Comprehensive analysis of the correlations of S100B with hypoxia response and immune infiltration in hepatocellular carcinoma

Cell culture

The human HCC cell line HepG2 and the immortalized human hepatic cell line HL-7702 (L02) were obtained from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). The cells were cultured with DMEM (Gibco, Waltham, MA, USA) including 10% fetal bovine serum (FBS) (Gibco, Waltham, MA, USA), 10 μg/mL streptomycin, and 10 μg/mL penicillin (Sigma, St. Louis, MO, USA). All cells were grown in an incubator at 37 °C in 5% CO₂.

Plasmid construction and transfection

To construct a plasmid containing shRNA to inhibit target gene expression. The shRNA sequences of S100B and HIF-1α were obtained from online websites (https://rnaidesigner.thermofisher.com/rnaiexpress). Subsequently, the annealed oligonucleotides of shRNA primers were subcloned into the PGPU6/GFP/Neo vector (Gene Pharma, Shanghai, China). These plasmid constructions were synthesized by the gene pharma company. Plasmids expressing NC and shRNA were designated, shNC, shS100B, and shHIF-1α, respectively.

In cell transfection assays, when the cell density was reached to 70%, Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA)-mediated cell transfection was executed according to the manufacturer’s protocol. The Lipofectamine 2000 (5 μL) was diluted in 250 μL DMEM, and then mixed well with plasmid. Subsequently, the above solutions were added to the cell culture and placed in the incubator 5% CO₂ at 37 °C. The efficiency of genetic silencing by the shRNA was evaluated by quantitative PCR (q-PCR).

The q-PCR analysis

RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) protocol. cDNA was obtained according to the protocol of the Prime Script™ RT Master Mix Kit (Takara, Kusatsu, Japan), and q-PCR assay was performed using SYBR Premix Ex Taq™II (Takara, Kusatsu, Japan) on a Thermal Cycler CFX6 System (Bio-Rad, Hercules, CA, USA). The relative gene expression levels were assessed using the 2−ΔΔCt method.

Western blot analysis

Total proteins were extracted from cells after transfection for 48 h using RIPA buffer with proteinase, and then BCA protein assay (Thermo Fisher Scientific, Waltham, MA, USA) were performed to assessed protein concentrations. The different concentrations of SDS-PAGE were used to separate target proteins. Subsequently, it was transferred on PVDF membranes and blocked with 5% nonfat milk. Protein-bound membranes were incubated with primary and secondary antibodies, respectively. The ECL substrates was used to detect the signals and photo record (Millipore, Burlington, MA, USA).

Hypoxia response element

We obtained the genomic sequence of human S100B from NCBI and screened for the presence of HREs 2,000 bp upstream of the TSS sequence. The HRE sequence was “(A/G)CGT(G/C)”.

Luciferase reporter assay

The luciferase reporter assay related plasmids were transfected into HepG2 cells. Subsequently, the cells were washed with PBS and treated using the Luciferase Reporter Assays Substrate Kit (ab228530; Abcam, Cambridge, MA, USA). Then, the Dual Glo Luciferase Assay System (Promega, Madison, WI, USA) and GloMax Multi Detection System (Promega, Madison, WI, USA) were used to confirm their luciferase activity. Their negative control was Renilla luciferase activity.

Chromatin immunoprecipitation (ChIP) assay

All collected cells were performed the assays according to the protocol from the Simple ChIP Enzymatic Chromatin IP Kit (Cell Signaling Technology, Danvers, MA, USA). DNA binding chromatin fragments were incubated with anti-HIF-1α antibodies and IgG (Sigma, St. Louis, MO, USA) and then the antibody and protein conjugate were immunoprecipitated with Protein G magnetic beads. Subsequently, the immunoprecipitated DNA was collected. The enrichment of particular DNA sequences was analyzed by q-PCR.

Cell viability assay (MTT assay)

Cell proliferation was measured by an MTT assay (Roche Diagnostics, Basel, Switzerland). The cells were digested and collected after transfection for 24, 48, 72, and 96 h, subsequently cultured in medium with 0.5 mg/ml MTS and kept in the dark for 4 h. Cell viability was assessed by the OD (Optical Density) value at 490 nm by an EnSpire Multimode Plate Reader (PerkinElmer, Suzhou, Japan).

Colony formation assay

The transfected cells were collected after digestion with trypsin. Then, the cells were seeded onto a six-well plate with a density of 1,000 cells/well. The cells grow in a humidified incubator (5% CO₂, 37 °C). After routine incubation for 14 d, the colonies were stained with 1% crystal violet for 30 min, after which the number of colonies formed was counted.

