Plasma CircCYP24A1 as a Novel Biomarker of Esophageal Squamous Cell Carcinoma

Specimen Collection

This study collected plasma samples from 70 patients with ESCC, 8 patients with esophagitis and 40 healthy individuals. CircCYP24A1 has been reported to be upregulated in prostate cancer (PCa) ¹⁹ and cutaneous squamous cell carcinoma (cSCC). ²⁰ In China, the age-standardized incidence rate for PCa is 9.68%, while non-melanoma skin cancer is much lower. ²¹ Therefore, we also collected the plasma of 5 PCa patients in order to verify the expression of circCYP24A1 in the plasma of patients with other tumor types. All the volunteers were sourced from the Second Clinical Medical College of North Sichuan Medical College. All the samples were obtained after getting informed consents of patients and/or their legal guardian, and the utilization of clinical specimens has been granted approval by the Ethics Committee of North Sichuan Medical College in accordance with the Declaration of Helsinki on June 22, 2021 (Approval number: [2021]42) in Nanchong, China. The staging of ESCC follows the eighth edition of the UICC/AJCC TNM staging system for esophageal cancer, established in 2017. The “LNM positivity criterion” was established based on histopathological findings and imaging examination. Pathological biopsy of lymph nodes revealing ESCC cells, or ultrasound examination of supraclavicular lymph nodes and CT scanning of neck, chest and abdomen regions revealing abnormal lymph node size, morphology, echo characteristics, blood flow, density, enhancement mode and other manifestations such as enlarged lymph nodes, a morphological index >0.5, irregular morphology, blurred edges, disappearance of the portal structure, uneven echo, rich blood flow, irregular circular enhancement with central low-density shadow, etc were considered positive for LNM. N0 indicates the absence of regional lymph node involvement, while N1-N3 represent varying degrees of LNM.

The inclusion criteria for the healthy control group were adult individuals who had no evident abnormalities in chest X-ray or chest CT scans, gastroscopy, as well as tumor marker tests. The patient group of ESCC patients consisted of untreated primary ESCC patients diagnosed by histopathological examination, aged 18 to 80 years, with a clear LNM area and a complete medical history. The exclusion criteria for ESCC patients were patients with an unclear medical record diagnosis, patients with non-primary ESCC or primary non-squamous cell carcinoma, those with serious underlying disease, those who had received radical resection surgery or chemoradiotherapy, patients with diseases causing pathological changes of lymph nodes such as infection, autoimmune diseases, tuberculosis, etc, and those unable to complete the follow-up. The inclusion and exclusion criteria for Pca patients were consistent with those of ESCC. And patients with esophagitis were diagnosed through gastroscopy and brush biopsy.

Peripheral Blood Collection

Using EDTA anticoagulant vacuum collection tubes (purple tubes), approximately 2 mL of fasting venous blood was collected from patients and healthy adults in the morning, followed by centrifugation in a low-temperature high-speed centrifuge at 4 °C, 1900 ×g (3000 rpm) for 10 min. After centrifugation, the supernatant (plasma) was aspirated with a pipette and transferred to RNase-free tubes, which were then stored in a −80 °C freezer for future use.

Cell Culture and Transfection

ESCC cell lines TE-1 (CSTR TCHu89, RRID: CVCL_1759) and KYSE-150 (CSTR TCHu236, RRID: CVCL_1348) were obtained from the Cell Bank/Stem Cell Bank, Chinese Academy of Sciences (Shanghai, China). KYSE-410 (DSMZ ACC-381, RRID: CVCL_1352) was purchased from Fenghui Biological Co. Ltd (Changsha, China), which introduced KYSE-410 from German Collection of Microorganisms and Cell Cultures. Human esophageal epithelial cell line Het-1A (ATCC CRL-2692, RRID: CVCL_3702) and ESCC cell line KYSE-510 (JCRB JCRB1436, RRID: CVCL_1354) were obtained from Mengjin Biological Co. Ltd (Shanghai, China), which introduced KYSE-510 from the Japanese Collection of Research Bioresources Cell Bank and Het-1A from the American Type Culture Collection. All cell lines were authenticated by STR profiling and free of mycoplasma contamination. Cells were cultured in RPMI-1640 medium or DMEM (GIBCO, USA) supplemented with 10% fetal bovine serum (GIBCO, USA) at 37 °C in a 5% CO₂ incubator. SiRNA targeting circCYP24A1 was designed and synthesized by GenePharma (Shanghai, China), the sequence of si-circCYP24A1 was as follows: sense AUAAUACGCCUCAGGGAAGTT, antisense CUUCCCUGAGGCGUAUUAUTT. After cells reached 60% confluency following plating, si-circCYP24A1 was transfected into the cells using Lipofectamine™ 2000 (Invitrogen, USA).

