Human esophageal cancer cell lines (KYSE-150, KYSE-170 and TE-1) and the human normal esophageal epithelial cell line (HEEC) were purchased from the China Center for Type Culture Collection and cultured in RPMI1640 medium (cat. no. 11879020; Gibco; Thermo Fisher Scientific, Inc.) containing 10% fetal bovine serum (FBS; cat. no. A5670201; Gibco; Thermo Fisher Scientific, Inc.). Cells were maintained under standard culture conditions (37°C and 5% CO₂) in culture medium. For thapsigargin (TG; cat. no. T9033; MilliporeSigma) treatment, KYSE-150 and TE-1 cells were cultured at 37°C and treated with 100 nM TG for 0 to 24 h. For recombinant human (rh-)CXCL8 (cat no. HY-P7379; MedChemExpress) treatment, KYSE-150 and TE-1 cells were cultured at 37°C and treated with rh-CXCL8 at varying concentrations (0, 10, 20, 40, 80 or 160 ng/ml) for 0 to 48 h. For TGF-β1 (cat. no. HY-P7118; MedChemExpress) treatment, KYSE-150 and TE-1 cells were cultured at 37°C and treated with 10 ng/ml TGF-β1 for 48 h.

Total RNA was extracted from KYSE-150, KYSE-170, TE-1 and HEEC cells (4×10⁶ cells per sample) using Triquick reagent (cat. no. R1100; Beijing Solarbio Science & Technology Co., Ltd.) and reverse transcribed into cDNA with the Transcriptor First Strand cDNA Synthesis Kit (cat. no. 04896866001; Roche Diagnostics) according to the manufacturer's protocol. SYBR Green real-time fluorescent quantitative PCR premix (cat. no. B110031; Sangon Biotech Co., Ltd.) was carried out in StepOnePlus real-time fluorescent quantitative PCR system (Applied Biosystems; Thermo Fisher Scientific, Inc.) with the following cycling conditions: Initial denaturation at 95°C for 10 min; 40 cycles of 95°C for 15 sec, 56°C for 30 sec and 72°C for 30 sec, and a final extension at 72°C for 7 min. The expression levels of genes were analyzed with the 2^−ΔΔCq method (33) and normalized to endogenous GAPDH. Primer sequences are listed in Table SI.

TE-1 cells were treated with 100 nM TG for 12 h under standard culture conditions (37°C; 5% CO₂), after which total RNA was extracted using TRIzol^™ Reagent (cat. no. 15596026CN; Thermo Fisher Scientific, Inc.). The quality and integrity of the RNA samples were assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Inc.). Library construction, sequencing and data analysis were carried out by Sangon Biotech Co., Ltd. Sequencing was carried out on an Illumina platform using paired-end sequencing with a read length of 150 bp. The sequencing was performed using the Illumina NovaSeq 6000 system with the NovaSeq 6000 S4 Reagent Kit (300 cycles; cat. no. 20027466; Illumina, Inc.). The final library was quantified using a Qubit 4.0 Fluorometer (Thermo Fisher Scientific, Inc.) and the concentration was adjusted to 10–20 pM prior to sequencing, based on molar concentration calculations. Bioinformatics analysis was performed using the programs FastQC (v0.11.9; http://www.bioinformatics.babraham.ac.uk/projects/fastqc/), Cutadapt (v2.10; http://cutadapt.readthedocs.io/), HISAT2 (v2.2.1; http://daehwankimlab.github.io/hisat2/) and featureCounts (v2.0.1; http://subread.sourceforge.net/featureCounts.html). Statistical analysis and visualization were conducted in R (v4.3.2; https://www.r-project.org/; The R Foundation). Differential expression analysis was conducted in R (v4.3.2; http://www.r-project.org) using DESeq2 (v1.30.1; http://bioconductor.org/packages/release/bioc/html/DESeq2.html). Differentially expressed genes were screened based on the following criteria: |log2foldchange|>1, adjusted P<0.05.

