CircRNA hsa_circ_0110102 Function as Anti-Oncogenic Gene in Hepatocellular Carcinoma Through Modulating miR-580-5p/CCL2 Pathway

Background: Circular RNAs (circRNAs) have been shown to have critical regulatory roles in tumor biology, whereas their contributions in hepatocellular carcinoma (HCC) still remains enigmatic. The purpose of this study was to investigate the molecular mechanisms involved in hsa_circ_0110102 in the occurrence and development of HCC. Methods: The expression levels of hsa_circ_0110102 in HCC cell lines and tissues were estimated by RT-qPCR assay. The proliferation, migration, and invasion of HCC cells were determined by CCK-8 and transwell assay. The western blot and ELISA were employed to examine the related-protein and cytokine expression. The association between miR-580-5p and hsa_circ_0110102 or CCL2 was predicted and armed by dual-luciferase reporter assay and RNA pull-down. Results: hsa_circ_0110102 was signicantly down-regulated in HCC cell lines and tissues, low hsa_circ_0110102 expression levels were associated with poor prognosis. Knockdown hsa_circ_0110102 signicantly inhibited cell proliferation, migration and invasion. In addition, luciferase assay and RNA pull-down assay indicating that hsa_circ_0110102 function as sponge for miR-580-5p. Moreover, miR-580-5p which could directly bind to the 3’-UTR of CCL2 and induce its expression, then active the COX-2/PGE2 pathway in macrophage via FoxO1 in p38 MAPK dependent manner. Furthermore, the Δ256 mutant of FoxO1 showed no activation effect. Conclusion: hsa_circ_0110102 act as a sponge for miR-580-5p and decreased CCL2 secretion in HCC cells, then inhibits pro-inammatory cytokine release from activated macrophage by regulating the COX-2/PGE2 pathway. These results indicating that hsa_circ_0110102 serves as a potential prognostic predictor or therapeutic target for HCC. The proposed model illustrates the role of hsa_circ_0110102/miR-580-5p/CCL2 signaling in the regulation of HCC metastasis. hsa_circ_0110102 is down regulated in HCC patients with metastasis or recurrence and binds miR-580-5p, which target CCL2. CCL2 in the tumor microenvironment could active the COX-2/PGE2 pathway in macrophage via p38 MAPK/FoxO1 dependent manner.


