Inhibiting roles of FOXA2 in hepatocellular carcinoma cell migration and invasion by transcriptionally suppressing microRNA-103a-3p and activating the GREM2/LATS2 axis

Background Forkhead box A2 (FOXA2) is a transcriptional activator for liver-specic genes. Hepatocellular carcinoma (HCC) is a prevalent fetal malignancy across the globe. This work focused on the role of FOXA2 in HCC cell migration and invasion and the involving molecules. Methods FOXA2 expression in HCC tissues and cells was determined using RT-qPCR. Altered expression of FOXA2 was introduced to identify its role in HCC cell migration and invasion using Transwell assays. The potential target microRNA (miRNA) of FOXA2 was predicted via online prediction and validated through a ChIP assay, and the mRNA target of miRNA-103a-3p was predicted and conrmed through a luciferase assay. The roles of miR-103a-3p and GREM2 in HCC cell invasion and migration were determined, and the downstream molecules mediated by GREM2 were analyzed. Results FOXA2 and GREM2 were poorly expressed while miR-103a-3p was abundant in HCC tissues and cells. Overexpression of FOXA2 or GREM2 suppressed migration and invasion of HepG2 and SK-HEP-1 cells, while up-regulation of miR-103a-3p led to reverse trends. FOXA2 transcriptionally suppressed miR-103a-3p to increase GREM2 expression, and silencing of GREM2 partially blocked the inhibitory effects of FOXA2 on cell migration and invasion. GREM2 increased LATS2 activity and YAP phosphorylation and degradation. Conclusion This study evidenced that FOXA2 inhibits migration and invasion potentials of HCC cell lines through suppressing miR-103a-3p transcription. The following upregulation of GREM2 plays key roles in migration inhibition by promoting LATS2 activity and YAP phosphorylation. This study may offer new insights into HCC treatment.


Introduction
Though most of the risk factors such as obesity, excessive alcohol consumption, smoking, and hepatitis B and C virus (HBV and HCV) infection are potentially preventable, the incidence rate of liver cancer (LC) is increased most rapidly among human cancers by an annual 2-3% increase during 2007 to 2016 [1]. LC ranks the second highest causes of cancer death worldwide with hepatocellular carcinoma (HCC) representing the most common type [2]. China has a particularly high mortality and morbidity rate with estimated 466,100 new diagnoses and 422,100 deaths according to the cancer statistics in 2015, accounting for over half of all cases worldwide [3,4]. In addition, due to the di culty in early diagnosis, high malignancy with rapid progression, and the lacking of effective targeted drugs, the survival rate of LC is seriously low [5]. Though surgical resection, liver transplantation and ablation are promising curative strategies for HCC patients, these operations are only available for patients at early stages characterized by limited tumor in liver [6,7]. Unfortunately, most patients are found at late stages with frequent malignant metastasis. Hence, developing novel ways in metastasis control may provide novel therapeutic options for HCC treatment.
Forkhead box A2 (FOXA2) is a member of the forkhead class of DNA-binding proteins that are transcriptional activators for the regulation of cell differentiation and metabolism [8,9]. FOX proteins may in uence multiple cellular processes that are correlated with the initiation, metastasis, and development of cancers by different regulation manners with diverse networks [10]. Interestingly, a previous study noted the silencing of FOXA2 by microRNA (miR)-92a could increase HCC cell proliferation and invasion [11]. However, the exact role of FOXA2 in HCC cell migration and invasion and the mechanisms remain largely unstudied. Transcription factors are capable of binding to promoter regions of downstream target genes [12] including miRNAs. miRNAs are the mostly studied non-coding RNAs that are often involved in FOXA-mediated events [13]. miRNAs comprise 17-25 nucleotides with a major role in gene degradation by binding to the target complementary mRNAs in a post-transcriptional level [14]. Aberrant expression of miRNAs has been broadly found in human malignancies, and they function either as oncogenes or tumor suppressors through the different post-transcriptional modi cations in genes [15]. Unsurprisingly, a large number of miRNAs have been noted to play key roles in HCC pathogenesis by mediating cell malignancy and tumor metastasis, thus may serve as potential therapeutic approaches [16]. Here, our study validated that FOXA2 could bind to miR-103a-3p, which was found to directly target gremlin 2 (GREM2). miR-103a-3p has been noted to work as a tumor promoter in oral squamous cell carcinoma [17]. On the other hand, it was suggested to suppress invasion and proliferation of prostate cancer cells [18]. But its function in HCC has not been investigated yet. As for GREM2, it is suggested to be lowly expressed in HCC through the data on GEPIA (http://gepia.cancer-pku.cn/). Taken together, we hypothesized that FOXA2 could inhibit HCC metastasis through suppressing miR-103a-3p and the following up-regulation of GREM2, with cell migration and invasion experiments performed to validate the hypothesis.

