Oxidative stress induces extracellular vesicle release by upregulation of HEXB to facilitate tumour growth in experimental hepatocellular carcinoma

Abstract

Extracellular vesicles (EVs) play a crucial role in triggering tumour‐aggressive behaviours. However, the energetic process by which tumour cells produce EVs remains poorly understood. Here, we demonstrate the involvement of β‐hexosaminidase B (HEXB) in mediating EV release in response to oxidative stress, thereby promoting the development of hepatocellular carcinoma (HCC). Mechanistically, reactive oxygen species (ROS) stimulate the nuclear translocation of transcription factor EB (TFEB), leading to the upregulation of both HEXB and its antisense lncRNA HEXB‐AS. HEXB‐AS can bind HEXB to form a protein/RNA complex, which elevates the protein stability of HEXB. The stabilized HEXB interacts with lysosome‐associated membrane glycoprotein 1 (LAMP1), disrupting lysosome‐multivesicular body (MVB) fusion, which protects EVs from degradation. Knockdown of HEXB efficiently inhibits EV release and curbs HCC growth both in vitro and in vivo. Moreover, targeting HEXB by M‐31850 significantly inhibits HCC growth, especially when combined with GW4869, an inhibitor of exosome release. Our results underscore the critical role of HEXB as a modulator that promotes EV release during HCC development.

The ROS promote the nuclear translocation of TFEB, increasing the expression of HEXB and its antisense lncRNA HEXB‐AS, which can bind HEXB to elevate the stability of HEXB. The stabilized HEXB interacts with LAMP1 and mediates the degradaiton of LAMP1, disrupting lysosome‐MVB fusion, which protects EVs from degradation.

1. INTRODUCTION

Cancer progression can be facilitated by intercellular communication, and such signalling relies heavily on the secretion of soluble factors and the exchange of extracellular vesicles (EVs). Previous studies revealed that EV secretion, cargo packaging, and the ability to interact with recipient cells can be affected by early genetic drivers of cancer (Kalluri & McAndrews, 2023; Qin et al., 2020). In comparison with non‐malignant cells, malignant cells typically secrete a higher quantity of EVs (Taylor et al., 2020). For example, in pancreatic cancer, recent research reported that pancreatic EVs released by cancer cells have been shown to drive non‐tumorigenic recipient cell transformation by transporting biomolecules that induce ER stress (Hinzman et al., 2022). This finding indicates that cancer cells may actively transfer oncogenic components to surrounding cells during malignant transformation by increasing EV secretion.

Oxidative stress has been identified to participate in various stages of cancer development. Through inducing malignant cell proliferation, reactive oxygen species (ROS) promote the pathological processes of tumorigenesis and subsequent neoplastic growth (Forman & Zhang, 2021; Hayes et al., 2020; Li et al., 2022). In recent years, the critical role of EVs in oxidative stress‐related pathologies, including cancer, has been well demonstrated, and the relationship between oxidative stress and EVs provides a novel perspective to understand ROS‐induced tumour development (Borras et al., 2020). Substantial experimental evidence has observed a marked increase in the number of EVs, especially exosomes, under oxidative stress. Consequently, interfering with EV release under oxidative stress may represent a potential strategy to disrupt EV‐driven pathological processes (Adams et al., 2021; Hirsova et al., 2016; Zhang et al., 2022). However, the mechanisms that account for the enhanced release of oxidative stress‐related EVs from carcinoma cells and the extent to which these EVs support cancer development remain to be clarified.

Here, we unveil the role of β‐hexosaminidase B (HEXB), a lysosomal enzyme, in accelerating HCC growth by facilitating EV release under oxidative stress. ROS promotes transcription factor EB (TFEB) nuclear translocation, which activates the transcription of HEXB and its antisense lncRNA HEXB‐AS. Then, the HEXB protein binds HEXB‐AS, forming a protein/RNA complex that enhances the stability of HEXB. In addition, the stabilized HEXB interacts with lysosome‐associated membrane glycoprotein 1 (LAMP1), disrupting lysosome fusion with MVB. This event protects EVs from degradation, resulting in increasing EV secretion. Furthermore, combined use of the HEXB inhibitor M‐31850 and the EV release inhibitor GW4869 exhibited robust antitumor effects in HCC. In summary, our findings identify HEXB as a key regulator of EV release in HCC development.

2. MATERIALS AND METHODS

Cell culture. Human liver cancer cell lines Huh7, PLC/PRF/5, Hep 3B, HepG2, HepG2.2.15, HepAD38 and MHCC97H were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). Huh7, PLC/PRF/5, HepG2, HepG2.2.15, MHCC97H and HepAD38 were maintained in DMEM (Gibco). penicillin and streptomycin (100 U/mL) (Sigma, Kenilworth, NJ, USA) and 10% FBS (Biowest, Paris, France) were added to the culture medium. For HepG2.2.15, G418 was further supplemented (200 µg/mL) (Thermo Fisher Scientific, 11811023), and tetracycline (Sigma, 87128) was added to suppress HBV replication in the HepAD38 cell line (Xie et al., 2016). 24 mm Transwell with 0.4 µm pore polycarbonate membrane insert (Merck, CLS3412) was used for co‐culture assays.

Reagents and antibodies. GW4869 (HY‐19363), tetracycline (HY‐A0107), cycloheximide (CHX) (HY‐12320), MG132 (HY‐13259), Chloroquine (CQ) (HY‐17589A), M‐31850 (HY‐104050), sorafenib (HY‐10201), regorafenib (HY‐10331) and lenvatinib (HY‐10981) were obtained from MedChemExpress (Monmouth Junction, NJ, USA). H₂O₂ (18304, Sigma‐Aldrich) and NAC (A7250, Sigma‐Aldrich). Antibodies for LAMP1 (#9091, Rabbit mAb), CD63 (#52090, Rabbit mAb), TSG101 (#72312, Rabbit mAb), Calnexin (#2679, Rabbit mAb), TFEB (#37785, Rabbit mAb), cleaved caspase‐3 (#9661, Rabbit mAb), β‐actin (#3700, Mouse mAb), Histone H3 (#4499, Rabbit mAb), GAPDH (#5174, Rabbit mAb), were obtained from Cell Signalling Technology. Antibody for HEXB (16229‐1‐AP, Rabbit mAb) was purchased from Proteintech (Wuhan, China). Antibodies were unconjugated and diluted using TBST (western blotting: 1:1000; immunofluorescence and immunohistochemistry: 1:200). Goat anti‐Mouse IgG (H+L) Secondary Antibody, HRP (C31430100), Goat anti‐Rabbit IgG (H+L) Secondary Antibody, HRP (C31460100), Goat anti‐ Rabbit IgG (H+L) Cross‐Adsorbed Secondary Antibody, Alexa Fluor™ 488 (A‐11008) were obtained from ThermoFisher scientific.

Animal model. BALB/c nude mice (6‐week‐old, male) were purchased from HFK Bioscience Co., Ltd (Beijing, China). Mice were raised in the SPF Laboratory Animal Center of Sichuan University. 1 × 10⁷ PBS‐suspended cells were injected subcutaneously to establish the subcutaneous model. Ten days postinjection, mice received EVs from HEXB overexpressing cell culture medium by tail vein injection (50 µg per mouse, every 2 days) for another 10 days. For the orthotopic xenograft mouse model, 1 × 10⁶ PBS‐suspended cells were injected into the left liver lobe. After 1 week, mice were randomly grouped and received HEXB overexpressing cell‐derived EVs by tail vein injection (50 µg per mouse, every 2 days) for another 8 days. For the HCC patient‐derived xenografts (PDX) model, tumours obtained from HCC cancer patients (P0) (61 years old, male, tumour size > 5 cm, TNM stage: IIb) were cut into 9 mm³ pieces and implanted subcutaneously into NSG mice (F1) for engraftment. When the tumours grew to a suitable size, they were further cut into 27 mm³ pieces and subcutaneously transplanted into each mouse. Once reaching a mean group size of approximately 60 mm³, the mice were randomly assigned to groups and treated with the corresponding lentiviral vector intratumoural injections. Subsequently, the mice were administrated with HEXB overexpressing cell‐derived EVs by tail vein injection (50 µg per mouse, every 2 days) for 5 times.