Background
Hepatocellular carcinoma (HCC) is the most frequent primary malignancy of the liver and the third leading cause of cancer death worldwide , which accounts for about 90% of all cases of primary liver cancers [1]. China accounts for more than 50% of the total HCC cases worldwide. The main risk factors for HCC are chronic infection with hepatitis B virus (HBV) and hepatitis C virus (HCV) and nonalcoholic fatty liver disease (NAFLD). Over the past decades, great efforts have made to reach diagnosis and treatment methods to new heights, but the mortality rate is still unsatisfactory. Therefore, it is urgently relevant to determine the risk factors of HCC occurrence and reveal the molecular mechanism of its promotion.
Circular RNAs (circRNAs) are a subclass of transcripts with covalently closed loop structures that lack classic 5' caps or 3' polyadenylated tails, which used to be regarded as by-products of erroneous splicing [2]. But recently, mounting studies have shown that circRNAs play important roles and serve various functions in nervous system, carcinogenesis, cardiovascular diseases, immunomodulation and metabolic disorder [3,4]. CircRNA could regulates target gene or protein expression in various manners: function as microRNA (miRNA) sponges to show miRNA inhibition effect by competing endogenous RNA (ceRNA) [5], interfere with their parental genes by modulation of alternative splicing or transcription [6], and interact with RNA-binding proteins (RBPs)as scaffolds for the assembly of protein complexes [7]. Recently, numerous circRNAs have found to be dysregulated in HCC. For example, CircBACH1 (hsa_circ_0061395) is signi cantly upregulated in HCC tissue andcircBACH1 knockdown suppresses the proliferation and increased apoptosis of HCC cells by regulating p27 repression via HuR [8]. CircASAP1 acts as a ceRNA for miR-326 and miR-532-5p to mediates tumor-associated-macrophage (TAM) in ltration [9]. As a consequence, emerging evidences support the critical roles of circRNA for HCC tumorigenesis and development.
C-C chemokine ligand 2 (CCL2) is overexpressed in HCC and is a prognostic factor for patients [10]. In tumor microenvironment, the interaction between CCL2 and C-C motif chemokine receptor 2 (CCR2) could regulates chemotaxis of TAMs and contributes to the cancer progression. Cyclooxygenase-2 (COX-2), as the key enzyme in the synthesis of prostaglandin E2 (PGE2), is an important in ammatory factor and over-expressed in many metastatic tumors [11]. COX-2 plays an important role in angiogenesis, apoptosis, in ammatory and metastatic [12] and is proposed as a potential tumor therapeutic target [13].
Nonsteroidal anti-in ammatory drugs (NSAIDs), the main available COX-2 inhibitors, have been widely used in cancer treatment [14].
In our study, we identi ed a novel circRNA hsa_circ_0110102 and investigated its biological roles in the HCC progress. hsa_circ_0110102 could target the miR-580-5p/CCL2 in HCC. Moreover, CCL2 secreted into the tumor microenvironment could induced the COX-2 expression and PGE2 release via FoxO1-dependent manner in macrophage, and then promote the proliferation of HCC cells. Overall, these ndings indicate that hsa_circ_0110102 might act as novel biomarkers for HCC prognosis and promising therapeutic targets.
Clinical tissue samples and cell culture HCC tissues used in this study involved 17 patients of both genders with previously diagnosed HCC (13 men, 4 women) with averaged age 57 ± 12 years at Shandong Provincial Hospital A liated to Shandong First Medical University, Jinan, China. Tumor tissues and adjacent tissue samples were immediately collected and placed into liquid nitrogen and stored at -80 °C until further processing. All samples used in this study were approved by the Committees for Ethical Review of Research Involving Human Subjects at Shandong Provincial Hospital.
The THP-1 and Huh-7 co-culture system were established to simulate the microenvironment of HCC as previously reported [15]. Brie y, THP-1 cells were seeded into 6-well plates and differentiated with 200 ng/mL PMA for 24 h. Meanwhile, Huh-7 cells were transfected with expression vector for 24 h then seeded into transwell at a density of 1.5×10 5 cells per chamber. The differentiated THP-1 cells were washed three times with PBS and then co-cultured with Huh-7 cells for another 24 h. At the end of the treatment, THP-1 cells were harvested for FCM, RT-qPCR and western blot assay, the medium was collected for ELISA assay.

RNA immunoprecipitation (RIP)
RIP was conducted with a Magna RIP TM RNA-binding protein immunoprecipitation kit (Millipore, USA) according to the manufacturer's guidelines. Brie y, anti-Ago2 antibody and rabbit IgG was incubated with magnetic beads at room temperature for 30 min to generate antibody-coated beads. Approximately 1×10 7 cells were lysed and mixed at 4°C overnight. After wash, co-immunoprecipitated RNA was extracted and detected using RT-qPCR.

Pull-down assay
Biotin-labelled hsa_circ_ 0110102 or oligo probes (GenePharma, China) were pre-incubated with Streptavidin-Dyna beads M-280 (#11206D, Invitrogen, USA). hsa_circ_0110102 overexpressing and control cells were lysed and incubated with the beads at 4°C overnight. Then, RNA was extracted and measured by RT-qPCR.

Cell proliferation assay
A Cell counting Kit-8 (CCK-8, #HY-K0301, MCE, Shanghai, China) method was utilized to assess the proliferation of cells. Transfected cells were seeded into 96-well plates at a density of 5×10 3 cells per well and incubated at 37°C for indicated time. 10 μL CCK-8 solution was added into each well and incubated for 2 h before detect, and then absorbance at 450 nm was measured using a microplate reader (Varioskan LUX, Thermo Fisher, USA).
For EdU staining assay, an Edu staining kit (iFluor 647, ab222421, Abcam) was used according to the manufacturer's protocol. In brief, cells were seeded into 6-well plates and transfected for 48 h. Then, cells were stained with EdU solution for 2 h and xed with 4% paraformaldehyde and cultured with 0.5% Triton X-100 for 10 min. Finally, cells were stained with DAPI for 10 min. Images were collected with a FV3000 microscope (Olympus, Japan).

