Long non-coding RNA LINC00092 inhibits esophageal squamous cell carcinoma progression by promoting ferroptosis through the MAZ/NFE2L2 axis

Abstract

Background

Esophageal cancer (EC) is the eighth most prevalent malignancy worldwide and exhibits the sixth poorest prognosis. Esophageal squamous cell carcinoma (ESCC) is the predominant pathological subtype. Ferroptosis, an iron-dependent form of cell death, plays a critical role in cancer progression. Long non-coding RNAs (lncRNAs) have emerged as key regulators in the initiation and progression of EC. However, the role of lncRNAs in modulating ferroptosis within EC remains poorly understood. Therefore, this study aimed to identify key ferroptosis-related lncRNAs in ESCC and to investigate the role and mechanism of a specific lncRNA, long intergenic non-protein-coding RNA 92 (LINC00092).

Methods

Bioinformatics analysis was conducted to identify ferroptosis-related lncRNAs, transcription factors (TFs), and genes associated with ESCC. The expression, function, tumor microenvironment, immunotherapy, and downstream molecular pathways were also determined. The expression levels of LINC00092, MYC-associated zinc finger protein (MAZ), and NFE2 like bZIP transcription factor 2 (NFE2L2) were detected using quantitative real-time polymerase chain reaction (qRT-PCR), immunohistochemical analysis, and western blotting. Fluorescence in situ hybridization (FISH) was performed to determine the subcellular localization of LINC00092. 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), colony formation, wound healing, Transwell, and flow cytometry apoptosis assays were performed to determine the phenotypes and functions of loss- and gain-of LINC00092. RNA immunoprecipitation (RIP) and luciferase reporter assays were used to evaluate interactions involving LINC00092. The expression of ferroptosis-related proteins was verified by western blotting.

Results

LINC00092 was found to be downregulated in ESCC datasets, cell lines, and tissue samples. Bioinformatics analysis revealed that LINC00092 was associated with ferroptosis and negatively correlated with NFE2L2 expression. Further investigations demonstrated that LINC00092 acted as a binder to the TF MAZ, modulating the expression of the ferroptosis-related gene NFE2L2. Overexpression of LINC00092 inhibited ESCC cell progression, whereas its downregulation promoted tumor progression. RIP and luciferase reporter assays confirmed that MAZ was a target of LINC00092, and NFE2L2 was a downstream target of MAZ. Western blot analysis showed that LINC00092 enhanced ferroptosis in ESCC cells. The LINC00092/MAZ/NFE2L2 axis appeared to inhibit cancer progression by promoting ferroptosis through the regulation of NFE2L2 and sequestration of the TF MAZ.

Conclusions

LINC00092 exerts tumor-suppressive effects in ESCC cells by inhibiting cancer progression through the LINC00092/MAZ/NFE2L2 axis and promoting ferroptosis. Therefore, LINC00092 may serve as a potential therapeutic target for ESCC.

Highlight box.

Key findings

• This study revealed that long intergenic non-protein-coding RNA 92 (LINC00092)/MYC-associated zinc finger protein (MAZ)/NFE2 like bZIP transcription factor 2 (NFE2L2) axis could inhibit esophageal squamous cell carcinoma (ESCC) progression by promoting ferroptosis.

What is known and what is new?

• Ferroptosis is an iron-dependent cell death linked to lipid peroxidation and various diseases, including cancer, where it affects metabolism and redox balance, making it a potential therapeutic target for EC. Long non-coding RNAs (lncRNAs), which regulate gene expression, can influence tumor progression and ferroptosis.

• LINC00092 exerts tumor-suppressive effects in ESCC cells by inhibiting cancer progression through the LINC00092/MAZ/NFE2L2 axis and promoting ferroptosis. Therefore, LINC00092 may serve as a potential therapeutic target for ESCC.

What is the implication, and what should change now?

• This study systematically revealed an innovative model of LINC00092/MAZ/NFE2L2 axis inhibiting ESCC progression by promoting ferroptosis. This provides a new non-coding RNA regulatory perspective for the pathogenesis of ESCC, enriching the theoretical role of ferroptosis in cancer. LINC00092 or downstream effectors (such as NFE2L2) may become new targets for ESCC therapy, especially in regulating the iron death pathway, with promising applications.

Introduction

Esophageal cancer (EC) is the eighth most common cancer worldwide and has the sixth poorest prognosis. Esophageal squamous cell carcinoma (ESCC) is the predominant pathological subtype, characterized by its aggressive nature and poor survival outcome (1,2). Annually, over 600,000 new cases of EC are diagnosed, and more than 500,000 deaths are attributed to the disease globally, with age-standardized incidence and mortality rates exceeding 6 and 5 per 100,000 individuals, respectively (3). Due to its highly invasive nature, the 5-year survival rate remains dismal, ranging from 15% to 25% (3). Despite recent advances in EC treatment, long-term survival rates have shown little improvement, primarily due to late-stage diagnosis and metastatic spread. Consequently, there is an urgent need to identify novel biomarkers and develop more effective therapeutic strategies to improve patient outcomes.

Ferroptosis is a newly identified form of iron-dependent cell death driven by lipid peroxidation (4). Its dysregulation is implicated in various diseases, including cancer (5). In tumor biology, ferroptosis is closely linked to altered metabolism and redox balance, as the accumulation of iron and reactive oxygen species (ROS) can influence cancer cell survival and proliferation (6-8). Emerging evidence highlights the role of key regulators such as GPX4, NRF2, and SLC7A11 in modulating ferroptosis, with their activity often disrupted in cancers (9-11). Notably, in ESCC, factors like METTL3 have been shown to inhibit ferroptosis, thereby promoting tumor progression (12). Thus, targeting ferroptosis pathways represents a promising therapeutic strategy for EC.

Long non-coding RNAs (lncRNAs) are transcripts longer than 200 nucleotides with limited protein-coding potential, which play important roles in gene regulation (13). In cancer, lncRNAs can modulate tumor development and progression, including by influencing ferroptosis. For instance, some lncRNAs promote ferroptosis resistance (e.g., by stabilizing SLC7A11), while others can enhance ferroptosis sensitivity (14-16). Recent studies have begun to identify specific lncRNAs, such as LINC01605, that are dysregulated in EC and may function through mechanisms like competing endogenous RNA (ceRNA) networks, highlighting their potential roles in carcinogenesis (17). Similarly, other non-coding RNAs like circular RNAs (circRNAs) are also implicated in EC; for example, the hsa_circ_0036722/miR-503-5p/PDCD4 axis has been identified as a novel regulatory pathway in this malignancy (18). However, the involvement of lncRNAs, particularly in the critical process of ferroptosis within EC, remains largely unexplored. However, the functions of most ferroptosis-related lncRNAs in EC remain largely unknown. Given their regulatory potential and the unmet need in EC, we aimed to identify key ferroptosis-related lncRNAs in this malignancy. Through an analysis of EC microarray data, long intergenic non-protein-coding RNA 92 (LINC00092) emerged as a significantly dysregulated candidate, leading us to focus on it for further investigation. We present this article in accordance with the MDAR reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2433/rc).

Methods

Screening differential expression ferroptosis-related genes (DEFGs)

RNA-seq transcriptome profiles were downloaded from The Cancer Genomic Atlas (TCGA) database (https://portal.gdc.cancer.gov/), comprising 11 normal and 161 EC tumor samples. We obtained 564 ferroptosis-related genes from the “FerrDb” database (http://www.zhounan.org/ferrdb/current/). The R package “limma” was utilized to identify DEFGs. Differentially expressed genes (DEGs) were identified with a threshold of |log₂fold change (FC)| >1 and an adjusted P value <0.05, a criterion consistent with previous transcriptomic studies in cancer (19). Subsequently, we validated these findings using the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/). The datasets for ESCC included GSE17351, which contained 5 normal and 5 tumor samples; GSE100942, which comprised 5 normal and 4 tumor samples; and GSE157804, which included 5 normal and 5 tumor samples (Table 1). Our research commenced in June 2022, and the TCGA data were downloaded in August 2022. We added the exact download date of the TCGA dataset (2022) in the methodology.

