LncRNA FAM83H‐AS1 promotes oesophageal squamous cell carcinoma progression via miR‐10a‐5p/Girdin axis

Abstract Long non‐coding RNAs (lncRNAs) have been well demonstrated to emerge as crucial regulators in cancer progression, and they can function as regulatory network based on their interactions. Although the biological functions of FAM83H‐AS1 have been confirmed in various tumour progressions, the underlying molecular mechanisms of FAM83H‐AS1 in oesophageal squamous cell carcinoma (ESCC) remained poorly understood. To address this, we treated human oesophageal cancer cell line Eca109 cells with TGF‐β and found FAM83H‐AS1 was notably overexpressed. In the present study, FAM83H‐AS1 was observed to be significantly up‐regulated in ESCC tissues and was associated with TNM stage, pathological differentiation and lymph node metastasis. FAM83H‐AS1 reinforced oesophageal cancer cell proliferation, migration and invasion, and participated in epithelial‐to‐mesenchymal transition (EMT) process at mRNA and protein levels. In addition, a concordant regulation between FAM83H‐AS1 and its sense strand FAM83H was detected at the transcriptional and translational levels. Furthermore, FAM83H‐AS1 could act as competing endogenous RNA to affect the expression of Girdin by sponging miR‐10a‐5p verified by RIP and luciferase reporter assays. Consequently, the study provided a unique perspective of FAM83H‐AS1 in ESCC progression, which may be considered as potential biomarker and therapeutic target for ESCC therapy.

chromatin modification and gene transcription in the nucleus, and modulating mRNA stability, translation and post-translational modification in the cytoplasm. 5 LncRNAs located in the cytoplasm can serve as competitive endogenous RNAs (ceRNAs), which could sponge miRNAs through competition for shared miRNAs, thereby imposing an additional regulation on miRNA targets at post-transcriptional level. 6 Epithelial-to-mesenchymal transition (EMT) is a typically fundamental transdifferentiation process in development which enables cancer cell invasion, contributes to cancer stroma formation, generates stem-like tumour-initiating cells and increases drug resistance.
Among a myriad of EMT-regulating factors discovered in the cancer microenvironment, transforming growth factor-β (TGF-β) has been shown to be a potent signal to initiate and drive EMT. 7  FAM83H-AS1 and its cognate sense strand FAM83H are head-tohead located on 8q24. Natural antisense transcripts (NATs) are defined as RNA sequences that originate from complements of their endogenous sense counterparts in cis or trans. 9,10 Notably, some natural antisense lncRNAs were reported to exert regulatory effects on expression of their sense protein-coding genes. 11,12 But the expression level and correlation between FAM83H-AS1 and FAM83H in ESCC were not well characterized. Accumulating evidence was illustrating that FAM83H-AS1 was overexpressed in various cancer types that promoted cell growth and metastasis by multiple molecular mechanisms, [13][14][15][16] while the exact mechanism of FAM83H-AS1 in ESCC was largely unclear and its prospect as therapeutic target for ESCC was still unexplored.
In the present study, we aimed at providing an integrated analysis on the expression and correlation between FAM83H-AS1 and FAM83H, the potential biological function of these antisense-sense strands and downstream regulatory mechanism of FAM83H-AS1 in the pathogenesis of ESCC, as well as its role in TGF-β-induced EMT.

| Patients and specimens
All the 67 pairs of ESCC tissues and corresponding normal tissues were taken from the surgical specimens of ESCC patients from the years of 2015 to 2017 in the Fourth Affiliated Hospital of Hebei Medical University. Informed consent was received from all patients who were not given any radiotherapy or chemotherapy before operation. According to the standard of American Joint Committee on Cancer system, histological grade was staged.
Information on clinical data and clinicopathological characteristics was available from hospital recordings and is summarized in Table S1. Smokers were defined as former or current individuals smoking at least five cigarettes per day for 2 years or longer. 17 Individuals with at least one first-degree relative or at least two second-degree relatives having oesophageal/cardia/gastric cancer were defined as having family history. Ethical consent was granted from the Ethics Committee of the Fourth Affiliated Hospital of Hebei Medical University.

| Cell culture and treatment
Human oesophageal cancer cell lines Kyse150, Kyse170, TE1 and Eca109 were purchased from American Type Culture Collection and were cultured in RPMI 1640 (Invitrogen) medium containing 10% foetal bovine serum (Invitrogen) at 37°C in an atmosphere containing 5% CO 2 . The cells were treated with 10 ng/mL of recombinant TGF-β1 (R&D Systems) for 7 days with the medium replenishment every 2 days.

