ZNF320 is a hypomethylated prognostic biomarker involved in immune infiltration of hepatocellular carcinoma and associated with cell cycle

Data collection and processing

Hepatocellular carcinoma (HCC) clinical and mRNA expression level was collected from the TCGA Database. In terms of the gene expression profile, this study included 374 LIHC samples and 50 normal samples, and the data type of mRNA expression profile was HTSeq-FPKM. We obtained clinical information from 377 patients. In addition, RNA-seq data and clinical information were also acquired from the ICGC website (https://dcc.icgc.org/projects/LIRI-JP). [LINC-JP] Liver Cancer - NCC, JP datasets included 202 normal samples and 243 tumor samples [14].

Patients and tumor specimens

Human HCC tissues and corresponding adjacent tissues were obtained from 35 patients undergoing nephrectomy in the Second Affiliated Hospital of Nanchang University from January 2018 to January 2021. The patients’ informed consent was obtained. At the same time, the research ethics committee of Second Affiliated Hospital of Nanchang University agreed to the experiment.

TIMER database analysis

The Tumor Immune Estimation Resource (TIMER) is a synthetic web used for analyzing the levels of immune invasion in diverse cancers, which includes 32 cancer types. In this work, we use “Diff-Exp module” to confirm the expression of ZNF320 in diverse tumors. Then the “gene module” was used to estimate the relationship of ZNF320 with immune infiltration. Moreover, the “SCNA module” was used to compare the level of tumor invasion among tumors with different somatic copy number changes for ZNF320. Finally, we considered the correlation between ZNF320 expression in TIICs and immune markers by using related modules. The difference between infiltration level and normal level was evaluated for each SCNA category through Wilcoxon rank-sum test. Finally, with the help of the “Correlation module”, considering the Spearman’s rho value and predicted statistical implications, we verified the relationship between ZNF320 and tumor infiltrating immune cell markers in HCC [15].

Immunohistochemistry

Cut formalin-fixed, paraffin-embedded HCC tissue into 4 μm-thick sections. After deparaffinization, rehydration, and microwave heating in microwave-heated antigen unmasking solution (EDTA, pH8.0) to repair the antigen, 30 minutes in total was used to seal the sections with goat serum. Next, incubate the sections overnight with anti-ZNF320 monoclonal antibody (RRID 24882-1-AP, Abcam, 1:500 dilution) at 4° C. After that, HRP-conjugated secondary antibody (Boster) was allowed to stand at room temperature for 2 hours. Subsequently, the two-step method (catalog no.: PV-9000; ZSGB-BIO Co., Ltd., Beijing, China) were used for immunostaining. Finally, three pathologists who are unaware of the clinical parameters assessed the staining intensity and the percentage of positive cells semi-quantitatively.

UALCAN database analysis

UALCAN database is a tumor data on-line analysis website that provides comprehensive cancer transcriptome and clinical patient data (extracted from TCGA) [16]. In our research, we assessed ZNF320 expressions by using the “Expression Analysis” module. We also studied the correlation between the ZNF320 expression and clinicopathological features.

LinkedOmics analysis

LinkedOmics, a comprehensive online site, is usually chosen to analysis multidimensional data within and across 32 kinds of cancer. Using it, we succeeded in mining the co-expressed genes linked to ZNF320 in the TCGA LIHC database through the results of analysis. Volcano plots and heat maps provided strong evidence for this. We define the top 500 related genes as co-expressed genes.

DAVID

Database for Annotation, Visualization and Integrated Discovery (DAVID) 6.8 is a comprehensive, functional annotation website, which enables us to study the biological functions and signal pathways of specific gene sets. Gene annotation includes Gene Oncology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis [17].

DNA methylation

The cBioPortal database provides visualization, analysis, and download of large-scale tumor genome data for a variety of cancers. We used it to study the association between ZNF320 promoter methylation level and expression in HCC with “Plot module” [18].

Gene set enrichment analysis (GSEA)

We used GSEA to conduct a group study on the RNA-sequencing data of 374 TCGA-LIHC patients downloaded from Genomic Data Commons, analyzing the pathways associated with ZNF320 expression in HCC specimen. The number of permutations: 1000; normalized enrichment score (NES) >1; false discovery rate (FDR) <0.25, p-val<0.05 [19].

