ncRNAs-mediated overexpression of STIL predict unfavorable prognosis and correlated with the efficacy of immunotherapy of hepatocellular carcinoma

STIL is overexpressed in HCC

To probe the potential roles of STIL in carcinogenesis, we investigated the STIL mRNA expression in HCC expression profiles from TCGA and GEO databases. Analysis of twelve HCC sets from the HCCDB database showed that the transcriptional level of STIL was remarkably upregulated in HCC tissues compared with normal adjacent tissues in eleven HCC cohorts (Fig. 1A). Similarly, the expression of STIL in HCC tissues exhibited an obviously higher level when compared with that in the normal adjacent tissues, in two different statistics from Oncomine (Fig. 1B). These results showed that STIL is significantly overexpressed in HCC tissues, implying that STIL may exert a crucial function in the oncogenesis of HCC.

Fig.1figure 1

The STIL expression in HCC. A Chart and scatter showing the transcription level of STIL in HCC and the adjacent normal tissues in HCCDB. B Box plot showing STIL mRNA expression in the Roessler Liver, and Wurmbach Liver datasets in Oncomine database, respectively

STIL expression serves as an independent prognostic indicator in HCC patients

The observed overexpression of STIL initiated us to further explore the clinical significance of STIL in HCC. The potential relationships of STIL expression with some clinicopathological parameters were examined using the Wilcoxon signed-rank test. As shown in Fig. 2, the mRNA expression of STIL was significantly correlated with age (P = 0.004), grade (P = 1.655e−05), T stage (P = 0.005), TNM Stage (P = 0.006) and AFP (P = 0.009). However, there were no significant associations between STIL expression and other clinical factors, including gender, N stage, and M stage (data not shown). Furthermore, we compared the clinical outcome of STIL-high and STIL-low groups separated by the cut-off value of 4.264 among TCGA and ICGC datasets. In the LIHC cohort from TCGA, KM survival curves revealed that the STIL-high expression group had an unfavorable overall survival (OS) than the STIL-low expression group (P = 6.285e−04) (Fig. 3A). Likewise, the STIL-high expression group exhibited poorer outcomes (P = 0.384e−05) in the ICGC-LIRI-JP cohort from ICGC (Fig. 3B). Besides, the univariate Cox regression analysis revealed that high STIL expression was highly associated with poor clinical outcomes in TCGA cohort (hazard ratio [HR] = 1.826, 95% confidence interval CI [1.424–2.342], P < 0.001; Fig. 3C). Meanwhile, in multivariate analysis, only high expression of STIL (HR = 1.683, 95% CI [1.280–2.215], P < 0.001) still implied an inferior prognosis independent of other clinical parameters (Fig. 3D). Collectively, these results confirmed that STIL expression is an independent prognostic indicator and that high levels of STIL expression predict the poor OS of HCC.

Fig. 2figure 2

The expression of STIL and its relationships with clinicopathologic factors. The median value of STIL expression was set as the cut-off value. The Wilcoxon signed-rank test was applied to examine the association between clinicopathological varieties and the expression of STIL in 374 HCC samples. A Age, B histological grade, C T stage, D TNM stage, and E AFP levels. STIL STIL centriolar assembly protein; HCC hepatocellular carcinoma

Fig. 3figure 3

Prognostic significance of STIL expression in HCC. Kaplan–Meier survival curves showed that the overall survival (OS) of patients with HCC with high STIL expression was poor than that of patients with HCC with low STIL expression in A TCGA-LIHC and B ICGC-LIRI-JP cohort; Association with OS and clinicopathological factors in HCC patients from TCGA dataset using C univariate Cox regression (D) multivariate Cox regression

The expression of STIL was distinct in the tumor immune microenvironment

The scRNA-seq data of 71,915 cells from 10 HCC patients with 21 samples, including 10 samples from the primary tumor (PT), 2 samples from the portal vein tumor thrombus (PVTT), 8 samples from the non-tumor liver (NTL), and 1 sample from metastatic lymph node (MLN), was abstracted from GSE149614 dataset. A total of 49 cell clusters were identified using canonical gene markers of major cell populations and found they consist of 10 T and natural killer (NK) cell clusters,13 hepatocyte clusters, 2 endothelial cell clusters, 15 myeloid cell clusters, 2 fibroblast clusters, and 7 B cell clusters (Fig. 4A–C). We then compared the expression of STIL among different tissue sources and cell types. Consistent with the above result, STIL expression was significantly increased in primary tumor tissue than in normal tissue (Fig. 4D). Besides, the expression of STIL was significantly distinct among different cell clusters (Fig. 4E).

