Globulol (PubChem CID: 12304985, Catalog Number: B29181, Purity: ≥ 98%) was supplied by Yuanye Biotechnology (Shanghai, China). Tislelizumab (TIS, BGB-A317) was supplied by BeiGene, Ltd. (Beijing, China). The cell lines used in this study were obtained from Wuhan Oyster Mushroom Biotechnology Co., Ltd., Bena Bioengineering Technology Research Center, and Wuhan Huofai Biotechnology Co., Ltd. The fetal bovine serum and Dulbecco’s Modified Eagle Medium (DMEM) were supplied by Gibco. Apexbio (K1018) provided the CCK8 reagent. Beyotime Biotechnology provided the EdU cell proliferation kit (C0075S), cell cycle analysis kit (C1052), CFDA-SE Cell Proliferation Detection Kit (C0051), LDH assay kit (C0016), PAK4 inhibitor (PF-03758309), Human CD8⁺T Cells Positive Selection Kit (071A103.11), and the apoptosis assay kit (40302ES50). IPHASE Co., Ltd, BeiGene, Ltd., and Yeasen Biotechnology (Shanghai) Co., Ltd also supplied various reagents. Abcam and Proteintech provided the antibodies used in this study. The antibodies against PAK4 (ab314765), AKT (60203-2-Ig), phospho-AKT (80455-1-RR), STAT3 (ab68153), phospho-STAT3 (ab76315), PD-L1 (28076-1-AP), and GAPDH (60004-1-Ig) were acquired. The Guangzhou Tianyi Huiyuan Biotechnology Co., Ltd. offered primers for genes such as GAPDH, CCL2, CCL3, CCL4, CCL5, CXCL9, CXCL10, IFN-γ, and TNF-α.

HCC cell lines Huh1 and Huh7, and the human hepatic stellate cell line LX-2 were purchased from Wuhan Procell Life Science and Technology Co., Ltd. (Wuhan, China). Primary human CD8⁺ T cells were obtained from Xiamen Immocell Biotechnology Co., Ltd. (Xiamen, China). All cell lines were authenticated by short tandem repeat (STR) profiling to confirm their identity, and no cross-contamination was detected in subsequent culture processes. Huh1, Huh7, and LX-2 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) high-glucose medium (Gibco) supplemented with 10% fetal bovine serum (FBS, Gibco, United States), 100 mg/mL penicillin, and 100 mg/mL streptomycin. Primary human CD8⁺T cells were maintained in RPMI-1640 medium (Gibco) containing 10% FBS, 100 mg/mL penicillin, and 100 mg/mL streptomycin. All cells were incubated in a humidified cell culture incubator at 37 °C with 5% CO₂. Prior to co-culture experiments with HCC cells, CD8⁺T cells were pre-activated to enhance their functional activity: CD8⁺T cells were stimulated with ImmunoCult™ CD3/CD28/CD2 T Cell Activator (STEMCELL Technologies) and IL-2 for 48 h at 37 °C with 5% CO₂.

In 96-well plates, cells were seeded at 3,000 cells per well and treated with varying doses of Globulol (0–320 μM) for a duration of 24 h. Post-treatment, the supernatant was removed, and each well received 100 μL of medium supplemented with 10% CCK8. Absorbance was measured at 450 nm using a spectrophotometer to assess cell viability.

The cells were then incubated with a 2x EdU solution for 2 hours after being treated with several doses of Globulol. After that, each well was fixed for 15 min with 4% paraformaldehyde. Permeation of the cells was carried out for a further 15 minutes using 0.25% Triton X-100 following fixation. Before the last staining stage with Hoechst for 10 min, cells were incubated with a click additive solution for 30 min. Afterwards, they were observed under a fluorescence microscope.

