RNF114-PACSIN3 signaling axis promotes hepatocellular carcinoma progression by enhancing GLUT1 membrane retention and glucose uptake

Abstract

Purpose

Hepatocellular carcinoma (HCC) is a leading cause of cancer-related death worldwide, with current therapies often limited by significant drug resistance. Owing to the Warburg effect, targeting cancer-specific metabolic vulnerabilities is a promising therapeutic strategy. This study aims to investigate the role of RNF114 in HCC progression and its regulatory mechanism, as well as its clinical translational potential as a therapeutic target.

Methods

We evaluated the clinical significance of RNF114 using tissue microarrays and database analysis. RNF114 function in promoting HCC progression by regulating glucose uptake was investigated using knockdown experiments in cell lines and subcutaneous xenograft models. Furthermore, a therapeutic xenograft model was employed to assess the potential of RNF114 knockdown in overcoming Sorafenib resistance.

Results

RNF114 was highly expressed in HCC and correlated with poor prognosis. Knockdown of RNF114 significantly suppressed HCC cell proliferation, migration, invasion, and glycolysis. Co-immunoprecipitation identified PACSIN3 as a key substrate of RNF114. RNF114 interacted with the SH3 domain of PACSIN3, promoting its ubiquitination and proteasomal degradation. Subcellular fractionation revealed that the F-BAR domain of PACSIN3 facilitated GLUT1 vesicular trafficking. Consequently, RNF114 impaired this process, leading to increased plasma membrane retention of GLUT1 and enhanced glycolytic flux. Consistently, in both HCC cells and subcutaneous xenograft models, RNF114 knockdown sensitized tumors to Sorafenib treatment.

Conclusions

Collectively, our findings reveal that the RNF114–PACSIN3–GLUT1 axis regulates glucose uptake and metabolic reprogramming in HCC, thereby promoting tumor progression and contributing to therapy resistance. Targeting this signaling axis provides a novel insight into metabolic therapy for HCC.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s13402-026-01259-6.

Introduction

Liver cancer ranks among the top five causes of cancer-associated death worldwide. Notably, over half of global cases and deaths occur in East Asia, with China alone accounting for 45.3% of cases and 47.1% of deaths [1]. Hepatocellular carcinoma (HCC) is the most common histological subtype. The poor prognosis and treatment difficulties in HCC are attributed to factors including late diagnosis, rapid progression, and resistance to most therapies. Enhanced glycolysis, a hallmark of the Warburg effect, is a key driver of these malignant features [2–4].

Multiple oncogenic pathways, including FTO-IT1/FTO, BMP4/SMAD, and SRSF10/MYB, converge to regulate the expression and post-translational modifications of the glucose transporter GLUT1, a pivotal player in HCC glycolysis, thereby enhancing glycolytic flux and lactate production [5–7]. This lactate-rich microenvironment drives immunosuppressive histone lactylation and IGF2BP3 lysine lactylation, collectively promoting tumor progression and conferring lenvatinib resistance [8]. As a key molecular target in the glycolytic pathway, the membrane-localized GLUT1—a primary glucose transporter—must be tightly regulated through various post-translational modifications to exert its metabolic function [9, 10]. However, whether GLUT1 activity is regulated by alternative mechanisms beyond these established pathways remains to be elucidated.

RNF114, also known as ZNF313, a member of the E3 ubiquitin ligase family, is an early identified zinc-binding protein with C2H2 and RING-finger structures. It is highly expressed in normal testes but reduced in azoospermic testes and other tissues, suggesting a role in spermatogenesis and male fertility. Initially identified as a novel psoriasis susceptibility gene, RNF114 has been implicated in immune-related disorders and plays a central role in regulating cell cycle progression, differentiation, and senescence [11, 12]. Located at 20q13.13, a genomic region frequently amplified in multiple cancers, RNF114 promotes malignant tumor proliferation, metastasis, and autophagy [13–17]. While a recent study revealed that circFADS1 recruits RNF114 to promote the ubiquitination-mediated degradation of GSK3β and drive HCC progression, its role in HCC metabolic reprogramming remains unexplored [18].

In this study, we identify RNF114 as a critical E3 ubiquitin ligase that targets the SH3 domain of PACSIN3 for ubiquitination and promotes its proteasomal degradation. The loss of PACSIN3 stability disrupts its role in mediating membrane invagination, thereby enhancing GLUT1 retention at the plasma membrane. Functional assays both in vitro and in vivo demonstrate that the RNF114–PACSIN3–GLUT1 axis promotes HCC progression by increasing glucose uptake, revealing a previously unrecognized link between regulation of membrane endocytosis and metabolic reprogramming in liver cancer.

Materials and methods

Analysis of TCGA, CPTAC and GEO databases

Data from the TCGA and CPTAC databases were analyzed using UALCAN and GEPIA2. The mRNA expression data, derived from the TCGA database and analyzed via the UALCAN platform, are presented as log2(TPM + 1) values to normalize the distribution and facilitate visualization of transcriptomic profiles. For the CPTAC data, Z-values represented standard deviations from the median across samples for a given cancer type. The log2 spectral count ratios from CPTAC were first normalized within each sample profile and then normalized across samples. Data obtained from the GEO database were processed and analyzed in RStudio. We identified upregulated differentially expressed genes (DEGs) with an adjusted P-value < 0.05 and |log2FC| > 0.3. Survival analysis was conducted based on optimal cut-off value.

Tissue microarray (TMA) analysis and immunohistochemistry (IHC)

A formalin-fixed tissue microarray containing 90 patients (paired tumor and adjacent-tissue) was purchased from Shanghai Outdo Biotech (Shanghai, China) and subsequently sent to Weiao Biotech (Shanghai, China) for immunohistochemical analysis. The chip was deparaffinized, rehydrated and blocked for endogenous peroxidase with H₂O₂ (3%), followed by sequential primary and secondary antibody incubation. The stained section was digitally analyzed using Visiopharm’s AI-powered pathology platform as follows: All nuclei were identified for total cell counting. The HDAB-DAB optical filter was applied to segment regions of interest (ROI) based on staining intensity, with corresponding labels assigned to segmented areas: negative (0), weak (1+), moderate (2+), and strong (3+). Final staining results were statistically analyzed using the H-score system, which is a method of converting the number of positive cells and their staining intensity in a stained section into corresponding values for semi-quantification. Histochemistry scores (H-scores) = Σ(pi × i) = (percentage of weak intensity×1)+(percentage of moderate intensity×2)+(percentage of strong intensity×3), pi = percentage of positive nuclei, i = intensity level. Final H-scores were divided into low (0-60.5) or high (60.5–300) expression groups using X-tile software. This study utilized a commercial tissue microarray (TMA) (HLivH180Su09) from Shanghai Outdo Biotech, with the use of human tissues approved by its Ethics Committee (approval number: YB M-05-01).

