From November 2020 to July 2021, 49 healthy controls (HCs), 20 benign liver disease patients (BLD) and 78 HCC patients were consecutively enrolled in this study. All serum samples were collected prior to surgery or radiation therapy and rapidly frozen at −80 °C until use, avoiding freezing and thawing cycles.

The diagnosis of HCC was verified by histological testing of the resected material from patients. Clinicopathological parameters of the subjects in this research, including age, gender, CNLC (China liver cancer staging) stage, distant tumor metastasis and tumor size were collected. The institutional ethics committee for human studies at the Second hospital affiliated with Chongqing Medical University granted this study. Informed consent was acquired from all volunteers following the related rules. Patient features were concluded in Table 1.

Serum was isolated from the specimens and stocked at −80 °C after centrifuging at 3,000 rpm for 10 min. ELISA Kit (Mlbio, Shanghai, China) was measured following the manufacturer's instructions. The automatic chemiluminescence immunoassay system LIAISON XL (Diasorin, Italy) was applied to detect AFP and Ferritin levels in serum samples.

Clinical data and expression profiles were downloaded from The Cancer Genome Atlas (TCGA) dataset (https://portal.gdc.cancer.gov/) (10) and the International Cancer Genome Consortium (ICGC) (https://dcc.icgc.org/releases/current/Projects). We performed consistency analysis following the expression of MRPs. Log-rank test was conducted to compare survival variance and time-dependent receiver operating characteristic (ROC) curves were performed to compare the predictive accuracy of MRPL9 for the prognosis of HCC patients. Nomogram and decision curve analyses (DCA) validated the prognostic effects. Univariate and multivariate cox regression analysis was applied to access the prognostic effect of risk scores. The half-maximal inhibitory concentrations (IC50) were downloaded from the Genomics of Drug Sensitivity in Cancer (GDSC, https://www.cancerrxgene.org/) (11). The immunophenoscore (IPS) and the tumor immune dysfunction and exclusion (TIDE) algorithm were applied to assess the immunotherapeutic response (12). OCLR algorithm predicted the stemness (13). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were analyzed by the differentially expressed genes (DEGs) in HCC patients. Threshold: fold-change (FC) = 3, p < 0.05. Gene Set Enrichment Analysis (GSEA) was conducted to investigate the potential mechanism and biological functions between two clusters. R packages were performed using R software version v4.0.3, including “rms”, “ggrisk”, “survival”, “survminer”, “timeROC”, “pRRophetic”, “clusterprofiler”, “ggplot2”, “immunedeconv”, “ConsensusClusterPlus” and “ggDCA”.

The liver tumor line cells Huh7 and HepG2 were grown in Dulbecco's modified Eagle's medium (Hyclone, Uinted States) containing 10% FBS (Lonsera, Uruguay) supplemented with 1% penicillin/-streptomycin at 37 °C in a humidified incubator containing 5% CO2. The overexpressed plasmid (EX-V0670-M02-MRPL9) and control plasmid (EX-V0670-M02-Control) were purchased from Labcell biotechnology (Chongqing, China). And they were transfected into Huh7 and HepG2 cells using H4000 (Engreen, China) based on the standard instruction.

Cell proliferation was detected by the cell counting kit-8 (Dojindo Laboratories, Kumamoto, Japan) according to the manufacturer's instructions. In brief, 1,000 cells transfected with MRPL9 overexpressed plasmid and control plasmid were plated into 96-well plates and cultured until the indicated time. Then 10 µl of CCK-8 agent was added to every well and incubated for 1.5 h. The absorbance at 450 nm was detected to calculate the number of viable cells.

Cell cycles were measured by CytoFLEX flow cytometer. Cells were harvested after different treatments and then washed twice with ice-cold PBS, dyed with propidium iodide (PI) followed by flow cytometry analysis, as previously described (14).

The MRPL9-overexpressed cells and control cells were resuspended in an FBS-free medium and seeded in the top chamber of transwell plates at a density of 5 × 10⁴ cells/well, 500 µl of DMEM containing 10%FBS was added in the bottom chamber. After incubating for 24 h at 37 °C with 5% CO₂, the cells were fixed with 4% formaldehyde and stained with crystal violet. As for the invasion assay, the chamber was coated with Matrigel (40 µl/well, 1 : 5 dilution). The cells were counted by 3 random fields of every well.

Western blotting was performed as described previously (14). The following primary antibodies were used: MRPL9 (15342-1-AP, 1:1000, Proteintech, China), β-Tubulin (10068-1-AP, 1 : 1,000, Proteintech, China), E-cadherin (24E10, 1:800, CST, US), β-catenin (sc-7199, 1:800, Sants Cruz Biotechnology, US), N-cadherin (22018-1-AP, 1:1000, Proteintech, China). The protein signals were visualized using ECL ((Millipore, Massachusetts, Uinted States) and the signal strength was analyzed using Image Lab V6.0 software.

