SNHG3/miR-148a-3p Axis–mediated High Expression of DNMT1 Correlates with Poor Prognosis and Tumor Immune Inltration of Hepatocellular Carcinoma

Background Epigenetic reprogramming plays an important role in the occurrence, development, and 31 prognosis of hepatocellular carcinoma (HCC). DNA methylation is a key epigenetic 32 regulatory mechanism, and DNA methyltransferase 1 (DNMT1) is the major enzyme 33 responsible for maintenance methylation. Nevertheless, the role and mechanism of 34 DNMT1 in HCC remains poorly defined. and Genome data Set Enrichment between high-and-low DNMT1 to

HCC, and DNMT1 might be a biomarker predicting unfavorable prognosis in HCC 48 patients. DNMT1 mRNA expression was statistically associated with age, histological 49 grade, and the level of serum AFP. Moreover, DNMT1 level was significantly and 50 positively linked to tumor immune cell infiltration, immune cell biomarkers, and 51 immune checkpoint expression. Meanwhile, Gene Set Enrichment Analysis (GSEA) 52 revealed that high-DNMT1 expression was associated with epithelial mesenchymal 53 transition (EMT), E2F target, G2M checkpoint, and inflammatory response. Finally, 54 through a combination series of computer analyses the SNHG3/hsa-miR-148a-55 3p/DNMT1 axis was confirmed as the potential regulatory pathway in HCC. and LUAD, indicating that DNMT1 might play a key regulatory role in the 184 carcinogenesis of these 10 cancers. 185

The prognostic values of DNMT1 in human cancer 187
We next performed survival analysis, including OS and RFS, for DNMT1 in UCEC, 188 BLCA, BRCA, HNSC, LIHC, LUSC, THCA, and LUAD. For OS, LIHC patients 189 with higher expression of DNMT1 had a poorer prognosis; however, those with a 190 higher expression with HNSC had a better prognosis (Fig.2). For RFS, increased 191 expression of DNMT1 was correlated with poor clinical outcomes in LIHC and 192 THCA patients (Fig. 3). Taken together, these data suggested that DNMT1 could be 193 used as a biomarker predicting unfavorable prognosis in HCC patients. 194 195

Relationship between DNMT1 expression and clinical characteristics in HCC 196
We downloaded clinical and gene expression data of HCC patients from TCGA 197 database, including gender, age, histologic grade, alpha-fetoprotein (AFP), OS,  Pugh grade, T classification, and pathologic stage. Then, associations between clinical 199 characteristics and DNMT1 mRNA expression in HCC were analyzed. Patients were 200 divided into high-and low-expression groups on the basis of median DNMT1 mRNA 201 expression. Our results revealed that DNMT1 mRNA expression was statistically 202 associated with age, histological grade, and the level of serum AFP (all P< 0.001,; Table  203 1). 204 205 Prediction and analysis of upstream miRNAs of DNMT1 206 ncRNAs are responsible for the regulation of gene expression and can be classified into 207 miRNAs, small nucleolar RNAs (snoRNAs), circular RNAs (circRNAs), and lncRNAs 208 (18). To determine whether DNMT1 was regulated by some ncRNAs, we first predicted 209 the upstream miRNAs that might bind to DNMT1 and finally found 14 miRNAs. For 210 better visualization, we established an miRNA-DNMT1 regulatory network using 211 Cytoscape software (https://cytoscape.org/) (Fig.4A). In theory, there should be a 212 negative correlation between miRNA and DNMT1 due to the action mechanism of 213 miRNA. Therefore, we performed expression correlation analysis for miRNA-DNMT1 214 pairs. As shown in Fig. 4B, DNMT1 was significantly and negatively associated with 215 hsa-miR-148a-3p and positively linked to the other 13 predicted miRNAs in HCC. Then, 216 we assessed the expression of hsa-miR-148a-3p in HCC and normal control samples 217 with starBase and evaluated the prognostic value of hsa-miR-148a-3p in HCC with a 218 Kaplan-Meier plotter. As shown in Fig. 4C and 4D, hsa-miR-148a-3p was significantly 219 downregulated in HCC, and its downregulation was correlated with poor prognosis. 220 Together, these findings suggest that hsa-miR-148a-3p might be the most influential 221 upstream miRNA of DNMT1 in HCC. 222 223