Fig. 4figure 4

The expression of STIL in HCC by single-cell analysis. A, B By t-SNE analysis of HCC single-cell data, 71,915 cells were divided into 49 clusters and six cell types, including T and natural killer (NK) cell (T/NK), hepatocyte, endothelial cell (Endo), myeloid cell, fibroblast (Fibro), and B cell. C, D The scatter plot revealed the expression of STIL was distinct among different tissues and cell types

Gene mutation profile between STIL-high and STIL-low expression group

To probe the genomic alterations between STIL-high and STIL-low subsets, we performed somatic mutation profiles from the TCGA-LIHC datasets. The top-5 highest mutated genes in the STIL-high subset (Fig. 5A) were TP53 (41%), TTN (21%), MUC16 (17%), CTNNB1 (13%), RYR2 (10%), whereas those in the STIL-low subset (Fig. 5B) were CTNNB1 (31%), TTN (23%), MUC16 (14%), TP53 (12%), and PCLO (10%). We found nine genes (TP53, DNAH3, SVIL, MCTP2, RB1, ASTL, ATP1A2, CHST3, and TECTA) have higher mutation prevalence in the STIL-high subset than in the STIL-low subset, while three genes (CTNNB1, LRP1, and APC) displayed highly mutated rates in STIL-low subset (Fig. 5C). Although the TMB was not significantly different between STIL-low and STIL-high subsets (Fig. 5D), patients from high-STIL subset exhibited significantly elevated MATH scores, suggesting a higher level of tumor heterogeneity in this subset (Fig. 5E, P = 0.039).

Fig. 5figure 5

Somatic mutation profiles between STIL-high and STIL-low expression group. Oncoplots exhibited the top-20 mutated genes in the A high-STIL and B low-STIL subset of the TCGA-LIHC cohort. C Forest plot displayed twelve mutated genes with statistic significant between the STIL-high and STIL-low group. The comparison of TMB (D) and MATH (E) scores in the STIL-high and STIL-low group. ***p < 0.001, **p < 0.01

Functional analysis of STIL

We further probed the potential mechanisms of STIL on the influence of survival by performing the GSEA and GSVA. GSEA of the KEGG gene set and Hallmark gene set were performed, and the results revealed that the STIL-high group was mostly enriched in several functional pathways related to tumor proliferation and DNA damage, such as KEGG_CELL_CYCLE, KEGG_BASE_EXCISION_REPAIR, HALLMARK_DNA_REPAIR, and HALLMARK_E2F_TARGETS (Fig. 6A, B). Consistently, GSVA analysis showed similar results that the cell proliferation gene sets (such as KEGG_CELL_CYCLE, KEGG_OOCYTE_MEIOSIS, HALLMARK_MITOTIC_SPINDLE, and HALLMARK_E2F_TARGETS) and DNA damage related sets (such as KEGG_MISMATCH_REPAIR, KEGG_HOMOLOGOUS_RECOMBINATION, and HALLMARK_DNA_REPAIR) were significantly overexpressed in the STIL-high group (Fig. 6C, D). These results suggested that the overexpression phenotype of STIL was positively associated with pathways enriched in cell proliferation and DNA damage response, which were reported to be associated with cancer initiation, progression, and immune response [47,48,49].

Fig. 6figure 6

The gene set enrichment analysis (GSEA) and Gene set variation analysis (GSVA) of STIL in the TCGA-LIHC dataset. A Top 5 pathways in the high-STIL expression phenotype predicted by GSEA analysis; B Top 5 hallmarks in the high-STIL expression phenotype predicted by GSEA analysis; C GSVA-derived clustering heatmaps of top 20 differentially expressed pathways for STIL expression; D GSVA-derived clustering heatmaps of top 20 differentially expressed hallmarks for STIL expression

Construction of ceRNA regulatory network of STIL

Non-coding RNAs (ncRNAs) have been widely acknowledged as the regulator of gene expression. To test whether STIL was regulated by some ncRNAs, we first used the ENCORI, a bioinformatics tool, to screen potentially upstream miRNAs of STIL and finally selected 19 miRNAs. A Cytoscape visualization of the miRNAs-STIL regulatory network was constructed (Fig. 7A). Mechanism of action relied on miRNA indicating that there should be a negative association between miRNA and STIL. We then performed the differential analysis and correlation analysis, and only hsa-miR-204-5p was retained, with a negative correlation of STIL and significantly downregulated expression in tumor tissue than normal tissue (Fig. 7B, C). Finally, we explored the prognostic value of hsa-miR-204-5p, and found that high levels of hsa-miR-204-5p predict a favorable prognosis for HCC (Fig. 7D). Taken together, hsa-miR-204-5p might be the most potential regulatory miRNA of STIL in HCC.

