Monoacylglycerol lipase promotes progression of hepatocellular carcinoma via NF-κB-mediated epithelial-mesenchymal transition

To investigate the effects of MAGL in HCC, we first detected MAGL expression in 27 tumor samples by immunohistochemical assay and found that MAGL expression was obviously elevated in HCC tissues compared with paired peritumor tissues (Fig. 1a). Consistently, western blot and qRT-PCR analysis of the indicated HCC patients demonstrated that the average expression of MAGL on both protein and mRNA levels were significantly higher in HCC tissues than peritumor tissues (Fig. 1b, c). Interestingly, immunohistochemical analysis for a large cohort of HCC samples (n = 170) showed that the expression of MAGL was significantly elevated in HCC patients with recurrence than those without recurrence (P = 0.001; Fig. 1d), which indicated a potential role of MAGL in HCC progression.

To explore the clinical significance of MAGL expression in HCC, we further assessed the relationship between MAGL expression and the clinicopathologic characteristics in a TMA of 170 HCC patients (Table 1). Immunohistochemical analysis showed that the average intensity of MAGL expression in tumor was significantly higher than that in corresponding peritumor tissues. Notably, elevated MAGL expression was identified to be associated with larger tumor size (P = 0.048), microvascular invasion (P = 0.026), poor tumor differentiation (P = 0.012), and advanced TNM stage (P = 0.001). Whereas the other clinicopathologic characteristics, including age, gender, HbsAg, HCV, liver cirrhosis, alpha-fetoprotein, tumor number, or tumor encapsulation, showed no correlation with the expression level of MAGL in HCC.

In the present study, all 170 HCC patients were dichotomized as MAGL^low expression (n = 77) and MAGL^high expression (n = 93), with MAGL^high expression accounting for 54.7% (93 of 170). Moreover, patients in the MAGL^high expression group exhibited shorter OS than those in the MAGL^low expression group (P = 0.003; Fig. 1e). Consistently, the 1-, 3-, and 5-year OS rates after surgery were much worse in the MAGL^high expression group than those in the MAGL^low expression group (66.7 vs. 80.5%, 32.5 vs. 57.7%, 29.7 vs. 47.8%, respectively). In addition, the TTR in the MAGL^low expression group was significantly lower than that in the MAGL^high expression group (P = 0.001; Fig. 1f), and the 1-, 3-, 5-year TTR rates were significantly higher in the MAGL^high group than in the MAGL^low group (13.4 vs. 9.4%, 66.7 vs. 44.6%, 87.4 vs. 58.5%, respectively). Univariate and multivariate analysis indicated that MAGL was an independent prognostic factor for both OS (HR = 1.628, P < 0.05) and TTR (HR = 1.593, P < 0.05; Tables 2 and 3). Thus, these data clearly revealed that MAGL is a valuable predictive factor for clinical outcome of HCC.

To investigate the exact biological role of MAGL in HCC, we evaluated MAGL expression in various HCC cell lines (HepG2, SMMC7721, Huh7, MHCC97L, MHCC97H, and HCCLM3) and one normal liver cell line (L0-2). Data showed that the protein and mRNA expression of MAGL were significantly upregulated in HCC cell lines compared to in L0-2 cells (P < 0.05; Fig. 2a, b). And the expression of MAGL was enhanced in parallel with the increase of the metastatic potential of HCC cells, with the lowest level in HepG2 cells and the highest level in HCCLM3 cells. Then, MAGL expression in HepG2, a MAGL^low HCC cell line, was successfully upregulated (HepG2-MAGL). Meanwhile, in HCCLM3, a MAGL^high HCC cell line was stably downregulated (HCCLM3-shMAGL) (Fig. 2c, d). Our data showed that the growth of HepG2-MAGL cells in vivo was significantly increased (3350.0 ± 250.0 vs. 2076.7 ± 302.7 mm³, P < 0.01), whereas the growth of HCCLM3-shMAGL cells was markedly restrained (626.7 ± 157.0 vs. 1756.7 ± 172.1 mm³, P < 0.01), compared to their respective controls (HepG2-vector and HCCLM3-vector) (Fig. 2e). Moreover, the wound healing and transwell Matrigel invasion assays demonstrated that upregulation of MAGL in HepG2 cells promoted its migratory and invasive capacities, while HCCLM3-shMAGL cells with decreased MAGL levels exhibited reduced motility and invasiveness, compared with their respective controls (HepG2-vector and HCCLM3-vector) (Fig. 2f, g). Evidently, our results revealed that MAGL plays an important role in enhancing proliferative and invasive abilities of HCC cells in vivo and in vitro.

