ATP6V1C1, associated with the tumor microenvironment and mTORC1 signaling pathway, is a potential diagnostic, prognostic, and therapeutic biomarker for hepatocellular carcinoma Authors: Yuhao Pan, Hao Chen, Chenhui Lv, Wei He, Yongpeng Xu, Qijia Xuan Published: 2024 ABSTRACT BackgroundHepatocellular carcinoma (HCC) is a global health challenge with high mortality. ATP6V1C1, one of the subunit genes of vacuolar adenosine triphosphatase (V-ATPase), is a potential oncogene. However, its role in HCC remains unclear.Materials and methodsDifferential analysis of mRNA and microRNA (miRNA), combined with machine learning, identified ATP6V1C1 as a potential biomarker for HCC. The expression and prognostic role of ATP6V1C1 in HCC were evaluated. Additionally, we explored the distribution of ATP6V1C1 in HCC tumor microenvironment (TME) at single-cell and spatial transcriptome levels. Furthermore, the association between ATP6V1C1 and malignant biological features, TME characteristics, and therapy response in HCC was investigated. Finally, in vitro experiments validated the effects of ATP6V1C1 on the malignant phenotype of HCC.ResultsATP6V1C1 had higher expression in HCC tissues compared to paired normal tissues. Upregulated ATP6V1C1 was associated with poor HCC prognosis. ATP6V1C1 was primarily expressed in malignant cells and the tumor region in HCC TME. A positive correlation was observed between ATP6V1C1 expression and the activation of cancer-related pathways. The high ATP6V1C1 expression group exhibited increased pro-tumorigenic immune infiltration, inhibited anti-tumor immune activity, and high tumor proliferation rate. HCC patients of low ATP6V1C1 expression group had more clinical response to anti-tumor therapies. Knockdown of ATP6V1C1 impaired the proliferation, migration, and invasion of HCC cells by downregulating the mTORC1 signaling pathway.ConclusionATP6V1C1 multifacetedly contributes to the oncogenesis and progression of HCC and is a promising diagnostic and prognostic biomarker with predictive value on therapy response.Supplementary InformationThe online version contains supplementary material available at 10.1007/s12672-024-01578-w. Introduction Hepatocellular carcinoma (HCC), accounting for over 80% of primary liver cancers, ranks fourth in cancer mortality worldwide [1]. Most HCC patients gained their diagnosis at an advanced stage, missing the opportunity for radical therapy due to diagnostic limitations [2]. Despite advances in treatment strategies, including surgical resection, liver transplantation, ablation, transcatheter arterial chemoembolization (TACE), targeted therapy, and immunotherapy, the 5-year relative survival rate for HCC remains inferior [3, 4]. Consequently, novel diagnostic and prognostic biomarkers for HCC, which guide personalized therapeutic strategies, are urgently needed. Vacuolar adenosine triphosphatase (V-ATPase), composed of a peripheral V1 domain and an integral V0 domain, participates in a collection of physiological processes, many of which become abnormal in cancer [5–7]. It has been reported that V-ATPase plays a part in cell signaling, such as Wnt, Notch, and mTOR signaling, and has a impact on tumor cell survival, invasion, migration, and metastasis [8, 9]. ATP6V1C1, one of the main genes regulating V-ATPase activity, has been shown to inhibit autophagy and enhance radiotherapy resistance in esophageal squamous cell carcinoma via the mTOR signaling pathway [10, 11]. Additionally, a study by Feng et al. manifested that ATP6V1C1 co-localized with F-actin in 4T1 cells with high expression contributing to breast cancer metastasis by regulating filament actin arrangement [12]. However, the role of ATP6V1C1 in HCC remains ambiguous. MicroRNAs (miRNAs), small noncoding RNAs composed of 21–22 nucleotides, modulate gene expression by binding to mRNA targets specifically and silencing them post-transcriptionally [13]. It is notable that miRNAs play a vital part in the pathogenesis of HCC, serving as potentially valuable diagnostic and prognostic biomarkers [14]. Therefore, further analysis combining differentially expressed genes (DEGs) and differentially expressed microRNAs (DEMs) between HCC and normal liver tissues is expected to identify novel biomarkers for HCC. In this study, we aimed to explore potential biomarkers for HCC through the analysis of DEGs and DEMs between HCC and normal liver tissues. DEGs and DEMs were obtained from three mRNA and three miRNA expression profiling datasets of HCC from the Gene Expression Omnibus (GEO) database respectively. The intersection of final DEGs (fDEGs) and target genes of final DEMs (fDEM-TGs), followed by selection using machine learning, identified ATP6V1C1 as a key gene. Comprehensive bioinformatics analysis reveals that ATP6V1C1 is not only a risk factor for HCC, with good performance in distinguishing between HCC and normal liver tissues, but also highly associated with tumor malignancy, tumor microenvironment (TME), and therapy response in HCC. In addition, in vitro experiments demonstrate that ATP6V1C1 facilitates HCC cell proliferation, migration, and invasion by regulating the mTORC1 signaling pathway. Therefore, ATP6V1C1 may serve as a novel diagnostic, prognostic and therapeutic biomarker for HCC. Materials and methods Data collection We downloaded three mRNA expression profiling datasets (GSE214846, GSE77314, and GSE105130) and three miRNA expression profiling datasets (GSE147889, GSE63046, and GSE176288) of HCC patients from the GEO database (https://www.ncbi.nlm.nih.gov/geo/) to gain DEGs and DEMs. The TCGA-LIHC, ICGC-LIRI-JP, and GSE116174 datasets, which contained mRNA expression levels and clinical information of HCC patients, were obtained from The Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/), International Cancer Genome Consortium (ICGC, https://dcc.icgc.org/), and the GEO databases. The GSE91061, GSE165252, GSE109211, GSE104580, and IMvigor210-Kidney cancer datasets, utilized to evaluate the response of HCC patients to anti-tumor