Quantitative reverse-transcription PCR (RT-qPCR)
Total RNA was extracted using TRIzol reagent (#15596-026, Invitrogen), nuclear and cytoplasmic RNA was extracted with the PARIS™ Kit (#AM1921, Thermo Fisher).RT-qPCR kits were purchased from TaKaRa Bio Inc (Dalian, China) and cDNA was synthesized according to the manufacturer's instruction. The QuantStudio 5 Real-Time PCR System (Applied Biosystems, USA) and the SYBR Green (no. B21202, Bimake, Houston, USA) were used for RT-qPCR assay. The expression levels of mRNA were normalized to GAPDH mRNA. Primers were purchased from Genscript biotech co., ltd. For miRNA assay, miRNA-speci c Taqman PCR primers were obtained from Life Technologies (CA, USA). Genes levels were normalized to U6 or GAPDH to generate a2 -ΔΔCt value for the relative expression of each transcript. Primer sequences used in the experiments were as follows: The isolated TIL was incubated with CD68 and CCR2 antibody for 1 h at 4°C; CCL2-stimulated PBMCs were incubated with CCR2 antibody for 1 h at 4 °C, then washed with PBS for three times and nally resuspended in 200 μL PBS and analyzed by ow cytometry.
For cell cycle analysis, HepG2 and Huh-7 cells were trypsinized and xed with 70% ethanol overnight at 4℃, followed by staining with PI reagent. The percentage of cells in the G0/G1, S, and G2/M phases was recorded using Flow cytometry (BD Biosciences, NY, USA). The results obtained were analyzed by the software FlowJo.
ELISA assay THP-1 cells were treated with 50 ng/ml rCCL2 or transfected with the expression vector, then cells were treated with or without 1 μM AS1842856 or infection with sh-FoxO1 for 24 h. The culture supernatant was collected, and were used to detect the concentration of VEGF (cat no. DVE00), COX-2 (cat no. AF4198), IL-6 (cat no. D6050) and PGE2 (cat no. KGE004B) secreted using ELISA kits (all from R&D, Minneapolis, MN, USA). The absorbance was measured at 450 nm using a microplate reader (Varioskan LUX, Thermo Fisher, USA).

In vitro migration and invasion assays
Cell migration and invasion were measured using Transwell assay as previously described [16].

Plasmid constructs and transfection
pSELECT-HA-mFoxO1-wild-type (Addgene plasmid # 83308) and Δ256 (Addgene plasmid # 83379) were a gift from Steven Abcouwer [17]. Full-length of CCL2 cDNA was ampli ed and cloned into a pcDNA3.1 expression vector. The CCL2 promoter sequence from positions 500 to +1 and COX-2 promoter sequence from positions −1,122 to +27 relative to the transcriptional start site subcloned into pGL3-Basic vector, the point mutation in the FoxO1 binding site of COX-2 promoter was generated by site-directed mutagenesis that splices by overlapextension. FoxO1 shRNA and scrambled shRNA were purchased from Obio technology Co., Ltd. (shanghai, China). Cells were seeded into 6-well plates for overnight culture, then transfected with hsa_circ_0110102 overexpression vector or siRNA, miR-NC, miR-580-5p mimic or inhibitor with lipofectamine 2000 for 48 h, blank vector-transfected cells were used as controls.

Chromatin immunoprecipitation assay
To detect the in vivo association of FoxO1 with human COX-2 promoter, chromatin immunoprecipitation (ChIP) analysis was performed as previously described using SimpleChIP Enzymatic Chromatin IP Kit (#9002, CST, USA) according to manufacturer's protocol. ChIP-PCR was used for validation with the forward primer 5'-CACCGGGCTTACGCAATTTT-3' and the reverse primer 5'-ACGCTCACTGCAAGTCGTAT-3', which were speci cally designed from the COX-2 promoter region (139 to +29).

