Super-enhancer-driven LncRNA MIR205HG promotes esophageal squamous cell carcinoma progression via glycolysis reprogramming Authors: Ze-Jun Zheng, Yin-Qiao Liu, Yan-Shang Li, Dong-Chen Han, Jun-De Zhu, Qi-Xin Su, Zhi-Ya Wang, Chun Li, Zhuo-Ying Kang, Jin-Cheng Guo, Ying-Hua Xie, Jing-Ru Ye, Lu-Shuang Mao, Jiang-Yun Peng, Xing-Dong Xiong, Jian-Jun Xie Published: 2026 ABSTRACT AbstractBackgroundSuper-enhancer-associated long non-coding RNAs (lncRNAs) have been shown to play key roles in the occurrence and development of malignant tumors, including esophageal squamous cell carcinoma (ESCC), yet their precise molecular mechanisms remain elusive.MethodsChIP-Seq, dual-luciferase reporter assays, HiChIP-Seq, and ChIP-qPCR were performed to investigate the transcriptional regulation mechanism of MIR205HG. The functions and downstream signal transduction mechanisms of MIR205HG in ESCC were explored by a series of in vitro and in vivo assays. Furthermore, comprehensive bioinformatics methods were used to analyze its correlation with ESCC patient survival.ResultsHere, we identified and characterized MIR205HG, a lncRNA driven by super-enhancer, as a crucial oncogene in ESCC. MIR205HG was up-regulated in SCCs and its high expression correlated with poor clinical outcomes. Both TP63 and KLF5, two important master transcription factors in ESCC, could simultaneously occupy the super-enhancer region at the MIR205HG locus to promote its transcription and overexpression. MIR205HG is essential for ESCC proliferation, migration, invasion, and the growth of xenograft tumors in vitro and in vivo. Mechanistically, MIR205HG directly bound PTBP3 and acted as a molecular scaffold to promote HIF-1α translation, leading to enhanced cellular glycolysis via up-regulating the expression of HK2 and LDHA. Moreover, survival and pseudotime analyses of ESCC scRNA-seq data revealed a significant positive correlation between MIR205HG/PTBP3 signaling and the stemness and malignancy of ESCC cells. Finally, we showed that specifically targeting MIR205HG-SE using a CRISPR interference method resulted in potent suppressive effects on ESCC malignant phenotypes.ConclusionWe identified a pivotal oncogenic super-enhancer-driven lncRNA, MIR205HG, which interacts with PTBP3 to promote glycolysis in ESCC. It may serve as a promising prognostic biomarker and therapeutic target for patients.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12967-026-08202-1.Keywords: Oncogenesis, Gene regulation, Super-enhancer, Glycolysis, Epigenetic regulation Background Esophageal squamous cell carcinoma (ESCC), a common malignant digestive tumor, is the primary histological subtype of esophageal cancer, accounting for approximately 85% of all cases [1]. However, its pathogenesis remains unclear. Current challenges in the diagnosis and treatment of ESCC include the lack of specific and sensitive biomarkers for early diagnosis, tumor recurrence, and resistance to chemoradiotherapy, leading to poor patient prognosis [2]. Therefore, there is an urgent need to explore the pathogenesis of esophageal cancer, identify specific and sensitive biomarkers, and develop novel and effective therapeutics. Depending on their tissue- and condition-specific expression patterns, long non-coding RNAs (lncRNAs) may hold great potential as clinical targets and biomarkers for the early detection of ESCC [3]. LncRNAs are a class of non-coding RNAs with a sequence length of more than 200 bp that are aberrantly expressed in various tumors and are involved in malignant phenotypes, including cell proliferation, chemoradiotherapy resistance, epithelial-mesenchymal transition (EMT), angiogenesis, stemness, and other pathological processes [4, 5]. LncRNAs can regulate gene expression and tumor progression not only through transcriptional regulation via chromatin remodeling, recruitment of histone modification complexes, and interaction with transcription factors (TFs), but also through post-transcriptional regulation by interacting with molecules such as RNA and RNA-binding proteins (RBPs) [6, 7]. Several studies have shown that lncRNA MIR205HG, previously named LINC00510 or NPCA-5, is abundant in various tumors, such as cervical cancer, lung squamous cell carcinoma (LUSC), head and neck squamous cell carcinoma (HNSCC), and hepatoblastoma [8–11]. These studies have also shown that MIR205HG plays a significant role in the initiation and progression of various cancers. However, the molecular mechanisms underlying its abnormal expression and the oncogenic role of MIR205HG remain unclear. Over the last decade, the concept of super-enhancers has been proposed, providing novel insights into aberrant gene expression in cancer. Super-enhancers are tissue-specific cis-regulatory elements composed of transcriptional activity enhancer clusters that can recruit key transcriptional activators, cofactors, and RNA polymerase II at high densities and that drive expression of genes associated with cell identity and malignancy [12, 13]. In ESCC, super-enhancers have been identified that were found to play critical roles in both oncogene activation and tumor suppressor gene inactivation [14]. Several studies have demonstrated that super-enhancers hijack promoters to upregulate lncRNAs, known as super-enhancer-associated lncRNAs (SE-lncRNAs) [15, 16]. Accordingly, our previous studies have shown that super-enhancers are particularly important at the upregulation of lncRNAs in ESCC [17]. Metabolic reprogramming is a defining feature of cancer cells, allowing them to reprogram energy metabolism, including that of glucose, glutamine, and fatty acids to support their rapid proliferation. Notably, glucose metabolism reprogramming underpins altered metabolism. In 1927, Otto Warburg proposed that cancer cells exhibit a unique metabolic phenotype involving the preferential conversion of glucose to lactate via fermentation, even when there is sufficient oxygen available for mitochondrial oxidative phosphorylation (OXPHOS) [18]. This phenomenon is known as the “Warburg Effect”. Numerous transcription factors, such as p53, c-Myc, and hypoxia inducible factor-1 (HIF-1), have been implicated in regulating this altered energy metabolism. For example, hypoxic conditions in cancer cells stabilize HIF-1α expression, thereby enhancing the expression of transporters and key enzymes involved in glycolysis. While this glycolysis reprogramming, which plays an important role in tumorigenesis, is observed across various tumors, specific studies on its role in esophageal cancer remain limited. In this study, we found that in ESCC, MIR205HG was a novel super-enhancer-driven oncogenic lncRNA. The core transcriptional circuitry (CRC) TP63/SOX2/KLF5 activated MIR205HG transcription through a super-enhancer and promoted its upregulation. Moreover, we showed that MIR205HG facilitated ESCC progression via cellular glycolysis reprogramming and targeting the MIR205HG-SE might be an efficient approach to suppress ESCC. Methods Cell culture and treatment The basic medium of HEK293T cells was high glucose Dulbecco’s Modified Eagle Medium (DMEM) (Gibco, Cat#C11995500BT), while the basic medium for TE7 cells was high glucose DMEM (Solarbio, Cat#11995). All the other cell lines used in this study were cultured in RPMI 1640 (Gibco, Cat#C11875500BT) with 10% FBS (TransGen, Cat#FS201-02). For hypoxia in vitro model establishment, ESCC cells were treated with 100 μM cobalt chloride (CoCl2) (Sigma-Aldrich, Cat#202185) for 6 h. Glycolysis related inhibitors, such as 2-Deoxy-D-glucose (2-DG, MCE, Cat#HY-13966), DICA (MCE, Cat#HY-B0486), and FX-11 (TargetMol, Cat# T15362), were added into ESCC cells after MIR205HG knockdown. Clinical specimens and microarray analysis The clinical ESCC tissue cDNA microarray (ID: HEsoS095Su01) was purchased from Outdo Biotech Co., Ltd. (Shanghai, China). The microarray included a total of 67 cases of ESCC, among which 14 cases had normal paired adjacent tissues. We obtained detailed clinical information of the cases from Outdo BioTech and the ethical approval for the study of the cDNA microarray was provided by the Clinical Research Ethics Committee, Taizhou Hospital of Zhejiang Province (Table S1). The relative expression of MIR205HG was measured via RT-qPCR. Transcriptome data from 179 pairs of ESCC and adjacent normal esophagus samples were downloaded from GSE53625 [19], which includes GSE526252 and GSE52624. Kaplan-Meier analysis was performed to analyze the clinical relevance of MIR205HG expression. Plasmids, small interfering RNAs, and cell transfection Expression plasmid pCMV-PTBP3 (Cat#HG21894-UT) was purchased from Sino Biological. Small interfering RNA (siRNA) against MIR205HG, TP63, SOX2, KLF5 and PTBP3 were obtained from Genepharma. DNA vectors and siRNA were transfected into cells using Lipo8000™ and LipoRNAi™ Transfection Reagent (Beyotime) respectively, according to the manufacturer’s instructions. Subsequent experiments were carried out 36 h after transfection. The siRNA sequences are listed in Table S2. The shRNA sequences targeting MIR205HG were