3.1 NOL7 expression levels

TCGA pan-cancer analysis showed high expression of NOL7 in several cancers, including bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), lung adenocarcinoma (LUAD), HCC, and colon adenocarcinoma (COAD), compared to normal tissues (Fig. 1A, all P < 0.05). To investigate NOL7 expression in HCC, we investigated TCGA database and found that NOL7 mRNA levels in HCC tissues increased significantly compared to non-carcinoma tissues. (Fig. 1B, P < 0.001) and matched normal tissues (Fig. 1C, P < 0.001). There was a statistically significant increase in NOL7 mRNA expression levels in HCC tissues, as compared to healthy adjacent tissues from the TCGA and GTEx databases (Fig. 1D, P < 0.001). When twelve HCC research cohorts in HCCDB database were examined, a similar trend in mRNA levels was observed (Fig. 1E, all P < 0.05). In our validation set (n = 20), elevated NOL7 mRNA levels were observed (Fig. 2A, P = 0.021; Fig. 2B, P = 0.007). Notably, the receiver operating characteristic (ROC) curve indicated that NOL7 expression exhibited high predictive capacity, boasting an area under the curve of 0.969 (95% confidence interval [CI] = 0.951–0.988). This suggests that HCC tissues are effectively differentiable from normal tissues (Fig. 1F).

Fig. 1

NOL7 mRNA and protein expression in HCC. A NOL7 expression levels in tumor tissues and their matched normal tissues in various cancers in the TCGA cohort. B The NOL7 mRNA expression level in HCC patients was elevated in TCGA databases. C NOL7 expression in tumor tissues and their matched normal tissues from HCC patients in TCGA databases. D Significant differences in NOL7 mRNA levels in HCC and the adjacent normal tissues in TCGA and GTEx databases. E Figure and plot of NOL7 expression in HCC and their matched normal tissues based on the HCCDB database. F ROC curves for classifying HCC versus normal liver tissue in TCGA databases. G NOL7 protein levels in HCC patents from UALCAN database (P < 0.001). *P < 0.05, **P < 0.01, ***P < 0.001

Fig. 2 

NOL7 expression from our validation set. A qRT-PCR, using t-tests of unpaired samples. B qRT-PCR, using t-test paired samples. C HE images of HCC tumor tissue and adjacent noncancerous tissue (left). D IHC images and expression of NOL7 protein in HCC tumor tissues and adjacent noncancerous tissues (right), Magnification =  × 400, The positive stain appears brown. E Using t-tests of unpaired samples. F Using a paired t-test sample. *P < 0.05, **P < 0.01, ***P < 0.001

To verify the expression of NOL7 in HCC, the NOL7 protein expression in HCC tissues was evaluated. This evaluation revealed that NOL7 was highly expressed in HCC tissues but displayed very low in healthy adjacent samples, as reported in the UALCAN database (Fig. 1G, P < 0.001). In our study, we utilized IHC staining techniques on 20 tumor samples along with their corresponding healthy tissue margins to evaluate the differential expression levels of NOL7. H&E staining of tumor tissue revealed a beam-like and pseudoglandular arrangement, with enlarged nuclei, coarse chromatin, irregular nuclear shape, partially visible eosinophilic nucleoli, and visible nuclear division (Fig. 2C). The highest expression of the NOL7 protein was detected in the nuclei of HCC tissues (Fig. 2D). We also observed higher expression of NOL7 protein in HCC tissues compared to matched noncancerous tissues by IHC (P = 0.0004, Fig. 2E; P = 0.0017, Fig. 2F). Together, these data demonstrate that NOL7 is more highly expressed in HCC tissues than in the adjacent normal tissues.

3.2 Relationship between NOL7 expression in HCC and clinicapathological parameters

To investigate the prognostic significance of NOL7 in HCC, we studied a cohort retrieved from the TCGA database, including 371 subjects. NOL7 expression and its correlation with the clinical characteristics of HCC patients are summarized in Table 1. NOL7 was expressed differently in different T stages, pathologic stages, and histologic stages in HCC (Fig. 3A–C). In addition, a significant association was observed between high NOL7 mRNA levels and several clinical factors, including elevated AFP (P < 0.001), vascular invasion (P = 0.007), poorer tumor differentiation (P < 0.001), more advanced T stage (P = 0.006), and higher TNM stage (P = 0.005) in the training set (Table 1). The Cox regression model revealed that patients with high NOL7 expression had poorer OS, DSS, and PFI than patients with low NOL7 expression (P = 0.005, P = 0.003, and P = 0.020, respectively, Fig. 3D–F).