PPI network construction

STRING database is an online tool for evaluating protein interactions in multiple ways. In our research, the top 500 genes with co-expression coefficients greater than 0.4 were collected from the STRING database. They were evaluated by Cytoscape 3.8.2 and its plug-in MCODE (Molecular Complex Detection). And the selection criteria are as follows: Max depth=100, node score cutoff=0.2, K-core=2 [20].

Validation of hub genes

Hub genes are highly connected to the nodes in the module and have been certificated to play an important role in function. The significance of the genes was measured by the absolute values of the Spearman’s correlation in our study. On the basis of the results, the top 500 genes with confidence > 0.9 were uploaded to the STRING database to construct protein-protein interaction (PPI). Then, we used Cytoscape 3.8.0 and its plug-in MCODE to evaluate the top 500 genes. Furthermore, a standard for hub genes (yellow nodes) was set with a degree cut-off = 2, node score cut-off = 0.2, k-core = 2, and max. depth= 100, which was screened with MCODE. In total, there were 6 genes scoring the highest and initially defined as hub genes. Ultimately, after digging out their backgrounds, three genes, closely related to the cell cycle, were considered to be a “real” hub gene among these genes.

GEPIA analysis

GEPIA [21], an interactive web server containing RNA sequencing data based on the TCGA and GTEx databases, which analyzed the mRNA expression. In this study, GEPIA was used to compare tumors with normal tissues, analyze pathological stages, and analyze m6A related prognosis. We used Spearman method to ascertain the correlation coefficient of relevance [22].

Statistical analysis

All statistical analysis in our work was done by R software (version 3.6.3). The detection of different ZNF320 expression levels between LIHC samples and normal samples was realized by using the “limma” and “beeswarm” packages of “R” and rank sum test method. We probed the relevance between ZNF320 expression and clinicopathological feature by Wilcoxon signed-rank test or Kruskal-Wallis test. Then, Kaplan-Meier survival curve using log-rank test to explore the survival distribution among patients with different levels of expression was drawn to evaluate whether the differential genes expression had a significant influence on the prognosis of patients (p<0.05). Univariate and Multivariate Cox regression analysis screened factors significantly related to prognosis (p<0.05) (Cox model uses the “survival” and “survminer” packages of “R”). Lastly, we used the ROC curve drawn by “survival ROC” to evaluate the predictive ability of ZNF320 expression level on one-year, three-year, or five-year survival. T-test was used to study the different expression of m6A interrelated genes between ZNF320 high and low expression groups.

Data availability

The expression profiling of genes from diabetics and diabetics nephropathy can be available from the GEO (GSE142153 and GSE26168). The other data can be obtained from the corresponding author (Bo Zhang).

ZNF320 is over expressed in HCC

To compare the differential expression of ZNF320, we first studied the expression of ZNF320 in LIHC and other tumors by using TIMER database. Expression of ZNF320 in liver hepatocellular carcinoma was predicted to be vitally upregulated (Figure 1A). Next, we used TCGA sequencing data to explore ZNF320 expression in 374 HCC samples and 50 normal samples. Our result revealed ZNF320 expression was vitally upregulated in HCC (Figure 1B). Analysis of the paired samples revealed a notable rise of ZNF320 expression in HCC tissues (n =50) (Figure 1C). What’s more, we used ICGC, another database, to do the same analysis (Figure 1D), and got a similar consequence with our preliminary result in TCGA. To further check the ZNF320 expression in HCC, we compared 35 HCC tissue samples with para cancerous samples by IHC analysis. Our results showed that ZNF320 expression protein in HCC tissues was conspicuously higher than that of adjacent tumor tissues (Figure 1E, 1F). In sum, these outcomes verified that ZNF320 is conspicuously overexpressed in HCC tissues.

Figure 1. ZNF320 is over expressed in HCC tissues. (A) The mRNA level of ZNF320 in 33 kinds of tumor types by TIMER. (*p<0.05,**p<0.01,***p<0.001). (B) Expression levels of ZNF320 were higher than the corresponding normal tissues in HCC samples (p=5.282e-06). (C) ZNF320 expression in HCC group downloaded from TCGA RNA-seq dataset (p=5.637e-06). (D) The mRNA expression level of ZNF320 in tumor and normal tissues in the UALCAN (E, F) Typical images of immunohistochemistry (IHC) in 35 pairs of HCC tissues showing the protein expression of ZNF320 in HCC and adjacent nontumor tissues.