cloned into the pLKO.1-TRC Vector (Addgene), while MIR205HG was cloned into the pLVX-DsRed-Monomer-N1 Vector (Clontech). The plasmids were co-transfected into 293T cells with packaging vectors, psPAX2 and pMD2.G. Then, the supernatant containing viral particles was harvested every 12 h for 60 h post-transfection. All ESCC cells were infected with the supernatant in the presence of 5 µg/mL Polybrene (Santa Cruz). Puromycin (Santa Cruz) was used to screen the stable expression cell lines. The cloning primers sequence are listed in Table S3. According to the ChIP-Seq profile, individual super-enhancer constituents were cloned into firefly luciferase reporter vector pGL3-Promoter (Promega). As for the experiment to measure HIF-1α translation efficiency, the 5′UTR sequence was cloned into the pGL3 Vector (Promega). The primer sequences for constructing pGL3-promoter luciferase plasmids are listed in Table S4. CRISPRi-mediated transcriptional repression CRISPRi is a transcriptional repression system using a catalytically inactive Cas9 (dCas9) fused to transcriptional repressors, such as KRAB domain, to block transcription initiation or elongation at targeted promoters, enabling reversible and specific gene silencing without DNA cleavage [20]. Here, target sequences of the MIR205HG-SE were designed using CHOPCHOP (http://chopchop.cbu.uib.no/), with high-scoring sgRNAs selected (Table S5). SgRNAs were cloned into the pLV hU6-sgRNA hUbC-dCas9-KRAB-T2a-Puro vector (Miaolingbio) expressing dCas9-KRAB. Lentiviruses were produced in HEK293T cells. TE1 cells were transduced and selected with 2 µg/mL puromycin to generate stable lines. The inhibition of the enhancer was verified by RT-qPCR for MIR205HG expression. Nude mouse tumor xenograft assay All animal procedures were conducted after approval from the Animal Ethics Committee of the Medical College, Shantou University (Approval Number: SUMC2020-366). Briefly, 3 × 106 KYSE510 cells stably transfected with pLKO.1-shNC or pLKO.1-shMIR205HG plasmids were subcutaneously injected into 4-week-old BALB/c male nude mice (n = 5). The tumor sizes were measured with a Vernier caliper every three days and calculated by using the following formula: tumor volume (mm3) = length × width2 /2. After the mice in each group were sacrificed by CO2, the xenograft tumors were removed, imaged, and weighed. The tumors were then fixed in 4% paraformaldehyde and embedded in paraffin for H&E staining. In addition, the RNA of each tumor tissue was extracted and the expression of MIR205HG was verified by qPCR after reverse transcription. Total RNA extraction and RT-qPCR Total RNA was extracted using Vazyme RNA-Easy Isolation Reagent Kit and reverse transcribed into cDNA according to HiScript® III RT SuperMix for qPCR (+ gDNA wiper) protocol. Primers were synthesized by BGI Tech Solutions (Beijing Liuhe), Beijing, China. Sequences were listed in Table S6. Real-Time PCR reactions were performed according to the ChamQ Universal SYBR qPCR Master Mix reagent instructions. Based on β-actin as the endogenous control, the relative quantity of each target was determined by comparative Ct (ΔΔCt) method. Western blotting analysis Protein extracts were obtained from the specified cells after lysis with 1× SDS buffer and boiled for 5 min to denature. The proteins were resolved by 10% separation gel, transferred to PVDF membranes, and blocked with 5% skim milk. The protein of interest was probed with antibodies against PTBP3 (Santa Cruz, Cat#sc-100845, and Invitrogen, Cat#PA598125), normal mouse IgG (Santa Cruz, Cat#sc-202), HIF-1α (Huabio, Cat# HA722778), HK2 (Abclonal, Cat#A0994), LDHA (Proteintech, Cat#21799-1-AP), PFKP (Proteintech, Cat#13389-1-AP), PKM2 (Santa Cruz, Cat# sc-365684), GAPDH (Proteintech, Cat#60004-1-Ig), Lamin A/C (Santa Cruz, Cat#sc-376248), and β-actin (Santa Cruz, Cat#sc-47778). Routinely, peroxidase-conjugated anti-mouse IgG (Invitrogen, Cat#C31430100) or anti-rabbit IgG (Invitrogen, Cat#C31460100) were used as secondary antibodies. The signal could be detected by ECL chemiluminescence Detection Kit (Vazyme). Chromatin immunoprecipitation-qPCR analysis Chromatin immunoprecipitation (ChIP) was performed as described previously [17]. Cells were cross-linked with 1% formaldehyde and neutralized by 1.25 M glycine followed by nuclear extraction. DNA was sheared by sonication (M200, Covaris). After washing, the sonicated lysates were incubated with magnetic beads (ThermoFisher, Cat#88803) coupled with one of the antibodies: H3K27ac (Abcam, Cat#ab4729), TP63 (CST, Cat#13109S), SOX2 (Santa Cruz, Cat#sc-365823), KLF5 (Abcam, Cat#ab137676), H3K27ac (Abcam, Cat#ab4729) and Normal Rabbit IgG (CST, Cat#2729S) antibody. The enriched DNA, after being eluted and purified with QIAquick PCR Purification Kit (QIAGEN), were analyzed by qPCR. The corresponding ChIP-qPCR primers are listed in Table S6. Cell proliferation assays After successful transfection, cells were serum-starved for 12 h. Then, 5000 cells were seeded in each well of a 96-wells plate, and the optical density (OD) value at 490 nm was measured after adding MTS and incubating for 1.5 h as part of the MTS assay. For the colony formation assay, 1000 cells/ well were seeded and grown in 6-well plates for 2 weeks. The colonies were fixed with methanol and stained with crystal violet. Transwell assays For the Transwell invasion assay, the upper chamber (Falcon) was pre-coated with Matrigel (BD Biosciences). Afterwards, 2 × 105 cells (KYSE510 and KYSE150) or 5 × 104 (KYSE30 and TE7) cells in FBS-free medium were seeded in the upper chamber, while medium containing 10% FBS was added as attractant in the bottom chamber. After culturing for 24 (migration) or 48 h (invasion) in an incubator at 37 °C with 5% CO2, unmigrated cells in the membranes of the upper chamber were wiped off with cotton swabs, and migrated ones were fixed with methanol, stained with crystal violet, and quantified at a 200× magnification. Dual-luciferase reporter assays The cells were co-transfected with pGL3-promoter or pGL3-basic plasmids containing corresponding sequences and Renilla luciferase plasmid. After 36 h, the luciferase reporter activities were measured by using the Dual-Luciferase Reporter Assay System (Promega). The activity of Renilla luciferase reporter was used as a normalization control for both. RNA subcellular isolation RNA Subcellular Isolation Kit reagent (Active motif, Cat#AM.25501) was used for the nuclear/cytoplasmic fractionation of RNA. The cultured cells were lysed by complete lysis buffer. The lysate was separated by centrifugation, with the supernatant containing cytoplasmic RNA and the pellet containing nuclear RNA. Each RNA fraction was loaded onto individual spin columns after being added guanidine-based buffer and ethanol gradually. The RNA fractions were eluted with DEPC water to wash off proteins. RT-qPCR was performed subsequently, where GAPDH was used as the cytoplasmic control for normalization, U6 was used as the nuclear one, and the percentage was calculated as follows: ΔCt = Ct nucleus − Ct cytoplasm, cytoplasm(%) = 1/ (1 + 2−ΔCт), nucleus(%) = 100% − cytoplasm(%). RNA fluorescence in situ hybridization (FISH) RNA FISH assays were carried out following the manufacturer’s instructions for the RNA Fluorescence in Situ Hybridization Kit (GenePharma). A set of hybridization probes labeled with FAM and targeting MIR205HG was designed by GenePharma (Table S7). Cells were seeded into glass culture slides (Corning) and incubated for 12 h. After fixation with 4% formaldehyde and permeabilization treatment with 0.1% Triton X-100, the cells were then incubated overnight in hybridization buffer containing 1 nM MIR205HG probe at 37 °C. Finally, cells were washed and counterstained with DAPI. Microscopically, the subcellular distribution and localization of MIR205HG were observed. RNA pull-down assay RNA pull-down experiments were performed following the manufacturer’s instructions for the TranscriptAid T7 High Yield Transcription Kit (Thermo Scientific, Cat#K0441), GeneJET RNA Purification Kit (Thermo Scientific, Cat#K0731), and Pierce™ Magnetic RNA-Protein Pull-Down Kit (Thermo Scientific, Cat#20164). Briefly, MIR205HG was cloned into the T7-cloning vector (TransGen), and then used to amplify DNA templates for RNA synthesis. RNA was transcribed in vitro with T7 RNA polymerase and labeled with biotin. Cell lysates were attained by lysing the cells with polysome extraction buffer (PEB) containing protease and RNase inhibitors. Cell lysates, purified biotinylated RNA and streptavidin-coupled Dynabeads were mixed and co-incubated overnight at 4 °C, which resulted in the immunoprecipitated complex. Elution buffer was added to the immunoprecipitated complex to collect the supernatant, and then discarded the magnetic beads. Interacting RBPs were detected by mass spectrometry (Table S8-9) and Western blotting analyses. RNA immunoprecipitation (RIP) and Crosslinking and Immunoprecipitation (CLIP) RNA immunoprecipitation (RIP) was performed as described previously [21]. Preincubated with