The original microarray datasets GSE45436 (19) and GSE121248 (20) of HCC were downloaded from the Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/) database (21). The sequencing platform for both microarrays were “GPL570” [HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array. The sequencing data from two microarrays of HCC tissues on the same sequencing platform were analyzed and processed using “GEOquery” and “tidyverse” package, and then merged after normalization. The ComBat function of the “sva” package was employed to eliminate batch effects (22). Principal component analysis (PCA) was used to compare data characteristics before and after normalization. While the raw dataset showed scattered distribution and ambiguous clustering, the normalized dataset exhibited significantly improved distribution and more distinct clustering patterns ( Supplementary Figure 1 ).

The expression profile data of the gene microarray were analyzed using the “limma” package in R Studio software to screen out the differentially expressed genes (DEGs). The screening criteria for DEGs were P < 0.05 and |log2 (fold-change) | > 1. Subsequently, the volcano plot and heatmap were drawn by using the “ggpubr” package and “Pheatmap” package of R, respectively.

The targets related to Tex were obtained from the GeneCards (https://www.genecards.org/) database and Online Mendelian Inheritance in Man (OMIM) (https://omim.org/) database (23). All databases used the keywords “T cell exhaustion”. Venny 2.1.0 (https://bioinfogp.cnb.csic.es/tools/venny/) (24) online tool was used to draw Venn diagram to realize the core target between ZJC, HCC and Tex.

By searching the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (25) (TCMSP, https://old.tcmsp-e.com/tcmsp.php) database, the ZJC chemical components related to related to the two Chinese herbs, namely Coptidis Rhizoma and Evodiae Fructus were retrieved. The retrieval results were screened, with parameters being the comprehensive oral bioavailability (OB) and drug-likeness (DL). OB represents the percentage of the drug reaching the systemic circulation at the same oral dose. DL is used to evaluate the “drug-likeness” degree of the required compound, which helps optimize the pharmacokinetics and drug properties of the drug, such as solubility and chemical stability (26). Subsequently, the chemical components of traditional Chinese medicine that simultaneously met the criteria of OB>30% and DL>0.18 (27, 28) were collected. Then the targets that met the criteria were selected. Moreover, these data were combined and duplicate items were deleted. Finally, the targets of traditional Chinese medicine collected from the database were obtained and regarded as the potential targets of ZJC.

The intersection targets of ZJC, HCC, and Tex were imported into the STRING (29) (https://cn.string-db.org/) database for PPI analysis, and a PPI network diagram was constructed. Subsequently, the PPI network diagram was imported into the Cytoscape software for visual analysis.

To further evaluate the immune cell infiltration in normal tissues and liver cancer tissues, we utilized xCell (30) (https://xcell.ucsf.edu/) to assess the composition of various immune cells in normal tissues and liver cancer tissues. The results of the correlation between gene expression and immune cells were presented as box plots. The xCell database integrates the advantages of gene enrichment analysis through back-folding and can evaluate 64 cell types, covering a wide range of adaptive and innate immune cells, hematopoietic progenitor cells, epithelial cells, extracellular stromal cells, and cells related to the TME. Subsequently, Spearman correlation analysis was further employed to analyze the correlation between 5 key genes and some key immune cells, and the results were displayed as lollipop plots.

To elucidate the functional roles and pathway involvement of the intersecting targets, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were performed using the R package “clusterProfiler” and The Database for Annotation, Visualization and Integrated Discovery (DAVID) (31) (https://david.ncifcrf.gov/). GO analysis encompassed three categories: biological process (BP), cellular component (CC), and molecular function (MF) (32). Terms with P<0.05 were considered statistically significant.

To identify survival-associated genes in HCC, RNA-seq expression data from tumor and normal tissues were retrieved from The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) databases using GEPIA 2 (http://gepia2.cancer-pku.cn/#index). Survival analysis was performed on HCC-related genes to evaluate their prognostic significance. Genes with statistically significant associations with survival outcomes were identified as candidate genes for further study.