Table 1. Summary of the ESCC microarray dataset from the GEO database.

Series	Platform	Affymetrix GeneChip	Tumor	Normal
GSE17351	GPL570	[HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array	5	5
GSE100942	GPL570	[HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array	5	4
GSE157804	GPL20115	Agilent-067406 Human CBC lncRNA + mRNA microarray V4.0 (Probe name version)	5	5

ESCC, esophageal squamous cell carcinoma; GEO, Gene Expression Omnibus.

Weighted gene co-expression network analysis (WGCNA)

WGCNA was performed for all ferroptosis-related genes using the WGCNA package in R (R Foundation for Statistical Computing, Vienna, Austria). The soft-thresholding power (β) was determined during module construction by applying the pickSoftThreshold function in WGCNA. Power values were screened using a gradient algorithm to assess independence, with the power values of different modules ranging from 1 to 20. Gene modules were constructed after selecting an optimal power value, where the reference dataset’s index value exceeded 0.8. A minimum module size of 30 was set, and the heatmap package was used to analyze and visualize the correlation strength between modules. A cut-off threshold of 0.25 was applied to generate a dendrogram plot.

Identification of ferroptosis-related lncRNA

To identify ferroptosis-related lncRNAs, we performed a co-expression correlation analysis between lncRNAs and DEFGs expression profiles using the limma package. To identify robust co-expression relationships, we applied thresholds of |Pearson correlation coefficient (R)| >0.4 and a P value <0.001. The threshold of |R| >0.4 is widely used to define moderate to strong correlations in transcriptomic studies (20), ensuring the selected relationships are biologically meaningful beyond random noise.

Identification of transcription factors (TFs)

The CatRAPID (http://s.tartaglialab.com/page/catrapid group) and AnimalTFDB (http://bioinfo.life.hust.edu.cn/AnimalTFDB4/#/) databases were used to predict potential binding TFs. The canonical binding motifs of TFs were identified using the JASPAR database (http://jaspar.genereg.net/), a curated, open-source resource for TF binding motifs.

Evaluation of tumor microenvironment and immunotherapy

ESTIMATE is a computerized algorithm that can be used to infer the level of stromal and immune cell infiltration in tumor tissues based on expression profiles (15,16). The tumor microenvironment was evaluated by ESTIMATE. We used the “ESTIMATE” package to calculate the immune score, matrix score and estimated score of the relevant samples, respectively, and explored the correlation between these scores and the expression of LINC00092. The Comprehensive Analysis on Multi-Omics of Immunotherapy in Pan-cancer (CAMOIP) database could explore the correlation between specific genes and immune checkpoints, as well as the differences in immune characteristic scores among different expression groups (17). In this study, the immune activity in LINC00092 high- and low-expression groups in EC was analyzed by CAMOIP database.

Functional analysis

Gene set enrichment analysis (GSEA) was conducted using GSEA v3.0 (http://www.broadinstitute.org/gsea/). GSEA is a method used to analyze genome-wide expression profiles at the gene set level and is widely used to annotate and predict gene functions. Additionally, the Comparative Toxicogenomics Database (CTD) (http://ctdbase.org/) was utilized to identify integrated chemical-disease, chemical-gene, and gene-disease interactions, facilitating the prediction of novel associations and the generation of expanded networks. Furthermore, the relationships between gene products and ventricular remodeling were examined using the CTD.

ESCC tissue specimens

Patients who had not received preoperative chemotherapy, radiotherapy, or molecular targeted treatment were selected as research subjects. Specimens of ESCC and paracancerous tissues were surgically excised from 40 patients at the Fourth Hospital of Hebei Medical University. The specimens were promptly placed in liquid nitrogen at −180 ℃ for preservation. Concurrently, clinicopathological data corresponding to each specimen were collected. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of the Fourth Hospital of Hebei Medical University (No. 2022KY259). Informed consent was provided by all individual participants included in the study. All tissue samples were histologically examined and confirmed by at least two independent pathologists to ensure the diagnosis of ESCC and to verify that adjacent tissues were free of tumor cells.

ESCC cell lines

ESCC cell lines (KYSE30, KYSE150, and KYSE170) were obtained from the Shanghai Cell Institute of the Chinese Academy of Sciences (Shanghai, China) and maintained in the laboratory with consistent cell passages. Cells were cultured in Roswell Park Memorial Institute (RPMI)-1640 medium (Thermo Fisher, Waltham, MA, USA; catalog, A4192301) supplemented with 10% fetal bovine serum (FBS; Thermo Fisher, catalog, 26010074) at 37 ℃ in a humidified atmosphere with 5% CO₂. The total RNA extraction kit and transfection reagent Lipofectamine 2000 were purchased from Invitrogen^TM Life Technologies (Carlsbad, CA, USA), and the reverse transcription kit and fluorescence quantification polymerase chain reaction (PCR) detection kit were purchased from Takara (Kusatsu, Shiga, Japan). Three independent technical replicates were conducted for each cell line experiment. Each repetition followed the same experimental procedure and conditions, including material preparation, experimental operations, and data analysis.

Immunohistochemistry

Endogenous peroxidase activity was blocked with 3% hydrogen peroxide in methanol at room temperature for 10 minutes. Non-specific binding was blocked with 5% bovine serum albumin (BSA; Sigma-Aldrich, St. Louis, MO, USA; catalog, 10711454001) in phosphate-buffered saline (PBS) for 1 hour at room temperature. Primary antibody incubation was performed overnight at 4 ℃ with rabbit MYC-associated zinc finger protein (MAZ) (1:500; catalog no. ab85725; Abcam, Cambridge, UK) and rabbit NFE2 like bZIP transcription factor 2 (NFE2L2) (1:100; catalog no. ab62352; Abcam) antibodies diluted in PBS containing 1% BSA. Sections were washed three times with PBS containing 0.1% Tween-20 (PBST) for 5 minutes each. Secondary antibody incubation was performed for 1 hour using goat anti-rabbit IgG antibody (1:100; catalog no. ab150077; Abcam). Negative controls were included by omitting the primary antibody. Signals were developed with 3,3’-diaminobenzidine (DAB; Dako, Glostrup, Denmark) for 5 minutes, and sections were counterstained with hematoxylin.

Fluorescence in situ hybridization (FISH)

FISH was performed on KYSE30 and KYSE170 cells grown on chamber slides. Cells were fixed with 4% paraformaldehyde for 15 minutes at room temperature. A specific probe (Cy5-labeled, 10 ng/µL) purchased from QIAGEN (Hilden, Germany) was diluted in hybridization buffer [50% formamide, 2× saline sodium citrate (SSC)] and applied to slides. Cells were washed three times with PBS containing 0.5% Triton X-100 and hybridized with probes overnight at 37 ℃ in a dark, humidified chamber. Post-hybridization, slides were washed with 2× SSC at 37 ℃ for 10 minutes and then with 1× SSC at room temperature for 5 minutes. Fluorescence signals were detected using a CyTM5-Streptavidin conjugate (ZyMaX Grade, Invitrogen) at a 1:500 dilution. Negative controls used scrambled probes.

