The expression levels of miR-330-5p in HCC cells and tissues have been documented by several research groups. However, the results have been inconsistent. In this study, combining in-house real time RT-qPCR, miRNA-sequencing, miRNA-microarray, and in-silico integrated analysis, we noted that miR-330-5p levels were apparently upregulated in HCC tissues based on 1,095 cases from multiple research centers. More interestingly, we also pointed out that miR-330-5p may accelerate hepatocarcinogenesis via metabolic pathways, cancer pathways, and the signaling pathways of alanine, aspartate, and glutamate metabolism.
Contradictory miR-330-5p levels have been found in HCC cells and tissues by different groups. Downregulation of miR-330-5p was observed in HCC HepG2, HuH-7, Bel-7402, and SMMC-772 cells, compared to L-02 (24). Two studies showed that ectopic miR-330-5p had a protective function to suppress the cell growth of HCC in vitro and in vivo. The restoration of miR-330-5p could also suppress the migration, invasion, and angiogenesis capacity of the HCC cells (24, 29). Hence, the above studies concluded that miR-330-5p may play a proactive role in the HCC process, as the increase of miR-330-5p could slow down or even block HCC development. However, higher expression levels of miR-330-5p have also been detected in HCC tissues and cells by three independent groups (26–28). Among these three studies, 35 pairs of HCC and non-HCC cases from TCGA were detected by RNA-sequencing and miRNA-microarray (26–28). More importantly, two groups also verified the oncogenic role of miR-330-5p in HCC through in vitro and in vivo experiments (27, 28). Overexpression of miR-330-5p by transfection of miRNA mimics could assist the cell proliferation of HCC and the formation of HCC tumors, as well as promote cell infiltration ability (27, 28).
In the previous studies, The insufficient sample size influenced the accuracy of the above incompatible results. To have a comprehensive view of the clinicopathological value of miR-330-5p levels in HCC, the current study combined various detecting methods, including in-house real time RT-qPCR, miRNA-sequencing, miRNA-microarray, and SMD and sROC calculation. The RT-qPCR with the clinical samples showed a significantly higher level of miR-330-5p in 26 cases of HCC, with the relative expression of 7.51, 2.52 folds from the non-cancerous controls. To verify this finding, we mined the data from high-throughput datasets. Upregulation of miR-330-5p levels was consistently noted, as the SMD was 0.3 for 1,095 cases of HCC. With the integration of all available data, the final SMD increased to 0.44, which was comparable to the reports of previous studies (26–28). Hence, together with previous work, the current findings confirm a marked increase in miR-330-5p expression levels in HCC tissues, compared to non-HCC liver tissues. This upregulation may play an oncogenic function in the tumorigenesis of HCC.
In previous studies, conflicting results existed in the expression of miR-330-5p in HCC tissues. Beyond that, the association between miR-330-5p and the clinical parameters of HCC were contradictory. In the studies in which miR-330-5p was down-regulated in HCC, the expression value of miR-330-5p was negatively related to the progression parameters of HCC. The higher the TNM stage of HCC, the lower the expression value of miR-330-5p was (24, 29). However, the expression of miR-330-5p was positively associated with the TNM stage of HCC patients in the studies where miR-330-5p was up-regulated in HCC; in these studies, the higher expression of miR-330-5p was connected with poorer prognosis of HCC patients (26–28). Although this contradiction existed in multiple studies, the researchers only used their own clinical data, including TNM stage and prognostic data; public data were not used in their studies. This lack of independent cohort verification was likely to be the cause of the contradictory results. In addition, Yu et al. reported that the expression of miR-330-5p in lower migrating HCC exosomes was greater than in higher ones (25).