Gene expression and staging analyses in normal liver tissues and HCC tissues were conducted using GEPIA 2 (33) and the University of ALabama at Birmingham CANcer data analysis Portal (UALCAN, http://ualcan.path.uab.edu/analysis.html) (34). The Kaplan-Meier Plotter database (35) (https://kmplot.com/analysis/) provides survival-related information for more than 54,000 genes in 21 types of cancer. It was used to analyze the relationship between five key genes and the overall survival (OS) of HCC patients. The screening criteria included hazard ratio (HR) within the 95% confidence interval (CI) and log-rank P value <0.05.

The GEPIA 2 (http://gepia2.cancer-pku.cn/#correlation) database integrates a large amount of gene expression data from tumors and normal tissues. In this study, gene correlation analysis was conducted using the GEPIA 2 database. During the research process, we conducted correlation analyses between cyclin-dependent kinase 1 (CDK1) and programmed death receptor-1 (PD-1), CDK1 and programmed death receptor ligand-1 (PD-L1), CDK1 and cytotoxic T lymphocyte-associated antigen-4 (CTLA-4), CDK1 and T-cell Ig and ITIM domain (TIGIT), CDK1 and T-cell immunoglobulin and mucin-domain containing-3(TIM-3), as well as CDK1 and lymphocyte-activation gene 3 (LAG-3). The relationship was determined based on the correlation coefficients and P values provided by the database.

To further verify the effect of ZJC on CDK1, we conducted a molecular docking verification between the main components of ZJC and CDK1. The 3D structure of CDK1 and the main components structure of ZJC were obtained from the RCSB PDB (http://www.rcsb.org/) and PubChem (https://pubchem.ncbi.nlm.nih.gov/) databases. Water molecules and the original ligands were removed by using the PyMOL software. Then, the target protein was imported into the AutoDock Tools 1.5.6 software for hydrogenation treatment. The results were stored in the PDBQT format. AutoDock Vina was run for molecular docking, and the results were visualized using PyMOL.

ZJC was purchased from Haihe Pharmaceutical Co, LTD (production batch number: 20191004, specification: 350mg/seeds) (Wenzhou, China) (9). Anti-PD-1 (BE0146) and anti-PD-1 (66220-1-Ig) were purchased from Bioxcell and Proteintech Group, Inc, respectively. Anti-Ki67 (HA721115) and anti-p-p70S6K (HA721803) were purchased from HUABIO. Anti-CD8 (29896-1-AP) and anti-CD163 (83285-4-RR) were purchased from Proteintech. Anti-β-actin (AC026), anti-mTOR (A2445), and anti-p-mTOR (AP0115) were purchased from Abclonal. Anti-CDK1 (CY5061), anti-eIF4E (CY8863), anti-p-eIF4E (CY5419), and anti-p70S6K (CY5365) were purchased from Abways. Goat Anti-Rabbit IgG (H+L) HRP (S0001) was purchased from Affinity. The BCA kit (P0011) was purchased from Beyotime. The ECL chemiluminescence substrate (BL520B) was purchased from Biosharp.

Hep3B2.1-7 (Hep3B) and HepG2.2.15 cells were purchased from icell and cultured in cell complete culture medium (icell-h092-001b, icell) supplemented with 10% fetal bovine serum (FBS, C04001-500, Vivacell), 100 U/ml penicillin and 100 μg/ml streptomycin at 37°C and 5% CO₂ in an atmosphere of 100% humidity.

Hep3B and HepG2.2.15 cells were seeded in 96-well plates (1×10⁴ cells/well). They were treated with ZJC (25, 50, 100, 150, 200, 300, 400μg/mL), at different concentration for 24 hours, and 48 hours, respectively. Cell viability was determined with Cell Counting Kit-8 (CCK-8, BS350B, Biosharp) according to the manufacture’s protocol. Finally, optical density was monitored IC50 values were obtained from the cytotoxicity curves using the ELx800 software at 450 nm. IC50 values were obtained from the cytotoxicity curves using the SOFTmax PRO software.

