The SNHG3/miR-148a-3p axis-mediated high expression of DNMT1 is correlated with poor prognosis and tumor immune infiltration in hepatocellular carcinoma

Introduction

According to 2024 edition of GLOBOCAN, published by the International Agency for Research on Cancer’s Global Cancer Observatory, hepatocellular carcinoma (HCC) is one of the leading causes of cancer-related death worldwide, and its incidence has significantly increased in both developed and developing regions (1). In China, the 5-year survival rate of HCC patients is only 14.1%, and the recurrence rate is about 70% (2). Chemotherapy, liver transplantation, and surgery are common treatments for HCC, but they are only suitable for early-stage HCC patients (3). As the molecular pathogenesis of HCC is yet to be elucidated, it is essential to identify and characterize novel cancer-promoting genes to better understand this deadly disease, identify promising prognostic biomarkers, and develop more effective clinical therapies.

Epigenetic reprogramming regulates the malignant properties of HCC. DNA methylation is a key epigenetic regulatory mechanism that helps maintain the cancer-stem-cell pool in HCC and possibly other solid tumors (4) that are usually catalyzed by DNA-methyltransferases (DNMTs) DNMT1, DNMT3a, and DNMT3b (5). Site-specific hypermethylation and silencing of putative tumor-suppressor genes associated with abnormal expression of DNMTs have been shown to contribute to carcinogenesis and tumor progression (6). DNMT1 is considered a maintenance DNMT, which mainly maintains cytosine-phosphate-guanine (CpG) methylation, and is involved in embryonic development and somatic cell survival (7). DNMT1 is universally overexpressed in proliferating cells. Numerous studies have shown that DNMTI is closely associated with tumorigenesis and metastasis in various cancers, including melanoma (8), prostate cancer (9), pancreatic cancer (10), head and neck squamous carcinoma (11), and breast cancer (12). The overexpression of DNMT1 in tumors indicates poor prognosis (13). Lou et al. demonstrated that DNMT1 directly targets and hypermethylates the SOCS1 promoter to maintain stemness properties in human liver cancer stem-like cells, highlighting the critical role of DNMT1-mediated epigenetic silencing in cancer stem cell biology (14). Further, DNMT1 can modulate the immune system by maintaining the methylation status of the forkhead box P3 (Foxp3) gene (15). Nevertheless, to date, no comprehensive studies on the expression, prognosis, and mechanisms of DNMT1 in HCC have been conducted, and the relationship between DNMT1 and tumor immune infiltration in HCC remains poorly defined.

In the present study, expression analyses and survival analyses of DNMT1 were first conducted across various human cancers. Subsequently, microRNA (miRNA)-related and long non-coding RNA (lncRNA)-related DNMT1 regulation in HCC was investigated. The comprehensive bioinformatics analysis identified hsa-miR-148a-3p as the key upstream regulatory miRNA that was downregulated in HCC and significantly negatively correlated with DNMT1. The present study ultimately clarified the relationship between DNMT1 expression and the tumor immune microenvironment, including immune cell infiltration and the expression of immune checkpoints. We present this article in accordance with the REMARK reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-641/rc).

Methods

The Cancer Genome Atlas (TCGA) data acquisition

Level 3 RNA-sequencing (RNA-seq) data [fragments per kilobase per million (FPKM)] of 33 types of human cancer, including those from the TCGA-LIHC dataset, and its corresponding clinical information, were obtained from TCGA Genomic Data Commons (GDC, https://portal.gdc.cancer.gov/). Artificial intelligence (AI)-assisted literature mining tools were used to comprehensively analyze TCGA-LIHC dataset (16). The RNA-seq data (FPKM values) were then transformed to transcripts per million (TPM) reads and normalized into log₂ (TPM+ 1). Patients were stratified into low and high DNMT1 mRNA expression groups (n=187 each) based on the median expression value. Associations between DNMT1 expression and clinicopathological characteristics were evaluated using Chi-squared and Fisher’s exact tests, as presented in Table 1.

Kaplan-Meier plotter analysis

Using Kaplan-Meier plotter (http://kmplot.com/analysis/), an online survival database, we analyzed the association between DNMT1 expression and survival across eight human cancers. Specifically, we evaluated overall survival (OS) and relapse-free survival (RFS) across 21 cancer types.

Gene Expression Profiling Interactive Analysis (GEPIA) database analysis

Using the GEPIA web server (http://gepia.cancer-pku.cn/) and TCGA datasets, the correlation was evaluated between DNMT1 expression and immune-cell biomarkers in HCC.

Clinical samples

A total of 102 patients with primary HCC who underwent surgical treatment at Affiliated Hospital of Nantong University between January 2022 and March 2023 were retrospectively enrolled in the study (Table 2). Inclusion criteria: (I) histologically confirmed primary HCC; (II) underwent curative surgical resection; (III) no prior anticancer treatment; (IV) complete clinicopathological data available. Exclusion criteria: (I) distant metastasis; (II) recurrent HCC; (III) combined hepatocellular- cholangiocarcinoma; (IV) insufficient tissue samples for analysis; (V) loss to follow- up within 3 months post-surgery. All HCC diagnoses were confirmed by histopathological examination of surgical resection specimens. Included patients underwent primary surgical resection without prior neoadjuvant therapy. Tumor tissues were retrospectively collected from formalin-fixed, paraffin-embedded (FFPE) surgical blocks. The clinicopathological data of the patients were collected, including their: (I) demographic characteristics: gender and age; (II) tumor characteristics: maximum diameter (as determined by postoperative pathological examination); and (III) pathological features: differentiation grade (the cancers were classified as well-, moderately-, poorly-, moderately-to-poorly, or well-to-moderately differentiated according to the Edmondson-Steiner grading system) and microvascular invasion status. This study was approved by the Ethics Committee of Affiliated Hospital of Nantong University (No. 2023-L038), and was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Individual consent for this analysis was waived due to the retrospective nature of the study.