RNA Sequencing (RNA-seq)

A total of six matched ESCC tissues and normal tissues surrounding the tumor were collected for RNA-seq,^22,23 the obtained fresh tissue had been cryogenically preserved in liquid nitrogen within 30 min after surgery for future use. And all the volunteers had received no prior treatment. Total RNA extraction and sequencing were performed by Aksomics Inc. (Shanghai, China) with standard procedures. The total RNA was checked by agarose electrophoresis and quantified by nanodrop. After removing rRNA and digesting linear RNA, the circRNA sequencing library was constructed by the KAPA Stranded RNA-Seq Library Prep Kit (Illumina, USA), and sequenced by Illumina Hiseq 4000 (Illumina, USA). The sequencing data were uploaded in the NCBI's Gene Expression Omnibus database (GSE263647). The junction sites of circRNAs were detected by comparing the sequencing data to the known transcriptome in the STAR, and then the expression quantification of circRNAs (the backsplice junction reads count) was calculated by CIRCexplorer2. The circRNAs with a CPM mean above 100 in each group were considered as expressed in the group.

Quantitative Real-Time Reverse Transcription-Polymerase Chain Reaction (qRT-PCR)

Plasma and cell total RNA were extracted using the MiRNeasy Serum/Plasma Kit (QIAGEN, Germany) and the FastPure Cell/Tissue Total RNA Isolation Kit (Vazyme, China), respectively. Reverse transcription of the RNA into complementary DNA (cDNA) was carried out using the Hiscript III first strand cDNA Synthesis Kit (Vazyme, China). PCR was conducted using the AceQ qPCR SYBR Green Master Mix (Vazyme, China).

Detection and Analysis of the Plasma circCYP24A1

Each 200 μL of plasma was extracted to obtain 10 μL of total RNA, all of which were reverse transcribed to obtain 20 μL of cDNA. The undiluted cDNA was amplified according to the cycle reaction parameters shown in Supplementary Table 2. In addition, total RNA from KYSE-510 cells was extracted, and 1μg of RNA was reverse transcribed, and the resulting cDNA was diluted 10 times and used for future use. The primer sequences were displayed in Supplementary Table 3. With GAPDH as the internal reference, total RNA of KYSE-510 cells was introduced into each batch of samples during the qRT-PCR process as the control group to facilitate data normalization. The relative changes in gene expression were determined using the 2^−ΔΔCt method. We collected clinical data of 70 ESCC patients, divided them into low expression group (n = 35) and a high expression group (n = 35) based on the median expression level of plasma circCYP24A1, and analyzed the correlation between clinical data and plasma circCYP24A1. Patients were followed up once a month from the time of blood collection for 13 months to analyze the effect of plasma circCYP24A1 content on patients’ survival, and the follow-up rate was 98.6%. The receiver operating characteristic (ROC) curve analysis was used to calculate the area under the curve (AUC), sensitivity, and specificity to evaluate the ability to distinguish ESCC patients from health control groups and predict LNM.

RNase R Assay

RNase R (Biosearch, USA) is used to digest RNA. 1 μg of RNA was treated with 3 U of RNase R at 37 °C for 30 min. The changes of circCYP24A1 and CYP24A1 abundance before and after digestion were detected by qRT-PCR.