Fig. 3

Multifaceted prognostic value of NOL7 in the training set. A Expression of NOL7 in different pathologic T stages in HCC. B Expression of NOL7 in different histologic stages in HCC. C Expression of NOL7 in different histologic grades in HCC. D Cox regression model estimates OS, DSS (E) and PFI (F) based on NOL7 mRNA levels in TCGA databases. *P < 0.05, **P < 0.01, ***P < 0.001

Univariate and multivariate Cox regression analyses were carried out to evaluate whether NOL7 had a independent predictive value for prognosis. Univariate Cox regression analysis showed a significant correlation between OS and T stage (P < 0.001), M stage (P = 0.018), and NOL7 expression (P = 0.015) (Table 2A) as well as between DSS and T stage (P < 0.001), M stage (P = 0.025), and NOL7 expression (P = 0.003) (Table 2B). Furthermore, we found that the T stage (P < 0.001), M stage (P = 0.037), vascular invasion (P = 0.003), and NOL7 expression (P = 0.033) were associated with PFI (Table 2C).

Table 2 Univariate Cox regression analysis of OS (A) , DSS (B), and PFI (C) related factors in TCGA databases

Multivariate analysis incorporated variables that were proven to be significant indicators of HCC prognosis. The findings revealed that NOL7 expression was an independent clinical prognostic factor linked to OS (P = 0.022, HR = 1.670) and DSS (P = 0.037, HR = 1.861) but not with PFI (P = 0.689, HR = 1.084) (Table 2D–F).

3.3 Nomogram construction and validation

To account for NOL7 expression and other predictive variables, a nomogram model was constructed to evaluate OS, incorporating pathological T stage and NOL7 mRNA expression (Fig. 4A). The C-index of the nomogram model was 0.635 (0.607–0.663). Subsequently, we assigned a score to each patient using the nomogram, and the predictive capability and agreement of the nomogram were assessed using ROC analysis and a calibration curve. The AUCs of the nomograms in term of the 1-,3-,and 5-years OS rates were 0.676, 0.627, and 0.604, respectively (Fig. 4B). The calibration plots demonstrated that the 1-, 3-, and 5-year OS rates were in similar agreement between the nomogram model and the ideal model (Fig. 4C). Moreover, a nomogram was devised to assess DSS, which incorporated pathological T stage and NOL7 mRNA expression (Fig. 4D). This nomogram had a C-index of 0.722 (0.691–0.753), indicating a good prediction performance. When evaluating the nomogram’s prediction ability in terms of ROC analysis and calibration cure, the nomogram AUCs for the 1-, 3-, and 5-year DSS rates were 0.732, 0.663, and 0.595, respectively (Fig. 4E). Furthermore, the agreement between the nomogram and ideal models was perfectly excellent as demonstrated through the calibration plots (Fig. 4F). Collectively, these findings validate the predictive potential of the nomogram in predicting the survival of patients with HCC.

Fig. 4

A Independent risk factors were evaluated for OS using multivariate Cox regression and were combined into the nomogram model. B ROC curves and nomogram AUC values for predicting 1-, 3-, and 5-year OS for HCC patients in the TCGA cohort. C Calibration charts for the nomogram depicting 1-year, 3-year and 5-year OS. D Independent risk factors were evaluated for DSS using multivariate Cox regression and were combined into the nomogram model. E ROC curves and nomogram AUC values for predicting 1-, 3-, and 5-year DSS for HCC patients in the TCGA cohort. F Calibration charts for the nomogram depicting 1-year, 3-year and 5-year DSS

