Identification of the down-regulated lncRNAs in HCC tissues
A total of 15 down-regulated lncRNAs were screened out in Hepatocellular Carcinoma tissues by GEPIA (http://gepia.cancer-pku.cn, FC>2, Padj<0.01, supplementary Table 1). Next, K-M survival analysis suggested that five lincRNAs (LINC01554, LINC01093, LINC01018, LINC01370, LINC00238) were associated with overall survival of patients after surgery by Median expression level of each lincRNA (supplementary Table 1 and supplementary Fig.1, P < 0.1). LINC01554 [7, 8], LINC01093 [9, 10], LINC01018 [11, 12] had been reported in Hepatocellular Carcinoma, whereas LINC01370 and LINC00238 had not. As shown in the Fig 1A&B, the expression of LINC00238 was significantly decreased in the HCC tissues compared with normal tissues from the analysis by GEPIA website. In addition, as indicated in the Fig 1C, LINC00238 was also down-regulated in 91.8% (45/49, P<0.0001) HCC tissues by paired student’s t-test analysis of matched and normal tissues from TCGA dataset. Kaplan-Meier (K-M) survival curves indicated that patients with high LINC00238 expression showed longer overall survival time (Figure 1D, median survival 80 months versus 50 months, hazard ratio for death = 0.74, P = 0.099).
Low levels of LINC00238 expression in HCC tissues correlated with bigger tumor size, early recurrence and poor survival of patients after surgery
To validate these results from TCGA, we qualified LINC00238 expression in our own 90 cases of HCC cohort. Consistently, LINC00238 expression was decreased in 83.33% (75/90) HCC tissues compared with adjacent tissues (Figure 1E&F, P < 0.001). To facilitate analysis, we divide patients into low and high LINC00238 group according to its median expression level in HCC tissues. As shown in Table I, low expression of LINC00238 was associated with bigger tumor size, early recurrence and shorter survival time. The proportion of low LINC00238 expression was higher in the tumor tissues of patients with bigger tumor size (56.92% vs 43.08%, c2 test, P=0.034), higher ratio of early recurrence within 2 years after surgery (66.67% versus 33.33%, c2 test, P=0.003), higher N stage (c2 test, P=0.011) and less overall survival time of 4 years (c2 test, P=0.030), indicating that low LINC00238 expression may be associated with tumor progression. However, there was no association between LINC00238 expression and other clinicopathological features, including sex, age, hepatic cirrhosis and venous invasion (c2 test, P>0.05, Table I). We then assessed the association between LINC00238 expression and OS of patients with HCC. Consistent with the above results from GEPIA website, Kaplan-Meier survival analysis revealed that patients with low LINC00238 expression exhibited a significantly shorter OS （Median 47.4 vs 62.4 months, P=0.050）and DFS (Median 12.7 vs 60.0 months, P=0.026) compared with those with high expression (Fig. 1G&H).
LINC00238 overexpression impairs HCC malignant phenotype in vitro
We then quantified LINC00238 expression in on human normal hepatic L02 cells and HCC cells, including HepG2, SMMC-7721, PLC/PRF/5, BEL-7402 and Huh7. The LINC00238 was significantly down-expressed in all HCC cells compared to L02 cells (Fig. 2A). We chose Huh7 cells (relatively low LINC00238 expression) for the overexpression experiments and HepG2 cells (relatively high LINC00238 expression) for the knockdown experiments. Compared to control cells, LINC00238 expression was increased by 24 fold times after overexpression as detected qPCR (Fig. 2B). Forced expression of LINC00238 suppressed Huh7 cell viability (Fig. 2C) and colony formation (Fig. 2D, about 47.9% inhibition). In addition, forced expression of LINC00238 significantly decreased the migratory and invasive ability of the Huh7 cells, as determined by Transwell assay (Fig. 2E, about 40.0% inhibition for cell migration and 36.9% inhibition for cell invasion). These in vitro findings suggested that LINC00238 might act as tumor suppressor in HCC.
Knockdown of LINC00238 promotes HCC malignant phenotype in vitro
We knocked down the expression of LINC00238 in HepG2 cells using three shRNAs targeting different sites. Three shRNAs had 82%, 88% and 72% knockdown efficiency respectively (Fig. 3A), and two silencers, shRNA1 and shRNA2, were choose in subsequent experiments. The CCK-8 assay and plate colony formation assay results demonstrated that LINC00238 depletion promoted cell viability (Fig. 3B) and colony formation ability of HepG2 (Fig. 3C). Furthermore, Transwell assay results showed that inhibition of LINC00238 significantly enhanced the migration and invasive ability of Hep2 cells with the two shRNAs (Fig. 3D).