Immunohistochemical detection of DNMT1 protein expression

Immunohistochemical staining was performed on 102 pairs of HCC tissue microarrays at Affiliated Hospital of Nantong University. Using known positive and negative control sections as the reference, the staining intensity was graded as negative (no staining), weak, moderate, and strong. The proportion of positive cells was calculated as the percentage of positively stained cells among the total cells in each microscopic field. A comprehensive scoring system that integrates both staining intensity and the proportion of positive cells was employed. Staining intensity was scored as 0 (negative), 1 (weak), 2 (moderate), or 3 (strong). A composite scoring system was applied that combined staining intensity and the positive cell ratio, and the cells were graded as follows: 1 (≤40% positive cells or <5% if intensity =0), 2 (41–60%), 3 (61–80%), and 4 (>80%). Based on the final score (intensity × proportion), DNMT1 expression was classified as low [0–4], medium [5–8], or high [9–12].

Candidate miRNA prediction and starBase database analysis

Upstream binding miRNAs of DNMT1 were predicted by the following miRNA-target prediction programs: PITA (https://tools4mirs.org/software/target_prediction/pita/), RNA22 (https://cm.jefferson.edu/rna22/Interactive/), miRmap (https://mirmap.ezlab.org/), microT (https://bio.tools/DIANA-microT), miRanda, PicTar (https://pictar.mdc-berlin.de/), and TargetScan (http://www.targetscan.org/vert_72/). Ultimately, only those with miRNAs appearing in two or more of the above programs were included as candidate miRNAs of DNMT1 for subsequent analysis. The StarBase project (version 3.0, http://starbase.sysu.edu.cn/) was used to analyze the expression correlation between the candidate miRNAs and DNMT1, and the expression level of hsa-miR-148a-3p in HCC. In addition, starBase was employed to predict the candidate lncRNAs that could potentially bind to hsa-miR-148a-3p. starBase was also used to analyze the expression correlation between hsa-miR-148a-3p and SNHG3, and between SNHG3 and DNMT1 in HCC.

Gene Set Enrichment Analysis (GSEA)

GSEA (http://software.broadinstitute.org/gsea/index.jsp) was conducted between high- and low-DNMT1 expression groups to identify DNMT1-related functional significance based on the Hallmark gene set (“h.all.v7.0.symbols.gmt”). Statistical significance was set at |normalized enrichment score (NES)| >1, adjusted P value <0.05, and false discovery rate <0.05.

Cell-composition fraction estimation

To make reliable immune infiltration estimations, the “immunedeconv” package was used (https://icbi-lab.github.io/immunedeconv/), an R package that integrates six state-of-the-art algorithms, including xCell (https://xcell.ucsf.edu/), MCP-counter, TIMER, CIBERSORT, EPIC, and quanTIseq. Visualization was performed using the software packages “ggplot2” and “pheatmap”.

Additional bioinformatics and statistical analysis

R software (version 3.6.3, https://www.r-project.org/) was used to analyze the data and plot the graphs. The Wilcoxon rank-sum test was used for unpaired samples, and the Wilcoxon signed-rank test was used for paired samples. The results were visualized using the R package “ggplot2” (http://ggplot2.org). Patient characteristics between groups were compared using the Chi-squared test and Fisher’s exact test. Correlations of all RNAs, and correlations between the levels of RNAs and immune checkpoints in HCC were analyzed using Spearman’s correlation and visualized using “ggplot2”. A survival analysis of all the RNAs in the competing endogenous RNA (ceRNA) network was carried out, and the results were visualized using the survival package in R (https://CRAN.R-project.org/package=survival). The differentially expressed genes (DEGs) were determined by limma tests based on |log₂(foldchange)| >1 and adjusted P value <0.05 (17). The Kruskal-Wallis test was used to analyze RNA expression among different histological grades. A P value <0.05 was considered statistically significant.

Results

Pan-cancer analysis of DNMT1 expression

To investigate the possible roles of DNMT1 in carcinogenesis, the analysis first examined its expression in 33 types of cancer using TCGA-ALL dataset (normal samples =730, tumor samples =11,363). As Figure 1A shows, compared with the corresponding normal tissue, DNMT1 was upregulated in 16 types of human carcinoma, and downregulated in kidney chromophobe, and not significantly altered in kidney renal papillary cell carcinoma, pancreatic adenocarcinoma, thymic carcinoma or prostate adenocarcinoma. Next, we verified the expression of DNMT1 in TCGA tumor tissues compared with paired normal tissues. The results demonstrated that the expression of DNMT1 was significantly increased in UCEC, LIHC, BRCA, ESCA, HNSC, STAD, READ, BLCA, KIRC, LUAD, and LUSC, and significantly reduced in THCA (Figure 1B-1M). In summary, DNMT1 was upregulated in UCEC, LIHC, BRCA, ESCA, HNSC, STAD, READ, BLCA, KIRC, and LUAD, indicating that DNMT1 might play a key regulatory role in the carcinogenesis of these 10 cancers.

