Comprehensive analysis reveals TSEN54 as a robust prognosis biomarker and promising immune-related therapeutic target for hepatocellular carcinoma

Data source and processing

This article mainly includes two major databases, The Cancer Genome Atlas (TCGA) database (https://cancergenome.nih.gov) and the International Cancer Genome Consortium (ICGC) database (https://dcc.icgc.org/projects/LIRI-JP). To date, TCGA has published data regarding 33 types of cancers, and our study researched TSEN54 expression in HCC using 374 tumor samples and 50 normal samples in the TCGA database with file type HTSeq-FPKM [15]. Other than this, through 377 clinical samples in TCGA, we drilled down the clinicopathological features of TSEN54 in HCC and the impact of TSEN54 on the survivorship of HCC patients. Moreover, we also made use of data from the ICGC [LINC-JP] Liver Cancer– NCC to verify the above conclusion, which involved 202 normal samples and 243 tumor samples.

TIMER database analysis

TIMER (https://cistrome.shinyapps.io/timer/) is a synthetic network that analyzes the immune invasion level in different cancers [16]. We used “Diff Exp” to identify the expression of TSEN54 in multiple tumors. “Genes” and “SCNA” were then applied to estimate the relation between tumor infiltrating-immune cells and the expression of mRNA and copy number variations of TSEN54 in HCC. Finally, we verified the relevance between TSEN54 expression and immune cell markers and immune checkpoints in HCC with “correlation”.

HCCDB database analysis

The Hepatocellular Cancer Database (HCCDB) (http://lifeome.net/database/hccdb/home.html) is a database dedicated to hepatocellular carcinoma, and it contains 15 publicly available expression datasets about HCC, altogether 3917 samples [17]. Using this database, we investigated the TSEN54 expression in HCC tissues and adjacent tissues in 10 datasets.

GEPIA database analysis

The Gene Expression Profiling Interactive Analysis (GEPIA) database (http://gepia.cancer-pku.cn/) uses data from TCGA and GTEx to present subscribers with customized functionality [18]. Together with the data from TCGA, we employed it to draw box plots for the analyses of TSEN54 expression in HCC and normal tissues. Besides, we exploited both Overall Survival (OS) and Disease-Free Survival (RFS) survival plots for survival analysis of TSEN54 in HCC patients and examined the impact of m6A-related regulators on the prognosis of HCC patients by OS curves.

HPA analysis

The Human Protein Atlas (HPA) (https://www.proteinatlas.org/) is a versatile database that analyzes gene expression in human proteins from different aspects [19]. From the HPA database, we obtained the outcome of staining tissues from diverse patients to verify the protein expression level of TSEN54 in normal tissues (Patient id:3402) and HCC tissues (Patient id:3477).

UALCAN database analysis

The UALCAN database (http://ualcan.path.uab.edu) is an interactive portal online tool that delves into TCGA gene expression data and can be applied to research multiple cancers [20]. In our research, we explored the clinicopathological features, methylation and survival of TSEN54 in HCC via it.

cBioPortal analysis

cBioPortal (https://www.cbioportal.org/) is an integrated web resource, presenting us with visual and multidimensional cancer genomics data [21]. Based on Liver Hepatocellular Carcinoma (TCGA, Firehose Legacy) dataset, we monitored TSEN54 methylation level linked to its mRNA expression.

MEXPRESS analysis

MEXPRESS (https://mexpress.be) online website is used to visualize DNA methylation and gene expression, as well as the correlation between them [22]. We studied the relationship between the methylation and expression level of some TSEN54 CpG sites by it.

SMART analysis

Shiny methylation analysis resource tool (SMART; http://www.bioinfo-zs.com/smartapp/) is a site which is friendly to subscribers, and it can be applied to evaluate DNA methylation data from the TCGA database [23]. By searching “TSEN54,” we researched the allocation of methylation probes associated with TSEN54 on the chromosome.

Construction of ceRNA regulatory network

The TargetScan database (http://www.targetscan.org) allows for the prediction of probable miRNA binding sites and contributes to the placement of miRNAs into gene regulatory networks [24]. RNAinter (http://www.rnainter.org) online site incorporates experimentally confirmed and computationally forecasted RNA interactome data, which facilitates us to study the regulatory environment of cellular RNA [25]. Our study extracted the data from these two databases to take the intersection for the purpose of screening out the target miRNAs of TSEN54. miRNet 2.0 (https://www.mirnet.ca/miRNet/home.xhtml) is a database that integrates user data with existing knowledge for the clarification of miRNAs [26]. starBase (https://starbase.sysu.edu.cn/) online website available for the decoding of RNA-RNA and protein-RNA interaction networks [27]. We exploited both two platforms to identify target lncRNAs for hsa-miR-125b-5p and hsa-miR-27b-3p.