LINC00238 functioned as tumor suppressor in vivo
To further confirm the suppression effect in HCC by LINC00238 in vivo, we performed a modified chick embryo chorioallantoic membrane (CAM) assay to assess the tumor growth and metastasis. Cells labeled with red fluorescence were inoculated onto CAM to monitor tumor growth and distance metastasis to lung tissues of chick embryo. Compared to the control cells, overexpression of LINC00238 resulted in a decrease of tumor weight by 64.8% (Fig. 4A). Moreover, the number of metastatic tumor colonies, which represented metastasis ability to lung tissues, were decreased by 80.9% in cells after LINC00238 over-expression (Fig. 2B). Consistently, knocking down of LINC00238 expression resulted in a significant increase in not only the tumor weight but also the extent of metastasis to the lungs (Fig. 3C and 3D) by CAM assay. Notably, metastatic nodules in lung tissues of the LINC00238 shRNA groups were significantly more than those of the scramble group (Fig. 3D). Hence, the reciprocal effects of LINC00238 overexpression and knockdown both in vitro and in vivo supported that LINC00238 played a tumor suppressor in HCC.
LINC00238 sponges miR-522 to regulate malignant phenotype in HCC
Cytoplasm lncRNAs can regulate cell phenotype by sponging miRNAs. The prediction by lncLocator (http://www.csbio.sjtu.edu.cn/cgi-bin/lncLocator.py) suggested that LINC00238 is predominantly located in the cytoplasm. Consistently, the nucleus cytoplasm separation experiments confirmed that LINC00238 distributed mainly in the cytoplasm (Figure 5A), which suggested that LINC00238 might sponge miRNAs. Bioinformatics software predicted that the seed regions of miR-522 was complementary with three isotypes of LINC00238 (Figure 5B). Furthermore, the expression of miR-522 showed decreased in LINC00238 over-expression Huh7 cells, but increased in the LINC00238 knock-down HepG2 cells (Figure 5C). Then, RNA pull-down experiments suggested that expression of miR-522 was enriched about 8.3 times by LINC00238 compared with control group (Figure 5D). In addition, the expression of LINC00238 in the biotin-miR-522 group was about 5.8 times that of the negative control group (Figure 5E). We constructed wild-type and mutant versions of LINC00238 downstream of the luciferase gene using PGL3-control vector (Figure 5G). By a dual-luciferase report assay, overexpression of miR-522 was associated with a decrease of luciferase activity of LINC00238 wild type of vector, but had no effect on the mutant type of vector (Figure 5H). Previously, it had reported that miR-522 contributes to cell proliferation of hepatocellular carcinoma by targeting two Wnt signaling inhibitors, DKK1 and SFRP2 . We therefore confirmed that whether LINC00238 release the inhibition effect on these target genes of miR-522. Consistent with the hypothesis, SFRP2 and DKK1 were increased in both Huh7 cells and tumors after over-expression of LINC00238, but decreased in HepG2 cells and tumors after interference expression of LINC00238. Taken together, LINC00238 might regulate malignant phenotype in HCC by sponging miR-522 followed by release the inhibition of down-stream targets of miR-522.
Rescue expression miR-522 reverse suppression effects of LINC00238 partially
To corroborate the key role of miR-522, we performed a gain-of-function experiment in Huh7-LINC00238 cells by using miR-522 mimics (Fig. 6A). The expression of miR-522 was enhanced by 186 times after transfection with mimics. Consistent with our prediction and previous report , rescue of miR-522 partially reversed the inhibitory effects of LINC00238 on Huh7 cell viability (Fig. 6B), plate colony formation ability (Fig. 6C), migration and invasion ability (Fig. 6D), tumor growth (Fig. 6E) and metastasis to the lungs (Fig. 6F) in the CAM assay. Two known downstream targets of miR-522，SFRP2 and DKK1, was inhibited in both Huh7-LINC00238 cells and tumors after transfection with mimics. These results consolidated that LINC00238 functioned as tumor suppressor through sequestering miR-522, thereby relieving the inhibition of SFRP2 and DKK1at least partially.