Human L-02 hepatocytes and human HCC cell lines, including Hep3B, SK-Hep-1, and hepatoblastoma HepG2, were purchased from the National Collection of Authenticated Cell Cultures (Shanghai, China). Human HCC cell line HCCLM3 (LM3) was purchased from ATCC (Manassas, VA, USA). Human HCC cell lines MHCC97H (97H) and Huh7 were purchased from BeNa Culture Collection (Beijing, China). HepG2, LM3, 97H, Huh7, and L-02 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Carlsbad, CA, USA), while Hep3B and SK cells were cultured in Eagle’s minimum essential medium (Gibco) replenished with 10% fetal bovine serum (FBS; HyClone Laboratories Inc., Novato, CA, USA) at 37°C with 5% CO₂. In addition, the HepG2, Hep3B, and LM3 cells were treated with human recombinant TGF-β1 (10 ng/ml, R&D Systems, Minneapolis, USA) for 48 h.

For the wound healing assay, HepG2, HepB3, and LM3 cells were seeded in 24-well plates (Corning, NY, USA) and cultured overnight to form a monolayer on the bottom of the plate. The straight line was scratched with the 200-μl pipette tip, and adherent cells were cultured in complete medium with TGF-β1 (10 ng/ml) for 48 h. The wound healing rate was analyzed by measuring the distance of migrated cell monolayer. Cells were imaged using a microscope (Nikon, Japan), and wound width was analyzed with Image pro plus 6.0 (Media Cybernetics, USA). For the invasion assay, a 60-μl Matrigel matrix was added to each well to evenly cover the bottom of the Transwell chamber and cultured for 60 min to a semi-solidified state. After that, 2×10⁴ HCC cells were seeded in the upper chamber. Next, a complete medium with or without TGF-β1 (10 ng/ml) was added to the 24-well plate (outer chamber). After 48 h, 4% paraformaldehyde and 0.4% crystal violet (Sigma-Aldrich, MO, USA) were used to fix and stain the cells, respectively. The number of invasive cells was measured under an inverted phase contrast microscope and calculated using ImageJ software.

Cells were treated according to certain conditions before sphere culture with or without TGF-β1. Treated cells (1×10²) were seeded in ultra-low attachment 96-well plates (Corning) in serum-free DMEM/F12 (Gibco) supplement with insulin (5 μg/ml, HY-P0035, MCE), epidermal growth factor (EGF, 20 ng/ml, PeproTech), 2% B27 supplement (Invitrogen, CA, USA), and fibroblast growth factor-1 (FGF-1, 20 ng/ml, PeproTech). The cells were cultured for 7 days and then photographed using a microscope. The number of spheres (diameter > 50 μm) in each well was finally counted.

Total RNA was extracted from HCC cells, transfected cells, and exosomes derived from the plasma of HCC patients using TRIzol reagent (Takara, Dalian, China). Bestar™ qPCR RT kit (DBI, Germany) was used for the complementary DNA (cDNA) synthesis of mRNA. Mir-X™ miRNA First-Strand Synthesis Kit (TaKaRa) was used to synthesize the cDNA of miRNA. Bestar^® SYBR Green kits (DBI) were used to measure the expression of mRNA and miRNA. Real-time quantitative PCR (qPCR) was performed in Agilent Stratagene Mx3000P (Agilent, CA, USA). 2^−△△Ct method was used to calculate the relative expression of genes. Intracellular mRNAs and miRNAs were normalized with glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and U6. Exosomal miRNAs were normalized with cel-miR-39. The related primers are listed in Table 1 .