The subcutaneous model, orthotopic xenograft mouse model and PDX were also used to assess the antitumour effect of M‐31850 and GW4869. When HCC cell or tissue was implanted, mice were randomly grouped and administered with vehicle, M‐31850 (0.2 mg/kg, intraperitoneal injection), GW4869 (2.5 mg/kg, intraperitoneal injection), sorafenib (50 mg/kg, oral gavage), regorafenib (10 mg/kg, oral gavage), lenvatinib (10 mg/kg, oral gavage) once daily. Each compound in combinational drug treatment was administrated at the same dose and scheduled as the single drug treatment. The subcutaneous tumour model was also used to analyze overall survival. Following animal ethics guidelines, mice were considered to have reached the endpoint when their subcutaneous tumour size exceeded 1000 mm³, at which point they were sacrificed. The study was terminated when all mice in the NC group had reached this endpoint. Then mice were sacrificed and tissues were excised and fixed in 4% PFA prior to undergoing immunohistochemistry staining. The research was approved by the Ethics Committee of Sichuan University.

Plasmid construction and stable cell line generation. HBV 1.3 plasmid was purchased from Addgene. Knockdown plasmids were generated by inserting specific oligonucleotides containing target gene sequences into pLKO.1 vector (Addgene). The cDNA of HEXB, CD63 were cloned into Flag‐tagged pCDH vector and mCherry‐tagged pCDH respective. To establish stable transfected cell lines, HEK293T cells were co‐transfected with pSPAX2 and pMD2.Gs. This co‐transfection enabled the production of lentivirus containing the genetic material from the indicated plasmids. After 48–72 h, supernatants were collected and indicated that HCC cells were infected with the lentivirus.

Isolation and validation of EVs. To collect EVs from the cell culture medium, cells were seeded at a density of 5 × 10⁷ cells per 150 mm plate. Subsequently, the cells were incubated with a medium supplemented with EV‐free FBS at a concentration of 10% for a minimum duration of 48 h. EV‐free FBS was procured by centrifuging FBS at 100,000 × g, 4°C, 18 h (Beckman Coulter, Optima XE‐90). The sequential ultracentrifugation method was used to isolate EVs from the cell culture medium. The collected cell culture medium underwent successive differential centrifugation steps at 300 × g, 4°C for 10 min, 2000 × g, 4°C for 10 min, and 10,000 × g, 4°C for 30 min. After filtering with a 0.22 µm syringe filter (Millipore), the supernatant was centrifuged at 100,000 × g, 4°C for 2 h using SW32 Ti rotor. The pellet obtained from the previous step was subjected to a wash with PBS and centrifuged at 100,000 × g, 4°C for 70 min. The isolated EV pellet was subsequently resuspended in PBS. Resuspended EVs were then purified by size‐exclusion chromatography (SEC). Briefly, Sepharose CL‐2B (Sigma, CL2B300) was packed in a SPE column to a final volume of 10 mL and equilibrated with PBS. Resuspended EVs were applied to the premade column, and the 5th, 6th, 7th and 8th fractions were collected for further detection or stored at −80°C until it was ready to be used.

To collect EVs from tumour tissue, 0.1 g of tissue was gently sliced using a sterile disposable scalpel. The tissue was dissociated until all pieces had a homogeneous size of approximately 2 × 2 × 2 mm. The tissue pieces were then incubated with 1 mL DMEM (containing 2 mg/mL DNase I [ThermoFisher, 18047019] and 40 U/mL collagenase D [Gibco, 17104019]) at 37°C for 30 min. The tissue pieces and DMEM medium were filtered using a 70 µm cell strainer (Corning, 431751), and the liquid was collected and used to isolate EVs. The subsequent steps are the same as the isolation of EVs from the cell culture medium.

The EV morphology and number were evaluated using a Hitachi HT7800 transmission electron microscope (Hitachi, Japan) and ZetaView nanoparticle tracking analysis (NTA) (Particle Metrix, German). The protein lysates of the EVs were probed using specific antibodies targeting TSG101 and CD63. An equal number of cells were used to extract the EVs, ensuring consistency. The intensity of EV markers detected in the western blot reflects the quantitative changes in the secretion of EVs (Ji et al., 2022). For visualizing recipient cell uptake EVs, mCherry‐CD63 labelled EVs (5 µg) were added to the medium and incubated with cells for 6 h at 37°C. Then the recipient cells were fixed with 4% PFA and visualized with confocal laser scanning microscopy (Carl Zeiss Microimaging) in the Pub‐lab of West China School of Basic Medical Sciences & Forensic Medicine, Sichuan University.

Cell viability and proliferation determination. The MTT assay was performed by plating 3000 cells per well in 96‐well plates. After adding MTT (5%) and culturing for 3 h, the cells were counted. The absorbance at 570 nm was measured for each well. In 12‐well plates, 1000 cells per well were seeded and treated for 2 weeks for the colony formation assay. After this, 4% paraformaldehyde (PFA) fixed cells were stained with crystal violet. Subsequently, images of the stained cells were captured using a Molecular Imager Gel Do XR+ System (BIO‐RAD, Hercules, CA, USA). For EdU assay, cells were plated in 96‐well plates with 3000 cells per well. To label all cells, a concentration of 50 µM EdU was used and the labelling process lasted for 2 h. Subsequently, fixation of the cells was carried out using 4% PFA. The images were captured using a fluorescence microscope (ZEISS, Oberkochen, Germany), and the quantification of the captured images was conducted following the instructions provided by the manufacturer.

RNA immunoprecipitation (RIP). The RIP assay was performed as the previous study (Zhou et al., 2022). HEXB antibody was incubated with whole cell lysate (supplemented with RNase inhibitors) for 4 h at 4°C. Then, an overnight incubation at 4°C was performed with prewashed protein A/G magnetic beads. After incubation with proteinase K, RNA was extracted using the phenol‐chloroform method and analyzed by qPCR.

RNA pulldown assay. The RNA pulldown assay was performed as the previous study (Zhou et al., 2022). Briefly, 10× Biotin RNA labelling mix (Roche, 1165597910) and T7 enzyme mix (New England Biolabs, M0251S) were used to transcribe the biotin‐labelled RNAs. The RNAs underwent a heat treatment at 65°C for 5 min and were then cooled to room temperature. The RNAs labelled with biotin were then incubated with magnetic beads coated with streptavidin (Beyotime Biotechnology, P2151) for 30 min. The whole cell lysate was added to the RNA‐beads complex and mixed for 1 h at 4°C. Subsequently, the pulldown complexes were washed and boiled at 100°C for 5 min, followed by immunoblotting.

Immunoblotting. Cells were collected and rinsed twice with chilled PBS. The cell pellet was then incubated with lysis buffer for 30 min on ice. The proteins were transferred to a PVDF membrane (0.2 µm pore size, Millipore, ISEQ85R) after reduced SDS‐PAGE. 5% non‐fat milk was used to block the membrane and blocked membrane were probed with specific antibodies.

