECA-109-shSREBP1, KYSE-150-Vector and KYSE-150-SREBP1 cells (2×10⁵) were suspended in 200 µl serum-free RPMI-1640 medium and seeded on the upper chamber. The lower chamber was filled with 600 µl RPMI-1640 containing 10% FBS as an attractant. After incubation for 24 h, the non-migrated cells in the upper chamber were removed using a cotton swab. Invaded cells were fixed with 4% cold paraformaldehyde and stained with 0.2% crystal violet (CoWin Biosciences) at room temperature for 15 min. The stained cells were counted at ×200 magnification under an light inverted microscope (Olympus Corporation). At least three different experiments were performed and data are presented as the mean number of stained cells from five randomly chosen fields.

The expression levels of MMP9, VEGF-A and VEGF-C in culture medium were determined using quantitative ELISA kits (cat. nos. EK1M09-96, BMS277-2, EK1154-96; Hangzhou Multisciences Biotech, Co., Ltd.) according to the manufacturer's instructions.

Immunofluorescence was used to detect the location and expression of target proteins (E-cadherin, Vimentin) as described previously (7,39). Cells were grown on glass coverslips and fixed in 4% paraformaldehyde at 4°C for 30 min, rinsed three times with PBS, permeabilized with 0.5% Triton X-100 for 10 min and blocked at room temperature for 30 min using PBS containing 5% BSA. The cells on the coverslips were incubated with the indicated primary antibodies against: E-cadherin (1:200; catalog no. 14472; Cell Signaling Technology, Inc.), Vimentin (1:200; catalog no. 5741; Cell Signaling Technology, Inc.) and Ki-67 (1:100; catalog no. 27309-1-AP; ProteinTech Group, Inc.) at 4°C overnight, followed by incubation at room temperature for 1 h with Cy3 or FITC-conjugated secondary antibodies (1:1,000; cat. nos. ab6717, ab6939 and ab97035; Abcam). The nuclei were counterstained using DAPI (Sigma-Aldrich; Merck KGaA) at room temperature in the dark for 10 min. Images were captured using a fluorescence microscope (BX51; Olympus Corporation; magnification, ×200) and analyzed using ImageJ version 1.46 software (National Institutes of Health).

A total of 20 female BALB/c-nu mice (age, 6 weeks; weight, 18–20 g) were purchased from Cavens Laboratory Animal Inc. (http://www.cavens.com.cn/). After 5 days of acclimatization in the Laboratory Animal Research Center of Jiangsu University at a constant temperature at 25°C and humidity from 50–70% on a 12/12 h light/dark cycle with free access to food and water, the mice were subcutaneously injected in the right thigh with 7×10⁶ ECA-109-sh-SREBP1, ECA-109-NC or ECA-109 cells suspended in 100 µl PBS and injected subcutaneously into the right dorsal flanks of mice, while mice only subcutaneously injected with 100 µl PBS were treated as the control group. All mice were randomized assigned into 4 groups (n=5) and the investigators were blinded to the group assignment. The mean fold-change in tumor size was plotted for each animal group. The tumor volume was measured every 4 days using a Vernier caliper and calculated as follows: Tumor size=(π × length × width × height)/6. Tumor tissues were harvested for molecular analysis and the standard deviation values were determined for the normalized values of each sample. For lung metastasis formation, ten mice were randomly selected and divided equally into two groups. The mice were injected via the lateral tail vein with 1×10⁶ ECA-109-NC cells or ECA-109-shSREBP1 cells suspended in 100 µl PBS. Tumor sizes were measured every 4 days, and mice were sacrificed on day 28. The mice with a weight loss >25% of total body weight and a maximum tumor size of 428 mm³ were defined as humane endpoints. The mice were euthanized 4 weeks after injection by CO₂ inhalation gradually increasing in concentration in sealed chamber (45×35×30 cm) with flow meter (3–5 min; with the flow rate of 2.5 l/min and the volume displacement rate of 28%/min), and death was confirmed by ceased breathing, cardiac arrest and pallid eyes. The lungs were removed, (10%, neutrally buffered, stored at 4°C overnight), embedded in paraffin, deparaffinized sections (5-µm thick) and incubated at 60°C for 2 h followed by deparaffinization with xylene and rehydration in concentrations of 100, 95, 85 and 75% alcohol, respectively. Subsequently, the sections were stained with hematoxylin and eosin (H&E). The metastatic foci in the lungs were counted manually under a light microscope (magnification, ×100; BX51; Olympus Corporation). All animal experiments were performed in strict accordance with the principles and procedures approved by The Animal Experimentation Ethics Committee of Jiangsu University (approval number, UJS-IACUC-AP-2020032559).