Fig. 7figure 7

has-miR-204-5p was screened out as a potential upstream miRNA of STIL in HCC. A Cytoscape software displayed the miRNAs-STIL regulatory network. B The expression relationship between has-miR-204-5p and STIL in HCC. C The differential expression of has-miR-204-5p in HCC tissues and normal tissues. D Kaplan–Meier (KM) survival analysis of has-miR-204-5p in HCC

Next, the 119 upstream lncRNAs of hsa-miR-204-5p were predicted respectively using ENCORI database (Additional file 1: Table S1). Then, the expression analysis, survival analysis, and correlation analysis were determined using the Wilcoxon test, log-rank test, and Spearman's correlation analysis, respectively, in R software. We considered the most potential upstream lncRNAs of has-miR-204-5p should meet the following criteria simultaneously. First, the expression of these lncRNAs was overexpressed in tumor tissue than normal tissue with P-value < 0.001 (Fig. 8A, B); Second, these lncRNAs were significantly associated with prognosis (Fig. 8C, D); Third, based on the ceRNA hypothesis, these lncRNAs should be a negative relationship with hsa-miR-204-5p and positive correlation with STIL (Fig. 8E–H). Last, there were two lncRNAs (CCNT2-AS1 and SNHG1) with the most potential upstream lncRNAs of hsa-miR-204-5p retained in this study, and a lncRNA-miRNA-mRNA regulatory network was constructed (Fig. 9).

Fig. 8figure 8

Differential expression analysis, KM survival analysis and Spearman’s correlation analysis. Differential expression analysis showed that CCNT2-AS1 (A) and SNHG1 (B) expression were significantly higher in tumor tissues than normal tissues; KM survival curves showed that the high expression of CCNT2-AS1 (C) and SNHG1 (D) exhibited a poor survival. The has-miR-204-5p expression possessed a significantly negative relationship with CCNT2-AS1 (E) and SNHG1 (F); The STIL expression possessed a significantly positive relationship with CCNT2-AS1 (G) and SNHG1 (H)

Fig. 9figure 9

The model of CCNT2-AS1/SNHG1- has-miR-204-5p -STIL axis in carcinogenesis of HCC

STIL expression correlates with the infiltration levels of immune cells in HCC

Dysregulated STIL is exert a crucial function in chromosomal instability which is found to be involved in the tumor immune microenvironment. Besides, tumor-infiltrating immune cells have a huge effect on the clinical outcomes of patients with multiple cancers [50, 51]. Therefore, we investigated the association between STIL expression and the tumor-infiltrating immune cells in HCC using the TIMER database. We observed that the levels of STIL expression were positively associated with tumor purity (cor = 0.144, P = 7.13e−03), indicating STIL expression was found mainly from the tumor cells. Moreover, the expression levels of STIL were positively associated with the infiltration levels of B cells (r = 0.442, P = 7.43e−18), CD8+ T cells (r = 0.333, P = 2.63e−10), CD4 + T cells (r = 0.445, P = 3.80e−18), Macrophages (r = 0.524, P = 1.90e−25), Neutrophils (r = 0.452, P = 8.21e−19), DCs (r = 0.511, P = 5.02e−24) in HCC tissues (Fig. 10A). These findings indicated that STIL is closely correlated with the levels of immune cells in HCC.

Fig. 10figure 10

Association between STIL expression and tumor-infiltrating immune cells (TIICs) and related functions or pathways in HCC. A Exploiting the TIMER database, the relationship between STIL and the infiltration levels of B cells, CD8 + T cells, CD4 + T cells, macrophages, neutrophils and dendritic cells was displayed in scatterplots based on the purity-corrected Spearman method; B, C Comparison of the ssGSEA scores between different STIL expression groups in the TCGA cohort

To further probe the relationship between the STIL expression and immune status, we computed the ES (enrichment scores) of sixteen immune cells and thirteen immune-related pathways and functions with ssGSEA. Interestingly, the high-STIL expression group has a higher level of scores in aDCs (activated dendritic cells) and MHC class I, which both related to the antigen presentation process, than the low-STIL expression group in the TCGA cohort (adjusted P < 0.05, Fig. 10B). Moreover, the score levels of type II IFN response, Neutrophils, B cells, NK cells, and Cytolytic activity were lower in the high-STIL expression group, while the activity of Tregs and Tfh was just the opposite (adjusted P < 0.05, Fig. 10C).

Correlation analysis between STIL and biomarkers of potential immunotherapy targets

The inhibitory molecules-targeted immunotherapies, such as anti-CTLA-4, -PD-1/L1, and -CD47 inhibitors, have offered new opportunities for treating patients with hard-to-treat malignancies and thus improving their survival rates [52,53,54]. We then assessed the relationship between STIL and multiple promising immunotherapy targets, including LAG3, CD274 (PD-L1), CTLA4, CD47, PDCD1 (PD-1), and HAVCR2 (TIM-3), and results showed STIL was positively associated with these immunotherapy targets (Fig. 11A–F). Moreover, these immunotherapy targets were significantly upregulated in the high-STIL expression group (Fig. 11G).