3.4 Analysis of NOL7 function enrichment and co-expression gene screening in HCC

To understand the mechanisms of NOL7, we used the TCGA database to identify the correlations between NOL7 and various HCC-associated genes. Volcano and heatmap plots showed genes that were highly correlated (Fig. 5A–C). We found that the most positively correlated genes with NOL7 included SNRPC, ABT1, TAF11, XPO5, and WDR46, whereas the negatively correlated genes with NOL7 included ETNPPL, HP, C8A, ACSM2A, and AL354872.1. The 30 genes that were most strongly associated with NOL7 were selected for enrichment analysis using GO and KEGG (Fig. 5D). GO and KEGG enrichment analyses revealed that NOL7 plays important roles in the regulation of stem cell proliferation, embryonic organ morphogenesis, kidney epithelium development, proximal/distal pattern formation, and digestion, as well as several other biological processes. GSEA analysis of the KEGG pathway showed a positive correlation between elevated NOL7 levels and several pathways, including the cell cycle and DNA replication, homologous recombination, axon guidance, and retinol pathogenic E. coli infection (Fig. 5E). Conversely, GSEA revealed a negative correlation between high levels of NOL7 and pathways such as complement and coagulation cascades, peroxisome, fatty acid metabolism, valine, leucine, and isoleucine degradation, and retinol metabolism (Fig. 5F). These results indicate that the pathways regulating DNA replication and cell cycle control are strongly associated with NOL7 expression.

Fig. 5

Functional validation of NOL7 and potential pathway enrichment analysis. A Volcano plot for co-expressed genes associated with NOL7 expression. B The top 30 genes positively associated with NOL7 and the top 30 genes negatively associated with NOL7 in HCC (C). D GO/KEGG pathway analysis of 300 genes that had the strongest positive association with NOL7. E KEGG pathway analysis revealed 5 positively correlated pathways, and (F) KEGG pathway analysis showed 5 negatively correlated pathways, P < 0.05. G TMB was significantly linked to NOL7 expression. H A PPI network of NOL7 using the STRING tool. I NOL7 and 9 immune checkpoints and 15 core cell cycle modulators in HCC. ICI-related gene: LAG3、IDO1、CD47、CD274、CD276、CTLA4、TIGIT、PDCD1、HAVCR2. Cell cycle modulator: CDK1、CDK4、CDK2、CDK6、CCNB1、CCNA1、CCND2、CCND1、CCND3、CDKN2A、CDKN2B、CDKN2C、CDKN2D、E2F1、RB1; P < 0.001

3.5 Relationship between NOL7 expression and TMB

An increasing number of reports indicate that TMB plays a vital role in predicting cancer patient responsiveness to immune therapies [20, 21]. Therefore, it is important to explore the relationship between TMB and NOL7 expression in HCC. Our results revealed that NOL7 expression was significantly and positively correlated with TMB in HCC (Fig. 5G).

3.6 Correlation between NOL7 expression and protein interaction

Based on the STRING database, a PPI network for NOL7 was constructed. The top ten functional partner proteins with a high degree of connectivity were selected. These proteins were UTP4, WDR43, WDR75, NOL11,UTP15, FCF1, FBL, NOL10, RPS24, and RPS5 (Fig. 5H).

3.7 Correlation between NOL7 expression and immune cell infiltration

Immune cells that permeate tumors play an important role in the tumor milieu, and their infiltration is related to cancer progression, metastasis, and spread [22]. Therefore, we investigated the relationship between the potential correlation between the expression of NOL7 and the extent of immune cell infiltration in HCC. First, we evaluated the relationship between NOL7 expression and 9 immune checkpoints and 15 core cell cycle modulators. According to the Spearman correlation analysis, NOL7 expression was significantly associated with immune checkpoints, including LAG3, CD47, CD276, CTLA4, TIGIT, PDCD1, and HAVCR2 (Fig. 5I). NOL7 expression was also significantly associated with the expression of cell cycle modulators such as CDK2, CDK1,CDK6, CDK4, CCNA1, CCNB1, CCND2, CCND3, CDKN2A, CDKN2B, CDKN2C, CDKN2D, E2F1, and RB1 (Fig. 5I). All p-values were < 0.001. In addition, we found that the expression of six immune checkpoints was elevated in the NOL7 high-expression group, and most of the cell cycle modulators were also elevated in the NOL7 high-expression group (Fig. 6A, B). A significant majority of immune checkpoint molecules have shown a positive correlation with cell cycle regulatory molecules. These findings support our hypothesis that NOL7, as a critical cell cycle regulator, exerts a significant influence on HCC tumor immunity. In addition, the study showed a negative correlation between the expression of NOL7 and StromalScore and ESTIMATEScore (R = − 0.276, P < 0.001; R = − 0.170, P = 0.001, respectively) in HCC (Fig. 7A). However, there was no significant associated with the ImmuneScore (R = − 0.075, P = 0.149) (Fig. 7A). Our results also showed a significant correlation between NOL7 and tumors infiltrating lymphocytes (TILs) abundance (Fig. 7B). A positive correlation was observed between NOL7 expression and TH2 (R = 0.330, P < 0.001, Fig. 7C). Neutrophils (R = − 0.330, P < 0.001, Fig. 7D), DC (R = − 0.288, P < 0.001, Fig. 7E), and cytotoxic cells (R = − 0.268, P < 0.001, Fig. 7F) were negatively correlated with the expression of NOL7. The NOL7 group showed significantly higher Th2 cell infiltration enrichment scores than the low expression NOL7 groups (Fig. 7G). Conversely, the high expression NOL7 groups displayed significantly lower neutrophil and DC cell counts and cytotoxic cell infiltration enrichment scores (Fig. 7H–J).

