The tumor suppressor role and prognostic value of PTEN in PRAD
According to HPA database analysis, PTEN was over-expressed in normal prostate tissue, but down-expressed in PRAD tissue (Figures 1A and 1B). Combined with the IHC from the HPA, the same conclusion was confirmed (Figure 1G). Through survival analysis, we also found that low expression of PTEN was associated with poor overall survival (OS) in PRAD patients cohort (Figure 1C). Given the above situation, in order to further explore the potential mechanism about abnormally low expression of PTEN in PRAD, the genome and copy number of PENT were analyzed. cBioPortal analysis indicated the deletion of PTEN in the TCGA PRAD dataset (Figure 1F). Further studies showed that over one-third of PRAD samples had PTEN deletion, and their mRNA expression level was significantly lower than that exhibited diploid PTEN (Figure 1E). In addition, we also found that the copy number of PTEN was positively correlated with the mRNA expression level (Figure 1F). Therefore, we believe that the deletion of PTEN copy number in PRAD may be related to the down-expressed of its expression level.
Screening results of DElncRNAs, DEmiRNAs and DEmRNAs
Based on the above analysis, we thought that the ceRNA network associated with PTEN might be a potential prognostic model in PRAD patients. Therefore, DElncRNAs, DEmiRNAs and DEmRNAs were identified in PRAD samples with the PTENhigh and PTENlow expression groups by TCGA database. Then, a total of 1918 DElncRNAs, 68 DEmiRNAs and 2920 DEmRNAs were screened (Figure 2A-C). Finally, the expression of 15 significant DERNAs were subjected to two-way cluster analysis, and the corresponding heat map is shown in Figure 2D–3F.
Construction of ceRNA regulatory network
The target mRNA set for DEmiRNAs was predicted by miRTarBase, MiRDB and TargetScan. Mircode and TargetScan database were used to predict the target lncRNA set for DEmiRNAs. In total, 8 lncRNAs, 7 miRNAs and 525 mRNAs in PRAD were selected to construct the ceRNA regulatory network (Figure 3A). Finally, the hub ceRNA regulatory network with three lncRNAs (MEGS, AGAP11, and ADAMTS9-AS2), five miRNAs (miR-93-3p, miR-141-3p, miR-222-3p, miR-590-3p and miR-3619-5a), and six mRNAs (NRG1, SHH, TIMP3, CLVS2, NFAT5 and ADAM22) was constructed and visualized with the help of Cytoscape plug-in cytoHubba were identified.(Figure 3B).
Enrichment analysis results of DEmRNAs
Enrichment analysis of 525 DEmRNAs was carried out by Metascape. The results of GO and KEGG showed the genes were primarily involved in chemotaxis, urogenital system development and cell morphogenesis involved in differentiation (Figure 3C).
Results of analysis between the prognosis and expression of PRAD and hub-DElncRNAs, DEmiRNAs and DEmRNAs in ceRNA network
Firstly, the expression levels of RNAs was analyzed, which from the hub ceRNA with PTENhigh and PTENlow expression groups and in PRAD and normal prostate tissues. Three up-regulated lncRNAs (MEGS, AGAP11, and ADAMTS9-AS2), one up-regulated (miR-222-3p) and two down-regulated (miR-590-3p, miR-141-3p) miRNAs and six up-regulated mRNAs (NRG1, SHH, TIMP3, CLVS2, NFAT5 and ADAM22) in PRAD samples was found with PTENhigh groups (Figure 4A). The three down-regulated lncRNAs (MEGS, AGAP11, and ADAMTS9-AS2), two up-regulated (miR-590-3p, miR-141-3p) and one down-regulated miRNAs (miR-222-3p), and six down-regulated mRNAs (NRG1, SHH, TIMP3, CLVS2, NFAT5 and ADAM22) were found in PRAD (Figure 4B). Kaplan Meier analysis was used to further explore the correlation between DERNAs in hub-ceRNA regulatory network and the prognosis of PRAD. The results showed that one DElncRNA (MEG3), one DEmiRNAs (miR-222-3p) and two DEmRNAs (NRG1 and CLVS2) were correlated with the prognosis of PRAD (Figure 5 and Figure 6).
The lncLocator was used to analyze the subcellular localization of MEG3, AGAP11, and ADAMTS9-AS2. As shown in Figure 7A-C, only MEG3 is mainly located in the cytoplasm. These results indicated MEG3, as a ceRNA, might enhance NRG1 and CLVS2 expression by spongy miR-222-3p. In addition, expression correlation analysis indicated that there was a positive correlation between the expression of MEG3, miR-220-3p, NRG1, CLVS2 and PTEN (Figure 7D-G). Therefore, the PTEN-related MEG3-miR-220-3p-NRG1/CLVS2 ceRNA network was constructed.
Relationship between NRG1 and CLVS2 expression and methylation
UALCAN analysis indicated that the methylation degree of CLVS2 in normal prostate tissue was lower than that in PRAD (Figure 8C-D). In addition, methylation sites related to the expression level of NRG1/CLVS2 in their DNA sequences were found (Figure 8A-B). Normalized mRNA expression data with corresponding patients’ information was collected in TCGA-PRAD. The differential expression of methylation of m6A related genes in PRAD tissues with high and low NRG1 and CLVS2 was shown in Figure 9A-B. It was suggested the expression of NRG1 and CLVS2 in prostate cancer may be regulated by methylation modification.
Relationship between the expression of NRG1, CLVS2 and immune cell infiltration
Timer database analysis showed that the expression level of NRG1 and CLVS2 in TCGA PRAD was significantly positively correlated with the infiltration level of 6 immune cells (Figure 10). These results suggest that NRG1 and CLVS2 may regulate the occurrence and development of prostate cancer by affecting immune cell infiltration.
Expression of NRG1 and CLVS2 protein in prostate cancer
As shown in Figure 11, the expression of NRG1 and CLVS2 protein was localized in the cell membrane, some nuclei and cytoplasm. The statistical results show that the comprehensive score of NRG1 and CLVS2 in prostate cancer were 4 and 5, respectively. All reached the positive expression standard.