Cell lines and culture conditions

Human HCC cell line HuH-1 was purchased from Meisen CTCC (Zhejiang, China). Cell lines of HCCLM3, MHCC97-H were obtained from Hycyte biotechnology (Jiangsu, China). All cell lines were maintained in DMEM (#SH30243. FS, Hyclone, USA), supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin (100 U/ml penicillin, 100 µg/ml streptomycin), cultured at 37 °C in 5% CO₂. Glucose deprivation was performed using D-glucose-free DMEM medium (#L160KJ, BasalMedia, China).

Reagents and antibodies

MG132 (#HY-13259), Sorafenib (#HY-10201), Cycloheximide (CHX, #HY-12320) and sodium pyruvate (#HY-W015913) were obtained from MedChemExpress (New Jersey, USA). Tunicamycin (#M4798) and dynasore (#M2569) were purchased from Abmole Bioscience (Houston, USA). Chloroquine (#S6999) and WZB117 (#S7927) were purchased from Selleck Chemicals (Houston, USA). Antibodies used in this study and their sources were listed in supplementary Table S1.

Cell model construction and plasmids

Two specific siRNAs (siRNF114-1: 5′-CCAUGGCUGCCGUAAGAAUUUtt-3′; siRNF114-2: 5′-GCUUAGAGGUGUACGAGAAtt-3′) targeting RNF114 were designed and constructed. To generate stable RNF114-knockdown HCC cells, RNF114-specific shRNA was synthesized based on the target sequence of siRNF114-1 and cloned into vector psi-LVRU6MP (GeneCopoeia, USA). Lentiviral particles were generated to infect HuH-1, HCCLM3 and MHCC97-H cells, followed by puromycin selection (#P8230, Solarbio, China) to establish stable RNF114-knockdown cell lines. The siRNAs used in this study were synthesized by Hzykang (Zhejiang, China), and the plasmids by Paiwei Biotechnology (Wuhan, China). The target sequences are as follows: Human PACSIN3 siRNA (siPACSIN3): 5′-GGUUGAGGCUUCCAAGAAA-3′; Human GLUT1 siRNA (siGLUT1): 5′-GGGCCAAGAGUGUGCUrAAA-3′. Human RNF114 and PACSIN3 overexpression plasmids were constructed by cloning the full-length RNF114, PACSIN3 gene into the pcDNA3.1 vector.

Cell proliferation, migration, invasion, colony formation and IC50 assay

Stable knockdown cells were plated directly, whereas transient overexpression cells were plated 24 h post-transfection. For proliferation assays, cells were seeded in 96-well plates and analyzed with CCK-8 reagent (#CW9101M, Cowin, China) at 0, 12, 24, and 48 h according to manufacturer’s protocol. Cells were seeded in Transwell chambers (#14341, Labselect, China) with matrigel (60 µl, 1 mg/ml; #356234, Corning, USA) pre-coated for 4 h in invasion assays (migration without matrigel). The upper chamber contained DMEM with 20% FBS, while the lower chamber contained serum-free medium. After 48 h incubation, chambers were fixed and stained with 0.1% crystal violet (#BL802A, Biosharp, China). For colony formation assay, cells were cultured for 14 days, fixed with 4% paraformaldehyde (#BL539A, Biosharp, China), and stained with crystal violet prior to colony counting. After 24 h of plating, cells were subjected to various treatments. Cell viability was assessed using the CCK-8 assay, and the IC50 value was calculated by nonlinear regression fitting.

Cell surface biotinylation assay

All procedures were performed at 4 °C unless otherwise stated. The cells were first labeled with Sulfo-NHS-SS-Biotin (#A2288815, Ambeed, China) for 30 min at 4 °C. The labeling reaction was then stopped with 50 mM glycine, and the unbound biotin was removed by washing. Pre-warmed complete medium was then added and the cells were transferred to a 37 °C incubator for internalization. At the indicated time points, cells were immediately placed on ice to terminate internalization. The cells were then incubated with a reducing stripping buffer (#HY-13679, 50 mM MESNA, pH 8.0) at 4 °C to remove surface-bound biotin. After stripping, cells were lysed, and the protein extracts were incubated with streptavidin magnetic beads (#P2153, Beyotime, China) to enrich biotin-labeled internalized membrane proteins for pull-down assays.

Co-immunoprecipitation, membrane-cytosol fractionation and western blotting

Cells were lysed in universal protein extraction buffer (#PP1801, Bioteke, China) supplemented with protease and phosphatase inhibitors (#HY-K0010, #HY-K0021, MCE, USA). Lysates underwent immunoprecipitation (IP) with specific antibodies coupled to Protein A/G magnetic beads (#HY-K0202, MCE, USA). Membrane and cytosol separation was performed by Membrane and Cytosol Protein Extraction Kit (#PC202, Epizyme, China). The concentrations of the extracted proteins were measured by BCA Kit (#KTD3001, Abbkine, China). Proteins were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis, transferred to polyvinylidene difluoride membranes (#IPVH00010, Merk-Millipore, Germany), and detected using specific primary antibodies followed by horseradish peroxidase-conjugated secondary antibodies. Protein signals were visualized by enhanced chemiluminescence (#BMU102, Abbkine, China) on an imaging system (Bio-Rad, USA).