Huh1 and Huh7 cells were treated with different concentrations of Globulol for 24 h, followed by resuspension in 1× binding buffer. The cells were stained with Annexin V-FITC and PI according to the directions of the apoptosis kit, and then left to incubate in darkness for 10–15 min. Afterwards, Fc was used for quantitative analysis of apoptotic rates. Cells were fixed with 80% ethanol, stained with PI, and cultured in darkness at 37 °C for 30 min prior to Fc evaluation in order to conduct cell cycle study. CFDA-SE was integrated into the co-culture system of 6-well plates treated with different doses of Globulol and TIS. CD8⁺T cells, post co-culture, were separated using the Human CD8⁺T Cells Positive Selection Kit protocol, and proliferation metrics were evaluated via FC.

Forty microliters of Cytoport Biotech’s Biomimetic Matrix Gel for three-dimensional (3D) Cell Culture (CulX I) and 50 μL of Huh1 and Huh7 cell suspensions were spread out in every well of a 96-well plate. One hundred microliters of culture media was added once the gelling process had culminated. The cells were cultured for 6 days, then stained with calcein-AM/PI, subjected to Globulol for 24 h, and finally evaluated under a fluorescence microscope.

A 50 μL suspension of Huh1 and Huh7 cells mixed with Biomimetic Matrix Gel was placed into each well of a 96-well plate. After solidification, 100 μL of medium was added and cells were cultured for 6 days. Post transfer to a 6-well plate, tumor spheroids were exposed to varied concentrations of Globulol for 12 h, stained with Calcein-AM, and visualized with a fluorescence microscope.

In 12-well plates equipped with Transwell inserts, 80 μL of Biomimetic Matrix Gel was applied to the bottom of the upper chamber. After solidification, CD8⁺T cells were seeded into the upper chamber at a density of 1 × 10⁵ cells/well, and stimulated with ImmunoCult™ CD3/CD28/CD2 T Cell Activator (STEMCELL Technologies) and IL-2 for 48 h. The stimulation was performed in RPMI-1640 medium (Hyclone) supplemented with 10% fetal bovine serum (FBS, Gibco) under the condition of 37 °C and 5% CO₂. Subsequently, hepatocellular carcinoma (HCC) cells (Huh1 and Huh7) were added to the lower chamber of the 12-well plate at a ratio of 25:1 (relative to CD8⁺T cells, approximately 4,000 cells/well) and cultured in serum-free Dulbecco’s Modified Eagle Medium (DMEM). At the initiation of co-culture, Globulol (80 μM) and TIS (57.14 mg/L) were added to the lower chamber (Figure 5A). Subsequently, the activated CD8⁺T cells and HCC cells were co-cultured in a humidified incubator at 37.0 °C with 5% CO₂ for 24 h. After the 24 h co-culture period, the supernatant and floating CD8⁺T cells were removed, and the biomimetic matrix was wiped off. Cells in the upper chamber were stained with Calcein-AM and imaged using a fluorescence microscope.

Using previously described techniques, a 3D tumor spheroid model comprising Huh7 cells was constructed. These spheroids were transferred to a 24-well plate, and CD8⁺T cells were seeded into the plate at a density of 1 × 10⁴ cells per well in RPMI-1640 medium containing 10% FBS. Subsequently, the co-culture system was treated with Globulol (80 μM) and TIS (57.14 mg/L), and maintained for 24 h. After the incubation period, the 3D tumor spheroids were observed under a bright field microscope.

In a 24-well format, CD8⁺T cells and Huh7 cells were co-cultured and then treated with various Globulol and TIS combinations for 24 h. The supernatant was subsequently collected for LDH activity measurement using a microplate reader.

RNA-seq was performed on Huh7 cells, both with and without drug treatment. Following extraction, total RNA was assessed for its purity, concentration, and integrity. The Illumina HiSeq platform’s paired-end sequencing technique was utilized for comprehensive sequencing of the samples. Differentially expressed genes (DEGs) were identified using the DESeq2 software package. The selection criteria for DEGs were defined as absolute log₂ fold change (|log₂FC|) ≥ 1 and a statistical significance of P value ≤0.05. Analyses were carried out utilizing the Kyoto Encyclopedia of Genes and Genomes (KEGG) database and the TopGO database, respectively, to implement gene ontology (GO) functions. The analysis’s visual enrichment plots were also made easier to create with the use of gene set enrichment analysis (GSEA) software.

