UBQLN4 promotes progression of HCC via activating wnt-β-catenin pathway and is regulated by miR-370

Pan-cancer tissue microarray (Pan-cancer TMA) contained ten solid tumors (each tumor contained 20 pairs of tumor tissues and adjacent non-tumor tissues). Tissue microarray was performed on 341 HCC specimens, as previously described [15]. The tissues used in this study were acquired between January 2012 and December 2016 from the First Affiliated Hospital of Zhengzhou University, Zhengzhou University, China. Cancer Genome Atlas Project (TCGA; http://​tcga-data.​nci.​nih.​gov/​) was used to analyze the relationship between UBQLN4 expression and development of solid tumors, according to the methods reported previously [16].

Normal liver cells (Chang Liver and L02) and HCC cell lines (HepG2, SMMC-7721, SK-Hep-3B, HCC-LM3 and MHCC-97H) were procured from Sibcb (Shanghai, China). Chang Liver and L02 were cultured in 1640 supplied with 10% fetal bovine serum (FBS), whereas HCC cells were cultured in MEM supplied with 10% fetal bovine serum. The cells were incubated at 37 °C in a humidified atmosphere of 5% CO₂.

Human-specific shUBQLN4 (1–3) and negative control vectors were used to develop stable UBQLN4 suppression cell lines. Human UBQLN4 overexpression plasmid and negative control were procured from Gene Pharma (Shanghai, China). MiR-370 mimics, inhibitors, other miRNA mimics, and the controls (NC) were procured from Gene Pharma (Shanghai, China). The cell transfection procedure was conducted according to the manufacturers’ instructions.

Protein isolation and western blotting were conducted as per the methods described previously [17]. Samples were probed with UBQLN4 (1:1000), β-catenin (1:1000), CyclinD1 (1:1000), Axin2 (1:1000), Wnt5a (1:1000), C-myc (1:1000), Actin (1:5000) and GAPDH (1:5000) monoclonal antibodies. Real-time polymerase chain reaction (RT-PCR) was conducted according to the methods described previously [18].

Pan-cancer Tissue microarray (Pan-cancer TMA) and TMA containing 341 HCC specimens were used to conduct immunohistochemistry analysis according to the methods described previously [19]. Also, immunohistochemistry analysis was conducted using tissues of nude mice, including sh-Ctrl, sh1-UBQLN4, and sh2-UBQLN4. The detection proteins included Ki67, UBQLN4, β-catenin, CyclinD1, Axin2, Wnt5a, and C-myc.

Hep-G2 and SMMC-7721 cell lines were transfected with UBQLN4 shRNA1–2, and negative control (or miR-370 mimics, miR-370 mimics + UBQLN4 and negative control). Cell viability test, EDU assay and transmembrane invasion assay were conducted according to the methods described previously [20].

The SMMC-7721 cells transfected with shUBQLN4 or negative control were divided into two groups, one group was used for colony-forming assay in vitro, the other was used for subcutaneous tumor formation in nude mice. The number of clones in vitro, tumor volume and relative photon flux in vivo was detected and analyzed according to the methods described previously [16].

The gene set enrichment analysis (GSEA) was conducted to evaluate correlated pathways based on UBQLN4 expression in HCC tissues. A total of 377 samples from TCGA data set were divided into two groups (UBQLN4 ^high and UBQLN4 ^low). The specific bioinformatics analysis of GSEA v2.0 was conducted according to the methods described previously [15].

Human Hep-G2 and SMMC-7721 cell lines were transfected with UBQLN4 shRNA1–2 and negative control (or miR-370 mimics, miR-370 mimics and UBQLN4 and negative control). Next, cells were co-transfected with the Wnt/β-catenin signaling reporter TOPFlash/FOPFlash (Addgene, Cambridge, MA, USA). The reporter-driven luciferase activity was quantified by analyzing the luminescence intensity using a Synergen 2 (BioTek) reader. The results are represented as normalized TOPFlash/FOPFlash values.

To investigate the expression of UBQLN4 in common solid cancers, the TCGA database was used to compare mRNA levels of UBQLN4 between tumor tissues and non-tumor tissues. The result showed that UBQLN4 mRNA levels were low in most solid cancer tissues, including KIRC, KIRP, PRAD and UCEC (Fig. 1a). Pan-cancer TMAs containing 21 pairs of cancer and normal adjacent tissues was performed to verify these results at the protein level. Ubiquilin-4 was overexpressed in most solid tumor tissues except in renal carcinoma (Fig. 1b). The increase in UBQLN4 expression suggested that UBQLN4 might play an essential role in the initiation and progression of solid tumors, including HCC.