β‐Hexosaminidase activity assay. The activity of β‐hexosaminidase activity was detected as the previous study (Plesch et al., 2018). Cell lysate was incubated with natrium citrate buffer (pH 4.5) and 4‐Methylumbelliferyl N‐acetyl‐β‐D‐glucosaminide (M1233, Sigma, 1 mM) for 20 min. Glycine buffer was used to stop the reaction and the activity of β‐hexosaminidase activity was detected using a plate reader (Excitation: 365 nm; Emission: 450 nm) (Tecan Spark M20).

RNA interference and quantification. Stable knockdown and overexpressed HCC cells were generated using pLKO.1 and pCDH vector, respectively. Here are the sequences of the primers for shRNA and overexpression: HEXB shRNA #1, F 5′‐CCGGGCCGGGCACAATAGTTGAAGTCTCGAGACTTCAACTATTGTGCCCGGCTTTTTG‐3′, R 5′‐AATTCAAAAAGCCGGGCACAATAGTTGAAGTCTCGAG ACTTCAACTATTGTGCCCGGC‐3′; HEXB shRNA #2, F 5′‐CCGG GCATCTGGCTTCCCTGTAATCCTCGAGGATTACAGGGAAGCCAGATGCTTTTTG‐3′, R 5′‐ AATTCAAAAAGCATCTGGCTTCCCTGTAATCCTCGAGGATTACAGGGAAGCCAGATGC‐3′; HEXB‐AS shRNA #1 F 5′‐CCGGGCCGAGATTGTACTACTGCACCTCGAG GTGCAGTAGTACAATCTCGGCTTTTTG‐3′, R 5′‐AATTCAAAAA GCCGAGATTGTACTACTGCACCTCGAGGTGCAGTAGTACAATCTCGGC‐3′; HEXB‐AS shRNA #2 F 5′‐CCGGGGTGACAGAACAAGACTCCATCTCGAGATGGAGTCTTGTTCTGTCACCTTTTTG‐3′, R 5′‐AATTCAAAAAGGTGACAGAACAAGACTCCATCTCGAG ATGGAGTCTTGTTCTGTCACC‐3′; HEXB overexpressing primer, F 5′‐ GGCTAGCAAAGCAGCCGAGCGGCCAT‐3′, R 5′‐ TTGCGGCCGCGCTGTGGCCTTTTTCCCCTCCA‐3′.The siRNA of TFEB, LAMP1 and miR‐21‐5p inhibitor were obtained from GenePharma (Shanghai, China). The sequence of siTFEB is as follows: Sense 5′‐ GAAAGACAAUCACAACUUAAU‐3′, Antisense 5′‐ UAAGUUGUGAUUGUCUUUCUU‐3′. The sequence of siLAMP1 is as follows: Sense 5′‐CGAGAAAUGCAACACGUUACA‐3′, Antisense 5′‐UAACGUGUUGCAUUUCUCGUG‐3′. The sequence of miR‐21‐5p inhibitor is as follows: 5′‐UCAACAUCAGUCUGAUAAGCUA‐3′. For quantitative RT‐qPCR assay, total RNA was extracted using TRIzol (Thermo Fisher Scientific, 15596018) and reverse transcribed using the RT kit (Takara, RR047A, Mountain View, CA, USA).

Immunofluorescence. Cells seeded on coverslips were fixed by 4% PFA and then permeabilized with 0.2% Triton X‐100 and blocked with 5% BSA. The coverslips were then incubated with corresponding primary antibodies at 4°C overnight. We then stained nuclei using DAPI for 10 min after incubating with secondary antibodies for 1 h. Confocal laser scanning microscopy was used to capture these images.

Immunohistochemistry. Tumour xenografts were fixed by 4% PFA, embedded in paraffin and then sliced. Sections were retrieved using citrate buffer, blocked with FBS for 1 h at room temperature, and incubated overnight with primary antibodies at 4°C. Following washing in PBS, secondary antibodies were incubated at room temperature for 30 min. EnVision Detection System (Agilent Technologies, K5007) was used to measure antigen levels after staining with DAB peroxidase substrate.

Fluorescence in situ hybridization assay. miR‐21 levels in tissues were detected by using RNA fluorescence in situ hybridization (FISH) assay kit (GenePharma), the procedure in the kit instruction manual. Briefly, the rehydrated tissue sections and then incubated with proteinase K (20 µg/mL) for 30 min. Then tissue sections were hybridized with Cy5 labelled miR‐21 oligonucleotide probe (1 µM) at 37°C overnight after incubation in prehybridization buffer for 1 h at 37°C. After three washes in 1×SSC, the nucleus was stained with DAPI.

Statistical analysis. All data are obtained from at least three independent experiments and presented as mean ± s.d. and. Two‐tailed Student t‐tests and one‐way ANOVA were used to analyze data in different groups. The p value was calculated using GraphPad 8 and was considered statistically significant if p < 0.05.

3. RESULTS

3.1. ROS‐induced EV release is mediated by the lysosomal protein HEXB

The development of liver cancer is linked to increased oxidative stress (Huang et al., 2024; Liu et al., 2020; Rebouissou & Nault, 2020). For example, both virus infection and alcohol can induce oxidative stress and then promote EV release to contribute to the development of liver cancer (Gallard et al., 2022; Ma et al., 2020b). In addition, in response to mitochondrial stress, adipocytes release extra small EVs which can trigger a burst of ROS and enhance fibrotic signalling in hepatocytes (Crewe et al., 2021; Pan et al., 2019). These studies indicate that EVs may play a role in the progression of HCC induced by oxidative stress. To investigate the key factors involved in oxidative stress promoting EV release, we first evaluated secreted EV concentration under different ROS levels. Our study utilized hydrogen peroxide (H₂O₂) treated cells, a CoCl₂ hypoxia model (Lan et al., 2012), and a stable HBV replication model (HepG2.2.15 cell line) (Yuan et al., 2016) as high ROS conditions, while the ROS scavenger N‐acetyl cysteine (NAC) was used to abolish increased ROS (Figure S1a‐c). Nanoparticle tracking analysis (NTA) (Figure 1a‐c) and changes in EV markers (Figure 1d‐f) revealed that oxidative stress promotes EV release in HCC cells. To identify potential mediators of the ROS‐induced increase in EV secretion, various datasets, including extracellular exosome‐related proteins (GO: 0070062), lysosomal genes (KEGG_LYSOSOME gene set), differentially expressed genes in H₂O₂ treated HCC cells (GSE47739) (Aravinthan et al., 2014) and liver diseases‐related genes (CTD Gene‐Disease Associations) (Davis et al., 2023) were employed to screen candidate molecules (Figure 1g). The overlap of these four datasets included 24 candidate genes, which were ranked according to their expression levels in HCC patients using The Cancer Genome Atlas (TCGA) dataset (Figure 1h) (Cerami et al., 2012). The relevance between the expression of the top five candidate genes and patient survival was analyzed using The Human Protein Atlas. The results showed that HEXB is the most significant gene simultaneously associated with the exosome and lysosome function, oxidative stress responsiveness, liver diseases, and patient survival (Figures 1i and S1d‐g). In addition, bioinformatics analysis revealed that HEXB is highly expressed in HCC (Figure 1j) (Rhodes et al., 2004). Therefore, HEXB might be an EV‐related protein involved in HCC progression.

FIGURE 1.