All in vitro experiments were repeated at least three times. The data were analyzed using GraphPad Prism 5.0 software (GraphPad Software, Inc.), and the values are presented as the mean ± standard deviation. Differences between two groups were analyzed using an unpaired Student's t-test or using a paired Student's t-test when comparing the SREBP1 expression between tumor and non-tumor tissues from the same patient. One-way ANOVA with Tukey's post hoc test were used for multiple group comparisons. The association between SREBP1 and clinicopathological features was assessed using χ² tests. P<0.05 was considered to indicate a statistically significant difference.

Expression levels of SREBP1 were investigated through bioinformatic analysis using Oncomine to determine whether SREBP1 is aberrantly expressed in ESCC. Results demonstrated that SREBP1 mRNA expression levels in ESCC tumors were significantly higher compared with normal esophageal tissues in two independent datasets (Fig. 1A) (37,38). Similarly, data from the IHC staining showed consistently higher levels of SREBP1 in primary ESCC tissues (32/77, 41.6%) compared with normal non-neoplastic tissues (5/77, 6.5%). As presented in the Fig. 1B, SREBP1 was primarily located in the cytoplasm of ESCC or normal cells. The association between SREBP1 expression levels and clinicopathological features was further analyzed. IHC of human ESCC samples revealed that SREBP1 expression was significantly associated with tumor differentiation, lymphatic metastasis and Ki-67 expression (Table I). In addition, the expression levels of SREBP were significantly higher in ESCC tumors compared with adjacent normal tissues, as detected using western blotting and RT-qPCR (P<0.001; Fig. 1C). The expression levels of SREBP1 and mature (m)SREBP1 were increased in ESCC tissues compared with the matched normal tissues, and the difference in SREBP2 expression was not significant (Figs. 1D and S1). SREBP1 expression levels in ESCC cell lines were measured to investigate the potential effect of SREBP1 in ESCC. The results demonstrated that SREBP1 protein expression was higher in all three ESCC cell lines (TE-1, ECA-109 and KYSE-150) compared with the normal immortalized cell line Het-1A (Fig. 1E). Quantitative protein analysis revealed that the relative expression of SREBP1 protein in TE-1, ECA-109, and KYSE-150 cells was 2.62, 2.41, and 1.95 times that of Het-1A cell, respectively (P<0.05; Fig. 1E). Notably, the ECA-109 and TE-1 cell lines had higher levels of SREBP1 expression, whereas KYSE-150 cells had relatively low expression. SREBP1 was then knocked-down in ECA-109 cells and overexpressed in KYSE-150 cells to functionally validate the role of SREBP1 in ESCC. Compared with the control and negative control groups, the relative expression level of SREBP1 was significantly decreased in the shRNA-transfected ECA-109 cells, and SREBP1 expression level was increased in the plasmid-treated KYSE-150 cells (Fig. 1F). According to the results presented in Fig. S2, the most effective shRNA (sh1), Bank Id NM_004176, was selected for the follow-up experiments. Collectively, these results demonstrated that SREBP1 is highly expressed in ESCC tumors and cells.

The proliferation changes in ECA-109 and KYSE-150 cells were detected using a CCK-8 assay, and Ki-67 expression levels were investigated to explore the role of SREBP1 in the progression of ESCC cells. Results demonstrated that knockdown of SREBP1 significantly inhibited the cell viability of ECA-109 cells, although this was not as notable at 2 days after seeding compared with 3 days after seeding (Fig. 2A). Furthermore, this suppressive effect of proliferation was further confirmed by analysis of Ki-67 expression (Fig. 2B and C). SREBP1 ectopic overexpression effectively increased the cell viability and proliferation of KYSE-150 cells compared with the control group (P<0.05; Fig. 2A-C).