Figure 1 Pan-cancer expression analysis for DNMT1. (A) The expression of DNMT1 in 33 types of human cancer based on TCGA cancer and normal data. (B-M) DNMT1 expression in TCGA-UCEC (B), LIHC (C), BRCA (D), ESCA (E), HNSC (F), THCA (G), STAD (H), READ (I), BLCA (J), KIRC (K), LUAD (L), and LUSC (M) tissues compared with paired normal tissues. *, P<0.05; **, P<0.01; ***, P<0.001; ns, P>0.05. DNMT1, DNA-methyltransferase 1; TCGA, The Cancer Genome Atlas; TPM, transcripts per million.

The prognostic values of DNMT1 in human cancer

A survival analysis was subsequently performed, including that of OS and RFS, for DNMT1 in UCEC, BLCA, BRCA, HNSC, LIHC, LUSC, THCA, and LUAD. In terms of OS, the LIHC patients with higher expression of DNMT1 had poorer prognosis; however, the HNSC patients with higher expression of DNMT1 had better prognosis (Figure 2). In terms of RFS, the increased expression of DNMT1 was correlated with poor clinical outcomes in LIHC and THCA patients (Figure 3). Taken together, these results suggest that DNMT1 could be used as a biomarker for predicting unfavorable prognosis in HCC patients.

Figure 2 OS of dichotomized DNMT1 expression in various human cancers. (A-H) The OS plot of DNMT1 in UCEC (A), BLCA (B), BRCA (C), HNSC (D), LIHC (E), LUSC (F), THCA (G), and LUAD (H). CI, confidence interval; DNMT1, DNA-methyltransferase 1; HR, hazard ratio; OS, overall survival.

Figure 3 RFS analysis of dichotomized DNMT1 expression in various human cancers. (A-H) The RFS plot of DNMT1 in UCEC (A), BLCA (B), BRCA (C), HNSC (D), LIHC (E), LUSC (F), THCA (G), and LUAD (H). CI, confidence interval; DNMT1, DNA-methyltransferase 1; HR, hazard ratio; RFS, relapse-free survival.

The expression of DNMT1 and clinical characteristics in HCC

Clinical and gene expression data of HCC patients were downloaded from TCGA database, including data on gender, age, histologic grade, alpha-fetoprotein (AFP) level, OS, Child-Pugh grade, T classification, and pathologic stage. The DNMT1 mRNA expression levels were analyzed with TCGA-LIHC unpaired and paired samples. The results demonstrated that DNMT1 mRNA was significantly upregulated in the tumor tissues compared to the adjacent non-tumor tissues (Figure 4A,4B). Additionally, the immunohistochemical analysis of the 102 pairs of primary HCC and corresponding non-tumor liver tissues showed that DNMT1 protein was overexpressed in the majority of the HCC samples examined (Figure 4C). Additionally, statistical analyses were performed to evaluate the associations between DNMT1 mRNA expression and various clinicopathological parameters in HCC patients. The patients were divided into high- and low-expression groups based on median DNMT1 mRNA expression. The results revealed that DNMT1 mRNA expression was statistically associated with age, histological grade, and the serum AFP level (all P<0.001; Table 1).

Figure 4 Identification of hsa-miR-148a-3p as a potential upstream miRNA of DNMT1 in HCC. (A,B) DNMT1 mRNA was significantly upregulated in the tumor tissues compared to the adjacent non-tumor tissues. (C) DNMT1 immunostaining was performed on a tissue microarray comprising 102 paired HCC tissues and adjacent normal samples, and DNMT1 was found to be frequently overexpressed in HCC. Representative immunostaining images from an HCC patient are shown. Magnifcation: ×20 (left), ×200 (right). The percentage of cases displaying low, medium, and high staining intensity of DNMT1 in the paired HCC samples and adjacent normal tissues is illustrated by the bar graph (Fisher’s exact test). In the tumor tissues, DNMT1 expression was categorized as high in 47 cases (46.08%), moderate in 33 cases (32.35%), low in 19 cases (18.63%), and negative in 3 cases (2.94%). In the adjacent normal tissues, DNMT1 expression was categorized as high in 21 cases (20.59%), moderate in 40 cases (39.22%), low in 35 cases (34.31%), and negative in 6 cases (5.88%). (D) The miRNA-DNMT1 regulatory network established by Cytoscape software. (E) The expression correlation between predicted miRNAs and DNMT1 in HCC analyzed by the starBase database. (F) The expression of hsa-miR-148a-3p in HCC and normal samples in TCGA-LIHC. (G) The prognostic value of hsa-miR-148a-3p in HCC. Statistical significance is indicated as: **, P<0.01; ***, P<0.001. CI, confidence interval; DNMT1, DNA-methyltransferase 1; HCC, hepatocellular carcinoma; HR, hazard ratio; mRNA, messenger RNA; miRNA, microRNA; TCGA, The Cancer Genome Atlas; TPM, transcripts per million.

Prediction and analysis of upstream miRNAs of DNMT1

Non-coding RNAs (ncRNAs) are responsible for the regulation of gene expression, and can be classified as miRNAs, small nucleolar RNAs (snoRNAs), circular RNAs, and lncRNAs (18). To determine whether DNMT1 was regulated by some ncRNAs, upstream miRNAs that might bind to DNMT1 were first predicted and ultimately 14 miRNAs were identified.