therapies, were extracted from the GEO database and IMvigor210 Core Biologies (http://research-pub.gene.com/IMvigor210CoreBiologies) [15]. All the above datasets enrolled in this work were publicly available (Table S1). DEGs/DEMs/fDEGs/fDEMs/fDEM-TGs analysis Using the “limma” R package, DEGs and DEMs were identified between HCC and paired normal tissue samples in mRNA/miRNA datasets. The RemoveBatchEffect function in the “limma” R package was utilized to eliminate the batch effect within datasets. The threshold for DEGs was set as false discovery rate (FDR) < 0.05 and |Fold Change (FC)|> 2, while for DEMs, the threshold was set as FDR < 0.05 and |FC|> 1.2. We then overlapped upregulated or downregulated DEGs/DEMs across all mRNA/miRNA datasets, which yielded fDEGs/fDEMs. Subsequently, fDEM-TGs were predicted using the online databases miRTarBase (https://mirtarbase.cuhk.edu.cn/) and miRmap (https://mirmap.ezlab.org/) [16, 17]. Finally, the intersection of fDEM-TGs and fDEGs identified key genes for further analysis. Identification of hub gene based on machine learning The Least Absolute Shrinkage and Selection Operator (LASSO) regression analysis and Random Survival Forest (RSF) analysis were employed to screen hub genes related to overall survival (OS). The intersection of prognostic hub genes identified by these two machine learning methods determined robust prognostic hub genes. Pan-cancer analysis of expression and prognostic value of ATP6V1C1 Gene expression data and clinical information of pan-cancer were downloaded from the TCGA and the Genotype-Tissue Expression Project (GTEx) databases (https://www.gtexportal.org/). Expression patterns of ATP6V1C1 in 17 HCC bulk RNA-seq datasets were extracted from the Integrative Molecular Database of Hepatocellular Carcinoma (HCCDB) database (http://lifeome.net:809/#/home/) [18]. Single-cell and spatial transcriptome analysis of ATP6V1C1 We used the Tumor Immune Single-cell Hub 2 (TISCH2) database (http://tisch.comp-genomics.org/) to decipher the expression of ATP6V1C1 in TME of HCC at single-cell resolution [19]. The expression of ATP6V1C1 in HCC in a spatial dimension was explored using the HCCDB database. Protein expression analysis and perturbation effect analysis of ATP6V1C1 Protein expression data for ATP6V1C1 in liver cancer and normal liver tissues were obtained from The Human Protein Atlas (HPA) database (https://www.proteinatlas.org/) and analyzed using the “HPAanalyze” R package [20]. Differences in ATP6V1C1 proteomic expression profiles between HCC and normal liver tissues were explored through The University of Alabama at Birmingham Cancer (UALCAN) database (https://ualcan.path.uab.edu/index.html), based on data from Clinical Proteomic Tumor Analysis Consortium (CPTAC) database (https://proteomics.cancer.gov/programs/cptac/). Perturbation effects of ATP6V1C1 in HCC cell lines were analyzed using the Dependency Map (DepMap) database (https://depmap.org/portal/). Analysis of malignant biological features of ATP6V1C1 Based on the study by Ally et al. [21], we explored the association between ATP6V1C1 and three molecular subtypes (iCluster 1, 2, and 3) of HCC. Gene set enrichment analysis (GSEA) was performed based on the HALLMARK gene set from the Human Molecular Signatures Database (MSigDB) database (https://www.gsea-msigdb.org/gsea/msigdb/index.jsp/) using the “clusterProfiler” R package. Correlations between ATP6V1C1 and ononcogenic pathways were evaluated using the “progeny” R package and the IMPACT database (http://www.brimpact.cn/ or http://impact.brbiotech.com/) [22, 23]. Fraction altered, aneuploidy score, and homologous recombination deficiency (HRD) of each TCGA-LIHC patient were gained from the study by Thorsson et al. [24] Exploration of TME We collected 29 TME signatures from the study by Bagaev et al. [25] Based on the “GSVA” R package, the single-sample gene set enrichment (ssGSEA) algorithm was then used to quantify TME signatures of each TCGA-LIHC patient. Subsequently, these signatures were further categorized into four TME scores. TIMER2.0 database (http://timer.cistrome.org/) was used to assess the correlation between ATP6V1C1 and immune cell infiltration [26]. Using Tracking Tumor Immunophenotype (TIP) database (http://biocc.hrbmu.edu.cn/TIP/), we explored the relationship between ATP6V1C1 and seven-step cancer-immunity cycle [27]. Additionally, the association between ATP6V1C1 and intratumoral immune states was evaluated based on the study by Thorsson et al. The predictive value of ATP6V1C1 on immunotherapy and potential drugs for ATP6V1C1 Tumor Immune Dysfunction and Exclusion (TIDE) algorithm (http://tide.dfci.harvard.edu/) was employed to predict the response of HCC patients to immunotherapy [28]. Furthermore, potential small molecule drugs targeting ATP6V1C1 was identified through the connectivity map (CMap) database (https://clue.io/) [29]. Cell lines and cell culture The human HCC cell lines Huh7, Hep3B, and HCCLM3 were purchased from the iCell, Shanghai, China. The cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Hyclone, USA) containing 10% fetal bovine serum (FBS; TransGen Biotec, China) at 37 °C and 5% CO2 in the incubator (Thermo Fisher Scientific, USA). Cell transient transfection According to the manufacturer’s instruction, Negative Control (NC) and ATP6V1C1 siRNA (Proteintech, China) were transfected into HCC cells using Lipofectamine 2000 (Invitrogen, USA). The sequence of the ATP6V1C1 siRNA was as follows: 5′ GCAGGAAGUUUGCUAACUATT 3′. RNA extraction and quantitative reverse transcription PCR (qRT‑PCR) Total RNA was extracted from the cells using TRIZOL reagent (Invitrogen, USA), followed by reverse transcription using RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, USA). The expression level of ATP6V1C1 was calculated using the comparative ΔΔCT method with β-actin as the reference gene. Primer sequences were listed as follows: ATP6V1C1, forward: 5′ ACATGGCTGATGTATTGGAAGA 3′, and reverse: 5′ GTCCCACTGGAACCTTGTTAT 3′; β-actin, forward: 