Kaplan-Meier plotter database analysis

Kaplan-Meier Plotter database (http://kmplot.com) is available to measure the connection between gene expression and survivorship, whose data is derived from GEO, EGA and TCGA databases [28]. Separating TSEN54 expression into high and low groups, our study not only inspected the relevance of TSEN54 expression to overall survival (OS), progression-free survival (PFS), relapse-free survival (RFS) and disease-specific survival (DSS) in HCC, but also explored the immune cells affecting HCC patients‘ survival.

LinkedOmics analysis

LinkedOmics database (http://linkedomics.org/login.php) can be applied to discuss the co-expression of genes and their corresponding functional enrichment analysis [29]. Our study used the LinkFinder module to structure a volcano map and heatmaps to illustrate the co-expression of TSEN54 genes and displayed their correlation by plotting scatter plots. The LinkInterpreter module was applied to establish Go annotations and KEGG pathways with the purpose of describing the pathways through which these genes function. The Spearman correlation test in this study is our statistical method.

Gene set enrichment analysis (GSEA)

GSEA can explain gene expression data by testing if the gene set was enriched at the top or bottom of a pre-set sequencing table [30]. Our research divided the TCGA-LIHC data into two groups with TSEN54 high and low expression to study the possible influence of the TSEN54 gene set on HCC pathways via GSEA v4.2.2. We configure some parameters as follows: gene set database: kegg. v7.5.1 symbols. gmt (Curated), meanwhile, P < 0.05 and FDR < 0.25 were regarded meaningful.

TISCH analysis

The Tumor Immune Single Cell Center (TISCH) (http://tisch.comp-genomics.org/) database is concentrating on the examination of the tumor microenvironment, gathering data from the Gene Expression Omnibus (GEO) and ArrayExpress for 27 cancer types and 76 tumor datasets [31]. It was utilized in our study for the visualization of the single-cell level expression status of TSEN54 in two immune cell datasets, which were LIHC_GSE140228_10× and LIHC_GSE140228_Smartseq2.

TISIDB analysis

TISIDB (http://cis.hku.hk/TISIDB/) is a comprehensive database could be employed to analyze interactivity between neoplasms and the immune system [32]. We predicted the prognosis of HCC patients and explored TSEN54 expression in diverse immune subtypes, as well as the relationship of TSEN54 expression with chemokines in HCC utilizing the TISIDB database.

Quantitative real-time PCR

The HCCLM3 cell line was obtained from the Cell Bank of Type Culture Collection of the Chinese Academy of Sciences and the Shanghai Institute of Cell Biology in China. We extracted total RNA from cell samples applying Trizol reagent (Thermo Fisher Scientific) by following the manufacturer’s guidelines, then performing reverse transcription reactions employing the PrimeScript RT Reagent Kit (Invitrogen, USA). And the qPCR was completed utilizing SYBR Premix Ex Taq (TaKaRa, China) [33]. This technique was adopted in this study to examine the relative mRNA expression of CCL15, CCL26, and CCL28.

GeneMANIA analysis

GeneMANIA (https://genemania.org/) database can be used to find genes that share functions with the gene of interest and to explore interactions and functions between them [34]. Through GeneMANIA database, we researched the connection between TSEN54 and other related genes.

Protein structure and docking analysis

The secondary structures of TSEN54 and its physically interacting proteins were acquired from the “Mutations” panel in the cBioPortal database (https://www.cbioportal.org/) for the dataset: Liver Hepatocellular Carcinoma (TCGA, Firehose Legacy) [35]. CSNK2B and ATM advanced structures were acquired by using the PDB database (https://www.rcsb.org/) (PDB ID: 3EED and 7SIC) [36]. The advanced structure of TSEN54 was predicted by AlphaFold Protein Structure Database (https://www.alphafold.ebi.ac.uk/) (UniProt ID: Q7Z6J9) [37]. Finally, the interaction docking patterns between TSEN54 and CSNK2B as well as TSEN54 and ATM were predicted by the HDOCK server (http://hdock.phys.hust.edu.cn/) and visualized using PyMOL software [38].

