The pathological specimens of 74 ESCC patients were obtained from Tianjin Medical University General Hospital. Each specimen included esophageal carcinoma and adjacent normal tissue (distance from cancer tissue >5 cm). None of the patients received chemotherapy, radiotherapy, or other related antitumor therapies before surgery. The final follow-up time was December 31, 2020. In this study, The pathological types of LGIN, HGIN, and esophageal squamous cell carcinoma in the present study were determined by independent diagnosis by two pathologists (21, 22). The tumor stage and grade classification were based on the 8th American Joint Committee on Cancer (AJCC). This research was in accordance with the Helsinki Declaration of the World Medical Association. Every patient signed an informed consent form. The experiment was approved by the Ethics Committee of Tianjin Medical University General Hospital. The clinicopathological characteristics of patients are shown in Table 1 .

Human ESCC cell lines (Eca-109 and KYSE-150) were purchased from the Cell Bank of the Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences. Eca-109 and KYSE-150 cells were cultured in DMEM containing 10% fetal bovine serum (Gibco, Life Technologies, Rockville, MD, USA) and incubated at 37°C in a 5% carbon dioxide incubator at constant temperature.

Eca-109 and KYSE-150 cells were seeded into 6-well plates at 5 × 10⁵ cells/well for transfection. The culture density was 50%–60%. According to the instructions of Lipofectamine 2000 (Life Technologies, Rockville, MD, USA), two different siRNAs against Transgelin and corresponding negative control siRNAs were transfected into Eca-109 and KYSE-150 cells. The final transfection concentration was 100 nmol/L. After 6 h of culture in serum-free medium, the culture was transferred to 10% serum medium for 24 h. Cells were collected for subsequent experiments. Transgelin-overexpressing Transgelin plasmids (Transgelin cDNA ORF Clone, Human, pCMV3-N-HA as carrier, SinoBiological, China) and no-load control plasmids were transfected with the same method. The final transfection concentration was 1mg/L. The experimental cells were divided into the following groups: Vector-NC (empty plasmid), Transgelin (Transgelin-overexpressing plasmid), si-NC (negative control siRNA), and si-Transgelin (Transgelin siRNAs). All cell phenotype experiments are completed within 48 hours after the completion of transfection.

The transfected cells in the growth phase were taken. Cells were seeded into 96-well plates with 5000 cells per well. After conventional culture for 24, 48, and 72 h, 10 μL of CCK-8 solutions (Beyotime, Shanghai, China) was added to each well. After incubation at 37°C for 1 h, the absorbance (A) value of each well at 450 nm was detected with a microplate analyzer (Multiskan EX, Lab systems, Helsinki, Finland), and the cell viability was calculated.

Before the wound scratch, incubate the cells with serum-free medium overnight to inhibit cell proliferation on migration. After digestion, the cells were collected and seeded into 6-well plates with 2 × 10⁵ cells/well for culture. When the fusion degree of cell culture reached 100%, the cell center was scratched linearly. Next, 1× phosphate buffer solution (PBS) was used to wash the cell suspension. Serum-free DMEM (Life Technologies) was added to each well. The cells were cultured at 37°C in a cell incubator under 95% relative humidity and 5% CO₂. Changes of cell migration were observed at 0, 12, 24, and 36 h. Each experiment was repeated three times.

Matrigel (BD, Biosciences, Franklin Lakes, NJ, USA) was dissolved overnight at 4°C and diluted in complete medium (1:3). About 50 μL of the mixture was added to the upper and lower compartment of each 24-well plate (Millipore, Billerica, MA, USA). Then, the samples were placed into a 37°C cell incubator and set for 30 min to coagulate. Logarithmic growth cells were digested and collected. About 2 × 10⁴ cells were washed in 200 μL of serum-free medium. Resuspended cells were added to the upper compartment. About 600 μL of complete medium was added to the lower chamber. The cell culture plates were cultured at 37°C in an incubator under 95% relative humidity and 5% CO₂. After 24 h, Matrigel and cells in the upper compartment were removed. The submembrane compartment surface was stained with crystal violet and photographed for cell counting. Nikon (Japan) frontal microscope was used to randomly select 10 fields to count the number of cells, and the average value was obtained. Each experiment was repeated three times.