To better visualize the regulatory network, we constructed a miRNA-DNMT1 network with Cytoscape (https://cytoscape.org/) (Figure 4D). Given the miRNA mechanism of action, a negative correlation between miRNAs and DNMT1 was expected. We then performed expression correlation analyses for the miRNA-DNMT1 pairs. As shown in Figure 4E, DNMT1 exhibited a significant negative correlation with hsa-miR-148a-3p and a significant positive correlation with the other 13 predicted miRNAs in HCC. Next, the expression of hsa-miR-148a-3p was assessed in the HCC and normal control samples using starBase, and the prognostic value of hsa-miR-148a-3p in HCC was evaluated using Kaplan-Meier plotter. As Figure 4F,4G show, hsa-miR-148a-3p was significantly downregulated in HCC, and its downregulation was correlated with poor prognosis. Together, these findings suggest that hsa-miR-148a-3p might be the most influential upstream miRNA of DNMT1 in HCC.

Prediction and analysis of upstream lncRNAs of hsa-miR-148a-3p

Subsequently, the upstream lncRNAs of hsa-miR-148a-3p were predicted using the starBase database and a total of 45 possible lncRNAs were obtained. For better visualization, the lncRNA-hsa-miR-148a-3p regulatory network was established using Cytoscape software (Figure 5A). A total of 1,548 DEGs were detected between TCGA-LIHC tumor samples and normal tissues using the “limma” R package. A Venn diagram was finally created showing three intersections of the 1,454 DEGs and 45 possible lncRNAs (Figure 5B), indicating that LINC01554, SNHG3, and H19 were the co-expressed differential lnRNAs found in both cohorts. As the volcano plot in Figure 5C shows, in HCC, SNHG3 was the upregulated gene, and LINC01554 or H19 was the downregulated gene. Subsequently, the prognostic value of SNHG3 in HCC was assessed, and the results revealed that the overexpression of SNHG3 indicated poor OS in patients with HCC (Figure 5D). Based on the above results, the regulatory relationship was further explored to see whether SNHG3 regulates hsa-miR-148a-3p expression as a ceRNA in HCC. According to the ceRNA hypothesis, lncRNAs usually serve as ceRNAs by binding to miRNAs, and the key qualified lncRNAs in the ceRNA subnet should be both negatively associated with miRNA and positively associated with mRNA. The expression correlation between SNHG3 and hsa-miR-148a-3p or DNMT1 in HCC is shown in Figure 5E,5F. Cumulatively, these findings suggest that SNHG3 might serve as a ceRNA to mediate DNMT1 by competitively binding to hsa-miR-148a-3p.

Figure 5 Identification of SNHG3 as a potential upstream lncRNAs of hsa-miR-148a-3p in HCC. (A) The lncRNA-miR-148a-3p regulatory network established by Cytoscape software. (B) Venn diagram showing the overlap of 45 possible upstream lncRNAs of miR-148a-3p (LIST2) with DEGs between TCGA-LIHC tumor samples and normal tissues. (C) Volcano plot of differentially expressed lncRNAs (|log₂ fold change| ≥1 and adjusted P value <0.05). Red represents the upregulated genes, and blue represents the downregulated genes. The three overlapping genes are highlighted and labeled. (D) The prognostic value of SNHG3 in HCC. (E) The correlation between hsa-miR-148a-3p expression and SNHG3 expression in HCC from the starBase project (version 3.0). (F) The correlation between SNHG3 expression and DNMT1 expression in HCC from the starBase project (version 3.0). CI, confidence interval; DEGs, differentially expressed genes; DNMT1, DNA-methyltransferase 1; HCC, hepatocellular carcinoma; HR, hazard ratio; lncRNA, long non-coding RNA; TCGA, The Cancer Genome Atlas.

DNMT1 expression was positively related to immune cell infiltration in HCC

Tumor-infiltrating immune cells are independent predictors of cancer survival. As Figure 6A shows, there was no significant change in the level of immune cell infiltration with the copy number alteration of DNMT1 in HCC. Thus, a correlation analysis was conducted between DNMT1 expression and immune cell infiltration in HCC using TIMER (http://timer.cistrome.org/). The expression of DNMT1 was significantly positively related to all the analyzed immune cells, including B cells (Cor =0.486; P=8.03e−22), CD8⁺ T cells (Cor =0.331; P=3.36e−10), CD4⁺ T cells (Cor =0.494; P=1.40e−22), macrophages (Cor =0.541; P=2.27e−24), neutrophils (Cor =0.467; P=4.67e−20), and dendritic cells (DCs, Cor =0.536; P=1.22e−26) in HCC (Figure 6B-6G).

Figure 6 Correlation between DNMT1 levels and immune cell infiltration in HCC. (A) The level of immune cell infiltration in HCC under different copy numbers of DNMT1. (B-G) The relationship between the DNMT1 expression level in HCC and the infiltration levels of (B) B cells, (C) CD8⁺ T cells, (D) CD4⁺ T cells, (E) macrophages, (F) neutrophils, and (G) DCs. DNMT1, DNA-methyltransferase 1; DC, dendritic cell; HCC, hepatocellular carcinoma; TPM, transcripts per million.