Prediction and analysis of upstream lncRNAs of hsa-miR-148a-3p 224
Next, we predicted the upstream lncRNAs of hsa-miR-148a-3p using the starBase 225 database and obtained a total of 45 possible lncRNAs. Similarly, for better visualization, 226 the lncRNA-hsa-miR-148a-3p regulatory network was established using Cytoscape 227 software (Fig. 5A). We then detected 1548 DEGs between TCGA-LIHC tumor samples 228 and normal tissues using the "limma" R package. A Venn diagram was finally created 229 showing 3 overlaps of the 1454 DEGs and 45 possible lncRNAs (Fig. 5B), indicating 230 that LINC01554, small nucleolar RNA host gene 3 (SNHG3), and H19 were the co-231 expressed differential lnRNAs found in both cohorts. As is presented in the volcano 232 plot in Fig. 5C, SNHG3 was the upregulated gene and LINC01554 or H19 was the 233 downregulated gene in HCC. Subsequently, the prognostic values of the SNHG3 in 234 HCC were assessed, which revealed that overexpressed SNHG3 indicated poor OS of 235 patients with HCC (Fig. 5D). Based on the above results, we next explored whether 236 SNHG3 could regulate hsa-miR-148a-3p expression as a ceRNA in HCC. According 237 to the ceRNA hypothesis, lncRNAs usually serve as ceRNAs by binding to miRNAs, 238 and the key qualified lncRNAs in the ceRNA subnet should be negatively associated 239 with miRNA and positively linked to mRNA at the same time. The expression 240 correlation between SNHG3 and hsa-miR-148a-3p or DNMT1 in HCC is shown in Fig.  241 5E-F. Furthermore, different expressions of DNMT1, has-miR-148a-3p, and SNHG3 242 were observed in normal and HCC tissues of different histologic grades (Fig. 5

DNMT1 expression was positively related to immune cell infiltration in HCC 252
Tumor-infiltrating immune cells are independent predictors of cancer survival. As  shows that DNMT1 expression was positively related to that of E2F1, E2F2, E2F3, 308 E2F4, E2F5, E2F6, E2F7, and E2F8 in HCC, with statistical significance. 309 310

Discussion 311
The occurrence and development of HCC is a complex, dynamic biological process that 312 involves genetic, epigenetic cell state, and microenvironment alterations. Clarifying the 313 molecular mechanism underlying HCC carcinogenesis may contribute to the 314 development of effective therapeutic targets or valuable prognostic biomarkers. 315 Accumulating evidence has shown that DNMT1 participates in the tumorigenesis and 316 progression of various human cancers, including HCC. However, to date, knowledge 317 about DNMT1 in HCC remains insufficient, and further research is needed. 318

319
This study was performed to identify the feasibility of DNMT1 as a promising 320 biomarker in HCC patients. We first performed pan-cancer analysis on the expression 321 of DNMT1 using TCGA database. Then, the survival analysis of DNMT1 in some of 322 the cancer types with statistical significance as analyzed above was conducted, 323 indicating increased expression of DNMT1 to be correlated with poor clinical outcomes 324 in HCC. Moreover, high DNMT1 expression was associated with histological grade. 325 This suggests that upregulated DNMT1 might be involved in malignant transformation, 326 which is in line with previously described results (14). 327

328
The ceRNA hypothesis consists of lncRNA mainly regulating mRNA through the 329 ceRNA regulatory mechanism and RNAs affecting each other's levels by competing 330 with a limited pool of miRNAs (21). It has been shown that DNMT1 can inhibit the 331 transcription of tumor-suppressive miRNAs in cancer progression by maintaining their 332 hypermethylation (22). In our study, we developed a lncRNA-miRNA-mRNA triple 333 regulatory network related to DNMT1 in HCC. Through candidate miRNA prediction 334 conducted by the prediction programs-PITA, RNA22, miRmap, microT, miRanda, 335 PicTar, and TargetScan-and correlation analysis-including expression analysis and 336 survival analysis-we finally identified has-miR-148a-3p as the most likely potential 337 upstream miRNA of DNMT1 in HCC. Has-miR-148a-3p is a member of the miR-338 148/152 family and has been reported to be a tumor suppressor for various human 339 cancers, including pancreatic cancer (23), esophageal cancer (24), and HCC (25). 340 Recent evidence has confirmed that the reciprocal negative regulation between hsa-341 miR-148a-3p and DNMT1 contributes to cell proliferation, cell cycle processes, and 342 maintaining cell stemness characteristics in HCC (26). 343 344 Next, upstream lncRNAs of has-miR-148a-3p/DNMT1 axis were also predicted, and 345 45 possible lncRNAs were found. We intersected these lncRNAs with 1548 DEGs 346 between TCGA-LIHC tumor samples and normal tissues to screen for differentially 347 expressed lncRNAs. In the end, on the basis of the volcano map, ceRNA hypothesis, 348 and correlation analyses, we identified SNHG3 as the upstream lncRNA. It has been 349 reported that SNHG3 is an oncogene in many kinds of malignancies (27). Meanwhile, 350 multiple miRNAs in HCC have been shown to promote tumor growth and metastasis 351 through targeting SNHG3 (28, 29). Considering these findings, we identified 352 SNHG3/hsa-miR-148a-3p/DNMT1 axis as the potential regulatory pathway in HCC. In summary, we established the SNHG3/hsa-miR-148a-3p/DNMT1 axis as the 390 potential regulatory pathway of hepatocarcinogenesis, which was also identified as a 391 biomarker of poor prognosis. We further found that DNMT1 might exert its oncogenic