magnetic beads for one hour, normal mouse IgG (Santa Cruz, Cat#sc-2025) or anti-PTBP3 antibody (Santa Cruz, Cat#sc-100845) were added into cell lysates and incubated overnight at 4 °C. After washing the immunoprecipitation complex carefully, precipitated RNAs were purified using Trizol buffer. RT-qPCR assay was done to measure the enrichment of MIR205HG by the PTBP3 protein. CLIP was performed following a previously reported protocol with some modifications [22]. Cells were seeded in four 15 cm plates and treated with 100 µM 4-thiouridine (GLPbio, Cat#GC42474) for 16 h. Next, the cells were crosslinked with 150 mJ/cm2 of 365 nm UVA light for 10 min, scraped in 5 mL of ice-cold PBS. The cell pellets were resuspended in NP-40 lysis buffer for 15 min at 4 °C and then digested with 1 U/µL of RNase T1 (Thermo Scientific, Cat#EN0541) at 22 °C for 10 min. Subsequent immunoprecipitation steps were performed as described for the RIP assay. Nuclear/cytoplasmic fractionation of proteins The cells were lysed in cytoplasmic lysate buffer (0.01 M Tris-HCl, 0.01 M NaCl, 3 mM MgCl2, 0.03 M Sucrose, 0.25% NP-40) and centrifuged after incubating on ice for 10 min. The lysate was separated by centrifugation, with the supernatant containing cytoplasmic fractions and the pellet containing nuclear fractions. The pellet was washed using cytoplasmic lysate buffer twice, and lysed with RIPA buffer. Protein concentrations were determined by BCA assay, and equal amounts (20 µg/well) were subjected to Western blot analysis. Lamin A/C was used as the nuclear marker, while GAPDH as the cytosolic one. The percentage of protein in each compartment was calculated as: cytoplasmic percentage = [cytoplasmic signal / (cytoplasmic signal + nuclear signal)] × 100%; nuclear percentage = 100% – cytoplasmic percentage. Detection of glucose uptake and lactate production The transfected ESCC cells were cultured in a 96-well plate for 48 h before the supernatant was collected. The glucose uptake assay was measured by using Glucose Uptake Colorimetric Assay Kit (Sigma, Cat#3091336) following the manufacturer’s protocol, while lactate production assay of supernatant was measured by using L-Lactate Assay Kit (AAT Bioquest, Cat#AAT-13815). The rate of glucose uptake and lactate production were analyzed by normalizing the values to the indicated control group. ECAR measurement Extracellular acidification rate (ECAR) was measured using a Seahorse XFe24 Flux Analyzer (Agilent) and the Seahorse XF Glycolysis Stress Test Kit (Cat#103020-100) [23]. Briefly, 1.3 × 10⁴ transfected cells were seeded in a 24-well Seahorse plate and incubated overnight. One hour before the assay, the medium was replaced with XF Base Medium (pH 7.4), and the plate was equilibrated in a non-CO₂ incubator at 37 °C for 60 min. During the assay, glucose (10 mM), oligomycin (1 µM), and 2-deoxy-D-glucose (50 mM) were sequentially injected. ECAR values were calculated using Seahorse software. Data analysis Sequencing reads were aligned to the reference genome (GRCh38/hg38) with Bowtie Aligner (1.2.2), and the ChIP-Seq files were converted into bigwig files using wigToBigWig tool (http://hgdownload.cse.ucsc.edu/admin/exe/) and visualized in Integrative Genomics Viewer (IGV). Super-enhancer regions were identified using Rank Ordering of Super-Enhancers (ROSE) algorithm. Briefly, H3K27ac peaks were stitched together if within 12.5 kb, excluding regions within 2 kb of transcription start sites to remove promoter-associated signals. Stitched enhancers were ranked by H3K27ac signal intensity (IP signal - Input signal), and a line with slope of 1 was fitted to the ranked curve. Super-enhancers were defined as those with signal intensity above the inflection point of the curve. Single-cell sequencing data of ESCC was obtained from GSE188900 [24], GSE160269 [25], and GSE197677 [26]. Genes expressed in less than 5 cells or with expressions below 500 genes per single cell were filtered out, and the percentage of mitochondria (10%) was calculated by PercentageFeatureSet function of the Seurat R package [27]. Transcriptional data from 81 ESCC samples were collected from the TCGA database, excluding samples lacking survival data and non-ESCC. The scRNA-Seq data of ESCC was reanalyzed using the Seurat package to provide features relevant to single-cell survival. The data preprocessing encompassed cells expressing 1,000 to 10,000 genes and underwent the “Modify the style” tool of the diagram. The FindIntegrationAnchors function was employed to merge the data, followed by dimensionality reduction using the t-distributed stochastic neighbor embedding (tSNE) method (resolution = 0.1), and selecting 50 principal components. Cells were classified into different subgroups using tSNE (dim = 50, resolution = 0.1), and the RuntSNE function was employed for tSNE dimensionality reduction. The Bioconductor/R software package Scissor (v2.0.0) [27] was used to identify cell subgroups associated with survival. Prepared datasets included the single-cell expression matrix from GSE188900, the bulk RNA expression matrix from TCGA, and survival information from bulk RNA-Seq samples. Scissor’s default parameters were utilized to identify positive and negative cells associated with prognosis. Single-cell pseudotime trajectories were generated using the monocle (2.24.0) package in R4.2.2 [28]. Functions such as newCellDataSet, estimateSizeFactors, and estimateDispersions were used for pseudotime analysis. The detectGenes function (min_expr = 0.1) was used to filter out low-quality cells, while functions like plot_cell_trajectory, plot_genes_jitter, plot_genes_violin, and plot_genes_in_pseudo-time were employed for data visualization. Statistical analysis Patients with ESCC were divided into high- and low-expression groups based on the optimal cut-points for MIR205HG or PTBP3 expression level, which were calculated using the X-tile software (version 3.6.1) [29]. Kaplan-Meier curve was performed for overall survival analysis using a Log-rank test. Student’s t-test was used for comparisons between two groups, while one-way ANOVA was employed for multiple-group comparisons. The values are shown as mean ± SD. Statistical charts were drawn by Prism 8 software. The differences with *P < 0.05, **P < 0.01, ***P < 0.001 were considered statistically significant. Results Super-enhancer-associated lncRNA MIR205HG is uniquely highly expressed in SCC tumors In our previous study, we identified super-enhancer-associated genes in several ESCC cell lines using chromatin immunoprecipitation sequencing (ChIP-Seq) with antibodies against H3K27ac (GSE106433) [17]. Here, we screened and discovered 44 lncRNAs that were potentially driven by super-enhancers in six ESCC cell lines (Fig. S1A and B). Notably, we found that only two lncRNAs, MIR205HG (ENSG00000230937) and LINC01451, were substantially upregulated in ESCC tissues compared to their levels in both esophageal adenocarcinoma (EAC) and normal tissues (Fig. 1A, Figure S1C and D). Moreover, besides ESCC, MIR205HG and LINC01451 showed uniquely high levels in other squamous cell carcinomas (SCCs), such as HNSCC, cervical squamous cell carcinoma (CESC), and LUSC, suggesting that the regulation of SE-lncRNAs may be SCC-specific (Fig. 1B, Figure S1E). MIR205HG was selected for further study due to its high expression level among the 44 SE-associated lncRNAs in ESCC and its large fold change in expression in ESCC tissues compared to adjacent tissues (Fig. 1A and D). MIR205HG shows uniquely high-expression in SCC tumors. (A) The expression heatmap of super-enhancer-associated lncRNAs in ESCC tissues compared with normal tissues or EAC tissues in TCGA and GTEx gene expression dataset. (B) Expression levels of MIR205HG in TCGA normal and tumor samples. Each point represents one tissue sample. (C) Expression levels of MIR205HG in cancer cell lines were derived from the CCLE project. From left to right, the expression of MIR205HG in each cell line decreases progressively. (D) Expression levels of MIR205HG in ESCC, EAC and normal esophageal tissues were analyzed in TCGA gene expression datasets. (E) RT-qPCR was used to detect the expression of MIR205HG in ESCC cell lines (red) and ECA (black) cell lines. (F-H) RNA levels of MIR205HG were derived from the indicated Gene Expression Omnibus datasets measuring matched ESCC and adjacent non-malignant esophageal epithelium. (I) The expression levels of MIR205HG were measured in 67 ESCC tissues and 14 matched adjacent non-malignant esophageal epitheliums using RT-qPCR. (J and K) Kaplan-Meier analysis showing the association between the expression of MIR205HG and the overall survival of patients with ESCC Complementary DNA microarray data from the Cancer Cell Line Encyclopedia (CCLE) project supported the RNA-Seq results from the TCGA dataset and showed that MIR205HG was expressed at the highest level in SCC cells, including HNSCC and ESCC cells (Fig. 1C). RT-qPCR analysis confirmed that MIR205HG expression was significantly higher in the majority of ESCC cell lines (red columns) compared to EAC cell