Fig. 6

Comparison of immune checkpoints (A) and cell cycle modulators (B) between high and low NOL7 groups in HCC. *P <0.05, **P <0.01, ***P <0.001, ****P <0.0001, ns no significant

Fig. 7

Infiltration of immune cells and NOL7 expression levels. A Correlation of NOL7 expression with StromalScore, ESTIMATEScore and ImmuneScore. B Relationship between NOL7 and TILs abundance. C–F Correlation of NOL7 expression with immune cell enrichment fraction.G–J Comparing groups with high and low expression of NOL7 in terms of immune infiltration levels. *P < 0.05, **P < 0.01, ***P < 0.001, ns no significant

3.8 Correlation between NOL7 and the TME heterogeneity

We used four datasets (LIHC_GSE140228_10X, LIHC_GSE140228_Smartseq2, LIHC_GSE146409 and LIHC_GSE166635) from the TISCH database to assess NOL7 expression in TME-associated immune cells. For example, NOL7 expression in proliferating Tprolif cells, innate lymphoid cells (ILC), and dendritic cells (DC) was relatively higher in the LIHC_GSE140228_10X and LIHC_GSE140228_Smartseq2 datasets (Fig. 8A). In the LIHC_GSE146409 and LIHC_GSE166635 datasets, NOL7 expression levels remained elevated in malignant cells, suggesting that NOL7 plays a significant role in HCC progression (Fig. 8A). The violin plot showed the same trend for NOL7 expression in the HCC microenvironment (Fig. 8B). Figure 8C illustrates the specific distribution of the immune cells. NOL7 expression levels were significantly higher in malignant cells from two datasets (LIHC_GSE146409 and LIHC_GSE166635). These results suggest that NOL7 expression levels are quite different in distinct cell types,which may be the source of heterogeneity in the HCC microenvironment.

Fig. 8

NOL7-related cell type distribution using the TISCH database. A Cell types and their distribution in the four datasets. B Distribution of NOL7 in different cells in the four datasets. C The specific distribution of immune cells

3.9 Correlation between NOL7 and the cell pathway score

The correlation between the pathway score and mRNA expression value of NOL7 was plotted (Fig. 9). NOL7 mRNA expression was positively correlated with tumor proliferation, DNA repair, DNA replication, MYC targets, PI3K AKT mTOR pathway, pyrimidine metabolism, cellular response to hypoxia, and sphingolipid metabolism.

Fig. 9

Correlation plot of the NOL7 mRNA expression with the cell pathway score. A–H mRNA expression of NOL7 was positively correlated with the expression of many cell pathway

3.10 Drug sensitivity analysis

Some drugs have been identified as related to NOL7. Our study demonstrated a tendency for higher IC50 values of 5 − fluorouracil, cyclophosphamide, afatinib, gefitinib, and osimertinib in specimens at lower risk than in those at higher risk (Fig. 10A–E). In addition, there was a tendency for higher IC50 values of oxaliplatin, cisplatin, irinotecan, sorafenib, and cytarabine in specimens at higher risk compared to those at lower risk (Fig. 10F–J). This result suggests that the commonly used anticancer drugs in clinical treatment, including oxaliplatin, cisplatin, and sorafenib, were all less effective (higher IC50) in the high expression NOL7 groups.

Fig. 10

Relationship between NOL7 expression and drug sensitivity. Box plots for estimated IC50 of (A) 5 − fluorouracil, B cyclophosphamide, C afatinib, D gefitinib, E osimertinib, F oxaliplatin, G cisplatin, H irinotecan, I sorafenib, J cytarabine between high and low-risk HCC specimens