ROS‐induced EV release is mediated by the lysosomal protein HEXB. (a–c) NTA of EVs isolated from corresponding treatment group (H₂O₂ (10 µM, 12 h), CoCl₂ (100 µM, 24 h), and NAC (1 mM, 12 h)). EVs were imaged by TEM (scale bar, 100 nm). (d–f) Western blotting analysis of EVs extracted from equal numbers of different treatment group cells. (g) Venn diagram showing overlap of genes among four datasets obtained from GO:0070062, KEGG_LYSOSOME, GSE47739, and CTD Gene‐Disease Association (liver diseases). (h) Heatmap showing 24 candidate genes mRNA levels in HCC tissues according to TCGA data set. (i) Overall survival of HCC patients according to The Human Protein Atlas. (j) mRNA levels of HEXB from GEPIA TCGA liver consisting of normal liver (n = 160) and HCC tissues (n = 369). (k–n) Correlation analysis of relative protein levels of HEXB and VAMP7, SNAP29, RAB5A, and VPS33A in NIH Proteomic Data Common (PDC000198, HBV‐Related Hepatocellular Carcinoma–Proteome). (o, p) Western blotting analysis was performed to detect the protein levels of HEXB in Huh7 and PLC/PRF/5 cells with/without HEXB stable knock‐down (o) or overexpression (p). (q) NTA of EV isolated from corresponding treatment group. EV was examined by TEM (scale bar, 100 nm). (r) Western blotting analysis of EVs extracted from equal numbers of different treatment group cells. (* p < 0.05).

To further explore the role of HEXB in ROS‐induced EV release in HCC, we first analyzed the relevance between HEXB and several exosome secretion‐related genes (Yang et al., 2019). The proteomic data from NIH Proteomic Data Common (PDC000198, HBV‐ Related Hepatocellular Carcinoma—Proteome) showed that the protein levels of VAMP7, SNAP29, RAB5A, VPS33A, RAB2B and RALB positively correlated with that of HEXB (Figures 1k‐n and S1h‐i). We then evaluated the effect of HEXB on EV secretion. Endogenous HEXB was stably knocked down by short hairpin RNA (shRNA), and pCDH‐Flag‐HEXB was used to construct a HEXB stable overexpressing cell line (referred to as HEXB OE) (Figures 1o,p and S1j). NTA and the intensity of EV markers indicated that loss of HEXB decreased the secretion of EVs in vitro (Figures 1q,r and S1k,l) and in the tumor tissue of HCC PDX model (Figure S1m,n). Moreover, HEXB knockdown inhibited the EV release promoted by H₂O₂ treatment (Figures 1q,r and S1k,l). These results suggested that HEXB mediates EV release in HCC cells in response to oxidative stress.

3.2. HEXB facilitates HCC cell growth by promoting EV release

To ascertain the function of HEXB‐mediated EV release in HCC progression, we examined the effect of HEXB on HCC cell proliferation and demonstrated that loss of HEXB attenuated HCC cell growth (Figures 2a‐d and S2a‐d). Consistent with the in vitro results, the subcutaneous xenograft model (Figure S3a‐h), orthotopic xenograft mouse model (Figures 2i‐l and S3h‐j) and HCC PDX model (Figures 2m‐p and S3m‐p) also showed that HEXB knockdown significantly decreased tumour burden and the positive rate of the proliferation marker Ki67, while increased the level of cleaved caspase‐3 in xenograft tissues. These results indicated that HEXB facilitates HCC cell growth both in vitro and in vivo.

FIGURE 2.

HEXB facilitates HCC cell growth by promoting EV release. (a, b) The growth of Huh7 and PLC/PRF/5 cells was detected by MTT assay over a 96‐h period with or without HEXB stable knock‐down. (c) Colony formation assays were used to evaluate the effects of HEXB knockdown on the proliferation of Huh7 and PLC/PRF/5 cells. (d) The proliferation rates of Huh7 and PLC/PRF/5 cells were detected by EdU assays. (e) The effects of HEXB OE‐EV on the growth of HEXB stable knockdown cells were detected by MTT assays. (f) Schematic diagram of co‐culture assay. HEXB OE cells and HEXB knockdown cells were co‐cultured by an indirect method using a transwell system. HEXB OE cells pretreated with/without GW4869 (10 µM, 4 h) were cultured in the upper chamber, and HEXB knockdown cells in the lower chamber to co‐cultured for 7 days. (g) Western blotting analysis was performed to detect the protein levels of Rab27a in Huh7 and PLC/PRF/5 cells with/without Rab27a siRNA treatment. (h) After 7 days of treatment, crystal violet staining was used to detect the proliferation of HEXB knockdown cells in the lower chamber. (i) Representative images of tissues from an orthotopic liver tumour model (dashed circles indicate orthotopic tumours). (j) Representative HE staining images of the livers in I (scale bar: 0.5 cm). (k) Percentage of tumour tissue to total area. (l) Representative images of HEXB and cleaved caspase‐3 staining of the orthotopic tumour (scale bar: above 50 µm). (m) The schematic illustrates the process of establishing a PDX model to provide patient information. (n) Image of isolated tumours from the PDX model with/without HEXB knockdown and EV treatment. (o) Tumour volume in PDX model at indicated time point. (p) Representative images of HEXB, cleaved caspase‐3, PTEN from IHC staining and miR‐21 from immunofluorescence in the PDX tumour (dashed lines mark miR‐21 expression and PTEN unexpressed regions) (scale bar: 50 µm). (** p < 0.01 and *** p < 0.001).

Given the notion that HEXB mediated EV secretion in HCC cells, we further examined whether HEXB‐induced HCC cell proliferation was mediated by released EVs. EVs from cultured supernatants of HEXB overexpressing cells were extracted and used to stimulate HEXB knockdown cells. We found that EVs from HEXB overexpressing cells can be actively taken up by HEXB knockdown cells (Figure S2e), and incubating with these EVs can almost completely rescue the growth defect induced by HEXB knockdown (Figure 2e). Co‐culture experiments using HEXB knockdown cells and HEXB overexpressing cells were performed with or without GW4869, an exosome‐release inhibitor, or Rab27a siRNA to suppress EV secretion (Figures 2f,g and S2f‐i). While HEXB knockdown cells co‐cultured with HEXB overexpressing cells showed a significantly increased proliferation, blocking EV secretion from HEXB overexpressing cells significantly reduced this proliferative benefit observed in HEXB knockdown cells (Figures 2h and S2j). To further validate the tumour‐promoting effects of HEXB‐mediated EV release in vivo, the subcutaneous xenograft model (Figure S3a‐h), orthotopic xenograft mouse model (Figures 2i‐l and S3h‐j) and HCC PDX model were conducted (Figures 2m‐p and S3m‐p) and received EVs from HEXB overexpressing cell culture medium by tail vein injection (50 µg per mouse, every 2 days). The results showed that HEXB overexpressing cell‐derived EVs could rescue the tumorigenic capacity of HEXB knockdown cells. Taken together, these results suggest that HEXB promotes HCC cell growth by facilitating EV secretion.

MiRNAs are important for intercellular communication in EVs and are considered useful biomarkers for cancer diagnosis and prognosis (Preethi et al., 2022). To further investigate the mechanism by which HEXB overexpressing cell‐derived EVs promote HCC cell growth, EVs isolated from Huh7 and Huh7 HEXB overexpressing cells were subjected to miRNA sequencing. Compared with Huh7 NC cell‐derived EVs, a total of 28 differentially expressed miRNAs were identified (FDR p‐value < 0.01), including 14 up‐regulated and 14 down‐regulated miRNAs in HEXB overexpressing EVs (Figure S4a). To screen for miRNAs most closely associated with HCC progression, datasets examining exosomal miRNA in liver disease (GSE85677) and miRNAs involved in liver carcinogenesis (GSE147889) were used to screen the candidate miRNAs. The results showed that miR‐21‐5p and miR‐100‐5p were upregulated candidates (Figure S4b). Subsequently, the levels of candidate miRNAs and several HCC‐related exosomal miRNAs in HEXB overexpressing EVs were detected, and the results showed both miR‐21‐5p and miR‐100‐5p were significantly increased in HEXB overexpressing EVs (Figure S4c). Of the two miRNAs, miRNA‐21‐5p exhibited a stronger correlation with the expression of HEXB in HCC patients (Figure S4d,e). It has been demonstrated that the exosomal miR‐21 induced by acidic microenvironment promotes both the proliferation and metastasis of HCC cells (Tian et al., 2019). To determine whether exosomal miRNA‐21‐5p played a role in HEXB‐induced HCC cell proliferation, we established co‐culture systems of HEXB knockdown cells and HEXB overexpressing cells that were pretreated with/without miR‐21 inhibitor (Figure S4f,g). The results demonstrated that knockdown miR‐21 significantly abolished HEXB overexpression‐induced HCC cell proliferation (Figure S4h). We also demonstrated that miR‐21 knockdown EVs rarely promoted HCC cell growth even if they were derived from HEXB overexpressing cells (Figure S4i‐l).