lines (black columns, OE33 and SKGT4 cells), with the exception of KYSE410 (Fig. 1E). The upregulation of MIR205HG in ESCC samples in comparison with matched non-malignant esophageal mucosa was confirmed by three additional public datasets (Fig. 1F-H). Moreover, we detected MIR205HG expression in a cDNA microarray of ESCC tissues (HEsoS095Su01, including 67 tumor and 14 non-tumor tissues) by using RT-qPCR and confirmed that MIR205HG expression was markedly higher in cancerous tissues (Fig. 1I). Notably, MIR205HG expression was already elevated in early-stage ESCC and remained sustained throughout progression, with no significant increase from Stage I to Stage IV (Figure S1F-H), suggesting it is an early oncogenic event. Finally, the clinical relevance of MIR205HG expression was investigated using Kaplan-Meier analysis. It is noteworthy that elevated MIR205HG expression was strongly correlated with poorer overall survival in ESCC patients (Fig. 1J and K). MIR205HG is a SCC-specific super-enhancer-driven gene To elucidate the transcriptional regulatory mechanisms underlying the abnormally high expression of MIR205HG, we analyzed the H3K27ac profiles in six ESCC and three EAC cell lines. An SE peak upstream of the MIR205HG gene locus was found in all six ESCC cell lines but was absent in the EAC cell lines (OE19, Flo-1, and OE33), suggesting that the regulation of MIR205HG by SE is specific to the tumor subtype (Fig. 2A). Moreover, we analyzed H3K27ac ChIP-Seq data from four ESCC/normal matched cases. Of note, SE was identified in two ESCC samples (ESCC#1 and ESCC#2) but was absent in the matched normal tissues. However, in the other two ESCC cases (ESCC#3 and ESCC#4), SE was not detected in either tumor or normal tissues, indicating tumor heterogeneity in ESCC (Fig. 2A). Subsequently, HiChIP-Seq was performed to determine the global chromatin interactions in ESCC cells. The HiChIP tracts revealed connected loops between the promoter region of MIR205HG and the enhancer constituents of the MIR205HG-SE, as well as regional interactions between different SE constituents, indicating strong enhancer activity and transcriptional regulation within this genomic region (Fig. 2B). Transcription of MIR205HG is regulated by a super-enhancer. (A) ChIP-Seq profiles of H3K27ac at MIR205HG gene locus in six ESCC cell lines, three ECA cell lines, and four paired ESCC/match normal tissue samples. Upper blue bar denotes the SE regions. Based on the H3K27ac binding peak, nine constituent enhancers were cloned into dual-luciferase reporter vectors, and Ctrl was used as a negative control. (B) Human genome tracks show the ChIP-Seq signals of H3K27ac, H3K4me1, and H3K4me3 at the MIR205HG-SE locus, as well as HiChIP-Seq data of MIR205HG in the TE5 cell line. The purple lines reveal the spatial interaction between the MIR205HG promoter and the identified super-enhancer. (C and D) The enhancer activities were detected using dual-luciferase reporter in KYSE510 (C) and TE7 (D) cells. (E) Expression levels of MIR205HG in KYSE510 and KYSE30 cell lines after treatment with different concentrations of THZ1, JQ1, CPI-637, and Birabresib were determined using RT-qPCR Then, nine individual enhancer constituents (E1→E9) of the MIR205HG-SE and a negative control region (Ctrl) were cloned into the pGL3-promoter luciferase reporter vector and the activities were measured using luciferase reporter assay (Fig. 2A). Consistent with the high H3K27ac signal peaks of ChIP-Seq, significant activity of almost all enhancer constituents was detected in both TE7 and KYSE510 cells (Fig. 2C and D). In contrast, all constituents displayed no activity in the EAC cell line (Figure S2), again highlighting the ESCC-specific characteristics of this SE, which is consistent with our previous finding that MIR205HG was highly and uniquely expressed in ESCC tumor samples. BET bromodomain inhibitors (JQ1, CPI-637 and Birabresib) and the CDK7 inhibitor THZ1 have been shown to effectively inhibit SE-driven gene transcription and demonstrated anti-tumor activity in several cancers [30–33]. In this study, ESCC cells were treated with these four compounds and the expression of MIR205HG was measured. As expected, expression levels of MIR205HG were suppressed in the KYSE510 and KYSE30 cell lines upon treatment with THZ1, JQ1, CPI-637, and Birabresib (Fig. 2E), thus validating the role of the MIR205HG-SE in the regulation of this lncRNA. ESCC master transcriptional regulators TP63 and KLF5 activate the transcription of MIR205HG via the super-enhancer Given the ESCC-specific feature of MIR205HG-SE, we hypothesized that ESCC-specific TFs might be responsible for establishing and maintaining its activity. Therefore, the ChIP-Seq profiles of TP63, SOX2, and KLF5, three master transcriptional factors (mTFs) recently identified by us in ESCC cells, were analyzed to map their genome-wide occupancy [34]. Notably, all these TFs exhibited certain prominent peaks within the MIR205HG-SE region (Fig. 3A), indicating that these mTFs were involved in the SE-mediated regulation of MIR205HG. Several binding motifs of TP63, SOX2, and KLF5 were consistently predicted in the ChIP-Seq peaks surrounding MIR205HG-SE using the JASPAR online method (http://jaspar.genereg.net/) (Figure S3A, Table S10). We then knocked down TP63, SOX2, and KLF5 in ESCC cells, respectively, and then detected MIR205HG expression. The results showed that, upon silencing of TP63, SOX2, or KLF5, MIR205HG expression was downregulated in both KYSE510 and TE7 cells (Fig. 3B and C). Moreover, we found that knocking down any one of the three mTFs consistently and significantly downregulated the expression of the other two mTFs, validating our previous study showing the co-regulation of TP63, SOX2, and KLF5 in ESCC [34]. TP63 and KLF5 regulate the transcription of MIR205HG by co-occupying the MIR205HG-SE region. (A) ChIP-Seq profile of TP63, SOX2, and KLF5 at the MIR205HG-SE across various cell lines. The ChIP-qPCR primers are indicated with short lines of different colors. (B and C) Expression levels of the three mTFs and MIR205HG after silencing of TP63, KLF5, or SOX2 in KYSE510 (B) and TE7 (C) cells. (D and E) ChIP-qPCR was performed using antibodies against TP63, KLF5, or SOX2 in KYSE510 (D) and TE7 (E) cells. IgG was used as a negative control antibody. The positions of qPCR primers are marked in panel (A). (F) CRISPRi-mediated targeting of enhancer constituents in TE1 cells was validated by RT-qPCR. sgRNA localization is shown in panel (A). (G) ChIP-qPCR using an anti-H3K27ac antibody was performed in cells expressing the indicated sgRNAs (sgNC, sgM2, sgM4, sgM7, and sgM8). H3k27ac enrichment was measured at multiple enhancer components of the MIR205HG-SE, including E2 (detected by primers PK1, PS1, and PT1), E4 (primer PK2), E7 (primer P4), and E8 (primers P5 and PK3) To verify the ability of these mTFs to bind to MIR205HG-SE, a ChIP-qPCR assay was performed. As expected, prominent co-occupancy of TP63 and KLF5 with the super-enhancer was observed in both KYSE510 and TE7 cell lines, underscoring the ChIP-Seq profiles (Fig. 3D and E). However, SOX2 did not specifically bind to the corresponding super-enhancer constituents of MIR205HG (E2 and E6). Importantly, silencing the three TFs using siRNAs decreased the luciferase activities of the corresponding enhancer binding constituents (E1, E2, E7 and E8) (Figure S3B). In addition, significant positive correlations were observed between the expression levels of TP63 and MIR205HG, as well as between the expression levels of KLF5 and MIR205HG (Figure S3C and D). But no correlation was observed between MIR205HG and SOX2 expression (Figure S3E). Moreover, the expression levels of TP63, SOX2, and KLF5 were not significantly higher in MIR205HG high-expressing tumors compared to low-expressing tumors (Figure S3F-H), suggesting that the establishment of the MIR205HG-SE is the key determinant of its tissue-specific expression. To further determine whether the interactions between mTFs and SE regions were involved in the transcriptional regulation, we implemented the CRISPR interference (CRISPRi) system to design four independent single guide RNAs (sgRNAs) targeting transcriptionally active enhancer constituents (E2, E4, E7, and E8) that were co-occupied by both TP63 and KLF5. Results showed that these sgRNAs demonstrated potent suppression of MIR205HG expression (Fig. 3F). ChIP-qPCR revealed that targeting individual enhancer components reduced H3K27ac enrichment not only at the targeted sites but also at distal constituents, indicating that the MIR205HG-SE functions as an integrated cooperative unit (Fig. 3G). All these findings suggest that both TP63 and KLF5 could directly bind to MIR205HG-SE and subsequently upregulate MIR205HG transcription and promote its overexpression in ESCC cells. MIR205HG promotes malignant phenotypes in ESCC To investigate the biological significance of MIR205HG in ESCC cells, both loss- and gain-of-function approaches were used. Specifically, siRNAs targeting MIR205HG were applied to ESCC