GSCALite analysis

Gene Set Cancer Analysis (GSCALite) (http://bioinfo.life.hust.edu.cn/web/GSCALite/) integrates information from TCGA, Drug Sensitivity in Cancer (GDSC), Cancer Therapeutics Response Portal (CTRP), GTEx (Genotype-Tissue Expression) to users [39]. Our research utilized data from TCGA-LIHC and GTEx_Liver to explore TSEN54-related cancer pathways and analyzed drug susceptibility in HCC by GDSC.

Statistical analysis

Our article performs all data analysis by R software (version 3.6.3/4.1.2). We adopted the rank-sum test to detect the expression of TSEN54 in HCC tissues and normal tissues according to the R packages “limma” and “beeswarm”, and the statistical methods of the Wilcox. test and Kruskal. test and logistic regression analysis were applied to probe the link between TSEN54 and clinicopathological characteristics. What is more, for the purpose of examining the connection between TSEN54 expression and the outcome of HCC sufferers, we drew Kaplan-Meier survival curves by log-rank test and created a univariate and multivariate Cox regression model through the R packages “survival” and “survminer” (P < 0.05), and we also built ROC curves by “survival ROC” to assess the accuracy of TSEN54 expression in predicting the survival rate of HCC patients at 1, 3 and 5 years.

Data availability

The TCGA data used to support the findings of this study have been deposited in the [https://cancergenome.nih.gov] repository (doi:10.1038/nmeth.2956).

TSEN54 expression is upregulated in HCC

With the purpose of assessing TSEN54 differential expression in HCC and normal samples, we attempted a variety of methods. In the beginning, we made a study of the TSEN54 expression level in pan-cancer on the TIMER database. The result demonstrated that the expression of TSEN54 was remarkably higher in multiple cancer tissues involving HCC in comparison with normal tissues (Figure 1A). Then, applying the HCCDB database to verify the above result, we also discovered that the finding was in line with our expectations (Figure 1B). After that, we further investigated the expression of TSEN54 via data from TCGA depending on “limma” and “beeswarm” packages from R software, and TSEN54 expression was detected meaningfully higher in HCC tissues than in normal tissues (Figure 1C). Moreover, we made a comparison of the TSEN54 expression of HCC between 50 pairs of tumor samples and normal samples, the same outcome was obtained (Figure 1D). What is more, as exhibited in Figure 1E, 1F, we still consistently ascertained identical results in both the GEPIA website and ICGC database. Ultimately, we obtained tissue slices through the HPA database to further corroborate our conclusion, and it can be observed from the IHC staining results that the TSEN54 protein expression in HCC tissues displayed higher levels versus normal liver tissues (Figure 1G). Above all, we can come to the conclusion that in contrast to normal tissues, the TSEN54 expression in HCC tumor tissues is upregulated.

Figure 1. The expression analysis of TSEN54 in HCC. (A) Differential expression of TSEN54 in pan-carcinoma and corresponding normal tissues in TIMER. (B) Charts and graphs present the TSEN54 expression in HCC and normal tissues downloaded from the HCCDB database. (C) Plots describe the mRNA expression level of TSEN54 in normal and HCC samples. (D) Paired differential expression of TSEN54 in normal and HCC tissues. The difference representation of TSEN54 in HCC and normal samples was plotted using the (E) GEPIA website and the (F) ICGC database. (G) IHC staining results show the protein expression level of TSEN54 in liver normal tissues and tumor tissues.

The correlation between the expression of TSEN54 and clinicopathological features

Digging out the connection between the expression of TSEN54 in HCC and clinical pathological features is of great practical importance for our research. Through TCGA-LIHC samples from the UALCAN database, we conducted a subgroup analysis of several clinical pathological characteristics. According to Figure 2A–2F, we did know that TSEN54 was related to the grade, stage, race, gender, TP53 mutation status, and histological subtypes in HCC. Interestingly, we also observed that the expression of TSEN54 in grade1 and stage1 had a certain degree of increase in HCC compared to normal, and it was also very significant between many other subgroups of grade and stage (Figure 2A, 2B). Next, we downloaded clinical samples from the TCGA database and then made use of the Wilcox. test and Kruskal. test in R software, again finding that TSEN54 expression in HCC was highly related to grade (p=0.012), stage (p=0.006) and T (p=0.01) (Figure 2G–2I). Lastly, we further verified our conclusion by logistic regression analysis and uncovered it to be consistent with our previously stated consequence (Supplementary Table 1). All findings point to an inextricable relationship between TSEN54 expression in HCC and multiple clinicopathological features.