Cell transfection

Vectors (pcDNA3.1) containing short hairpin RNA (shRNA-1: 5'-GCACUCUACGCCUGGCCAATT-3'; shRNA-2: 5'-GGACUCCAGCGCAAAGCAUTT-3'; and shRNA-3: 5'-CCGAUUCGGUUUGCCCUUUTT-3') were designed and synthesized by Shanghai GeneChem Co., Ltd. (Shanghai, China). The MISSION shRNA Control Vector was purchased from Sigma-Aldrich (catalog no. SHC001; Merck KGaA). The overexpression (pcDNA3.1) and empty vectors (pcDNA3.1) were provided by Shanghai Sangong Pharmaceutical Co., Ltd. (Shanghai, China). For transfection, 1 µg of vector DNA was mixed with 2 µL of Lipofectamine^® 2000 (catalog no. 11668-019; Invitrogen) in Opti-MEM medium (Gibco, Carlsbad, CA, USA) and incubated for 20 minutes at room temperature. Lipofectamine^® 2000 (catalog no. 11668-019; Invitrogen; Thermo Fisher) was used to transfect 15 nm vectors into 5×10⁵ KYSE30 and KYSE170 cells. Transfections were performed at 37 ℃ for 6 hours. Non-transfected cells served as controls, whereas cells transfected with empty vectors were used as negative controls (NC). Transfection efficiency was determined by quantitative real-time polymerase chainreaction (qRT-PCR). Cells were harvested 24 hours post-transfection prior to subsequent experiments.

qRT-PCR

Total RNA extraction from ESCC tissue and matched adjacent normal tissue samples and in vitro cultivated cells (1×10⁵) was performed using the Monarch^® Total RNA Miniprep Kit (New England Biolabs, Ipswich, MA, USA) according to the manufacturer’s instructions. RNA concentration and purity were assessed using a NanoDrop spectrophotometer (Thermo Fisher), with A260/A280 ratios between 1.8 and 2.0 considered acceptable. Reverse transcription was conducted using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, Inc.) to synthesize complementary DNA (cDNA) in a 20 µL reaction volume with 1 µg of RNA, incubated at 25 ℃ for 10 minutes, 37 ℃ for 120 minutes, and 85 ℃ for 5 minutes. PCR reaction mixtures were prepared using the Luna^® universal One-Step qRT-PCR Kit (SYBR; New England Biolabs). All qPCR reactions were performed using the CFX96 Touch^™ Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA, USA). The qPCR thermocycling conditions were as follows: Initial denaturation at 95 ℃ for 52 seconds, followed by 40 cycles of 95 ℃ for 14 seconds and 58.5 ℃ for 26 seconds. The sequences of the primers used were as follows: LINC00092: forward 5'-TGAGGGCATTTTAGGAGGTGAG-3' and reverse 5'-AAATCATAGGGCAGGTCATCCC-3'. This experiment was performed in triplicate, and all quantitation cycle values were normalized to β-actin. Relative expression was quantified using the 2^−ΔΔCq method.

Detection of ESCC cell strain proliferation, cell migration and invasion detection

Cells were inoculated into 96-well plates at a concentration of 3×10³ cells per well and transfected with siRNA. At 0, 24, 48, 72, and 96 hours post-transfection, 20 µL of MTT solution (1.55 g/L) was added to each well. After incubating the cells for 4 hours at 37 °C, 150 µL of dimethyl sulfoxide was added to each well. Absorbance values were measured to construct cell growth curves. Cells from the experimental and treatment groups were inoculated into 6-well plates at a density of 1,000 cells per well. The culture medium was replaced after 4 days, and cells were fixed with formalin after 14 days, followed by crystal violet staining (0.5% crystal violet in 20% methanol for 30 minutes). Colonies with >50 cells were counted manually at a magnification of 1×. The wound healing and transwell assays were used to determine the migration and invasion. For wound healing assay, cells were cultured in 6-well plates at 5×10⁵ cells/well, then cells were scraped by a 200-µL pipette tip and created a linear regions. The scratch distance of the linear regions was evaluated by ImagePro 6.0 (Media Cybernetic, Rockville, Maryland, USA) at a magnification of 100×. For transwell assay, cells were transferred into upper Transwell chambers and cultured with serum-free medium. Then, culture medium was added to the lower chamber. After 24 hours, the cells were fixed with 4% paraformaldehyde and staining with 0.25% crystal violet solution. Then cells were counted at a magnification of 200×. Each experiment was performed in triplicate.

Flow cytometry

The concentration of ESCC cells in the logarithmic growth phase was adjusted to 3×10⁵/mL, and the cells were inoculated into 6-well plates at 2 mL/well. At 48 hours post-transfection, the cells were divided into experimental and control groups. After an additional 48-hour incubation, the cells from both groups were collected and washed with PBS. For apoptosis assessment, cells were double-stained using Annexin V/propidium iodide (PI) and incubated in the dark for 10 minutes. For apoptosis assessment, cells were resuspended in 100 µL of binding buffer and double-stained using Annexin V-FITC (5 µL; Invitrogen) and PI (10 µL of 50 µg/mL; Sigma) for 15 minutes in the dark. Flow cytometry was employed to determine the apoptosis rate in each group under consistent methodological conditions. Cells were resuspended in pre-cooled 75% ethanol and fixed at −20 ℃ overnight. Unstained and single-stained controls were included for compensation.

Western blot

Total protein was extracted from in vitro-cultivated cells (1×10⁵) using a Total Protein Extraction kit (catalog no. 2140, Merck KGaA, Darmstadt, Germany) according to the manufacturer’s protocol. The lysis buffer contained protease and phosphatase inhibitors. Protein concentrations were determined using a bicinchoninic acid (BCA) assay (Sangon Biotech Co., Ltd., Shanghai, China). Proteins (30 µg per lane) were separated by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and subsequently transferred to polyvinylidene fluoride membranes. Blocking was performed using PBS containing 5% skim milk for 1 hour at room temperature. Primary antibodies against MAZ (1:2,000; catalog no. ab85725, Abcam) and NFE2L2 (1:200; catalog no. ab62352; Abcam) were incubated at 4 ℃ for 18 hours. Membranes were washed three times with TBST for 10 minutes each. The secondary antibodies used were horseradish peroxidase-conjugated goat anti-rabbit IgG (1:1,000; catalog no. mBS435036; myBioSource, Inc., San Diego, CA, USA). Electrochemiluminescence (ECL)^™ Western Blotting Reagents (Sigma-Aldrich, Merck KGaA) were applied to develop signals. Densitometric analysis was conducted using ImageJ software v.1.46 (National Institutes of Health, Bethesda, MD, USA).

RNA immunoprecipitation (RIP)

RIP assays were conducted using the Magna RNA Immunoprecipitation Kit (Millipore, Burlington, MA, USA) according to the manufacturer’s instructions. Cells were lysed in RIP buffer containing magnetic beads conjugated with an IgG antibody (negative control) or anti-MAZ antibody (5 µg; Abcam). After incubation at 4 ℃ for 4 hours, beads were washed five times with RIP wash buffer. Subsequently, the immunoprecipitated RNA was isolated using TRIzol reagent, and RNA was treated with DNase I to remove genomic DNA. Input RNA (10% of total) was used as a positive control. LncRNA enrichment was analyzed using qRT-PCR, as described in section “qRT-PCR”.