Next, to further investigate the role of DNMT1 in tumor immunity, the HCC patients from TCGA were divided into DMNT1-high and DNMT1-low subgroups. XCell was used to compare the differences in the abundance of tumor-infiltrating immune cells and extracellular matrix cells between the two groups (Figure S1). The results illustrated that the stromal score was higher in the low-DNMT1 expression group than the high-DNMT1 expression group (P<0.001). The high-DNMT1 expression group had a significantly higher abundance of CD4⁺ memory T cells, Th2 cells, gamma delta T cells, natural killer T cells, monocytes, and most DCs, and B cells, but a significantly lower abundance of M2 macrophages, CD4 central memory T cells, CD8⁺ naive T cells, hematopoietic stem cells, granulocyte-macrophage progenitor cells, and endothelial cells compared to the low-DNMT1 expression group (P<0.05). Finally, we used the GEPIA database to determine the expression correlation between DNMT1 and immune cell biomarkers in HCC. DNMT1 expression was significantly positively correlated with the gene markers of B cells (CD19 and CD79A), CD8⁺ T cells (CD8A and CD8B), CD4⁺ T cells (CD4), M1 macrophages (IRF5 and PTGS2), M2 macrophages (CD163, VSIG4, and MS4A4A), neutrophils (ITGAM and CCR7), and DCs (HLA-DPB1, HLA-DRA, HLA-DPA1, CD1C, NRP1, and ITGAX) in HCC (Table 3). All these results indicate that DNMT1 is significantly positively related to immune cell infiltration.

DNMT1 expression was positively associated with immune checkpoints in HCC

The relationship between DNMT1 and tumor immune escape was plotted. The expression of DNMT1 was significantly positively correlated with all the three immune checkpoints, CTLA-4, PD-1, and PD-L1, showing statistical significance in HCC (Figure 7A-7C). Similar to the GEPIA, the analysis also revealed that DNMT1 expression was significantly positively corelated with CTLA-4, PD-1, and PD-L1 in HCC, with statistical significance as indicated by TIMER (Figure 7D-7F). These findings suggest that the carcinogenic effects mediated by DNMT1 might be related to the dysfunctional state of T cells and tumor immune escape.

Figure 7 Correlation between DNMT1 expression and immune checkpoint molecules in HCC. (A-C) Correlation analysis between DNMT1 and CTLA-4, PD-1 (PDCD1), and PD-L1 (CD274) expression using GEPIA. (D-F) Spearman’s correlation analysis between DNMT1 and CTLA-4, PD-1 (PDCD1), and PD-L1 (CD274) expression adjusted for tumor purity using TIMER. CTLA-4, cytotoxic T lymphocyte antigen 4; DNMT1, DNA-methyltransferase 1; HCC, hepatocellular carcinoma; PD-1, programmed cell death protein 1; PD-L1, programmed death ligand 1; TPM, transcripts per million.

GSEA identified DNMT1-related hallmark pathways

To determine the potential function of DNMT1 in HCC, GSEAs were conducted between the high- and low-DNMT1 expression groups. The top four hallmark pathways (adjusted P value <0.05) involved in the high-DNMT1 expression group were epithelial-mesenchymal transition (EMT), the G₂/M checkpoint, E2F_targets, and the inflammatory response (Figure 8A), while the top four hallmark pathways (adjusted P value <0.05) involved in the low-DNMT1 expression group were bile acid metabolism, fatty acid metabolism, oxidative phosphorylation, and xenobiotic metabolism (Figure 8B). Cyclin-dependent kinase 1 (CDK1) is the key regulator of the G₂/M checkpoint (19). Further analysis examined the expression correlation between DNMT1, and CDK1 and E2Fs (E2F1, E2F2, E2F3, E2F4, E2F5, E2F6, E2F7, and E2F8) in HCC. As Figure 8C shows, DNMT1 expression was significantly and positively correlated with CDK1 expression. As Figure 8D-8K further show, DNMT1 expression was significantly positively related with E2F1, E2F2, E2F3, E2F4, E2F5, E2F6, E2F7, and E2F8 in HCC.

Figure 8 GSEA revealed the Hallmark pathways related to DNMT1 in HCC. (A) GSEA enrichment plots of Hallmark gene sets that are significantly up-regulated in the high-DNMT1 subgroup. (B) GSEA enrichment plots of Hallmark gene sets that are significantly up-regulated in the low-DNMT1 subgroup. (C) Scatter-plot showing the significantly positive correlation between DNMT1 and CDK1 mRNA levels in HCC samples. (D-K) The correlation between DNMT1 expression and individual E2F family members in HCC: E2F1 (D), E2F2 (E), E2F3 (F), E2F4 (G), E2F5 (H), E2F6 (I), E2F7 (J), and E2F8 (K). DNMT1, DNA-methyltransferase 1; FPKM, fragments per kilobase per million; GSEA, Gene Set Enrichment Analysis; HCC, hepatocellular carcinoma.

Discussion

The occurrence and development of HCC is a complex, dynamic biological process that involves genetic factors, the epigenetic cell state, and microenvironment alterations. Clarifying the molecular mechanism underlying HCC carcinogenesis may contribute to the development of effective therapeutic targets and valuable prognostic biomarkers. Accumulating evidence has shown that DNMT1 participates in the tumorigenesis and progression of various human cancers, including HCC (8,9,13). However, knowledge about DNMT1 in HCC remains insufficient, and further research is needed.