Tissue Microarrays (TMA) and Fluorescence in Situ Hybridization (FISH) Assay

The TMA (HEsoS180Su11) was obtained from Shanghai Outdo Biotech Co., Ltd (Shanghai, China) with the tissue samples obtained from the National Human Genetic Resources Sharing Service Platform. The chip contained data on 108 cases of ESCC with 72 cases having corresponding adjacent non-cancerous tissue. The human tissues obtained in TMA were approved by the Ethics Committee of Shanghai Outdo Biotech company in accordance with the Declaration of Helsinki on April 1, 2020 (Approval number: SHYJS-CP-2004003) in Shanghai, China. The data was anonymous, and thus the requirement for informed consent was waived. Probes mixes were designed for tissue FISH assay targeting circCYP24A1. The probe mixes sequences are displayed in Supplementary Table 3. The RNA FISH kit SA-Biotin System (paraffin sections) was purchased from GenePharma (Shanghai, China) and used according to the provided instructions. Panoramic multispectral scanning of slides was performed by the Tissue- FAXS Spectra (TissueGnostics, Austria) and we imported the data into Strata-Quest analysis software to calculate the positive rate of circCYP24A1. The results of FISH showed that one of the 108 cancer tissues lacked ESCC tissue and one of the 72 corresponding adjacent non-cancer tissues lacked normal esophageal epithelial tissue. So, there were a total of 107 ESCC tissues and 71 normal esophageal epithelial tissues, with 70 pairs of matched cancerous and adjacent normal epithelial tissues in our subsequent statistical analysis. Taking the median of the positive rate of circCYP24A1 as the cutoff value, ESCC patients were stratified by low (n = 54) and high (n = 53) expression of circCYP24A1, and correlation analysis was performed according to the collected clinical data and follow-up information. The 5-year survival rate is defined as the time from the date of surgery to the date of death or 60 months. The follow-up rate was 99.0%. The impact of different circCYP24A1 expression groups on the 5-year survival rate of ESCC patients was analyzed by survival analysis.

Actinomycin D Assay

To assess RNA stability, 2 μg/mL Actinomycin D was added to KYSE510 and KYSE150 cells when their fusion degree reached approximately 50%. The cells were treated with Actinomycin D for 4, 8, 12 and 24 h, respectively. A control group without intervention was set at each time point. After reaching the corresponding time points, the cells were collected and total RNA was extracted. Reverse transcription was performed to synthesize cDNA, followed by qRT-PCR to detect the molecular levels of circCYP24A1 and CYP24A2.

Cell Proliferation Assay (CCK-8)

Cell viability was assessed via the CCK-8 assay (Kyegen, China) following kit directions. In brief, 100 μL cells/well (approximately 3 × 10³ cells) were seeded in 96-well plates and incubated at 37 °C in a 5% CO₂ cell culture incubator. At different time points (0, 24, 48, 72, 96 h), after preparing the culture medium and CCK-8 reagent in a ratio of 10:1, 100 μL of the mixture was added to each well and incubated for 1 h in the dark at 37°C. Absorbance was measured at 450 nm via microplate reader.

Wound Healing Assay

Cells treated with siRNA were seeded in a 6-well plate. After 24 h of incubation, two perpendicular baseline lines were drawn in the cells using a 10 μL tip. Photographs were taken at 0 and 24 h after scratching. Scratch healing rate (%) = [(0 h scratch width-24 h scratch width)/ 0 h scratch width] × 100%.

Trans-Well Assay

100 μL of serum-free culture medium containing 3 × 10⁴ cells was added to the upper chamber of the trans-well. In the lower chamber, 600 μL of 1640 culture medium containing 20% fetal bovine serum was added. For the invasion assay, Matrigel (BD biosciences, USA) was diluted 1:4 with serum-free culture medium, and 40 μL of the mixture was evenly spread on the upper chamber. After being kept at a temperature of 37 °C for 1 h, the trans-well chamber containing the mixture added 100 μL of serum-free medium containing 5 × 10⁴ cells. After 24-48 h of incubation (48 h for KYSE-510, 24 h for KYSE-150), the cells were fixed with 20% methanol for 30 min, followed by staining with 0.1% crystal violet solution. The cells in the upper chamber were gently wiped off, photographed, and counted.

Western Blot Analysis

ESCC cells were lysed using RIPA protein lysis buffer, followed by a 30-min incubation on ice, centrifugation and collection of the supernatant. Proteins quantitated by BCA were subjected to 10% SDS-PAGE analysis. The proteins were transferred onto a PVDF membrane (Millipore, USA). The membrane was blocked with 5% skim milk for 1 h and incubated overnight at 4 °C with antibodies against N-cadherin (1:500, 22018-1-AP, Proteintech, China), E-cadherin (1:10000, #ET1607-75, HUABIO, China), snail (1:1000, #13099-1-AP, HUABIO, China), vimentin (1:500, #ET1610-39, HUABIO, China), and GAPDH (1:50000, ET1601-4, HUABIO, China). After incubation, the membrane was treated with HRP-conjugated secondary antibodies at room temperature for 1 h, washed with PBST and exposed in a darkroom after adding the ECL chromogenic solution.