Association of ZNF320 expression and clinicopathological characteristics in HCC patients

After the expression of mRNA and protein were found to increase in HCC patients, we next used UALCAN to integrate various clinic factors of HCC samples, including patients’ individual tumor grade, cancer stages, gender nodal metastasis, age, and weight. Then we compare the expression levels of ZNF320 in each group. The result exhibited that with grade increasing, the patients had statistically higher expression of ZNF320. (Figure 2A) Additionally, ZNF320 were associated with patients’ cancer stages, and patients in more advanced cancer stages has a tendency to have higher expression of ZNF320. (Figure 2B) We also caught that the expression of ZNF330 increased in tumor tissue with nodal metastasis. (Figure 2C) The analysis result revealed significant differences between male and female. (Figure 2D) What’s more, compared with the normal subjects, older and heavier patients maintained a high expression level of ZNF320. (Figure 2E, 2F) The above findings further prove that ZNF320 is increased expression in HCC and related to clinicopathological characteristics.

Figure 2. ZNF320 expression is a correlation with clinicopathological characteristics. Increased ZNF320 expression was significantly with (A) higher grade, (B) stage, (C) nodal metastasis, (D) gender, (E) age, and (F) weight.

ZNF320 expression is an independent prognostic factor which is associated with poorer prognosis of HCC patients

To further find the potential role of ZNF320 expression in TCGA patients with Hepatocellular Carcinoma, we used Kaplan Meier survival method for survival analysis. We found via the KM plotter that high expression of ZNF320 patients had shorter OS and DSS times than low expression (OS:HR= 1.82, P<0.01; DSS:HR=1.66, P<0.05) (Figure 3A, 3B). The results also manifested that patients with high ZNF320 expression had a obviously shorter OS time proportionate to low expression group, which meant that ZNF320 high expression predicted a poorer prognosis (p = 0.017) (Figure 3C). Next, we constructed the ROC curve to detection the sensitivity and specificity to predict one-, three-, and five- year survival in HCC patients, The AUC of the ROC curve is significant, (one-year AUC:0.645, three-year AUC:0.569, five-year AUC:0.544), which indicates that the expression of ZNF320 can availably predict the survival time of patients (Figure 3D). Then, to screen out deeply the connections between the ZNF320 expression and clinical characteristics, we conducted univariate and multivariate Cox regression (Table 1). The univariate cox analysis revealed ZNF320 (HR: 1.298, 95% CI: 1.127-1.495, p <0.001) is substantially related with OS. The multivariate cox analysis exposed the variables of ZNF320 expression (HR: 1.294, 95% CI: 01.108-1.521, p = 0.001) could regard as an independent prognostic indicator for patients with HCC. The forest map also demonstrated this point (Figure 3E). In brief, our study implied that ZNF320 expression can be an independent prognostic parameter, and cases with elevated ZNF320 expression tend to be communicate with a worse prognosis.

Figure 3. ZNF320 expression in tumor tissues is associated with poor survival in HCC patients. (A, B) HCC patients with lower expression level of ZNF320 had favorable OS and DSS results from KM plotter (p=0.00084; p=0.027). (C) HCC patients with lower expression level of ZNF320 had favorable OS results from R(p=0.017) (D) ROC curves for the 1-, 3-, and 5-year survival according to the expression level of ZNF320. AUC, area under the curve; ROC, receiver operating characteristic. (E) A forest plot of the results of the multivariate analysis. *, P<0.05; **, P<0.01; ***, P<0.001. HR, hazard ratio; CI, confidence interval; T, tumor; N, node, M, metastasis; OS, overall survival; AIC, Akaike’s information criterion.