Materials And Methods
Clinical tissue sample collection From January 2018 to February 2019, 50 pairs of HCC tissues and the adjacent normal tissues (at least 3 cm away) were collected from HCC patients who treated in the Second People's Hospital of Liaocheng.
The tissues were collected during surgery and instantly frozen at -80°C. There were 39 male (78%) and 11 female (22%) patients with an average age of 51 ± 11.3 years. All enrolled patients were diagnosed by pathological examination without surgery history or other malignancies.

Reverse transcription quantitative polymerase chain reaction (RT-qPCR)
Total RNA in cells and tissues was extracted using the TRIzol Reagent (Invitrogen). The RNA was reversely transcribed into cDNA utilizing a SuperScript RT kit (Invitrogen) as per the kit's protocol. Next, real time qPCR was conducted using SYBR Green (Applied Biosystems, Foster City, CA, USA) on a CFX96 System (Bio-Rad, Hercules, CA, USA) as per the manufacturer's protocols. U6 was served as the internal reference for miR-103a-3p, while GAPDH for other genes. Relative gene expression was determined using the 2 -ΔΔCt method. Each procedure was performed 3 times. The primers are presented in Table 1.

Transwell assay
Invasion and migration of HCC cells was determined using Transwell assays. As for cell migration, 24 hours after transfection, a total of 2 × 10 4 cells were seeded into the apical chamber supplemented with FBS-free RMPI-1640, while each basolateral chamber was loaded with 600 µL DMEM with 10% FBS. Then Transwells were placed in an incubator at 37°C for 24 hours. The cells in the apical chamber and on the upper surface were wiped out by cotton swabs. The migrated cells were xed in 4% paraformaldehyde (Beyotime Biotechnology Co., Ltd., Shanghai, China) for 20 minutes and then stained with 0.5% crystal violet (Beyotime) for 15 minutes. The invasion of cells was determined in a similar manner with the apical chambers pre-coated with Martrigel (BD-Biosciences, CA, USA) before cell seeding. The number of invaded and migrated cells were observed and counted under an IX81 microscope (Olympus, Tokyo, Japan) with 5 elds randomly selected and the average value was calculated.
Dual luciferase reporter gene assay Luciferase reporter plasmids were constructed with the ligation of oligonucleotides containing the wide type (WT) and mutant type (MT) binding sequences based on the putative target sites of GREM2 3'UTR and miR-103a and inserted into GV272 vectors (Genechem Co., Ltd., Shanghai, China). Well-constructed vectors were co-transfected with miR-103a mimic or mimic NC into HepG2 and SK-HEP-1 cells using the Lipofectamine 2000 (Invitrogen). Forty-eight hours later, the cells were lysed, and the luciferase activity was measured using a Dual Luciferase Reporter Assay Kit (Promega, Madison, WI, USA) as per the manufacturer's protocols. Three ndependent experiments were performed.
Chromatin immunoprecipitation (ChIP) assay Cells were treated with 37% formaldehyde, collected and subjected to ultrasonication using a VCX750 (SONICS, USA) at a power of 25%. The cells were ultrasonicated 4.5 seconds each time for a total of 14 times at a 9-second interval. Thereafter, the cells were centrifuged at 10000 g at 4℃ for 10 minutes to discard the insolubles. Next, the antibody against FOXA2 (ab256493, Abcam, Cambridge, UK) and the control IgG antibody (ab6721, Abcam) were added to combine with the target protein-DNA compound. Next, Protein A was further administrated to precipitate with the antibody-target protein-DNA compound.

Statistical analysis
Measurement data were exhibited as mean ± standard derivation (SD). SPSS 21.0 (IBM Corp. Armonk, NY, USA) was applied for data analysis. Data were analyzed using the paired t test, Pearson's Correlation Analysis, one-way or two-way analysis of variance (ANOVA) and Tukey's multiple comparison test. The p value was obtained from two-tailed tests, and p < 0.05 was considered to present signi cant difference.