Radioimmunoprecipitation assay (RIPA) lysate (Beyotime, Nanjing, China) and nuclear extract kit (Active Motif, CA, USA) were used to extract the total protein from exosomes and cells, and the nuclear and cytoplasmic extracts, respectively. Protein concentration was quantitatively analyzed using a bicinchoninic acid (BCA) kit (Beyotime). The same amount of protein samples were taken from each group for electrophoresis. Then, the proteins were transferred to the polyvinylidene fluoride (PVDF) membrane and sealed with a rapid blocking solution for 30 min. The membranes were incubated with primary antibodies at 4°C overnight and then incubated with second horseradish peroxidase (HRP)-conjugated antibodies for 1 h. Finally, membranes were incubated with ECL-Plus reagent, and the blots were observed using Gel Imaging System. Related antibodies were as follows: CD44 (37259, CST, USA), CD133 (64326, CST, USA), OCT4 (2750, CST, USA), CD63 (ab134045, Abcam, USA), ALIX (92880, CST, USA), TSG101 (ab125011, Abcam, USA), Calnexin (ab133615, Abcam, USA), STK25 (ab157188, Abcam, USA), ZO-1 (8193T, CST, USA), N-cadherin (13116T, CST, USA), E-cadherin (3195T, CST, USA), p-YAP (s297) (13619, CST, USA), YAP (14074, CST, USA), TAZ (72804, CST, USA), PCNA (ab280088, Abcam, USA), GAPDH (A00227, BOSTER, China), Histone H3 (ab1791, Abcam, USA), HRP goat anti-rabbit IgG (BA1054, BOSTER, China), and HRP goat anti-mouse IgG (BA1051, BOSTER, China).

HCC cells were cultured in the corresponding medium with 10% exosome-depleted FBS for 48 h. The cell-conditioned medium (10 ml) or serum (250 μl, peripheral blood from HCC patients or healthy volunteers) were collected. The exosomes were isolated using ExoQuick-TC™ (System Biosciences, USA) and identified by transmission electron microscopy (TEM) and Western blot analysis, respectively. For TEM, exosomes prefixed with 4% paraformaldehyde (PFA) were loaded onto formvar carbon-coated grids and incubated for 20 min; then, the grids were refixed with 3% glutaraldehyde and 1% osmium tetroxide, respectively. After these, the exosomes were stained with uranyl oxalate, contrasted, and embedded in a mixture of 2% methylcellulose and 4% uranyl acetate with a ratio of 900 μl/100 μl, respectively. Finally, the exosomes were observed with a TEM (Hitachi 600, Hitachi, Japan). Furthermore, the expression of exosome-specific proteins CD63, TSG101, Alix, and Calnexin was detected by Western blotting analysis.

The isolation of serum (250 μl, peripheral blood from HCC patients or healthy volunteers)-derived exosomes was conducted using the total exosome-isolation kit following the manufacturer’s recommendation. The treated serum was harvested using twice centrifugation (3,000 rpm at 4°C for 15 min; 15,000×g for 30 min) to remove microvesicles.

Plasma was collected from 20 HCC patients and 20 healthy volunteers who provided informed consent. The clinical information of patients is shown in Table 2 . The miR-4800-3p expression was detected from plasma-derived exosomes. This research followed the ethical guidelines of the 1975 Declaration of Helsinki and was performed with the approval of the Sun Yat-sen Memorial Hospital, Sun Yat-sen University Institutional Review Board.

The PKH67 Green Fluorescent Cell Linker Mini Kit (Sigma) was used to label exosomes. The labeled exosomes were added to HepB3 and LM3 cells in a 24-well plate. After incubation for 0.5 min, 2 h, 12 h, and 24 h, images were captured to observe the presence of exosomes within the HCC cells.

To change the expression of miR-4800-3p and STK25, the miR-4800-3p mimic, the miR-4800-3p inhibitor, and their negative control (NC), the overexpression of the STK25 plasmid (pcDNA3.1-STK25 plasmid, OE-STK25) and the vector plasmid pcDNA3.1 (OE-NC) was transfected into HepB3 and LM3 cells using Lipofectamine 2000 (Invitrogen, USA). The mimic, inhibitor, and miRNA-4800-3p antagomir were synthesized by Ribobio (Guangzhou, China), and the plasmids were obtained from GeneChem (Shanghai, China). Transfection efficiencies 48 h after transfection were explored by real-time PCR. In addition, to explore the effect of exosomes derived from Huh7 cells on HCC cells, HepB3 and LM3 cells were treated with exosomes (10 μg/ml) for 48 h.