Correlation between ZNF320 mutation, hypermethylation, and prognosis in HCC

Significant increases of ZNF320 expression in HCC were observed. In consequence, we will study the reason for the overexpression of ZNF320. As methylation plays an important role in gene expression. we studied the methylation and expression of ZNF320 through the cBioPortal dataset. The finding was that ZNF320 expression was highly positively correlative with methylation (R = 0.1, p =0.022) in HCC (Figure 4A). We detected the methylation level of FARSB promoter in liver cancer tissues through the MethSurv website, and the correlation heat map showed that three methylation probes 16204618, cg03067828, and cg24484296 were in a hypermethylated state (Figure 4B) What’s more, the MethSurv analysis indicated that patients with high ZNF320 methylation had a worse overall survival than those with low ZNF320 methylation (p < 0.05) (Figure 4C–4E).

Figure 4. Correlation between ZNF320 mutation, hypermethylation, and prognosis in HCC. (A) The correlation between ZNF320 methylation and its expression level. (B) The visualization between the methylation level and the ZNF320 expression. (C–E) the Kaplan–Meier survival of the promoter methylation of ZNF320.

In conclusion, the methylation level of ZNF320 was positively associated with its expression. It is supposed that the increased methylation level of ZNF320 promoter leads to the high ZNF320 expression in liver cancer, which in turn causes poor prognosis in patients with liver cancer.

Function and pathway enrichment analyses and co-expression genes of ZNF320 in HCC

To better realize the biological meaning of ZNF320 in HCC, the LinkedOmics was used to find the co-expression pattern of ZNF320. As displayed in Figure 5A, it reveals that 13934 genes (red dots) positively associated with ZNF320, and 5989 genes (green dots) was negative correlation with ZNF320 (p-value < 0.05). The top 50 obvious gene set positive (left) and negative (right) correlation with ZNF320 were present in the heatmap. (Figure 5B, 5C) The top 200 genes that were correlated most clearly with ZNF320 were extracted for enrichment analysis. Moreover, Supplementary Table 1 detailed lists the co-expressed genes. We used David database to analyze the function and pathway enrichment of gene co-expressed with ZNF320 (version 6.8). Bubble plots demonstrate the top 10 enriched biological functions and pathways (based on P-value) were present by the hipplot. Analysis using the GO database showed that ZNF320 was functionally associated with G2/M transition of mitotic cells, Cell-cell adhesion, Regulation of transcription, Cell division, Cadherin binding involved in cell-cell adhesion, DNA binding, Transcription factor activity (Figure 5D–5F). Pathway analyses were conducted by the KEGG database. Results revealed that genes were enriched mostly in pathways in cancer, Wnt signaling pathway. These pathways, which were related to the tumorigenesis and progression of tumors, indicated that ZNF320 may be closely-related to HCC tumorigenesis and progression. (Figure 5G) Furthermore, GSEA was conducted to seek out KEGG pathways, which exposed FC epsilon pathway, FC gamma R mediated phagocytosis, cell cycle, WNT signaling pathway, and pathways in cancer, leukocyte transendothelial migration, mismatch repair, tight junction (Figure 6A–6I) These results suggested that ZNF320 worked by participating in cell cycle, DNA mismatch repair, WNT signaling pathway and immune-related pathways in HCC.

Figure 5. Genes differentially expressed in correlation with ZNF320 and enriched GO annotations, KEGG pathway of ZNF320 correlated genes in HCC. (A) Pearson test was used to analyze the association between ZNF320 and genes differently expressed in HCC, red indicates positively correlated genes, and green indicates negatively correlated genes. (B, C) The top 50 genes positively and negatively relative to ZNF320 in HCC were shown by heat maps. (D–F) Significant Gene Ontology terms of the top genes most positively relative to ZNF320, including biological processes (BP), molecular function (MF), and cell component (CC). (G) Significant KEGG pathways of the top genes most positively associated with ZNF320.

Figure 6. GSEA was used to validate the gene signatures, including positive regulation of (A) FC epsilon pathway, (B) FC gamma R mediated phagocytosis, (C) cell cycle, (D) pathway in cancer, and (E) leukocyte transendothelial migration, (F) WNT signaling pathway, (G) mismatch repair, (H) TGF-beta signaling pathway, (I) tight junction.