Figure 2. TSEN54 expression is associated with multiple clinicopathological features. Box-whisker plots reveal the variability of gene expression level ground on subgroup (A) tumor grade; (B) individual cancer stages; (C) patient’s race; (D) patient’s gender; (E) TP53 mutation status; (F) histological subtypes. The relationship between TSEN54 expression and (G) grade; (H) stage; (I) T stage in HCC according to data from the TCGA database. (*, p < 0.05; **, p < 0.01; ***, p < 0.001).

High expression of TSEN54 is affected by hypomethylation

There is growing evidence that illustrates methylation has a crucial part in cancer emergence and progression [40]. Firstly, we applied the UALCAN website and revealed that TSEN54 promoter methylation level in HCC tissues were lower than those in normal samples (Figure 3A). Additionally, we identified that the TSEN54 promoter methylation level was linked to clinical stage, histological grade and lymph node metastasis (Figure 3B–3D). Moreover, HCC patients with TP53 mutations have lower methylation level compared to those without mutations (Figure 3E). Besides, in the sex subgroup of patients, TSEN54 methylation level in the HCC tissues of female patients were lower than those of male patients (Figure 3F). The subtle contrast in clinicopathological features between the methylation results and the expression results in Figure 2 impelled us to directly detect the relation between mRNA expression and methylation of TSEN54 using the cBioPortal website, and unsurprisingly, the two behave as a negative correlation (Figure 3G).

Figure 3. The TSEN54 promoter methylation level is correlated with clinicopathological features. (A) sample type; (B) stage; (C) grade; (D) lymph node metastasis; (E) TP53 mutation status; (F) patient’s gender. (G) The correlation between TSEN54 methylation and its mRNA expression.

Then, through the MEXPRESS online website, we found that the methylation level of 9 CpG sites were correlated with TSEN54 expression. Among them, the methylation level of 8 CpG sites including cg05606039, cg09173924, cg23601586, cg00866186, cg20863668, cg02970545, cg06665453, cg20366832 were negatively associated with TSEN54 expression level (Figure 4A). Furthermore, we utilized the SMART website to lucubrate the distribution of methylation probes related to TSEN54 on chromosome (Figure 4B). The result shown that cg05606039, cg00866186, and cg20863668 are on the island, cg02970545, cg06665453, and cg23601586 are on N_Shore, and cg09173924 is on S_Shore (Figure 4C). In conclusion, we discovered hypomethylation of TSEN54 in HCC, which contributes to its high expression.

Figure 4. Distribution of CpG sites on TSEN54. (A) Correlation between TSEN54 expression and methylation of the CpG sites. (B) Distribution of TSEN54 methylation probes on chromosome. (C) Location of CpG sites associated with TSEN54.

The construction of a ceRNA regulatory network of TSEN54 in HCC

In an attempt to better gain an understanding of the possible regulatory mechanisms at the level of TSEN54 expression, we predict and build a lncRNA-miRNA-mRNA ceRNA network that involves TSEN54 in HCC employing multiple bioinformatics databases. Firstly, we jointly predicted 12 miRNAs targeting TSEN54 by TargetScan and RNAinter databases, which were hsa-miR-135b-5p, hsa-miR-135a-5p, hsa-miR-125b-5p, hsa-miR-214-3p, hsa-miR-27a-3p, hsa-miR-27b-3p, hsa-miR-3202, hsa-miR-4319, hsa-miR-125a-5p, hsa-miR-6510-5p, hsa-miR-1200, hsa-miR-761 (Figure 5A). Since miRNAs usually have the capacity to negatively regulate mRNA, two miRNAs (hsa-miR-27b-3p and hsa-miR-125b-5p) that were negatively relevant to TSEN54 at the expression level were filtered from those 12 miRNAs, and their correlation scatter plots were given in Figure 5B, 5C. Afterward, utilizing miRNet and starBase tools, a series of 26 lncRNAs that may bind to hsa-miR-27b-3p and 12 lncRNAs that combine with hsa-miR-125b-5p were further predicted (Figure 5D, 5E). Similarly, as lncRNAs are usually able to negatively regulate miRNAs, we continued to screen for lncRNAs that correspond to a negative association with the two miRNAs at the expression level. From our findings, for hsa-miR-27b-3p, the eligible lncRNAs were SCAMP1-AS1, HCP5, and RMST (Figure 5D); while for hsa-miR-125b-5p, NOP14-AS1, DANCR, KCNQ1OT1 and LINC00943 were qualified (Figure 5E). In the end, based on the hypothesis that there is an inverse relationship between miRNA and lncRNA or mRNA, we constructed a ceRNA regulatory network, and the results are exhibited in the Sankey diagram (Figure 5F).