FOXA2 is poorly expressed in HCC and inhibits HCC cell migration and invasion
FOXA2 is a transcriptional activator for liver-speci c genes. Importantly, FOXA2 has been noted to be expressed at a lower level in HCC tissues than that in normal liver [19]. Here, our study measured the FOXA2 expression in 50 pairs of HCC tissues and the adjacent normal tissues using RT-qPCR, and the results suggested that the FOXA2 expression was much lower in tumor tissues than that in the paired normal ones (Fig. 1A). In terms of FOXA2 in cells, it was found that the HCC cell lines HepG2 and SK-HEP-1 also presented lower levels of FOXA2 (Fig. 1B). Then, pcDNA-FOXA2 was administrated into HCC cell lines to arti cially up-regulate FOXA2 expression in cells (Fig. 1C). Importantly, the invasion and migration abilities of cells were inhibited (Fig. 1D).

FOXA2 transcriptionally suppresses miR-103a-3p
FOXA2 is a transcription factor that can speci cally bind to the upstream sequences of genes at the 5'end to suppress gene expression. miR-103a-3p has been suggested to play oncogenic roles in human cancers and is linked to worse prognosis of cancer patients [17]. Here, we determined miR-103a-3p expression in HCC and normal tissues, which suggested that the HCC tissues showed higher expression of miR-103a-3p ( Fig. 2A). The similar trend was also found in HCC cell lines, where HepG2 and SK-HEP-1 cells presented increased miR-103a-3p expression as relative to THLE-2 cells (Fig. 2B). Then, miR-103a-3p mimic/inhibitor or the NC was transfected into cells, after which the miR-103a-3p expression in cells was correspondingly increased or decreased (Fig. 2C). Next, the Transwell assay results suggested that the migration and invasion of cells was increased following miR-103a-3p up-regulation, while decreased after miR-103a03p down-regulation (Fig. 2D).
According to the prediction on the Bio-Information System Jasper (http://jaspar.genereg.net/), it was suggested that FOXA2 has binding sites with miR-103a-3p at the promoter region (Fig. 2E). The sequence with highest relative score was selected for ChIP assay, and the following RT-qPCR suggested that miR-103a-3p was enriched in the FOXA2 antibody rather than IgG, indicating that FOXA2 could bind with the promoter region of miR-103a-3p (Fig. 2F). In addition, the RT-qPCR results revealed an inverse correlation between miR-103a-3p and FOXA2 expression (Fig. 2G). Following pcDNA-FOXA2 administration, the miR-103a-3p expression was notably inhibited (Fig. 2H). The above experiments suggested miR-103a-3p promotes HCC cell migration and invasion, while FOXA2 transcriptionally suppresses miR-103a-3p expression.
miR-103a-3p directly targets GREM2 GREM2 was suggested as a poorly expressed gene in HCC according the prediction on GEPIA (Fig. 3A). Then, GREM2 was found as a putative target of miR-103a-3p in StarBase (Fig. 3B). We then supposed that miR-103a-3p could exert functions through binding with GREM2. First, RT-qPCR found that GREM2 was poorly expressed in HCC tissues as relative to the normal ones (Fig. 3C), and lower level of GREM2 was also found in HCC cell lines than that in THLE-2 cells (Fig. 3D). Then, pcDNA-GREM2 and siRNA-GREM2 were transfected into HCC cell lines, and then the GREM2 expression was successfully up-or down-regulated, accordingly (Fig. 3E). The Transwell assay results suggested that pcDNA-GREM2 notably inhibited the migration and invasion of HCC cell lines, while the invasion and migration potential of cells was promoted when they were treated with siRNA-GREM2 (Fig. 3F).
In addition, GREM2 presented a negative correlation with miR-103a-3p expression in HCC tumor tissues (Fig. 3G). After miR-103a-3p mimic transfection, it was found that the GREM2 expression was signi cantly decreased (Fig. 3H). To further validate the binding relationship between miR-103a-3p and GREM2, a dual luciferase reporter gene assay was performed, which suggested that the relative reduced luciferase activity in cells co-transfected with miR-103a-3p mimic and GREM2-WT sequence, while the luciferase activity in cells subjected to other co-transfection showed little differences (Fig. 3I).