Correlation of ZNF320 expression with cell cycle

To further investigate the potential function of ZNF320 in HCC, STRING database was performed on ZNF320 co-expressed genes to make the protein–protein interaction (PPI) network, and Cytoscape (MCODE plug-in) was applied to find the most vital module. What’s more these genes were highlighted in yellow (Figure 7A). Based on the degree score, the module with the highest score consisted PRIM2, SPDL1, CKAP5, GINS4, KIF23, KIF18A. (Figure 7B) And we investigate that there existed an obvious correlation coefficient between ZNF320 and PRIM2 through GEPIA analysis (Figure 7C). What’s more, we have done prognosis analysis of these genes by Kaplan-Meier Survival Method, which showed that all of these 6 genes were oncogenes that were related to poor prognosis. (Figure 7D) The results of pathway analysis proved that all 6 genes in the module were related to cell cycle, and based on the above analysis, we inferred that the impact of ZNF320 on HCC prognosis may be connected to the cell cycle.

Figure 7. Protein–protein interaction network of related genes (Top200) and analysis of hub genes in HCC. (A) Protein–protein interaction (PPI) network (B) MCODE analysis (C) Correlation between ZNF320 and the mRNA expression of SPDL1, KIF18A, PRIM2, GINS4, KIF23, CKAP5 in HCC determined using GEPIA (D) Prognosis analysis of correlational genes.

ZNF320 correlates with tumor purity and immune infiltration level in HCC

It has been confirmed that the occurrence and development of tumors and their prognosis depend on the number of immune cell infiltration. Meanwhile, KEGG and GSEA suggest that ZNF320 is associated to immune infiltration. Through analyzing whether ZNF320 expression was associated to immune infiltration levels in HCC, we discovered a positive correlation between ZNF320 expression and tumor purity, infiltrating levels of B cells, CD8+ T cells, CD4 + T cells, Macrophage, Neutrophils, and Dendritic cell (Figure 8A). Furthermore, ZNF320 CNV was significantly correlate with infiltrating levels of B cells and CD4 + T cells. (Figure 8B). Additionally, we detected the expression of ZNF320 in different immune subgroups in LIHC by TISIDB. We found higher ZNF320 expression in the Cl (wound healing) and C2 (IFN-gamma dominant) subgroups, whereas lower in the C6 (TGF-b dominant) subgroup (Figure 8C). Altogether, ZNF320 is closely related with the main immunity cells in HCC.

Figure 8. Correlations of ZNF320 expression with immune infiltration level in HCC (TIMER). (A) ZNF320 expression is positively related to tumor purity, infiltrating levels of CD8 +T cells, CD4 + T cells, macrophages, neutrophils and dendritic cells in HCC. (B) ZNF320 CNV affects the infiltrating levels of CD8 + T cells, macrophages, and dendritic cells in HCC. *P < 0.05, ***P < 0.001. (C) Distribution of ZNF320 expression across immune subtypes in LIHC (TISIDB). The different color plots represent the five immune subtypes (C1: wound healing; C2: IFN-gamma dominant; C3: inflammatory; C4: lymphocyte-depleted and C6: TGF-b dominant).

The correlation between ZNF320 and markers of immune infiltrates in HCC

To further explore the underlying correspondence of ZNF320 and different immunocyte, TIMER was used to analyze correlations between ZNF320 levels and multiple gene markers of immune cells, which included TAM, M1 and M2 macrophages, CD8+ T cell, T cell and B cell in HCC (Table 2). Because B cells, T cells, CD8+ T cells, and macrophages are the most relevant immune cell types to ZNF320 expression, the connection between ZNF320 and the sets of immune markers of these cells were further investigated through TIMER. We found positive correlations between ZNF320 and the expression of these specific immune markers of these cells, e.g., B cell markers, TAM markers, M1 markers, M2 markers (Figure 9A–9F). In conclusion, our results showed that there is an association between ZNF320 and tumor cell infiltration in HCC.

Figure 9. ZNF320 expression correlates with marker sets of immune cells in HCC. The association between the expression levels of ZNF320 and B cell (A), T cell (B), M1 macrophages (C), TAM (D), CD8+T cell (E), and M2 macrophages (F) using TIMER. (G) The association between the expression levels of ZNF320 and the HCC-related chemokines.