ERN1, EIF2AK3 and ATF6 small interfering (si)RNA, alongside si-negative control were purchased from Shanghai GenePharma Co., Ltd., sequences are listed in Table SII. pcDNA3.1-XBP1 and pcDNA3.1-ATF6 were purchased from GenScript Biotech Corporation. The cDNA encoding ATF4 was amplified and inserted into the pcDNA3.1 vector (cat. no. V79520; Invitrogen; Thermo Fisher Scientific Inc.), named pcDNA3.1-ATF4, sequences are listed in Table SIII. KYSE-150 and TE-1 cells were seeded in 6-well plates at a density of 1×10⁶ cells/well and cultured for 24 h to allow adherence. Upon reaching 70–80% confluency, transfection was carried out according to the manufacturer's protocol for Lipofectamine^® 2000 (cat. no. 11668500; Invitrogen; Thermo Fisher Scientific, Inc.). Specifically, 3 µg of target overexpression plasmid and empty vector control plasmid or 100 pmol of siRNA and negative control siRNA was mixed with transfection reagent at room temperature for 20 min and added to the cell culture system. Post-transfection, cells were maintained in a 37°C, 5% CO₂ incubator. The culture medium was replaced with fresh medium at 6 h. After 48 h of total incubation, cells were harvested for transfection efficiency analysis by RT-qPCR and subsequent experiments.

Full-length human CXCL8 cDNA was obtained via RT-qPCR and inserted into the lentiviral expression vector pCDH-CMV-MCS-EF1-Puro. The recombinant plasmid (10 µg), along with the packaging plasmid psPAX2 (7.5 µg) and the envelope plasmid pMD2.G (5 µg), was mixed and incubated at room temperature for 20 min to form the transfection complex. The complex was then co-transfected into 293T cells using Lipofectamine 2000^® to produce lentiviral particles at 37°C for 6 h, after which the medium was replaced with fresh medium. After 48 h, viral supernatants were harvested and subsequently used to transduce KYSE-150 cells. Infected cells were cultured in medium containing 1.0 µg/ml puromycin for 14 days to generate a stable KYSE-150 cell line with CXCL8 overexpression.

KYSE-150 and TE-1 cells were washed in ice-cold PBS and RIPA lysis buffer (cat. no. R0010; Beijing Solarbio Science & Technology Co., Ltd.). PMSF (cat. no. P0100; Beijing Solarbio Science & Technology Co., Ltd.) was used to extract proteins from cells. Protein concentrations were measured using the BCA Protein Assay Kit (cat. no. PC0020; Beijing Solarbio Science & Technology Co., Ltd.) per the manufacturer's instructions. Briefly, samples were mixed with working reagent (1:50 ratio) and incubated at 37°C for 30 min before absorbance measurement at 562 nm. Protein (50 µg/lane) was then separated by 10% SDS-PAGE and transferred to PVDF membranes (cat. no. 1.15546; MilliporeSigma). Membranes were blocked with 5% non-fat milk in TBS with 0.05% Tween-20 for 1 h at room temperature, followed by incubation with primary antibodies overnight at 4°C and corresponding HRP-conjugated secondary antibodies for 1 h at room temperature. The enhanced chemiluminescence detection reagent (cat. no. WBULS0100; MilliporeSigma) was used to detect the protein bands by Chemi XT 4 (Syngene). The antibodies are listed in Table SIV.

Expression profile analysis of CXCL8 across various types of cancer and normal tissues was carried out using the GEPIA web server (http://gepia.cancer-pku.cn/), which integrates RNA sequencing data from The Cancer Genome Atlas and Genotype-Tissue Expression projects. The ‘Expression DIY’ module was used to assess CXCL8 mRNA expression in pan-cancer types, and to compare its expression levels between ESCC and normal esophageal tissues. The thresholds were set at |log₂ fold change (FC)|>1 and P<0.01 to define significant differential expression. Box plots were generated directly through GEPIA for visualization.