Luciferase assays
Cells were transfected with luciferase reporter plasmids and subsequently incubated for 24 h in the complete medium. Luciferase activity was measured using the Dual-luciferase reporter assay System (Promega, Madison, WI) and normalized to Renilla luciferase values. Measurements for three biological replicates were taken in triplicate and averaged.

Statistical analysis
Data are expressed as mean ± SEM. SPSS 21.0 was used for statistical analysis. Unpaired Student's ttest were used for comparation between two groups. Two-way ANOVA analysis using the Tukey test was employed to compare three groups or more. P<0.05 was considered statistically signi cant.

Results
hsa_circ_0110102 is down-regulated in HCC cell lines and tissues hsa_circ_0110102, with the 1252 bp spliced mature sequence length, is located at chr19:9763628-9770143 and derived from exons 3 to 6 of the ZNF562 gene ( Figure 1A). We used the published GSE135806 data set to identify circRNAs differentially expressed between ve HCC and ve adjacent normal tissues from HBV-related male HCC patients, aged 43-57 [18], and found that, compared with the control tissues, the expression of hsa_circ_0110102 was down-regulated in HCC tissues ( Figure 1B).
Compared with matched normal control tissues, hsa_circ_0110102 was low-expressed in 17 HCC tissues collected from our hospital ( Figure 1C). In the HCC cell lines (Hep3B, MHCC-97H, Huh-7, HepG2 and SMCC-7721), the expression of hsa_circ_0110102 was signi cantly lower than human normal LO2 hepatocytes ( Figure 1D). As Huh-7 and HepG2 showed the lower hsa_circ_0110102 levels than other HCC cells, we would use these two cell lines in the next studies to reveal the tumor suppressive effect in HCC.
hsa_circ_0110102 knockdown promotes HCC cell growth, migration and invasiveness Next, we examined the relative abundance of hsa_circ_0110102 in the nucleus and cytoplasm of Huh-7 and HepG2 cells via nuclear mass separation assays ( Figure 2A) and FISH ( Figure 2B), concluded that hsa_circ_0110102 was mainly located in the cytoplasm.
To analyze the role of has_circ_0110102 in HCC, has_circ_0110102-speci c small interfering RNAs (siRNAs) were used to down-regulate the expression levels of has_circ_0110102.Thus, siRNAs targeting the backsplice junction sequenceofhas_circ_0110102 and the full length of ZNF562 were designed. We found that siRNA targeting the backsplice junction knocked down only the circular transcript and did not affect the expression of linear species. Contrarily, siRNA targeting linear transcript knocked down only the ZNF562 linear transcript but not the circular transcript ( Figure S1).Huh-7 and HepG2 cells were transfected with hsa_circ_0110102 expression vector and siRNA then followed with CCK-8 assay, migration and invasion assay and found that compared with control cells, hsa_circ_0110102 knockdown dramatically induced the proliferation, cell migration and invasiveness, which were inhibited by the hsa_circ_0110102 overexpression ( Figure 2C-2E).
Next, we investigated the effect of miR-580-5p on cell proliferation in HCC. Transfected with miR-580-5p inhibitor signi cantly decreased the miR-580-5p expression in Huh-7 and HepG2 cells( Figure 4D). CCK-8 assay indicated that miR-580-5p inhibitor also inhibited cell proliferation rate ( Figure 4E), which was con rmed with EdU staining assay to detect DNA synthesis levels ( Figure 4F and 4G), cell cycle assay showed that less cells were in S phase after miR-580-5p inhibitor transfection ( Figure 4H). These results indicating for the rst time that miR-580-5p may function as an oncogene in the development of HCC.
CCL2 is a direct target of miR-580-5p By searching on Miranda (http://www.microrna.org/) and Targetscan (http://www.targetscan.org/), we found that there exists binding site between miR-580-5p and the 3'-UTR of CCL2 ( Figure 5A). Overexpression of miR-580-5p increased the mRNA and protein levels of CCL2 both in Huh-7 and HepG2 cells and the content of CCL2 in the culture medium, indicating that miR-580-5p overexpression increased the synthesis and secretion of CCL2 in HCC cells. But co-transfected with hsa_circ_0110102 signi cantly down-regulated the CCL2 mRNA, protein and the concentration in the medium ( Figure 5B and 5C), indicating that hsa_circ_0110102 functions as a miR-580-5p sponge to inhibit the expression of CCL2.
We then constructed pGL3 plasmids containing the CCL2 3'-UTR wile type and mutant sequence. Compared with the NC group, the miR-580-5p overexpression showed a signi cant increase in relative Rluc activity in both Huh-7 and HepG2 cells, hsa_circ_0110102 showed no effect on the CCL2 promoter activation. However, co-transfection with hsa_circ_0110102 abolished the effect. However, mutating the CCL2 3'-UTR binding site blocked the effect of miR-580-5p and hsa_circ_0110102 ( Figure 5D). The CCL2 mRNA and protein levels were detected with RT-qPCR and western blotting, the results showed that, miR-580-5p mimic signi cantly. We then used miR-580-5p inhibitor and siRNA targeting hsa_circ_0110102 to examine whether hsa_circ_0110102 inhibit the CCL2 expression levels in HCC cell by antagonist the miR-580-5p effect. The results showed that si-hsa_circ_0110102 could abolish the miR-580-5p inhibitorinduced downregulation of CCL2 mRNA and protein expression levels ( Figure 5E and 5F). miR-580-5p overexpression increased the migration and invasion of Huh-7 cells, hsa_circ_0110102 signi cantly inhibited the effect of miR-580-5p ( Figure 5G). These results showed that, CCL2 is identi ed to be a direct target of miR-580-5p, and hsa_circ_0110102 act as a sponge for miR-580-5p to block the CCL2 induce effect.