Luciferase reporter assay

Based on the genomic sequence of NFE2L2 from the National Center for Biotechnology Information GenBank, a putative promoter region spanning from −2,000 bp to +200 bp relative to the transcription start site was amplified by PCR. This fragment was cloned into the pGL3-Basic vector (Promega, Madison, USA) to construct the luciferase reporter plasmid for the NFE2L2 wild-type promoter (pGL3-NFE2L2-WT). A mutant reporter plasmid (pGL3-NFE2L2-MUT) was generated from the wild-type plasmid using site-directed mutagenesis to disrupt the predicted MAZ binding sites. KYSE30 cells were seeded in 24-well plates. After 24 hours of culture, cells were co-transfected using Lipofectamine 2000 with the following: 100 ng of either pGL3-NFE2L2-WT or pGL3-NFE2L2-MUT reporter plasmid, 10 ng of pRL-TK vector (Promega) expressing Renilla luciferase as an internal control, and 50 ng of a MAZ overexpression plasmid (or its corresponding empty vector, pcDNA3.1, as a negative control). The total DNA amount was kept constant for all transfection groups. Forty-eight hours post-transfection, cells were lysed, and luciferase activity was measured using the Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s protocol. Firefly luciferase activity was normalized to Renilla luciferase activity for each sample. The relative luciferase activity of cells transfected with the empty vector control was set to 1.

Statistical analysis

Statistical analysis was performed using the software SPSS 26.0 (IBM Corp., Armonk, NY, USA). Data are presented as mean ± standard deviation, and t-tests were employed to compare groups. Diagnostic analyses were performed using receiver operating characteristic (ROC) curves. Differences with a P value of less than 0.05 were considered statistically significant.

Results

Identification of the ferroptosis-related LINC00092/MAZ/NFE2L2 axis in ESCC

A total of 161 DEFGs were identified from the differential expression analysis, comprising 90 downregulated and 71 upregulated genes (Figure 1A). WGCNA was performed on all DEFGs using the WGCNA package in R. Firstly, a hierarchical clustering tree of all DEFGs was constructed, resulting in the generation of 14 significant modules (Figure 1B). The dendrogram and heatmap of DEFGs indicated no significant differences in interactions among the modules, suggesting a high degree of independence (Figure 1C). As shown in Figure 1D, NFE2L2 was located in the tan module, which exhibited the highest correlation with the tumor. Co-expression correlation analysis of the lncRNAs and DEFG expression profiles, utilizing the limma package in R with criteria of |R| >0.4 and P<0.001, was conducted to identify ferroptosis-related lncRNAs. LINC00092 was identified as that most closely associated with NFE2L2 (Figure 1E,1F). A Venn diagram was constructed, revealing 11 intermediate TFs (Figure 1G), of which NFE2L2 showed the strongest association with LINC00092, according to catRAPID predictions. Additionally, we predicted LINC00092 downstream TFs using catRAPID (http://s.tartaglialab.com/page/catrapid group) and upstream TFs of NFE2L2 using the AnimalTFDB website (http://bioinfo.life.hust.edu.cn/AnimalTFDB4/#/). Three potential binding sites for NFE2L2 on LINC00092 were identified (Figure 1H-1J). Subsequent validation confirmed that LINC00092 binds to the transcription factor (TF) MAZ, thereby regulating the expression of NFE2L2.

Figure 1.

Identification of ferroptosis-related genes and regulatory networks in EC via bioinformatics analysis of TCGA data. (A) Volcano plot of differentially expressed ferroptosis-related genes between EC tissues (n=161) and normal tissues (n=11). Black dots represent non-significant genes; red and green dots indicate significantly up-regulated and down-regulated genes, respectively. (B) Hierarchical clustering tree of ferroptosis-related genes based on expression similarity. The colored bars beneath the dendrogram represent the gene modules identified by WGCNA. (C) Cluster dendrogram and heatmap showing the co-expression patterns of all analyzed genes. The color scale (blue to red) indicates low to high expression levels. (D) Module-trait relationships heatmap quantifying the correlations (color scale: −1 to 1) between gene modules (rows) and key clinical traits of EC (columns). (E) Venn diagram identifying 156 overlapping genes between the ferroptosis-related gene set (pink) and the genes within the significant ‘tan’ module (blue) from WGCNA. (F) Sankey diagram (flow diagram) visualizing the co-expression network, illustrating the connections between the ferroptosis process (left) and the specific genes (right) within the key module. (G) Correlation heatmap depicting the interrelationships (red indicates high positive correlation) between the core gene NFE2L2 and its associated lncRNAs. (H) Venn diagram showing the overlap of 11 TFs predicted to bind both NFE2L2 (pink circle) and LINC00092 (blue circle). The list of overlapping TFs is provided. (I) Sequence logo representing the canonical DNA binding motif for the transcription factor MAZ, as predicted by the JASPAR database. The height of each nucleotide represents its frequency at that position. (J) Schematic diagram of the NFE2L2 gene promoter region (−2,000 to +100 bp relative to TSS), showing the predicted binding sites for NFE2L2 itself and the transcription factor MAZ. EC, esophageal cancer; LINC00092, long intergenic non-protein-coding RNA 92; lncRNA, long non-coding RNA; MAZ, MYC-associated zinc finger protein; NFE2L2, NFE2 like bZIP transcription factor 2; TCGA, The Cancer Genome Atlas; TF, transcription factor; TSS, transcription start site; WGCNA, weighted gene co-expression network analysis.

Validation of LINC00092, MAZ, and NFE2L2 in ESCC

The ROC curves indicated that the areas under the curve (AUCs) for LINC00092, MAZ, and NFE2L2 were 0.830, 0.833, and 0.540, respectively (Figure 2A-2C), suggesting that these markers might serve as early diagnostic indicators for ESCC patients. Additionally, the CTD showed that LINC00092 was associated with neoplasm invasiveness, adenoma, and necrosis (Figure 2D); MAZ was linked to neoplasm invasiveness, death, and inflammation (Figure 2E); NFE2L2 was associated with carcinogenesis, ESCC, and neoplasm metastasis (Figure 2F). Furthermore, LINC00092 effectively predicted overall survival (log-rank P=0.04, Figure 2G) and disease-free survival (log-rank P=0.048, Figure 2H) in ESCC. Notably, LINC00092 was negatively correlated with the expression of the ferroptosis-related gene GPX4 (Figure 2I).

Figure 2.

Bioinformatic analysis of the LINC00092/MAZ/NFE2L2 regulatory axis in EC. (A-C) ROC curves evaluating the diagnostic value of LINC00092 (A), MAZ (B), and NFE2L2 (C) for distinguishing EC tissues (n=161) from normal adjacent tissues (n=11) based on TCGA-ESCA dataset. The AUC value for each gene is indicated. The analysis was performed using the R package “pROC”. (D-F) CTD analysis showing the association scores of LINC00092 (D), MAZ (E), and NFE2L2 (F) with cancer invasion and related diseases. The scores were retrieved from the CTD. (G,H) Survival analysis of LINC00092 expression in the TCGA-ESCA cohort using the GEPIA2 online tool. (G) Overall survival curve. (H) Disease-free survival curve. Patients were stratified into high- and low-expression groups based on the median expression level of LINC00092. The log-rank test was used for statistical comparison, and the HRs with their 95% confidence intervals and P values are displayed on the graph. (I) Correlation analysis between the expression levels of LINC00092 and the ferroptosis-related gene GPX4 in EC samples from the TCGA-ESCA dataset (n=172). The statistical significance and strength of the linear relationship were assessed using Pearson correlation analysis, and the correlation coefficient (r) and P value are shown. AUC, area under the curve; CTD, Comparative Toxicogenomic Database; EC, esophageal cancer; GEPIA2, Gene Expression Profiling Interactive Analysis 2; HR, hazard ratio; LINC00092, long intergenic non-protein-coding RNA 92; MAZ, MYC-associated zinc finger protein; NFE2L2, NFE2 like bZIP transcription factor 2; ROC, receiver operating characteristic; TCGA-ESCA, The Cancer Genome Atlas-esophageal carcinoma; TPM, transcript per million.