| RNA isolation and quantitative real-time polymerase chain reaction (qRT-PCR) assay
Total RNA from the tissues and cells was isolated using TRIzol reagent (Invitrogen) in accordance with the manufacturer's instructions. Transcriptor First Strand cDNA Synthesis Kit (Roche) was used to generate cDNA according to the manufacturer's protocol.
qRT-PCR was performed in the StepOne Real-Time PCR System (Applied Biosystems) using GoTaq ® qPCR Master Mix (Promega).
GAPDH and U6 snRNA were employed as endogenous controls for mRNA/lncRNA and miRNA, respectively. The relative expression level of RNAs was calculated using the 2 −ΔΔCT method. 18 Each specimen was tested in triplicate. Primer sequences are displayed in Table S2.

| Subcellular fractionation
The nuclear and cytoplasmic fractions of oesophageal cancer cell lines were isolated by PARIS™ Kit Protein and RNA Isolation System (Invitrogen) according to the manufacturer's protocol.

| Cell transfection
The shRNAs targeting FAM83H-AS1 and the pcDNA3.1-FAM83H-AS1 were designed and synthesized by GenePharma and Sangon Biotech, respectively. The miR-10a-5p mimics, inhibitor and negative control were purchased from GenePharma. The FAM83H siRNAs and si-NC were synthesized by General Biosystems. Transfections were performed using Lipofectamine 2000 Reagent (Invitrogen) according to the manufacturer's protocol. The sequences of four shRNAs, three siRNAs and miR-10a-5p mimics and inhibitor are listed in Table S3.

| Cell proliferation assay
The ability for cellular proliferation was detected by MTS assay and clone formation assay. The MTS assay was measured using CellTiter96 ® AQ ueous One Solution Cell Proliferation Assay kit (Promega). For MTS assay, the transfected cells were seeded into 96-well plate with 1 × 10 3 per well. After incubation at 0, 24,48,72 and 96 hours, cells of each well were added with 20 μL (500 μg/mL) of MTS reagent and incubated at CO 2 incubator for 2 hours. The optical density was measured with a microplate reader at a wavelength of 490 nm. For clone formation assay, 3 × 10 3 cells per well following transfection for 24 hours were inoculated into a six-well plate and regularly cultured for 1 week. More than 50 cells were considered to be one clone, and the numbers of clone were counted under a microscope.

| Transwell migration and invasion assays
Cell migration assay was conducted using non-Matrigel-coated chambers (Corning) with 8-μm pore membranes. A total of 1 × 10 5 cells per well were seeded into the upper compartment of chamber.
After 24 hours of incubation at 37°C, the invasive cells located on the lower surface of the membrane were counted in five randomly selected visual fields using a Leica DMI4000B microscope. For invasion assay, the upper surface of the membrane was pre-coated with 50 μL 1× Matrigel ® Basement Membrane Matrix (Corning) to form a matrix barrier; the remaining steps were used for invasion assay as described above.

| Western blot analysis
Total proteins were extracted from transfected cells using RIPA lysis buffer containing PMSF (Solarbio) and protease inhibitor cocktail

| Statistical analysis
All data were expressed as mean ± SD. The significance of differences between two groups or among multiple groups was determined by Student's t test or one-way ANOVA, respectively.
Bivariate correlations between study variables in tissues were calculated by Spearman correlation analysis. All statistical tests were two-sided, and P < .05 was considered to be statistically significant.

| FAM83H-AS1 emerges as a potential oncogenic lncRNA and is associated with clinicopathological characteristics in ESCC
Based on scanning the NCBI and GEPIA data set, the relative expression levels of FAM83H-AS1 in different normal tissues and in most of the tumour types were detected ( Figure S1A,B). By evaluating FAM83H-AS1 expression in 67 pairs of ESCC tissues and corresponding normal tissues, it was confirmed that FAM83H-AS1 expression level was significantly elevated in ESCC tissues ( Figure 1A).
Additionally, the expression level of FAM83H-AS1 in a panel of human oesophageal cancer cell lines was performed, which was remarkably higher in all oesophageal cancer cell lines, especially in Kyse150 and TE1 cells ( Figure 1B). It was identified that high expression level of FAM83H-AS1 was closely associated with lymph node metastasis, TNM stage and pathological differentiation ( Figure 1C). LncRNAs have been shown to play functional roles in both the nucleus and cytoplasmic compartments. FAM83H-AS1 was found to be predominantly located in the cytoplasm of oesophageal cancer cells by subcellular fractionation assay ( Figure 1D). Coding Potential Calculator and Coding Potential Assessment Tool were further used to analyse the coding potential of FAM83H-AS1, and no protein-coding potential of FAM83H-AS1 was found ( Figure S1C,D).