The sequences for PAK4 overexpression and knockdown were acquired from Shanghai Jikai Gene Chemical Technology. Huh7 cells were transfected with these plasmids using PEI 40K Transfection Reagent (Servicebio, Wuhan, China) following the manufacturer’s guidelines. Post-transfection, cells were exposed to Globulol for 24 h. Subsequently, gene and protein samples from these cells were harvested for further analysis.

The co-culture approach was used to isolate CD8⁺T cells by employing a Human CD8⁺T cell Positive Selection Kit. To prepare for qPCR, total RNA was extracted using the Eastep® Super Total RNA Extraction Kit and then converted to cDNA using Hifair 1/3 1st Strand cDNA Synthesis SuperMix. Using a thermal cycler, the qRT-PCR was carried out with SYBR Green. Table 1 lists the primers that were used. The 2^−ΔΔCt technique was used to process the data after normalizing the expression levels of particular genes to GAPDH.

SDF files of Globulol (CAS number: 489–41-8) were retrieved from the Pubchem database. Structural details of pertinent target proteins were obtained from the PDB database. These protein structures were refined using Pymol-2.1.0 software, followed by hydrogen addition and charge settings using AutoDockTools-1.5.6. Molecular docking was executed with the vina-2.0 module within the pyrx software suite, calculating the binding energies. Visual representations were analyzed using Pymol and the Discovery Studio 2019 Client tools.

A RIPA buffer with 1% protease inhibitor, purchased from Beyotime Biotechnology Co., Ltd., was used to break down the cells. Afterwards, the protein concentration was measured using the supernatant. After the protein samples were prepared, they were transferred to PVDF membranes and separated using SDS-PAGE. After 1 hour of blocking with 5% skim milk, the membranes were incubated with primary antibodies at 4 °C throughout the night. After three washes with 1% TBST the next day, the membranes were incubated with secondary antibodies conjugated with HRP for 1 hour. Protein bands were seen using a Tanon imager and band intensity was assessed using ImageJ software after three more washes.

At least three independent experiments were conducted for each experimental condition. The mean ± standard deviation is the result of data analysis performed using GraphPad Prism 8 program. A two-tailed paired Student’s t-test, with a significance level of P < 0.05, was used to conduct comparative analyses between the two datasets.

The chemical structure of Globulol, sourced from the TCMSP database, is depicted in Figure 1A. The CCK8 assay revealed that Globulol significantly diminished the viability of both Huh1 and Huh7 cells in a dose-responsive manner. Treatment with Globulol for 24 h resulted in IC50 values of 136.6 μM and 141.5 μM for Huh1 and Huh7 cells, respectively, demonstrating statistically significant effects (P < 0.001, Figure 1B). Conversely, LX-2 cells (human hepatic stellate cell line) exposed to Globulol concentrations ranging from 30 to 160 μM exhibited no notable viability alterations (P > 0.05, Figure 1C). For subsequent experiments, dosages of 40 μM, 80 μM, and 120 μM were chosen. Furthermore, a marked decrease in EdU-positive Huh1 and Huh7 cells was observed following Globulol treatment, suggesting a suppression of cellular proliferation at these concentrations (P < 0.01, Figures 1D–G).

FC was employed to analyze cell-cycle distribution changes in Huh1 and Huh7 cells after treatment with 40 μM, 80 μM, and 120 μM Globulol for 24 h. Data from Figures 2A,B highlight a substantial increase in the G0/G1 phase fraction in Huh1 cells compared to controls (P < 0.001). Similarly, a significant elevation in the G0/G1 phase of Huh7 cells was observed post-treatment (Figures 2C,D) (P < 0.001). These findings suggest that Globulol effectively arrests cell cycle progression in these cell lines, thus inhibiting HCC cell proliferation. The impact of Globulol on apoptosis in Huh1 and Huh7 cells was also evaluated using FC. Figures 2E and F illustrate that varying concentrations of Globulol led to an increase in apoptosis rates, particularly at 40 μM (P < 0.01), 80 μM (P < 0.001) and 120 μM (P < 0.001). This dose-dependent induction of apoptosis suggests that Globulol effectively promotes cell death in these cell lines, further inhibiting HCC cell proliferation.