The immune activity in LINC00092

The immune microenvironment of tumors was characterized by ESTIMATE. The immune cell populations in each sample are shown in Figure 3A. The heatmap reflecting the distribution of the immune activity in EC was shown in Figure 3B. And the correlational heatmap of 21 immune cells is shown in Figure 3C. Macrophages (M0) and resting mast cells were significantly different (P<0.05) in LINC00092 high- and low-expression groups (Figure 3D). Principal component analysis (PCA) indicated that the immune cell components of different expression subgroups of LINC00092 were different (Figure 3E). Moreover, the expression of LINC00092 was positively correlated with dendritic cells resting (R=0.27, P=0.03), macrophages M1 (R=0.27, P=0.03) and mast cells resting (R=0.38, P=0.02), and negatively correlated with macrophages M0 (R=−0.27, P=0.03, Figure 3F). Microsatellite Analysis for Normal-Tumor InStability (MANTIS) is a tool for identifying microsatellite instability in paired tumor-normal patient samples. In this study, the MANTIS was employed to estimate microsatellite instability in LINC00092 high- and low-expression groups. In this study, the MANTIS in LINC00092 high-expression group was higher than low-expression group, indicating that LINC00092 expression was negatively correlated with MSI-H (Figure 3G). Moreover, Immune escape potentials were statistically significant at different levels of LINC00092 expression (Figure 3H). The high-expression subgroup of LINC00092 had a lower Tumor Immune Dysfunction and Exclusion (TIDE) score, indicating a better potential immunotherapeutic effect. The correlation analysis between linc00092 and immune checkpoint was shown in Figure 3I, and the expression of LINC00092 was positively correlated with BTN2A2, BTLA and TNFSF18, and negatively correlated with BTNL3, PVR and HLA-G. Finally, the difference in scores of immune features between high and low expression groups of Linc00092 was shown in Figure 3J. And the high-expression subgroup has a higher score in stromal fraction, intratumor heterogeneity, macrophage regulation lymphocyte infiltration signature score, B cell receptor (BCR) Shannon, transforming growth factor (TGF)-beta response, BCR Richness, T cell receptor (TCR) Shannon, and TCR Richness, and a lower score in wound healing and number of segments.

Figure 3.

Comparative analysis of tumor immune microenvironment activity between LINC00092 high- and low-expression groups in EC. All analyses were performed using transcriptomic data from the TCGA-ESCA cohort. The cohort was dichotomized into LINC00092 low-expression (n=81) and high-expression (n=80) groups based on the median expression value for all analyses. (A) Stacked bar plot showing the relative fractions of 22 types of immune cells in each sample within the low- and high-expression groups, as calculated by the CIBERSORT algorithm. (B) Heatmap displaying the standardized enrichment scores of various immune-related functions or cell populations (rows) across individual tumor samples (columns), stratified by LINC00092 expression level. (C) Correlation heatmap of the infiltration levels of 21 immune cell types across all EC samples. Correlation coefficients were calculated using Spearman’s rank correlation analysis. (D) Violin plots comparing the infiltration levels of 21 immune cell types between the LINC00092 low- (blue) and high-expression (red) groups. Statistical significance was assessed using the Wilcoxon rank-sum test. (E) Principal component analysis plot based on the transcriptomes of samples from the low- and high-expression groups. (F) Scatter plots illustrating the Spearman correlation between LINC00092 expression level and the infiltration abundance of specific immune cells. (G) Box plot comparing the MANTIS score (a measure of microsatellite instability) between the two groups. The P value was calculated using the Student’s t-test. (H) Box plot comparing a quantitative measure of immune escape potential between the two groups. Statistical analysis was performed using the Student’s t-test. (I) Scatter plots showing the Spearman correlation between LINC00092 expression and the expression levels of key immune checkpoint molecules. (J) Box plots comparing the scores of various immune signature pathways between the two groups. Statistical significance was determined using the Student’s t-test. *, P<0.05; **, P<0.01; ***, P<0.001. BCR, B cell receptor; EC, esophageal cancer; LINC00092, long intergenic non-protein-coding RNA 92; MANTIS, Microsatellite Analysis for Normal-Tumor InStability; PC, principal component; TCGA-ESCA, The Cancer Genome Atlas-esophageal carcinoma; TCR, T cell receptor; TGF, transforming growth factor; TIDE, Tumor Immune Dysfunction and Exclusion.

GSEA

Next, the potential functions and molecular mechanisms of LINC00092 were explored using GSEA. Patients with ESCC were divided into low and high LINC00092 expression groups for comparison via GSEA. Gene Ontology (GO) analysis indicated that cell-cell signaling by Wnt, epidermal development, and morphogenesis of a polarized epithelium were upregulated in the high LINC00092 expression group (Figure 4A). Conversely, the B cell receptor signaling pathway and calcium-independent cell adhesion via plasma membrane cell adhesion molecules (CAMS) were downregulated (Figure 4B). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed that the mitogen-activated protein kinase (MAPK) and mammalian target of rapamycin (mTOR) signaling pathways were upregulated in the high LINC00092 expression group (Figure 4C), whereas antigen processing and presentation, as well as CAMS, were downregulated (Figure 4D).

Figure 4.

Functional enrichment analysis of LINC00092 via GSEA. The analysis was performed using RNA-seq data from the TCGA-ESCA cohort, where samples were dichotomized into LINC00092 high-expression (n=80) and low-expression (n=81) groups based on the median expression level. (A,B) GO analyses conducted using GSEA. (C,D) KEGG analyses performed using GSEA. GO, Gene Ontology; GSEA, gene set enrichment analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; LINC00092, long intergenic non-protein-coding RNA 92; RNA-seq, RNA sequencing; TCGA-ESCA, The Cancer Genome Atlas-esophageal carcinoma.

Expression of LINC00092 in clinical ESCC samples

qRT-PCR was employed to assess the expression of LINC00092 in ESCC tissues obtained from patients. The correlation between LINC00092 expression and various clinicopathological parameters, including age, sex, tumor stage, histological grade, lymph node metastasis, and metastasis stage, was analyzed (Table 2). As shown in Table 2, LINC00092 expression was significantly associated with tumor stage (P=0.045) and tumor, node, metastasis (TNM) stage (P=0.03).

Table 2. Relationships between LINC00092 expression and the clinical-pathological features in ESCC tissues.

Groups	n	Expression of LINC00092		P value
		Low-expression	Over-expression
Age, years				0.70
<50	16	7	9
≥50	24	12	12
Gender				0.79
Female	7	3	4
Male	33	16	17
Differentiation				0.79
Undifferentiated	6	3	3
Poorly differentiated	20	10	10
Intermediate differentiation	3	2	1
Highly differentiated	11	4	7
T stage				0.047
T0	12	4	8
T1	7	1	6
T2	6	3	3
T3	11	7	4
T4	4	4	0
N stage				0.23
N0	23	9	14
N1	6	2	4
N2	10	7	3
N3	1	1	0
TNM stage				0.03
I	19	6	13
II	4	3	1
III	12	5	7
IVA	5	5	0

ESCC, esophageal squamous cell carcinoma; LINC00092, long intergenic non-protein-coding RNA 92; N, node; T, tumor; TNM, tumor, node, metastasis.