Prediction of transcription factor binding sites was performed using the PROMO database (http://alggen.lsi.upc.es/cgi-bin/promo_v3/promo/promoinit.cgi?dirDB=TF_8.3), which is based on the TRANSFAC transcription factor binding site database. The promoter region (−2,000 bp upstream of the transcription start site) of the CXCL8 gene was analyzed, with the species set to Homo sapiens and the maximum matrix dissimilarity rate set to 15%. Candidate transcription factors were selected based on their known relevance to ERS for further validation.

Cellular proliferation capability was detected using an MTS assay. Briefly, 1×10³ cells pretreated with rh-CXCL8 for 24 h under standard culture conditions (37°C, 5% CO₂) were seeded per well in 96-well plates. At 0, 24, 48, 72 and 96 h post-seeding, CellTiter 96^® AQueous One Solution Reagent (cat. no. G3582; Promega Corporation) was added, followed by a 2-h incubation at 37°C before measuring absorbance at 490 nm.

Wound healing and Transwell assays were carried out to detect the migration and invasion ability of cells. For the wound healing assay, treated cells (KYSE-150 and TE-1 cells subjected to 24 h treatment with NC or 10 ng/ml rh-CXCL8, 12 h treatment with 100 nM TG, 48 h transfection with 100 pmol siCXCR1 and combined treatment with 100 nM TG and 100 pmol siCXCR1) were inoculated in 6-well plates and cultured in medium containing 10% FBS. A straight scratch was made in each well when the cell monolayer reached near complete confluence, and cells were then maintained in serum-free medium. Images were taken at the same position at 0 and 24 h using an inverted optical microscope (Nikon Eclipse 50i; Nikon Corporation). Subsequently, ImageJ (National Institutes of Health) software was employed to analyze the acquired images. The calculation of scratch area percentage was defined as the ratio of scratch area at 24 h to the scratch area at 0 h for each group.

For the Transwell invasion assay, 1×10⁵ indicated cells in 200 µl serum-free medium were placed in the upper chamber (Corning, Inc.) and cultured at 37°C for 24 h. The chamber had been previously coated with 200 µl Matrigel (incubated at 37°C for 6 h), and 0.7 ml medium containing 10% FBS was placed in the lower chamber. Cells on the upper surface were wiped off after 24 h of incubation and 4% paraformaldehyde was used to fix the invasive cells on the lower surface of the chamber which was then stained with 0.1% crystal violet solution for 20 min at room temperature. Subsequently, the number of invading cells was counted in five random fields selected by an inverted microscope, and the average was calculated. The migration assay was carried out following the same protocol as the invasion assay, except that Transwell membranes were not coated with Matrigel.

In KYSE-150 cells, the promoter-reporter gene plasmids (containing promoter regions derived from CXCL8, SNAI2 and ZEB1) were separately co-transfected with pcDNA3.1-XBP1, pcDNA3.1-ATF4, pcDNA3.1 empty plasmid or treated with TGF-β1. Luciferase activity was measured 48 h after transfection and treatment under standard culture conditions (37°C; 5% CO₂), using the Dual-Luciferase^® Reporter Assay System (cat. no. E1910; Promega Corporation) according to the manufacturer's instructions and normalized to Renilla luciferase activity. The primer sequences for the dual-luciferase reporter plasmids are detailed in Table SV.

A total of 10 BALB/c-nude mice (6-week-old male and body weight of 18–22 g) were purchased from the Beijing Huafukang Biotechnology Co., Ltd. Mice were housed under standard conditions and had unlimited access to sterilized food and distilled water. Room temperature was maintained at 24±2°C, relative humidity was maintained at 60±10% and a 12-h light/dark cycle was used. After 7 days of acclimation, the mice were randomly divided into two experimental groups (n=5 per group): i) Control group, ii) CXCL8-overexpressed group. Group i) mice served as the normal control group and were subcutaneously injected with 5×10⁶ control (KYSE-150) cells into the dorsal flanks. Group ii) was the CXCL8-overexpressed group and the mice were injected with a total of 5×10⁶ CXCL8-overexpressing KYSE-150 cells into the dorsal flanks. The total duration of the animal study was 30 days. The mice were monitored daily for food and water intake, weight, body posture, behavior, distress and response to external stimuli.