Previous studies have indicated that the exosomal miR‐21 can regulate the expression of PTEN in multiple ways and promote HCC cell growth (Cao et al., 2019). To further verify the targets of exosomal miR‐21 under oxidative stress, we detected the level of miR‐21 and PTEN in the PDX model (Figures 2p and S3q,r) and cell (Figure S4m) treated with HEXB‐OE cell‐derived EVs. The results showed that HEXB‐OE cell‐derived EVs can significantly increase miR‐21 level in HCC cell and tissue and decrease PTEN expression, while knockdown miR‐21 in EVs can reverse EV‐induced PTEN decrease, indicating that miR‐21 in HEXB‐OE cell‐derived EVs may promote HCC cell proliferation by decreasing PTEN.

3.3. ROS promote HEXB expression and stability

Based on the observation that HEXB is a differentially expressed gene in H₂O₂‐treated HCC cells (GSE47739) and HEXB mediates ROS‐induced EV secretion, we wondered if ROS stimulates the upregulation of HEXB in HCC cells. H₂O₂ treatment, CoCl₂‐induced hypoxia model, and HBV replication models (HBV 1.3 plasmid transfected cell, HepG2.2.15 cell line, and HepAD38 cell line [Figure S5a‐f]) were used to detect the level of HEXB under oxidative stress condition. NAC was utilized to abolish increased ROS, and tetracycline was used to inhibit HBV replication in HepAD38 cells, thereby reducing ROS generation (Figure S5g‐i). The results indicated that the protein level of HEXB was increased in a ROS‐dependent manner (Figure 3a‐e). To ascertain how ROS induces high levels of HEXB in HCC cells, we detected HEXB mRNA levels and protein stability in H₂O₂‐treated HCC cells. The results showed that H₂O₂ treatment could not only increase the transcription of HEXB (Figure 3f), but also enhanced the protein stability of HEXB when HCC cells were treated with cycloheximide (CHX) (Figure 3g).

FIGURE 3.

ROS promote HEXB expression and stability. (a) Western blotting showing the levels of HEXB in Huh7 and PLC/PRF/5 cells treated with corresponding concentrations of H₂O₂ (12 h). (b) Western blotting showing the levels of HEXB in the CoCl₂ hypoxia model treated with/without NAC (1 mM, 12 h). (c) The protein levels of HEXB in HBV replication cell model. Left, tetracycline was added to inhibit HBV replication in HepAD38 cells. Right, HepG2.2.15 was treated with NAC (1 mM, 12 h) to abolish ROS produced by HBV replication. (d) The protein levels of HEXB in HBV 1.3 plasmid transfected cells treated with/without NAC (1 mM, 12 h). (e)The protein levels of HEXB in Huh7 and PLC/PRF/5 cells treated with H₂O₂ (10 µM, 12 h) and NAC (1 mM, 12 h). (f) The mRNA levels of HEXB in Huh7 and PLC/PRF/5 cells treated with H₂O₂ (10 µM, 12 h) and NAC (1 mM, 12 h). (g) Cells were treated with cycloheximide (CHX) (100 µg/mL) for indicated time followed by detection of HEXB expression by immunoblotting. (h) Cell nucleus/cytoplasm fractionation and western blot analysis to show TFEB translocation. GAPDH and histone H3 were used as cytoplasmic and nuclear markers, respectively. (i) The subcellular localization of TFEB after transfecting HBV 1.3 plasmid with/without NAC treatment was displayed by immunofluorescence (scale bar, 5 µm). (j) The mRNA levels of TFEB after transfecting TFEB siRNA. (k) Western blotting analysis showed the protein levels of TFEB after silencing TFEB. (l, m) The protein levels (l) and mRNA levels (m) of HEXB silencing cells treated with/without H₂O₂. (** p < 0.01 and *** p < 0.001).

Previous reports have identified that HEXB is one of the target genes of TFEB, and ROS can lead to the nuclear translocation of TFEB (Fang et al., 2021; Ivankovic et al., 2016). We aimed to verify whether ROS promotes HEXB expression via inducing TFEB nuclear translocation in HCC cells. In H₂O₂‐treated HCC cells, the nuclei‐cytoplasm fractionation assay showed consistent results (Figure 3h), and immunofluorescence staining showed the increased translocation of TFEB to the nucleus, while NAC treatment attenuated TFEB nuclear translocation in HBV 1.3 transfected cells (Figure 3i). To ascertain whether ROS regulates HEXB expression via TFEB nuclear translocation, we detected decreased mRNA and protein levels of HEXB in TFEB knocked‐down cells (Figure 3j,k) when treated with/without H₂O₂ (Figure 3l,m). Collectively, these results suggest that increased ROS can promote HEXB expression by inducing TFEB nuclear translocation, while also enhancing the stability of HEXB protein, resulting in an increase of HEXB in HCC cells.

3.4. HEXB‐AS maintains the protein stability of HEXB

To elucidate the mechanism underlying enhanced HEXB protein stability, we noticed that a gene encoding the lncRNA HEXB antisense RNA (HEXB‐AS) is located on the negative strand of the HEXB gene on chromosome 5 (Figure S6a). The nuclear/cytoplasmic RNA fractionation assay showed that HEXB‐AS is located in both the nucleus and cytoplasm (Figure S6b,c). This suggests that HEXB‐AS may play a role in regulating the expression and function of HEXB. Given the notion that most antisense lncRNAs perform their function via regulating the transcription of neighbouring genes, we first examined whether HEXB‐AS is involved in ROS‐induced HEXB expression. The results showed that oxidative stress could increase HEXB‐AS level and NAC can abolish this increase (Figure 4a). In addition, knockdown of TFEB inhibited HEXB‐AS expression (Figure 4b). Interestingly, knockdown of HEXB‐AS (Figure S6d) decreased the protein level of HEXB (Figure 4c), indicating that HEXB‐AS also contributed to the high level of HEXB. However, we found that knocked‐down HEXB‐AS did not impinge on HEXB mRNA level (Figure 4d). These results suggest that HEXB‐AS promotes overexpression of HEXB via enhancing protein stability, which was validated by the CHX assay (Figure 4e). By treating with the proteasome inhibitor MG132 or the lysosome inhibitor CQ, we also demonstrated that HEXB is degraded via lysosomal protein hydrolysis (Figure 4f,g). Together, these results suggested that ROS‐induced TFEB nuclear translocation activates the transcription of both HEXB mRNA and HEXB‐AS. A high level of HEXB mRNA directly leads to HEXB protein overexpression, while HEXB‐AS enhances HEXB protein stability.

FIGURE 4.