Real-time quantitative PCR

Total RNA was extracted using the RNA extraction kit (#AG21023, Aikori, China), then reverse-transcribed into cDNA with Strand cDNA Synthesis SuperMix (#11141ES60, Yeasen, China) according to manufacturers’ protocols. Quantitative PCR was performed with pre-mixed SYBR Green Master Mix (#11184ES08, Yeasen, China) using a two-step program on a real-time qPCR system (Bio-Rad, USA). Primer sequences are available in Supplementary File-Table S2.

2-NBDG assay

Glucose uptake capacity was assessed by 2-NBDG Glucose Uptake Assay Kit (#S0561S, Beyotime, China). After seeding, cells underwent glucose-free starvation for 12 h, followed by 1-hour incubation with 2-NBDG and quantification via flow cytometry (Beckman, USA).

Lactate assay and extracellular flux (XF) analysis

Following cell lysis by sonication, protein concentration was quantified using a BCA assay kit (#KTD3001, Abbkine, China). Lactate levels were measured with the Lactate Assay Kit (#KTB1100, Abbkine, China), normalized to total protein content (µmol/mg), and compared with the first group (typically the control group). Extracellular flux analysis was performed using a Seahorse XF96 analyzer. Cells were seeded at 4 × 10³ cells per well in XF96 plates (#101085-004, Agilent, USA) and incubated for 24 h. One hour before assay, the medium was replaced with XF Base Medium. Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured using the XF Cell Mito Stress Test Kit (#103015-100, Agilent, USA) and the XF Glycolytic Rate Assay Kit (#103344-100, Agilent, USA), respectively, following the manufacturer’s instructions. All measurements were normalized to cell number per well.

Immunofluorescence

Cells grown on coverslips were fixed with ice-cold methanol, permeabilized, blocked, and incubated with primary antibody at 4 °C overnight. After incubation with fluorophore-conjugated secondary antibody (Yeasen, China) for 1 h at room temperature, nuclei were stained with DAPI (#C0065, Solarbio, China). Samples were imaged by confocal microscopy (Leica, Germany).

Metabolomics and metabolic flux analysis

Metabolomic analysis was performed by Ultra-Performance Liquid Chromatography Tandem Mass Spectrometry (UPLC-MS/MS) technology. The workflow comprised: sample collection, metabolite extraction, LC-MS/MS detection, and data processing. Acquired raw spectral data were interrogated against a database to establish metabolite identities and abundances, followed by data curation including missing value imputation and quality control. Processed data were subjected to statistical analysis and visualization. For isotope-labeled samples, metabolic flux analysis was conducted via Gas Chromatography-Mass Spectrometry (GC-MS). Bar plots for metabolomic results were generated using median-normalized and log-transformed peak area data, while flux analysis bar graphs utilized preprocessed peak area data.

Animal tumor models and in vivo treatments

Female BALB/c nude mice (4 weeks old) were purchased from GemPharmatech (Jiangsu, China). Animals were weight-matched and randomized into groups. Subcutaneous injections of RNF114-knockdown or control stable cells were performed. After 3 weeks, daily intraperitoneal injections of drug or vehicle control (PEG300) were given. Tumors were measured every 2–3 days, with volume calculated as (length × width²) × 0.52. Upon reaching the endpoint (1500 mm³), mice were euthanized. Tumors were fixed in 4% paraformaldehyde, sectioned, and analyzed by IHC. For metabolomics, tumors were snap-frozen in liquid nitrogen, followed by metabolite extraction. All animal experiments were approved by the Animal Ethics Committee of West China Hospital (20241122010).

Statistical analysis

Image J software was used to quantify grayscale values of protein bands, IHC staining intensity, cellular fluorescence intensity, and crystal violet-stained cell counts. Data preprocessing and analysis of the GEO database were performed using RStudio. Statistical analysis used GraphPad Prism 8.0 and SPSS Statistics 26, with all values expressed as mean ± standard deviation (SD). GraphPad Prism 8.0 and BioGDP were used to generate plots [19]. Survival curves were generated by Kaplan-Meier analysis with between-group comparisons via log-rank test. The association between RNF114 protein levels and clinical pathological indexes was analyzed using the chi-square test. When expected frequencies were less than 5, either the Yates’s correction for continuity or Fisher’s exact test was applied. Clinical data underwent univariate Cox analysis, after which selected variables meeting the criterion of P < 0.25 were added to the multivariate Cox regression model based on the model’s parameter requirements. IC50 values were determined by nonlinear regression curve fitting. Unless otherwise stated in figure legends, statistical significance between two independent groups was determined using unpaired, two-tailed Student’s t-test. Datasets with more than two groups were analyzed using one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test. For datasets involving two independent factors, two-way ANOVA was performed followed by Sidak post hoc test. All P values below 0.05 were considered statistically significant (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). All experiments were repeated at least three times unless stated otherwise.

Results

RNF114 is upregulated in HCC, and its high expression is associated with poor prognosis in HCC patients

The RNA-seq data from The Cancer Genome Atlas (TCGA) and protein data from the Clinical Proteomic Tumor Analysis Consortium (CPTAC) of liver cancer and paracancerous tissues were analysed [20, 21]. The RNF114 RNA and protein levels were increased in various cancers (Fig. 1A, B). Notably, in liver cancer, RNF114 mRNA and protein levels are significantly upregulated. Analysis of TCGA data further demonstrated a pronounced elevation of its mRNA in HCC (Figure S1A, B). Analysis of a Gene Expression Omnibus (GEO) dataset [22] revealed significant upregulation of RNF114 in HCC compared to non-tumor tissue across two independent cohorts, and ROC curves from both cohorts supported its potential as a biomarker (Fig. 1C-G). Furthermore, survival analysis of TCGA, GSE16757 [23], and GSE116174 [24] showed that high RNF114 expression was significantly associated with poor prognosis in HCC patients (Fig. 1H-J). Immunohistochemistry (IHC) of a tissue microarray (TMA) containing 86 paired HCC samples demonstrated that high RNF114 expression was significantly associated with a lower survival rate in HCC patients [25, 26] (Fig. 1K-M, Figure S1C, D). In addition, we found that pathological grade was significantly associated with RNF114 expression. However, no significant association was found between RNF114 expression and patients’ gender, age, clinical stage, liver cirrhosis nodules, tumor capsule or recurrence (Table S3). Cox univariate and multivariate analyses revealed that high RNF114 expression was an independent predictor of poor prognosis in HCC, which is consistent with our previous findings (Table S4 and Table S5). In general, our results indicate that RNF114 is highly expressed in HCC, associated with the pathological grade and poor prognosis, suggesting its potential as a novel prognostic biomarker.