To determine whether EMT contributes to the increased HCC cell growth and invasion induced by MAGL, we first identified the cellular morphology of HCC cell lines with different MAGL expression levels (HepG2-vector, HepG2-MAGL, HCCLM3-vector, and HCCLM3-shMAGL). Results showed a significant difference in morphology of the indicated cell lines (Fig. 3a). In this study, HCC cells with higher MAGL levels (HepG2-MAGL and HCCLM3-vector) exhibited a typical mesenchymal appearance which was more dispersed and presented a spindle-like morphology, while HepG2-vector and HCCLM3-shMAGL cells showed a distinct epithelial morphology which took on a clustered and cobblestone-like appearance. These results suggested that MAGL might be closely correlated with EMT process in HCC. To confirm whether EMT is essential in the enhanced invasiveness of HCC cells mediated by MAGL, we explored EMT markers by western blot and qRT-PCR analysis in the indicated HCC cells. We found that E-cadherin expression was markedly downregulated in HepG2-MAGL and HCCLM3-vector cells, while mesenchymal markers, including N-cadherin and the key EMT regulator Snail, were obviously upregulated compared with their respective controls (HepG2-vector and HCCLM3-shMAGL) (Fig. 3b–d and f). However, no significant differences were observed in other EMT markers such as vimentin and twist between groups with different MAGL levels (Fig. 3b, e, and g). Consistently, this phenomenon has also been observed in subcutaneous tumor samples by immunohistochemistry assay. Our data revealed an obvious reduction of E-cadherin expression in HepG2-MAGL group, with markedly upregulated expression of MAGL, N-cadherin, and Snail, compared to HepG2-vector group (Fig. 3h). Collectively, our data verified that MAGL overexpression might help HCC cells acquire EMT-like biochemical traits, which may contribute to MAGL-induced HCC cell growth and invasiveness.

It is increasingly appreciated that NF-κB signaling pathway plays a vital role in the tumor EMT process [20‐22]. To uncover how MAGL affects the process of EMT, we assessed the activity of EMT-related molecules by western blot analysis in HCC cells. We found that the expression of NF-κB p65 was markedly increased in HepG2-MAGL cells but significantly reduced in HCCLM3-shMAGL cells compared to their respective controls (Fig. 4a, b), suggesting that NF-κB signaling might be responsible for MAGL-regulated HCC progression. To identify this hypothesis, we treated HepG2-MAGL cells with MAGL-inhibitor JZL184 for various time periods and found the total protein expression of NF-κB p65 was downregulated in a time-dependent manner (Fig. 4c). Consistently, the expression of phosphorylated NF-κB p65 was also reduced in a time-dependent manner when treated with JZL184 (Fig. 4d), suggesting that MAGL could affect the activation of NF-κB signaling in HCC cells. Furthermore, our data showed that the level of NF-κB p65 protein in the nucleus was detected to be downregulated in a time-dependent manner, whereas this phenomenon was not observed in cytoplasm (Fig. 4e, f), suggesting that MAGL-inhibitor JZL184 could inhibit the translocation of NF-κB p65 protein into the nucleus, thus decreased the total expression level of NF-κB p65. Evidently, our results clearly displayed that MAGL-inhibitor JZL184 could inhibit upregulation, phosphorylation, and nuclear translocation of NF-kB p65 in HCC cells, which means that MAGL overexpression could promote the activity of NF-kB p65 in HCC cells.

To identify whether NF-κB pathway participated in the effects of MAGL on HCC cells, we blockaded the activation of NF-κB p65 using NF-κB p65-shRNA in HepG2-MAGL cells (P < 0.001; Fig. 5a, b). As expected, we found that the increased expression of Snail, a key EMT regulator, could markedly be reversed at both protein and mRNA levels by NF-κB p65-shRNA treatment (P < 0.01; Fig. 5a, c), while the enhanced expression levels of MAGL displayed no obvious changes (Fig. 5a, d). Furthermore, the wound healing and transwell Matrigel invasion assays showed that downregulation of NF-kB p65 in HepG2 cells with high MAGL expression could significantly inhibited MAGL-induced motility and invasiveness of HCC cells (P < 0.01; Fig. 5e, f). Collectively, these results suggested that NF-κB pathway was involved in MAGL-mediated EMT and contributed to the effects of HCC cells induced by MAGL.