cell lines to inhibit high MIR205HG expression (KYSE510 and KYSE30), whereas an overexpression vector was applied to KYSE150 cells with a low expression of this lncRNA. Importantly, silencing of MIR205HG significantly impaired the growth, colony formation, migration, and invasion abilities of ESCC cells (Fig. 4A-I). However, the ectopic expression of MIR205HG resulted in the converse effects (Fig. 4J-M). Next, we established KYSE510 ESCC cells with stable knockdown of MIR205HG and determined the oncogenic effects of this lncRNA using in vivo xenograft assays. As expected, both xenograft volume and tumor weight were decreased upon MIR205HG depletion (Fig. 4N-R). Collectively, these results identify MIR205HG as a potentially crucial oncogenic lncRNA in ESCC cells. MIR205HG promotes the malignant progression of ESCC. (A) siRNA-mediated knockdown of MIR205HG in KYSE510 and KYSE30 were detected using RT-qPCR. (B-E) MTS assays (B and C) and colony-formation assays (D and E) were used to determine the effect of MIR205HG knockdown on cell growth. (F-I) The change in migration (F and G) and invasion (H and I) abilities of KYSE510 and KYSE30 cells upon MIR205HG knockdown. (J-M) After MIR205HG overexpression (J), the abilities of proliferation, migration, and invasion of KYSE150 were detected using MTS (K), colony formation and Transwell assays (L and M). (N-P) Nude mouse tumor xenograft assay following MIR205HG stable knock down. n = 5 per group. (Q) RT-qPCR was used to measure the expression of MIR205HG in the tumor tissues of nude mice. (R) Representative images of H&E staining of sections from mice-bearing tumors MIR205HG interacts with polypyrimidine tract binding protein 3 (PTBP3) in ESCC cells The functions of lncRNAs are closely related to their subcellular localization [35]. RNA fluorescence in situ hybridization (FISH) showed that MIR205HG was predominantly localized in the cytoplasm, with minimal expression in the nucleus (Fig. 5A). RT-qPCR of the fractionated nuclear and cytoplasmic RNA confirmed its principal presence in the cytoplasm (Fig. 5B and C, Figure S4A). MIR205HG interacts with PTBP3 in the cytoplasm of ESCC cells. (A) RNA FISH was used to localize MIR205HG in KYSE510 and TE7 cells. Green: MIR205HG; Blue: DAPI; Scale bar: 20 μm. (B and C) Subcellular localization of MIR205HG in KYSE510 (B) and TE7 (C) cells were detected using RT-qPCR after RNA subcellular isolation assay. GAPDH and U6 were used as the cytoplasmic and nuclear RNA internal reference, respectively. (D) Silver staining of biotinylated MIR205HG-associated proteins after RNA pull-down. MIR205HG-antisense (MIR205HG-AS) was used as negative control. (E) List of the top ten MIR205HG-bound proteins identified using mass spectrometry. (F) Western blotting analysis of PTBP3 after RNA pull-down to verify the binding of MIR205HG with PTBP3 protein. (G) RIP-qPCR assay was employed to confirm enrichment of MIR205HG with PTBP3 antibody. (H) RIP-qPCR assay on fractionated nuclear and cytoplasmic protein to detect the subcellular location of the interaction between MIR205HG and PTBP3. (I) A schematic diagram showing the probable binding sites between MIR205HG and PTBP3 protein predicted using the catRAPID method. (J) Top: The schema of truncated MIR205HG constructions. Bottom: Immunoblot detection of the PTBP3 protein after RNA pull-down with different truncations of MIR205HG Generally, lncRNAs located in the cytoplasm play vital roles by interacting with specific proteins. To investigate this phenomenon in ESCC cell models, we performed in vitro RNA pull-down of MIR205HG and then subjected the precipitants to mass spectrometry (MS) analysis. Silver staining revealed noticeable specific bands in the MIR205HG pull-down products (Fig. 5D). Among the top 10 proteins identified, several have been reported to have RNA binding capacity, including IGF2BP3 and IGF2BP1 [36], supporting the validity of the MS analysis (Fig. 5E, Table S8-9). We were particularly interested in PTBP3, which showed the highest number of peptide-spectrum matches (PSMs) in the list, because PTBP3 is involved in regulating malignant tumor proliferation and metastasis [37, 38]. Moreover, analysis of the PXD021701 proteome dataset revealed that PTBP3 was significantly upregulated in ESCC tumors and that high PTBP3 expression correlated with poor prognosis (Figure S4B-C), further highlighting its potential relevance in ESCC. The interaction between MIR205HG and PTBP3 was validated using an RNA pull-down assay followed by Western blotting (Fig. 5F). Moreover, qPCR analysis following RNA immunoprecipitation (RIP) assay confirmed the enrichment of MIR205HG with the PTBP3 antibody (Fig. 5G). We also conducted a RIP-qPCR assay on the fractionated nuclear and cytoplasmic proteins and found that the interaction mostly occurred in the cytoplasm (Fig. 5H). To describe the precise mechanisms of the MIR205HG/PTBP3 interaction, the catRAPID method (http://service.tartaglialab.com/page/catrapid_group) was used to predict the specific nucleotide (nt) sequence of MIR205HG that interacts with PTBP3. Three MIR205HG fragments ranking 150–250 nt, 520–620 nt, and 680–750 nt were found to be possibly responsible for this interaction (Fig. 5I, Figure S4D). Then, truncated MIR205HG was constructed and subjected to RNA pull-down assays. The results showed that deleting the RNA fragment ranking from 1 to 246 dramatically reduced the binding ability of MIR205HG to PTBP3, suggesting that this region was critical for their interaction (Fig. 5J). To investigate the effects of PTBP3 on the cancer phenotype of ESCC, functional studies were carried out on PTBP3. Knockdown of PTBP3 significantly inhibited the growth, colony formation, migration, and invasion of ESCC cells (Figure S5A-J), whereas the ectopic expression of PTBP3 resulted in opposite alterations (Figure S5K-O). Moreover, PTBP3 was silenced using siRNA in KYSE150 cells that stably overexpress MIR205HG. Results showed that PTBP3 could antagonize the increased growth, colony formation, migration, and invasion mediated by MIR205HG overexpression (Fig. 6A-I). PTBP3 knockdown MIR205HG phenocopies depletion in ESCC progression. (A and B) The expression of PTBP3 and MIR205HG was detected by Western blotting (A) and RT-qPCR (B) after silencing PTBP3 in KYSE150 cells stably transfected with lentivirus containing MIR205HG overexpression vectors. (C) MTS assay showing the effect of PTBP3 depletion on the growth of MIR205HG-overexpressing KYSE150 cells. (D-I) Colony formation assay (D and E) and Transwell migration (F and G) and invasion assays (H and I) were used to determine the alterations of ESCC malignant phenotypes after PTBP3 knock down in KYSE150 cells stably overexpressing MIR205HG MIR205HG/PTBP3 interaction activates HIF-1α mRNA translation LncRNAs located in the cytoplasm may play important roles in post-transcriptional regulation, including stabilizing protein conformations, mediating posttranslational protein modifications, determining their subcellular localization, and serving as scaffold molecules to facilitate target molecules [39]. Here, we determined whether MIR205HG affects the expression and subcellular localization of PTBP3. Western blotting and immunofluorescence staining revealed that PTBP3 was present in both the cytoplasm and nucleus of ESCC cells (Fig. 7A and B, Figure. S6A-C). Quantitative nuclear-cytoplasmic fractionation showed that PTBP3 exhibited a nuclear-to-cytoplasmic ratio of approximately 60:40 in control cells, which significantly increased to approximately 75:25 upon MIR205HG knockdown, while the total PTBP3 protein level remained unchanged, indicating that MIR205HG specifically influences PTBP3 subcellular distribution. MIR205HG/PTBP3 complex activates HIF-1α mRNA translation and promotes anaerobic glycolysis in ESCC cells. (A and B) Western blotting (A) and densitometric analysis (B) showing changes in PTBP3 subcellular distribution after MIR205HG knockdown in KYSE30 cells. (C) CLIP-qPCR assay was performed to detect the direct interaction between PTBP3 and MIR205HG or the HIF-1α 5′UTR. (D and E) RIP-qPCR assay was used to detect changes in the interaction between PTBP3 and the HIF-1α 5′UTR following MIR205HG knockdown. (F) Polysome profiling reveals HIF-1α translation upon MIR205HG knockdown. (G) The polysomal fractions were quantified upon MIR205HG knockdown. (H) Translational activity of the HIF-1α 5′UTR was detected using a dual-luciferase reporter assay. (I-N) Glucose consumption (I, K, M) and lactate production capacity (J, L, N) were detected in the indicated ESCC cells. (O and P) The effect of MIR205HG or PTBP3 knockdown on real-time ECAR in KYSE30 cells was measured using a Seahorse XF24 analyzer. 