5′ GGAAATCGTGCGTGACATTAAG 3′, and reverse: 5′ AGCTCGTAGCTCTTCTCCA 3′. Western blotting The cells were lysed in RIPA buffer mixed with PMSF (Beyotime, China) to extract total proteins. Protein samples with exact amount were separated by SDS-PAGE and transferred to the PVDF membrane. Next, the PVDF membrane was incubated overnight with primary antibodies containing skim milk powder at 4 °C, followed by being rinsed three times and incubated with secondary antibodies at room temperature for 30 min. Subsequently, Tanon-5200 Chemiluminescent Image System (Tanon, China) was used for visualizing. Primary antibody: ATP6V1C1 (Proteintech, China), mTOR (Proteintech, China), p-mTOR (Proteintech, China), p70-S6K (Proteintech, China), p-p70-S6K (Proteintech, China), 4E-BP (Affinity Biosciences, USA), p-4E-BP(Affinity Biosciences, USA), and GAPDH (Proteintech, China). Cell counting kit-8 (CCK-8) assay Cell proliferation capacity was detected using the CCK-8 Cell Counting Kit (Vazyme, China). HCC cells were seeded at 37 °C overnight in a 96-well plate with 2000 cells per well. After incubating for 72 h, 10 μl CCK-8 reagent was added to the wells. Absorbance was measured at 450 nm after the cells were cultivated in the dark at 37 °C for 2 h. Colony formation assay HCC cells were inoculated in the 6-well plate with 1200 cells per well and incubated at 37 °C for 7 days. After washed in PBS twice, the cells were fixed with 4% paraformaldehyde (biosharp, China) for 20 min and stained with crystal violet (Solarbio, China) for 10 min. The colonies were counted using ImageJ software. Wound healing assay HCC cells were inoculated as a single-layer in a 6-well plate and cultured with serum-free DMEM. A wound was created using a 200 μL pipette tip. After incubating for 24 h, the wound was observed using a microscope and wound healing rate was analyzed using ImageJ software. Transwell assay Matrigel and 24-well Transwell plates with 8μm pore size (Corning, USA) were used for invasion experiments. The upper chamber of the Transwell chamber was wrapped with Matrigel at 37 °C for 30 min. Later, HCC cells in 200 μl serum-free DMEM were plated onto the upper chamber and 600 μL 20% FBS DMEM was loaded into the lower chamber. After incubating for 72 h, the cells were fixed with 4% paraformaldehyde for 20 min and stained with crystal violet for 10 min. The inverted microscope (Olympus Corporation) was used to photograph three randomly selected images. Statistical analysis All statistical calculations were performed using R software (Version 4.3.1) or GraphPad Prism 9.5 software. Differences between two groups were assessed using the unpaired Student’s t-test or Wilcoxon rank sum test. The prognostic value of ATP6V1C1 in pan-cancer was assessed using univariate Cox regression analysis. Multivariate Cox regression analysis was performed to confirm whether ATP6V1C1 was an independent prognostic factor when age, sex, and stage were included as covariates. The “survival” and “survminer” R packages were used to perform survival analysis and determine the optimal cut-off point. Kaplan‒Meier (KM) curves were used for survival analysis and the statistical significance was evaluated using the log-rank test. Correlation analyses were performed using the Spearman method. Categorical variables were analyzed using the Chi-square test or Fisher exact test. Experiments were repeated three times and experimental data were expressed as mean ± standard deviation (SD). P < 0.05 was indicated as statistical significance (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001). Results Identification of DEGs/DEMs/fDEGs/fDEMs/fDEM-TGs As for mRNA, 3940, 974, and 715 DEGs were screened from GSE214846, GSE77314, and GSE105130 datasets respectively (Fig. 1A). After overlapping upregulated or downregulated DEGs, 475 fDEGs were gained, of which 265 genes were upregulated and 210 genes were downregulated (Fig. 1B). In terms of miRNA, 67, 46, and 46 DEMs were identified from GSE147889, GSE63046, and GSE176288 datasets respectively (Fig. 1C). The intersections of upregulated or downregulated DEMs shared 14 fDEMs, of which 8 miRNAs were upregulated and 6 miRNAs were downregulated (Fig. 1D). Subsequently, 2547 and 2761 target genes of fDEMs were predicted using miRTarBase and miRmap respectively. 781 fDEMs-TGs were then obtained after intersecting these target genes (Fig. 1E). Finally, the intersection of fDEMs-TGs and fDEGs shared 23 key genes for further analysis (Fig. 1F).Fig. 1Identification of DEGs, DEMs, and ATP6V1C1. (A) Volcano plots of DEGs from the GSE214846, GSE77314, and GSE105130 datasets. (B) Venn plots of the intersection of upregulated and downregulated DEGs from the GSE214846, GSE77314, and GSE105130 datasets. (C) Volcano plots of DEMs from the GSE147889, GSE63046, and GSE176288 datasets. (D) Venn plots of the intersection of upregulated and downregulated DEMs from the GSE147889, GSE63046, and GSE176288 datasets. (E) Venn plot shows the intersection of target genes of fDEMs predicted by mirtarbase and miRmap. (F) Venn plot shows the intersection of fDEM-TGs and fDEGs. (G) Identification of prognosis-related genes using LASSO and RSF algorithms. (F) Identification of ATP6V1C1 through overlapping prognostic hub genes obtained from LASSO and RSF algorithms Identification of DEGs, DEMs, and ATP6V1C1. (A) Volcano plots of DEGs from the GSE214846, GSE77314, and GSE105130 datasets. (B) Venn plots of the intersection of upregulated and downregulated DEGs from the GSE214846, GSE77314, and GSE105130 datasets. (C) Volcano plots of DEMs from the GSE147889, GSE63046, and GSE176288 datasets. (D) Venn plots of the intersection of upregulated and downregulated DEMs from the GSE147889, GSE63046, and GSE176288 datasets. (E) Venn plot shows the intersection of target genes of fDEMs predicted by mirtarbase and miRmap. (F) Venn plot shows the intersection of fDEM-TGs and fDEGs. (G) Identification of prognosis-related genes using LASSO and RSF algorithms. (F) Identification of ATP6V1C1 through overlapping