To further evaluate the anti-tumor efficacy of ZJC in vivo, a syngeneic tumor model was established in male C57BL/6 mice (4–5 weeks old, 18–20 g) purchased from Jiangsu Ailingfei Biotechnology Co., Ltd. (License No. SCXK (Su) 2023-0020). The experimental protocol was approved by the Institutional Animal Care and Use Committee of the Affiliated Hospital of Chengdu University of Traditional Chinese Medicine (No. 2021DL-019). All mice were housed in a specific pathogen-free (SPF) environment under controlled conditions (temperature: 22 °C, humidity: 65%, 12-hour light/dark cycle) with free access to food and water. After a 7-days acclimatization period, the mice were fasted for 12 hours (water permitted) prior to the experiment. This fasting step was intended to minimize the interference of food residues in the abdominal cavity on the accuracy of injection site localization during subsequent subcutaneous tumor cell injection, while ensuring that all experimental animals were in a consistent baseline state for the experiment. A total of 1×10⁷ H22 cells were subcutaneously injected into the right flank of the mice to induce tumor formation. The mice were randomly divided into four groups with five mice in each group. The control group was given normal saline by gavage. According to the rules for converting equivalent doses between mice and humans, the dosage per kilogram of body weight for mice is calculated based on 9.01 times the dosage for adults (60 kg) (36). In the low-dose ZJC group (ZJC-L), 0.364 g/kg/d of ZJC was administered by gavage. The high-dose ZJC group (ZJC-H) was given ZJC by gavage at 0.728 g/kg/d. The anti-PD-1 group was intraperitoneally injected with anti-PD-1 monoclonal antibody (10 mg/kg, dissolved in normal saline) every 3 days. Body weight and tumor volume were measured every three days. Tumor volume was calculated by the formula: V (mm³) = width² (mm²)×length (mm)×0.5 (37). During the treatment, no mice died from loading tumor. For body weight and tumor volume measurements, the researcher conducting the measurements was blinded to group allocation. After 21 consecutive days of treatment, the mice were euthanized, and tumors were excised, photographed, and weighed. All data were recorded for subsequent analysis.

All the organs were promptly removed and then washed with cold phosphate-buffered saline (PBS). The liver tissues were fixed in 4% formaldehyde for 24 hours prior to being embedded in paraffin. The thickness of the sections was 5 µm. The sliced sections were stained with hematoxylin-eosin (H&E), and the changes in histopathology were observed under a microscope (Pannoramic SCAN II).

Paraffin-embedded tissue samples were cut into sections with a thickness of 5 micrometers. Immunohistochemical staining was then carried out for anti-Ki67, anti-CD8, anti-CD163, anti-p-eIF4E, anti-p-p70S6K. The sections were developed using a 3,3-diaminobenzidine kit and subsequently stained with hematoxylin. After the staining process, the sections were differentiated. They were then rinsed with an antiblue solution, followed by dehydration and clarification steps. Finally, the sections were sealed. Ultimately, the sections were examined under a microscope at magnifications of 20× and 40× to observe the presence of brown peroxidase in the liver tissue. The immunohistochemistry (IHC) results were evaluated by two senior histopathologists. Cells with their cell membranes or cytoplasm stained in light yellow or tan colors were regarded as positive cells. The optical density of the stained cells was measured using the Image-Pro Plus 6.0 software, which was manufactured by Media Cybernetics in the United States. The results of immunohistochemistry were examined by 2 senior histopathologists using the double blind method.

Hep3B and HepG2.2.15 cells were seeded in 6-well plates at a density of 1×10⁶ cells per well and then treated as described previously. Briefly, after being washed with PBS, the cells were directly lysed using the lysate. The primary antibodies used included anti-β-actin, anti-CDK1, anti-eIF4E, anti-p-eIF4E, anti-p70S6K, anti-p-p70S6K, anti-mTOR, and anti-p-mTOR. The secondary antibodies were Goat Anti-Rabbit IgG (H+L) HRP. The protein concentration was quantified by using the BCA kit. These protein bands were detected by an enhanced chemiluminescence (ECL) detection system (Bio rad, ChemiDoc XRS⁺). The results were measured by the Image-Pro Plus 6.0 software (from Media Cybernetics).

All data were analyzed and collated using R (version 4.2.1) software and GraphPad Prism 8 software. A P-value less than 0.05 was considered to indicate a statistically significant difference. Each experiment was performed in triplicate, with a minimum of three replicates per sample. Data are presented as means ± SD. For comparisons involving more than two groups, one-way ANOVA was applied, followed by post-hoc pairwise comparisons. P<0.05 was considered statistically significant.