Statistical Analysis

The experimental data were analyzed using GraphPad Prism 9.0 statistical software. Normally distributed measurement data were expressed as ( x ¯ ±s). Independent samples t-test was used for comparison between two groups, while ANOVA was used for comparison among multiple groups. χ2 test was used to evaluate the correlation between circCYP24A1 expression level and clinical data. Overall survival curves were obtained by Kaplan-Meier method and assessed with the log-rank test. P < .05 was considered statistically significant.

Identification and Characterization of the circCYP24A1

We had obtained the expression profile of circRNAs in ESCC tissues by conducting RNA-seq analysis on six pairs of ESCC tissues and adjacent normal tissues. A total of 2228 circRNAs were identified, and 131 down-regulated circRNAs and 54 up-regulated circRNAs were selected by the criteria of |Log₂FC|≥1 and P < .01 (Figure 1A-B). We focused on the up-regulated circRNAs in ESCC, as they exhibited higher expression levels which makes them more detectable in tissues and blood. CircCYP24A1 emerged as the most significantly up-regulated among the differentially expressed circRNAs (Figure 1B).

OPEN IN VIEWER

Figure 1.

Identifcation of the circCYP24A1 up-regulated in ESCC and its characterization. A Clustered heatmap for differentially expressed circRNAs. Cut of is |Log2FC|≥1 and P < .01. B Volcano plot compared the expression fold changes of circRNAs for six paired samples of ESCC by RNA-Seq. Right upper panel represented the up-regulated circRNAs, the dot pointed by the arrow indicate circCYP24A1. C The expression of circCYP24A1 in Het-1A and ESCC cells was detected by qRT-PCR. D The schematic representation of circCYP24A1 formation and Sanger sequencing results of the circCYP24A1 backsplice junction site. E The gel data indicating the detection of circCYP24A1 using divergent and convergent primers from complementary DNA or genomic DNA . F The resistance of circCYP24A1 and CYP24A1 to RNase R cleavage was evaluated by qRT-PCR. G The relative RNA levels of circCYP24A1 and CYP24A1 treated with Actinomycin D (5 μg/mL) was detected by qRT-PCR. Data are presented as the mean ± SD (n = 3). *P < .05, **P < .01, ***P < .001, ****P < .0001. P values were determined using the unpaired Student's t-test.

Due to the limitations of high-throughput sequencing, the identification of circRNAs often comes with a higher rate of false positives. ²⁴ Thus, to demonstrate the authenticity of circCYP24A1 as a up-regulated circRNA in ESCC, a series of validations were conducted on its characteristics. Firstly, the sequencing results were further validated by qRT-PCR, which confirmed that circCYP24A1 was highly expressed in ESCC lines compared to Het-1A (Figure 1C). According to circBase annotation, CircCYP24A1 is located at chr20: 52773707-52788209 and is derived from the exons 3-11 of cytochrome P450 family 24 subfamily A member 1 (CYP24A1), with a length of 1106 bp. Divergent primers were used to amplify circCYP24A1, Sanger sequencing confirmed the splicing junctions which consistent with the annotation in circBase (Figure 1D). Next, convergent and divergent primers were used to amplify cDNA and genomic DNA from KYSE-510, Agarose gel electrophoresis revealed that both primer pairs yielded products when using cDNA as the template, while only the convergent primer pair showed amplification when using genomic DNA as the template, indicating that the detected circCYP24A1 was generated through reverse splicing rather than genomic recombination (Figure 1E). Further, the stability of circCYP24A1 was assessed through RNase R digestion and actinomycin D treatment. The results demonstrated that circCYP24A1 exhibited resistance to RNase R while CYP24A1 was degraded by RNase R (Figure 1F). Finally, after treatment with actinomycin D, circCYP24A1 displayed greater stability compared to its parental host gene CYP24A1 (Figure. 1G). Collectively, these results indicated that circCYP24A1 is a real circRNA.