prognostic hub genes obtained from LASSO and RSF algorithms Machine learning identifies prognostic hub genes In order to obtain prognostic hub genes, we put key genes previously selected into LASSO regression analysis and RSF analysis, and screened 7 hub genes related to OS of patients in the TCGA-LIHC dataset respectively (Fig. 1G). By overlapping these prognostic hub genes, four genes, namely DEPDC1, CCDC137, ATP6V1C1, and LAPTM4B, were then gained (Fig. 1H). In our study, the role of ATP6V1C1 in HCC was explored for it remained unclear. The differential expression of ATP6V1C1 in different GEO datasets was shown in Table S2. Additionally, the expression patterns of hsa-miR-130a-3p and hsa-miR-542-3p, which regulated ATP6V1C1, were displayed in Table S3. The AUC values of two miRNAs were both > = 0.70 in the GSE147889, GSE63046, and GSE176288 datasets, suggesting a good ability to discriminate between HCC and normal liver tissues (Fig. S1A,B). The diagnostic and prognostic value of ATP6V1C1 We further explored the diagnostic value of ATP6V1C1. As shown in Fig. 2A, the AUC values of ATP6V1C1 were all > = 0.90 in the GSE214846, GSE63046, GSE176288, and TCGA-LIHC datasets, indicating good performance in distinguishing between HCC and normal liver tissues. Across 17 HCC datasets from the HCCDB database, ATP6V1C1 showed significant upregulation in HCC tissues except for five datasets (Fig. S1C). As for pan-cancer analysis, ATP6V1C1 displayed differential expression across various tumor types except for kidney chromophobe (KICH), pancreatic adenocarcinoma (PAAD), testicular germ cell tumors (TGCT), and uterine carcinosarcoma (UCS; Fig. 2B). Collectively, these findings suggest that ATP6V1C1 may be a promising molecular tool for diagnosis in HCC and even pan-cancer.Fig. 2The diagnostic and prognostic value of ATP6V1C1. (A) ROC curves of ATP6V1C1 in the GSE214846, GSE77314, GSE105130, and TCGA-LIHC datasets. (B) Pan-cancer analysis of the difference in expression of ATP6V1C1 between tumor and normal tissues. (C) KM curves show the difference in OS between HCC patients with high and low ATP6V1C1 expression in the TCGA-LIHC, ICGC-LIRI-JP, and GSE116174 datasets. (D) Univariate Cox regression analysis of ATP6V1C1 in pan-cancer. (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001) The diagnostic and prognostic value of ATP6V1C1. (A) ROC curves of ATP6V1C1 in the GSE214846, GSE77314, GSE105130, and TCGA-LIHC datasets. (B) Pan-cancer analysis of the difference in expression of ATP6V1C1 between tumor and normal tissues. (C) KM curves show the difference in OS between HCC patients with high and low ATP6V1C1 expression in the TCGA-LIHC, ICGC-LIRI-JP, and GSE116174 datasets. (D) Univariate Cox regression analysis of ATP6V1C1 in pan-cancer. (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001) To assess the prognostic value of ATP6V1C1, survival analysis was then performed. As depicted in Fig. 2C, HCC patients with high ATP6V1C1 expression showed shorter OS than those with low ATP6V1C1 expression in the TCGA-LIHC (P < 0.001), ICGC-LIRI-JP (P = 0.002), and GSE116174 (P = 0.003) datasets. Furthermore, HCC patients with high ATP6V1C1 expression had shorter disease free interval (DFI; P = 0.02), disease specific survival (DSS; P = 0.002), and progression free interval (PFI; P = 0.015) compared to their low counterparts in the TCGA-LIHC dataset (Fig. S1D). Multivariate Cox regression analysis showed that ATP6V1C1 was an independent prognostic factor in the TCGA-LIHC (HR = 1.64, P < 0.001), ICGC-LIRI-JP (HR = 1.73, P = 0.01), and GSE116174 (HR = 1.61, P = 0.024) datasets with age, sex, and stage included as variables (Fig. S2A). These results indicate that ATP6V1C1 is a poor prognostic factor for HCC patients. To better predict OS of patients with HCC, we established a predictive nomogram after inputting ATP6V1C1 and clinical features into stepwise Cox regression analysis (Fig. S2B). The nomogram displayed excellent predictive performance in three HCC datasets (Fig. S2C). In terms of pan-cancer analysis, ATP6V1C1 was found to be associated with OS of KICH, breast invasive carcinoma (BRCA), uveal melanoma (UVM), head and neck squamous cell carcinoma (HNSC), kidney renal clear cell carcinoma (KIRC), and glioma (GBMLGG) patients (Fig. 2D). The expression of ATP6V1C1 in HCC TME at single-cell and spatial transcriptome levels To further understand the distribution of ATP6V1C1 in HCC TME, we collected analysis of ATP6V1C1 at single-cell and spatial transcriptome levels from public databases. As for single-cell transcriptome analysis, the distribution of different cell types in three single-cell sequencing HCC datasets from TISCH2 database was shown in Fig. 3A–C. ATP6V1C1 was primarily expressed in Tprolif cells, Mono/Macro cells, and malignant cells (Fig. 3D). Regarding spatial transcriptome analysis, hematoxylin and eosin (H&E) staining and spatial cluster distribution of five HCC patients obtained from the HCCDB database were visualized in Fig. 3E–I. ATP6V1C1 had higher expression in the tumor region than in the adjacent, stromal, and immune regions in five HCC patients (Fig. 3J).Fig. 3Single-cell and spatial transcriptome analysis of ATP6V1C1. (A–C) The cell types and distribution of ATP6V1C1 in different cells types in the GSE125449 (A), GSE146115 (B), and GSE166635 (C) datasets. (D) The normalized expression levels of ATP6V1C1 across different cell types in the GSE125449, GSE146115, and GSE166635 datasets. (E–I) H&E staining, spatial cluster distribution, and distribution of ATP6V1C1 in different regions in spatial transcriptome data. (J) The normalized expression levels of ATP6V1C1 in spatial transcriptome data Single-cell and spatial transcriptome analysis of ATP6V1C1. (A–C) The cell types and distribution of ATP6V1C1 in different cells types in the GSE125449 (A), GSE146115 (B), and GSE166635 (C) datasets. (D) The normalized expression levels of ATP6V1C1 across different cell types in the GSE125449, GSE146115, and GSE166635 datasets. (E–I) H&E staining, spatial