Humane endpoints of the present study were loss of >20% body weight, severe dehydration, refusal of food, severe pain or distress or a moribund state, no animals reached these endpoints or were found dead during the present study. All mice were anesthetized with an intraperitoneal injection of 50 mg/kg pentobarbital sodium and 0.5–1 ml blood was collected by cardiac puncture followed by sacrifice by exsanguination. Indexes such as breathing, heartbeat, pupils and nerve reflexes were assessed to confirm death. Mice were euthanized after 30 days, and terminal volume and weight of tumor tissues were measured. The present study was approved by the Institutional Animal Care and Use Committee (IACUC) of the Fourth Hospital of Hebei Medical University (approval no. IACUC-4th Hos Hebmu-2023001).

In the CXCL8-overexpressing mouse model, both esophageal carcinoma and adjacent normal tissues were collected and fixed in 4% formaldehyde at 4°C for 24–48 h. After washing with PBS, the specimens were dehydrated through a graded ethanol series, embedded in paraffin and sectioned at a thickness of 4-µm. The sections were then blocked with 3% hydrogen peroxide (15 min at room temperature) to inhibit endogenous peroxidase activity and subjected to antigen retrieval in citrate buffer (pH 6.0) at >95°C for 5 min. Primary antibodies against CXCR1/2, p-SMAD2/3, SNAI2 and ZEB1 (all at 1:100 dilution) were applied and incubated overnight at 4°C, followed by a 2 h room temperature incubation with secondary antibodies. The stained sections were visualized using an inverted optical microscope. The antibody information is identical to that used for western blotting (Table SIV).

To identify candidate miRNAs that may be transcriptionally regulated by SMAD3 and simultaneously target ZEB1, bioinformatics screening was conducted using the TransmiR v3.0 (http://www.cuilab.cn/transmir) and TargetScan Release 7.1 (http://www.targetscan.org/vert_71/). First, TransmiR v3.0 was queried to identify miRNAs either known or predicted to be regulated by SMAD3. Subsequently, TargetScan was used to predict miRNAs with conserved binding sites in the 3′-untranslated region (3′-UTR) of ZEB1. The predicted miRNAs from both TransmiR v3.0 and TargetScan databases were compared, and the overlapping miRNAs were identified as potential candidates linking SMAD3 and ZEB1.

GraphPad Prism 8.0 (Dotmatics) was used to carry out data analysis and graphing. Statistical significance was determined using an unpaired Student's t-test or one-way ANOVA with Tukey's post hoc test. All in vitro experiment data came from three independent experiments carried out in triplicate and presented as mean ± SD. P<0.05 was considered to indicate a statistically significant difference.

ESCC cells were treated with 100 nM TG (an ERS inducer) for 0–24 h (34), after which expression of effectors of the UPR were assessed. The mRNA and protein expression levels of XBP1, ATF4 and ATF6 peaked at 12 h after TG treatment (Fig. 1A and B). To explore the potential mechanism of ERS in ESCC progression, RNA sequencing analysis was employed to compare the gene expression profiles between TE-1 cells treated with either TG or DMSO. CXCL8 was one of the top ranked differentially altered genes in response to ERS (Fig. 1C). The elevated mRNA and protein expression levels of CXCL8 were further verified at 12 h in TG-treated KYSE-150 and TE-1 cells by RT-qPCR and western blot analysis (Fig. 1D and E). CXCL8 was significantly upregulated in a variety of types of cancer, including esophageal cancer by searching the GEPIA database (Fig. 1F and G). The mRNA expression level of CXCL8, as well as its receptors, CXCR1 and CXCR2, was increased in ESCC cells compared with that in HEEC cells (Fig. 1H).