Unlike the invasion assay, the Transwell migration assay did not require Matrigel-embedded chambers for 1 h before cell inoculation. The remaining steps are the same as those in the Transwell invasion experiment.

Cells in each group were collected, and total RNA was extracted by Trizol method. cDNA was synthesized using a reverse transcription kit (Takara, Japan). The amplification was performed using a real-time fluorescent quantitative PCR apparatus in accordance with the instructions of the SYBR Green Kit (Invitrogen, Carlsbad, CA, USA). PCR was used to amplify Transgelin, E-cadherin, Vimentin, and GAPDH. The sequence was as follows: Transgelin upstream primer: 5′-GGTGGAGTGGATCATAGTGC-3′, downstream primer: 5′-ATGTCAGTCTTGATGACCCCA-3′; E-cadherin upstream primer: 5′-GACGCG GACGATGATGTGAAC-3′, downstream primer: 5′-TTGTACGTGGTGGGATTG AAGA-3′; Vimentin upstream primer: 5′-GACAATGCGTCTCTGGCACGTCTT-3′, downstream primer: 5′-TCCTCCGCCTCCTGCAGGTTCTT-3′; internal reference upstream primer: 5′-CCCTTCATTGACCTCAACTACATGG-3′, downstream primer: 5′-CATGGTGGTGAAGACGCCAG-3′. The reaction conditions were as follows: 95°C for 30 s; 95°C for 8 s, 60°C for 32 s, 40 cycles; 95°C for 1 min, 60°C for 30 s, 95°C for 30 s. The 2^-ΔΔCT method was used to calculate the relative gene expression. A single experiment was repeated for three times, and three duplicate wells were set in each experiment.

The total cell protein was extracted and quantified by BCA method. The sample quantity was 40 μg. The sample was separated by 10% SDS-PAGE. The membrane was transferred to the PVDF membrane by using the wet method (1 kDa/min), and the primary antibody was added after 1 h with TBST sealant containing 5% skim milk powder. Primary antibody was added and incubated overnight at 4°C. Anti-TAGLN/Transgelin (ab233971), abcam; Vimentin (D21H3) XP^® Rabbit mAb #5741, CST; E-Cadherin (24E10) Rabbit mAb #3195, CST; N-Cadherin (D4R1H) XP^® Rabbit mAb #13116, CST; Claudin-1 (D5H1D) XP^® Rabbit mAb #13255, CST; Rabbit anti-Catenin-gamma Polyclonal Antibody, abs131596, absin; Mouse Anti-β actin mAb, TA-09 ZSGB-BIO, Beijing, China; Mouse Anti-GAPDH mAb, TA-08 ZSGB-BIO, Beijing, China. The dilution ratio of antibody was 1:1000. The secondary antibody (Horseradish enzyme labeled goat anti-rabbit IgG (H+L), ZB-2301, ZSGB-BIO; Horseradish enzyme labeled goat anti-mouse IgG (H+L) ZB-2305, ZSGB-BIO) was added after washing the membrane with TBST and incubated at room temperature for 1 h. ECL glowed after washing the TBST film. The semi-quantitative analysis of the absorbance of the strips was carried out with ImageJ software. The ratio of the absorbance value of the target protein band to the absorbance value of the internal reference GAPDH and β-actin protein band was used to represent the relative amount of the target protein.

All specimens were fixed in 10% neutral formaldehyde. Paraffin embedded, 4 μm continuous section. Xylene dewaxing, gradient ethanol hydration. Antigen thermal repair with 0.01 mol/L sodium citrate buffer. Endogenous peroxidase was sealed at 37°C for 20 min. Primary antibody (Transgelin, 1:200, Abcam; CD31, 1:1000, Abcam) was added and incubated overnight at 4°C. The secondary antibody was added and incubated at 37°C for 30 min. Horseradish peroxidase was added and incubated at 37°C for 20 min. DAB staining, hematoxylin redyeing, gradient ethanol dehydration. Xylene transparent, neutral gum sealed sheet, light microscope detection and photography. The staining results were graded according to the staining intensity and percentage of positive cells. The percentages of positive cells <5%, 5%–25%, 26%–50%, 51%–75%, and >75% were 0, 1, 2, 3, and 4, respectively. The cell staining intensity was scored as follows: 0 for non-staining, 1 for light yellow, 2 for brownish yellow, and 3 for yellowish brown. The positive intensity is the product of two scores: 0–2 is negative, 3–5 is weakly positive, 6–8 is positive, and 9–12 is strongly positive. The results were determined by two pathologists, and the average value was taken.