HEXB‐AS maintains the protein stability of HEXB. (a) qPCR showing the levels of HEXB‐AS in cells treated with H₂O₂ (10 µM, 12 h) and NAC (1 mM, 12 h). (b) The levels of HEXB‐AS in cells transfected with TFEB siRNA. (c) Effects of HEXB‐AS knockdown or overexpression on the protein levels of HEXB. (d) qPCR was performed to determine the relative mRNA levels of HEXB in HEXB‐AS knockdown cells. (e) Cells were treated with CHX (100 µg/mL) for indicated time followed by detection of HEXB expression by immunoblotting. (f) Western blotting showing the protein degradation of HEXB by using CQ (10 µM, 24 h). (g) Western blotting showing the protein degradation of HEXB in HEXB‐AS stable knockdown cells using MG132 (10 µM, 24 h). (h) RIP analysis of the binding of HEXB to HEXB‐AS. (i) The binding of indicated truncated HEXB to HEXB‐AS in Huh7 cell was measured by RIP. (j) PRIdictor was used to predict the binding sites of HEXB‐AS interacting with HEXB. (k) The binding of HEXB to indicated truncated HEXB‐AS in Huh7 cell was measured by RNA pulldown. (l) Cells were treated with CHX (100 µg/mL) for indicated time followed by the detection of HEXB expression by immunoblotting. (*** p < 0.001, NS, not significant).

To investigate the mechanism underlying HEXB‐AS‐induced HEXB protein stabilization, an RNA immunoprecipitation (RIP) assay was performed to determine whether HEXB‐AS and HEXB interact. The results showed that HEXB could bind to HEXB‐AS directly (Figure 4h). To map the fragment of HEXB responsible for binding to HEXB‐AS, we generated truncated mutants of HEXB (122‐556 aa (M1), 122–311 aa (M2), and 315–556 aa (M3)). Further, RIP assays revealed that HEXB chain A (315‐556 aa (M3)) was responsible for interacting with HEXB‐AS (Figure 4i). To further verify the interaction, we predicted the potential binding site of HEXB‐AS using PRIdictor (Tuvshinjargal et al., 2016), and the results demonstrated that there are several binding sites in 150–600 bp of HEXB‐AS (Figure 4j). We then generated truncated mutants of HEXB‐AS (162‐705 bp (△1), and 740–1073 bp (△2)) to identify the binding fragment via RNA pulldown assay. The results demonstrated that fragment △1 (162–705 bp) of HEXB‐AS is critical for the binding of HEXB‐AS with HEXB (Figure 4k).

To ascertain the effect of the interaction between HEXB and HEXB‐AS on HEXB protein stability, we detected HEXB degradation rates in HEXB‐AS knockdown cells with or without fragment △1 overexpression (Figure 4l). The results suggest that overexpressing fragment △1 can prolong the half‐life of HEXB protein in HEXB‐AS knockdown cells. Taken together, these data indicate that in addition to promoting HEXB mRNA transcription, ROS‐mediated TFEB nuclear translocation can also induce HEXB‐AS expression, and HEXB‐AS can bind with HEXB to maintain protein stability.

3.5. HEXB‐AS‐mediated HEXB protein stabilization is required for EV release

After identifying the effect of HEXB‐AS on HEXB protein stability, we investigated whether HEXB‐AS is involved in HEXB‐induced HCC cell growth. We found that the knockdown of HEXB‐AS expression could attenuate HCC cell proliferation (Figure S6e‐l). Moreover, overexpression of HEXB significantly rescued the proliferation of HEXB‐AS knockdown HCC cells, while overexpression of HEXB‐AS failed to rescue the proliferation of HEXB knockdown HCC cells (Figure 5a‐d). In addition, overexpression of a truncated mutant fragment △1 of HEXB‐AS could also rescue the proliferation in HEXB‐AS knockdown HCC cells (Figure S6m‐o). These results showed that HEXB is accountable for HEXB‐AS‐mediated HCC cell growth.

FIGURE 5.

HEXB‐AS‐mediated HEXB protein stabilization is required for EV release. (a) The growth of HEXB knockdown cells with or without HEXB‐AS overexpression was detected by MTT assay. (b) The growth of HEXB‐AS knockdown cells with or without HEXB overexpression was detected by MTT assay. (c) The proliferation rates of HEXB‐AS knockdown cells with or without HEXB overexpression were detected by EdU assays. (d) Colony formation assays were used to evaluate the effect of HEXB overexpression on the proliferation of Huh7 and PLC/PRF/5 HEXB‐AS knockdown cells. (e–h) NTA of EVs isolated from corresponding treatment group (e, g) and western blotting analysis of EVs extracted from equal numbers of different treatment group cells (f, h). (i) The effects of HEXB OE‐EV on the growth of HEXB‐AS stable knockdown cells were detected by MTT assay. (j) Schematic diagram of co‐culture assay. HEXB OE cells and HEXB‐AS knockdown cells were co‐cultured by an indirect method using a transwell system. HEXB OE cells pretreated with/without GW4869 (10 µM, 4 h) were cultured in the upper chamber, and HEXB‐AS knockdown cells in the lower chamber to co‐cultured for 7 days. (k) After 7 days of treatment, crystal violet staining was used to detect the proliferation of HEXB‐AS knockdown cells in the lower chamber. (* p < 0.05, ** p < 0.01 and *** p < 0.001).

To explore whether HEXB‐AS was also involved in EV secretion, we performed NTA and assessed EV markers. Knockdown of HEXB‐AS inhibited EV release while replenishment of HEXB could rescue the secretion of EVs in HEXB‐AS knockdown cells (Figure 5e‐h). To further verify that HEXB‐AS knockdown‐induced proliferation inhibition is attributed to HEXB‐mediated EV secretion decrease, EV from cultured supernatants of HEXB overexpressing cells were provided to HEXB‐AS knockdown cells. As shown in Figure S7a, EVs from HEXB overexpressing cells can be actively taken up by HEXB‐AS knockdown cells, and the proliferation of HCC cells was significantly increased after EVs stimulation (Figure 5i). Co‐culture experiments were performed (Figure 5j) and the results showed that HEXB‐AS knockdown cells co‐cultured with HEXB overexpressing cells exhibited significantly increased proliferation, while blocking EV secretion from HEXB overexpressing cells significantly reduced the proliferative benefit observed in HEXB‐AS knockdown cells (Figures 5k and S7b,c). Together, these data indicated that HEXB‐AS regulates HEXB‐mediated EV release, thus promoting HCC cell growth.

3.6. HEXB attenuates MVB‐lysosome fusion to increase EV release

Given the demonstrated role of HEXB in regulating EV release, we attempted to investigate the mechanistic basis for the impact of HEXB on EV secretion. Given the fact that HEXB is located in the lysosomal lumen, we speculated that HEXB may interact with some EV release‐related proteins. The potential binding proteins of HEXB were co‐immunoprecipitated and subjected to mass spectrometric analysis. We also analyzed potential HEXB binding protein using the OpenCell database (Cho et al., 2022). Considering that HEXB is regulated by ROS, GSE47739 was used to screen the potential binding protein of HEXB. Only LAMP1 was found in the overlap of the three datasets (Figure 6a). In addition, co‐IP experiments were conducted to confirm the interaction between HEXB and LAMP1 in HCC cells (Figure 6b).

FIGURE 6.

HEXB attenuates MVB‐lysosome fusion to increase EV release. (a) Venn diagram showing overlap of protein among three datasets obtained from HEXB interacting proteome, Opencell‐interaction HEXB and GSE47739. (b) Co‐IP assays indicate the interaction between HEXB and LAMP1. (c) Western blotting showing the protein levels of LAMP1 in HEXB knockdown and overexpressing cells. (d) Western blotting showing the protein levels of LMAP1 in HEXB knockdown cells treated with/without H₂O₂ (10 µm, 12 h). (e) Cells were treated with CHX (100 µg/mL) for an indicated time followed by detection of HEXB expression by immunoblotting. (f) Western blotting analysis was performed to detect the protein levels of LAMP1 in cells treated with/without LAMP1 siRNA silencing. (g, h) Colocalization of MVB and lysosome was determined by immunofluorescence assays (scale bar, 20 µm). (i–l) NTA of EVs isolated from cell supernatants of corresponding treatment group (i, k) and western blotting analysis of EVs extracted from equal numbers of different treatment group cells (j, l).