The treated HepB3 and LM3 cells were labeled with 50 μM 5-ethynyl-2′-deoxyuridine (EdU; Solarbio, Beijing, China) for 2 h, and then, the cell nucleus was stained with 4′,6-diamidino-2-phenylindole (DAPI). Finally, the cells were imaged by a fluorescence microscope (Olympus, Tokyo, Japan). The proportion of EdU-positive cells in the whole visual field was analyzed.

HepB3 and LM3 cells were fixed, permeabilized, and blocked with 1% bovine serum albumin (BSA). After these, the cells were incubated with antibodies against STK25 (ab157188, Abcam, USA) and Alexa Fluor488-conjugated goat anti-rabbit IgG (ab150077, Abcam, USA) antibody, and post-stained with DAPI. The images were photographed using fluorescence microscopy.

The exact mechanism of miR-4800-3p that promotes the development of HCC remains unclear. Here, the bioinformatic tools (TargetScan, StarBase, and miRDB database) were used to predict the target genes of miR-4800-3p, and the data showed that STK25 was a target gene of miR-4800-3p. Then, the luciferase reporter assay was performed to verify the combination of the untranslated region (3′-UTR) of STK25 mRNA with miR-4800-3p. HepB3 and LM3 cells were cotransfected with dual-luciferase reporter vectors [psiCHECK-2 reporter plasmids bearing STK25 3′-UTR containing wild-type (MT-STK25) or mutated (MUT-STK25) predicted miR-4800-3p binding sites] and miR-4800-3p mimics or NC. After 48 h, luciferase activities were detected using a dual-luciferase reporter assay kit (Promega, Madison, USA). Firefly luciferase activity (F) was used as the internal control.

HepB3 and LM3 cells with overexpressed miR-4800-3p or NC were collected, and the assay was performed strictly according to the procedures of the miRNA Target IP Kit (Active Motif, USA). Briefly, cells were lysed in a complete lysis buffer and then incubated with protein G-coupled magnetic beads to immunoprecipitate miRNA/mRNA complexes. Next, complexes were treated with proteinase K digestion to elute RNA from beads, and RNA was extracted using phenol:chloroform:isoamyl alcohol (25:24:1). The relative expression of STK25 was detected by qRT-PCR.

To explore the role of exosomes and miR-4800-3p in vivo, 2×10⁶ LM3 cells per mouse were subcutaneously injected to establish xenografts (on the right). The mice were then randomly divided into four groups (N=5): Control, Exo, miR-4800-3p antagomir, and Exo+miR-4800-3p antagomir. Sterile saline (50 μl), exosomes (50 µg in 50 μl), and/or miR-4800-3p antagomir (20 nmol in 50 μl) were injected every 3 days for 4 weeks at 3-4 points in the subcutaneous tumors, respectively. The volume of implanted tumors was then calculated using the formula V =  a × b²/2 (a is the long axis, and b is the short axis). The male BALB/c nude mice (6 weeks old, 18 ± 2 g) were used in this study. Mice were sacrificed after 4 weeks of Exo and/or miR-4800-3p antagomir treatment, and the tumor tissues were removed and weighed. All animal studies were approved by the Animal Care Committee of Sun Yat-sen Memorial Hospital, Sun Yat-sen University in accordance with Institutional Animal Care and Use Committee guidelines.

In this study, the cell experiment had three biological repetitions, and the Transwell invasion experiment had six biological repetitions (visual field statistics). Statistical analysis was carried out using GraphPad Prism 7.0 software (GraphPad, CA, USA), and results were represented as mean ± standard deviation. Unpaired t-tests and one-way ANOVA were performed to compare the two and multiple groups, respectively. A p-value < 0.05 was defined as statistically significant.

