The ESCC cell lines (KYSE30, KYSE150 and KYSE510) utilized in the present study were procured from the Cell Resource Center at the Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences. All cells were cultured in RPMI-1640 medium (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum and maintained in an incubator at 37°C with 5% CO₂. The cells were passaged a maximum of 15 times.

The gene expression profiles were obtained from The Cancer Genome Atlas (TCGA) (https://portal.gdc.cancer.gov/) and the Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/gds/). RNA sequencing (RNA-seq) data from 82 ESCC tumor samples and 11 adjacent normal tissues were retrieved from the TCGA database. Concurrently, gene expression profiles GSE23400 and GSE75241 were downloaded from the GEO database. Differential gene analysis was conducted on the GSE20347 dataset from the GEO database. To ascertain differentially expressed genes, the criteria |logFC|>1 and adjusted P<0.05 were employed and the differences were visualized using volcano plots.

To generate KYSE30 and KYSE150 cells characterized by FOXM1 overexpression, ESCC cells were transfected with the FOXM1-overexpression plasmid (GV-FOXM1) or negative-control plasmid (GV–Vector) using Lipofectamine™ 3000 (Thermo Fisher Scientific, Inc.) at 37°C for 6 h, with 2 µg DNA and 5 µg HG-TransGene per well in a 6-well plate, according to the supplier's protocols. The overexpression plasmids utilized in the current study were engineered and constructed by Genomeditech (Shanghai) Co., Ltd. After the culture medium was replaced with fresh complete medium, the efficiency of the FOXM1 overexpression was confirmed. ESCC cells exhibiting FOXM1 overexpression were employed for subsequent experiments 24 h post-medium change.

The IGF2BP2 RNA interference lentiviral vector and the corresponding negative control lentiviral vectors were obtained from Genomeditech (Shanghai) Co., Ltd. The shRNA was synthesized, cloned and inserted into pLKO.1-puro vector. The generation system is the third system. Subsequently, pLKO.1-puro-shRNA plasmid (20 µg), virus packaging plasmids (psPAX2, 15 µg) and envelope plasmids (pMD2.G, 5 µg) were cotransfected into 293T cells (China Center for Type Culture Collection) using Lipofectamine™ 3000 (Thermo Fisher Scientific, Inc.) at 37°C for 6 h. Then medium was replaced with fresh DMEM (Thermo Fisher Scientific, Inc.) containing 10% FBS and incubated at 37°C with 5% CO₂. Supernatants were collected at 48 and 72 h post-transfection, filtered through 0.45 µm PVDF membrane and concentrated by ultracentrifugation (100,000 × g). MOI for lentivirus transfection was 10. KYSE30 and KYSE150 cells in the exponential growth phase were seeded in 6-well plates for 24 h. The IGF2BP2 RNA interference and negative control lentivirus were infected into KYSE30 and KYSE150 cells, respectively. Infection was enhanced with 6 µg/ml Polybrene. Following a 24-h infection, the medium was replaced with complete medium, and the cells were cultured. Puromycin (cat. no. P8230; Beijing Solarbio Science & Technology Co., Ltd.) at 5 µg/ml was utilized to select stably transduced cell populations at 72 h post-infection. The subsequent experiment was performed 10 days post-medium change. The sequences of all shRNAs were as follows: shRNA-IGF2BP2-1: 5′-GCCAGAACACUUCCAAGAUAU-3′; shRNA-IGF2BP2-2: 5′-CAGUUACUGGAGAUGAUUAAU-3′; shRNA-IGF2BP2-3: 5′-GCUAUCCACAAGGUCAGUAUU-3′; and non-targeting control shRNA: 5′-CCUAAGGUUAAGUCGCCCUCG-3′.

Small interfering RNA (siRNA) of FOXM1 and negative control were obtained from Genomeditech (Shanghai) Co., Ltd. 5 µl siRNA (20 µM) was diluted in 50 µl Opti-MEM (Gibco; Thermo Fisher Scientific, Inc.) and 5 µl of Lipofectamine™ 3000 (Thermo Fisher Scientific, Inc.) was diluted in 250 µl Opti-MEM. After incubating separately for 5 min, the mixtures were combined and further incubated for 20 min. The transfection complex was then added to ESCC cells seeded in 6-well plates for transfection at 37°C for 6 h. Knockdown efficiency was assessed 24 h after transfection, followed by subsequent experiments. The sequences of all siRNAs were as follows: siRNA-FOXM1-1 sense, 5′-GAAGCGGACCUCAAUGUGAATT-3′ and antisense, 5′-UUCACAUUGAGGUCCGCUUCTT-3′; siRNA-FOXM1-2 sense, 5′-GGAGCUCAAGAAUCUGACUAUTT-3′ and antisense, 5′-UAGUCAGAUUCUUGAGCUCCTT-3′; and siRNA-Ctrl sense, 5′-UUCUCCGAACGUGUCACGUTT-3′ and antisense, 5′-ACGUGACACGUUCGGAGAATT-3′.