cluster distribution, and distribution of ATP6V1C1 in different regions in spatial transcriptome data. (J) The normalized expression levels of ATP6V1C1 in spatial transcriptome data Protein expression and perturbation effects of ATP6V1C1 in HCC Based on the Immunohistochemistry (IHC) analysis of ATP6V1C1 from the HPA database, we found that ATP6V1C1 was upregulated in liver cancer tissues compared to normal liver tissues at the protein level (Fig. 4A, B). Additionally, protein expression of ATP6V1C1 in HCC tissues was significantly higher than in normal liver tissues in the CPTAC database (P < 0.001; Fig. 4C). As for perturbation effects of ATP6V1C1, analysis from the DepMap database showed that knockout of ATP6V1C1 using CRISPR screen inhibited cell proliferation in 21 HCC cell lines (Fig. 4D).Fig. 4Protein expression and perturbation effects of ATP6V1C1. (A, B) IHC analysis of ATP6V1C1 in the HPA database. (C) Difference in protein expression of ATP6V1C1 between tumor and normal tissues in the CPTAC database. (D) Perturbation effects of ATP6V1C1 in 21 HCC cell lines. (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001) Protein expression and perturbation effects of ATP6V1C1. (A, B) IHC analysis of ATP6V1C1 in the HPA database. (C) Difference in protein expression of ATP6V1C1 between tumor and normal tissues in the CPTAC database. (D) Perturbation effects of ATP6V1C1 in 21 HCC cell lines. (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001) ATP6V1C1 is associated with malignant clinicopathological and cancer features of HCC According to the median expression level of ATP6V1C1, we divided HCC patients into high and low ATP6V1C1 expression groups. In the GSE116174 dataset, HCC patients with vascular invasion had significantly higher ATP6V1C1 expression than those without (P < 0.01) and a larger proportion of vascular invasion was observed in the high ATP6V1C1 expression group (62.5% vs 28.1%, P < 0.05; Fig. 5A). Although the difference in ATP6V1C1 expression between HCC patients with and without vascular invasion had no statistical significance (P = 0.078), the high ATP6V1C1 expression group was also found to have a larger proportion of vascular invasion in the TCGA-LIHC dataset (40.3% vs 28.3%, P < 0.05; Fig. 5B). Besides, HCC patients with stage I had significantly lower ATP6V1C1 expression compared to stage III in the GSE116174 dataset (Fig. 5C), consistent with the result that lower HCC ATP6V1C1 expression was found in HCC patients with stage I and stage II than stage III & stage IV in the ICGC-LIRI-JP dataset (Fig. 5D). However, there was no association between ATP6V1C1 expression and tumor stage of HCC patients in the TCGA-LIHC dataset due to the heterogeneity of HCC datasets (Fig. 5E). Additionally, with the increase in tumor grade, the expression of ATP6V1C1 gradually got higher (Fig. S3A), while no correlation was observed between ATP6V1C1 expression and tumor T stage in the TCGA-LIHC dataset (Fig. S3B). As for three iClusters of HCC, a larger proportion of iCluster 1 while a smaller proportion of iCluster 2 was found in the high ATP6V1C1 expression group (43.5% vs 22.7%, 23.5% vs 38.6%, P < 0.01; Fig. 5F). Compared to iCluster 2, iCluster 1 exhibited higher tumor grade, more macrovascular invasion, and worse prognosis. These results show that ATP6V1C1 contributes to malignant clinicopathological features of HCC.Fig. 5Relationship of ATP6V1C1 with malignant clinicopathological and cancer features. (A, B) Difference in ATP6V1C1 expression levels between HCC patients with and without vascular invasion, and percentage of HCC patients with vascular invasion in high and low ATP6V1C1 expression groups in the GSE116174 (A) and TCGA-LIHC (B) datasets. (C–E) Differences in ATP6V1C1 expression among HCC patients with different tumor stages in the GSE116174 (C), ICGC-LIRI-JP (D), and TCGA-LIHC (E) datasets. (F) Percentage of three molecular subtypes in high and low ATP6V1C1 expression groups. (G) GSEA shows the enrichment of cancerogenic pathways in high ATP6V1C1 expression group. (H) Correlations between ATP6V1C1 expression and oncogenic pathways assessed by progeny algorithm. (I) Correlations between ATP6V1C1 expression and cancer-related pathways in the IMPACT database. (J) Correlations between ATP6V1C1 expression and DNA damage measures. (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001) Relationship of ATP6V1C1 with malignant clinicopathological and cancer features. (A, B) Difference in ATP6V1C1 expression levels between HCC patients with and without vascular invasion, and percentage of HCC patients with vascular invasion in high and low ATP6V1C1 expression groups in the GSE116174 (A) and TCGA-LIHC (B) datasets. (C–E) Differences in ATP6V1C1 expression among HCC patients with different tumor stages in the GSE116174 (C), ICGC-LIRI-JP (D), and TCGA-LIHC (E) datasets. (F) Percentage of three molecular subtypes in high and low ATP6V1C1 expression groups. (G) GSEA shows the enrichment of cancerogenic pathways in high ATP6V1C1 expression group. (H) Correlations between ATP6V1C1 expression and oncogenic pathways assessed by progeny algorithm. (I) Correlations between ATP6V1C1 expression and cancer-related pathways in the IMPACT database. (J) Correlations between ATP6V1C1 expression and DNA damage measures. (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001) To further understand the malignant biological characteristics of ATP6V1C1 in HCC, we first performed a GSEA based on the HALLMARK gene set. The results showed that cancerogenic pathways, including cell cycle-related pathways, TNFα/NF-κB signaling pathway, TGF-β signaling pathway, mTORC1 signaling pathway, Wnt/β-catenin signaling pathway, hypoxia signaling pathway, etc., were more enriched in the high ATP6V1C1 expression group (Fig. 5G). The highly compatible results with those from GSEA were then obtained through the progeny algorithm, which showed that ATP6V1C1 expression was positively correlated with the activation of several oncogenic pathways (TNFα/NF-κB