To explore the effect of TGF-β1 on HCC cells, HepG2 hepatoblastoma cells and low-metastatic HCC cells (HepB3 and LM3) were selected. The wound healing and Transwell assay results showed that TGF-β1 could markedly increase the migration and invasion abilities of HepG2, HepB3, and LM3 cells ( Figures 1A, B ). In addition, the sphere-forming assay further confirmed that TGF-β1 could significantly promote the stemness of HepG2, HepB3, and LM3 cells ( Figure 1C ). Furthermore, we also obtained the upregulation of tumor stem cell markers (CD44, CD133, and OCT4) in HepG2, HepB3, and LM3 cells with the TGF-β1 treatment at both RNA and protein levels ( Figures 1D, E ). All these data demonstrated that TGF-β1 promotes the malignant progression of HCC cells.

Exosome-encapsulated tumorigenic miRNAs could regulate tumor progression and cell communication. Here, we isolated exosomes from L-02 hepatocytes and six HCC cell lines (HepG2, HepB3, LM3, 97H, Huh7, and SK) and explored the role of related exosomal miR-4800-3p in the progression of HCC. As shown in Supplementary Figure S1A , exosomes were verified by TEM to have a typical cup-shaped morphology with a diameter of approximately 30–100 nm and further confirmed by Western blotting with positive exosomal markers (CD63, ALIX, and TSG101) in the exosomal fraction in both the control group and TGF-β1 group ( Supplementary Figure S1B ). Furthermore, the results also showed that, as a negative control, Calnexin was absent in the exosomal fraction ( Supplementary Figure S1B ). The expression of miR-4800-3p in HCC cells and cells-derived exosomes were further detected and markedly upregulated in TGF-β1-treated HCC cells (HepG2, HepB3, and LM3) and related cell-derived exosomes ( Figure 2A ). Furthermore, the expression of exosomal miR-4800-3p in HCC cells (HepG2, HepB3, LM3, 97H, Huh7, and SK) was higher than in L-02 hepatocytes, and the highest levels of miR-4800-3p were detected in Huh7-derived exosomes ( Figure 2B ). However, the expression of TGF-β1 demonstrated that endogenous TGF-β1 was not differentially expressed in exosomes derived from HCC cells and HCC cells (HepB3 and LM3) ( Supplementary Figure S1C ). Furthermore, higher levels of miR-4800-3p were detected in plasma exosomes from HCC patients than from healthy volunteers ( Figure 2C ). These data indicate that miR-4800 is elevated in cells and exosomes from HCC cell lines and patients.

Since our previous data showed the inverse expression pattern of STK25 and miR-4800-3p, it is still uncertain whether STK25 is a target of miR-4800-3p. To identify this prediction, the TargetScan, StarBase, and miRDB databases were used to predict the potential targets of miR-4800-3p, and the data showed the binding sites between miR-4800-3p and STK25 in 3′UTR ( Figure 2D ). The luciferase reporter plasmids containing wild-type (WT) or mutant (MUT) 3′-UTR of STK25 and cotransfected HepB3 and LM3 cells with miR-4800-3p mimic or NC mimic and WT or MUT STK25 3′-UTR were further constructed to verify the combination of STK25 and miR-4800-3p. The results showed that luciferase activity was markedly reduced when cotransfected with miR-4800-3p mimic and luc-WT-STK25 in HepB3 and LM3 cells, while luciferase activity did not change when cotransfected with miR-4800-3p mimic and luc-MUT-STK25, indicating that miR-4800-3p could directly interact with STK25 3′-UTR ( Figure 2E ). The miRNA target immunoprecipitation assay showed that STK25 was enriched in HepB3 and LM3 cells with miR-4800-3p overexpression ( Figure 2F ), further confirming that miR-4800-3p could exert its role by competitively binding STK25 to inhibit STK25 expression.

