DOI: https://doi.org/10.21203/rs.3.rs-1016997/v1
Background: Hepatocellular carcinoma (HCC) is a common malignancy with a poor prognosis worldwide. However, the pathogenesis of HCC remains poorly understood.
Methods: Through data mining and analyses of The Cancer Genome Atlas (TCGA) datasets, the NOL12 expression in HCC was determined and the associations between its expression and patient survival and clinicopathological parameters were evaluated. The pro-tumorigenic roles of NOL12 on HCC in vitro were further verified by loss-of-function assay. The correlation between NOL12 expression and tumor-infiltrating immune cells (TICs) was analyzed by CIBERSORTx method. In addition, the risk signature based on 8 NOL12-related genes was established to accurately evaluate the prognosis of patients with HCC and to further predict the efficacy of immune checkpoint inhibitors (ICIs) in HCC
Results: We found that NOL12 was significantly overexpressed in independent HCC datasets from TCGA database. High expression of NOL12 is associated with worse reduced overall survival (OS), high pathological grade, node metastasis and advanced clinical stage in patients with HCC. Moreover, NOL12 knockdown significantly inhibited cell proliferation, migration and invasion. CIBERSORTx analysis revealed that twelve types of TICs are correlated with NOL12 expression. The risk signature based on 8 NOL12-related genes is an independent prognostic factor for patients with HCC. The OS rate of patients in the low-risk score group was better than that in the high-risk score group. In addition, the total tumor mutation burden (TMB) in the high-risk score group increased significantly, and the risk scores could be used as an alternative indicator of ICI response.
Conclusions: Our findings indicated that NOL12 might be involved in the progression of HCC and can be used as a potential therapeutic target. Moreover, the NOL12-related risk signature may have predictive relevance with regard to ICI therapy.
Liver cancer has become the fourth largest cause of cancer-related death in the world, with the sixth highest incidence of all cancers. It is estimated that approximately 1 million people will die of liver cancer each year [1, 2]. Hepatocellular carcinoma (HCC), as the most common primary malignant tumor of the liver, progresses rapidly, and patients are usually diagnosed at an advanced stage. At present, the treatment of HCC is limited and mainly depends on hepatectomy, chemotherapy and immunotherapy [3]. However, due to early metastasis, postoperative recurrence, and the emergence of drug resistance and other factors, the 5-year survival rate of patients has not been significantly improved [4, 5]. Moreover, there is still a lack of effective methods to predict the prognosis of patients and provide individualized treatment [6]. Therefore, it is urgent to find more reliable diagnostic biomarkers and develop new indicators to predict patient survival to provide individualized treatment strategies.
The transcription and regulation of the human genome is a sophisticated process, in which RNA binding proteins participate and play an important role [7, 8]. In recent years, an increasing number of studies have found that RNA binding proteins are also involved in tumor progression and immune regulation [9–12]. Nucleolar protein 12 (NOL12), a multifunctional RNA binding protein, has been found to be related to genomic stability, DNA damage repair and apoptosis [13]. Pinho et al. [14] reported that NOL12 can affect the cell cycle and eventually lead to cell senescence in an RPL11-dependent manner. In addition, some studies have shown that NOL12 may be a potential oncogene for a variety of tumors and may be related to the prognosis of patients [15, 16].
Tumor microenvironment (TME) is an important factor affecting the progression and treatment of HCC [17]. Some studies have shown that, when all kinds of immune cells in the TME reach a certain balance, they will spur immune activation, thus promoting the occurrence of immune escape [18]. Increasing evidence has shown that tumor-infiltrating immune cells (TICs) affect the biological behavior of HCC cells and ultimately affect the prognosis of patients [19, 20]. However, the mechanism of NOL12 in the progression and immune infiltration of HCC has not been reported.
In this study, we first confirmed that the expression of NOL12 is upregulated in HCC through TCGA database, and it is associated with poor prognosis. Second, through CIBERSORTx and functional enrichment analysis, we found that NOL12 is related to a variety of TICs and tumor-related signaling pathways. In addition, we developed a risk signature based on NOL12-related genes and established a nomogram that can independently predict the clinical outcome of HCC. Patients with different risk scores have been shown to have different therapeutic potential for immune checkpoint inhibitors (ICIs). Finally, we determined the tumorigenic roles of NOL12 by a loss-of-function assay in vitro. Overall, our results indicated that NOL12 can act as a novel prognostic biomarker and a potential therapeutic target for HCC.