Cell migration assay

After transfection for 24 h, the cells were scratched with 100 μL pipette tips, and lines were drawn in parallel. Subsequently, they were washed out with PBS buffer twice times. After 48 h of wound formation, the wound size was photographed and measured. The wound healing rate was calculated as follows: wound healing rate = [(scratch width at 0 h) − scratch width at 48 h/(scratch width at 0 h)] × 100%.

Trans well invasion assays

The cells were starved for 6 h in serum-free medium, trypsinized and adjusted 5 × 10⁴ cells after counting. A total of 200 µL of the cell suspension was in the upper chamber, while 600 μL medium containing 10% serum was in the lower chambers. Then, the migrated cells in the lower chambers were fixed with 4% paraformaldehyde and stained with 0.5% crystal violet. The stained cells were observed under a Carl Zeiss microscope system (ZEISS, Jena, Germany) and recorded cell numbers after 24 h later.

S100B expression-related genes analysis in LIHC

The S100B expression related genes were obtained using the LinkedOmics based on TCGA_LIHC cancer cohort (HiSeq RNA platform) data. GO (Gene Ontology) according to biological processes, and KEGG (Kyoto Encyclopedia Genes and Genomes) analysis based on gene set enrichment analysis was performed in TCGA_LIHC. In addition, the relationships between S100B expression and cancer pathways activation in LIHC were assessed via GSCA database (Liu et al., 2018a).

Construction of PPI (protein–protein interaction) networks

The STRING database was used to analyze PPI networks between S100B and EMT-related genes. The networks were constructed by Cytoscape version 3.8.0. Furthermore, S100B directly interacting proteins were predicted using GeneMANIA.

The immune cell infiltration analysis

HPA (Human Protein Atlas) was used to analyze S100B level in normal live cells based on single cell sequencing results. Genes analyze module of CancerSCEM (Cancer Single-cell Expression Map) was used to confirm the expression of S100B in different cell subtypes based on single cell RNA-Seq datasets. S100B expression related tumor-infiltrating immune cell signatures in LIHC (Liver Hepatocellular Carcinoma) was investigated through TIMER (Tumor IMmune Estimation Resource) and GSCA (Gene Set Cancer Analysis) database (Li et al., 2017). The relationship between S100B expression and immune-related molecules were explored using TISIDB (Tumor-Immune System Interactions) portal based on the TCGA data (http://cis.hku.hk/TISIDB/) (Ru et al., 2019). Potential effects of S100B expression and immune cell infiltration level on overall survival (OS) were evaluated in TIMER2.0 via Kaplan–Meier plotter in HCC.

Statistical analysis

All experiments were independently performed at least three times. Their results were expressed as the mean ± SD. Two-tailed Student’s t-test was performed for comparing the data between two groups. Correlations were analyzed using Pearson’s correlation coefficient analysis. Differences were considered as statistically significant P values < 0.05.

S100B-related functions enrichment analysis

To confirmed the mechanism of S100B during HCC progression, we firstly explored its functions through online tool TISIDB. As shown in Table 1, biological processes, such as response to hypoxia and I-kappaB kinase/NF-kappaB signaling, were remarkably regulated by S100B. In molecular function, S100B was involved in protein binding, calcium ion binding and RAGE receptor binding. In addition, pathways, including TLR4 signaling, glycosylation endproduct receptor signaling and immune related pathways, were found to be associated with S100B alterations. Therefore, we focused on the functions of S100B in hypoxia responses and immune related signaling pathways in liver cancer.

Hypoxia enhances S100B expression via HIF-1α signaling in HCC

To further confirm the function of S100B in hypoxia biological process, the expression of S100B under hypoxic conditions was assessed. We found that S100B was significantly upregulated in HepG2 cells under hypoxic conditions (Figs. 1A, 1B). It is well known that HIF-1 is a master regulator of the hypoxia response in many cancers. We also fund that HIF-1α was upregulated in HepG2 cells under hypoxic conditions (Fig. 1C). To assess whether HIF-1α is a driver factor of S100B upregulation in hypoxia, we constructed the shHIF-1α to knock down its expression in HepG2 cell (Fig. 1D). Furthermore, hypoxia-induced S100B expression could be abolished after knockdown of HIF-1α in HepG2 cells under hypoxic conditions (Figs. 1E, 1F). Therefore, 100B is a direct transcriptional target of HIF-1α in HCC.