| FAM83H is significantly up-regulated in ESCC patients
A NCBI search identified that FAM83H-AS1 was in a head-to-head orientation relative to FAM83H ( Figure 1E). As indicated by NCBI and GEPIA data set, the relative expression levels of FAM83H in normal tissues and in various tumour types were similar to FAM83H-AS1 ( Figure S1E,F). Subsequently, qRT-PCR analysis detected increased mRNA expression level of FAM83H in ESCC tissues and FAM83H exhibited concordant co-regulation with FAM83H-AS1 ( Figure 1F,G).
Meanwhile, the expression level of FAM83H was significantly higher in Kyse150 and TE1 cells than other tested cell lines, Kyse150 and TE1 cells were selected for subsequent experiments ( Figure 1H). In addition, analysis of the correlation between FAM83H expression and clinicopathological characteristics showed that FAM83H expression level was intimately associated with pathological differentiation ( Figure 1I).

| The effect of FAM83H-AS1 and FAM83H on oesophageal cancer cell proliferation, migration and invasion
To determine the biological function of FAM83H-AS1 and If FAM83H in regulating biological processes was in accordance with the oncogenic role of FAM83H-AS1, the functional experiments of FAM83H were conducted in Kyse150 and TE1 cells. We screened si-FAM83H-1 with high interference efficiency compared with the non-targeting control si-NC group ( Figure 3A). As displayed in Figure 3B-D, siRNA-mediated FAM83H knockdown notably inhibited cell proliferation, migration and invasion, which largely phenocopied sh-FAM83H-AS1 inhibition in oesophageal cancer cells.

| FAM83H-AS1 regulates FAM83H at mRNA and protein levels
To identify the correlation between the levels of FAM83H-AS1 and

| FAM83H-AS1 is up-regulated in TGFβ-induced Eca109 cells and contributes to the process of EMT
Because EMT is a crucial step of metastasis, it is of great interest to examine whether FAM83H-AS1 regulates the migration and invasion of oesophageal cancer cells via EMT. We first measured the cell phenotype after incubation with TGF-β and found that TGF-βtreated Eca109 cells underwent morphological changes to a spindle-shaped appearance ( Figure 4A). Moreover, the cells displayed decreased expression of E-cadherin, as well as up-regulated expression of N-cadherin, vimentin, Snail and Twist1 ( Figure 4B). These results suggested that the cells displayed EMT-associated signatures and exhibited a proper biological response to TGF-β treatment.
Additionally, the expression level of FAM83H-AS1 was assessed and up-regulation of FAM83H-AS1 was detected in TGF-β-treated cells compared with untreated cells ( Figure 4C). Subsequently, knockdown of FAM83H-AS1 was found to promote the expression of E-cadherin and inhibit the expression of N-cadherin, vimentin, Snail The correlation between FAM83H-AS1 and miR-10a-5p expression. F, Relative expression of miR-10a-5p detected by transfection with miR-10a-5p mimics or inhibitor. G, MTS assay and H, clone formation assay were conducted by transfection with miR-10a-5p mimics or inhibitor. I, Transwell migration and J, invasion assays were performed by transfection with miR-10a-5p mimics or inhibitor (magnification, ×200). Data are shown as mean ± SD; *P < .05 and **P < .01

F I G U R E 4 FAM83H-AS1 is up-regulated in TGF-β-treated Eca109 cells and contributes to EMT process. A, Cell morphology in TGF
and Twist1 at transcriptional level ( Figure 4D), while overexpression of FAM83H-AS1, compared with negative control, could reduce E-cadherin expression and enhance expression level of N-cadherin, vimentin, Snail and Twist1 ( Figure 4E). As shown in Figure 4F, FAM83H-AS1 also regulated EMT-related markers at protein level.
In summary, FAM83H-AS1 may be an EMT-related lncRNA and participate in EMT of oesophageal cancer cells.