To investigate the role of UBQLN4 in HCC progression, TCGA and GEO database analysis were performed to determine the variation in UBQLN4 mRNA levels between HCC and normal tissues. Expression of UBQLN4 was higher in most HCC tissues than normal tissues (Fig. 2a). Besides, UBQLN4 expression was higher in HCC tissues than in cirrhosis liver, low-grade dysplastic liver and high-grade dysplastic liver tissues (Fig. 2b). Kaplan–Meier analysis of TCGA LIHC cohort showed that high expression of UBQLN4 was correlated with poor overall survival and disease-free survival rates of HCC patients (Fig. 2c, d). Moreover, patients with high expression of UBQLN4 in TNM I-II group were associated with poor overall survival rate. However, there was no significant difference in overall survival time among the patients in TNM III-IV group (Fig. 2e, f). High expression of UBQLN4 observed in the TNM III-IV group was associated with poor disease-free survival; however, a similar observation was not made in TNM I-II group (Fig. 2g, h). To verify these results, UBQLN4 protein levels in 8 pairs of liver cancer and normal adjacent tissues were evaluated. Ubiquilin-4 was highly expressed in HCC tissues (Fig. 3a). Besides, high expression of UBQLN4 was associated with poor TNM stage, poor histological grade, large tumor size, vascular invasion and poor overall survival time (Fig. 3b–e) (Tables 1, 2). Collectively, these results showed that UBQLN4 is upregulated in HCC tissues and its expression is correlated with poor prognosis of HCC patients.

Evaluation of UBQLN4 expression in normal liver cell lines (Chang liver and LO2) and HCC cell lines (HepG2, SMMC-7721, SK-Hep-3B, HCC-LM3 and MHCC-97H) was conducted to investigate the role of UBQLN4 in HCC cell progression. Ubiquilin-4 was upregulated in HCC cell lines, especially in HepG2 and SMMC-7721 (Fig. 4a, b). Subsequently, cell lines were transfected with UBQLN4 knockdown vectors, and the knockdown efficiency was verified (Fig. 4c). Suppression of UBQLN4 inhibited the proliferation and clonal formation abilities of HCC cells (Fig. 4d–f). Also, UBQILN4 suppression reduced the cell invasion ability of HCC cells (Fig. 4g). Collectively, these findings revealed that UBQLN4 was crucial for HCC cell proliferation and invasion.

Subcutaneous tumor formation analysis in nude mice was conducted to further verify the role of UBQLN4 in HCC progression in vivo. The shUBQLN4 group had smaller tumor volume and weaker relative photon flux than the negative control group (Fig. 5a–c). Next, expression levels of UBQLN4 and Ki67 were evaluated in tumor tissues of nude mice. Ubiquilin-4 suppression led to low levels of UBQLN4 in shUBQLN4 group, and significantly reduced the expression of Ki67 (Fig. 5d–f). Thus, these results demonstrated that UBQLN4 suppression might inhibit tumor formation ability of HCC cells in vivo.

A comprehensive bioinformatics analysis was performed based on TCGA database to investigate the mechanism underlying UBQLN4 roles in HCC progression. Gene Set Variation Analysis (GSVA) showed that UBQLN4 overexpression was correlated with the wnt-β-catenin pathway activation (Fig. 6a). Consistently, the Gene Set Enrichment Analysis (GSEA) further revealed the extensive positive correlation between UBQLN4 overexpression and Wnt pathway-related gene expressions (Fig. 6b, c). Wnt pathway and UBQLN4 expressions were significantly associated with stemness of solid tumors and signature genes, respectively (Fig. 6d, e). Furthermore, the results of the TOP/FOP experiment showed that UBQLN4 suppression significantly inhibited wnt-β-catenin pathway activation (Fig. 6f). Besides, suppression of UBQLN4 inhibited the expression of downstream genes involved in the wnt pathway, including β-catenin, CyclinD1, Axin2, Wnt5a and C-my (Fig. 6g–i). Mechanistically, UBQLN4 silencing impeded HCC cell growth, whereas reintroduction of β-catenin significantly reversed the inhibitory effects of UBQLN4 knockdown (Fig. 6j, k). Based on these findings, we verified that UBQLN4 might regulate proliferation and invasion abilities of HCC cells by affecting the activation of the wnt-β-catenin pathway.

Target Scan, miRTarBase, miRDB and miRcode were performed to investigate further the upstream regulatory gene of UBQLN4. Seven candidate miRNAs were complementary with 3′UTR of UBQLN4. Subsequently, miRNA mimics were used to suppress UBQLN4 expression in HCC cells. MiR-370 expression significantly reduced the UBQLN4 mRNA levels. Furthermore, luciferase experiment showed that miRNA-370 induced the degradation of UBQLN4 (Fig. 7a, b). MiR-370 mimics and inhibitors were co-cultivated with HCC cells to confirm the results further. MiR-370 mimics decreased mRNA and protein levels of UBQLN4 whereas miR-370 inhibitors enhanced UBQLN4 overexpression (Fig. 7c, d). Also, the expression of UBQLN4 was found to be negatively correlated with the miR-370 expression, as shown by TCGA database analysis (Fig. 7e).

Ubiquilin-4 overexpression plasmid was used to verify further the relationship between UBQLN4 and miR-370 in HCC progression. The plasmid significantly reversed the inhibitory effect of miR-370 on UBQLN4 expression in HCC cells (Fig. 8a). Through the following cell function experiments, we observed that miR-370 expression inhibited the viability, clonal formation and invasion abilities of HCC cells, whereas overexpression of UBQLN4 partially reversed tumor suppressive functions of miR-370 (Fig. 8b–d). Moreover, results of TOP/FOP experiments showed that overexpression of UBQLN4 reduced the inhibitory effect of miR-370 on wnt signaling pathway. And this phenomenon was directly verified by a change in expression levels of wnt pathway associated genes, such as β-catenin, CyclinD1 and Axin2 (Fig. 8e). Collectively, the results of the present study showed that UBQLN4 overexpression could partially reverse the anti-tumor effects of miR-370 in HCC cells.