CancerSEA (http://biocc.hrbmu.edu.cn/CancerSEA/home.jsp) is a comprehensive database that provides the single-cell functional states of cancer cells at different sites. In this study, we investigated the correlation between Transgelin gene and different functional states of tumors and different cancer types through this website. The average correlation between Transgelin and functional status in different cancers, including invasion, differentiation, metastasis, cell cycle, stemness, and proliferation, was analyzed through this website.

The data were constructed using the Kaplan–Meier Plotter (https://kmplot.com/analysis/) database based on the GEO, EGA, and TCGA public database of gene chip and RNA-seq. The effects of 54,675 genes on survival in 21 types of cancer were assessed. The Kaplan–Meier Plotter database integrates gene expression information and clinical prognostic value for meta-analysis and survival-related molecular marker research, discovery, and validation. This website was used to analyze the prognostic value of Transgelin in different tumor types. In the analysis, the Kaplan–Meier Plotter database divided patients into two groups on the basis of different quantiles of Transgelin expression. The Kaplan–Meier survival chart was used to compare two cohorts and calculate the HR, 95% CI, and log rank P values.

The study of the interaction network between proteins is helpful to mine the core regulatory genes. In this study, STRING (https://string-db.org/) was used to analyze the PPI network relationship of Transgelin interacting proteins. The STRING database searches for known and predicted interactions between proteins. The database can be applied to 2031 species, containing 9.6 million proteins and 13.8 million protein interactions. In addition to experimental data, mined results from PubMed abstracts, and integrated data from other databases, the STRING database also contains the predicted results using bioinformatics methods.

In this study, the LinkedOmics database was used to analyze the co-expressed genes of Transgelin and their participation in GO and KEGG enrichment analysis. Login LinkedOmics Database (http://linkedomics.org/login.php), the first to register to obtain access to the database. Then, screening conditions were selected successively in the database retrieval interface, and data related to Transgelin gene expression level and prognosis in ESCC were downloaded.

GraphPad Prism 7 software (GraphPad Software, Inc., San Diego, CA, USA) was used for statistical analysis. The measurement data were expressed as mean ± standard deviation. Two-tailed unpaired t-test was used for the comparison between two groups. One-way ANOVA was used for the comparison between multiple groups, and Tukey’s test was used for further pairwise comparison. The Pearson correlation coefficient was used for correlation analysis. The Kaplan–Meier method was used to calculate the relationship between Transgelin expression and survival prognosis of tumor patients. Statistical significance was considered at P < 0.05.

We analyzed the average correlation between Transgelin and functional status, including invasion, differentiation, metastasis, cell cycle, stemness, and proliferation, in different cancers through the CancerSEA database. The analysis results show that the expression level of Transgelin is negatively correlated with invasion, differentiation, metastasis, cell cycle, stemness, and proliferation ( Figures 1A, B ). These results indicated that the high expression of Transgelin could reduce the invasion and metastasis of tumors. Furthermore, we analyzed the distribution of Transgelin expression in tumors and the t-SNE diagram of all individual cells, in which the color represented the expression level of the input gene. The analytical results showed that cells with low Transgelin expression were clustered together ( Figure 1C ). The expression results of Transgelin in ESCC and paracancerous tissues in the GEPIA Database (http://gepia.cancer-pku.cn/index.html) are shown in Figure 1D . The results showed that the expression level of Transgelin in ESCC tissues was lower than that in normal esophageal tissues in the dataset of ESCC collected by TCGA (P < 0.05). Through the above database mining information, we found that the expression of Transgelin was decreased in ESCC tissues. To further clarify the relationship between Transgelin expression and prognosis of ESCC, we analyzed the correlation between Transgelin expression and prognosis in patients with ESCC in the Kaplan–Meier Plotter database. Kaplan–Meier analysis showed that the expression level of Transgelin gene was correlated with the survival prognosis of ESCC patients. Patients with high Transgelin expression had a lower overall mortality, while patients with low Transgelin expression had poorer prognosis, and the difference was statistically significant ( Figure 1E , P < 0.05). Prognostic analysis of Transgelin in breast cancer, liver hepatocellular carcinoma, and sarcoma also showed that patients with high Transgelin expression had a long survival time ( Supplementary Figure 1 ).