| FAM83H-AS1 sponges miR-10a-5p through direct binding in oesophageal cancer cells
FAM83H-AS1 was mainly located in the cytoplasm of oesophageal cancer cells; therefore, we hypothesized that FAM83H-AS1 might also function as a molecular sponge to competitively bind certain miRNAs. According to the prediction in online databases (RAID v2.0), miR-10a-5p was found to contain two potential binding sites to the FAM83H-AS1 sequence ( Figure 5A) and chosen for subsequent experiments for its tumour-suppressive role in ESCC.
The expression level of miR-10a-5p was down-regulated in ESCC tissues and oesophageal cancer cells, as well as closely associated with TNM stage and lymph node metastasis ( Figure 5B-D). In addition, a significant inverse correlation between FAM83H-AS1 and miR-10a-5p expression in ESCC tissues was found ( Figure 5E). The efficiency of miR-10a-5p mimics and inhibitor was validated prior to further analysis ( Figure 5F). As shown in Figure 5G-J, the overexpression of miR-10a-5p by transfection with miR-10a-5p mimics hindered the proliferation, migration and invasion of Kyse150 and Eca109 cells, whereas transfection of miR-10a-5p inhibitor in Kyse150 and Eca109 cells displayed opposite effects.
Consistently, the relative luciferase activity of FAM83H-AS1 wild-type was obviously decreased after cotransfection with miR-10a-5p mimics, but did not affect the activity of mutant type, which further verified that miR-10a-5p is a direct target of FAM83H-AS1 ( Figure 6D).

| miR-10a-5p directly targets Girdin in oesophageal cancer cells
By using four independent miRNA target-predicting algorithms (DIANA, TargetScan, Starbase and miRDB), potential downstream target genes of miR-10a-5p were predicted ( Figure 6E). Among these 50 predicted target genes, Girdin (also named CCDC88A) attracted our attention because of its critical role in the migration and invasion of cancer cells. Girdin regulates actin reconstruction and Akt-dependent cell motility, and involves in remodelling actin cytoskeleton which is essential for cell migration. 19 The conserved binding site of Girdin 3' UTR for miR-10a-5p is illustrated in Figure 6F.
Girdin expression was found to be higher in ESCC tissues than that in corresponding normal tissues ( Figure 6G) and was negatively correlated with miR-10a-5p ( Figure 6H). Meanwhile, Girdin expression was correlated with TNM stage, lymph node metastasis and pathological differentiation in ESCC tissues ( Figure 6I). Subsequently, overexpression of miR-10a-5p dramatically decreased the expression level of Girdin, while down-regulation of miR-10a-5p markedly exhibited the opposite effect in oesophageal cancer cells ( Figure 6J).
Luciferase reporter assay manifested that enforced expression of miR-10a-5p reduced the luciferase activity of pmirGLO-Girdin-3′ UTR wild-type vector while showed no obviously effect on the luciferase activity of pmirGLO-Girdin-3′ UTR mutant type in Eca109 cells ( Figure 6K), indicating the indeed regulatory role of miR-10a-5p on Girdin mRNA expression through direct binding to its 3′ UTR.

| FAM83H-AS1 positively regulates Girdin in a miR-10a-5p-dependent manner
Due to the fact that FAM83H-AS1 shared common binding sites of miR-10a-5p with Girdin, we wondered whether FAM83H-AS1 could modulate Girdin dependent on miR-10a-5p. In oesophageal cancer cells, down-regulation of FAM83H-AS1 significantly decreased Girdin expression, whereas miR-10a-5p inhibitor overcame such a decrease. Similarly, miR-10a-5p mimics could abrogate the increased effect of FAM83H-AS1 overexpression on Girdin expression ( Figure 7A). Besides, a dramatically positive correlation F I G U R E 6 FAM83H-AS1 sponges miR-10a-5p and miR-10a-5p directly targets Girdin in oesophageal cancer cells. A, Relative expression of miR-10a-5p in FAM83H-AS1 knockdown or overexpression cells. B, Relative expression of FAM83H-AS1 in miR-10a-5p mimics or inhibitor transfected cells. C, The MS2-RIP method identified the direct binding between FAM83H-AS1 and miR-10a-5p. D, The effect of miR-10a-5p mimics on luciferase activity of wild-type and mutant-type FAM83H-AS1 vectors observed by dual-luciferase reporter assay. E, The numbers of miR-10a-5p targeting the potential same genes (including Girdin) drawn by Venn diagram. F, Schematic representation of the potential binding sites of miR-10a-5p on Girdin 3′ UTR. G, Relative expression of Girdin in 67 pairs of ESCC tissues and corresponding normal tissues confirmed by qRT-PCR method. H, The correlation between Girdin and miR-10a-5p expression. I, Relative expression of Girdin in different subgroups. J, The regulation of miR-10a-5p on Girdin expression detected by qRT-PCR method. K, The effect of miR-10a-5p mimics on luciferase activity of wild-type and mutant-type Girdin 3′ UTR vectors observed by dual-luciferase reporter assay. Data are shown as mean ± SD; *P < .05 and **P < .01 between FAM83H-AS1 and Girdin expression was identified in 67 pairs of ESCC tissues ( Figure 7B). In gain-and loss-of-function experiments, miR-10a-5p inhibitor could partially rescue the inhibitory effect of FAM83H-AS1 knockdown on cell proliferation, migration and invasion capacity. Reciprocally, miR-10a-5p mimics could abolish biological functions caused by FAM83H-AS1 overexpression ( Figure 7C-F). Overall, these results revealed a vital role of FAM83H-AS1 in modulating Girdin by competitively binding with miR-10a-5p.