To obtain a comprehensive conclusion, clinicopathological parameters and prognostic data were collected from TCGA and GEO databases. In the TCGA data, the expression of miR-330-5p had no relationship with the TNM stage. However, there was a positive correlation between miR-330-5p levels and TNM stage in our PCR data. Because the TNM stage information existed in some GEO data, a meta-analysis was computed based on TCGA, GEO, and RT-qPCR data. The result showed that a higher expression of miR-330-5p was correlated with a more advanced TNM stage. Moreover, the AUC of miR-330-5p to differentiate advanced TNM stages from early stages showed that the specificity was 0.94, meaning the expression of miR-330-5p could be considered an indicator for tumor progression. Neither the included GEO data nor our PCR data had prognostic information, so the prognostic analysis of miR-330-5p in HCC was performed based on TCGA data. The results showed that miR-330-5p was closely related to patients’ overall survival in HCC, and high expression of miR-330-5p led to poorer prognosis, but the expression of miR-330-5p did not affect patients’ RFS. Previously, a relationship between miR-330-5p and RFS of HCC patients had been reported, but the literature used only its own data (27) and no public data. Hence, according to TCGA data and previous studies (27, 28), the overexpression of miR-330-5p could cause poorer overall survival of HCC patients.
The molecular function of miR-330-5p in HCC was still not clear, so we used various bioinformatics platforms to inquire into the putative molecular mechanism of miR-330-5p in HCC. First, the target genes of miR-330-5p were forecasted via miRWalk2.0, and the differential expression genes were calculated via 63 high-throughput RNA-sequencing and microarrays. All of the following operations were based on the intersection genes of the two gene sets. The result of KEGG pathway analysis revealed that miR-330-5p targeted genes might be involved in alanine, aspartate, and glutamate metabolism pathways, metabolic pathways, and pathways in cancer. In the top three pathways, there were two pathways associated with metabolism, particularly in glutamate metabolism. Numerous reports had indicated that the glutamate metabolic pathway was extremely important in the development of cancer. Glutamine can be broken down into glutamate, and glutamate can be further decomposed into α-ketoglutarate to promote the TCA cycle (44–46). Thus, miR-330-5p may play an essential part in liver cancer; miR-330-5p may affect the occurrence and development of liver cancer by targeting related genes of glutamate metabolism to change the metabolic pathway. To further reveal this phenomenon, the hub genes (GLUD1,GOT1,GLS2) of alanine, aspartate, and glutamate metabolism pathways were screened. GLUD1, glutamate dehydrogenase 1, converted glutamate to alpha-ketoglutarate to promote the TCA cycle (47). Glazer et al. found that the expression of GLUD1 in HCC tissues was lower than in non-tumorous liver tissues; combined with our study, we could conclude that targeting of GLUD1 by miR-330-5p led to low expression of GLUD1 in HCC (48). Similarly, as a glutamate metabolism-related gene, GOT1 (glutamic-oxaloacetic transaminase 1), also called AST1, is an extremely important indicator of liver function and could also convert glutamate to alpha-ketoglutarate (49, 50). Nwosu et al. reported that the expression of GOT1 was down-regulated in poorly differentiated hepatocellular carcinoma cell lines, and the researchers found that GLUD1 was also down-regulated in HCC cell lines (51). GLS2 (Glutaminase 2) is located on human chromosome 12, and it mainly encodes glutaminase (GA) which converts glutamine into glutamate in the liver (52). The expression of GLS2 in HCC was evidentiarily lower than in normal liver cells, and the high expression of GLS2 could suppress the growth of HCC cells (47). GLS2 could also be regarded as a molecular marker to assess the outcome of HCC patients (53). These results show that all three hub genes could encode key enzymes in glutamine and glutamate metabolic pathways, and all three genes had low expression in HCC. Therefore, the highly expressed miR-330-5p was highly likely to target these three hub genes to form the regulatory axis of miR-330-5p-GLUD1/GOT1/GLS2.
There were some shortcomings in our study. First, although the three key genes (GLUD1, GOT1, GLS2) were highly likely to be the target genes of miR-330-5p, a dual-luciferase experimental confirmation was still deficient. Second, in vivo and in vitro work must still be conducted to assess the regulatory axis of miR-330-5p–GLUD1/GOT1/GLS2.