The RNA transcriptome dataset, the corresponding clinical data and the mutation profiling from TCGA (https://tcga-data. nci.nih.gov/tcga/) database for the liver hepatocellular carcinoma (LIHC) project were downloaded. The limma package for R software was employed to further process RNA expression data. When combining clinical information, missing and incomplete samples were deleted. The remaining TCGA dataset contains 370 tumor samples and 50 normal samples for further analysis.
The abundant distribution of TICs in all tumor samples was estimated using the CIBERSORTx [21] computational method, and the corresponding R package was used to visualize them.
We used Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis with Gene Set Enrichment Analysis (GSEA) software to explore the potential biological function of NOL12 in HCC. Pathways with p < 0.05 were considered significantly enriched. The visualization was performed by the ggplot2 package. In addition, a protein-protein interaction (PPI) network highly related to the NOL12 encoded protein was obtained using the Search Tool for the Retrieval of Interacting Genes (STRING; http://string- db. org), with the highest confidence (0.9).
We screened prognosis-related coexpressed genes by univariate Cox regression analysis. To avoid overfitting, all the selected genes were involved in the subsequent least absolute shrinkage and selection operator (LASSO) penalized Cox regression analysis. The risk score for each patient with HCC was calculated based on the following formula: Riskscore =∑(Coefi * Expxi), where Expxi represents each gene expression and Coefi represents the coefficient of each gene.
All TCGA LIHC patients were randomly divided into training sets (186) and testing sets (184). The patients in the training and testing sets were further divided into high- and low-risk groups, with the median risk score as the cutoff value. To evaluate the accuracy of the risk signature, Kaplan-Meier survival analysis and time-dependent receiver operating characteristic (ROC) curves were performed on the training sets and testing sets. The complete dataset with corresponding clinical information was used for subsequent stratified survival analysis. To identify the independence of the risk signature, univariate Cox regression analysis between the risk signature and the clinicopathological features was performed, and a P < 0.05 was considered to be a significant independent prognostic factor.
The nomogram was constructed based on the risk score and independent clinical factors by using the “regplot” R package to quantify risk evaluation and predict the clinical outcomes of patients with HCC. The calibration plot was applied to assess the accuracy of the nomogram for predicting the 1-, 3-, and 5-year overall survival of patients with HCC. Decision curve analysis (DCA) was used to assess the clinical practicability of the nomogram.
The somatic variant data of patients with HCC were analyzed and visualized by package maftools. Then, difference, correlation and survival analyses were used to evaluate the effects of TMB and risk scores on the prognosis of patients with HCC.
In addition, we also compared the expression levels of common immune checkpoints in the high-risk score and low-risk score groups and the correlation between immune checkpoint expression level and risk score.
Human liver cancer cells (HepG2 and Huh-7) were obtained from Shanghai Cell Bank (Shanghai, China) or ATCC (the American Type Culture Collection). All the cell lines used in this study were approved by the Ethics Committee of The First Affiliated Hospital of Nanchang University. All cell lines were cultured in 90% DMEM and 10% FBS at 37 °C in a 5% CO2 incubator humidified atmosphere as recommended.
Lentiviral vectors encoding short hairpin RNAs (shRNAs) that target NOL12 were purchased from GenePharma (Shanghai, China). NOL12 shRNAs were transfected into HepG2 and Huh-7 cells according to the manufacturer’s instructions. To generate the stable cell line, the transduced cells were then selected in culture medium containing puromycin (5.0 µg/mL).
The MTT assay was used to evaluate cell viability. The steps were as follows: 1000 cells were placed in a 96-well plate. From day 1 to day 5, 20 μL of MTT solution (5 mg/mL) was added to each well, and the wavelength of 570 nm was measured with a 96-well plate reader. For the cell colony formation assay, 1000 HepG2 and Huh-7 cells/well from the control and experimental groups were inoculated into 6-well plates, and colony formation was detected 10 days later.
In the migration assay, 5-10 × 104 HepG2 and Huh-7 cells were added to the top 8 µm chamber without Matrigel, while in invasion assays, Matrigel was added. After 24-48 h of incubation, the inserts were rinsed with PBS and stained with 0.1% crystal violet solution for 5 min. Images were taken with an Olympus IX-70 microscope.
Data between two groups were examined using a two-tailed paired Student’s t-test or ANOVA (Bonferroni post hoc test). Correlation analysis was performed using the Pearson chi-square test or Spearman rank correlation test. All statistical analyses were performed by GraphPad Prism 7 or R (4.0.0); p values < 0.05 denoted statistically significant differences.