| D ISCUSS I ON
There is obvious evidence that the proverbial lncRNAs take up a significant portion operating as either oncogene or tumour suppressor in the pathological development of ESCC. In our study, we verified that FAM83H-AS1 and its sense transcript FAM83H were consistently up-regulated, and concordant co-regulation was  to regulate APC expression and the Wnt/β-catenin pathway. Huang F I G U R E 7 FAM83H-AS1 positively regulates Girdin in a miR-10a-5p-dependent manner. A, Relative expression of Girdin following cotransfection with miR-10a-5p inhibitor or mimics in FAM83H-AS1 knockdown or overexpression cells. B, The correlation between FAM83H-AS1 and Girdin expression. C, MTS assay and D, clone formation assay were rescued by cotransfection with miR-10a-5p inhibitor or mimics in FAM83H-AS1 knockdown or overexpression cells. E, Transwell migration and F, invasion assays were confirmed following cotransfection with miR-10a-5p inhibitor or mimics in FAM83H-AS1 knockdown or overexpression cells (magnification, ×200). Data are shown as mean ± SD; *P < .05 and **P < .01 et al 39 demonstrated that TRPM2-AS took important regulatory parts in gastric carcinoma development by functioning as a ceRNA to regulate HMGA1 via sponging miR-195. However, there are no reports concerning that FAM83H-AS1 acted as a ceRNA of miRNAs in ESCC, and then we investigated this potential regulatory mechanism of FAM83H-AS1 in the cytoplasm. In the present study, miR-10a-5p was selected as the potential miRNA due to possessing the complementary binding sites with FAM83H-AS1. MiR-10a has been previously reported to be a tumour suppressor by analysing the miRNA microarray in ESCC. 40 Combined with RIP assay and luciferase reporter assay, it confirmed the direct binding between FAM83H-AS1 and miR-10a-5p, implying that FAM83H-AS1 acted as a molecular sponge of miR-10a-5p.
Generally, miRNAs are ubiquitous post-transcriptional regulators that impact RNA stability and translation rate by binding to mRNAs in a sequence-specific manner [41][42][43] ; therefore, this regulation for target mRNA becomes an important part of the ceRNA network.
In this study, Girdin was finally screened out as the specifically target mRNA of miR-10a-5p proved by luciferase reporter experiment.
Girdin is a novel component of the PI3K/Akt signalling pathway that is a core-signalling transduction pathway in cancer progression.
Previous study has confirmed that Girdin exhibited an enhanced expression in ESCC and presented a positive role in oesophageal cancer cell proliferation, migration and invasion. 44 Meanwhile, we detected that FAM83H-AS1 positively regulated Girdin expression abrogated by ectopic expression of miR-10a-5p. Taken together, these results supported that FAM83H-AS1, miR-10a-5p and Girdin formed a ceRNA regulatory network in the progression of ESCC.
In conclusion, the current study demonstrated that TGF-βinduced FAM83H-AS1 served as a novel oncogene in ESCC and marked concordant expression with its cognate sense counterpart FAM83H. Additionally, FAM83H-AS1 was proved to regulate EMT process, and acted as a ceRNA in competitively sponging miR-10a-5p to enhance Girdin expression. Furthermore, these findings provided novel insights into the underlying mechanism of the aggressive biological behaviour of ESCC, which highlighted a potential target for ESCC therapy.

ACK N OWLED G EM ENTS
This study was supported by Grants from the National Natural

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.