Since HIF-1α binds to HREs in the promoters of hypoxia-responsive genes to induce their transcription, we analyzed the potential HREs in the S100B promoter. Two potential HREs were found in the S100B promoter (Fig. 2A). According to the HREs locations, three S100B promoter expression vectors, including promoter 1 containing two HREs and two truncated mutations, promoter 2 containing one HREs and promoter 3 no HREs were constructed and the dual luciferase reporter assays were performed. As expected, HIF-1α overexpression dramatically enhanced S100B promoter activity but did not affect its activity of truncated promoter with deletion of two HREs (Fig. 2B). In addition, ChIP assays also confirmed that HIF-1α bound to the HREs of the S100B promoter. Moreover, HIF-1α also enriched in the S100B promoter under hypoxic conditions (Fig. 2C). Further regression analysis using cBioPortal revealed that S100B mRNA level was positively correlated with HIF-1α expression in liver cancer (Fig. 2D).

Hypoxia-related HepG2 cell migration and invasion partially depend on S100B

It is well known that the hypoxic condition drives EMT leading to aggravate cancer progression in solid tumors. We analyzed the biological functions of S100B in HepG2 cells under hypoxic conditions. A wound healing assays showed that the migratory ability of HepG2 cells was significantly increased under hypoxic conditions (Figs. 3A, 3B). We then assessed the migratory ability of HepG2 cells with S100B knockdown (Figs. S1A, S1B). As expected, the migratory rate was significantly hindered following downregulation of S100B, which is similar to the inhibitory effect observed in HIF-1α downregulated HepG2 cells (Figs. 3C, 3D). Consistent with the cell migration results, the invasion assays also showed that hypoxia’s promoting effect was eliminated by HIF-1α and S100B knockdown (Figs. 3E–3H). These results indicated that knockdown of S100B may inhibit hypoxia-induced HepG2 cell migration and invasion.

The enrichment analysis and co-expression profiles of S100B in HCC

We further investigated the functions of S100B and its co-expressed genes using GSVA enrichment analysis in HCC. We found that S100B expression was positively correlated with epithelial mesenchymal transition (EMT) and DNA Damage activity, and negatively correlated with TSC mTOR activity in LIHC (Fig. 4A). To confirm the functional downstream targets of S100B in HCC, the co-expressed genes of S100B were identified in LIHC. The results showed that 538 genes were positively correlated with S100B mRNA level (R > 0.5, P < 0.05), while no negatively correlated genes (R < −0.5 P < 0.05) were found in LIHC (Table S1). The top 50 genes that were significantly associated with S100B expression in LIHC were showed by heat map (Figs. S2A, S2B). Furthermore, the top 50 S100B-related genes were also positively associated with EMT activity in LIHC (Fig. 4A). Therefore, S100B is involved in EMT process of HCC cell.

Subsequently, we obtained the hallmark EMT-related genes set from MSigDB. Veen analysis showed that eight EMT-related genes, such as BASP1, Col8a2, EMP3, ENO2, PLAUR, TGFB1, VIM, and WIPF1, were positively correlated with S100B expression in LIHC (Fig. 4B). Moreover, PPI network analysis results indicated that S100B only directly interacted with ENO2, and indirectly interacted with EMP3, VIM, and TGFB1 (Fig. 4C). Furthermore, the PPI network was also constructed via GeneMANIA. As shown in Fig. 4D, S100B had a positive correlation with 20 genes, such as CACYBP, CAPZA1/2, HMGB1, ANXA6, and other S100 members (S100A1, S100A11, S100P, S100A6, S100A12, and S100A4). Therefore, S100B likely promote cells invasion and migration to regulate these EMT-related genes in HCC progression.

S100B expression is involved in tumor immune infiltration of HCC

Function enrichment analysis of S100B associated gene sets were performed in HCC. We found that they were mainly involved in the biological processes, including adaptive immune response, leukocyte differentiation and activation, immune response-regulating signaling pathway, and response to chemokine (Fig. 5A). KEGG results indicated that these genes were primarily enriched in allograft rejection, Th17 cell differentiation, NF-kappa B pathway, T cells and B cells receptor, and chemokine signaling pathways in HCC (Fig. 5B). This result is consistent with the pathways analysis results through Reactome database, suggesting that S100B may be involved in regulating of inflammation and immune response in HCC.

Next, we investigated the relationship between S100B expression and infiltration levels of different immune cell subtypes in HCC. A significantly positive correlations were found between mRNA level of S100B and the degree of various immune cells infiltration in HCC. As shown in Fig. 5C, S100B was significantly correlated with four immune cells (cor > 0.45), including B cells, CD8+ T cells, macrophage cells, and myeloid dendritic cells. It was weakly associated with CD4+ T cells and neutrophil cells in HCC. Furthermore, we also investigated their correlations via the GSCA database. A positive relationship was found between S100B expression and the immune infiltration of CD8+ T cells, NK cells, DC cells and macrophage cells in HCC (Fig. 5D). Therefore, all the results suggested that S100B may be involved in immune cells infiltration of HCC.