ESCC cells were placed in a 6-well plate at 1×10⁵ cells per well and exposed to Actinomycin D (2 µg/ml; Abcam) for 12 h. Subsequently, the measurement of the FOXM1 mRNA level was conducted using a reverse transcription-quantitative polymerase chain reaction (RT-qPCR) assay.

According to the manufacturer's instructions, total RNA was extracted from tissues using TRIzol (cat. no. 15596026; Thermo Fisher Scientific, Inc.). The extracted RNA was subsequently converted to cDNA employing the Evo M-MLV RT Premix (cat. no. AG11706; Accurate Biotechnology Co., Ltd.) according to the supplier's instructions. The SYBR^® Green Premix Pro Taq HS qPCR kit (cat. no. AG11701; Accurate Biotechnology Co., Ltd.) was used for qPCR, according to the manufacturer's instructions, and the target sequence was amplified using a LightCycler 480II device (Roche Diagnostics). The thermocycling parameters were set as follows: Preincubation for 5 min at 95°C, followed by 45 cycles of 10 sec at 95°C, 10 sec at 60°C, and 10 sec at 72°C. β-actin was utilized as an internal control, and relative RNA expression was analyzed using the 2^−ΔΔCq method (20). The primer sequences were as follows: FOXM1 forward, 5′-GCAGCGACAGGTTAAGGTTGAG-3′ and reverse, 5′-AGTGCTGTTGATGGCGAATTGTAT-3′; IGF2BP2 forward, 5′-TCGTCAGAATTATCGGGCACTTC-3′ and reverse, 5′-CCTGCTGCTTCACCTGTTGTA-3′; and β-actin forward 5′-TGGCACCCAGCACAATGAA-3′ and reverse, 5′-CTAAGTCATAGTCCGCCTAGAAGCA-3′.

To extract proteins, ESCC cells were lysed in PMSF-containing RIPA buffer solution (cat. no. R0010; Beijing Solarbio Science & Technology Co., Ltd.) and centrifuged at 12,000 × g at 4°C for 20 min. The protein concentration was determined using the Bicinchoninic acid reagent test kit (Beijing Solarbio Science & Technology Co., Ltd.), according to the manufacturer's instructions. Protein samples (30 µg) were separated by SDS-PAGE on 10% gels (cat. no. PG112; Shanghai EpiZyme Co., Ltd.) and then transferred onto PVDF membranes (cat. no. WGPVDF22; Servicebio Co., Ltd.). Subsequently, the membranes were blocked with high-efficiency western blocking reagent (cat. no. GF1815; Shanghai Genefist Co., Ltd.) for 15 min at room temperature and maintained overnight at 4°C with the primary antibodies, including anti-IGF2BP2 (1:2,000; cat. no. 11601-1-AP; Proteintech Group, Inc.), anti-FOXM1 (1:2,000; cat. no. 13147-1-AP; Proteintech Group, Inc.) and anti-GAPDH (1:5,000; cat. no. 10494-1-AP; Proteintech Group, Inc.). After rinsing the membrane three times with TBST (cat. no. G2150-1L; Wuhan Servicebio Technology Co., Ltd.), the PVDF membranes were exposed to secondary rabbit antibodies (1:5,000; cat. no. SA00001-2; Proteintech Group, Inc.) for 1 h at ambient temperature. An enhanced chemiluminescence reagent (cat. no. SQ201; Shanghai EpiZyme Co., Ltd.) was employed to detect protein signals. Band intensities were analyzed utilizing ImageJ software (V1.54p; National Institutes of Health), with GAPDH serving as the internal control for normalizing target protein expression.

ESCC cells in the exponential growth phase from both the control and treated groups were inoculated into a 96-well plate at 3,000 cells per well and maintained for 24, 48, 72 and 96 h. After removing the culture medium, each well was washed with phosphate-buffered saline (PBS), and the CCK-8 reagent (MedChemExpress) was combined with serum-free medium to achieve a concentration of 10%. Subsequently, a volume of 100 µl of the prepared solution was dispensed into each well. Following incubation of the 96-well plate at 37°C for 1 h, the absorbance of each well was measured at 450 nm.