Fig. 1.

RNF114 is highly expressed and associated with poor prognosis in HCC. (A) Pan-cancer analysis of RNF114 mRNA expression levels in TCGA samples. (B) Pan-cancer analysis of RNF114 protein expression levels in CPTAC database. (C) Venn diagram showing the overlap of upregulated genes between GPL571 and GPL3921 of GSE14520. Paired scatter plot of GSE14520-GPL571 (D) and GSE14520-GPL3921 (E). ROC curves of GSE14520-GPL571 (F) and GSE14520-GPL3921 (G), AUC is the area under the curve, CI is the confidence interval. (H) Overall survival and disease-free survival analysis of RNF114 in liver cancer based on the TCGA database. Kaplan-Meier curves for overall survival in GSE16757 (I) and GSE116174 (J). (K) Representative IHC images of RNF114 staining from high- and low-expression groups. Scale bars: 200 μm (overview), 100 μm (inset). (L) IHC images of paired tumor tissues (T) and adjacent non-tumor tissues (N) from RNF114 high-expression group. Scale bars: 200 μm (overview), 100 μm (inset). (M) Kaplan-Meier analysis of overall survival for HCC patients from the TMA. Log-rank test, P = 0.0274. Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 and ns (not significant), as determined by paired t-test (D, E) and log-rank test (H, I, J, M)

Knockdown of RNF114 suppresses HCC cell proliferation and metastasis in vitro

In order to elucidate the role of RNF114 in HCC, two siRNA knockdown sequences targeting RNF114 were constructed. The efficiency of knockdown was verified by Western blot and qPCR (Figure S2A, B). Subsequently, stable lentivirus was generated using the sequence of siRNF114-1. Stable RNF114-knockdown cell lines were established in HuH-1, HCCLM3, and MHCC97-H cells, while RNF114-overexpressing HuH-1 cells were generated by transient plasmid transfection. CCK-8 assays demonstrated that decreasing RNF114 significantly suppressed proliferation, whereas overexpressing RNF114 increased cell growth (Fig. 2A). Consistent with this, crystal violet staining showed that knockdown of RNF114 led to a decrease in colony formation, while overexpression of RNF114 resulted in enhanced tumorigenicity (Fig. 2B). These results suggested RNF114 acts as a critical driver of HCC cell growth. Furthermore, the migratory and invasive capacities of the cells were significantly suppressed by RNF114 knockdown, and enhanced by RNF114 overexpressing (Fig. 2C, D), suggesting RNF114 critically regulates metastatic potential in HCC cells.

Fig. 2.

RNF114 promotes cell proliferation, migration, invasion and glycolysis in HCC. Experimental systems: transient transfection with siNC, siRNF114-1 and siRNF114-2 in HuH-1, HCCLM3 and MHCC97-H cell lines; transient transfection with RNF114-overexpression or empty vector plasmids in HuH-1 cells; stable RNF114-knockdown in HuH-1, HCCLM3 and MHCC97-H cell lines. (A) CCK-8 assay quantified proliferation rates in HCC cells (n = 4). (B) Representative colony formation images of HCC cells (n = 3). (C, D) Crystal violet quantified transwell migration and invasion in HCC cells (n = 3). (E) Measurement of 2-NBDG uptake by flow cytometry (MFI: median fluorescence intensity; n = 3). (F) Quantification of intracellular lactate levels (n = 3). (G) Measurement of ECAR (n = 3). (H) Measurement of OCR (n = 3). (I-L) The treatment of glycolysis inhibitor WZB117 reduced the effect of RNF114 knockdown on suppressing HuH-1 cells from growing and spreading (proliferation, clonogenicity, migration and invasion) compared to untreated controls (n = 4 in CCK-8 assay; n = 3 in other assays). Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 and ns (not significant), as determined by two-way ANOVA (A, I), one-way ANOVA (C, D, E, F, J-L) and two-tailed Student’s t-test (B-D, E-H)

RNF114 promotes glycolytic metabolism in HCC cells

Flow cytometry analysis with 2-NBDG uptake showed that RNF114 significantly promoted glucose uptake in HCC cells (Fig. 2E, Figure S3). Moreover, lactate production, a critical indicator of tumor glycolysis, was elevated with RNF114 overexpressing but reduced upon RNF114 knockdown (Fig. 2F). Our data showed that RNF114-knockdown dramatically inhibited ECAR and OCR, confirming a positive correlation between glycolytic activity and RNF114 expression (Fig. 2G, H). To further confirm the role of RNF114-enhanced glycolytic metabolism in HCC progression, HuH-1 cells were treated with glycolytic inhibitor WZB117 and subjected to CCK-8, colony formation and transwell assays (Fig. 2I-L). Our findings suggested that RNF114 increases cellular glycolysis to promote the growth and metastasis of HCC cells.