2-DG, 2-deoxyglucose Previous studies have reported that PTBP3 enhances the growth and metastasis of colorectal cancer by directly binding to the 5′UTR of HIF-1α mRNA and thereby facilitating HIF-1α translation [40]. To investigate whether MIR205HG is involved in this process in ESCC, we first predicted the potential interactions among MIR205HG, PTBP3, and the HIF-1α 5′UTR. Secondary structure prediction using RNAfold (http://rna.tbi.univie.ac.at/) revealed that the HIF-1α 5′UTR contains a pyrimidine-rich UCUU motif, the consensus binding sequence of PTBP3, but this motif is located within a stem region, suggesting limited accessibility for direct PTBP3 binding (Figure S6D). In contrast, the PTBP3-binding region of MIR205HG (1-246 nt) was predicted to adopt a structure with the UCUU motif positioned in a single-stranded loop, favoring protein recognition (Figure S6E). Consistent with these predictions, CLIP-qPCR demonstrated that PTBP3 directly bound MIR205HG but not the HIF-1α 5′UTR (Fig. 7C). We further performed RIP-qPCR assay using a PTBP3 antibody and confirmed enrichment of the HIF-1α 5′UTR in the PTBP3 complex (Fig. 7D), indicating that PTBP3 associates with HIF-1α 5′UTR despite lacking direct binding. Surprisingly, after silencing MIR205HG, this interaction was attenuated (Fig. 7E). To explore whether MIR205HG could directly interact with the HIF-1α 5′UTR, we performed RNA-RNA interaction prediction using IntaRNA (http://rna.informatik.uni-freiburg.de/IntaRNA/Input.jsp), which revealed potential base pairing between MIR205HG and the HIF-1α 5′UTR (Figure S6F). These results indicated that MIR205HG may serve as a molecular scaffold in linking the binding of PTBP3 to HIF-1α mRNA. Furthermore, RT-qPCR and Western blotting showed that MIR205HG did not affect HIF-1α mRNA levels but significantly regulated its protein expression (Figure S6H and I). To further validate the participation of MIR205HG in the PTBP3-enhanced translation efficiency of HIF-1α mRNA, polysome profiling was employed and confirmed that MIR205HG knockdown significantly reduced the association of PTBP3 with polysomal fractions (Figure S6J), decreased HIF-1α mRNA loading onto polysomes, and diminished overall polysome fractions (Fig. 7F and G). Moreover, dual-luciferase reporter assays further confirmed that both PTBP3 and MIR205HG enhance the translational activity of the 5′UTR (Fig. 7H). Collectively, these findings support a model in which MIR205HG serves as a molecular scaffold that directly engages PTBP3, enabling its functional interaction with the HIF-1α 5′UTR and promoting its translation. MIR205HG promotes ESCC glycolysis via HIF-1α signaling pathway Accumulating studies have suggested that stabilization of HIF-1α, independent of oxygen concentration, leads to the inactivation of tumor-suppressive functions and the activation of carcinogenesis, thereby allowing the upregulation of glycolytic enzyme transcription [41, 42]. Therefore, we next explored the effect of the MIR205HG/PTBP3 complex on the signaling pathway downstream of HIF-1α. First, gene set enrichment analysis (GSEA) of TCGA-ESCC RNA-seq data revealed that significant enrichment of the glycolysis pathway in genes overexpressed in tumors compared to normal epithelia (NES = 1.4614, P = 0.0018) (Figure S7G). Then, the correlation between PTBP3 or MIR205HG expression and glycolytic enzymes was determined using multiple datasets of ESCC samples. Results suggested that the expression of MIR205HG or PTBP3 was positively correlated with the expression of certain key glycolytic enzymes, such as HK2 and LDHA, at both protein levels (PXD021701) [43] (Figure S7A) and mRNA levels (Figure S7B). Furthermore, correlation analysis using the same cDNA microarray (HEsoS095Su01) confirmed that MIR205HG and PTBP3 expression levels were significantly positively correlated with HK2 and LDHA expression in ESCC tissues (Figure S7C-F), suggesting a potential impact of MIR205HG/PTBP3 axis on glycolysis. To validate this hypothesis, glucose consumption was measured in cell lysates and lactate production in cell supernatants. The results revealed that both glucose uptake and lactate production were reduced in ESCC cells depleted of MIR205HG or PTBP3 (Fig. 7I-L). Forced expression of MIR205HG dramatically increased glucose uptake and lactate production and these effects were reversed by PTBP3 silencing (Fig. 7M and N), indicating that MIR205HG promotes glycolysis via PTBP3. Furthermore, ECAR assays revealed that MIR205HG or PTBP3 silencing significantly decreased ECAR levels in KYSE510 and KYSE30 cells, particularly glycolytic capacity and glycolytic reserve (Fig. 7O and P, Figure S7H and I). Taken together, these results suggest that the MIR205HG/PTBP3 complex promotes glycolysis in ESCC cells. Western blotting was performed to identify the key glycolytic enzymes involved in MIR205HG-mediated glycolysis. Consistent with the positive relationship observed in ESCC samples, the expression of HK2 and LDHA was markedly reduced in KYSE510 and KYSE30 cells following MIR205HG or PTBP3 depletion (Figure S8A and B). To further elucidate this phenomenon, ESCC cells were treated with 100 µM CoCl2 to mimic a hypoxic condition in vitro by inducing HIF-1α protein expression. Results showed that, upon MIR205HG knockdown, the expression levels of HIF-1α, HK2, and LDHA were all significantly decreased under both normoxic and hypoxic conditions (Fig. 8A and B, Figure S8C and D). In contrast, MIR205HG overexpression had the opposite effect (Figure S8G). Similar results were observed in the ESCC cells with PTBP3 silencing or overexpression (Fig. 8C and D, Figure S8E, F, and H). A rescue experiment demonstrated that decreasing PTBP3 effectively restored the upregulation of HIF-1α, HK2, and LDHA proteins caused by stable MIR205HG overexpression, which was more pronounced under hypoxic conditions (Fig. 8E and F). To further establish the epistatic relationship between MIR205HG and HIF-1α, HIF-1α knockdown in MIR205HG-overexpressing KYSE150 cells significantly reversed the upregulation of HK2 and LDHA, cell proliferation, and lactate production (Figure S8K-M), confirming that HIF-1α acts as a critical downstream mediator of the MIR205HG/PTBP3 axis. Moreover, RT-qPCR analysis of the xenograft tumor tissues showed that the expression levels of HK2 and LDHA were downregulated in the MIR205HG knockdown groups (Figure S8I and J). Truncation mutants of MIR205HG (1-550 nt and 247–895 nt) partially restored HIF-1α/HK2/LDHA expression, cell proliferation, and lactate production in KYSE410 cells, albeit to a lesser extent than wild-type MIR205HG (Figure S8N-P), These results suggest that while the 1-246 nt region is critical for full oncogenic function, additional domains and mechanisms, such as potential self-interaction of MIR205HG (Figure S6G) or direct binding to the HIF-1α 5′UTR, also contribute to the regulation of tumor progression. MIR205HG/PTBP3 complex increases the expression of HK2 and LDHA via HIF-1α in ESCC cells. (A-D) Expression levels of HIF-1α protein, HK2 and LDHA were assessed in KYSE510 cells following knockdown of either MIR205HG (A and B) or PTBP3 (C and D) under both normoxic and hypoxic conditions (CoCl2 treatment). (E and F) Knockdown of PTBP3 partially abolished the upregulation of HK2 and LDHA mediated by MIR205HG overexpression in KYSE150 cells. E, Western blotting; F, RT-qPCR. (G and H) The growth rates of KYSE510 cells were detected upon HK2 (G, 2-DG and DICA) and LDHA (H, FX-11) inhibitor treatment with or without MIR205HG silencing. (I and J) Colony formation assay was used in ESCC cells upon combined treatment of MIR205HG siRNA and glycolysis inhibitors. (K) Kaplan-Meier analysis displaying the correlation between the expression of PTBP3 and overall survival (OS) in ESCC tissue samples from the GSE53625 dataset (n = 179). (L) In patients with high MIR205HG expression in GSE53625 dataset, the association between the expression of PTBP3 and the OS of patients with ESCC was explored Additionally, to explore the effect of anaerobic glycolysis on the ESCC cell growth, we performed MTS and colony formation assays in ESCC cells treated with HK2 inhibitors (2-DG and DICA, which have already been tested in Phase 2 and 3 clinical trials, respectively), as well as an LDHA inhibitor (FX-11). These 3 inhibitors all effectively suppressed ESCC cell growth in a concentration-dependent manner (Figure S9A-F). Furthermore, combined treatment with MIR205HG siRNA and glycolysis inhibitors led to slower cell proliferation and colony formation than treatment with either single approach (Fig. 8G-J, Figure S9G-J). These findings imply that the MIR205HG/PTBP3 cascade could be a promising therapeutic target for patients with ESCC. MIR205HG axis could serve as prognostic biomarker and gene therapy target in ESCC To explore the clinical significance of the MIR205HG/PTBP3 signaling cascade, we analyzed the clinical relevance of MIR205HG and PTBP3 expression in patients with ESCC (GSE53625). Kaplan-Meier analysis revealed a significant association between higher PTBP3 expression and poorer overall survival (Fig. 8K). Further subtype analysis revealed that higher PTBP3 expression was markedly correlated with poorer survival in patients with ESCC with high MIR205HG expression (P = 0.0032) (Fig. 8L). However, no significant correlation was found between the expression of PTBP3 and the survival time of patients