signaling pathway, TGF-β signaling pathway, Wnt signaling pathway, hypoxia signaling pathway, etc.; Fig. 5H). Additionally, analysis from the IMPACT database also displayed that ATP6V1C1 expression had a positive correlation with the activation of the above cancer-related pathways (e.g. mTORC1 signaling pathway, cell cycle-related pathways, and Wnt/β-catenin signaling pathway; Fig. 5I). Besides, a positive correlation was observed between ATP6V1C1 expression and DNA damage measures, including fraction altered, aneuploidy, and HRD (Fig. 5J). In terms of the mutation landscape, a larger proportion of HCC patients with gene mutations was found in the high ATP6V1C1 expression group (Fig. S3C). Further, HCC patients with CTNNB1 mutation and/or TP53 mutation exhibited higher ATP6V1C1 expression than those without (Fig. S3D). Nevertheless, there was no association between ATP6V1C1 expression and tumor mutation burden (TMB; Fig. S3E). Overall, these findings indicate that ATP6V1C1 is highly linked to tumor malignancy, which partially explains the poor prognosis of HCC patients with high ATP6V1C1 expression. Association between ATP6V1C1 and TME characteristics in HCC To further investigate the role of ATP6V1C1 in TME status, we initially quantified 29 TME signatures using the ssGSEA algorithm to evaluate their correlations with ATP6V1C1 expression. As shown in Fig. 6A, ATP6V1C1 expression showed primarily positive correlations with granulocyte traffic, matrix remodeling, myeloid cells traffic, and tumor proliferation rate. In terms of four TME scores, positive correlations were observed between ATP6V1C1 expression and the angiogenesis_fibroblast, pro-tumorigenic immune infiltration, and EMT_Proliferation TME scores, while there was no correlation between ATP6V1C1 expression and the anti-tumorigenic immune infiltration TME score (Fig. 6B). The correlation between ATP6V1C1 expression and pro-tumorigenic immune infiltration was further validated using TIMER2.0 database. Among four immune infiltration algorithms, the infiltration levels of regulatory T cells (Tregs) were positively correlated with ATP6V1C1 expression, except for the xCell algorithm. Additionally, the TIDE algorithm showed that the infiltration level of myeloid-derived suppressor cells (MDSCs) had a positive correlation with ATP6V1C1 expression (Fig. 6C). Regarding seven-step cancer-immunity cycle, although the high ATP6V1C1 expression group gained higher scores in step 1 (release of cancer antigens), step 4 (neutrophil recruiting, eosinophil recruiting, basophil recruiting, Th17 cell recruiting, and MDSC recruiting), the low ATP6V1C1 expression group had better performance in step 6 (recognition of cancer cells by T cells) and step 7 (killing of cancer cells; Fig. 6D). As for intratumoral immune states, ATP6V1C1 expression was positively correlated with TGF-β response, wounding healing, and tumor cell proliferation, whereas no correlation with IFN-γ response (Fig. 6E). Besides, C1 (wounding healing) subtype had higher ATP6V1C1 expression compared to C3 (inflammatory) and C4 (lymphocyte depleted) subtypes, and a larger proportion of C1 subtype and a smaller proportion of C3 subtype was observed in the high ATP6V1C1 expression group (Fig. 6F). C1 subtypre had higher tumor proliferation rate and worse prognosis than C3 subtype. In conclusion, these findings suggest that ATP6V1C1 is associated with TME features in HCC, which are characterized by increased pro-tumorigenic immune infiltration, inhibited anti-tumor immune activity, and high tumor proliferation rate.Fig. 6The effects of ATP6V1C1 on TME status. (A) Correlations of ATP6V1C1 expression with 29 TME signatures. (B) Scatter plots show the correlations of ATP6V1C1 expression with 4TME scores. (C) Scatter plots show the correlations of ATP6V1C1 expression with the infiltration levels of Tregs and MDSCs. (D) Differences in scores of seven-step cancer-immunity cycle between high and low ATP6V1C1 expression group. (E, F) The association between ATP6V1C1 expression and intratumoral immune states. (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001) The effects of ATP6V1C1 on TME status. (A) Correlations of ATP6V1C1 expression with 29 TME signatures. (B) Scatter plots show the correlations of ATP6V1C1 expression with 4TME scores. (C) Scatter plots show the correlations of ATP6V1C1 expression with the infiltration levels of Tregs and MDSCs. (D) Differences in scores of seven-step cancer-immunity cycle between high and low ATP6V1C1 expression group. (E, F) The association between ATP6V1C1 expression and intratumoral immune states. (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001) Potential of ATP6V1C1 to predict the response of HCC patients to anti-tumor therapies and drug candidates for ATP6V1C1 We further explored the predictive value of ATP6V1C1 on the response of HCC patients to anti-tumor therapies, including immune checkpoint blockades (ICBs), targeted drugs, and transcatheter arterial chemoembolization (TACE). Initially, TIDE database was used to evaluate the response of HCC patients to ICBs in the TCGA-LIHC dataset. The results showed that HCC patients with low ATP6V1C1 expression had lower TIDE scores (P = 0.001) and more clinical response to ICBs than those with high ATP6V1C1 expression (53.6% vs 32.6%, P < 0.001; Fig. 7A). The consistent results, the low ATP6V1C1 expression group tended to have a higher percentage of responders, were observed in the immunotherapy datasets GSE91061 (35.7% vs 10.7%, P = 0.058 in on-treatment cohort; 33.3% vs 8.0%, P = 0.037 in pre-treatment cohort), IMvigor210-Kidney cancer (25.0% vs 7.1%, P = 0.143), and GSE165252 (46.2% vs 16.7%, P = 0.202; Fig. 7B–E). In terms of targeted drug (Sorafenib), the low ATP6V1C1 expression group exhibited a larger proportion of responders in the GSE109211 dataset (45.5% vs 17.6%, P = 0.029; Fig. 7F). As for TACE, HCC patients with low ATP6V1C1 expression obtained relatively more clinical benefits than their high counterparts in the GSE104580 dataset