The qRT-PCR results showed that Transgelin mRNA expression level in Eca-109 and KYSE-150 cells transfected with Transgelin-overexpressing plasmid was significantly higher than that in the empty plasmid group, and the difference was statistically significant ( Figure 2A , P < 0.01). Conforming to the qRT-PCR results, the Western blot results showed that Transgelin protein expression levels in Eca-109 and KYSE-150 cells transfected with Transgelin-overexpressing plasmids were also significantly higher than those in the control group, and the difference was statistically significant ( Figure 2B , P < 0.01). The results indicated that the Transgelin plasmid used in the experiment could effectively promote the expression of Transgelin in vitro. The CCK-8 results showed that Transgelin overexpression in Eca-109 and KYSE-150 cells significantly reduced the proliferation capacity of cells, and the difference was statistically significant ( Figures 2C, D , P < 0.05). With the extension of transfection time, the inhibitory effect of Transgelin overexpression on the proliferation of Eca-109 and KYSE-150 cells increased continuously, and the difference was statistically significant (P < 0.05). These results indicated that the overexpression of Transgelin gene could effectively inhibit the proliferation ability of Eca-109 and KYSE-150 cells in vitro. The scratch test results showed that after Transgelin overexpression of Eca-109 and KYSE-150 cells, the migration ability of the overexpression group was significantly decreased compared with that of the blank control group, and the difference was statistically significant ( Figures 2E, F , P < 0.05). Transgelin experiment results showed that after Transgelin overexpression in Eca-109 and KYSE-150 cells, the number of transmembrane-penetrating cells decreased significantly compared with the blank control group ( Figures 2G, H , P < 0.05). These results indicated that the promotion of Transgelin expression could effectively inhibit the invasion and migration of Eca-109 and KYSE-150 cells in vitro.

The RT-PCR results showed that the expression of E-cadherin in Eca-109 cells was significantly increased with the increase of E-cadherin expression. However, the expression level of Vimentin was significantly decreased ( Figure 3A ). The results in KYSE-150 cells were consistent with those in Eca-109 ( Figure 3B ). Furthermore, the protein levels of E-cadherin, Claudin-1, N-cadherin, γ-Catenin and Vimentin after Transgelin overexpression were detected by Western blot. The experimental results showed that after Transgelin overexpression in Eca-109 and KYSE-150 cells, the content of E-cadherin and γ-Catenin protein in the cells increased, whereas the expression of Vimentin, N-cadherin and Claudin-1 protein in the cells decreased accordingly; the difference was statistically significant ( Figures 3C, D , P < 0.05). These results indicated that Transgelin could inhibit EMT by regulating the expression of epithelial phenotype and mesenchymal phenotype in Eca-109 and KYSE 150 cells.

Transgelin mRNA and protein expression levels in Eca-109 and KYSE-150 cells transfected with siRNA were significantly lower than those in the negative control group, and the difference was statistically significant ( Figures 4A, B , P < 0.01). This finding indicated that the Transgelin siRNA used in the experiment could effectively inhibit the expression of Transgelin gene in vitro. The CCK-8 results showed that the cell proliferation rate was significantly enhanced after transfection of Transgelin siRNA in Eca-109 and KYSE-150 cells, and the difference was statistically significant ( Figures 4C, D , P < 0.05). Moreover, with the extension of transfection time, the promotion of Transgelin siRNA on the proliferation of Eca-109 and KYSE-150 cells was enhanced. These results indicated that inhibiting Transgelin gene expression could effectively promote the proliferation ability of Eca-109 and KYSE-150 cells in vitro. The scratch test results showed that after Eca-109 and KYSE-150 cells were transfected with Transgelin siRNA, the migration ability of the interference group was significantly enhanced compared with that of the negative control group, and the difference was statistically significant ( Figures 4E, F , P < 0.05). The transwell experiment results showed that Transgelin siRNA transfection in Eca-109 and KYSE-150 cells significantly increased the number of transmembrane cells ( Figures 4G, H , P < 0.05). These results indicated that inhibiting Transgelin gene expression could effectively promote the invasion abilities of Eca-109 and KYSE-150 cells in vitro. To avoid the off-targets effects, another siRNAs against Transgelin was also used and completed related experiments, including cell transfection, qRT-PCR, Western Blot, CCK-8 assay, Cell scratch test, and Transwell assay. The results have been added to the Supplementary Figure 3 . Those results showed that inhibit of Transgelin promoted the proliferation, migration, and invasive abilities of esophageal squamous cell carcinomas cells in vitro.