Figure 5. The predicted ceRNA network of TSEN54 in HCC. (A) The Venn diagram demonstrates the miRNAs targeting TSEN54 predicted by TargetScan and RNAinter databases. The correlation scatter plots reveal the expression relevance of TSEN54 with (B) hsa-miR-27b-3p and (C) hsa-miR-125b-5p. Venn diagram illustrates the prediction of lncRNAs targeting (D) hsa-miR-27b-3p and (E) hsa-miR-125b-5p through miRNet and starBase tools. (F) Sankey diagram presents the ceRNA network of TSEN54 in HCC.

High expression of TSEN54 is a separate factor for poor prognosis in HCC patients

For the sake of analyzing the prognostic value of TSEN54 expression in HCC patients, we made use of several online websites. The association between TSEN54 expression and prognosis of HCC patients was initially investigated using the GEPIA website. In LIHC datasets, patients with high TSEN54 expression had shorter OS and DFS compared to those with low expression (Supplementary Figure 1). Next, the Kaplan-Meier Plotter site indicated that HCC suffers with higher TSEN54 expression possessed poorer OS, PFS and RFS, while the differential expression of TSEN54 in the DSS group had no significant impact on survival (Figure 6A–6D). Besides, an analogous outcome was likewise backed up by the TISIDB website (Figure 6E). Deepening, we hired the UALCAN website and identified that not only can the differential expression of TSEN54 impinge on patient prognosis, but TSEN54 expression combined with clinicopathological features (e.g., grade and race) can also affect the survival of HCC patients (Figure 6F–6H).

Figure 6. Relationship between TSEN54 expression and prognosis of HCC patients. The relation between TSEN54 expression and (A) OS; (B) PFS; (C) RFS; (D) DSS based on the Kaplan-Meier plotter database. (E) The correlation between TSEN54 expression and the survival of HCC patients based on the TISIDB website. Survival of HCC patients with (F) differential TSEN54 expression; (G) differential TSEN54 expression and grade; (H) differential TSEN54 expression and race.

In parallel, we optimized patient survival data collected from the TCGA database and again demonstrated a shorter prognosis within the TSEN54 expression upregulated group (Figure 7A). Thereafter, the ROC curve was created through the TCGA data to reflect the prognostic value of TSEN54. The AUC of the ROC curve was significant (one year AUC: 0.645, three years AUC: 0.593, five years AUC: 0.591), indicating that the expression of TSEN54 was a valid predictor for prognosis in HCC patients (Figure 7B). Eventually, the connection between clinicopathological features and the prognosis of patient data extracted from the TCGA database was calculated by constructing univariate and multivariate cox models. Univariate analysis revealed a close correlation between stage, T stage, M stage, TSEN54 expression, and deterioration of OS. Multivariate analysis demonstrated TSEN54 expression to be a separate prognostic factor (Supplementary Table 2). This is directly reflected in the forest map (Figure 7C). In conclusion, the expression of TSEN54 is a standalone prognostic biomarker of HCC patients, while hyper-expression of TSEN54 is linked to the unfavorable prognosis of HCC patients.

Figure 7. TSEN54 is an independent prognostic factor for the survival of HCC patients. (A) The association between TSEN54 expression and HCC patients based on data downloaded from TCGA. (B) ROC curve for one-, three-, and five-year survival according to TSEN54 expression level. (C) The forest plot displays the result of the multivariate analysis. *p< 0.05; **p< 0.01; ***p< 0.001.

Co-expression network of TSEN54 and its functional enrichment analysis

To detect co-expressed genes of TSEN54 and probe their biological functions in HCC, we did a series of works. We first utilized the LinkedOmics database to investigate the co-expression genes of TSEN54. As indicated in the volcano map, red dots represented genes positively associated with TSEN54, totaling 6,679, while green dots meant genes negatively correlating with the target genes, amounting to 5,702 (Figure 8A). The 50 genes most strongly positively and negatively related to TSEN54 were exhibited in heatmaps (Figure 8B, 8C). In the following, we opted the four genes that were most prominently positively correlated with TSEN54 called the hub genes to create scatter plots, which were AZI1, CCDC137, THOC4, and BIRC5 (Figure 8D–8G). Subsequently, we made a deeper exploration of the potential functions of TSEN54 and its co-expressed genes in HCC. From the GO outcome, we could conclude that TSEN54 was mainly concentrated in the pathways like chromosome segregation, DNA replication, cell cycle G2/M phase transition, interstrand cross-link repair, RNA catabolic process, etc. (Figure 9A). Meanwhile, the KEGG findings as revealed in Figure 9B suggested that TSEN54 may function through spliceosome, ribosome, cell cycle, DNA replication, homologous recombination, Fanconi anemia pathway, pyrimidine metabolism, PPAR signaling pathway, etc.