RNF114 modulates metabolic reprogramming in HCC

Based on UPLC-MS/MS analysis, 12 different metabolites (combined positive/negative ionisation modes) were identified between the RNF114-knockdown and control groups. The principal component analysis (PCA) showed distinct metabolic separation, a finding that was supported by PLS-DA, where the first principal component explained 51% of the total variance with a Q2 value of 0.651, thus confirming the reliability of the model (Fig. 3A, B). The knockdown of RNF114 decreased the levels of certain metabolites, especially glycolytic metabolites such as D-fructose 6-phosphate (D-F6P) and pyruvate (Fig. 3C, D). In murine ectopic xenograft models, 54 differential metabolites (positive and negative ionisation modes) were identified in RNF114-knockdown tumors compared with control tumors. PCA revealed metabolic reprogramming, with PLS-DA showing 73% variance explained by the first component (with Q2 = 0.753) (Fig. 3E, F). Consistent with cellular findings, results of tumors demonstrated a reduction in certain glycolytic metabolites, especially D-F6P, pyruvate and lactate, suggesting a disruption in glucose metabolism (Fig. 3G-I). To determine whether metabolic alterations resulted from enzymatic changes, ¹³C₆-glucose tracing was performed in glucose-starved cells, followed by GC-MS analysis. While the total uptake of D-glucose increased, there were no alterations in uptake kinetics. This suggested the increasing number of functional glucose transporters, rather than the enhancing transporter activity (Fig. 3J). The activities of key glycolytic enzymes (Hexokinase, Phosphofructokinase-1, Pyruvate kinase M2, and Lactate dehydrogenase A) remained unchanged (Figure S4). Integrative analysis of cellular metabolomics, tumor tissue metabolomics, and ¹³C metabolic flux demonstrated that RNF114 enhances glucose utilisation in cells and tumor tissue without modulating catalytic enzyme activity, but rather through elevated total glucose uptake.

Fig. 3.

Cellular metabolomics, animal metabolomics and ¹³C metabolic flux tracing in HCC. Cell models: Stable RNF114-knockdown and control HuH-1 cells. Tumor: Heterotopic xenografts in nude mice established from HuH-1 cell line. FC: fold-change. PCA (A) and PLS-DA (B) score plots of the metabolomic model for RNF114-knockdown cells (R2Y = 0.951, PR2Y = 0.01, Q2 = 0.651, PQ2 = 0.01; n = 6). (C) Volcano plot (Wilcoxon test): Differentially abundant metabolites (P < 0.05) in shRNF114 cells vs. shNC cells (n = 6). (D) Heatmap: log₂FC of differential metabolites (P < 0.05, Wilcoxon) in shRNF114 cells vs. shNC cells (n = 6). Column chart: quantitative analysis of differential metabolites in cells (models-internal standard-normalized and median-scaled data; n = 6). The PCA (E) and PLS-DA (F) of metabolomic profiles in xenograft tumors derived from stable RNF114-knockdown cells vs. control cells (R2Y = 0.953, PR2Y = 0.02, Q2 = 0.753, PQ2 = 0.02; n = 7). (G) Heatmap: log₂FC of differentially abundant metabolites (P < 0.05) in RNF114-knockdown vs. control xenografts (n = 7). (H) Quantitative analysis of differential metabolites in xenograft tumor models. Multi-step processing: BCA-normalization; Primary log₂; Median-scaling; Final log₂ (n = 7). (I) Volcano plot (Wilcoxon test): Intratumoral metabolites differing in RNF114-knockdown vs. control xenografts (P < 0.05; n = 7). (J) ¹³C₆-glucose tracing analysis of metabolite labeling patterns in HuH-1 cells (n = 6). Total = Σ M_i (i = 1 to n). Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 and ns (not significant), as determined by two-way ANOVA (left in J) and two-tailed Student’s t-test (D, H, J)

RNF114 promotes glycolysis by enhancing plasma membrane localization of GLUT1

To investigate the specific mechanism by which RNF114 regulates glycolysis, the mRNA and protein levels of rate-limiting enzymes involved in glycolysis and lactate production were examined. RT-qPCR results showed that knockdown of RNF114 did not affect the transcriptional levels of the glycolytic rate-limiting enzymes SLC2A1, HK1, HK2, PFKM (PFK1), PKM2, GAPDH and LDHA (Fig. 4A, B). Meanwhile, Western blot analysis revealed that while HK1, HK2, PFK1, PKM2, GAPDH and LDHA protein levels remained unchanged, GLUT1 protein levels were altered upon RNF114 knockdown or overexpressing, exhibiting a positive correlation (Fig. 4C, Figure S5A, B). Immunoprecipitation analysis showed that RNF114 knockdown specifically reduced N-glycosylated GLUT1 in HuH-1 cells, suggesting that RNF114 may modulate GLUT1 membrane localization (Fig. 4D). Previous studies demonstrate that N-glycosylation of proteins facilitates their membrane localization, which is a prerequisite for GLUT1 to transport extracellular glucose into cells [27–30]. So we asked whether RNF114 modulates GLUT1 membrane localization. Immunofluorescence confirmed that membrane localization of GLUT1 was reduced upon RNF114 knockdown (Fig. 4E). Membrane-cytosol fractionation assays further showed a decrease in the membrane-to-cytosol ratio of GLUT1 following RNF114 depletion (Fig. 4F). Collectively, these results demonstrated that RNF114 enhanced the membrane localization of GLUT1. Furthermore, the knockdown of GLUT1 significantly abolished the suppressive effect of RNF114-knockdown on lactate production, confirming that RNF114 regulates lactate through a hierarchical signaling cascade where GLUT1 acts as the essential downstream mediator (Fig. 4G).

Fig. 4.

RNF114 enhances membrane localization of GLUT1. (A) RT-qPCR analysis of RNF114-knockdown and control cells (HuH-1), with mRNA levels normalized to ACTB (n = 3). (B) Schematic representation of glucose fermentation to lactate. (C) Protein levels of key glycolytic enzymes examined by Western blotting in HuH-1 cells (n = 3). (D) N-glycosylated and unmodified GLUT1 protein levels in tunicamycin-treated (5µM, 24 h) and untreated stable RNF114-knockdown HuH-1 cells (n = 3). (E) Immunofluorescence analysis of GLUT1 subcellular localization in RNF114-knockdown cells vs. control cells. Scale bars: 10 μm. (F) Western blotting of GLUT1 protein levels in membrane (CM) and cytoplasmic (Cy) fractions of HCC cells (n = 3). (G) Knockdown of GLUT1 abolished the suppressive effect of RNF114-knockdown on lactate production (n = 3). Data are presented as mean ± SD. **P < 0.01, ***P < 0.001, ****P < 0.0001 and ns (not significant), as determined by two-tailed Student’s t-test (A, C, F) and one-way ANOVA (D, G)

In summary, RNF114 regulates glycolysis by influencing the levels of functional GLUT1 in the cells, which was achieved by modulating GLUT1’s membrane localization.