Following the counting of ESCC cells in both the control and treated groups, the cells were placed into 6-well plates at 1,000 cells per well and subsequently cultured in an environment maintained at 37°C with 5% CO₂. When the count of cells within a single clone colony exceeded 50 or the culture duration surpassed 14 days, the culture medium was aspirated, and the plate underwent PBS washing. Following fixation in 4% paraformaldehyde (Beyotime Institute of Biotechnology) for 30 min, staining was performed using 0.1% crystal violet (Beyotime Institute of Biotechnology) at ambient temperature for 15 min. Colonies were counted utilizing ImageJ software.

Exponential growth ESCC cells from both the control and treated groups were inoculated into 96-well plates at 5,000 cells per well and maintained for 24 h. The prepared EdU solution (cat. no. C10310-1; Guangzhou RiboBio Co., Ltd.) was incorporated into the culture medium and incubated for 2 h for EdU labeling. Following the removal of the medium and washing with PBS, 4% paraformaldehyde (Beyotime Institute of Biotechnology) was introduced, and the plates were incubated for 30 min at ambient conditions. Subsequently, glycine solution (MedChemExpress), PBS, and an osmotic agent (Beyotime Institute of Biotechnology) were sequentially added to immobilize the cells. The wells were then subjected to Apollo and DNA staining for 30 min at ambient temperature respectively, with images acquired and analyzed using a fluorescence microscope (magnification, ×200).

The Annexin V-FITC Cell Apoptosis Detection Kit (cat. no. E-CK-A211; Wuhan Elabscience Biotechnology Co., Ltd.) was employed to evaluate the level of cell apoptosis. ESCC cells underwent trypsin digestion to obtain a single-cell suspension (the density of cells was 5×10⁵/ml). Following a 15-min dual staining procedure with V-FITC and PI in the 200-µl single-cell suspension, the cell specimens were examined via flow cytometry using BD LSRFortessa (BD Biosciences), BD FACSDiva^TM software (V7.0; BD Biosciences) and FlowJo software (V10.8.1; FlowJo LLC).

A glycolysis assay was performed utilizing a lactate assay kit (cat. no. 600450; Cayman Chemical Company). Exponentially growing ESCC cells from both the control and treated groups were inoculated into 96-well plates at 5,000 cells per well. After 24 h, the culture medium was harvested for lactate content assessment. The specimens were examined following the instructions outlined in the reagent kit, the values were noted, and calculations were performed.

Following transfection of IGF2BP2-knockdown KYSE30 and control cells, the cell samples were collected by scraping and subsequently lysed in TRIzol (cat. no. 15596026; Thermo Fisher Scientific, Inc.) for total RNA extraction. DNase (cat. no. AG12001; Accurate Biotechnology Co., Ltd.) was used to remove genomic DNA contamination. The purity of the sample was determined by NanoPhotometer^® (IMPLEN, USA). The concentration and integrity of RNA samples were detected by Agilent 2100 RNA nano 6000 assay kit (cat. no. 5067-1511; Agilent Technology Co., Ltd.). Sequencing libraries were generated using VAHTS Universal V6 RNA-seq Library Prep Kit for Illumina^® (cat. no. NR604-01/02; Vazyme Co., Ltd.) following the manufacturer's recommendations. Poly-T magnetic beads were used to isolate mRNA, which was then fragmented. First- and second-strand cDNA were synthesized, purified, and subjected to end repair, adenylation, and adapter ligation. The library was size-selected and enriched by PCR prior to sequencing. The library was quantified using the Bio-RAD CFX 96 fluorescence quantitative PCR instrument with the iQ™ SYBR^® Green kit (cat. no. 1708880; Bio-Rad Laboratories, Inc.) to the effective concentration of >10 nM. Sequencing was performed on the Illumina NovaSeq 6000 platform (Illumina, Inc.) with the NovaSeq 6000 S4 Reagent Kit V1.5 (cat. no. 20028312; Illumina, Inc.), generating 150 bp paired-end reads. DESeq2 (V1.44.0) and EdgeR (V3.44.0) were used for Difference analysis (21,22), and the Gene Ontology (GO; http://geneontology.org/) was used for function enrichment analysis.