First, we downloaded the transcriptome data of LIHC and the corresponding clinical data from TCGA database, including 50 normal and 370 tumor samples. To evaluate the significance of NOL12 in HCC, these datasets were used to study the expression of NOL12 in HCC. The results showed that the mRNA expression of NOL12 in HCC tissues increased significantly (Figure 1A). This result was also fully verified in the paired analysis of HCC tissue and adjacent normal tissue in the same patient (Figure 1B). In addition, we found that the overall survival rate of the NOL12 high expression group was lower than that of the low expression group, which was statistically significant (Figure 1C). We further examined the correlation of NOL12 with the clinicopathologic features of HCC. In a subgroup analysis based on race, sex, age, grade, metastasis status and stage in the UALCAN database, the transcription level of NOL12 in patients with LIHC was significantly higher than that in normal controls, and the higher the expression of NOL12 was, the later the tumor grading and staging. (Supplementary Figure 1). The above results show that NOL12 may be a potential diagnostic marker for HCC.
We next systematically described the pattern of tumor-infiltrating immune cells (TICs) by processing the signature gene expression profile in HCC with the CIBERSORTx algorithm. The results of Supplementary Figure 2 show the distribution landscape and correlation of infiltrating immune cells in LIHC specimens. According to the results from the difference and correlation analyses, we concluded that NOL12 might regulate the immune activity of the TME in HCC mainly through twelve kinds of TICs (Figure 2). Among them, seven kinds of TICs were negatively correlated with NOL12 expression, including naïve B cells, resting CD4+ T cell memory, activated NK cells, monocytes, M2 macrophages, resting mast cells and activated mast cells. Five kinds of TICs were positively correlated with NOL12 expression, including memory B cells, M0 macrophages, activated CD4+ T cell memory, follicular helper T cells and regulatory T cells. In addition, the results from GO enrichment analysis indicated that the functions of NOL12 correspond to immune-related activities (Figure 3A). These results suggested that the level of NOL12 might be a key indicator to reflect the state of the TME. Figure 3B shows that a variety of tumor-related signaling pathways were enriched, including the MAPK, PI3K-Akt cAMP and Ras signaling pathways.
We constructed a PPI network of NOL12 using the Search Tool for the Retrieval of Interacting Genes (STRING) database, with the highest confidence (0.9) (Figure 3C).
We obtained the 20 genes most closely related to NOL12 and then performed univariate Cox regression analysis to further evaluate the relationship between these coexpressed genes and the survival of patients with HCC. The results showed that 16 NOL12 coexpressed genes were closely related to the prognosis of HCC (Figure 3D). All patient samples were randomly assigned to a training set (n = 186) or a testing set (n = 184). Least absolute shrinkage and selection operator (LASSO) proportional Cox regression analysis was performed to construct an eight-gene risk signature model (Figure 3E, F). The risk score of each patient in the training and testing sets was calculated as follows: risk score = (expWDR43 * 0.00322) + (expNOP58 * 0.08182) + (expRCL1 * -0.13789) + (expUTP3 * 0.01855) + (expEBNA1BP2 * 0.16428) + (expNOP56 * 0.19824) + (expBMS1 * 0.08048) + (expRBM28 * 0.55925).
The patients in the training and testing sets were further divided into high- and low-risk groups, with the median risk score as the cutoff value. The results showed that in both the training and testing sets, the number of deaths in the high-risk group was significantly higher than that in the low-risk group (Figure 4A, B). The results of K-M survival analysis also confirmed that the prognosis of the low-risk group was better than that of the high-risk group (Figure 4C, D). Time-dependent receiver operating characteristic (ROC) curves were generated to estimate the efficacy of the risk signature for predicting the survival of patients with HCC. The areas under the curve (AUCs) of the 1-, 3- and 5-year risk scores were 0.762, 0.705 and 0.728 in the training sets, respectively. In the testing sets, the AUC values were 0.713, 0.662, and 0.609 at the 1-, 3- and 5-year time points, respectively (Figure 4E, F). In addition, to further evaluate the value of the risk signature in clinical application, we conducted a stratified survival analysis of patients with different clinicopathological features, including age, sex, grade, T stage, N stage, M stage, and clinical stage. For different stratifications, patients in the low-risk group had a higher overall survival rate than those in the high-risk group (Supplementary Figure 3). These results indicated that the NOL12-related risk signature can accurately predict the prognosis of patients with HCC without considering clinical factors. Univariate Cox regression analyses also showed that the risk score was an independent prognostic factor for patients with HCC (Figure 4G, H).