First, we used RT-PCR to detect the changes in the expression levels of E-cadherin and Vimentin after Transgelin knockdown. The results showed that Transgelin inhibition significantly reduced the expression of E-cadherin in Eca-109 and KYSE-150 cells. By contrast, the expression level of Vimentin was significantly increased ( Figures 5A, B ). Furthermore, the protein levels of E-cadherin, Claudin-1, N-cadherin, γ-Catenin and Vimentin after Transgelin overexpression were detected by Western blot. The results showed that Transgelin knockdown in Eca-109 and KYSE-150 cells reduced the expression of E-cadherin and γ-Catenin protein. However, the expression of Vimentin, N-cadherin and Claudin-1 proteins in cells was upregulated, and the difference was statistically significant ( Figures 5C, D , P < 0.05). This finding suggested that Transgelin knockdown promoted EMT in ESCC. The same results were obtained using si-TAGLN-2 to knock out TAGLN ( Figure S4 ).

The expression of Transgelin in different genders, ages, invasion ranges, and clinical stages is shown in Table 1 . Statistical analysis showed that the expression of Transgelin in ESCC was not significantly correlated with age, gender, and invasion range of patients (P > 0.05), but was significantly correlated with lymph node metastasis, distant metastasis, and clinical stage ( Table 1 , P < 0.05). We further detected the expression changes of Transgelin in normal esophageal tissues, LGIN, HGIN, and tumor by immunohistochemistry ( Figure 6A ). The results showed that the expression of TGLN expression is lower in tumors compared to normal, LGIN and HGIN tissues. The expression level of Transgelin protein in these four tissues showed a continuous declining trend. However, no statistically significant difference was observed in the expression of LGIN in normal esophageal mucosa tissues ( Figure 6B , P > 0.05). To further clarify the relationship between Transgelin expression and prognosis of ESCC, we followed up the patients using the information collected by the research team. Kaplan–Meier analysis showed that the expression level of Transgelin gene was correlated with the survival prognosis of ESCC patients. Patients with high Transgelin expression had a longer overall survival, whereas those with low Transgelin expression had a poorer prognosis, and the difference was statistically significant ( Figure 6C , P < 0.05).

In this study, the public dataset from LinkedOmics database was used to further analyze the expression of Transgelin gene in ESCC and its clinical significance. First, we analyzed the co-expressed genes of Transgelin. Figure 7A shows a volcanic map information of Transgelin co-expressed genes. Figures 7B, C show heat maps of genes positively and negatively correlated with Transgelin expression, respectively. Through further analysis, we found that Transgelin co-expression was correlated with CNN1, DACT3, MRVI1, and PRELP ( Figure 7D ). The results of Transgelin interaction protein network analysis showed that Transgelin was at the center of the interaction. This finding indicates that Transgelin plays an important role ( Supplementary Figure 1 ). Furthermore, we used GO and KEGG analysis functions in LinkedOmics database to enrich Transgelin co-expressed genes. The results of GO enrichment analysis showed that the enriched biological processes mainly included developmental process, metabolic process, cell proliferation, and cell communication. Moreover, the affected molecular functions mainly included protein binding, ion binding, nucleotide binding, transporter activity, and transferase activity ( Figures 7E–G ). In addition, the pathways affected by Transgelin co-expressed genes mainly included the cGMP–PKG signaling pathway, Wnt signaling pathway, and TGF-β signaling pathway ( Figure 7H ).