GREM2 activates LATS2 to promote YAP phosphorylation
The downstream effector YAP of the Hippo signaling pathway is a publicly accepted oncogene. The nuclear translocated YAP can bind with the transcription factor TEAD to promote the transcription of proliferation and invasion-related genes [20]. Here, we determined the expression of LATS2, a core kinase of the Hippo signaling pathway, and the phosphorylation of YAP in cells, to explore the potential mechanical molecules involved in the miR-103a-3p/GREM2-mediated events. The pcDNA-GREM2 and siRNA-GREM2 and the corresponding NC vectors were transfected into HCC cells, after which the western blot analysis results found that LATS2 expression and YAP phosphorylation were increased when GREM2 was up-regulated, though the total YAP expression was decreased. Accordingly, siRNA-GREM2 led to a signi cant increase in LATS2 expression and YAP phosphorylation, but a decline in YAP expression (Fig. 4). The phosphorylation of YAP blocks its potential in nuclear translocation, leading to a degradation in cytoplasm. These results indicated that GREM2 increases the LATS2 kinase expression to activate the Hippo signaling and to trigger YAP phosphorylation in HCC cells.

FOXA2 inhibits migration and invasion of HCC cells through the miR-103a-3p/GREM2/LATS2 axis
Then, pcDNA-FOXA2, pcDNA-FOXA2 + siRNA-LATS2 or the NC vectors were transfected into cells, and then it was found that the pcDNA-FOXA2 led to an increase in LATS2 expression and YAP phosphorylation but a decrease in YAP expression, but the siRNA-LATS2, accordingly, resulted in reversed trends in cells (Fig. 5A). In addition, a group of cells were co-transfected with pcDNA-FOXA2 and siRNA-LATS2. Then, it was found that the migration and invasion abilities of HCC cells inhibited by pcDNA-FOXA2 were increased following further administration of LATS2 (Fig. 5B). These ndings inferred that the LATS2 is at least partially implicated in the functions of FOXA2.

Discussion
Owing to the frequent late diagnosis and lack of effective treatments, HCC patients at advanced stages presented very unsatisfactory prognosis, leaving the development of new therapeutic options of great urgency. FOXA factors were initially known as hepatocyte nuclear factor 3 and as a family of pioneer transcription factors for liver-speci c genes [21,22]. Studies have demonstrated the suppressing effects of FOXA2 in several human neoplastic diseases [23,24]. The present study identi ed that by suppressing miR-103a-3p transcription, FOXA2 played inhibiting effects on HCC cell invasion and migration with the further involvement of GREM2 and LATS2 activation and the following YAP phosphorylation.
The FOXA subfamily participates in the differentiation and regulation of metabolic tissues including liver, pancreas and adipose tissues [8]. The initial nding of the study was that FOXA2 was expressed at low levels in tissues from HCC patients as well as in HCC cell lines HepG2 and SK-HEP-1 as relative to the paired normal liver tissues or the normal cell line. This was partly in line with a previous report where researchers found a decrease in FOX2 expression in HCC cells and HepG2 cell line as compared to normal liver cells [19]. Importantly, this selective regulation in cellular gene by the FOXA pioneer factors may offer the chance for the speci c suppression in HBV gene expression and the following resolution of chronic HBV infections that causes approximately one million of deaths each year around the world by liver cirrhosis and HCC [25,26]. In addition, our study con rmed that FOXA2 was capable of inhibiting migration and invasion abilities of HCC cell lines. The FOXA factors may differ from each other in terms of tumorigenesis in different cancer types [10,27,28]. In particular, FOXA2 has been noted as a tumor suppressor in several cases. For example, unlike FOXA1 and FOXA3, FOXA2 was suggested to be signi cantly decreased in lung cancer cell lines and was positively linked to prognosis and better survival rate of lung cancer patients [29]. More relevantly, FOXA2 has been documented to inhibit matrix metalloproteinase-9 expression to suppress HCC metastasis [30]. FOXA2 was also suggested to ful ll key roles at enhancer regions of epithelial genes to maintain the enhancer structure and function, thus suppressing the epithelial-mesenchymal transition, a key process during invasion and metastasis [31,32].
The preliminary nding above triggered us to explore the possible downstream molecules. FOX factors are capable of recognizing some speci c patterns in DNA sequences and consequently bind to chromatin and promote the activities of other regulators [10]. Here, the present study identi ed that FOXA2 can bind to the promoter region of miR-103a-3p through integrated online prediction and ChIP assay. In addition, miR-103a-3p expression was found to be highly expressed in HCC tissues and cell lines, presenting an inversive correlation with FOXA2. Then, it was found that up-regulation of miR-103a-3p by miRNA mimic inhibited the migration and invasion potential of HCC cell lines. miR-103a-3p has been noted to be involved with cancer progression in several cancer types. For instance, it presented a high-expression pro le in the tissues and plasma of colorectal cancer patients [33]. An in vitro study suggested miR-103a-3p promotes proliferation of gastric cancer cells by targeting ATF7 [34]. Likewise, inhibition of miR-103a-3p was found to inhibit the proliferation of oral squamous cell carcinoma [17]. Our nding here indicates the inhibiting potential of miR-103a-3p in HCC. Moreover, our study further identi ed that GREM2 as a putative target of miR-103a-3p. GREM2 was suggested to be lowly expressed in HCC according to the data on GEPIA bioinformation system. The role of GREM2 in cancers has hardly been investigated, though, it is a well-known antagonist of bone morphogenetic proteins (BMP) [35], which have been revealed to promote the progression of several human cancers [36][37][38]. Here, our study identi ed that GREM2 was decreased in HCC tissues and cells, with its arti cial up-regulation inhibited cell migration and invasion.
Intriguingly, our study found that GREM2 increased the activity of LATS2 and the further phosphorylation of YAP, namely. LATS2 activation and the following YAP phosphorylation are typical signals of Hippo activation, whose dysfunction is involved in a large number of human diseases including cancer [39]. By phosphorylation, YAP is prevented from nuclear accumulation and consequently degraded in cytoplasm, while the non-phosphorylated YAP is localized in nuclear and that promotes cell proliferation and organ growth, which was inhibited by LATS1/2 [40]. Activation of the Hippo signaling has been found to inhibit the malignant behaviors including growth, migration and invasion of HCC cell lines [41,42]. The experimental ndings that pcDNA-FOXA2 increased LATS2 expression and YAP phosphorylation, and siRNA-LATS2 recovered the migration and invasion abilities inhibited by pcDNA-FOXA2 evidenced that the LATS2 was involved in the FOXA2-mediated events.