To provide a quantitative method with clinical value to predict the probability of 1-, 3- and 5-year OS in HCC, we constructed a nomogram based on the independent prognostic factors (Figure 5A). The total scores were calculated according to the points of all variables in the nomogram, and then the 1-, 3- and 5-year survival rates of each patient with HCC could be predicted by drawing a vertical line from the total points to the survival prediction axis. The prediction values of the 1-, 3- and 5-year nomograms in the calibration plot were close to the 45-degree line in the complete dataset, which indicates that the nomogram demonstrates good prediction ability (Figure 5B). The ROC curves showed that, compared with other clinical factors related to prognosis, the nomogram had the highest accuracy in predicting HCC survival, with an AUC of 0.804 (Figure 5C). In addition, decision curve analysis (DCA) was performed to evaluate the net clinical benefits of multiple prognostic factors (Figure 5D). The results showed that using a nomogram to predict survival probability can bring more benefits to patients.
We analyzed the mutation profiles of each LIHC patient and associated the patient's risk score with TMB for difference and correlation analyses. The results showed that the level of TMB in the high-risk group was higher than that in the low-risk group and was positively correlated with the risk score (Figure 6A, B). Subsequently, we performed a K-M survival analysis to evaluate the effects of high and low TMB and risk scores on the prognosis of patients with HCC. As shown in Figures 6C and D, the survival rate of patients with h-TMB was significantly lower than that of patients with l-TMB. Among all the groups, the survival rate of patients with h-TMB + h-risk was the worst, while that of patients with l-TMB + l-risk was the best. Since patients with elevated TMB tend to benefit more from ICIS treatment, we further investigated the relationship between ICIS and risk scores. The results showed that the expression levels of PD-1, PD-L1, CTLA-4 and LAG3 in the high-risk group were significantly higher than those in the low-risk group and were positively correlated with the risk score, indicating that the high-risk group could benefit more from ICI treatment (Figure 6E-L).
Considering that the previous data results supported the potential tumor-promoting effect of NOL12 in HCC, a series of functional experiments were carried out in vitro. We selected HepG2 and Huh-7 cells with relatively higher endogenous NOL12 in HCC cells for knockdown experiments. Two shRNAs targeting human NOL12 (shNOL12#1, shNOL12#2) were introduced into HepG2 and Huh-7 cells to detect changes in endogenous NOL12 protein. As shown in Figure 7A, we confirmed NOL12 silencing by comparison with the corresponding negative control (shCtrol). The results of the cell proliferation assay showed that the number of cells in the NOL12-knockdown group was significantly lower than that in the control group (Figure 7B). The results of the MTT assay also showed that cell viability was significantly inhibited after NOL12 knockdown in vitro (Figure 7C), and the colony numbers of HepG2 and Huh-7 cells decreased significantly (Figure 7D, E). To further determine whether NOL12 participates in the regulation of migration and invasion processes in HCC cells, Transwell experiments were executed to examine the effect of NOL12 on cell migration and invasion. The results of Transwell assays showed that knockdown of NOL12 in HepG2 and Huh-7 cells inhibited their migration and invasion ability (Figure 8).
In recent years, although some new drugs have been approved for the treatment of HCC, the therapeutic effect of patients with advanced HCC is not satisfactory, and the 5-year survival rate has not been significantly improved [22]. How to realize the early diagnosis of HCC is still a great challenge for clinicians, so it is necessary to find new biomarkers to provide new targets for the early diagnosis and treatment of HCC [23, 24]. Previous studies have shown that RNA-binding proteins play an important role in tumorigenesis and development. Iino et al. [9] reported that the RNA binding protein NONO can promote the proliferation of breast cancer by regulating cell proliferation-related genes and may become a potential therapeutic target for breast cancer. Li et al. [25] constructed a prognostic model that can accurately predict the prognosis of patients with lung cancer by using a variety of RNA-binding proteins. In our study, we identified a new RNA binding protein, NOL12, that may be an oncogene of HCC. Through bioinformatics analysis, we found that the expression of NOL12 in HCC was significantly increased, and high expression was associated with a low overall survival rate. In addition, the high expression of NOL12 is closely related to high pathological grade, node metastasis and advanced clinical stage. To verify the tumor-promoting effect of NOL12 in HCC, we carried out a series of functional experiments in vitro. The results showed that NOL12 gene knockout could significantly inhibit the proliferation, migration and invasion of HCC cells. In addition, KEGG analysis also showed that NOL12 was associated with a variety of tumor-related signaling pathways, including the MAPK, PI3K-Akt, cAMP and RAS signaling pathways. These results strongly suggest that NOL12 can be used as a new diagnostic and prognostic marker and promote the development of HCC.