To identify HCC-related targets, we integrated two HCC expression microarray datasets, GSE45436 and GSE121248. GSE45436 contains 39 normal liver tissues and 95 HCC tissues, and GSE121248 contains 37 normal liver tissues and 70 HCC tissues. After batch effect correction and normalization using the R package “limma”, the merged dataset included 76 normal liver tissues and 165 HCC tissues. Differential expression analysis identified 20,815 significantly dysregulated HCC targets. Volcano plots showed up-regulated and down-regulated genes in HCC and normal tissues, while heat plots showed different expression patterns in the combined cohort ( Figures 1A, B ).

Genes related to Tex were retrieved from the GeneCards database and OMIM database, with 3,330 and 436 targets, respectively. After merging and removing duplicates, 3,726 targets related to Tex were retained ( Figure 1C ).

The effective components in ZJC (Coptidis Rhizoma and Evodiae Fructus) were screened by TCMSP. According to the screening criteria of OB> 30% and DL> 0.18, 14 compounds (11 with targets) were identified from Coptidis Rhizoma and 30 compounds (24 with targets) were identified from Evodiae Fructus. The corresponding targets were input into the UniProt database for standardization, resulting in 165 targets for Coptidis Rhizoma and 191 for Evodiae Fructus. After merging and deleting duplicate data, 193 unique targets were retained as potential therapeutic targets for ZJC. The ZJC target, HCC-related target and Tex-related target were analyzed using Venny 2.1.0, and 136 intersection targets were obtained (Figure1C). These intersection targets showed that ZJC might be a candidate mechanism for exerting anti-HCC effects by regulating Tex. A PPI network consisting of 136 targets and 2910 edges was constructed using the STRING database. This network was visualized in Cytoscape software and key genes were determined by analyzing the degree values ( Figure 1D ).

The antiproliferative effect of ZJC on liver cancer lines Hep3B and HepG2.2.15 was evaluated by CCK-8. In Hep3B cells, ZJC treatment significantly inhibited cell proliferation in a dose- and time-dependent manner. At 24 hours, compared with the control group, ZJC at concentrations of 25, 50, 100, 150, 200, 300 and 400μg/mL reduced cell viability. At all test concentrations, this inhibitory effect was further enhanced within 48 hours. In HepG2.2.15 cells, high concentrations (200,300 and 400μg/mL) of ZJC showed significant inhibitory effects at both 24 and 48 hours ( Figure 2A ). The half-maximal inhibitory concentration (IC50) of Hep3B and HepG2.2.15 cells treated with ZJC for 24 hours was approximately 310μg/mL and 530μg/mL, respectively ( Figure 2B ).

To further clarify the mechanism by which ZJC inhibits HCC, the intersection targets were further imported into the DAVID database for GO and KEGG analyses. Only the top 33 results are presented in the Figure 3 . The results in the BP of GO indicated that the intersection targets were mainly enriched in GO:0045944~positive regulation of transcription from RNA polymerase II promoter, GO:0045893~positive regulation of transcription, DNA-templated, etc. In the CC of GO, the intersection targets were mainly enriched in GO:0005634~nucleus, GO:0005829~cytosol, etc. In the MF of GO, the intersection targets were mainly enriched in GO:0005515~protein binding, GO:0042802~identical protein binding, etc. ( Figure 3A ). KEGG pathway analysis showed that the intersection targets were mainly enriched in hsa04150: mTOR signaling pathway, hsa04621: NOD-like receptor signaling pathway, hsa04151: PI3K-Akt signaling pathway, etc. ( Figure 3B ).

To further identify the key pathways of ZJC in the treatment of HCC, in vitro experiments were carried out for verification. The protein results showed that in Hep3B cells, compared with the control group, the mTOR inhibitor (INK128) and ZJC decreased the expression of p-mTOR, as well as the expression of two key molecules downstream of the mTOR pathway, p-eIF4E and p-p70S6K. When ZJC was used in combination with the mTOR activator (MHY1485), this inhibitory effect was restored (38–40) ( Figure 4A ). Similarly, in HG2215 cells, compared with the control group, INK128 and ZJC decreased the expression of p-mTOR, as well as the expression of p-eIF4E and p-p70S6K. When ZJC was combined with MHY1485, this inhibitory effect was restored ( Figure 4B ).