3D tumor spheroids, which better mimic in vivo solid tumor architecture through tight cell-cell and cell-matrix interactions (Ishiguro et al., 2017), were used to assess the impact of Globulol on the migratory capacity of Huh1 and Huh7 cells. Figures 3A and B reveal that in the control group, tumor spheroids exhibited evident radial cell migration. Conversely, in groups treated with increasing concentrations of Globulol, the density of migrating cells decreased. At the highest concentration tested (120 μM), the boundary of the tumor spheroids was well-defined, with few migrating cells observed. These findings suggest that Globulol significantly suppresses the migratory capacity of both Huh1 (P < 0.01) and Huh7 cells (P < 0.001).

To further investigate Globulol’s penetration and cytotoxicity in a 3D model, tumor spheroids were used to simulate the drug-resistant TME. In the control group’s Huh1 and Huh7 models, tumor spheroids displayed uniform green fluorescence (live cells). However, in groups treated with Globulol, red fluorescence (dead cells) intensified significantly as the concentration increased. At the highest concentration (120 μM), the tumor spheroids showed signs of structural disintegration (Figures 3C,D). These results demonstrate Globulol’s ability to kill Huh1 and Huh7 cells within a 3D structure and disrupt tumor spheroid integrity.

RNA-seq was employed to analyze gene expression changes in Globulol-treated Huh7 cells compared to untreated controls. Heat map were used to visualize these changes (Figure 4A). Additionally, GO analysis of all differentially expressed genes (Figure 4B) showed significant enrichment in pathways related to T cell proliferation and the regulation of T cell differentiation. Based on the initial 50 genes displayed in the heat map, p21-activated kinase 4 (PAK4) expression was significantly downregulated. The GSEA analysis (Figures 4C,D) revealed significant enrichment in pathways related to lymphocyte-mediated immunity, immune response regulation in tumor cells, T cell migration, cytoskeleton organization, and the mitotic cell cycle. KEGG analysis (Figure 4E) indicated significant enrichment in the cell cycle pathway, PD-L1 expression, and the PD-1 immune checkpoint pathway. qRT-PCR was used to validate the differential gene expression in Huh7 cells following Globulol treatment (Figure 4F), which demonstrated a significant reduction in PD-L1 transcription (P < 0.01). Given that PAK4 is overexpressed in various tumor tissues and is linked to malignancy and poor prognosis in HCC, it was selected for further investigation.

Transwell coculture assays were performed to investigate the influence of Globulol on CD8⁺T-cell chemotaxis towards HCC cell lines treated with either Globulol or the anti-PD-1 drug TIS, as depicted in Figure 5A. Enhanced migration of CD8⁺T cells was observed in both Huh1 and Huh7 cells post-treatment (Figures 5B–E), respectively. Further augmentation of CD8⁺T-cell migration was noted in the combination group with TIS. To identify potential chemokines involved in this process, qRT-PCR was performed to detect the expression levels of multiple chemokines in Globulol- or TIS-treated Huh1 and Huh7 cells. Among the detected chemokines, only CCL4 showed a significant upregulation in both Huh1 (Figure 5F) and Huh7 cells (Figure 5G) after Globulol treatment, while the expression of other chemokines remained unchanged.

To assess the combination effect of Globulol combined with TIS, an in vitro 3D tumor spheroid co-culture model of Huh7 cells and CD8⁺T cells was established (Figure 6A). Figure 6B reveals that in the control group, the tumor spheroids maintained intact structures with clear edges and no significant cell detachment. Treatment with Globulol resulted in slight loosening of the spheroid edges and some cell detachment. Following TIS treatment, the spheroid structure remained largely intact with localized disintegration. However, after combined treatment with Globulol and TIS, significant disintegration occurred, the core area collapsed, and numerous fragmented cells detached. These observations indicate that the combination of Globulol and TIS exerts a combination effect, which significantly outperforms either monotherapy alone. LDH assay results (Figure 6C) demonstrated that Globulol enhanced the cytotoxicity of CD8⁺T cells in a concentration-dependent manner, and this pharmacological effect was further potentiated when Globulol was combined with TIS (P < 0.001).