As the highest levels of miR-4800-3p in HCC cell-derived exosomes, Huh7 cell-derived exosomes were selected to detect the effect of exosomal miR-4800-3p on HCC progression in subsequent experiments. To determine whether exosomal miR-4800-3p can enhance proliferation, migration, and invasion in low-metastatic HCC cells and the effects of miR-4800-3p on HCC cells, HepB3 and LM3 cells were treated with Huh7 cell-derived exosomes or transfected with miR-4800-3p mimic or miR-4800-3p inhibitor, respectively. After being treated with PKH67-labeled exosomes for 24 h, strong green fluorescent signals were detected, indicating that the recipient HCC cells (HepB3 and LM3) could successfully absorb Huh7-derived exosomes ( Supplementary Figure S2A, B ). Furthermore, after being treated with PKH67-labeled exosomes for 24 h, the expression of cellular miR-4800-3p was increased ( Supplementary Figure S2C ). The results showed that miR-4800-3p mimic promoted the expression of miR-4800-3p in exosomes, while miR-4800-3p inhibitor inhibited the expression of miR-4800-3p ( Supplementary Figure S2D ). The proliferation abilities and stemness of HCC cells were further observed by EdU staining and the sphere formation assay, respectively ( Figures 3A, B ). The results showed that exosomes derived from Huh7 with high-metastatic activity could greatly upregulate the proliferation of HepB3 and LM3 cells. Besides, the proliferation of HCC cells was enhanced when treated with miR-4800 mimics, but the knockdown of miR-4800-3p exerted an opposite effect ( Figure 3A ). Furthermore, both Huh7-derived exosomes and overexpressed miR-4800-3p could notably improve the proliferation, migration and invasion abilities, and stemness of HCC cells, while miR-4800-3p knockdown exerted an opposite effect ( Figures 3B-D ). These data indicated that high-metastatic Huh7 cells could endow low-metastatic HCC cells (HepB3 and LM3) with more powerful proliferation, migration, and metastasis abilities by secreting exosomal miR-4800-3p.

To further explore the mechanism through which miR-4800-3p exhibited its role, HepB3 and LM3 were treated with Huh 7 cell-derived exosomes or transfected with miR-4800-3p mimic or miR-4800-3p inhibitor, respectively; the transfection effect and also the treatment of Huh 7 cell-derived exosomes were satisfactory ( Figure 4A ). A previous study showed that STK25 is an upstream activator of LATS kinases, whose loss notably promotes YAP/TAZ activity and enhanced cellular proliferation (25). However, it was not clear whether STK25-regulated LATS-YAP activity was involved in promoting miR-4800-3p in the malignant progression of HCC cells. Here, the expression of STK25, tumor stem cell markers, EMT, and YAP/TAZ activity in HCC cells were detected when treated with exosomal miR-4800-3p or transfected with miR-4800-3p mimic or miR-4800-3p inhibitor, respectively. Results showed that exosomal miR-4800-3p and miR-4800-3p mimic inhibited the expression of STK25 in HepB3 and LM3 cells, while miR-4800-3p inhibitor upregulated STK25 expression ( Figures 4B, C ). These results were further confirmed by immunofluorescence assay ( Figure 4D ). The expression of tumor stem cell markers (CD44, CD133, and OCT4) were also notably increased by exosomal miR-4800-3p and miR-4800-3p mimic, respectively ( Figures 4E, F ), while knockdown of miR-4800-3p decreased these tumor stem cell markers. For the detection of EMT markers, we found that exosomal miR-4800-3p or miR-4800-3p mimic treatment in HepB3 and LM3 cells could downregulate the expression of E-cadherin and ZO-1 but increase the expression of N-cadherin ( Figures 4G, H ). Similarly, opposite effects were observed when miR-4800-3p was silencing ( Figures 4G, H ). Next, it was investigated whether STK25-regulated ATS-YAP activity was involved in miR-4800-3p promoting the malignant progression of HCC cells. The activity of YAP/TAZ and the expression of proliferating cell nuclear antigen (PCNA) were explored via Western blot analysis. The results showed that the p-YAP in the nuclear fraction was downregulated, while the expression of YAP, TAZ, and PCNA increased in HepB3 cells under the exosomal miR-4800-3p or miR-4800-3p mimic treatment, respectively ( Figure 4I ). In contrast, silencing miR-4800-3p exerted the opposite effects ( Figure 4I ). Similar results were also obtained in LM3 cells. These data indicated that the STK25-regulates Hippo pathway participated in proliferation, migration, invasion, and EMT regulated by miR-4800-3p in HCC cells.