with ESCC with low MIR205HG expression (P = 0.7439) (Figure S9L). These findings suggest that ESCC patients with high expression of both MIR205HG and PTBP3 have worse prognosis. Furthermore, survival and pseudotime analyses were performed on single-cell RNA sequencing (scRNA-Seq) data obtained from seven ESCC samples (GSE188900) to further explore the potential clinical applications of MIR205HG. Cell clusters were identified by using principal component analysis (PCA) and the tSNE algorithm implemented in Seurat software (Figure S10A). Each cluster was then assigned a cell-type annotation based on the expression patterns of well-known marker genes (CellMarker 2.0 database, http://117.50.127.228/CellMarker/) and nine major cell types were obtained (Figure S10B-C). Subsequently, SCISSOR was used to distinguish subpopulations related to worse survival. SCISSOR is an algorithm that integrates bulk RNA data and corresponding phenotype information to automatically identify and map cell subpopulations most relevant to specific phenotypes (e.g. disease stage or survival outcomes) [27]. The Scissor+ subset was associated with worse survival, and the subset with no correlation was designated as background cells (0). Importantly, the Scissor+ subset predominantly comprised epithelial cells (98.75%) (Figure S10D and E). The expression pattern of certain genes involved in SE-driven MIR205HG signaling was then analyzed across these two clusters. A significantly higher percentage of all these genes was observed within the Scissor+ subtype, whereas background cells exhibited scarce expression, suggesting that the MIR205HG signaling axis might be activated in the cell subpopulation with poorer survival (Figure S10F). Additionally, given our previous observation that most cells associated with poorer prognosis were primarily clustered within epithelial cells, we further refined the classification of this epithelial population (Figure S10G). Based on the expression levels of esophageal cancer stem cell markers (SOX9 and CD44), clusters 0, 2, 3, and 4 were designated as epithelial stem cells (or malignant epithelial cells), whereas the remaining clusters were characterized as typical epithelial cells (Figure S10H and I). Further analysis using pseudotime modeling revealed a transitional trajectory from typical epithelial cells towards malignant epithelial cells over time (Figure S10J-L). Similarly, Scissor+ cells represented malignant epithelial stem cells (Figure S10M). Interestingly, the expression pattern of MIR205HG resembled that of SCC stem cell marker genes from CellMarker database, particularly CD44, KLF4 [44], SLC3A2, and SOX2 [45], implying an association between MIR205HG and both cellular stemness and malignancy (Figure S10N). Taken together, the comprehensive scRNA-Seq analysis indicated that MIR205HG could potentially serve as an effective biomarker for predicting malignancy and prognosis in patients with ESCC. Finally, to investigate the therapeutic potential of the transcriptional reprogramming, we performed functional assays to evaluate the impact of MIR205HG-SE inhibition on ESCC malignancy. Strikingly, CRISPRi-mediated silencing of the MIR205HG-SE (using sgRNAs sgM7 and sgM8) significantly attenuated malignant phenotypes in TE1 cells. This was as evidenced by reduced proliferation rates (Fig. 9A), diminished colony formation capacity (Fig. 9B and C), markedly impaired migration and invasion capabilities (Fig. 9D and E), as well as suppressed impaired glucose uptake and lactate production (Fig. 9F). Moreover, combined silencing of the MIR205HG-SE and pharmacological inhibition of HK2/LDHA effectively suppressed ESCC cell growth (Figure S9K). These findings strongly suggest that targeting of the MIR205HG-SE is a promising therapeutic strategy for ESCC. Inhibition of the MIR205HG-SE demonstrates potent suppressive effects on ESCC tumorigenesis. (A) Effect of CRISPRi-mediated MIR205HG-SE silencing on TE1 cell proliferation assessed by CCK-8 assay. (B-C) Colony formation assay (B) and quantitative analysis (C) of TE1 cell clonogenicity. (D-E) Transwell migration and invasion assays were used to determine metastatic potential in TE1 cells. (F) Both glucose uptake and lactate acid production capacity were detected in TE1 cells. (G) Mechanism diagram of this study showing that SE-driven MIR205HG acts as a molecular scaffold, bridging PTBP3 to the HIF-1α 5′UTR, thereby promoting HIF-1α translation, and subsequently enhancing glycolysis and malignant progression of ESCC cells Discussion Over the past decade, studies have identified and verified lncRNAs driven by super-enhancers with tissue-specific expression patterns in cancer [46–48]. In our previous studies, we identified two SE-driven lncRNAs (CCAT1 and LINC01503) that are overexpressed in ESCC [17, 49]. Here, we demonstrate that MIR205HG, a novel SE-driven lncRNA, is uniquely highly expressed in SCC tissues and cells. Moreover, higher expression of MIR205HG significantly correlated with the malignancy and poorer survival, suggesting its potential as an early prognostic biomarker in patients with ESCC. Additionally, the mTFs TP63 and KLF5 directly bind to the SE region of MIR205HG, synergistically activating its transcription. Mechanistically, MIR205HG interacts with PTBP3 and acts as a scaffold that bridges PTBP3 to HIF-1α, thereby promoting the translation of HIF-1α and accelerating the progression of ESCC (Fig. 9G). Traditionally, MIR205HG serves as a precursor for miR-205 biogenesis through Drosha-dependent processing [50]. However, emerging evidence reveals that the majority of MIR205HG isoforms lack the intron sequence context required for miR-205 maturation [51]. Therefore, MIR205HG can functionally decouple from miR-205 and operate autonomously as a critical lncRNA, underscoring the necessity to investigate its miRNA-independent mechanisms in cancer. Emerging evidence suggests that MIR205HG acts as a tumor booster in various types of carcinomas, particularly in cervical [52, 53] and breast cancer [54]. It exerts its oncogenic effects mechanistically through distinct molecular pathways, including acting as a competing endogenous RNA (ceRNA) and directly interacting with key protein partners. These multifunctional characteristics significantly expand our understanding of the multifaceted roles of lncRNAs in cancers. Intriguingly, in esophageal cancer, recent studies have further revealed that MIR205HG exhibits histological subtype-specific expression patterns. It is downregulated and exerts suppressive functions in ECA [55, 56], while it is upregulated and exhibits oncogenic functions in ESCC [57]. Given the cell-context-dependent biological properties of MIR205HG, it is necessary to analyze its expression pattern and functional role of MIR205HG according to the histological subtype. Herein, we showed that MIR205HG is uniquely highly expressed in SCC, including ESCC. Through the analysis of clinical specimen and multi-omics datasets, we verified its specific upregulation in ESCC tissues when compared with the matched adjacent normal tissues. Additionally, functional assays validated that MIR205HG promoted the proliferation, migration, and invasion of ESCC cells in vitro and in vivo, establishing MIR205HG as a functionally oncogenic lncRNA. Another point that should be highlighted is that by performing survival and pseudotime analyses on ESCC scRNA-Seq data, we identified a significantly positive association between MIR205HG and the stemness and malignancy of ESCC cells, again underscoring the oncogenic potential of this lncRNA. SEs establish tissue/cell-specific transcriptional programs by recruiting lineage-determining TFs, particularly those involved in core regulatory circuits, as well as chromatin modifiers to activate key genes of cellular identity [12]. In ESCC, TP63, SOX2, and KLF5 have been identified as TFs that cooperatively regulate oncogenic genes by binding to super-enhancers [34]. Our study revealed that TP63 and KLF5 directly activate MIR205HG transcription by co-occupying the MIR205HG-SE region, while SOX2 exerts its effect indirectly by regulating TP63 and KLF5 through the CRC, suggesting that MIR205HG is a key downstream target of the mTFs-mediated transcriptional network. Gene regulation is controlled at different levels by cytoplasmic lncRNA complexes, including mRNA turnover, translation, protein stability, sponging of cytosolic factors, and signaling pathways modulation. Here, we found that MIR205HG interacted with PTBP3 and subsequently affected the translation of HIF-1α. Polypyrimidine tract binding proteins (PTBs) are well-characterized RNA-binding proteins that recognize pyrimidine-rich motifs (UCUU/CUU) to directly interact with target mRNAs, acting as multifunctional post-transcriptional regulators of RNA splicing, stability, and translation, with significant functional overlap among PTBP1, PTBP2, and PTBP3 [58–60]. Accumulating evidence has shown that PTBP3 is highly expressed in a wide variety of tumors, leading to the enhancement of cell growth, epithelial