AS1842856 inhibits CCL2 induced CCR2 expression and cytokine secretion in macrophages
Previously studies found that, CCL2 secreted from hepatocytes triggers macrophage recruitment and could induce liver brosis, and even HCC [19]. In tumor microenvironment, CCL2 interacts with CCR2, which on the surface of macrophage, to mediate chemotaxis of monocytes and TAM to facilitates cancer progression [20]. Our results have shown that, hsa_circ_0110102 act as a sponge for miR-580-5p and inhibited the CCL2 expression and secretion from HCC cells. Next, we use the THP-1 and Huh-7 co-culture system to detect their effect on the macrophage phenotype and function. The FCM results showed that CD68 + macrophages cell developed polarized CCR2 phenotypes in THP-1, and that the percentage of CCR2 + CD68 + cells increased signi cantly in the miR-580-5p mimic group, but co-transfection with hsa_circ_0110102 blocked the effect of miR-580-5p ( Figure 6A).
Previous studies have shown that COX-2 and PGE2 in the tumor microenvironment which released from macrophage was critical in the HCC development. As shown in Figure 6B and 6C, miR-580-5p signi cantly induced the COX-2 mRNA and protein expression in THP-1 cells and increased the content of COX-2 in the co-culture medium. As COX-2 is a rate-limiting enzyme in regulating PGE2-generation and its expression is highly inducible in macrophages [21]. PGE2, which induces the secretion of vascular endothelial growth factor (VEGF) to accelerate neovascularization in HCC [22]. We also found that, miR-580-5p induced and hsa_circ_0110102 blocked the PGE2 expression in the co-culture system ( Figure 6D).
Previously studies found that, FoxO1 could promote pro-in ammation cytokine release of macrophage [23]. The activation of FoxO1 was also involved in the regulation of COX-2 expression [24]. So, we suspected whether CCL2-induced COX-2 expression increasement was mediated through FoxO1 dependent manner in macrophage. As shown in FigureS3, 6E and 6F, 50 ng/ml rCCL2 signi cantly upregulated the mRNA and protein levels of COX-2 in THP-1 cells, and also the released COX-2, PGE2, VEGF and IL-6 in the medium, either FoxO1 knockdown or inhibitor AS1842856 signi cantly inhibited the effect of rCCL2, instead of blocked it. These results showed that CCL2 induce the COX-2/PGE2 pathway partly through FoxO1.