All statistical analyses were conducted using GraphPad Prism software7.0 (GraphPad Software, CA, Uinted States) and the SPSS program (version 23.0). Median values with interquartile ranges were used to describe serum protein concentrations. The Mann–Whitney test was used to compare two groups and the Kruskal–Wallis test to compare serum markers in more than two groups. The chi-square test examined categorical data. ROC analysis was performed to evaluate the diagnostic role. p-value < 0.05 was considered as statistical significance. p <0.05 (*); p < 0.01(**); p < 0.001(***).

The expression of 79 MRPs in HCC was detected according to the TCGA and GTEx profiles. The overall survival (OS) prognosis roles of MRPs in HCC patients were presented in Supplementary Table S1 and high levels of 37 MRPs indicated poor outcomes. Our results signified that 78 MRPs were remarkably augmented in tumor tissues than in normal individuals (Figure 1A). Functional enrichment profiling suggested that MRPs mainly got involved in the cellular adheren junction and innate immune response (Figure 1B). Protein-protein interaction (PPI) plot revealed the interactivity of MRPs (Figure 1C).

371 HCC patients were enrolled and we performed a consensus clustering analysis to sort them into two clusters based on the expression of MRPs (Figures 2A,B, Supplementary Table S2). Principal component analysis (PCA) demonstrated the superior intergroup distribution (Figure 2C). Heatmap showed that the genomic levels of MRPs were higher in cluster 1 than in cluster 2 (Figure 2D). The bar chart displayed that the T stage and grade were significantly different between the two subtypes (Figure 2E). Additionally, as determined in Figure 2F, the KM curve indicated that patients in cluster 1 had a poorer prognosis than those in cluster 2.

Subsequently, we investigated the underlying biological activity and pathways in two subtypes. GSEA of biological processes implied that cluster 1 was majorly enriched in peptide and amide biosynthetic processes, and the regulation of gene expression (Figure 3A). Catabolic processes were involved in cluster 2 (Figure 3B). In the following pathways analysis, we derived DNA repair, E2F targets and MYC targets (Figure 3C). As for cluster 2, we mainly observed metabolism-associated pathways (Figure 3D). Moreover, we performed KEGG and GO analyses on the differentially expressed genes in two molecular subtypes and found cell cycle was the key term in upregulated genes (Supplementary Figure S1A). Downregulated genes mainly participated in the metabolism-related signal transduction (Supplementary Figure S1B). GO analysis of upregulated genes was primarily enriched in organelle fission and nuclear division (Supplementary Figure S1C), and downregulated molecules were concerned with multiple metabolism processes (Supplementary Figure S1D).

Resistance to chemotherapy is a major obstacle to effective cancer treatment (15). Considering the diverse clinical features and prognosis, we further calculated the IC50 values of 16 common anticancer drug agents in two MRPs clusters by applying the pRRophetic algorithm. Among the 16 drugs, 13 of them might be more sensitive for patients in cluster 1, including Temozolomide, Shikonin, Tubastatin A, Cyclopamine, Bosutinib, Pazopanib, Phenformin, Belinostat, Sorafenib, Docetaxel, Methotrexate, 5-Fluorouracil, Vorinostat (Figures 4A–M). As regards to the patients in cluster 2, Cetuximab, Nutlin-3a (−) and Trametinib might be more efficient for the treatment (Figures 4N–P).

Then, we estimated the characteristics of the tumor microenvironment between the two clusters. As shown in Figure 5A, the majority of immune checkpoints were notably elevated in cluster 1. TIMER and MCPCOUNTER algorithms uncovered that the abundance of B cells, CD8+ T cells, Macrophage and Myeloid dendritic cells were enriched in cluster 1 (Figures 5B,C). Simultaneously, the stemness scores were significantly increased in cluster 1 (Figure 5D). Besides, immunophenoscore (IPS) confirmed that patients in cluster2 might benefit more from anti-PD1 and anti-CTLA4 therapy (Figures 5E,F). The TIDE algorithm validated that patients in subtype2 were conferred higher microsatellite instability and exclusion scores (Figures 5G,H), while having lower TIDE scores (Figure 5I). Our findings suggested that cluster 2 may have a higher likelihood of a response to ICB response, thereby achieving a better clinical outcome.