RNF114 promotes PACSIN3 degradation via enhanced SH3 ubiquitination to suppress endocytosis

To examine the mechanistic role of RNF114 participating in altering the subcellular localization of GLUT1, HuH-1 cell lysates were subjected to an IP-MS screening followed by co-IP validation, revealing a protein-protein interaction between RNF114 and PACSIN3 (Fig. 5A, Figure S6A-D). A study demonstrated that PACSIN3 promotes the translocation of GLUT1 in adipocytes [31]. Furthermore, several studies have also shown that PACSIN proteins are associated with endocytic vesicle formation [32, 33]. So we would like to know whether RNF114 regulates GLUT1 by affecting endocytic trafficking. In order to investigate the effect of PACSIN3 on GLUT1, a series of experiments were conducted into both the mRNA and protein levels of GLUT1. RT-qPCR results demonstrated that neither RNF114 nor PACSIN3 transcriptionally regulated GLUT1 (Figure S6E). Western blot analysis of whole-cell lysates (WCL), membrane (CM) and cytosolic (Cy) fractions revealed that RNF114 increased total protein level of GLUT1 and its membrane-to-cytosol ratio. Meanwhile, transient transfection of siPACSIN3 in RNF114-knockdown cells restored GLUT1 levels and membrane localization, while OE-PACSIN3 transfection reduced them in RNF114-overexpressing cells (Fig. 5B). Immunofluorescence assay results confirmed the retention of GLUT1 on the membrane in response to the rescue of siPACSIN3 in RNF114-knockdown cells, with the same recovery of lactate production (Fig. 5C, D). Western blot analysis demonstrated that RNF114-knockdown resulted in an increase in PACSIN3 protein levels, while RNF114-overexpressing led to a decrease in PACSIN3 (Fig. 5E, Figure S6F). Cycloheximide (CHX) chase assays (0–6 h) showed sustained PACSIN3 stability in RNF114-knockdown cells but progressive degradation in the control group (Fig. 5F). MG132, but not chloroquine, reversed the accumulation of PACSIN3 induced by the knockdown of RNF114 (Fig. 5G). These findings suggested that RNF114 regulates PACSIN3 stability. Given RNF114’s role as a RING-domain E3 ubiquitin ligase, it was hypothesized that RNF114 ubiquitinates PACSIN3 to promote its proteasomal degradation. Western blot revealed that co-transfection of Flag-PACSIN3 and RNF114 in HEK-293T cells resulted in enhanced ubiquitination of PACSIN3 (Fig. 5H). We constructed plasmids for the two major domains of PACSIN3 (Fig. 5I) [34]. Membrane fractionation revealed that F-BAR domain mediates membrane binding and endocytosis, resulting in GLUT1 internalization, while the SH3 domain is primarily cytosolic (Fig. 5J). This localization of SH3 facilitated its interaction with RNF114, as shown by co-IP in HEK-293T cells. RNF114 binds to the SH3 domain, promoting its ubiquitination and consequently triggering PACSIN3 degradation via the proteasome pathway (Fig. 5K, L). Additionally, the pharmacological inhibition of endocytosis using dynasore enhanced GLUT1 membrane retention. The combination of dynasore and siPACSIN3 provided no additive effect, and dynasore also restored the OE-PACSIN3-induced decrease in membrane GLUT1, suggesting that PACSIN3 participated in dynamin-dependent endocytic trafficking (Fig. 5M, N, Figure S6G). We further performed cell surface biotinylation‑based internalization assays to directly demonstrate that PACSIN3 knockdown decreased GLUT1 internalization in HuH-1 cells, whereas PACSIN3 overexpression enhanced it (Fig. 5O). Notably, we did not detect a direct interaction between PACSIN3 and GLUT1; instead, our results indicated that PACSIN3 promoted GLUT1 internalization through increased membrane curvature mediated by its F‑BAR domain (Figure S6H, I). Clathrin-mediated endocytosis is a sequential process. It involves the assembly of the clathrin coat, membrane invagination, constriction of the vesicle neck, and finally, dynamin-mediated vesicle scission from the plasma membrane. Previous studies have shown that the F-BAR domain of PACSIN family proteins participates in the membrane invagination step [35, 36]. Consistently, re‑expression of full‑length PACSIN3 or its F‑BAR domain alone in PACSIN3‑knockdown cells rescued GLUT1 internalization (Fig. 5P). Our fluorescence imaging revealed that overexpression of PACSIN3 enhanced membrane internalization. However, this enhancement was largely abolished when dynamin was inhibited by Dynasore, which prevents the vesicle scission process (Fig. 5Q). Likewise, lactate level assays concluded that RNF114 modulated GLUT1-mediated glycolysis by regulating the level of PACSIN3, and the endocytosis was rescued by dynasore (Fig. 5R). N-glycosylated GLUT1 is unstable in the cytoplasm. Our results further indicated that GLUT1 internalized into the cytoplasm was preferentially degraded via the lysosomal pathway rather than the proteasomal pathway, as evidenced by the accumulation of GLUT1 in the cytoplasm upon treatment with the lysosomal inhibitor chloroquine, but not with the proteasome inhibitor MG132 (Fig. 5S, Figure S6J).

Fig. 5.