To investigate the regulatory effect of HEXB on LAMP1, we detected the levels of LAMP1 in HEXB knockdown and overexpressing cells. As shown in Figure 6c, the expression of LAMP1 was increased in HEXB knockdown cell and decreased in HEXB overexpressing cell. In addition, HEXB knockdown disrupted H₂O₂‐induced LAMP1 decrease (Figure 6d), and CHX treatment showed that HEXB knockdown accelerated LAMP1 protein degradation (Figure 6e). Previous studies have indicated that dysfunction of the lysosome can alter the fate of MVBs that are intended for degradation, leading them to fuse with the plasma membrane, thus giving rise to an increase in EV secretion (Latifkar et al., 2019). To investigate whether HEXB‐induced LAMP1 degradation could disrupt the fusion between MVB and lysosome, we performed immunofluorescence to label MVBs with CD63 and used lysotracker to label functional lysosomes. The results showed that the knockdown of HEXB can increase the co‐localization of MVB and lysosome, and the knockdown of LAMP1 in HEXB knockdown cells could decrease MVB‐lysosome co‐localization (Figure 6f‐h). Consistently, the knockdown of LAMP1 rescued EV secretion in HEXB knockdown cells (Figure 6i‐l). Moreover, we also verified that HEXB‐AS prevented MVB‐lysosome fusion in a way similar to HEXB (Figure S8a). Taken together, our data suggest that high‐level HEXB inhibits lysosomal function by interacting with LAMP1, preventing MVB‐lysosome fusion. This event protects EVs from degradation, leading to an increase in EV secretion.

3.7. HEXB inhibitor enhances the efficacy of GW4869 in HCC cells

To investigate the potential clinical benefits of targeting HEXB to inhibit tumour growth, HCC cells were treated with M‐31850, a potent and selective inhibitor of β‐hexosaminidase. As demonstrated in the results, M‐31850 could inhibit the growth of HCC cells with efficacy and low toxicity on liver immortalized epithelial cells, THLE‐3 (Figures 7a and S9a), and M‐31850 (5 µM) significantly decrease the activity of β‐hexosaminidase (Figure 7c) and disrupted the interaction between HEXB and LAMP1 (Figure 7d), ultimately leading to a decrease in EV release (Figure 7e‐h).

FIGURE 7.

HEXB inhibitor enhances the efficacy of GW4869 in HCC cells. (a, b) MTT assay showing the viability of cells treated with indicated concentration of M‐31850 (a) and GW4869 (b) for 24 h. (c) HEXB activity in cells treated with M‐31850. (d) Co‐IP assays indicate the interaction between HEXB and LAMP1 in cells treated with M‐31850. (e–h) NTA of EVs isolated from corresponding treatment group (e, g) and western blotting analysis of EVs extracted from equal numbers of different treatment group cells (f, h). (i) Heatmaps showing cell viability of Huh7, PLC/PRF/5 and THLE‐3 treated with indicated concentration of M‐31850 and GW4869 for 24 h. (j) The schematic diagram illustrates the process of establishing a PDX model and providing patient information. (k) Image of isolated tumours from the PDX model treated with M‐31850, GW4869 or three targeted agents (sorafenib, lenvatinib, regorafenib). (l) Tumour volume in PDX model at indicated time point. (m) The relative positive rate for Ki67. (n) Representative images of cleaved caspase‐3 staining of the PDX tumour (scale bar: above: 50 µm) and the average optical density of cleaved caspase‐3. (*** p < 0.001).

Considering the distinct mechanisms of action of M‐31850 and GW4869 on EV release, we were curious about the potential enhanced synergistic effects that could arise from their combination. HCC cells were subjected to a combined treatment of M‐31850 and GW4869 at their half inhibitory concentration (Figures 7a,b, and S9a). As shown in Figure 7e‐h, the combination of M‐31850 and GW4869 further suppressed EV secretion. MTT assays revealed that M‐31850 enhanced the inhibitory efficacy of GW4869 in HCC cells and the toxicity of M‐31850 and GW4869 on THLE‐3 is much less than in HCC cells (Figure 7i). To further evaluate the synergistic effect of M‐31850 and GW4869 in vivo, subcutaneous xenograft model (Figure S9b‐g), orthotopic xenograft mouse model (Figure S9h‐m) and HCC PDX model (Figure 7j‐n) were administered with GW4869 alone or in combination with M‐31850 or three target agents (sorafenib, lenvatinib, regorafenib). The results of tumour growth, relative positive rate of Ki67, and level of cleaved caspase‐3 showed that M‐31850 synergized with GW4869 in HCC treatment, and mice following the combination of M‐31850 and GW4869 have similar anti‐tumour effect as those treated with sorafenib, lenvatinib and regorafenib. In addition, the combination treatment of M‐31850 and GW4869 demonstrated no noticeable adverse effects in mice (Figure S9n,o). These findings highlight the potential therapeutic value of such a combined approach in HCC treatment.

4. DISCUSSION

Existing studies have revealed that the progression of HCC is a stepwise process driven by multiple cell‐intrinsic genetic changes, such as TP53, YAP and others (Shiraha et al., 2013; Zender et al., 2006). However, a more in‐depth exploration is required to elucidate the detailed mechanisms by which these cell‐intrinsic driver events contribute to HCC development. EVs have become widely recognized as significant regulators of various liver diseases, including HCC, and inhibiting the secretion of EVs holds immense significance for the treatment of these diseases. (Kostallari et al., 2021; Liu et al., 2021; Thietart & Rautou, 2020). Some studies have suggested the link between EVs and cell‐intrinsic drivers of HCC. For example, under stress conditions, the p53 activation enhances the production of exosomes and these vesicles can communicate with adjacent cells (Yu et al., 2006). Exosomal secretion of miR‐30e‐3p is regulated by the p53 signalling pathway in HCC cell lines treated with sorafenib, and higher exosomal miR‐30e‐3p level is associated with subsequent resistance to sorafenib (Gramantieri et al., 2020). Another study has demonstrated that HCC‐derived exosomal ASMTL‐AS1 reinforces the malignant behaviours of HCC cells via miR‐342‐3p/NLK/YAP signalling (Ma et al., 2020a). These pieces of evidence indicate that EVs may play certain roles in HCC development caused by several cell‐intrinsic drivers.

The major risk factors associated with HCC contribute to an elevation in the level of ROS, which is closely linked to the progression of HCC from early‐ to advanced‐ grade (Ko et al., 2018; Szabo & Momen‐Heravi, 2017), and previous studies have shown that endogenous overproduction of ROS is associated with enhanced EV release (Zhang et al., 2022). Elaborating on the mechanism of EV secretion under oxidative stress may provide novel therapeutic targets for HCC.