S100B expression is involved in the functional network of immune molecules

We further investigated the relationship between S100B expression and immune molecules in HCC. S100B was positively associated with most of immune molecules in HCC. S100B expression had significant positive correlations with immuneostimulators in HCC (Figs. 6A, 6B). The immune-inhibitors, containing CTLA4 (R = 0.639), TIGIT (R = 0.638), and HAVCR2 (R = 0.631), and immune-stimulators, including CD48 (R = 0.693), CD86 (R = 0.673), and TNFRSFB (R = 0.66), were positively correlated with S100B expression, respectively (Figs. S3A, S3B). Moreover, it was closely associated with MHC genes (Fig. 6C), especially HLA-DPB1 (R = 0.623), HLA-DOA (R = 0.62), and HLA-DPA1 (R = 0.593) (Fig. S3C). Additionally, the positive relationships also were found with many chemokines and all listed chemokine receptors (Figs. 6D, 6E). The top three chemokines included CCL5 (R = 0.598), CCL22 (R = 0.59), and CCL4 (R = 0.57), and the top three chemokines’ receptors were CCR5 (R = 0.652), CXCR3 (R = 0.598), and CXCR4 (R = 0.591) (Figs. S3D, S3E). Additionally, the correlation of S100B expression and immune checkpoints, such as PDCD-1, CTLA4, HAVCR2, and TIGIT were confirmed in HCC. The results showed that expression of these genes was significantly correlated with S100B expression in HCC (Figs. 6F–6I). Thus, S100B expression maybe result in the migration of immune cells in HCC tumor microenvironment.

S100B expresses in immune cells at the tumor tissues of HCC

Hypoxia is a common characteristic of solid tumors. To further explore S100B expression in immune cells at the hypoxia microenvironment, we assessed its expression in immune cells based on the single cell sequencing results. HPA database showed that S100B was mainly centralized in T cells and Kupffer cells in normal liver (Figs. 7A, 7B). Subsequently, the expression of S100B in different cell subtypes was further assessed via CancerSCEM database based on the single-cell RNA-Seq results in HCC. We found that S100B was observed in the immune cell at the tumor tissues, especially NK cell in HCC (Figs. 7C, 7D).

In addition, the prognosis of patients with HCC was investigated based on S100B mRNA expression and different immune cells infiltration levels. The OS outcome of low S100B/ high CD8+ patients were significantly better than other groups (Fig. 7E). While higher infiltration levels of CD4+Th2, macrophage, neutrophil, and DC cells with high S100B expression were connected with shorter OS (Figs. 7F–7I). However, there was no significant different was found in the group with B cells infiltration and S100B expression. Therefore, these results implicated that S100B may affect the prognosis of patients with HCC in part due to immune cells infiltration.

The hub genes of S100B co-expression and hypoxia response in HepG2 cell are associated with immune infiltration

To validate how up-regulation of S100B by hypoxia might be correlated to the immune cell infiltration, we performed an association analysis between the S100B significantly co-expressed genes and the differential expression profiles of HepG2 cells in hypoxia response based on the microarray data of GSE15366. Venn diagrams showed that 53 genes were differentially expressed in the hypoxia and associated with S100B expression in HCC (Fig. 8A). These genes were showed in Table S2. Furthermore, enrichment analysis results revealed that these genes were strongly associated with immune-related terms/pathways, such as regulation of T cell activation, adaptive immune response, modulators of TCR signaling and T cell activation, mast cell activation, and neutrophil degranulation (Fig. 8B). Moreover, we found that the expression of these gene sets has a closely association with tumor immune cell infiltration (Cor = 0.78), including Tfh, Th1, CD8 T cell, exhausted cell, NK cell, macrophage, DC, and CD4 T cell (Fig. 8C). Due to hypoxia enhances S100B expression via HIF-1α signaling in HCC, HIF-1α-S100B axis related differentially expressed genes were identified in HCC. As shown in Table S2, 34 genes were obtained in two groups. They also involved in regulation of immune response. Next, we used the GSCA database to analyze the correlation between their expression and immune infiltrating cells in LIHC. The results showed that the expression of these genes was positively associated with the degree of infiltration of immune cells (Cor = 0.59), including central memory, macrophage, Tfh, CD4 T cell, DC and iTreg (Fig. 8D). Therefore, these results suggested that up-regulation of S100B by hypoxia might regulate these genes expression to affect the immune cell infiltration of LIHC.