although not significant (63.0% vs 47.3%, P = 0.08; Fig. 7G). In a word, these findings suggest that ATP6V1C1 is a potential biomarker to predict the response of HCC patients to anti-tumor therapies.Fig. 7Predictive role of ATP6V1C1 on the response of HCC patients to anti-tumor therapies and potential drugs for ATP6V1C1. (A) Difference in TIDE scores between high and low ATP6V1C1 expression groups and proportion of HCC patients responding to immunotherapy. (B–E) Percentage of clinical response to ICBs in high and low ATP6V1C1 expression groups in the GSE91061 (B, on-treatment; C, pre-treatment), IMvigor210-Kidney cancer (D), and GSE165252 (E) datasets. PD progressive disease, SD stable disease, CR complete response, PR partial response. (F) Proportion of HCC patients responding to Sorafenib in high and low ATP6V1C1 expression groups in the GSE109211 dataset. (G) Rate of clinical response to TACE in high and low ATP6V1C1 expression groups in the GSE104580 dataset. (H) CMap identified potential drugs for ATP6V1C1. (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001) Predictive role of ATP6V1C1 on the response of HCC patients to anti-tumor therapies and potential drugs for ATP6V1C1. (A) Difference in TIDE scores between high and low ATP6V1C1 expression groups and proportion of HCC patients responding to immunotherapy. (B–E) Percentage of clinical response to ICBs in high and low ATP6V1C1 expression groups in the GSE91061 (B, on-treatment; C, pre-treatment), IMvigor210-Kidney cancer (D), and GSE165252 (E) datasets. PD progressive disease, SD stable disease, CR complete response, PR partial response. (F) Proportion of HCC patients responding to Sorafenib in high and low ATP6V1C1 expression groups in the GSE109211 dataset. (G) Rate of clinical response to TACE in high and low ATP6V1C1 expression groups in the GSE104580 dataset. (H) CMap identified potential drugs for ATP6V1C1. (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001) Taking into consideration the oncogenic role of ATP6V1C1, CMap was used to identify drug candidates for ATP6V1C1 based on DEGs between high and low ATP6V1C1 expression groups. Among the top 50 potential drugs, GDC-0879, a RAF inhibitor, was considered the optimal candidate (Fig. 7H). ATP6V1C1 promotes the proliferation, migration, and invasion of HCC cells via the mTORC1 signaling pathway To confirm the effect of ATP6V1C1 on the malignant phenotype of HCC, in vitro experiments were further conducted. ATP6V1C1 expression was knocked down with siATP6V1C1 in HCC cell line Huh7, which had higher expression of ATP6V1C1 compared to the other two HCC cell lines (Hep3B and HCCLM3; Fig. S4A). The knockdown efficiency of ATP6V1C1 was verified by qRT-PCR (Fig. S4B). In CCK-8 and colony formation assays, inhibited ATP6V1C1 significantly suppressed the proliferation ability of HCC cells (Fig. 8A, B). In addition, wound-healing assay and Transwell assay indicated that ATP6V1C1 knockdown significantly hindered the migration and invasion capacities of HCC cells (Fig. 8C, D). The previous results showed that ATP6V1C1 expression had a significant positive correlation with the activation of the mTORC1 signaling pathway. Therefore, we surmised that ATP6V1C1 might contribute to HCC progression through the regulation of the mTORC1 signaling pathway. Subsequently, the activation of the crucial proteins participating in the mTORC1 signaling pathway was evaluated in Huh7-siATP6V1C1 cells to verify our speculation. As shown in Fig. 8E, phosphorylation of mTOR, p70S6K, and 4E-BP was inhibited after the knockdown of ATP6V1C1 in Huh7 cells. Overall, these findings illustrate that ATP6V1C1 promotes HCC cell proliferation, migration, and invasion via the mTORC1 signaling pathway.Fig. 8ATP6V1C1 aggravates in vitro proliferation, migration, and invasion of HCC cells via the mTORC1 signaling pathway. (A–D) CCK-8 (A), colony formation (B), Transwell (C), and wound healing (D) assays performed in the control, si-NC, and si-ATP6V1C1 groups. (E) The phosphorylation levels of mTOR, p70S6K, and 4E-BP in HCC cells with ATP6V1C1 knockdown detected by western blotting. (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001) ATP6V1C1 aggravates in vitro proliferation, migration, and invasion of HCC cells via the mTORC1 signaling pathway. (A–D) CCK-8 (A), colony formation (B), Transwell (C), and wound healing (D) assays performed in the control, si-NC, and si-ATP6V1C1 groups. (E) The phosphorylation levels of mTOR, p70S6K, and 4E-BP in HCC cells with ATP6V1C1 knockdown detected by western blotting. (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001) Discussion MiRNAs serve as guide molecules in RNA silencing, affecting tumor initiation, growth, and progression when altered [30, 31]. There is increasing evidence that miRNAs have become promising diagnostic and prognostic biomarkers for cancer [32–35]. Thus, we integrated miRNAs and mRNAs to identify novel potential biomarkers for HCC. In our work, through the intersection of tDEGs and tDEM-TGs followed by the selection using machine learning, ATP6V1C1 was finally uncovered. ATP6V1C1 had higher expression in HCC tissues than normal liver tissues both at mRNA and protein levels in multiple datasets. The ROC curves revealed the excellent performance of ATP6V1C1 in distinguishing between HCC and normal liver tissues. Meanwhile, the pan-cancer analysis displayed that the expression of ATP6V1C1 significantly varied in a group of cancers. Additionally, high ATP6V1C1 expression was found to represent poor OS, DFI, DSS, and PFI in HCC in several independent datasets. Multivariate Cox analysis further confirmed ATP6V1C1 was a independent risk factor for HCC. However, the prognostic role of ATP6V1C1 in pan-cancer showed not significant. Besides, ATP6V1C1 was associated with malignant clinicopathological features of HCC, which may be a contribution to the poor prognosis related to ATP6V1C1. Notably, ATP6V1C1 was highly expressed in malignant cells at single-cell transcriptome level and in the tumor region