Figure 8. Co-expressed genes of TSEN54 in HCC (LinkedOmics). (A) The volcano map reveals the positive and negative co-expression genes of TSEN54 and roughly illustrates the correlation between them. Heatmaps show the first fifty genes most (B) positively and (C) negatively linked to TSEN54. (D–G) The scatter plots detail the link between TSEN54 and its hub genes. GO: Gene Ontology; KEGG: Kyoto Encyclopedia of Genes and Genomes.

Figure 9. Functional enrichment study of TSEN54 co-expressed genes. (A) Go annotations and (B) KEGG pathways with a notable enrichment of the TSEN54 co-expression genes in HCC by LinkedOmics. (C–H) Analysis of the respective pathways enriched in high and low expression of TSEN54 and its co-expressed genes according to GSEA.

Additionally, we made further investigation based on GSEA to probe the pathways by which TSEN54 and some of its relative genes may exert influence on HCC. The findings implied that it may have a connection with DNA repair and metabolism-related pathways, such as base excision repair, pyrimidine metabolism, nucleotide excision repair and insulin signaling pathway (Figure 9C–9H). From the outcomes mentioned above, we draw a conclusion that the expression of TSEN54 may function critically in HCC by modulating the cell cycle, substance metabolism and DNA repair.

Expression of TSEN54 in immune cells of the tumor microenvironment

Aiming to characterize TSEN54 expression in tumor microenvironment immune cells at the single cell level, we drew on two-single cell sequencing datasets from the TISCH database, which were LIHC_GSE140228_10× and LIHC_GSE140228_Smartseq2. We noticed that the expression of TSEN54 translated by log (TPM/10+1) manifested heterogeneity among distinct cell clusters with respect to diverse datasets, and the highest expression level of TSEN54 was evident in Mast cells of the LIHC_GSE140228_Smartseq2 dataset (Figure 10A). Later on, we had further exploration of two datasets, the results were as follows: the LIHC_GSE140228_10× dataset was classified into 12 cell types, of which the CD8T cell count was the most abundant (19969) (Figure 10B), plus the distribution and quantity of varied immune cells were described in Figure 10C; Also, Figure 10D consisted of 10 cell types from the LIHC_GSE140228_Smartseq2 dataset, whose distribution was given in Figure 10E, and it was noted that Mono/Macro cell and NK cell had the largest and similar numbers of cell count, with the respective count of 1699 and 1650. Additionally, the distribution of TSEN54 gained employing single-cell resolution for all kinds of immune cells in the LIHC_GSE140228_10× and LIHC_GSE140228_Smartseq2 datasets were plotted in Figure 10F, 10G, which were consistent with the finding in Figure 10A. In summary, we are able to discover that TSEN54 is expressed in almost all immune cells, especially NK cells and Mast cells by single-cell sequencing.

Figure 10. Expression of TSEN54 in HCC at the single cell level. (A) Heatmap reveals the expression level of TSEN54 in a variety of immune cells on LIHC_GSE140228_10× and LIHC_GSE140228_Smartseq2 datasets. (B–E) Types and distribution of cells in the LIHC_GSE140228_10× and LIHC_GSE140228_Smartseq2 datasets. The t-SNE plots describe the distribution of TSEN54 expression in (F) LIHC_GSE140228_10× and (G) LIHC_GSE140228_Smartseq2 datasets of individual cell clusters.

Association between TSEN54 expression and immune infiltration in HCC

Since there is a strong association between immunological characteristics of cancer and prognosis, the relation between TSEN54 expression and immune cell infiltration was first investigated utilizing the TIMER database. Consequences indicated that the TSEN54 expression level in HCC was positively correlated with B cells, CD8+T cells, CD4+T cells, macrophages, neutrophils, and dendritic cells infiltration level (Figure 11A). Then we investigated the copy number variation (CNV) of TSEN54 in relation to immune cell infiltration level. Inexplicably, high amplification of TSEN54 had lower level of infiltration of CD4+ T cells, macrophages and neutrophils (Figure 11B). To delve into the explanations between TSEN54 expression and immune cell infiltration, we examined the linkage between TSEN54 expression and relevant immune cell markers. The correlation coefficient was regulated by tumor purity. The outcomes demonstrated a positive correspondence between the TSEN54 expression and a multitude of immune cell markers (Supplementary Figure 2) (Supplementary Tables 3, 4). Separately we noticed a higher level of B cell and macrophage infiltration in the TP53 mutant group versus the wild-type TP53 subgroup (Figure 11C).