Whether SREBP1 affects the EMT process in ESCC was also investigated. Notably, SREBP1-overexpressing KYSE-150 cells became scattered and adopted the typical spindle-like morphology of mesenchymal cells (Fig. S3). The expression levels of multiple EMT-related factors post-SREBP1-knockdown or overexpression were analyzed using western blotting. In agreement with the phenotypic changes associated with SREBP1-overexpression, the expression levels of mesenchymal markers, including N-cadherin and Vimentin, were increased and the expression levels of the epithelial marker E-cadherin were decreased. Opposite results were observed on SREBP1-knockdown (Fig. 2D). The expression patterns of two other typical EMT-associated transcription factors, Snail1 and Slug, were also significantly reduced in SREBP1-knockdown cells and significantly increased in SREBP1-overexpressing KYSE-150 cells (P<0.01; Fig. 2E). The colocalization of E-cadherin and Vimentin, compared to each other under the same views was assessed using immunofluorescence analysis as described previously (7,40). Following SREBP1-knockdown and overexpression to further confirm that EMT is regulated by SREBP1 ECA-109 cells with SREBP1-knockdown exhibited increased E-cadherin expression levels compared with the negative control group, whereas the upregulation of SREBP1 in KYSE-150 cells enhanced Vimentin expression levels compared with the control group (Fig. 2F). These results suggested that SREBP1 promoted proliferation and induced EMT in ESCC cells.

Gene expression levels in ESCC cells were analyzed using RT-qPCR to elucidate the mechanisms by which SREBP1 modulates metastasis. As presented in Fig. 3A, the expression levels of the vascularization-related proteins VEGF-A and VEGF-C and the invasion and metastasis-related proteins MMP9, but not MMP2, were significantly enhanced after SREBP1-overexpression in KYSE-150 cells, but significantly downregulated after SREBP1-knockdown in ECA-109 cells (P<0.05; Fig. 3A). Furthermore, these expression levels were detected using ELISA, and the results revealed that the expression levels of MMP9, VEGF-A and VEGF-C were significantly suppressed in SREBP1-knockdown ECA-109 cells, but significantly upregulated in SREBP1-overexpressing KYSE-150 cells (P<0.001; Fig. 3B). Wound healing and Transwell assays were then performed to ascertain whether SREBP1 affected the migration and invasion abilities of ESCC cells. The wound healing assay revealed that ECA-109 cells with SREBP1-knockdown demonstrated significantly slower closure of the wound area compared with the control cells. Conversely, SREBP1-overexpression increased the migration of KYSE-150 cells (P<0.001; Fig. 3C). Moreover, the Transwell invasion assay revealed that silencing SREBP1 significantly reduced the number of cells on the membrane filters compared with those in the control, whereas overexpressing SREBP1 significantly increased the number of invaded cells (P<0.01; Fig. 3D). Collectively, these results suggested that SREBP1 contributes to the migration and invasion capacity of ESCC cells.

SREBP1 is associated with the regulation of fatty acid synthesis in response to inhibition of β-catenin (30). Thus, SREBP1 may be involved in cancer cell proliferation and metastasis via the Wnt/β-catenin signaling pathway. The activity of TCF/LEF in Wnt signaling was analyzed to confirm this hypothesis. TCF1 and LEF1 were markedly decreased in SREBP1-knockdown ECA-109 cells compared with the negative control, suggesting that knockdown of SREBP1 inhibited TCF/LEF activity. By contrast, overexpression of SREBP1 markedly elevated TCF1 and LEF1 levels in KYSE-150 cells (Fig. 3E). The expression levels of GSK-3β and β-catenin were further investigated to evaluate the association between SREBP1 and Wnt/β-catenin signaling. As illustrated in Fig. 3E, the expression level of p-GSK-3β was reduced by the knockdown of SREBP1 compared with the control. The expression levels of β-catenin were also markedly decreased in ECA-109 SREBP1-depleted cells compared with the control. The observations in ECA-109 cells demonstrated that knockdown of SREBP1 notably decreased the expression levels of p-GSK-3β, whereas expression levels of total GSK-3β were not changed (Fig. 3E). Furthermore, the expression levels of Wnt pathway downstream targets, including c-Myc, c-Jun, Met, MMP7, CD44 and cyclin D1, were also examined. Downregulating SREBP1 in ECA-109 cells reduced the expression levels of cyclin D1 and CD44 (Fig. 3E) but had no observable effect on c-Myc, c-Jun, MMP7 and Met levels (Fig. S4). Conversely, SREBP1-overexpression in KYSE-150 cells increased the expression levels of CD44 and cyclin D1. Moreover, the aforementioned hypothesis was further confirmed in ECA-109 cells treated with the Wnt/β-catenin pathway-specific antagonist ICG-001. In ECA-109 cells, ICG-001 markedly inhibited the Wnt/β-catenin pathway, as indicated by the reduced expression levels of nuclear β-catenin, Snail, cyclin D1 and Ki-67. As expected, the expression levels of these proteins were further decreased after adding ICG-001 to the SREBP1 interference group (Fig. 3F). Furthermore, ICG-001 markedly inhibited the viability of KYSE-150 cells overexpressing SREBP1 in a dose-dependent manner (Fig. 3G). Overall, these observations indicated that the Wnt/β-catenin signaling pathway serves a functional role in SREBP1-induced proliferation, migration and invasion in ESCC.