There is growing evidence that oncogenes or tumor suppressor genes can recruit different immune cells in the TME to promote or inhibit tumor progression [26, 27]. RNA-binding proteins have been shown to affect the composition of the TME through immune activation and immune regulation [10, 12]. Zhao et al. [28] reported that the RNA binding protein SORBS2 can inhibit the TME and ultimately inhibit the metastasis of ovarian cancer by affecting the polarization of M2-like macrophages. In our study, through GO analysis, we found that NOL12 may be related to immune activation. Further research showed that NOL12 can regulate the immune activity of the TME in HCC mainly through twelve kinds of TICs. Therefore, we speculate that NOL12 can promote the immune infiltration of HCC to promote its proliferation and metastasis, but this finding needs further study to be confirmed.
Currently, access to public high-throughput gene expression datasets has contributed to the discovery of potentially reliable biomarkers of HCC [29–31]. Many research groups have found that signals based on specific gene expression can accurately predict the prognosis of patients with HCC. Dai et al. predicted the survival time of patients with HCC and the efficacy of immunotherapy by constructing an immune-related gene-based prognostic index [32]. Tang et al. constructed a prognostic signature based on four ferroptosis-related genes. This prognostic signature shows superior diagnostic and predictive performance and provides a new possibility for individualized treatment of patients with HCC [33]. Since our previous studies have shown that NOL12 can be used as a reliable biomarker for the diagnosis and prognosis of HCC, we wondered whether the signature based on NOL12-related genes can accurately predict the prognosis of patients. Here, we constructed a NOL12-related gene risk signature by LASSO regression analysis and verified it using a training set and test set. The ROC curve results showed that the risk signature has high accuracy in predicting 1-, 3- and 5-year survival rates. Independent prognostic analysis showed that the risk signature can be used as an independent prognostic determinant of patients with HCC. In addition, the results of a stratified survival analysis showed that the risk signature could accurately predict the prognosis of patients with HCC without considering clinical factors. Because nomograms can directly show prognosis, they are widely used to predict the survival and prognosis of patients with tumors [34]. Subsequently, a nomogram was constructed based on the NOL12-related gene risk signature and other independent clinical features to predict the 1-, 3- and 5-year overall survival rates of individual patients with HCC. The results of the calibration curve, ROC curve and decision curve showed that the nomogram we constructed can provide clinicians with a more accurate, convenient and practical prediction tool.
Numerous studies have shown that immune checkpoint is an important factor affecting the prognosis and treatment of advanced HCC [35]. PD-L1 expression, TIC density, TMB, and mismatch repair deficiency have been associated with the effect of ICI treatment and used to screen patients before ICI treatment [36]. In the present study, we found that the predictive power of risk signatures was related to TMB. In addition, TMB was significantly higher in the high-risk score group and predicted a poor prognosis. Therefore, we speculated that risk scores, such as TMB, may help to improve the efficacy of ICIs in patients with high risk scores. The results showed that the expression level of immune checkpoint molecules increased in patients with high risk scores, indicating that patients with high risk scores may be more sensitive to ICI treatment and benefit more. To date, this study demonstrates for the first time the important role of NOL12 in the prognosis of HCC. However, more clinical samples are needed to validate these results. In addition, more work on finding out the exact NOL12 regulatory network need be down in the future as well as novel inhibitors targeting NOL12.
In conclusion, our findings revealed the expression pattern, prognostic and tumorigenic roles of NOL12 and identified NOL12 as a novel oncogene and potential therapeutic target in HCC that may be achieved by affecting the TME. In addition, the risk signature based on NOL12-related genes can guide clinicians to judge the prognosis of patients with HCC and evaluate the effect of ICI treatment.
HCC: Hepatocellular carcinoma; NOL12: Nucleolar protein 12; LIHC: Liver hepatocellular carcinoma; TICs: Tumor-infiltrating immune cells; ICIs: Immune checkpoint inhibitors; TME: Tumor microenvironment; TMB: Tumor mutation burden; LASSO: Least absolute shrinkage and selection operator; DCA: Decision curve analysis; ROC: Receiver operating characteristic; TCGA: The Cancer Genome Atlas; GSEA: Gene set enrichment analysis; GO: Gene ontology; KEGG: Kyoto encyclopedia of genes and genomes; OS: Over survival; PPI: Protein-protein interaction.
We thank all the contributors for this work
CCY, JZG and YCJ designed the research study; HJF wrote and revised the manuscript; KWB analyzed the data; PSB provided professional advice about the research. All authors read and approved the final manuscript.
This work was supported by the National Natural Science Foundation of China (No.81960503), Postgraduate Innovation Special Foundation of Jiangxi Province (YC2020-B054).
The datasets used during the current study are available from the corresponding author on reasonable request.
Not applicable
Not applicable
The authors declare that they have no competing interests.