This study was performed to identify the feasibility of DNMT1 as a promising biomarker in HCC patients. The present study first performed a pan-cancer analysis to examine the expression of DNMT1 using TCGA database. Next, we conducted a survival analysis of DNMT1 in some of the statistically significant cancer types, as described above, and found that the increased expression of DNMT1 was correlated with poor clinical outcomes in HCC. Moreover, high DNMT1 expression was associated with histological grade. This suggests that upregulated DNMT1 might be involved in malignant transformation, which is in line with previous findings (13).

The ceRNA hypothesis asserts that lncRNA mainly regulates mRNA through the ceRNA regulatory mechanism, and RNAs affect each other’s levels by competing with a limited pool of miRNAs (20). DNMT1 has been shown to inhibit the transcription of tumor-suppressive miRNAs in cancer progression by maintaining their hypermethylation (21). In our study, a lncRNA-miRNA-mRNA regulatory network was developed with DNMT1 in HCC. Through candidate miRNA prediction conducted by prediction programs (i.e., PITA, RNA22, miRmap, microT, miRanda, PicTar, and TargetScan) and a correlation analysis, including an expression analysis and survival analysis, the analyses ultimately identified has-miR-148a-3p as the most likely potential upstream miRNA of DNMT1 in HCC. has-miR-148a-3p is a member of the miR-148/152 family, and has been reported to be a tumor suppressor for various human cancers, including pancreatic cancer (22), esophageal cancer (23), and HCC (24). Recent evidence has confirmed that the reciprocal negative regulation between hsa-miR-148a-3p and DNMT1 contributes to cell proliferation, cell-cycle processes, and the maintenance of cancer stem cell characteristics in HCC (25).

Subsequently, upstream lncRNAs of the hsa-miR-148a-3p/DNMT1 axis were predicted, yielding 45 possible lncRNAs. These lncRNAs were intersected with 1,548 DEGs between TCGA-LIHC tumor samples and normal tissues to screen for differentially expressed lncRNAs. Based on volcano plot analysis, ceRNA hypothesis, and correlation analyses, SNHG3 was ultimately identified as the upstream lncRNA. SNHG3 has been reported to function as an oncogene in many types of malignancies (26). Meanwhile, multiple miRNAs in HCC have been shown to promote tumor growth and metastasis by targeting SNHG3 (27,28). Based on these findings, we identified SNHG3/hsa-miR-148a-3p/DNMT1 axis as the potential regulatory pathway in HCC.

At present, compared with other tumors, immunotherapy for liver cancer is still in its infancy (29). The increased infiltration of immune cells in tumors and the high expression of immune checkpoints contribute to the efficacy of immunotherapy (30). Epigenetic modulation enhances immunotherapy for HCC by upregulating previously repressed neoantigens and increasing cytotoxic T-cell infiltration in the immunosuppressive tumor microenvironment (31). Research has shown that increased T-cell methylation is beneficial and that the epigenetic control of Foxp3 in intratumoral T-cells can regulate the growth of HCC (32). The results ultimately revealed DNMT1 expression to be positively associated with immune cell infiltrates, such as DCs, CD4⁺ T cells, and B cells. Meanwhile, the analysis also revealed positive correlations between DNMT1 expression and immune checkpoints, which affect T-cell exhaustion and immune escape. Taken together these findings indicate that DNMT1 may be a valuable immunotherapy biomarker, and targeting DNMT1 might enhance the efficacy of immunotherapy in HCC. Recent studies have further supported the role of DNMT1 in immune regulation. Tao et al. demonstrated that DNMT1-mediated DNA methylation contributes to HCC resistance to immune checkpoint inhibitors by silencing immune surveillance genes, which aligns with our findings of positive correlations between DNMT1 expression and immune cell infiltrates (33). Additionally, Shen et al. identified specific tumor-associated macrophage subtypes (C1QA+ and THBS1+ macrophages) associated with poor prognosis, providing mechanistic insights into how DNMT1-regulated pathways may influence immune cell recruitment and function in the heterogeneous HCC tumor microenvironment (34).

To further explore the biological functions of DNMT1, GSEAs were conducted between the high- and low-DNMT1 expression groups. The GSEA results revealed that the hallmark gene sets enriched in the high-DNMT1 expression group were mainly related to EMT, E2F targets, the G₂M checkpoint, and the inflammatory response. The EMT process is critical for epithelial cell invasion, resistance to apoptosis, tumor dissemination, and drug resistance (acquired ability to survive therapeutic intervention) (35). It might be that the DNMT1-induced epigenetic silencing of SFRP1 causes the activation of the Wnt signaling pathway and increases the aggressiveness of HCC by the induction of EMT (36). Both the E2F and G₂/M checkpoints are targets associated with the cell cycle. Additionally, E2F activators regulate the transition from the G1 to S phase in the cell cycle, and control cell apoptosis and differentiation (37). The key member of the hallmark G₂M checkpoint gene set was CDK1. Meanwhile, the results showed that DNMT1 significantly increased the expression of E2Fs and CDK1. These results showed that increased DNMT1 expression participates in tumor progression via a deleterious interaction with the cell cycle-related molecules. The GSEAs also showed the enrichment of genes involved in the inflammatory response, which might account for the increased infiltration of immune cells in tumors. Future mechanistic studies incorporating protein-protein interaction network analyses using tools such as STRING or Cytoscape would further elucidate DNMT1-centered signaling modules and provide deeper insights into its role in cell cycle regulation and immune modulation. There are several limitations in this study that should be acknowledged. First, we did not collect tumor-node-metastasis (TNM) stage, Child-Pugh class, nor histological grade/stage, which limits evaluation of stage-related differences and may affect the generalizability of our findings. Second, although our computational predictions and immunohistochemistry (IHC) analyses are supportive, functional validation—such as luciferase reporter assays, siRNA-mediated knockdown, and rescue experiments—is required to conclusively establish the SNHG3/miR-148a-3p/DNMT1 regulatory axis. Future studies should incorporate functional validation through in vitro and in vivo experiments to substantiate the mechanistic relationships identified in this bioinformatics analysis. Additionally, key research priorities include developing synergistic DNMT1 inhibitor-immunotherapy combinations, characterizing epigenetic-metabolic networks regulating DNMT1 activity, and deploying AI-powered models for therapy response prediction. These targeted investigations will expedite the translation of DNMT1 findings into clinically applicable HCC treatments. Cumulatively, these findings elucidate the way in which DNMT1 participates in HCC, which will help develop targeted therapy research in the future.