Figure 11. The relationship between TSEN54 expression and immune infiltration in hepatocellular carcinoma. (A) The correlation between the immune cell infiltration level and the expression of TSEN54. (B) The connection between the CNV of TSEN54 and the immune cell infiltration level. (C) The relation between the TP53 mutation status and the immune cell infiltration level.

Relationship between TSEN54 expression and immune subtype and chemokine expression in HCC

Later, we researched the association between TSEN54 expression and different immune subtypes and immune chemokines expression in HCC by TISIDB. It was found that the expression of TSEN54 was higher in C1 (would healing) and C2 (IFN-gamma dominant) subtypes, but was lower in C3 (inflammatory) and C6 (TGF-B dominant) (Supplementary Figure 3A). The expression level of chemokines XCL1, CCL15, CCL20, CCL26, CCL28, and TSEN54 were positively correlated (Supplementary Figure 3B–3F). In addition, our assay of chemokines expression in HCCLM3 cells with different treatments demonstrated that the expression of CCL15, CCL26, and CCL28 was downregulated in shTSEN54 cells (Supplementary Figure 3G–3I).

Role of TSEN54 in the regulation of immune response to HCC

Aiming to study the action of TSEN54 in the immune response, we again exploited the TIMER website to conduct the relation between the expression of TSEN54 and common immune checkpoints. The consequence reflected that TSEN54 expression was positively linked to most of the expression of immune checkpoints, especially CD276 (B7-H3), CTLA4, HAVCR2, and PDCD1 (PD-1) (Supplementary Figure 4). This outcome implied a possible involvement of TSEN54 in suppressing the immune response and helping tumors to undergo immune escape.

The prognostic impact of TSEN54 expression is related to immune infiltration

Since TSEN54 expression was remarkably associated with immune infiltration in HCC, we then utilized the Kaplan-Meier Plotter website to explore whether the prognostic impact of TSEN54 expression was also associated with immune infiltration. As it turns out, the poor prognostic ending of patients with high TSEN54 expression was not altered irrespective of the enrichment or decrease of B cells, CD8+ T cells, macrophages, mesenchymal stem cells, natural killer T cells and type 1 T helper cells (Figure 12A, 12C–12F, 12H). Excitingly, when CD4+ memory T-cells were enriched, changes in TSEN54 expression could no longer impinge on survival (Figure 12B). Similarly, when regulatory T-cells were decreased, changes in TSEN54 expression also failed to affect survival (Figure 12G). This result implied that TSEN54 expression affected patient survival in relation to the degree of infiltration of CD4+ memory T-cells and regulatory T-cells.

Figure 12. The prognostic influence of TSEN54 expression in HCC is linked to the infiltration level of immune cells. (A) B cells; (B) CD4+ memory T-cells; (C) CD8+ T-cells; (D) Macrophages; (E) Mesenchymal stem cells; (F) Natural killer T-cells; (G) Regulatory T-cells; (H) Type 1 T-helper cells. *p < 0.05, **p < 0.01, ***p < 0.001.

Relationship between TSEN54 and m6A modification in HCC

Previous studies have implicated aberrant regulation of m6A modification is associated with a number of human malignancies including HCC, which prompted us to explore its relation to TSEN54 in HCC [41]. In a preliminary step, when we researched the correlation between TSEN54 and the expression of 20 m6A-related regulators through heatmaps on the basis of TCGA and ICGC databases in HCC, we detected that among them, 19 regulators revealed a significantly positive correlation with TSEN54 expression in each of TCGA and ICGC databases (P<0.05) (Figure 13A, 13B). In the meantime, dividing the samples into two groups of high and low TSEN54 expression, we studied the associations of these m6A-related regulators with the differential expression of TSEN54, and it was interesting to note that these same 19 regulators had shown remarkable results from Figure 13C. Next, from the Venn plot, we could find three m6A-related regulators whose expression correlation with TSEN54 was greater than 0.55 in ICGC and TCGA databases and presented meaningful differential expression, which were YTHDF1, RBM15B, HNRNPA2B1 (Figure 13D). Immediately after, we plotted the scatter plots of the correlation between these three regulators and TSEN54 (Figure 13E). As a final step, through plotting survival curves on the GEPIA website, we observed that high expression of YTHDF1, RBM15B, and HNRNPA2B1 could lead to low survival of HCC suffers, and it could signal that they may have association with poor prognosis of HCC patients (Figure 13F). Taken together, TSEN54 may have a strong relationship with m6A modification in HCC, and it may play a possible influential part in the progression of HCC patients.