Fig. 11figure 11

Relationship between the expression of STIL and several promising immunotherapeutic targets via the TIMER database. AF The scatterplots of correlation between STIL expression and the promising immunotherapeutic targets (CD274, PDCD1, CTLA-4, LAG3, HAVCR2, and CD47. G The levels of promising immunotherapeutic targets between the high- and low-STIL expression groups

STIL predicts immunotherapy/chemotherapy efficacy

The strong positive relationship between STIL and immune cell infiltrations and multiple promising immunotherapy molecules that were overexpressed in patients with high-STIL expression prompted us to address whether the STIL expression level predicts immunotherapy efficacy. An immunotherapy cohort (IMvigor210 cohort) integrating the therapeutic effects of patients who underwent immunotherapy (anti-PD-L1) and transcriptome data was utilized to explore the prognostic value of the STIL expression for immune blockade therapy. Patients with high-STIL expression were associated with a more favorable prognosis than those with low-STIL expression (Fig. 12A). The complete/partial response (CR/PR) has a higher STIL expression than the SD/PD efficacy subgroup after dividing the immunotherapeutic efficacy in a binary mode (Fig. 12B). We also found that the high-STIL expression group has a higher percentage of CR/PR than the low-STIL group (Fig. 12C). Moreover, the treatment-related IC50 of six agents (including cisplatin, paclitaxel, gemcitabine, doxorubicin, sorafenib, and sunitinib) in patients from TCGA-LIHC cohort was performed to evaluate the sensitivity of the two risk subsets to these drugs, and we observed that STIL-high patients were more sensitive to the gemcitabine, doxorubicin, while more resistant to the sorafenib and sunitinib (Fig. 12D–I).

Fig. 12figure 12

The expression level of STIL can predict immunotherapy/chemotherapy efficacy. A KM survival curves showed that patients with high-STIL expression have a favorable prognosis than those with low-STIL expression in IMvigor 210 cohort. B The levels of STIL expression were grouped by immunotherapy efficacy. C Comparison of immunotherapy efficacy between the high- and low-STIL expression groups. DI The comparison of the IC50 values of six common chemotherapeutic drugs between the high- and low-STIL expression groups

STIL knockdown inhibits HCC growth, invasion, and migration

We first performed an immunohistochemistry assay with STIL antibody to validate our findings, using a tissue microarray (TMA). Figure 13A showed the representative image of STIL expression in the tumor and adjacent tissues. In differential analysis, the expression of STIL protein was significantly upregulated in tumor tissue compared with adjacent normal tissue (Fig. 13B). Survival analysis suggested that patients with high-STIL expression were associated with poor prognosis (Fig. 13C). We then investigate the biological effects of siRNA-induced knockdown of STIL expression on HCC cell lines. Knockdown of STIL in HCC cell lines, as verified by western blot assay (Fig. 14A, Additional file 2: Figure S1), significantly reduced the proliferation of HCC cells, using Immunofluorescence assay, CCK8 assay, and colony-forming assay (Fig. 14B–E). Evaluation of the impacts of STIL knockdown on the invasion and migration ability of HCC cells, as determined by transwell (Fig. 15A, B) and wound healing assays (Fig. 15C), showed that knockdown of STIL inhibited cell invasion and migration. Collectively, these results showed that STIL was highly expressed in tumor tissue, high-STIL expression predicts poor OS, and was significantly associated with HCC cell proliferation, invasion, and migration.

Fig. 13figure 13

The expression of STIL in HCC tissues and its prognostic significance were assessed by immunohistochemistry. A Representative IHC image of STIL expression in tumor and adjacent tissues. B showed that the expression of STIL is distinct between tumor tissues and adjacent liver tissues. C High STIL expression in HCC was associated with poor clinical prognosis

Fig. 14figure 14

STIL-silencing impeded HCC cell proliferation properties in vitro. A Expression levels of STIL were determined by western blot in HCC cells treated with siRNA-STIL and siRNA-con. STIL knockout hindered HCC cell growth assessed by immunofluorescence assay (B), CCK-8 assay (C, D) and colony-forming assay (D)

Fig. 15figure 15

STIL-silencing blocked HCC cell migration and invasion in vitro. A, B Representative images and computed of the number of cells that migrated or invaded in transfected HCC cell lines. Experiments were repeated three times with similar results, and error bars represent the mean ± SEM, ***p < 0.001 **p < 0.01. C Wound-healing assays were used to detect the migration ability of transfected cells

留言 (0)

沒有登入
gif