Total RNA was obtained and purified from ESCC samples with TRIzol (cat. no. 15596026; Thermo Fisher Scientific, Inc.), according to the manufacturer's instructions. Following extraction, RNA was incubated with m6A antibodies utilizing the Magna Methylated RNA Immunoprecipitation m6A Kit (MilliporeSigma) for immunoprecipitation. The concentration of m6A-modified mRNA underwent examination through m6A-qPCR or m6A-seq methodologies (Haplox Biotechnology Co., Ltd.). Primers targeting the negative region of FOXM1 m6A served as negative controls, whereas primers targeting the positive region of FOXM1 m6A were employed as positive controls. For m6A-seq, the enriched m6A RNA underwent reverse transcription to generate cDNA, and the library was prepared per the instructions provided by Illumina's NEBNext Ultra RNA Library Preparation Kit (New England Biolabs, Inc.). High-throughput library sequencing was conducted to acquire sequence information about m6A-modified RNA using the Illumina HiSeq X™ Ten platform (Illumina, Inc.).

Statistical analysis was performed utilizing GraphPad Prism 9 (Dotmatics). Data are presented as the mean ± standard deviation (SD). Comparisons among several groups were performed using one-way ANOVA followed by Least Significance Difference test was used for the post hoc test, whereas evaluations between paired groups employed paired Student's t-test. P<0.05 was considered to indicate a statistically significant difference.

Differential gene analysis was performed utilizing the GSE20347 dataset, which revealed a total of 389 differentially expressed genes, encompassing 226 downregulated and 163 upregulated genes (Fig. 1A). Among these, the 10 most markedly upregulated genes were identified as COL11A1, ZIC1, COL1A2, ECT2, ACKR3, LUM, MFHAS1, FNDC3B, IGF2BP2 and KIF23. Furthermore, a comparison of IGF2BP2 expression levels between ESCC tumor tissues and normal tissue samples from the TCGA database indicated a significant increase in IGF2BP2 expression (P=0.0014; Fig. 1B). Data obtained from GSE23400 and GSE75241 confirmed the upregulation of IGF2BP2 in ESCC (P<0.0001; Fig. 1C and D). Regarding the ESCC cell lines, RT-qPCR (Fig. 2A) demonstrated a significant upregulation of IGF2BP2 in KYSE30 and KYSE150 relative to HET-1A (the normal esophageal cell line). Consequently, both cell lines were chosen for the following investigations.

To explore the role of IGF2BP2 in ESCC, alterations in cellular phenotypes were assessed following the inhibition of IGF2BP2 expression in cell lines. Efficient knockdown of IGF2BP2 was achieved using two short-hairpin (sh)RNAs (sh-IGF2BP2-1 and sh-IGF2BP2-2) in both KYSE30 and KYSE150 cells (Fig. 2B). Given that sh-IGF2BP2-1 exhibited the highest efficiency, this sequence was selected for subsequent depletion experiments.

CCK-8 assays were performed to evaluate the influence of IGF2BP2 on the proliferation of ESCC cells. The results demonstrated that IGF2BP2 knockdown resulted in significantly decreased cell proliferation (Fig. 2C). Images obtained from colony formation assays illustrated that silencing IGF2BP2 significantly impaired the formation of colonies of ESCC cells (Fig. 2D). Moreover, EdU assays validated the substantial suppressive effect of IGF2BP2 knockdown on ESCC cell proliferation (Fig. 2E). Additionally, flow cytometry assays indicated that the knockdown of IGF2BP2 markedly enhanced the apoptosis of ESCC cells (Fig. 2F). Notably, it was observed that silencing IGF2BP2 resulted in a reduction of anaerobic glycolysis in KYSE30 and KYSE150 cells (Fig. 2G). Moreover, the depletion of IGF2BP2 significantly enhanced ESCC cell responsiveness to paclitaxel (Fig. 2C-F). Collectively, these findings indicated that IGF2BP2 serves a crucial function in anaerobic glycolysis and resistance to paclitaxel in ESCC cells.

To elucidate the mechanism through which IGF2BP2 contributes to paclitaxel resistance in ESCC, RNA-seq was conducted using both IGF2BP2-knockdown KYSE30 and control cells. It was demonstrated that the expression of 1,139 genes was globally modified following IGF2BP2 knockdown, with 424 genes exhibiting upregulation and 715 genes displaying downregulation (Fig. 3A). GO analysis suggested that several pathways, encompassing the MAPK signaling pathway, TGF-β signaling pathway, and protein processing in the endoplasmic reticulum, were markedly enriched (Fig. 3B). These findings collectively highlight the oncogenic role of IGF2BP2 in ESCC.