RNF114 promotes PACSIN3 ubiquitination to stabilize GLUT1 membrane localization. (A) Co-immunoprecipitation of RNF114 in HCC cells with special antibodies. (B) Analysis of protein levels by Western blot in WCL, CM and Cy fractions of HuH-1 cells (n = 3). (C) Representative immunofluorescence images showing GLUT1 subcellular localization in HCC cells transfected with siNC, siRNF114, or siRNF114 plus siPACSIN3. Scale bars: 10 μm. (D) Relative lactate levels in HuH-1 cells (n = 3). (E) PACSIN3 protein levels upon RNF114-knockdown or RNF114-overexpressing in HuH-1 cells (n = 3). (F) Quantification of PACSIN3 protein levels in control and RNF114-knockdown HuH-1 cells treated with 40µM CHX over 0–6 h (n = 3). (G) Analysis of protein expression in HuH-1 cells treated with 10 µM MG132 or chloroquine (CQ) for 24 h (n = 3). (H) Enhanced ubiquitination of Flag-PACSIN3 by RNF114. Analysis of Flag-PACSIN3 ubiquitination by co-IP in HEK-293T cells co-transfected with Flag-PACSIN3 and either RNF114 plasmid or empty vector control. (I) Domain structure of PACSIN3. (J) Analysis of protein levels by Western blot in WCL, CM, and Cy fractions of HuH-1 cells transfected with various plasmids for 48 h (n = 3). (K) Domain-specific Co-IP assay in HEK-293T cells. (L) Analysis of domain ubiquitination by Co-IP in HEK-293T cells. (M) Analysis of protein levels by Western blot in WCL, CM, and Cy fractions of HuH-1 cells under various interventions. Dynasore (Dy): 10µM for 12 h; siRNA: 100nM for 48 h (n = 3). (N) Relative lactate levels in HuH-1 cells under various interventions. Dy: 10µM for 12 h; siRNA: 100nM for 48 h (n = 3). (O) Cell surface biotinylation-based internalization assay over 0–30 min (n = 3). (P) Cell surface biotinylation-based internalization assay (n = 3). (Q) Representative immunofluorescence images of GLUT1 subcellular localization in HuH-1 cells with plasmid transfection and pharmacological interventions. WT: wild type. Dy: 10µM for 12 h; plasmid intervention: 72 h. Scale bars: 10 μm. (R) Protein-normalized lactate production (vs. WT group; n = 3). WT: wild type. (S) Western blot analysis of CQ (10 µM, 24 h) rescue (n = 3). CQ: chloroquine. Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 and ns (not significant), as determined by two-tailed Student’s t-test (E), two-way ANOVA (F) and one-way ANOVA (B, D, G, J, M, N, O, P, R, S)

Collectively, RNF114 ubiquitinates PACSIN3 to trigger its degradation, thereby suppressing PACSIN3-mediated endocytic trafficking. This pathway stabilizes GLUT1 membrane retention to potentiate glycolytic flux in HCC.

Knockdown of RNF114 suppresses glycolysis-dependent tumor growth and Sorafenib resistance in xenograft models

To investigate the role of RNF114 in HCC in vivo, stable RNF114-knockdown HuH-1 cells were subcutaneously injected into 6-week-old nude mice, with control plasmid-infected HuH-1 cells as control. Three weeks after implantation, mice were randomly divided into different groups for daily intraperitoneal injections of glycolysis inhibitor WZB117 (or PEG300) at a dose of 10 mg/kg (Fig. 6A). Results demonstrated significantly reduced tumor volumes in mice injected with RNF114-knockdown cells compared to those injected with control cells. Notably, WZB117 treatment reduced mean tumor size in mice implanted with wildtype HuH-1 cells, showing a significant difference from the untreated control group. But there was less difference in the treatment group and the control group implanted with RNF114 knockdown stable cells. Also, treatment of WZB117 can reduce the difference between RNF114 knockdown stable cells and wildtype cells. These findings indicated that the role of RNF114 in HCC growth is associated with glycolysis (Fig. 6B). Additionally, IHC analysis revealed that RNF114 knockdown reduced GLUT1 expression in membrane. However, the effect of RNF114 knockdown on GLUT1 was reduced when WZB117 was added (Fig. 6C).

Fig. 6.

RNF114 enhances tumor growth and Sorafenib resistance in xenograft model. (A) Establishment of ectopic xenograft models in BALB/c-Nude mice (implanted with RNF114-knockdown stable cells or control cells) and drug treatment (injected with 10 mg/kg WZB117 or PEG300) scheme. (B) Representative photos of three tumor tissues from each group and tumor growth rate. Volume (mm³) = (Length × Width²)×0.52. Black arrows: the first day of dosing initiation (once daily). Six mice per group (n = 6). (C) Representative immunohistochemical images of RNF114 and GLUT1 in tumor tissues (n = 6). Scale bar, 100 μm. (D) Determination of IC50 values in HuH-1 cells treated with Sorafenib among different groups (n = 3). (E) Ectopic xenograft models in BALB/c-Nude mice and drug treatment (injected with 10 mg/kg Sorafenib or PEG300) scheme. (F) Representative photos of three tumor tissues from each group and tumor growth rate. Volume (mm³) = (Length × Width²)×0.52. Black arrows: the first day of dosing initiation (once daily). Six mice per group (n = 6). (G) Determination of IC50 values in HuH-1 cells treated with Sorafenib among different groups (n = 3). Na Pyr: Sodium pyruvate. (H) Lactate production measurement in HuH-1 cells under different treatments (n = 3). Na Pyr: Sodium pyruvate. (I) Representative tumor images and growth curves from nude mice treated with Sorafenib (10 mg/kg) plus WZB117 (10 mg/kg) in conjunction with shRNF114 (n = 6). Volume (mm³) = (Length × Width²)×0.52. Black arrows: the first day of dosing initiation (once daily). Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 and ns (not significant), as determined by one-way ANOVA (C, H), two-way ANOVA (B, F, I) and nonlinear regression curve fitting (D, G)

Accumulating evidence suggests that tumor-specific glycolysis and lactate production drive Sorafenib resistance. Inhibition of glycolysis has been demonstrated to enhance the anti-hepatocellular carcinoma effects of Sorafenib [37, 38]. Consistent with this, our results also showed that stable RNF114-knockdown sensitized HCC cells to Sorafenib, while transient RNF114 overexpression conferred resistance (Fig. 6D). To examine whether RNF114 plays a role in Sorafenib resistance in vivo, xenograft models were established and injected with a low-dose (10 mg/kg, daily) of Sorafenib [39] (Fig. 6E). As expected, the control group showed no significant reduction in tumor size under low-dose of Sorafenib treatment. However, the growth of RNF114-knockdown tumors was significantly suppressed (Fig. 6F).