Immunohistochemistry and FISH

Immunohistochemistry results indicated that MAZ was highly expressed in ESCC tissues, whereas NFE2L2 exhibited low expression levels (Figure 5). FISH was conducted to determine the distribution of LINC00092 in KYSE30 and KYSE170 cells. As shown in Figure 6, LINC00092 was predominantly localized in the nucleus of KYSE30 cells. Similarly, LINC00092 was primarily localized in the nucleus of KYSE170 cells.

Figure 5.

Immunohistochemical analysis of MAZ and NFE2L2 in EC tissue (n=3 independent technical replicates per group). EC, esophageal cancer; MAZ, MYC-associated zinc finger protein; NFE2L2, NFE2 like bZIP transcription factor 2.

Figure 6.

FISH of LINC00092 in KYSE30 cell line (n=3) and KYSE170 cell line (n=3) revealed LINC00092 dual signal (red, Cy3 channel), and the nuclei were stained with DAPI (blue) in the images. DAPI, 4',6-diamidino-2-phenylindole; LINC00092, long intergenic non-protein-coding RNA 92.

Overexpression of LINC00092 in KYSE30 cells

To investigate the influence of LINC00092 on the biological function of ESCC cells, we employed qRT-PCR to assess LINC00092 expression in KYSE30, KYSE150, and KYSE170 cells. LINC00092 exhibited the highest expression in KYSE170 cells and the lowest in KYSE30 cells (Figure 7A). Subsequently, we overexpressed LINC00092 (ov-LINC00092) in KYSE30 cells (Figure 7B). The MTT assay was utilized to evaluate the impact of LINC00092 overexpression on the proliferation activity of KYSE30 cells, revealing a significant decrease in proliferation capacity in the experimental group compared to the control group (Figure 7C). Results from the colony formation assay corroborated those obtained from the MTT assay (Figure 7D). Furthermore, a wound healing assay was conducted to determine whether modulation of LINC00092 affected ESCC cell invasion and metastatic potential. The results indicated that the migratory and invasion capacities of the ov-LINC00092 group were markedly decreased relative to the control group (Figure 7E). ESCC cells were obtained using the same method and seeded into Transwell chambers. After 48 hours, the cells that migrated across the chamber were counted, and the results were consistent with those of the wound healing assay (Figure 7F). To assess the influence of LINC00092 on apoptosis in KYSE30 cells, cells were harvested, and apoptosis was evaluated via flow cytometry. The findings demonstrated an increased apoptosis rate in the experimental group compared to the control group (Figure 7G).

Figure 7.

Overexpression of LINC00092 in the KYSE30 cell line. (A) Relative mRNA expression levels of LINC00092 in various esophageal cancer cell lines (KYSE30, KYSE150, KYSE170). Data are presented as mean ± SE from three independent experiments (n=3). (B) Validation of LINC00092 overexpression efficiency in KYSE30 cells transfected with an ov-NC or a LINC00092 overexpression plasmid (ov-LINC00092), compared to a non-transfected control. Data are from n=3 independent experiments and were analyzed by one-way ANOVA. (C) Cell proliferation assessed by MTT assay in KYSE30 cells under different treatments (control, ov-NC, ov-LINC00092) at 0, 24, 48, and 72 hours. Data are presented as mean ± SE within a representative experiment, which was independently repeated three times. Statistical analysis was performed using two-way ANOVA. (D) Colony formation ability of KYSE30 cells under different treatments (n=3) (crystal violet staining). Data were analyzed by one-way ANOVA. (E) Cell migration ability evaluated by wound healing assay at 0- and 48-hours post-wounding (n=3) (×100). Data were analyzed by one-way ANOVA. (F) Cell invasion ability measured by Transwell invasion assay (n=3) (crystal violet staining, ×200). Data were analyzed by one-way ANOVA. (G) Apoptosis rate analyzed by flow cytometry in KYSE30 cells under different treatments (n=3). Data were analyzed by one-way ANOVA. *, P<0.05; **, P<0.01; ***, P<0.001. ANOVA, analysis of variance; FITC, fluorescein isothiocyanate; LINC00092, long intergenic non-protein-coding RNA 92; mRNA, messenger RNA; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; OD, optical density; ov, overexpression; ov-NC, empty vector negative control; PI, propidium iodide; SE, standard error.

LINC00092 knockdown in KYSE170 cells

We selected KYSE170 cells, which exhibited the highest relative expression levels of LINC00092, as model cells. A specific interference sequence targeting LINC00092 was designed and synthesized, then transiently transfected into KYSE170 cells using Lipofectamine^® 2000. After 48 hours, cellular RNA was extracted from both the experimental (si-LINC00092) and control groups to assess the knockdown efficiency of si-LINC00092 (Figure 8A), and sequence-3 was chosen for further experimentation. The results from the MTT and colony formation assays demonstrated that the proliferative capacity of the si-LINC00092 group increased (Figure 8B,8C). Additionally, the migration and invasion capacities in the si-LINC00092 group were markedly enhanced (Figure 8D,8E). Flow cytometry revealed that the si-LINC00092 group exhibited reduced levels of apoptosis (Figure 8F).

Figure 8.

Knockdown of LINC00092 in the KYSE170 cell line. (A) Validation of LINC00092 expression in the knockdown cell line (n=3). KYSE170 cells were transfected with a si-NC or a si-LINC00092, with untransfected cells as an additional control. The relative mRNA expression level of LINC00092 was quantified by qRT-PCR. Data were analyzed by one-way ANOVA. (B) Cell viability assessed by MTT assay in control, si-NC, and si-LINC00092 groups at 0, 24, 48, and 72 hours post-transfection (n=3). Data were analyzed by two-way ANOVA. (C) Results of the colony formation assay in LINC00092 knockdown cells (n=3) (crystal violet staining). Data were analyzed by one-way ANOVA. (D) Results of the wound healing scratch assay in LINC00092 knockdown cells (n=3) (×100). Data were analyzed by one-way ANOVA. (E) Results of invasion assays in LINC00092 knockdown cells (n=3) (crystal violet staining, ×200). Data were analyzed by one-way ANOVA. (F) Results of flow cytometry analysis of apoptosis in LINC00092 knockdown cells (n=3). Data were analyzed by one-way ANOVA. **, P<0.01; ***, P<0.001. ANOVA, analysis of variance; FITC, fluorescein isothiocyanate; LINC00092, long intergenic non-protein-coding RNA 92; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; OD, optical density; PI, propidium iodide; si-LINC00092, LINC00092-targeting siRNA; si-NC, non-targeting siRNA negative control; siRNA, small interfering RNA; qRT-PCR, quantitative real-time polymerase chain reaction.

Regulatory relationship of the LINC00092/MAZ/NFE2L2 axis

RIP assays using an anti-MAZ antibody revealed a significant reduction in the enrichment of LINC00092 in the precipitates from the si-LINC00092 group compared to the control group (Figure 9A). Upon overexpression of MAZ, NFE2L2 levels exhibited a significant increasing trend (Figure 9B). Overexpression of LINC00092 in KYSE30 cells resulted in a decrease in MAZ and NFE2L2 expression (Figure 9C). Conversely, in KYSE170 cells with LINC00092 knockdown, opposite effects were observed (Figure 9D).

Figure 9.