Conclusions

In summary, this study established the SNHG3/hsa-miR-148a-3p/DNMT1 axis as a potential regulatory pathway of hepatocarcinogenesis, which was also identified as a biomarker of poor prognosis. The present study further found that DNMT1 might exert its oncogenic effect through modulating cell-cycle progression by regulating transcription, and increasing tumor immune cell infiltration and immune checkpoint expression in HCC. However, these results should be validated by more basic hepatocarcinogenesis-related experiments in the future.

Acknowledgments

We would like to thank the patients and investigators who participated in TCGA for providing the data.

Footnote

Reporting Checklist: The authors have completed the REMARK reporting checklist. Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-641/rc

Data Sharing Statement: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-641/dss

Peer Review File: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-641/prf

Funding: This study was supported by grants from the National Natural Science Foundation (No. 82070622), the Jiangsu Provincial Research Hospital (No. YJXYY202204), the Key Research and Development Plan of Jiangsu Province (No. BE2020668), the Beijing Life Oasis Cancer Research Fund (No. BH004506), the Nantong Science and Technology Project (No. MSZ2023174), and the Jiangsu Provincial Postgraduate Research and Practice Innovation Program (No. SJCX24_2068).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-641/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Affiliated Hospital of Nantong University (No. 2023-L038), and individual consent for this analysis was waived due to the retrospective nature of the study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin 2024;74:12-49. [Crossref] [PubMed]
2. Allemani C, Matsuda T, Di Carlo V, et al. Global surveillance of trends in cancer survival 2000-14 (CONCORD-3): analysis of individual records for 37 513 025 patients diagnosed with one of 18 cancers from 322 population-based registries in 71 countries. Lancet 2018;391:1023-75. [Crossref] [PubMed]
3. Forner A, Reig M, Bruix J. Hepatocellular carcinoma. Lancet 2018;391:1301-14. [Crossref] [PubMed]
4. Raggi C, Factor VM, Seo D, et al. Epigenetic reprogramming modulates malignant properties of human liver cancer. Hepatology 2014;59:2251-62. [Crossref] [PubMed]
5. Jeltsch A, Jurkowska RZ. Allosteric control of mammalian DNA methyltransferases - a new regulatory paradigm. Nucleic Acids Res 2016;44:8556-75. [Crossref] [PubMed]
6. Arechederra M, Daian F, Yim A, et al. Hypermethylation of gene body CpG islands predicts high dosage of functional oncogenes in liver cancer. Nat Commun 2018;9:3164. [Crossref] [PubMed]
7. Iwanami N, Lawir DF, Sikora K, et al. Transgenerational inheritance of impaired larval T cell development in zebrafish. Nat Commun 2020;11:4505. [Crossref] [PubMed]
8. Maric H, Supic G, Kandolf-Sekulovic L, et al. DNMT1 and DNMT3B genetic polymorphisms affect the clinical course and outcome of melanoma patients. Melanoma Res 2019;29:596-602. [Crossref] [PubMed]
9. Lee E, Wang J, Yumoto K, et al. DNMT1 Regulates Epithelial-Mesenchymal Transition and Cancer Stem Cells, Which Promotes Prostate Cancer Metastasis. Neoplasia 2016;18:553-66. [Crossref] [PubMed]
10. Wong KK. DNMT1 as a therapeutic target in pancreatic cancer: mechanisms and clinical implications. Cell Oncol (Dordr) 2020;43:779-92. [Crossref] [PubMed]
11. Cui J, Zheng L, Zhang Y, et al. Bioinformatics analysis of DNMT1 expression and its role in head and neck squamous cell carcinoma prognosis. Sci Rep 2021;11:2267. [Crossref] [PubMed]
12. Vernier M, McGuirk S, Dufour CR, et al. Inhibition of DNMT1 and ERRα crosstalk suppresses breast cancer via derepression of IRF4. Oncogene 2020;39:6406-20. [Crossref] [PubMed]
13. Saito Y, Kanai Y, Nakagawa T, et al. Increased protein expression of DNA methyltransferase (DNMT) 1 is significantly correlated with the malignant potential and poor prognosis of human hepatocellular carcinomas. Int J Cancer 2003;105:527-32. [Crossref] [PubMed]
14. Lou L, Deng T, Yuan Q, et al. Targeted silencing of SOCS1 by DNMT1 promotes stemness of human liver cancer stem-like cells. Cancer Cell Int 2024;24:206. [Crossref] [PubMed]