CCL2 induces FoxO1 activation via p38 MAPK pathway in THP-1 cells
Previous studies found that, MAPK pathway can active FoxO family [25]. Interestingly, autocrine CCL2 is associated with activation MAPK pathway [26]. We next investigated the relationship between MAPK pathway and FoxO1 activity in response to rCCL2 stimulation in macrophage. ChIP assay found that, rCCL2 stimulation signi cantly promoted the binding of FoxO1 on COX-2 promoter in a time-dependent manner in THP-1 cells ( Figure 7A), which was reversed by p38 inhibitor SB203580, while the JNK inhibitor SP600125 and ERK1/2 inhibitor SCH772984 shows no effect on the COX-2 promoter activity ( Figure 7B).
These results demonstrate that CCL2 induced FoxO1 activation via p38 MAPK.
Using luciferase assay, we found that, both rCCL2 or forced overexpression of CCL2 increased COX-2 promoter transcription activity, FoxO1 co-operate with CCL2 caused a more signi cant increase, while pretreated with AS1842856 inhibited the transcription level ( Figure 7C). FoxO1-Δ256, lacking the transactivation domain of FoxO1, was used to detect whether this site function in the regulation COX-2 activity. THP-1 cells were co-transfected with FoxO1 WT and Δ256 expression vectors, via luciferase assay, we found that WT enhanced COX-2 transcription activity, but no similar enhancement was observed in Δ256 mut. AS1842856 and p38 MAPK inhibitors treatment both decreased the COX-2 transcription activity, but ERK and JNK inhibitor shows no effect ( Figure 7D). Implying that, in the physiologic state, Δ256 site of FoxO1 was quite important for the transcript activation effect of COX-2.
Next, a COX-2 promoter upstream region mutation was subcloned into pGL3-Basic to de ne FoxO1 target site on the COX-2 promoter, after co-transfection with FoxO1 into THP-1 cells, the activity of promoter variants was determined. As shown in Figure 7E, the activity of promoter variants with DNA deletions up to -111 nt and -101 nt shows no difference in the presence or absence of FoxO1, but FoxO1 signi cantly induced the WT promoter activity. These results indicating that, the FoxO1 target site was localized to the -111/-101-nt region of the COX-2 promoter.