Meanwhile, IC50 assays showed that inhibiting glycolysis reduced RNF114-mediated resistance to Sorafenib. However, adding sodium pyruvate to restore lactate levels led to the restoration of drug resistance (Fig. 6G, H, Figure S7A, B). In our Sorafenib-treated nude mouse model (10 mg/kg), the anti-proliferative effect of RNF114 knockdown was reduced by WZB117-mediated glycolytic inhibition, confirming that RNF114 promotes Sorafenib resistance by modulating glycolysis. Notably, tumor volume was significantly reduced in both the shRNF114/Sorafenib and shRNF114/WZB117/Sorafenib groups compared to the Sorafenib-treated group, with no significant difference between the two combination-treated groups. These results suggest that targeting RNF114 is a pivotal strategy for overcoming Sorafenib resistance (Fig. 6I).

Discussion

Previous studies have shown that cancer cells frequently depend on aerobic glycolysis, a metabolic phenomenon known as the Warburg effect. Notably, combination therapies targeting upstream glycolytic regulators such as PTBP1, HEXB, and NBS1 have been shown to enhance the sensitivity of HCC to radiotherapy and chemotherapy, suggesting that targeting glycolytic pathways represents a promising therapeutic strategy for HCC [40–43]. Therefore, elucidating the signaling mechanisms that regulate glycolytic flux is critical. GLUT1, a central glucose transporter overexpressed in HCC, is a key target of pro-tumorigenic signals like the BMP4/SMAD and HIF-1α pathways. Inhibition of GLUT1 expression by targeting these pathways suppresses HCC progression and increases drug sensitivity [6, 44]. Here, we identified a novel mechanism whereby RNF114 promotes HCC progression: it enhances glycolysis by inhibiting PACSIN3-dependent endocytosis of GLUT1.

In this study, we found that RNF114 is an independent factor of poor prognosis and associated with the pathological grade in HCC patients. Here, we reported that, in HCC, knockdown of RNF114 significantly suppressed glycolytic levels, consequently inhibiting cell proliferation, migration and invasion, revealing its critical role in HCC. Cellular metabolomics, tumor tissue metabolomics, and metabolic flux analyses consistently indicated that RNF114 enhances glycolytic flux and increases glycolytic metabolite production in both cellular and tumor contexts. However, this alteration was not attributable to enhanced activity of rate-limiting glycolytic enzymes. Intriguingly, our data revealed that RNF114 promotes GLUT1 membrane localization to elevate glycolysis. RNF114-induced GLUT1 membrane localization enhanced extracellular glucose uptake rather than cellular glucose utilization efficiency. And the regulation of GLUT1 by RNF114 was associated with PACSIN3-mediated endocytosis.

The stability and function of membrane proteins, including GLUT1, are critically regulated by endocytic trafficking [45–47]. GLUT1 is primarily internalized via clathrin-mediated endocytosis (CME), a process coordinated by numerous proteins. Among these, PACSIN family proteins contribute to CME in part through their F-BAR domain, which promotes membrane curvature and facilitates the invagination essential for the process [35, 36].

Our work revealed a novel regulatory mechanism that stabilizes GLUT1 at the plasma membrane by targeting its endocytic removal. For the first time, we demonstrated that RNF114 interacts with the SH3 domain of PACSIN3, promoting its ubiquitin-mediated degradation. The loss of PACSIN3 suppresses F-BAR domain-dependent membrane invagination, thereby inhibiting GLUT1 endocytosis. This leads to GLUT1 accumulation at the plasma membrane, which boosts glucose uptake and glycolytic flux. The resulting increase in glycolytic metabolites, including lactate, contributes to enhanced drug resistance. In line with this, our mouse xenograft models showed that RNF114-driven glycolysis promotes HCC growth and confers Sorafenib resistance (Fig. 7).

Fig. 7.

Working model of RNF114 in promoting HCC progression

Enhanced glycolysis reflects the high glucose dependency of HCC cells to sustain energy metabolism, and simultaneously shapes a tumor metabolic microenvironment that favors drug resistance. In China, North America and Europe, over 60% cases of HCC are diagnosed at intermediate or advanced stages with limited therapeutic options [48]. Sorafenib is the primary agent used in the clinical management of advanced HCC, yet its effectiveness is significantly reduced due to drug resistance [49, 50]. Targeting metabolic vulnerabilities to overcome such resistance has thus emerged as a compelling therapeutic strategy. Here, we reveal that RNF114 drives glycolytic reprogramming in HCC by promoting GLUT1 plasma membrane localization via PACSIN3 degradation, defining a novel RNF114–PACSIN3–GLUT1 regulatory axis. Disruption of this axis sensitizes HCC cells to Sorafenib both in vitro and in vivo by suppressing glycolytic metabolism. Given that a lactate-enriched microenvironment arising from enhanced glycolysis contributes broadly to multidrug resistance, targeting RNF114 could have broader therapeutic implications beyond Sorafenib, warranting further investigation.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

HCC

Hepatocellular carcinoma

TMA

Tissue microarray

IHC

Immunohistochemistry

ROI

Regions of interest

CHX

Cycloheximide

IP

Immunoprecipitation

UPLC-MS/MS

Ultra-Performance Liquid Chromatography Tandem Mass Spectrometry

GC-MS

Gas Chromatography-Mass Spectrometry

SD

Standard deviation

ANOVA

Analysis of variance

H-scores

Histochemistry scores

ECAR

Extracellular acidification rate

PCA

Principal component analysis

CME

Clathrin-mediated endocytosis

MFI

Median fluorescence intensity

Author contributions

Conceptualization: YS; Methodology: YS; Investigation: YS, QXL, LY, WZ, GL, XYH, XYL, YZ; Visualization: YS, YLL; Supervision: YS, YLL, YXM; Writing—original draft: YS; Writing—review & editing: YS, YLL, YXM.

Funding

This work was supported by National Natural Science Foundation of China (grant 31970698) and Sichuan Science and Technology Program (grant 2023YFS0109).

Data availability

All data generated or analysed during this study are included in this published article and its supplementary information files.

Declarations

Compliance with ethical standards

The TMA chip HLivH180Su09 was reviewed and approved by the Ethics Committee of Shanghai Outdo Biotech Company (Approval No. YB M-05-02). All procedures involving human participants were performed in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki Declaration and its later amendments. All animal experiments were approved by the Animal Ethics Committee of West China Hospital (Approval No. IACUC-20241122010).

Competing interests

The authors declare no competing interests.