Afterward, we implemented LASSO regression analysis to develop a prognostic signature following DEGs (Figures 6A,B, Supplementary Table S3). The model apparently differentiated the status of HCC patients in TCGA and we validated it in ICGC databases (Figure 6C). KM plots revealed that patients with high risk had an adverse OS than the low-risk group (Figure 6D). ROC plots exerted satisfactory effects in predicting prognosis for short and long terms (Figure 6E). Univariate and multivariate Cox analysis demonstrated our risk score presented as an independent predictive factor (Supplementary Table S4). Additionally, we incorporated clinical parameters to construct a nomogram for the prediction of 1-, 3- and 5-year prognosis (Figure 6F). The calibration curve approximated the desired curve (Figure 7G) and DCA demonstrated a greater advantage in the prediction of the prognosis in 1-, 3- and 5-year than other clinical factors (Figure 6H).

Given the aberrant expression and prognostic roles of MRPL9, we further validated the promising clinical values and the carcinogenic effects in HCC. The expression of MRPL9 in multiple tumors based on TCGA and GTEx datasets were explored and the results indicated that MRPL9 was highly expressed in adrenocortical carcinoma (ACC), bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), cholangio carcinoma (CHOL), colon adenocarcinoma (COAD), esophageal carcinoma (ESCA), lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), esophageal carcinoma (ESCA), glioblastoma multiforme (GBM), head and neck squamous cell carcinoma (HNSC), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), brain lower grade glioma (LGG), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), ovarian serous cystadenocarcinoma (OV), prostate adenocarcinoma (PRAD), pancreatic adenocarcinoma (PAAD), pheochromocytoma and paraganglioma (PCPG), skin cutaneous melanoma (SKCM), rectum adenocarcinoma (READ), testicular germ cell tumors (TGCT), stomach adenocarcinoma (STAD), thyroid carcinoma (THCA), thymoma (THYM), uterine corpus endometrial carcinoma (UCEC) and uterine carcinosarcoma (UCS); while was decreased in kidney chromophobe (KICH) and acute myeloid leukemia (LAML) compared with normal tissues (Figure 7A). Similar results were observed when compared with corresponding adjacent normal tissues (Figure 7B). Besides, we found that MRPL9 could exhibit excellent diagnostic function in diverse tumors (Figure 7C).

Earlier analysis implied the potential of MRPL9 in diagnosis for HCC patients, we further validated the serum diagnostic values. AFP and Ferritin were widely used serum biomarkers to detect HCC worldwide. We analyzed the serum MRPL9, AFP and Ferritin levels in HCC, benign liver disease and healthy groups. As shown in Figures 8A–C, serum MRPL9, AFP and Ferritin levels were significantly higher in HCC patients than in benign liver disease patients and healthy individuals (p < 0.01). ROC analysis was conducted to test the diagnostic accuracy of serum MRPL9. The area under curve (AUC) value was calculated to evaluate the diagnostic efficacy (16). The curve analysis indicated that MRPL9 had a high AUC value (0.867) to distinguish HCC patients from non-cancerous individuals (Figure 8D). When the cut-off value was 1581.3 pg/ml, the sensitivity and specificity were 76.9% and 91.3%, respectively. In contrast, both AFP and Ferritin exhibited low AUC values (0.705 and 0.74, respectively) and low sensitivity (53.8% and 43.6%, respectively), which suggested that MRPL9 was superior to AFP and Ferritin in distinguishing HCC from healthy and benign groups. Most importantly, the combination of these three markers resulted in a considerable increase in AUC value (0.948), with high sensitivity and specificity of 85.9% and 92.8%, respectively. Similarly, MRPL9 also had the best diagnostic potency in distinguishing HCC from the healthy (AUC = 0.89) and benign liver disease groups (AUC = 0.812) than AFP (AUC = 0.695, 0.729, respectively) and Ferritin (AUC = 0.734, 0.757 respectively; Supplementary Tables S5–S7).

Given the increased level of serum MRPL9 in HCC patients, we further determined the correlation between MRPL9 and different clinicopathological factors, including gender, age, tumor size, CNLC stage and HCC diagnostic markers including AFP, Ferritin. The serum MRPL9 was associated with tumor size and distant tumor metastasis. And there was no statistical difference in other clinical parameters. These results were shown in Table 2.

Next, the MRPL9 overexpressed and control plasmids were transfected into Huh7 and HepG2 cells, and the higher expression level of MRPL9 was identified by western blot (Figure 9A). Then, the CCK8 assay showed an enhanced proliferation rate in the overexpressed group compared to the control group (p < 0.01, Figure 9B). Upregulated MRPL9 promoted the transition of the G1/S phase (Figure 9C). Transwell assay showed that overexpressed MRPL9 promoted the ability of migration and invasion in Huh7 and HepG2 cells (p < 0.05, Figure 9D). Furthermore, western blot demonstrated overexpressed MRPL9 could promote the epithelial-mesenchymal transition (EMT) process in Huh7 and HepG2 cells (Figures 9E,F).