at spatial transcriptome level in HCC TME, indicating the potential association between ATP6V1C1 and malignancy of HCC. The perturbation effects of ATP6V1C1 in HCC cell lines further confirmed the contribution of ATP6V1C1 to the malignant phenotype of HCC cells. To better understand the malignant biological role of ATP6V1C1 in HCC, a GSEA was performed. The high ATP6V1C1 expression group had more enrichment of TNFα/NF-κB signaling pathway, TGF-β signaling pathway, mTORC1 signaling pathway, Wnt/β-catenin signaling pathway, and hypoxia signaling pathway, which were highly related to tumor initiation, progression, and treatment [36–41]. Similarly, the consistent results were gained by the progeny algorithm and the IMPACT database. In particular, we noticed mTORC1 signaling pathway as McConnell et al. found that ATP6V1C1 promoted breast cancer growth by activating the mTORC1 pathway. The activation of mTORC1 causes cap-dependent translation, leading to an increase in cell size and proliferation, which are two typical characteristics of cancer [42]. Additionally, aberrant activation of mTORC1 signaling pathway emerges frequently in HCC [43]. In our work, knockdown of ATP6V1C1 restrained HCC cell proliferation, migration, and invasion, as well as the activation of the mTORC1 signaling pathway. In addition, the association between ATP6V1C1 and DNA damage was explored. ATP6V1C1 expression had positive correlations with fraction altered, aneuploidy, and HRD. HCC patients with high ATP6V1C1 expression showed more CTNNB1 mutations and/or TP53 mutations. Chromosomal instability, found in over 90% of cancer, is linked to tumor cell malignancy and enhances adaptation capabilities of tumors [44]. CTNNB1 is the most frequently mutated proto-oncogene in liver neoplasms and encodes β-catenin, a major carcinogenic driver for the development of liver tumors [45]. HCC with TP53 mutation exhibits aggressive behavior, which might be due to the acquisition of stem cell-like gene expression traits [46]. Together, ATP6V1C1 facilitates high malignancy of HCC, thus leading to the inferior prognosis. TME status plays nonnegligible roles in tumor initiation, progression, metastasis, and response to therapies [47]. In our study, we found that ATP6V1C1 expression was positively correlated with the infiltration levels of pro-tumorigenic immune cells including Tregs and MDSCs. Oncogenic driver alterations found in many cancer types can cause the selective recruitment and activation of Tregs to evade anti-tumor immunity, facilitating tumor progression and resistance to immunotherapy [48]. A meta-analysis for HCC reveled that high levels of Tregs were associated with unfavourable OS and DFS [49]. MDSCs strongly suppress anti-tumor functions of T cells and natural killer cells, thus contributing to tumor immune escape. Additionally, MDSCs play important roles in promoting angiogenesis and tumor metastasis [50]. There is evidence from mouse models that MDSCs participate in the early phases of carcinogenesis [51–53]. Besides, the high ATP6V1C1 expression group showed more activation of TGF-β signaling pathway. With tumor progression, TGF-β can be leveraged to inhibit anti-tumor responses and thus promote immune evasion [54]. It was consistent with the result that in the seven-step cancer-immunity cycle, high ATP6V1C1 expression group had worse performance in step 6 and step 7, which represented lower anti-tumor activity. Moreover, we found that high ATP6V1C1 expression was associated with high tumor proliferation rate. Collectively, the prognostic value of ATP6V1C1 partially comes from the formation of TME status, which is characterized by increased pro-tumorigenic immune infiltration, inhibited anti-tumor immune activity, and high tumor proliferation rate. Anti-tumor therapies, including ICBs, targeted drugs, and TACE, have improved the prognosis of patients with advanced-stage HCC to a certain extent [55]. In our study, ATP6V1C1 exhibited predictive value on the response of HCC patients to anti-tumor therapies. HCC patients with low ATP6V1C1 expression had better response to ICBs, Sorafenib, and TACE. Hence, anti-tumor therapies are more suitable for HCC patients with low ATP6V1C1 expression. It may result from the high tumor malignancy and inhibited anti-tumor immune activity caused by ATP6V1C1. GDC-0879, with the best great disruptive effect on ATP6V1C1 expression, was further screened using CMap. GDC-0879 improves the survival rate of mice with the B-Raf mutant tumor MEXF989 [56]. As GDC-0879 is a B-Raf inhibitor, we hypothesized that ATP6V1C1 expression may be associated with B-Raf mutant in HCC. However, the association and mechanism involved need to be further explored. Some limitations remain in our study. The diagnostic and prognostic role of ATP6V1C1 and its potential to predict the response to anti-tumor therapies in HCC need to be further confirmed in prospective cohorts. Besides, the single-cell and spatial data used in our study come from public datasets with different sampling depths and spatial resolutions, which may affect ATP6V1C1 expression in different cell types and regions in HCC TME. In addition, more in vivo, in vitro, and clinical experiments are required to investigate the effects of ATP6V1C1 on HCC, as well as the underlying biological mechanism. Conclusion ATP6V1C1 is a potential diagnostic, prognostic, and therapeutic biomarker for HCC. Besides, ATP6V1C1 contributes to the oncogenesis and progression of HCC through its effects on the TME status and regulation of the mTORC1 signaling pathway. Supplementary Information Additional file1Additional file2Additional file3Additional file4Additional file5Additional file6Additional file7Additional file8Additional file9Additional file10Additional file11Additional file12Additional file13 Additional file1 Additional file2 Additional file3 Additional file4 Additional file5 Additional file6 Additional file7 Additional file8 Additional file9 Additional file10 Additional file11 Additional file12 Additional file13