HepB3 and LM3 cells were further cotransfected with miR-4800-3p mimic and OE-STK25 plasmid or NC plasmids, and STK25 expression increased markedly in HepB3 and LM3 cells when transfected with OE-STK25 plasmid alone ( Figures 5A, B ) or cotransfected with miR-4800-3p mimic ( Figures 5C, D ), demonstrating the combination of miR-4800-3p and STK25.

EdU staining showed that HepB3 and LM3 cell proliferation under miR-4800-3p mimic stimulation was markedly inhibited when treated with the OE-STK25 plasmid ( Figures 6A, D ). Invasion and wound healing assays also indicated that invasion ( Figures 6B, E ) and migration ( Figures 6C, F ) of HepB3 and LM3 cells were obviously decreased when cotransfected with the miR-4800-3p mimic and OE-STK25 plasmid. These data demonstrated that overexpression of STK25 could inhibit proliferation, invasion, and migration of HepB3 and LM3 cells.

The sphere-forming assay showed that the stemness of HepB3 and LM3 cells was markedly inhibited by overexpression of STK25 ( Figure 7A ). Correspondingly, the expression of tumor stem cell markers (CD44, CD133, and OCT4) upregulated by miR-4800-3p was also decreased by the overexpression of STK25 in HepB3 and LM3 cells ( Figures 7B, C ). Furthermore, the overexpression of STK25 could promote miR-4800-3p-suppressed E-cadherin and ZO-1 expression in HepB3 and LM3 cells but suppress miR-4800-3p-induced N-cadherin ( Figures 7D,E ). The effect of miR-4800-3p on YAP phosphorylation and YAP, TAZ, and PCNA expression were also largely reversed by the overexpression of STK25 in HepB3 and LM3 cells ( Figure 7F ). All these results indicated that the miR-4800-3p enhances malignant phenotypes of HCC cells by targeting STK25 and activating the Hippo pathway.

To further explore the roles of exosomes through miR-4800-3p in HCC, we built a nude mice model with LM3 cells and treated it with Huh7-derived exosomes and/or miR-4800-3p antagomir by injection. As the data revealed, miR-4800-3p antagomir treatment markedly suppressed tumor growth promoted by Huh7-derived exosomes ( Figures 8A–C ). The data showed that miR-4800-3p antagonist treatment significantly inhibited the expression of miR-4800-3p ( Figure 8D ). In addition, the EMT markers and the Hippo signaling pathway in tissues were also analyzed. As shown in Figure 8E , knockdown of miR-4800-3p could promote the Huh7-derived exosomes-suppressed E-cadherin and ZO-1 expression in tumor tissues but suppress the exosome-induced N-cadherin. In addition, the effect of Huh7-derived exosomes on the phosphorylation of YAP and the expression of STK25, YAP, TAZ, and PCNA were also largely reversed via knockdown of miR-4800-3p in HCC tumors tissues ( Figure 8E ). These results demonstrated that Huh7-derived exosomes promoted the growth and EMT of implanted tumors in vivo through miR-4800-3p.