Discussion
CircRNAs are endogenous noncoding RNAs that been regarded as potential novel diagnostic and prognostic molecular biomarkers in cancer [27]. Recently, mounting evidence have identi ed the important roles of circRNAs in the carcinogenesis and development of HCC, but little focus on its functional role in regulating the crosstalk between cancer cells and microenvironment. Microenvironment composed of broblasts, endothelial cells and immune system cells, that interact with tumor mass could promote the progression of cancer [28]. Among them, macrophages account for a large proportion in tumors, and also called TAM. Notably, TAM acts directly on tumor cells and acts indirectly on the tumor microenvironment through cytokines, like VEGF and IL-6, which enhancing cancer cell invasion and metastasis, angiogenesis ability [29].
Macrophages participate in the in ammatory environment of mutagenesis in the early stage of tumor development, and as the tumor progresses to malignant, it also can stimulate angiogenesis and promote tumor migration, invasion and inhibition of tumor immunity [30]. Many clinical studies have shown that macrophages may promote tumorigenesis. CircRNAs have also shown potential effect to in uence the tumor microenvironment, Circular RNA CDR1as may play a speci c role in immune and stromal cell in ltration in tumor tissue, especially those of CD8 + T cells, activated NK cells, M2 macrophages [31]. Circular RNA CircASAP1 could mediates tumor-associated-macrophage in ltration by regulating the miR-326/miR-532-5p-CSF-1 pathway [9]. So, targeting circular RNA on the tumor microenvironment may be a potential direction in the future.
This study focused on the hsa_circ_0110102 to discover novel HCC therapy targets. Our research found that hsa_circ_0110102 was down-regulated in HCC tissues and knockdown of hsa_circ_0110102 inhibits cell proliferation, invasion and migration in vitro. As circular RNAs may act as sponge of miRNA to regulate the target gene function, we use luciferase reporter assays and bioinformatics prediction revealed that hsa_circ_0110102 may compete for binding with miR-580-5p.
Aberrant miR-580-5p expressions in many tumor has been investigated, like glioma and breast cancer [32], but until now, it's exactly mechanistic contribution to HCC progression has not been explored. Our study indicated that, miR-580-5p levels in tumor tissues were dramatically higher than that of the adjacent normal tissue, and miR-580-5p could promotes the proliferation, invasion and migration of HCC. We further con rmed that hsa_circ_0110102 act as an endogenous sponge on miR-580-5p to regulates the expression of CCL2.
CCL2 as a chemokine, bind to CCR2 and chemotactic for monocytes and macrophages, is an important human CC chemokine which shows a strong tendency to monocytes. It plays an important role in physiopathological reactions such as in ammation, pathogen infection, and tumor formation [33]. Many research has shown that, CCL2 is closely related to tumorigenesis and related to the poor prognosis in various cancers such as HCC [34], lung cancer and prostate cancer [35]. Blocking the CCL2-CCR2 signaling pathway can inhibit malignant tumor growth, metastasis, reduce postoperative recurrence, and improve survival [36].
A growing number of cellular signaling pathways have been implicated in orchestrating the process of COX-2/PGE2 pathway in tumor development. COX-2 acts as the rate-limiting enzyme of PGE2, is considered to be an important mediator in in ammatory responses and is highly expressed in a variety of tumors. It interacts with many cytokines and other members of the prostaglandin family to constitute a complex network system. Several studies suggest that high levels of PGE2 are also involved in the in ammatory response and plays a predominant role in promoting tumor growth and is associated with poor prognosis. Moreover, in vivo studies have further indicated that the speci c inhibition of COX-2 shows a protective effect on HCC development and slow down the tumor progression in mouse models [37].
FoxO1, an important transcription factor which has been shown to be involved in various cellular functions, activated in response to a wide range of external stimuli, like growth factors, insulin and oxidative stress [38]. Recently, FoxO1 activation have been shown to participate in COX-2/PGE2dependent cell migration [39]. To comprehensively understand the effect of CCL2 on macrophage, we performed in vitro assays to testify the effect of transcription factor FoxO1 on CCL2 induced COX-2/PGE2 pathway activation, our results showed that FoxO1 knockdown and inhibitor decreases the COX-2 and PGE2 expression and inhibits the migration and tube formation. To further explore the role of FoxO1 pathway in the regulation of COX-2/PGE2 pathway in macrophage, ChIP assay was used to investigate the binding effect of FoxO1 on COX-2 promoter, luciferase assay was obtained to analyze the inhibition effect of CCL2 on COX-2 promoter transcription activity. FoxO1 WT were used to active the endogenous activity and Δ256 mut-FoxO1, a loss of function FoxO1 mutation, was used as a negative control. As we expected, FoxO1 increased the COX-2 transcription activity, but Δ256 mut-FoxO1 showed no similar effect. We further detected the binding site of FoxO1 on COX-2 promoter using a point mutation promoter luciferase vector.
In Conclusion, our data indicated that, hsa_circ_0110102 shows critical regulatory role as an oncogenic circRNA through the sponge effect of miR-580-5p to inhibit the expression of CCL2 in HCC cells. CCL2 could further actives COX-2/PGE2 pathway in macrophage via FoxO1-dependent manner (Fig. 8). Overall, the present research throw light on targeting the hsa_circ_00110102/miR-580-5p/CCL2 axis may be a promising therapeutic strategy for HCC.

Notes
Xinxing Wang and Wei Sheng contributed equally to this work.      Luciferase activity was normalized by Renilla luciferase values. Data are expressed as fold-change relative to the level of control. Data were presented as mean ± S.E., *, P<0.05.