To explore the signaling pathways which may be responsible for induction of CXCL8, three transmembrane sensors (IRE1α, PERK and ATF6) were respectively knocked down in KYSE-150 and TE-1 cells using siRNAs, and the knockdown efficiency was validated by RT-qPCR assay (Fig. S1A). As shown in Fig. 2A and B, under the DMSO-treated condition, knockdown of the three sensors resulted in reduced expression of their downstream transcription factors (XBP1, ATF4, and ATF6). A decrease in CXCL8 expression was observed only in the siIRE1α + DMSO and siPERK + DMSO groups under the DMSO-treated condition. Under the TG-treated condition, the siNS + TG group exhibited significantly increased mRNA and protein levels of the three sensors, their downstream transcription factors, and CXCL8, compared with the siNS + DMSO group. In the siRNA + TG groups, the mRNA and protein levels of the three sensors, their downstream transcription factors, were decreased compared with those in the siNS + TG group; however, a reduction in CXCL8 mRNA and protein expression was observed only in the siIRE1α + TG and siPERK + TG groups. Elevated levels of CXCL8 mRNA and protein in KYSE-150 and TE-1 cells overexpressing XBP1 and ATF4, the downstream transcription factors of IRE1α and PERK, respectively were detected (Fig. 2C and D). In contrast, the expression level of CXCL8 was not influenced by the overexpression of ATF6 (Fig. S1B and C), indicating that upregulation of CXCL8 in ERS may be regulated by the IRE1α/XBP1 and PERK/ATF4 signaling pathways. As transcription factors, XBP1 and ATF4 carry out key roles by regulating the transcription of various genes which are involved in ERS. A possible binding site of XBP1 on the promoter region of CXCL8 (−181 bp/-175 bp) was predicted by the PROMO database, and the binding effect was further verified by dual-luciferase reporter assay in KYSE-150 cells (Fig. 2E). A possible binding site of ATF4 on the promoter region of CXCL8 (−40 bp/-28 bp) was also predicted and the transcriptional regulatory effect of ATF4 on CXCL8 was also verified (Fig. 2F).

In order to study the biological function of CXCL8 in ESCC, KYSE-150 and TE-1 cells were treated with rh-CXCL8, and the expression levels of CXCR1 and CXCR2 were assessed. As shown in Fig. 3A-D, the mRNA and protein expression levels of both receptors peaked at 24 h with 10 ng/ml rh-CXCL8 treatment. ESCC cells were then treated with 10 ng/ml rh-CXCL8 for 24 h to detect the variation of cell function. CXCL8 had no effect on the viability of ESCC cells (Fig. 3E) but significantly promoted their migration and invasion (Fig. 3F and G).

Roles of CXCL8/CXCR1 in the progression of ESCC cells under ERS state were further examined. The knockdown efficiency of CXCR1 was validated by an RT-qPCR assay (Fig. S1D). When ESCC cells were treated with 100 nM TG for 24 h, cell migration and invasion were significantly enhanced, while knockdown of CXCR1 partially prevented this effect (Fig. 4A and B).