Elevated Expression of circCYP24A1 is Related to LNM, Advanced Clinical Stage, and Poor Outcomes of ESCC

To assess the expression level of circCYP24A1 in a larger population, we designed probes specifically targeting circCYP24A1 and utilized FISH to detect the expression of circCYP24A1 in tissues based on TMA. The results demonstrated a significant up-regulation of circCYP24A1 expression in ESCC tissues compared to adjacent normal tissues (Figure 2A-B). The median of positivity rate of circCYP24A1 was 2.24% as a cut-off value, the circCYP24A1 positivity rate ≤ 2.24% was categorized as the “Low level circCYP24A1” group, while the positivity rate >2.24% was classified as “High level circCYP24A1” group. A high level of circCYP24A1 was associated with N and TNM stage, but not with age, tumor size, and distant metastasis in ESCC tissues (Table 1). Kaplan-Meier analysis indicated that a high level circCYP24A1 had a shorter 5 – year survival rate (Figure 2C). The aforementioned findings indicated that circCYP24A1 exhibited elevated expression in ESCC tissues, which was significantly associated with LNM, advanced clinical stage, and poor prognosis.

OPEN IN VIEWER

Figure 2.

CircCYP24A1 is up-regulated in ESCC tissues and associated with poor prognosis. A The representative fluorescence images of circCYP24A1 (red) in ESCC and adjacent normal tissues by FISH analysis based on TMA. Scale bar, 100 μm. B The relative expression of circCYP24A1 in paired and unpaired ESCC tissues and adjacent normal tissue in TAM. C Kaplan-Meier curves show the association between circCYP24A1 expression and 5-year overall survival in 106 ESCC patients. Data are presented as the mean ± SD. *P < .05, **P < .01, ***P < .001, ****P < .0001. P values were determined with the two-tailed paired and unpaired Student's t-test (B), log-rank test (C).

OPEN IN VIEWER

Table 1.

Correlation Between circCYP24A1 Expression in Tissue Microarrays and Clinical Datas.

Variables	CircCYP24A1 level		χ2	P value
	Low	High
Age (year)			0.2346	.6282
≤65	28	24
>65	26	28
Sex			7.024	.008
Male	35	46
Female	19	7
Size (cm)			0.6187	.4315
≤3	12	38
>3	9	42
Grade			3.477	.0622
I/II	39	46
III	15	7
T stage			1.957	.1619
T1 + T2	13	7
T3 + T4	39	43
N stage			15.89	<.0001*
N0	34	15
N1	19	39
TNM stage			16.4	<.0001*
Ι + II	35	14
III + IV	17	37

Plasma circCYP24A1 is a Promising Potential Biomarker for the Diagnosis and Prediction of LNM in ESCC

According to the exoRBase database (http://www.exorbase.org/), we learned that circCYP24A1 is an exosomal circRNA that can be detected in human body fluids. We assessed the expression levels of circCYP24A1 in the plasma of ESCC patients, PCa patients, esophagitis patients and healthy individuals. Compared with the healthy controls, the expression of circCYP24A1 in plasma of ESCC patients was significantly upregulated (Figure 3A), while there was no significant difference in circCYP24A1 levels in the plasma of PCa patients and esophagitis patients (Figure 3B). The expression of circCYP24A1 in plasma was normalized, the median expression circCYP24A1 in the plasma of 70 ESCC patients was 1.565 as the cut-off value, samples with circCYP24A1 expression ≤ 1.565 were classified as the “low expression” group, while those >1.565 were classified into the “high expression” group. Notably, it was found that the high expression level of circCYP24A1 was associated with LNM and advanced clinical stage of the ESCC patients, while being independent of age, gender, smoking history and tumor size (Table 2). Further analysis revealed that patients in advanced stages (III/ IV) had significantly higher expression level of circCYP24A1 than patients in earlier stages (I/ II) (Figure 3C), and the level of circCYP24A1 in the LNM patients (N1-3) was significantly higher than patients non-LNM (N0) (Figure 3D). When ESCC patients were stratified according to the median expression of circCYP24A1 during the follow-up period of 13 months, Kaplan-Meier analysis revealed that patients with high circCYP24A1 levels had significantly lower overall survival compared to those with low levels (Figure 3E).

OPEN IN VIEWER

Figure 3.

