探讨circZKSCAN1抑制肝癌细胞侵袭转移的机制。

利用前期构建的差异表达circZKSCAN1基因的HepG2肝癌细胞进行实验。采用克隆形成实验、划痕实验、Transwell细胞迁移和侵袭实验分别检测OE组（过表达circZKSCAN1的HepG2）、SH-1和SH-2组（沉默circZKSCAN1的HepG2）、NC组（阴性对照的HepG2）肝癌细胞的迁移和侵袭能力。采用生物信息学方法分析circZKSCAN1上富集RNA结合蛋白（RBP）的结合位点，并通过RNA免疫共沉淀（RIP）和RNA pull-down实验验证其靶标RBP，分析靶标RBP与肝癌生存预后的关系。4组克隆形成数、迁移和穿膜细胞数比较采用单因素方差分析和LSD-t检验。两组富集倍数比较采用t检验。生存分析采用Kaplan-Meier法和Log-rank检验。

克隆形成实验显示，NC组、OE组、SH-1组、SH-2组克隆形成数分别为（10.5±1.1）、（4.7±0.8）、（21.7±1.8）、（22.6±2.6）个，与NC组相比，OE组克隆形成数明显减少，而SH-1组、SH-2组克隆形成数明显增多（LSD-t=-7.386，9.196，7.424；P<0.05）。划痕实验结果显示，与NC组相比，OE组24、48 h的迁移能力明显下降，而SH-1组、SH-2组24、48h的迁移能力则明显升高。Transwell细胞迁移实验表明，NC组、OE组、SH-1组、SH-2组迁移细胞数分别为（28.5±4.3）、（14.3±2.6）、（47.2±7.7）、（49.6±8.0）个，与NC组相比，OE组迁移细胞数明显减少，而SH-1组、SH-2组迁移细胞数明显增多（LSD-t=-4.895，3.673，3.474；P<0.05）。Transwell细胞侵袭实验显示，NC组、OE组、SH-1组、SH-2组穿膜细胞数分别为（21.3±2.1）、（7.6±1.5）、（38.6±5.8）、（41.3±7.1）个，与NC组相比，OE组穿膜细胞数明显减少，而SH-1组、SH-2组穿膜细胞数明显增多（LSD-t=-9.195，4.858，4.679；P<0.05） 。生物信息学分析显示，circZKSCAN1上富集RBP结合位点包括EIF4A3、FMR1、hnRNPC、U2AF65等。RIP实验显示circZKSCAN1可将细胞内的EIF4A3显著富集，与NC组相比，OE组富集倍数为（7.82±0.25）倍（t=1.057，P<0.05）。RNA pull-down实验显示EIF4A3可与circZKSCAN1结合。在线网站（http://gepia.cancer-pku.cn）分析显示，与癌旁组织相比，EIF4A3基因在肝癌组织中高表达，并主要出现在较晚分期的肝癌中（F=7.72，P<0.05），而高表达EIF4A3基因的肝癌患者具有更差的总体生存率（HR=1.90，P<0.05）。

EIF4A3是circZKSCAN1的靶标RBP，circZKSCAN1可能通过与EIF4A3结合抑制肝癌细胞的侵袭转移能力。

To investigate the mechanism of circZKSCAN1 inhibiting the invasion and metastasis of hepatocellular carcinoma (HCC) cells.

HepG2 cells which differentially expressed circZKSCAN1 gene were utilized in this experiment. The migration and invasion capability of HCC cells in the OE group (HepG2 overexpressing circZKSCAN1), SH-1 and SH-2 groups (HepG2 silencing circZKSCAN1) and NC group (negative control) was assessed by colony formation assay, scratch test and Transwell chamber test, respectively. The binding sites of enriched RNA-binding protein (RBP) in circZKSCAN1 were identified by bioinformatics analysis. The target RBP was verified by RNA immunoprecipitation (RIP) and RNA pull-down experiment. The relationship between the target RBP and survival of HCC was analyzed. The number of colony formation, migration and number of membrane-penetrating cells among 4 groups were analyzed by single-factor analysis of variance and LSD-t test. The enrichment multiples between two groups were compared by t test. Survival analysis was performed by Kaplan-Meier method and Log-rank test.

Colony formation assay showed that the number of colony formation in the NC, OE, SH-1 and SH-2 groups were 10.5±1.1, 4.7±0.8, 21.7±1.8 and 22.6±2.6, respectively. Compared with the NC group, the number of colony formation in the OE group was significantly reduced, whereas those in the SH-1 and SH-2 groups were significantly increased (LSD-t=-7.386, 9.196, 7.424; P<0.05). Scratch test demonstrated that compared with the NC group, the 24- and 48-h migration ability in the OE group was significantly decreased, whereas those in the SH-1 and SH-2 groups were significantly increased. Transwell chamber test found that the number of migrating cells in the NC, OE, SH-1 and SH-2 groups was 28.5±4.3, 14.3±2.6, 47.2±7.7 and 49.6±8.0, respectively. Compared with the NC group, the number of migrating cells in the OE group was significantly reduced, while those in the SH-1 and SH-2 groups were significantly elevated (LSD-t=-4.895, 3.673, 3.474; P<0.05). Transwell chamber test showed that the number of membrane-penetrating cells in the NC, OE, SH-1 and SH-2 groups was 21.3±2.1, 7.6±1.5, 38.6±5.8 and 41.3±7.1, respectively. Compared with the NC group, the number of membrane-penetrating cells in the OE group was significantly reduced, whereas those in the SH-1 and SH-2 groups were significantly increased (LSD-t=-9.195, 4.858, 4.679; P<0.05). Bioinformatics analysis revealed that enriched RBP-binding sites in circZKSCAN1 consisted of EIF4A3, FMR1, hnRNPC and U2AF65, etc. RIP experiment indicated that circZKSCAN1 could significantly enrich intracellular EIF4A3. Compared with the NC group, the enrichment multiple in the OE group was (7.82±0.25) times (t=1.057, P<0.05). RNA pull-down experiment showed that EIF4A3 could bind with circZKSCAN1. Online website (http://gepia.cancer-pku.cn) analysis demonstrated that compared with the adjacent tissues, EIF4A3 gene was highly expressed in HCC tissues, mainly in HCC of advanced stage (F=7.72, P<0.05). However, the overall survival rate of HCC patients with high expression level of EIF4A3 gene was even worse (HR=1.90, P<0.05).

EIF4A3 is the target RBP of circZKSCAN1. circZKSCAN1 may suppress the invasion and metastasis of HCC cells by binding with EIF4A3.