Moreover, ZNF320 was also associate with the HCC-related chemokines including CCL15, CCL8, and CCL26. It could be known from the results that ZNF320 was negatively associated with CCL15 and CXCL2 while positively associated with CCL26 (Figure 9G). Therefore, ZNF320 constitutes an immunosuppressive microenvironment by affecting the expression levels of relevant chemokines, helping tumor cells to participate in immune escape, thereby enhancing tumor cell invasion and migration capabilities.

These results indicated that ZNF320 could affect immune cell infiltration partly by regulating these chemokines expression.

Prognostic analysis of ZNF320 expression in HCC in view of immune cells

Because ZNF320 expression was significantly correlated with immune infiltration, which is related to prognosis in HCC, we analyzed whether ZNF320 expression influences the prognosis of HCC by influencing immune infiltration. The prognosis was analyzed based on the expression level of ZNF320 in HCC related immune cell subgroup. As shown in Figure 10, the expression of ZNF320 in different B cell, CD4+ memory T-cells, CD8 + T cell levels had no significant correlation with the prognosis of HCC. Whether or not these immune cells are enriched or not high expression of ZNF320 leads to a poor prognosis for HCC patients (Figure 10A–10C). However, when Natural killer T-cells, Regulatory T-cells, Type 1 T-helper cells, and Type 2 T-helper cells are enriched, ZNF320 high expression leads to a poor prognosis in patients with HCC (Figure 10D–10G). These outcomes indicated that differential gene expression can affect patient outcomes at different levels of cell infiltration. In conclusion, these results proclaimed that ZNF320 was related to tumor cell infiltration in HCC.

Figure 10. Kaplan-Meier survival curves according to high and low expression of ZNF320 in immune cell subgroups in liver cancer. (A–G) Correlations between ZNF320 expression and OS in different immune cell subgroups in HCC patients were estimated by Kaplan-Meier plotter.

Relationship between ZNF320 Expression and m6A modification in HCC

Modification of m6A plays a significant role in HCC. By analyzing TCGA and ICGC HCC data, we detected the correlation between the expression of ZNF320 and the expression of 21 m6A related genes in HCC, and ZNF320 expression significantly positively correlated with ZNF320, RBMX, RBM15B, LRPPRC, YTHDF1, HNRNPC (Figure 11A, P<0.01) in the TCGA database. Furthermore, ZNF320 expression significantly positively correlated with RBM15B, LRPPRC, YTHDF1, HNRNPC in ICGC data sets (Figure 11B, P < 0.01).

Figure 11. Correlation of ZNF320 expression with m6A related genes in HCC. (A, B) TCGA HCC data set and ICGC data set analyzed the correlation between the ZNF320 and m6A related genes expression in HCC. (C) The differential expression of glycolysis related genes between high and low ZNF320 expression groups in HCC tumor samples. (D) Venn diagram showed correlation coefficient greater than 0.39 with ZNF320 in two databases, including RBMX, YTHDF1, METTL3, IGF2BP2. (E) Draw a scatter plot to show the correlation between the ZNF320 and glycolysis related genes expression, including RBMX, YTHDF1, METTL3, IGF2BP2 (F) Kaplan-Meier curve of RBMX, YTHDF1, METTL3 *P < 0.05; *P < 0.05.

We divided TCGA samples into two groups according to the expression of ZNF320. We tried to exam the differential expression of genes related to M6A between the high and low ZNF320 groups. As shown in Figure 11, the m6A modification was not the same between high and low groups with the ZNF320 expression in HCC (Figure 11C). The intersection of genes with correlation coefficient greater than 0.39 with ZNF320 two databases was calculated, genes were presented in Venn’s diagram, including RBMX, YTHDF1, METTL3, IGF2BP2. (Figure 11D). The scattering plot shows the correlation between the expression of the genes related to ZNF320 and m6A (Figure 11E) Compared to the group of low expression, the expression of 19 genes in the high expression group of ZNF320 were increased (P <0.001). Then, we used Kaplan-Meier curve to reveal that the high expression of RBMX, YTHDF1, METTL3 were intensely associated with a poor prognosis of HCC (P<0.001) (Figure 11F). These results claim that ZNF320 may be closely related to the m6A modification of HCC, especially through its regulation with RBMX, YTHDF1, METTL3, which eventually influent the progression and prognosis of HCC.