LINC00092 binds to and sequesters the transcription factor MAZ, thereby attenuating its ability to regulate the expression of its target gene, NFE2L2. MAZ, inhibiting its function and activating the expression of NFE2L2. (A) Anti-MAZ RIP assays were performed after the KYSE170 cell line was transfected with either si-NC or si-LINC00092 (n=3). Data were analyzed by two-way ANOVA. (B) The effect of MAZ mimics on luciferase activity of wild-type and mutant-type NFE2L2 vectors observed by dual-luciferase reporter assay (n=3). Data were analyzed by two-way ANOVA. (C) KYSE30 cells were transfected with an ov-NC or a ov-LINC00092 and compared to a non-transfected control. (D) KYSE30 cells were transfected with a si-NC or a si-LINC00092 and compared to a non-transfected control. Representative blots from three independent experiments (n=3) are shown. Protein levels were quantified by densitometry, normalized to β-actin. Data in the bar graphs are presented as mean ± SD and were analyzed by one-way ANOVA. *, P<0.05; **, P<0.01; ***, P<0.001. ANOVA, analysis of variance; LINC00092, long intergenic non-protein-coding RNA 92; MAZ, MYC-associated zinc finger protein; Mut, mutation; NFE2L2, NFE2 like bZIP transcription factor 2; ov, overexpression; ov-NC, empty vector negative control; RIP, RNA immunoprecipitation; SD, standard deviation; si-LINC00092, LINC00092-targeting siRNA; si-NC, non-targeting siRNA negative control; siRNA, small interfering RNA; WT, wild type.

LINC00092 exerts a tumor suppression effect via the LINC00092/MAZ/NFE2L2 axis in ESCC cells

To further explore the roles of the LINC00092/MAZ/NFE2L2 axis in ESCC cells, three cell models were established using ov-LINC00092, ov-NFE2L2, and a negative control. As shown in Figure 10A, overexpression of LINC00092 resulted in a decrease in NFE2L2 expression, an effect that was reversed by ov-NFE2L2. The MTT and colony formation assays demonstrated that cell proliferation was inhibited in the ov-LINC00092 group, whereas proliferation was enhanced in the ov-NFE2L2 group (Figure 10B,10C). Consistent findings were obtained in the wound healing and Transwell assays, where ov-NFE2L2 mitigated the suppressive impact of ov-LINC00092 on cell migration and invasion in ESCC cells (Figure 10D,10E). Collectively, these results indicate that LINC00092 exerts its tumor suppression effect via the LINC00092/MAZ/NFE2L2 axis in ESCC cells.

Figure 10.

NFE2L2 reverses the suppressive effects of LINC00092 on malignancy phenotypes in the KYSE30 cell line. KYSE30 cells were subjected to three treatments: (I) co-transfected with empty vectors (NC-LINC00092 + NC-NFE2L2), (II) transfected with ov-LINC00092 and an empty vector (NC-NFE2L2), (III) co-transfected with both ov-LINC00092 + ov-NFE2L2. All experiments were independently repeated three times (n=3). Data are presented as mean ± SEM. (A) Expression levels of NFE2L2 in the KYSE30 cell line following transfection. Data were analyzed by one-way ANOVA. (B) Cell viability measured by MTT assay in the KYSE30 cell line after co-transfection with ov-LINC00092 and ov-NFE2L2. Data were analyzed by two-way ANOVA. (C) Results of the colony formation assay in the KYSE30 cell line following co-transfection with ov-LINC00092 and ov-NFE2L2 (crystal violet staining). Data were analyzed by one-way ANOVA. (D) Results of the wound healing scratch assay in the KYSE30 cell line after co-transfection with ov-LINC00092 and ov-NFE2L2 (×100). Data were analyzed by one-way ANOVA. (E) Results of invasion assays in the KYSE30 cell line following co-transfection with ov-LINC00092 and ov-NFE2L2 (crystal violet staining, ×200). Data were analyzed by one-way ANOVA. **, P<0.01; ***, P<0.001. ANOVA, analysis of variance; LINC00092, long intergenic non-protein-coding RNA 92; NC, negative control; NFE2L2, NFE2 like bZIP transcription factor 2; ov, overexpression; SEM, standard error of the mean.

Effect of LINC00092 on the tricarboxylic acid (TCA) cycle

We investigated the impact of LINC00092 on TCA cycle activity in both ov-LINC00092 and si-LINC00092 groups. In the ov-LINC00092 group, levels of key TCA cycle enzymes and intermediates, including α-ketoglutarate dehydrogenase (α-KGDH), succinate dehydrogenase (SDH), pyruvic acid (PA), and isocitric dehydrogenase (ICDH), were significantly reduced (Figure 11A). Conversely, knockdown of LINC00092 resulted in an increase in α-KGDH, SDH, PA, and ICDH levels (Figure 11B).

Figure 11.

Effects of LINC00092 manipulation on key TCA cycle components in esophageal cancer cells. (A) Relative expression levels of α-KGDH, SDH, pyruvic acid, and IDH in KYSE30 cells. Cells were subjected to three treatments: untreated control, transfected with ov-NC, or transfected with ov-LINC00092. Data are presented as mean ± SEM from three independent experiments (n=3). Statistical significance was determined using one-way ANOVA. (B) Relative expression levels of α-KGDH, SDH, pyruvic acid, and IDH in KYSE170 cells. Cells were subjected to three treatments: untreated control, transfected with si-NC, or transfected with si-LINC00092-3. Data are presented as mean ± SEM from three independent experiments (n=3). Statistical significance was determined using one-way ANOVA. **, P<0.01; ***, P<0.001. α-KGDH, α-ketoglutarate dehydrogenase; ANOVA, analysis of variance; IDH, isocitrate dehydrogenase; LINC00092, long intergenic non-protein-coding RNA 92; ov, overexpression; ov-NC, empty vector negative control; SDH, succinate dehydrogenase; SEM, standard error of the mean; si-LINC00092, LINC00092-targeting siRNA; si-NC, non-targeting siRNA negative control; siRNA, small interfering RNA; TCA, tricarboxylic acid.

Discussion

In this study, we analyzed a microarray dataset of EC to identify lncRNAs associated with ferroptosis. LINC00092 was identified as a ferroptosis-related lncRNA. LINC00092 is downregulated in ESCC datasets, cell lines, and tissue samples. Overexpression of LINC00092 inhibits ESCC cell progression, whereas downregulation of LINC00092 promotes ESCC cell proliferation and migration. Therefore, we propose that LINC00092 could exert its tumor-suppressive effects, at least in part, through a pathway involving MAZ and NFE2L2. This potential axis warrants further investigation to confirm the direct transcriptional regulation and in vivo relevance.

LINC00092, located on chromosome 9q22.32, plays a significant role in various types of cancer. Previous studies have shown that LINC00092 acted as a nodal driver of metastatic progression mediated by cancer-associated fibroblasts in epithelial ovarian cancer. Mechanistically, LINC00092 binds to the glycolytic enzyme fructose-2, 6-bisphosphatase 2 (PFKFB2), promoting metastasis by altering glycolysis and maintaining local support functions (21). Zhao et al. reported that LINC00092 expression is downregulated in invasive breast ductal carcinoma and correlates with poor prognosis. Overexpression of LINC00092 inhibits cell proliferation, colony formation, migration, and invasion by modulating the miR-1827/SFRP1 axis (22). Despite these findings in other cancers, the role of LINC00092 in ESCC remains unclear. In this study, we investigated the expression of LINC00092 in ESCC. LINC00092 was found to be downregulated in both ESCC cell lines and tissue samples, suggesting a potential role in ESCC development. Subsequent in vitro experiments on multiple ESCC cell lines revealed that LINC00092 overexpression inhibited ESCC cell progression, whereas its downregulation promoted cell progression. To the best of our knowledge, this is the first study to elucidate the crucial role of LINC00092 in ESCC. The mechanism by which LINC00092 potentially regulates ferroptosis through MAZ/NFE2L2 exemplifies the diverse modes of action employed by non-coding RNAs in ESCC pathogenesis. This regulatory model—involving direct interaction with a TF—complements other recently described mechanisms. For example, the lncRNA MALAT1, which is highly expressed in TNBC, can act as ceRNA. By adsorbing miR-106a-5p, it up-regulates the expression of REEP5, forming the MALAT1/miR-106a-5p/REEP5 regulatory axis, promoting the proliferation, invasion, and metastasis of TNBC cells. And its expression level is positively correlated with the degree of lymph node metastasis (23). Moreover, Studies have identified lncRNA PVT1 as a key positive regulatory factor for copper mortality. The mechanism lies in the fact that PVT1 can epigenetically activate the transcription of the copper death core protein FDX1, thereby significantly enhancing the sensitivity of tumor cells to copper death. This provides a new idea for targeting the PVT1-FDX1 axis for the treatment of colorectal cancer (24). Our study contributes to this landscape by implicating ferroptosis as a functional outcome of lncRNA activity.