Figure 13. TSEN54 may be relevant to m6A modification in HCC. Heatmaps illustrate the association between TSEN54 and regulators relevant to m6A by (A) TCGA and (B) ICGC databases in HCC. (C) The linkage between high and low expression of TSEN54 and the m6A-related regulators in HCC. (D) Regulators with expression correlation greater than 0.55 and exhibiting meaningful differential expression are shown by the Venn plot. (E) The correlation scatter plots between TSEN54 and 3 regulators in the cross set, including YTHDF1, RBM15B and HNRNPA2B1. (F) Survival curves of YTHDF1, RBM15B and HNRNPA2B1 in HCC.

Network, structure, pathway, and drug sensitivity analysis of TSEN54 and its interacting genes

Proteins have specific spatial conformations that can be used to achieve biological functions through interactions. Therefore, we first constructed a network of gene interactions for TSEN54 using the GeneMANIA online website. The eight genes connected to TSEN54 by pink straight lines mean that there exist protein-protein interactions for their coding products (Figure 14A). Surprisingly, we noticed the existence of protein-level physical interactions between TSEN54 and the renowned CSNK2B and ATM. Thus, we further obtained the secondary structures of the proteins encoded by these three genes from the cBioPortal online portal and analyzed their structural domains and protein modification sites. The results showed that TSEN54 has a tRNA_int_end_N2 structural domain with several modification sites, including the phosphorylation site, acetylation site, ubiquitination site, and methylation site. CSNK2B has a CK_II_beta structural domain with a phosphorylation site, acetylation site, ubiquitination site, glutathionylation site, and sumoylation site. ATM has TAN, FAT, PI3_PI4_kinase, with phosphorylation, acetylation, and methylation sites (Figure 14B). Then, we downloaded the spatial structures of CSNK2B, ATM from the PDB database and predicted the TSEN54 spatial structure using AlphaFold Database, and finally predicted their protein-protein interaction docking models using HDOCK Server (Figure 14C–14F). Hydrogen bonds hold a vital position in the binding of protein molecules. In the TSEN54-CSNK2B docking model, we found a total of 11 hydrogen bonds (Figure 15A), and in the TSEN54-ATM docking model, we found a total of 17 hydrogen bonds (Figure 15B).

Figure 14. Analysis of the interaction network and docking structures of TSEN54. (A) Interaction network of TSEN54 with other genes (GeneMANIA). The different colored lines represent different linkage information. (B) Protein secondary structures of TSEN54, CSNK2B and ATM. (C–F) Cartoon view and Surface view of the overall conformation of TSEN54 combined with CSNK2B or ATM. TSEN54 is colored in light pink, CSNK2B is colored in aquamarine, and ATM is colored in blue-white.

Figure 15. Analysis of the docking structures, pathways, and drug sensitivity of TSEN54 and its interacting genes. Detailed interaction structures between (A) TSEN54 and CSNK2B and (B) TSEN54 and ATM. The key residues of TSEN54 are shown in light pink, CSNK2B in aquamarine, and ATM in blue-white. The hydrogen bonds are shown as yellow dashed lines and the distance between two atoms is marked. (C) Cancer-related pathways of TSEN54 and its interacting genes are presented in the form of a global percentage. (D) Drug susceptibility analysis of TSEN54 and its interacting genes based on the GDSC database is presented as a bubble chart.

Furthermore, as we learned from the” pathway activity” module in GSCALite, both TSEN54 and its physical interactions genes could activate notable pathways linked to cancers such as apoptosis, cell cycle, DNA damage response, while inhibiting RAS/MAPK, RTK, and other pathways (Figure 15B). Finally, we performed a gene set resistance analysis of TSEN54 and its physical interactions genes through small molecules collected by the GDSC database. The Spearman correlation gave an indication of genetic expression relating to drug tolerance, where red represented a positive correlation, signifying that cells with high expression of a gene have resistance to a specific drug and blue on the contrary. Thus, we could understand from Figure 15C that cells with high expression of TSEN54 had resistance to 7 drugs such as Bleomycin, and CHIR-99021, while it was sensitive to Vorinostat and Navitoclax. In short, it was possible that these findings may lead to a new way of treating patients with HCC accompanied by high TSEN54 expression.