In humans, HEXB is an important lysosomal enzyme involved in the degradation of various cellular substrates (Jia et al., 2021). Accumulating studies have shown that HEXB is associated with various central nervous system diseases (Kostallari et al., 2021; Sierksma et al., 2020). In the present study, we demonstrated that HEXB is a key regulator promoting EV secretion in HCC and exhibits high expression levels in human HCC tissue, and is significantly associated with a poor prognosis. Knockdown of HEXB reduced ROS‐induced EV release in HCC cells, inhibiting the proliferation of HCC cells in vitro and in vivo. In addition, the exosomal microRNAs analysis of HEXB overexpressing cell‐derived EV revealed that high‐level HEXB not only promotes EV secretion, but also influences the levels of several oncogenic miRNAs in EVs, and miR‐21‐5p is the most significantly increased miRNA. It has been well demonstrated that miR‐21 is significantly increased in oxidative stress‐induced EVs, and HCC cells‐derived exosomal miR‐21‐5p promotes tumour growth (Cao et al., 2019; Xiao et al., 2016). Our findings remind that HEXB may facilitate exosomal miR‐21‐5p increase and then promote HCC cell growth. Therefore, our data indicate that HEXB is a key factor mediating EV release under oxidative stress and a potential therapeutic target for HCC treatment.

Previous studies have shown that HEXB is highly expressed in several cancers, such as laryngeal cancer (Olszewska et al., 2009) and glioblastoma multiforme (Jia et al., 2021). However, the precise mechanisms underlying the high level of HEXB in cancer cells still require further clarification. Here, we found that the levels of HEXB in HCC cells were upregulated by oxidative stress‐mediated TFEB nuclear translocation. On the one hand, HEXB has been identified as the target gene of TFEB which can directly lead to HEXB overexpression (Meireles et al., 2018). On the other hand, TFEB nuclear translocation also increases the levels of HEXB‐AS, which binds with and stabilizes the HEXB protein. It has been demonstrated that R‐loop structure, the RNA:DNA hybrids with concomitant displacement of the coding strand as single‐stranded DNA, can act as intrinsic Pol II promoters to initiate antisense lncRNA synthesis (Tan‐Wong et al., 2019). Given that HEXB‐AS overlaps with DNA sequence of HEXB in the genome (Figure S5a), we speculate that an R‐loop structure may exist in the HEXB gene locus, initiating the synthesis of HEXB‐AS when TFEB activates HEXB transcription.

Growing evidence highlights the lysosome as a dynamic regulator of cellular and organismal homeostasis and an appealing therapeutic target for multiple diseases (Ballabio & Bonifacino, 2020). Intriguingly, the lysosome in the EV release process showed a Janus face. On one hand, inhibiting lysosomal function can disrupt the MVB‐lysosome fusion, leading to enhanced transport of MVB towards the plasma membrane to release EVs (Hao et al., 2021). On the other hand, TFEB nuclear translocation enhances lysosomal function and promotes lysosomal exocytosis, an unconventional secretion process to release EVs (Buratta et al., 2020). In general, the increase of TFEB nuclear translocation promotes the expression of multiple lysosomal proteins, including LAMP1 (Meireles et al., 2018). However, a previous study has shown that hypoxia conditions induced EV release in ovarian cells, and a real‐time increase of TFEB expression accompanies this process and increased LAMP1 mRNA level but decreased protein level (Dorayappan et al., 2018). Moreover, the study on cancer‐associated fibroblasts has revealed that under hypoxic conditions, oxidative stress‐activated ataxia telangiectasia‐mutated gene affected lysosomal function (enhanced lysosomal pH and reduced LAMP1 level) to facilitate EV release (Xi et al., 2021). In this study, we found that in HCC cells, oxidative stress promoted TFEB nuclear translocation, leading to an upregulation of HEXB. Overexpression of HEXB decreased the LAMP1 protein level, and knockdown of HEXB led to the increase of the LAMP1 protein level. This result is consistent with a previous study on the Sandhoff disease (HEXB‐/‐) mouse model, showing that the LAMP1 protein level in the cerebellar of HEXB‐/‐ mice increased significantly compared with wild‐type mice (Dardis et al., 2020). Further, we demonstrated that the knockdown of HEXB attenuated LAMP1 degradation, and the dissociation of HEXB and LAMP1 is critical for maintaining LAMP1 stability. Our results also indicate that in response to M‐31850‐induced HEXB inhibition, the interaction between HEXB and LAMP1 is significantly reduced. HEXB can hydrolyze proteins, neutral glycolipids, and certain non‐reducing oligosaccharide residues of mucopolysaccharides, while LAMP1 is characterized by highly glycosylated. Consequently, we speculate that the interaction between HEXB and LAMP1 may cause changes in LAMP1 glycosylation, thereby affecting the protein stability of LAMP1. It has been demonstrated that LAMP1 plays a role in the fusion of lysosomes with phagosomes (Huynh et al., 2007) or autophagosomes (Luzio et al., 2007). By modulating the glycosylation level of LAMP1, HEXB may affect the protein stability of LAMP1, ultimately diminishing the fusion of lysosomes with other intracellular vesicles, including MVB. Our data showed that both knockdown HEXB directly and knockdown HEXB‐AS to influence the stability of HEXB can increase the fusion of lysosomes with MVBs, and this process can be abolished by silencing LAMP1, indicating the involvement of LAMP1 in MVB‐lysosome fusion and subsequent EV degradation. However, the detailed mechanisms of how HEXB mediates LAMP1 degradation still need further investigation.

The pharmacological inhibitor of EV release, GW4869, has been used in several studies to suppress tumour growth. However, its potential off‐target effects still need further investigation (Cheng & Hill, 2022). In our study, we found that the potent, selective β‐hexosaminidase inhibitor, M‐31850, can enhance the antitumor effect of GW4869. The combination of the two drugs showed a synergistic effect and significantly reduced the dose of GW4869, potentially minimizing off‐target effects. However, we also noticed that the anti‐HCC effect of combined use of GW4869 and M‐31850 did not improve the outcomes compared to targeted agents (sorafenib, lenvatinib, regorafenib). Given the notion that M‐31850 is currently underexplored in tumour therapy, the administration strategy of GW4869 and M‐31850 may require additional optimization before clinical application. Current studies on EV secretion and tumour progression are mainly based on experimental models which do not faithfully reflect the scenarios of EV inhibition in cancer patients, and further studies are needed to clarify the significance of targeting EV release for tumour therapy in the clinical setting. Additionally, it is important to consider the potential side effects of agents used to inhibit EV release and how to selectively deliver these inhibitors to cancer cells.

In summary, our study highlights how oxidative stress induces upregulation of the lysosomal protein HEXB, which attenuates the MVB‐lysosome fusion by promoting LAMP1 degradation, leading to an increase in EV secretion and promoting HCC cell growth. This finding reveals a mechanism by which oxidative stress promotes EV secretion in cancer cells and provides a potential target for the treatment of HCC.

AUTHOR CONTRIBUTIONS

Jiufei Duan: Investigation; methodology; visualization; writing—original draft; writing—review and editing. Zhao Huang: Investigation; methodology; visualization; writing—original draft; writing—review and editing. Siyuan Qin: Investigation; methodology; visualization; writing—review and editing. Bowen Li: Supervision; validation; writing—review and editing. Zhe Zhang: Supervision; validation. Rui Liu: Conceptualization; supervision; writing—review and editing. Kui Wang: Conceptualization; investigation; supervision; writing—review and editing. Edouard C. Nice: Writing—review and editing. Jingwen Jiang: Conceptualization; project administration; supervision; writing—review and editing. Canhua Huang: Conceptualization; funding acquisition; project administration.

CONFLICT OF INTEREST STATEMENT

None of the authors have any potential conflict of interest to disclose.

Supporting information

Supporting Information

Duan, J. , Huang, Z. , Qin, S. , Li, B. , Zhang, Z. , Liu, R. , Wang, K. , Nice, E. C. , Jiang, J. , & Huang, C. (2024). Oxidative stress induces extracellular vesicle release by upregulation of HEXB to facilitate tumour growth in experimental hepatocellular carcinoma. Journal of Extracellular Vesicles, 13, e12468. 10.1002/jev2.12468