The influence of Globulol on the proliferation of CD8⁺T cells was investigated using FC. Results presented in Figure 7A and B demonstrate that Globulol facilitated a dose-dependent increase in CD8⁺T cell proliferation, which was further potentiated when combined with TIS (P < 0.001). This combined treatment also markedly increased the expression levels of TNF-α and IFN-γ (P < 0.001) (Figures 7C,D), while reducing the expression of PD-1 and TIM-3 (P < 0.01) (Figures 7E,F). Notably, CD8⁺T cells from the combination treatment group showed elevated levels of TNF-α and IFN-γ compared to those from the untreated group.

RNA-seq differential gene analysis identified PAK4 as the primary target. Figure 8A illustrates the binding mode of Globulol with PAK4 (binding energy: −6.8 kcal/mol). PAK4 is shown as a gray ribbon, and Globulol fits snugly into PAK4’s binding pocket. The complex is stabilized by a hydrogen bond with residue A402, hydrophobic contacts (G401, F397, L398), and polar interactions (D458, D444, E396, K350). These favorable binding energy and stable interactions support Globulol’s potential as a PAK4-targeted agent. PAK4 regulates the WNT/β-catenin and promotes PI3K/AKT signaling by directly binding to PI3K, thus enhancing AKT phosphorylation (Naїja et al., 2021). This activation augments STAT3-dependent PD-L1 expression downstream (Zhang et al., 2018). Western blot analysis was conducted to measure key protein expressions in Huh7 cells treated with varying Globulol concentrations. Figure 8B and C show that Globulol significantly reduced the protein levels of PAK4, p-AKT (Ser473), STAT3, p-STAT3 (Y705), and PD-L1, without notably affecting total AKT levels (compared with Ctrl group). Additionally, a Huh7 cell model with PAK4 overexpression (PAK4-OE) was established (Figures 8D,E). This model, treated with a combination of Globulol and the PAK4 inhibitor PF-3758309, demonstrated that PAK4 overexpression increased p-AKT (Ser473) protein levels but did not affect total AKT levels (P > 0.05). Treatment with Globulol significantly decreased the protein expressions of PAK4, p-AKT (Ser473), STAT3, p-STAT3 (Y705), and PD-L1. Further combination with the PAK4 inhibitor resulted in additional reductions in these proteins (compared with the PAK4-OE group) (Figures 8F,G). Similarly, silencing PAK4 (sh-PAK4) markedly reduced these protein levels, and additional treatment with Globulol further decreased the levels of PAK4, STAT3, p-STAT3 (Y705), and PD-L1 (Figures 8H–J). These findings confirm that Globulol can effectively inhibit PD-L1 expression and improve the immunosuppressive microenvironment by modulating the PAK4-pAKT-STAT3 pathway.

Immunohistochemistry analysis was employed to evaluate the expression of PAK4 protein in both normal and cancerous tissues. The findings revealed a notable upregulation of PAK4 in cancerous tissues (Figures 9A,B). The expression of PAK4 also varies across different cancer types and different pathological stages of HCC (Figure 9C). Notably, elevated PAK4 levels were significantly correlated with reduced overall survival and disease-free survival rates (Figures 9D,E). An AUC value close to 1.0 suggests a reliable detection method, with the AUC for PAK4 being 0.883, indicating its reliability as a clinical diagnostic marker for HCC (Figure 9F). The expression level of the PAK4 gene was positively correlated with genes such as STAT3, CD274 (PD-L1), and PDCD1 (PD-1) (Figure 9G). The TIMER database was used to explore the relationship between PAK4, PDCD1 (PD-1) gene expression levels and the infiltration of six types of immune cells in LC tissues, showing significant correlations (Figure 9H).