The expression levels of SCD1, which is the target gene of SREBP1 and known regulatory factor of the Wnt/β-catenin pathway, were examined to further elucidate the regulatory effect of SREBP1 on Wnt/β-catenin signaling (31,32). As presented in Fig. 4A, SCD1 was elevated in KYSE-150 cells overexpressing SREBP1, and SREBP1-knockdown resulted in a lower expression level of SCD1 in ECA-109 cells. The effect of SCD1 siRNA on the activation of Wnt/β-catenin induced by SREBP1 was also assessed. SCD1-knockdown markedly reduced the nuclear β-catenin expression levels regardless of whether SREBP1 was overexpressed or not, whereas there was no effect on the expression levels of SREBP1 (Fig. 4B). KYSE-150 cells were cultured in the absence or presence of Fatostatin (10 µM) and Mevastatin (10 µM) for 24 h to verify whether pharmacological inhibition of SREBP1 downregulates SCD1. Western blotting confirmed that Fatostatin, an inhibitor of SREBP that directly binds SREBP cleavage-activating protein and blocks its endoplasmic reticulum-to-Golgi transport (45), reduced the expression levels of mSREBP1 and SCD1 in the nucleus but not in the cytoplasm, and blocked the nuclear translocation of β-catenin (Fig. 4C). In addition, regardless of whether SREBP1was overexpressed, mevastatin that inhibits HMGCR downstream of SREBP1 decreased the expression levels of SCD1 and β-catenin in the nucleus (Fig. S5). Furthermore, the activation of the Wnt/β-catenin pathway was measured when SREBP1 and SCD1 were inhibited by Fatostatin. As expected, Fatostatin reversed SREBP1-mediated growth promotion in a dose-dependent manner (Fig. 4D), and the expression levels of VEGF-A and MMP9 were significantly decreased (Fig. 4E) (46). Hence, these results suggested that activation of the Wnt/β-catenin pathway by SCD1 is essential for the proliferation, EMT and metastasis of ESCC cells driven by constitutively active SREBP1.

A xenograft tumor model in nude mice was established using the SREBP1-knockdown ECA-109 cells to investigate whether SREBP1 silencing inhibits ESCC cell proliferation and metastasis in vivo. As presented in Fig. 5A and B, xenograft models inoculated subcutaneously using cells transfected with shSREBP1 had reduced abilities to form tumors compared with those in control mice, and the mean final volume of tumors transfected with shSREBP1 was 3.64-fold smaller compared with the control group 28 days after implantation. The results demonstrated that the differences in tumor growth curves were significant between the ECA-109-shSREBP1 and control groups after inoculation for 20 days (P<0.01), and the maximum tumor diameter (14.84 mm) and volume (428 mm³) were observed in the control group (Fig. 5C). In general, these findings demonstrated that SREBP1-knockdown suppressed proliferation in ESCC. IHC analysis was also performed on xenograft sections. As indicated in Fig. 5C, lower SREBP1 expression levels were observed in the tumors derived from ECA-109-shSREBP1 cells, indicating that SREBP1-knockdown was maintained in vivo. Moreover, the expression levels of β-catenin, MMP9, VEGF-C and Ki-67 were repressed in the SREBP1-knockdown xenografts compared with the ECA-109-shNC xenografts (Fig. 5D). This phenomenon suggested that SREBP1-knockdown induced a reduction in MMP9, VEGF-C and Ki-67 expression levels, promoting tumor growth and metastasis in vivo, and these data were with the aforementioned in vitro findings. Even the expression levels of SCD1 and p-GSK-3β were elevated in the partial region of tumors, but the overall trend showed no difference between the two groups. Furthermore, ECA-109-shSREBP1 and negative control cells were injected into the tail veins of nude mice to reveal whether SREBP1-knockdown could affect the metastatic potential of ESCC cells in vivo. The mice were sacrificed 6 weeks after injection, and the number of metastatic nodules in the lung was counted. H&E staining demonstrated that the silencing of SREBP1 significantly decreased the number of metastatic nodules in the lung compared with that the vector group (P<0.001; Fig. 5E). These findings suggested that SREBP1-knockdown suppressed the proliferative and metastatic properties of ESCC cells in vivo.