Subsequently, to further identify the key targets of ZJC in the treatment of HCC, we obtained 500 candidate targets using the GEPIA 2 database, and these targets were different survival-related genes. By taking the intersection, 5 survival-related key targets were obtained, including CDK1, checkpoint kinase 1 (CHEK1), secreted phosphoprotein 1 (SPP1), baculoviral inhibitor of apoptosis repeat containing 5 (BIRC5), and matrix metalloproteinase-1 (MMP1) ( Figure 5A ). We further explored the expression of these genes. The results from the UALCAN database showed that the expression of the 5 key targets was upregulated in cancer tissues ( Figure 5B ). The results also indicated that in the GEPIA 2 database, except for stage IV, the expression of these 5 key genes increased with the increase of the stage ( Figure 5C ). Kaplan-Meier Plotter survival analysis demonstrated that high expression of these 5 key genes was associated with poor OS ( Figure 5D ). Studies have shown that mTOR activation promotes the upregulation of CDK1 (41), and rapamycin downregulates the expression of CDK1 by inhibiting the mTOR/p70S6K pathway (42). The results of in vitro experiments showed that compared with the control group, INK128 and ZJC decreased the expression of CDK1. The expression of CDK1 was restored after the combined use of ZJC and MHY1485 ( Figures 5E, F ). The GEPIA 2 database showed that the expressions of CDK1 and PD-L1 were positively correlated (P = 2.9E-09, r = 0.28) ( Figure 5G ). Through the TCMSP database, we obtained multiple components of ZJC. Among them, the top 4 components in terms of degree values were quercetin, beta-sitosterol, berberine and isorhamnetin. It is generally believed that a docking score of < 0 kcal/mol indicates that the component and the target can spontaneously bind, a score of -4.25 kcal/mol represents good docking affinity, and a score of -7 kcal/mol is considered to have a very strong docking affinity (43). The molecular docking results show that quercetin, beta-sitosterol, berberine and isorhamnetin have good interactions with CDK1. The detailed information of molecular docking is shown in Supplementary Table 1 . Quercetin forms three hydrogen bonds with CDK1 at the amino acid residues ASP-86, LEU-83. Isorhamnetin forms ten hydrogen bonds with CDK1 at the amino acid residue SER-166, ARG-23, SER-182, GLY-154, LEU-125, ARG-151 ( Supplementary Figure 2 ).

To further evaluate the anti-tumor effect of ZJC in vivo, we established a mouse syngeneic tumor model using H22 cells. Compared with the control group, the ZJC-L, ZJC-H, and anti-PD-1 groups inhibited tumor weight and volume ( Figures 6A, C, D ). This result demonstrated that both ZJC treatment and anti-PD-1 treatment exerted an inhibitory effect on tumor growth. There was no significant difference in the final body weight of the mice at the end of the experiment ( Figure 6B ). In addition, there were no significant changes in liver weight, spleen weight, liver index, and spleen index among the four groups of mice ( Figures 6E-H ). In summary, our research results indicated that ZJC inhibited the growth of tumor.

The results of H&E staining showed that in the control group, the cells in the tumor tissue were arranged closely and disorderly, with large and deeply stained nuclei and obvious atypia. In the ZJC-L, ZJC-H and anti-PD-1 group, the tumor tissue cells were arranged loosely, and the nuclear atypia was significantly decreased. The immunohistochemical staining results of Ki67 also showed that compared with the control group, the proportion of Ki67-positive cells in the ZJC-L group, ZJC-H group, and anti-PD-1 group was significantly lower than that in the control group. This showed that the proliferative activity of tumor cells was reduced, further demonstrating the inhibitory effects of ZJC and anti-PD-1 on tumor growth ( Figure 6I ).

These results showed that there were differences in the expression of p-eIF4E and p-p70S6K in tumor tissues of different groups. In the control group, the activity of p-eIF4E in tumor cells was relatively high. In contrast, in the ZJC-L, ZJC-H, and anti-PD-1 groups, the intensity of p-eIF4E positive staining was weakened, and the number of positive cells decreased. In the control group, p-p70S6K also showed strong positive staining, and a large number of brown positive signals could be seen in the tumor cells. In the ZJC-L, ZJC-H and anti-PD-1 groups, the positive staining intensity of p-p70S6K was weakened, the number of positive cells decreased, and the expression of p-p70S6K was inhibited ( Figure 7 ).