It is widely acknowledged that IGF2BP2 functions as an m6A reader, capable of recognizing and binding to m6A-methylated transcripts to regulate target genes. Consequently, m6A-seq was conducted in KYSE30 cells, resulting in the identification of 8,104 m6A modification peaks across 3,990 genes. The majority of these m6A modifications were located within the 3′-untranslated regions (3′-UTRs; 43.36%; Fig. 3C). Functional annotation indicated that these mRNAs were associated with several distinct gene clusters, including the regulation of RNA stability and RNA splicing (Fig. 3D). Given that IGF2BP2 has been recognized for its role in maintaining RNA stability (14), the attention was directed towards transcripts that were downregulated following IGF2BP2 knockdown and exhibited the top 100 highest peak m6A modifications. By overlapping the results from RNA-seq and m6A modification analyses, two candidate genes, FOXM1 and SYNCRIP, were identified as meeting the criteria. Among these, FOXM1 mRNA levels declined more markedly upon IGF2BP2 knockdown in ESCC cell lines. Furthermore, emerging evidence suggested that FOXM1 contributes to paclitaxel resistance (23). These findings prompted a shift in focus towards FOXM1, which is considered a critical factor in the mediation of paclitaxel resistance through IGF2BP2 in ESCC (Fig. 3E).

To ascertain whether IGF2BP2 modulates FOXM1 expression in ESCC cells, FOXM1 expression was assessed at both transcriptional and translational levels following IGF2BP2 knockdown, utilizing RT-qPCR and WB analysis, respectively. As anticipated, both the RNA expression levels (Fig. 4A) and protein levels (Fig. 4B and C) of FOXM1 exhibited a significant decrease correlating with IGF2BP2 depletion. Given that IGF2BP2 deficiency impaired the levels of FOXM1 mRNA, it was hypothesized that IGF2BP2 may regulate FOXM1 by maintaining the stability of FOXM1 mRNA. Subsequently, 2 g/ml Actinomycin D was administered to treat KYSE30 cells transfected with IGF2BP2-shRNA or control shRNA at various time points. As illustrated in Fig. 4D, a significantly accelerated decay of FOXM1 mRNA was observed in the absence of IGF2BP2 (Fig. 4D), indicating that IGF2BP2 can sustain the stability of FOXM1 mRNA.

The oncogenic function of FOXM1 has been documented in ESCC (21–23). To further validate the function of FOXM1 in ESCC, the phenotypes of ESCC cells, encompassing cell growth and proliferation, were examined through gene-loss experiments following FOXM1 depletion (Fig. 5A). Notably, a significant reduction in cell growth (Fig. 5B) was observed through the CCK-8 assays, alongside a diminished formation of colonies (Fig. 5C) of the KYSE30 and KYSE150 cells observed in the EdU assays. Consequently, it was hypothesized that FOXM1 facilitates oncogenesis in ESCC cells.

To further investigate whether FOXM1 mediates the oncogenic effects of IGF2BP2, FOXM1 expression was upregulated in ESCC cells that were silenced for IGF2BP2. The efficacy of FOXM1 overexpression was validated through WB analysis (Fig. S1A and B). Furthermore, the efficacy of ectopic FOXM1 expression in conjunction with silenced IGF2BP2 was assessed at both transcriptional and translational levels through RT-qPCR and WB analysis, respectively. The results indicated that both RNA expression (Fig. 6A) and protein abundance (Fig. 6B) of ectopically expressed FOXM1 were markedly increased in IGF2BP2-silenced ESCC cells. Subsequently, a series of functional assays were conducted to evaluate ESCC cell activity. CCK-8 assays suggested that the proliferation rate of ESCC cells exhibiting ectopic FOXM1 expression was significantly higher than that of cells lacking ectopic FOXM1, irrespective of the presence of paclitaxel (Fig. 6C). Furthermore, EdU assays illustrated that the reduced cell proliferation activity due to IGF2BP2 depletion was significantly enhanced through ectopic FOXM1 expression (Fig. 6D). The results of colony formation assays corroborated the findings observed in the EdU assays (Fig. 6E). Moreover, flow cytometric analysis revealed that ectopic FOXM1 expression mitigated the apoptosis induced through IGF2BP2 silencing (Fig. 6F). These findings demonstrated that ectopic FOXM1 expression improved cell proliferation, enhanced the formation of colonies and alleviated apoptosis associated with IGF2BP2 deficiency in ESCC cells. Notably, it was observed that the decrease in anaerobic glycolysis resulting from IGF2BP2 silencing in ESCC cells was reversed upon FOXM1 upregulation (Fig. 6G). Therefore, it is proposed that IGF2BP2 promotes anaerobic glycolysis and diminishes sensitivity to paclitaxel in ESCC cells by regulating FOXM1 (Fig. 7).