15. Sun X, Cui Y, Feng H, et al. TGF-β signaling controls Foxp3 methylation and T reg cell differentiation by modulating Uhrf1 activity. J Exp Med 2019;216:2819-37. [Crossref] [PubMed]
16. Prelaj A, Miskovic V, Zanitti M, et al. Artificial intelligence for predictive biomarker discovery in immuno-oncology: a systematic review. Ann Oncol 2024;35:29-65. [Crossref] [PubMed]
17. Ritchie ME, Phipson B, Wu D, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 2015;43:e47. [Crossref] [PubMed]
18. Wang J, Zhu S, Meng N, et al. ncRNA-Encoded Peptides or Proteins and Cancer. Mol Ther 2019;27:1718-25. [Crossref] [PubMed]
19. Liu S, Liu H, Kang J, et al. The Severe Fever with Thrombocytopenia Syndrome Virus NSs Protein Interacts with CDK1 To Induce G(2) Cell Cycle Arrest and Positively Regulate Viral Replication. J Virol 2020;94:e01575-19. [Crossref] [PubMed]
20. Salmena L, Poliseno L, Tay Y, et al. A ceRNA hypothesis: the Rosetta Stone of a hidden RNA language? Cell 2011;146:353-8. [Crossref] [PubMed]
21. Ning X, Shi Z, Liu X, et al. DNMT1 and EZH2 mediated methylation silences the microRNA-200b/a/429 gene and promotes tumor progression. Cancer Lett 2015;359:198-205. [Crossref] [PubMed]
22. Fu X, Hong L, Yang Z, et al. MicroRNA-148a-3p suppresses epithelial-to-mesenchymal transition and stemness properties via Wnt1-mediated Wnt/β-catenin pathway in pancreatic cancer. J Cell Mol Med 2020;24:13020-35. [Crossref] [PubMed]
23. Wang Y, Hu Y, Guo J, et al. miR-148a-3p Suppresses the Proliferation and Invasion of Esophageal Cancer by Targeting DNMT1. Genet Test Mol Biomarkers 2019;23:98-104. [Crossref] [PubMed]
24. Jung KH, Zhang J, Zhou C, et al. Differentiation therapy for hepatocellular carcinoma: Multifaceted effects of miR-148a on tumor growth and phenotype and liver fibrosis. Hepatology 2016;63:864-79. [Crossref] [PubMed]
25. Li X, Wang L, Cao X, et al. Casticin inhibits stemness of hepatocellular carcinoma cells via disrupting the reciprocal negative regulation between DNMT1 and miR-148a-3p. Toxicol Appl Pharmacol 2020;396:114998. [Crossref] [PubMed]
26. Xu B, Mei J, Ji W, et al. LncRNA SNHG3, a potential oncogene in human cancers. Cancer Cell Int 2020;20:536. [Crossref] [PubMed]
27. Zhao Q, Wu C, Wang J, et al. LncRNA SNHG3 Promotes Hepatocellular Tumorigenesis by Targeting miR-326. Tohoku J Exp Med 2019;249:43-56. [Crossref] [PubMed]
28. Zhang PF, Wang F, Wu J, et al. LncRNA SNHG3 induces EMT and sorafenib resistance by modulating the miR-128/CD151 pathway in hepatocellular carcinoma. J Cell Physiol 2019;234:2788-94. [Crossref] [PubMed]
29. Johnston MP, Khakoo SI. Immunotherapy for hepatocellular carcinoma: Current and future. World J Gastroenterol 2019;25:2977-89. [Crossref] [PubMed]
30. Yang Y. Cancer immunotherapy: harnessing the immune system to battle cancer. J Clin Invest 2015;125:3335-7. [Crossref] [PubMed]
31. Hong YK, Li Y, Pandit H, et al. Epigenetic modulation enhances immunotherapy for hepatocellular carcinoma. Cell Immunol 2019;336:66-74. [Crossref] [PubMed]
32. Liu Q, Du F, Huang W, et al. Epigenetic control of Foxp3 in intratumoral T-cells regulates growth of hepatocellular carcinoma. Aging (Albany NY) 2019;11:2343-51. [Crossref] [PubMed]
33. Tao S, Liang S, Zeng T, et al. Epigenetic modification-related mechanisms of hepatocellular carcinoma resistance to immune checkpoint inhibition. Front Immunol 2022;13:1043667. [Crossref] [PubMed]
34. Shen KY, Zhu Y, Xie SZ, et al. Immunosuppressive tumor microenvironment and immunotherapy of hepatocellular carcinoma: current status and prospectives. J Hematol Oncol 2024;17:25. [Crossref] [PubMed]
35. Kahlert UD, Nikkhah G, Maciaczyk J. Epithelial-to-mesenchymal(-like) transition as a relevant molecular event in malignant gliomas. Cancer Lett 2013;331:131-8. [Crossref] [PubMed]
36. Quan H, Zhou F, Nie D, et al. Hepatitis C virus core protein epigenetically silences SFRP1 and enhances HCC aggressiveness by inducing epithelial-mesenchymal transition. Oncogene 2014;33:2826-35. [Crossref] [PubMed]
37. Rowland BD, Bernards R. Re-evaluating cell-cycle regulation by E2Fs. Cell 2006;127:871-4. [Crossref] [PubMed]

(English Language Editor: L. Huleatt)