In addition, to further clarify the characteristics of immune cell infiltration in normal liver tissues and HCC tissues, we evaluated the infiltration ratios of immune-related cells and the correlations among immune cells. The results of the stacked bar chart indicated that compared with normal tissues, in HCC tissues, the proportion of M1 macrophages decreased, while the proportions of M2 macrophages and exhausted T-cells increased ( Figure 8A ).

To further clarify the correlation between the target genes and immune cells, we conducted a correlation analysis between the top five key genes and immune cells. The results of the immune infiltration analysis indicated that CDK1 had a significant negative correlation with naïve CD4⁺ T cells (P = 6.0×10^-3, r = -0.14), and significant positive correlations with activated memory CD4⁺ T cells (P = 5.7×10^-3, r = 0.14) and M0 macrophages (P = 5.9×10^-5, r = 0.21). CHEK1 showed a significant negative correlation with resting memory CD4⁺ T cells (P = 4.0×10^-2, r = -0.11), and significant positive correlations with activated memory CD4⁺ T cells (P = 3.5×10^-3, r = 0.15) and M0 macrophages (P = 1.3×10^-4, r = 0.20). SPP1 had significant negative correlations with CD8⁺ T cells (P = 2.0×10^-2, r = -0.13) and naïve CD4⁺ T cells (P = 5.0×10^-3, r = -0.15), and significant positive correlations with M0 macrophages (P = 4.4×10^-12, r = 0.35) and M2 macrophages (P = 4.0×10^-2, r = 0.11). BIRC5 had significant negative correlations with naïve B cells (P = 1.2×10^-3, r = -0.17), naïve CD4⁺ T cells (P = 1.0×10^-2, r = -0.13), and resting memory CD4⁺ T cells (P = 8.6×10^-3, r = -0.14), and significant positive correlations with memory B cells (P = 5.4×10^-3, r = 0.15), activated memory CD4⁺ T cells (P = 4.2×10^-4, r = 0.18), and M0 macrophages (P = 5.9×10^-6, r = 0.24). MMP1 had a significant negative correlation with resting memory CD4⁺ T cells (P = 4.0×10^-2, r = -0.11) and a significant positive correlation with M0 macrophages (P = 4.9×10^-5, r = 0.21) ( Figure 8B ).

The results showed that there were significant differences in the number of CD8 and CD163 among different treatment groups (control, ZJC-L, ZJC-H, anti-PD-1). There were fewer CD8 positive cells in the control group, while there were more positive cells in the ZJC-L, ZJC-H and anti-PD-1 groups. This indicated that ZJC and anti-PD-1 promoted the expression of CD8 cells ( Figure 9A ).

Previous studies have shown that high levels of M2-specific CD163 are associated with an increase in tumor nodules and poor prognosis in HCC patients (44). In this study, CD163 was used to label M2-type macrophages. The results showed that the positive cells of CD163 in the control group were relatively higher. Compared with the control group, there were fewer positive staining in the ZJC-L, ZJC-H and anti-PD-1 groups. This indicated that ZJC and anti-PD-1 inhibited the expression or aggregation of CD163-positive cells ( Figure 9A ).

Exhausted T cells often express high levels of inhibitory receptors, including PD-1, CTLA-4, TIGIT, TIM-3, LAG-3 (45–47). They may also express multiple inhibitory receptors simultaneously and suppress the T-cell immune response to tumor antigens (45–47). Therefore, we further analyzed the relationship between CDK1 and these inhibitory receptors. The GEPIA 2 database showed that there is a positive correlation between CDK1 and PD-1 (P = 5.6×10^-6, r = 0.22), CDK1 and LAG-3 (P = 1.3×10^-7, r = 0.25), CDK1 and CTLA-4 (P = 7.5×10^-8, r = 0.26), CDK1 and TIGIT (P = 1.3×10^-7, r = 0.25), CDK1 and TIM-3 (P = 1.9×10^-3, r = 0.15) ( Figure 9B ).

The results of immunohistochemical staining showed that in the tumor tissues of the control group, the number of PD-1 positive staining was relatively large. Compared with the control group, the positive expression of PD-1 in the ZJC-L, ZJC-H and anti-PD-1 group also showed a downward trend in PD-1 expression ( Figure 9C ). These results showed that ZJC improves the tumor immune microenvironment in HCC.