mesenchymal transition (EMT), and metastasis of cancer cells [37, 60, 61]. Previous studies showed that PTBP3 enhanced HIF-1α protein expression by directly binding to the 5′UTR of HIF-1α mRNA and activating its translation [40]. Notably, while PTBP3 directly binds the HIF-1α 5′UTR in colorectal cancer, our CLIP-qPCR analysis revealed that this direct interaction does not occur in ESCC cells, highlighting the cell-type-specific nature of PTBP3-mediated regulation. Instead, we found that MIR205HG directly binds PTBP3 and serves as a molecular scaffold to facilitate PTBP3-mediated translational activation of HIF-1α. Additionally, this direct interaction may influence PTBP3 subcellular distribution by masking its nuclear localization signal or promoting nuclear export. Together, these findings reveal that the MIR205HG/PTBP3 complex plays an important role in ESCC progression. Under hypoxic conditions, the oxygen-dependent prolyl hydroxylase domain (PHD)-mediated ubiquitination degradation of HIF-1α is impaired, enabling its stabilization and dimerization with HIF-1β [62]. This heterodimer then translocates into the nucleus and binds to hypoxia response elements (HREs) of the enhancer and/or promoter region to activate the transcription of numerous genes involved in hypoxia signaling, such as growth factors and glycolytic enzymes (e.g., HK2, LDHA), which are essential for tumor proliferation and metabolism [63, 64]. Our study suggests that HK2 and LDHA are upregulated in both ESCC cells and xenograft tumor tissues due to activation of HIF-1α signaling by the MIR205HG/PTBP3 complex, particularly under hypoxic conditions. This further elucidates the prevalence of the Warburg effect in tumor cells and provides a new molecular basis for this phenomenon in cancer. Notably, treating cells with HK2/LDHA inhibitors and/or using CRISPRi to silence the MIR205HG-SE both potently suppress ESCC cell growth, validating the crucial role of the MIR205HG/PTBP3 signaling cascade as a therapeutic vulnerability. However, several aspects of this study warrant further investigation. First, direct perturbation of loop anchors is needed to establish the functional requirement of chromatin contacts for MIR205HG transcription. Second, both the predicted intramolecular self-interaction of MIR205HG and its potential base pairing with the HIF-1α 5′UTR require experimental validation via RNA pull-down or mutational disruption in reporter assays. Third, the precise PTBP3-interacting domain of MIR205HG should be mapped using RIP assays with PTBP3 truncation constructs, which may enable molecular docking to identify small-molecule inhibitors. Addressing these questions will provide deeper mechanistic insights and facilitate the development of therapeutic strategies targeting the MIR205HG-PTBP3-HIF-1α axis in ESCC. Conclusions In conclusion, our findings revealed an mTFs-dependent, super-enhancer-driven mechanism underlying the overexpression of an oncogenic lncRNA, MIR205HG, in ESCC cells. MIR205HG directly binds to PTBP3 and acts as a molecular scaffold to promote HIF-1α translation and enhance glycolysis in ESCC cells, thereby driving unfavorable progression. Given its significant correlation with the survival and the aggressive stage of patients with ESCC, MIR205HG may serve as a potential biomarker and/or therapeutic target in this cancer. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements Not applicable. Abbreviations Esophageal squamous cell carcinoma Long non-coding RNA Squamous cell carcinoma Chromatin immunoprecipitation sequencing Esophageal adenocarcinoma Transcription factor Core regulatory circuitry Master transcription factor Polypyrimidine tract binding protein 3 RNA-binding protein Hexokinase Hypoxia response element Prolyl hydroxylase domain CRISPR interference Author contributions ZJ.Z., YQ.L., YS.L. and DC.H. contributed equally to the conceptualization and design of this study. ZJ.Z. wrote the original draft and revised the manuscript. YQ.L. revised the manuscript. YS.L. performed mass spectrometry and bioinformatics analysis. DC.H. and JC.G. performed bioinformatics analysis of scRNA-Seq data. JD.Z. performed gene cloning and statistical analysis. C.L and ZY.K. assisted with the xenograft mice assay and immunohistochemical staining. QX.S., ZY.W., YH.X., LS.M., JR.Y., and JY.P. helped perform the experiments. XD. X. and JJ.X. provided overall direction. JJ.X. supervised the research and revised the manuscript. All authors read and approved the final manuscript. Funding The study was supported by the National Natural Science Foundation of China (81871921), the Natural Science Foundation of Guangdong Province-Outstanding Youth Project (2019B151502059), and the Guangdong Basic and Applied Basic Research Foundation (2023A1515111094). Data availability The ESCC RNA expression profiles and detailed clinical information were downloaded from the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/gds/?term = GSE) database including GSE53624, GSE53622, and GSE53625 datasets (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA232519). The esophageal cancer lncRNA expression data of TCGA were obtained from MiTranscriptome (http://www.mitranscriptome.com/). The gene expression of kinds of cancer cell lines were downloaded from Cancer Cell Line Encyclopedia (CCLE, https://sites.broadinstitute.org/ccle/). The PXD021701 proteome datasets used by proteomic correlation heat map analysis were obtained from PRIDE database (https://www.ebi.ac.uk/pride/archive/projects/PXD021701). H3K27ac-ChIP-Seq data of six ESCC cell lines had been deposited to GSE106433 before (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA416868). EAC cell lines’ H3K27ac-ChIP data were obtained from GSE132686 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA548754). The ChIP-Seq data of TP63, KLF5 and SOX2 about ESCC cell lines were downloaded from GSE167367 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA704397) and GSE106565 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA417249). Published ChIP-Seq datasets in ESCC tissues were downloaded from the GEO dataset: PRJNA665151 (https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA665151). Single-cell sequencing data of ESCC was obtained from GSE188900 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA780580), GSE160269 (https://ngdc.cncb.ac.cn/gsa-human/browse/HRA000195), and GSE197677 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA811547). Hi-ChIP and other additional raw data and code are available from the corresponding authors upon reasonable request. Declarations Ethical approval The human study was conducted in accordance with the principles of the Declaration of Helsinki. All animal procedures were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Animal Ethics Committee of the Medical College, Shantou University (Approval Number: SUMC2020-366). Consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no potential conflicts of interest. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Ze-Jun Zheng, Yin-Qiao Liu, Yan-Shang Li and Dong-Chen Han contributed equally to this work. Contributor Information Xing-Dong Xiong, Email: xiongxingdong@126.com. Jian-Jun Xie, Email: xiejj0816@foxmail.com. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The ESCC RNA expression profiles and detailed clinical information were downloaded from the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/gds/?term = GSE) database including GSE53624, GSE53622, and GSE53625 datasets (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA232519). The esophageal cancer lncRNA expression data of TCGA were obtained from MiTranscriptome (http://www.mitranscriptome.com/). The gene expression of kinds of cancer cell lines were downloaded from Cancer Cell Line Encyclopedia (CCLE, https://sites.broadinstitute.org/ccle/). The PXD021701 proteome datasets used by proteomic correlation heat map analysis were obtained from PRIDE database (https://www.ebi.ac.uk/pride/archive/projects/PXD021701). H3K27ac-ChIP-Seq data of six ESCC cell lines had been deposited to GSE106433 before (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA416868). EAC cell lines’ H3K27ac-ChIP data were obtained from GSE132686 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA548754). The ChIP-Seq data of TP63, KLF5 and SOX2 about ESCC cell lines were downloaded from GSE167367 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA704397) and GSE106565 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA417249). Published ChIP-Seq datasets in ESCC tissues were downloaded from the GEO dataset: PRJNA665151 (https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA665151). Single-cell sequencing data of ESCC was obtained from GSE188900 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA780580), GSE160269 (https://ngdc.cncb.ac.cn/gsa-human/browse/HRA000195), and GSE197677 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA811547). Hi-ChIP and other additional raw data and code are available from the corresponding authors upon reasonable request.