Considering that both ERS and CXCL8 promoted the migration and invasion of ESCC cells, the mRNA expression levels of a set of EMT-related genes were screened in rh-CXCL8-treated ESCC cells, including SNAI2, ZEB1, TWIST1, SNAIL, FN1, TWIST2, CDH2, VIMENTIN and ZEB2. Among these genes, SNAI2 and ZEB1 demonstrated the most significant trend of increased expression in KYSE-150 (Fig. 5A) and TE-1 cells (Fig. S2) when treated with 10 ng/ml rh-CXCL8. Nevertheless, further increases in the rh-CXCL8 concentration led to a progressive decline in their expression levels. The protein expression levels of SNAI2 and ZEB1, which were similar to their mRNA expression levels, were further verified in rh-CXCL8-treated ESCC cells by western blot analysis (Fig. 5B). Given that SNAI2 and ZEB1 are important genes involved in EMT process, we hypothesized whether CXCL8 promoted the expression of SNAI2 and ZEB1 by regulating the activation of EMT-related pathways. The PI3K/AKT, JAK2/STAT3, Wnt/β-catenin and TGF-β/SMAD pathways are important pathways that regulate the EMT process, therefore whether CXCL8 promoted the activation of these pathways was further investigated. As shown in Fig. 5C and D, rh-CXCL8 treatment promoted phosphorylation of AKT, STAT3 and SMAD2/3, whereas its effect on β-catenin appeared to be minimal although this was not quantitatively assessed, suggesting that CXCL8 may be involved in the progression of EMT by regulating the activation of AKT, STAT3 and SMAD2/3. The present study revealed several signaling pathways that may be involved in the action of CXCL8. Since activation of AKT and STAT3 by CXCL8 has been reported (35–37), which is consistent with the findings of the present study, the TGF-β/SMAD signaling pathway was the focus of the present study.

CXCR1 was knocked down to study the activating effect of CXCL8 on SMAD2/3. As shown in Fig. 6A, the protein expression levels of p-SMAD2/3 were significantly reduced when CXCR1 was knocked down, suggesting that CXCL8 may activate SMAD2/3 by binding to CXCR1. The regulatory role of CXCL8/CXCR1 in the ERS state was further analyzed. As shown in Fig. 6B, the expression levels of p-SMAD2/3 were significantly increased in TG-treated ESCC cells, whereas knockdown of CXCR1 partially alleviated this effect.

Analysis revealed that 10 ng/ml CXCL8 significantly promoted the expression of SNAI2 and ZEB1 and whether this effect was achieved through its activation of SMAD2/3 was further investigated. KYSE-150 and TE-1 cells were treated with 10 ng/ml TGF-β1 for 48 h (38). As shown in Fig. 6C and D, increased mRNA and protein expression levels of SNAI2 and ZEB1 were detected in TGF-β1-treated cells, potentially indicating the regulatory effect of TGF-β/SMAD signaling pathway on the expression of SNAI2 and ZEB1. A binding site for SMAD3 was identified in the promoter region of SNAI2, and dual-luciferase reporter assay confirmed that exogenous addition of TGF-β1 promoted transcription of SNAI2 (Fig. 6E). Two binding sites for SMAD3 were also found in the promoter region of ZEB1, but ZEB1 transcription was not affected by exogenous incorporation of TGF-β1 (Fig. 6F). miR-3146, miR-3163 and miR-3171 were predicted to be transcriptionally regulated by SMAD3 and further regulated the expression of ZEB1 by binding to its 3′-UTR (Fig. 6G).

To evaluate the oncogenic role of CXCL8 in vivo, KYSE-150 cells stably overexpressing CXCL8 or control cells were Xeno transplanted subcutaneously into the flanks of mice. Quantitative analysis at 30 days post-inoculation revealed a statistically significant augmentation in both tumor volume and weight in the CXCL8-overexpressing group compared with control group (Fig. 7A-C), suggesting a pro-tumorigenic function of CXCL8. To further validate the role of the CXCL8-CXCR1/2-SMAD2/3-SNAI2/ZEB1 axis in tumor metastasis, the protein expression levels of CXCR1/2, p-SMAD2/3, SNAI2 and ZEB1 were examined in CXCL8-overexpressing ESCC tissues. Immunohistochemical analyses revealed enhanced positive staining for CXCR1/2, p-SMAD2/3, SNAI2 and ZEB1 in the CXCL8-overexpression group compared with control group (Fig. 7D).

The aforementioned results suggested that CXCL8, which was highly secreted by ESCC cells in the ERS state, promoted the EMT process by activating SMAD2/3 after binding to its receptors. Activated SMAD2/3 directly or indirectly regulated the transcription of SNAI2 and ZEB1 (Fig. 7E).