LncRNAs are known to exert their functions in a manner that is closely related to their precise subcellular localization. In this study, the subcellular distribution of LINC00092 was examined using FISH, revealing its localization predominantly within the nucleus. This finding is significant, as previous studies have demonstrated that lncRNAs can directly bind to TFs of target genes, thereby regulating gene expression (25,26). For instance, lncRNA HOXD-AS1, which is upregulated in cervical squamous cell carcinoma, promotes the disease’s progression. Mechanistically, HOXD-AS1 enhances FRRS1 expression by interacting with the TF ELF1, thereby influencing cervical cancer cell proliferation and apoptosis (27). Similarly, He et al. reported that lncRNA ZNF503-AS1 expression is downregulated in bladder cancer. Overexpression of ZNF503-AS1 attenuates bladder cancer cell proliferation, invasion, and migration, while promoting apoptosis, accompanied by reduced Ca²⁺-ATPase activities and elevated intracellular Ca²⁺ concentrations. Mechanistically, ZNF503-AS1 recruits the TF GATA6, upregulating SLC8A1 expression and thereby exerting tumor-suppressive effects in bladder cancer (28). In the current study, bioinformatic analysis identified MAZ as a TF potentially interacting with LINC00092. MAZ, located on chromosome 16p11.2, plays a pivotal role in gene expression regulation, signal transduction, and transcription by RNA polymerase II. Previous studies have reported that MAZ is upregulated in gastric cancer cells and inhibits cancer cell proliferation and migration while inducing apoptosis by modulating autophagy-related protein expression (29). In this study, we observed that MAZ was upregulated in ESCC. A luciferase reporter assay further demonstrated that MAZ directly binds to the ferroptosis-related gene NFE2L2. In this study, FISH experiments revealed that LINC00092 was primarily localized in the nucleus. This finding is crucial as it provides a localization basis for LINC00092’s function as a nuclear transcriptional regulator. Its nuclear localization coincides with the distribution area of the TF MAZ, which spatially supports the direct interaction between the two, as confirmed by RIP experiments. More importantly, this localization indicates that LINC00092 can directly act on chromatin and the transcriptional machinery, thereby providing a logical explanation for its enhancing effect on NFE2L2 transcriptional activity, as demonstrated in the luciferase reporter gene assays.

The NFE2L2 gene is located on human chromosome 2q31.2 and encodes a member of the basic leucine zipper (bZIP) protein family, which regulates genes containing antioxidant response elements (ARE) in their promoters. NFE2L2 plays a critical role in various cancers by modulating ferroptosis. For example, analysis of the FerrDb database suggested that NFE2L2 is associated with ferroptosis. It is overexpressed in gliomas and significantly correlates with poor prognosis in glioma patients. Furthermore, NFE2L2 is implicated in immune infiltration and tumor immunity during carcinogenesis (30). In glioma cells, LNC01564 inhibits ferroptosis and confers resistance to temozolomide by upregulating NFE2L2 expression (31). Ye et al. demonstrated that in hypopharyngeal squamous cell carcinoma, ALKBH5-mediated demethylation of N6-methyladenosine causes the posttranscriptional suppression of NFE2L2/NRF2. NFE2L2/NRF2 is crucial for regulating cellular antioxidant molecules. Knockdown of ALKBH5 subsequently elevates NFE2L2/NRF2 expression and enhances hypopharyngeal squamous cell carcinoma cells’ resistance to ferroptosis (32). Moreover, Vokshi et al. found that itaconic acid-induced NFE2L2 expression and activation serve as a defense mechanism limiting ferroptosis by upregulating antioxidant genes. Disruption of the NFE2L2 pathway increases susceptibility to itaconic acid-induced ferroptosis (33). These studies underscore the pivotal role of NFE2L2 in cancer and ferroptosis. However, the role of NFE2L2 in ESCC and its involvement in ferroptosis remain unexplored. In this study, RIP and luciferase reporter assays confirmed that LINC00092 regulates NFE2L2 expression by sponging the TF MAZ. These findings align with previous studies, reinforcing the function of NFE2L2. Our results demonstrate that the LINC00092/MAZ/NFE2L2 axis suppresses cancer progression in ESCC by promoting ferroptosis via NFE2L2 through the regulation of MAZ. Moreover, it is worth noting that gain-of-function mutations in NFE2L2, such as those in exon 2, are known to constitutively activate the pathway and influence ESCC prognosis (34). A parallel can be drawn with a recent study in hepatocellular carcinoma (HCC), which revealed that PIP5K1A promotes NRF2 stability and transcriptional activity by competitively binding to KEAP1, thereby inhibiting ferroptosis and contributing to sorafenib resistance. This mechanism reinforces our finding that the NRF2 axis is a common vulnerability point for therapeutic intervention in ferroptosis-resistant cancers (35). Further underscoring the broad relevance of ferroptosis induction in cancer therapy, research on gastric cancer revealed that a natural product, Mori Folium ethanol extract (MFEE), inhibits tumor growth by promoting ferroptosis. Mechanistically, MFEE acts by suppressing the PI3K/AKT pathway, which in turn modulates the AKT/GSK3β/NRF2 axis, leading to the downregulation of xCT and GPX4 (36).

While our study provides evidence supporting the role of the LINC00092/MAZ/NFE2L2 axis in ESCC ferroptosis, several limitations should be acknowledged. First, the precise molecular details of how MAZ binding influences NFE2L2 transcription require further elucidation. Second, although we observed phenotypes consistent with ferroptosis (e.g., changes in lipid ROS and cell viability upon specific inducers), more direct validation using a comprehensive set of ferroptosis markers would strengthen our conclusions. Third, the functional significance of this axis needs to be confirmed in vivo using animal models. Finally, the clinical relevance and therapeutic potential of targeting LINC00092 in ESCC remain to be explored. Future studies addressing these points will be crucial to understand the biological and translational impact of our findings fully.

Conclusions

This study revealed that LINC00092 is downregulated in ESCC datasets, cell lines, and tissue samples. Our data further suggest that LINC00092 may bind to the TF MAZ, thereby modulating the expression of NFE2L2. Collectively, these findings support a model wherein the LINC00092/MAZ/NFE2L2 axis could inhibit ESCC progression by promoting ferroptosis. However, this study has certain limitations, including the need for direct in vivo validation and more comprehensive ferroptosis marker analysis. Future research should focus on confirming this regulatory axis in animal models, performing ferroptosis rescue experiments, and exploring the translational potential of targeting LINC00092 or its downstream effectors in ESCC therapy.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of the Fourth Hospital of Hebei Medical University (No. 2022KY259). Informed consent was provided by all individual participants included in the study.

Data Sharing Statement

Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2433/dss

jtd-18-01-39-dss.pdf ^{(78.3KB, pdf)}

DOI: 10.21037/jtd-2025-1-2433