CircCYP24A1 showed elevated expression in the plasma of ESCC patients, exhibiting a positive correlation with LNM, advanced stage, and poorer survival rates. A The relative expression levels of circCYP24A1 in the plasma of ESCC patients (n = 70) and healthy controls (n = 40). B The relative expression levels of circCYP24A1 in the plasma of PCa patients (n = 5), esophagitis patients (n = 8) and healthy controls (n = 40). C The relative expression levels of circCYP24A1 in the plasma of patients with different stages of ESCC and healthy controls. Early stage (I/II, n = 49) and advanced stage (III/IV, n = 20). D The relative expression levels of circCYP24A1 in the plasma of patients with non-LNM (n = 46) and LNM (n = 23) compared to the healthy control group. E The correlation between the expression of circCYP24A1 in the plasma of 70 ESCC patients and the overall survival rate was demonstrated by the Kaplan-Meier curve. F The ROC curve analysis of circCYP24A1 differentiating ESCC patients and healthy controls. G The ROC curve analysis of circCYP24A1 differentiating ESCC patients with LNM and without LNM. **P < .01, ***P < .001, ****P < .0001. Data are presented as the mean ± SD. P values were determined with the unpaired Student's t-test(A-D), log-rank test (E).

OPEN IN VIEWER

Table 2.

The Correlation Between circCYP24A1 Expression in Plasma and Clinical Data.

Variables	CircCYP24A1 expression		χ2	P value
	Low (n = 35)	High (n = 35)
Age (year)			0.3571	.5501
≤65	8	6
>65	27	29
Sex			0.07224	.7881
Male	25	26
Female	10	9
smoking			0.05714	.8111
No	17	18
Yes	18	17
Size (cm)			0.5143	.4733
≤3	19	16
>3	16	19
N stage			10.47	.0012*
N0	29	17
N1 + N2 + N3	5	18
M stage			0.1587	.6903
M0	32	31
M1	3	4
TNM stage			5.530	.0187*
I + II	29	21
III + IV	5	14

We evaluated the ability of circCYP24A1 to diagnose ESCC and predict LNM through ROC analysis. As shown in Figure 3F, the ROC curve analysis of circCYP24A1 in distinguishing ESCC patients and healthy controls showed an AUC of 0.94 (95% CI 0.90-0.98), the sensitivity and specificity were 0.84 and 0.93, respectively. And Figure 3G showed that circCYP24A1 could be used as a significant parameter to discriminate between non-LNM and LNM ESCC patients with an AUC of 0.76 (95% CI: 0.65∼0.87), the sensitivity and specificity both were 0.70.

In conclusion, elevated circCYP24A1 in the plasma of ESCC patients was significantly related to LNM and advanced clinical stages as well as shorter overall survival, and circCYP24A1 was a potential biomarker for diagnosis and predicting LNM.

CircCYP24A1 Promotes ESCC Cells Migration, Invasion and EMT

In order to investigate the biological functions of circCYP24A1 in ESCC, we successfully knocked down circCYP24A1 in KYSE-510 and KYSE-150 cells (Figure 4A). Regrettably, we found that the knockdown of circCYP24A1 did not affect the proliferative capacity of ESCC cells (Supplementary Figure 1). To determine whether the expression level of circCYP24A1 influences cell motility, we conducted the wound healing assay, in which the control group showed significantly smaller gap sizes than the knockdown group, with the healing rate being the opposite (Figure 4B). The transwell assay also indicated that the depletion of circCYP24A1 reduced the migratory activity of cells and decreased the invasive activity when a matrigel was added (Figure 4C). Furthermore, we assessed whether circCYP24A1 could enhance the epithelial-mesenchymal transition (EMT) in ESCC cells. Western blot analysis revealed that the knockdown of circCYP24A1 in ESCC cells resulted in an increase in the expression of the epithelial marker E-cadherin, while N-cadherin, vimentin and snail significantly decreased (Figure 4D). Taken together, circCYP24A1 promotes ESCC cells’ migration, invasion, and EMT.

OPEN IN VIEWER

Figure 4.

CircCYP24A1 promotes migration, invasion and EMT of ESCC cell. A QRT-PCR analysis of circCYP24A1 after knocking down circCYP24A1 in KYSE510 and KYSE150. B Scratch healing assay was performed on ESCC cells from different treatment groups. C Transwell assay was used to determine cell migration and invasion abilities. D Levels of EMT marker proteins in ESCC cells from different treatment groups were assessed. Data are presented as the mean ± SD (n = 3). **P < .01, ***P < .001, ****P < .0001.P values were determined with the unpaired Student's t-test (A, C and D), two-tailed paired Student's t-test (B).