Conclusion
To sum up, our study provided evidence that FOXA2 ful lls tumor suppressing roles in HCC by binding to the promoter region of miR-103a-3p to repress its transcription, after which GREM2 was up-regulated, which triggers LATS2 activation and YAP phosphorylation and degradation (Fig. 6). Although the paper was limited in exploring the migration and invasion in the cell perspective, we hope the current nding may offer new insights into HCC control, and we would like to further investigate the roles of above molecules in tumor metastasis in animals and to evaluate other cell behaviors such as proliferation and apoptosis in our future studies.        miR-103a-3p directly targets GREM2. A, GREM2 expression in HCC predicted on GEPIA (http://gepia.cancer-pku.cn/); B, putative binding sites of miR-103a-3p and GREM2 predicted on Starbase; C, GREAM2 expression in HCC tumor tissues and in the paired adjacent normal tissues determined by RT-qPCR (n = 50, paired t test, **, p < 0.01); D, GREM2 expression in THLE-2, HepG2, SK-HEP-1 cells determined by RT-qPCR (one-way ANOVA, **, p < 0.01); E, GREM2 expression in HCC cell lines after pcDNA-GREM2 or siRNA-GREM2 or the corresponding NC vector transfection determined by RT-qPCR (one-way ANOVA, **, p < 0.01); F, number of migrated and invaded HCC cells after pcDNA-GREM2 or siRNA-GREM2 transfection determined by Transwell assays (one-way ANOVA, *, p < 0.05); G, correlation between GREM2 expression and miR-103a-3p expression in 50 pairs of HCC tissues and the adjacent tissues analyzed by Pearson's Correlation Analysis (r = -0.427, **, p < 0.01); H, GREM2 expression in HCC cell lines after miR-103a-3p transfection determined by RT-qPCR (one-way ANOVA, *, p < 0.05); I, binding relationship between miR-103a-3p and GREM2 validated through a dual luciferase reporter gene assay (two-way ANOVA, * p < 0.05); Repetition = 3.

Figure 5
FOXA2 inhibits migration and invasion of HCC cells through the miR-103a-3p/GREM2/LATS2 axis. A, protein levels of LATS2 and YAP and phosphorylation of YAP in each group of cells determined by western blot analysis (one-way, *, p < 0.05); B, migration and invasion abilities of cells transfected with pcDNA-FOXA2 and siRNA-LATS2 determined by Transwell assays (one-way ANOVA, *, p < 0.05).