TMEM161B‐AS1 suppresses proliferation, invasion and glycolysis by targeting miR‐23a‐3p/HIF1AN signal axis in oesophageal squamous cell carcinoma

Abstract Mounting data have shown that long non‐coding RNAs (lncRNAs) widely participate in tumour initiation, development, progression and glycolysis in a variety of tumours. However, the clinical prognosis and molecular mechanisms of TMEM161B‐AS1 in oesophageal squamous cell carcinoma (ESCC) remain still unknown. Here, TMEM161B‐AS1 and HIF1AN were significantly lower in ESCC tissues than in normal samples, and their low expressions were both related to TNM stage, lymph node metastasis and poor prognosis of ESCC patients. Functionally, TMEM161B‐AS1 overexpression or miR‐23a‐3p depletion suppressed the proliferation, invasion and glycolysis as well as reduced glucose consumption and lactate production in ESCC cells. Mechanistically, TMEM161B‐AS1 manipulated HIF1AN expression by competitively sponging miR‐23a‐3p in ESCC cells. MiR‐23a‐3p mimic and HIF1AN siRNA partly reversed cell phenotypes mediated by TMEM161B‐AS1 in ESCC cells. Collectively, TMEM161B‐AS1, miR‐23a‐3p and HIF1AN may be tightly involved in ESCC development and progression as well as patients’ prognosis, and TMEM161B‐AS1/miR‐23a‐3p/HIF1AN signal axis may be a promising target for the treatment of ESCC patients.


| INTRODUC TI ON
Oesophageal cancer (ESCA) is the sixth worst prognosis and the eighth most frequent occurring tumour with high aggressiveness and poor survival in the world. [1][2][3][4] The two main histological subtypes of ESCA are oesophageal adenocarcinoma (EAC) and oesophageal squamous cell carcinoma (ESCC), 5 and ESCC is still the predominant histological type of ESCA worldwide, accounting for more than 80% of all ESCA cases. 6 ESCC is highly prevailing in Asia, especially in Henan Province, China. 7,8 Although a great number of therapeutic strategies have been developed for ESCA patients, the overall fiveyear survival rate is still far from satisfactory. 9,10 Thus, considering the high morbidity and mortality of ESCA, it is badly in need to seek for new biomarkers for early diagnosis and prognostic determination as well as therapeutic strategies for ESCA patients.
Recently, long non-coding RNAs (lncRNAs) have been identified to be involved in tumorigenesis. 11 LncRNAs as a class of noncoding RNAs are more than 200 nucleotides in length, but be lack of protein-coding probability, and are widely implicated in regulation of gene expression at three different levels, including transcriptional level, post-transcriptional level and epigenetic level. 12,13 Functionally, lncRNAs manipulated multiple different complex biological processes by various molecular mechanisms, such as sponging of microRNAs (miRNAs), alternative splicing, recruitment of chromatin-related proteins and the regulation of gene transcription. [14][15][16][17] Clinically, many lncRNAs have been shown to be underlying diagnostic and prognostic biomarkers in a majority of different type tumours. [18][19][20][21] Metabolically, many lncRNAs are considered to be important for tumour cells to survive and grow by altering glucose and lipid metabolisms. [22][23][24][25][26] These data highlighted the biological significance and clinical value of lncRNAs in a variety of tumours, and thus, lncRNAs and its related regulatory pathways may be novel therapeutic strategies for tumour patients.
Currently, TMEM161B-AS1 has been verified to be poor expression in endometrial cancer cell HEC-50 derivatives exhibiting high invasive ability, 27 implying its close association with tumour invasion. However, the functions and exact molecular mechanisms of TMEM161B-AS1 in ESCC are not well defined. In this study, the TMEM161B-AS1 expression patterns in ESCC tissues and cells were detected, and its biological functions in cell proliferation, invasion and glycolysis were confirmed by gain of function (GOF) and loss of function (LOF). Mechanistically, TMEM161B-AS1 served as a molecular sponge by absorbing miR-23a-3p, a vital tumour oncogene that promotes ESCC proliferation, invasion and glycolysis by directly targeting HIF1AN and glycolysis-related proteins. Collectively, our data highlighted the essential glycolysis-related signal axis TMEM161B-AS1/miR-23a-3p/HIF1AN that reprograms ESCC glucose metabolism, and thus, targeting this signal axis may be a promising strategy for the treatment of ESCC patients.

| Bioinformatics assay
TMEM161B-AS1 and miR-23a-3p levels were investigated by StarBase v3.0 online tool, a web-based tool for searching for expression of non-coding RNA or coding RNA. The microarray expression GEO data set related to ESCC (GSE43732) was downloaded from the GEO database for investigation for miR-23a-3p expression. The coding probability of TMEM161B-AS1 was predicted using online tool CPAT from the website http://lilab.resea rch.bcm.edu/cpat/. The binding sites of miR-23a-3p in TMEM161B-AS1 transcript were predicted using DIANA-LncBase v2 online tool. TargetScan and miRDB were performed to predict the possible downstream target genes of miR-23a-3p.

| Cell proliferation assay
Cell proliferation of Eca109 and KYSE30 cells was performed as described in the manufacturer's protocol, and the experiment was con-

| Cell invasion assay
The Transwell assay was employed to examine cell invasion of ESCC cells

| Glucose uptake and lactate production assays
Glucose assay kit (Shanghai Rongsheng Biotech Co., Ltd) and lactate assay kit (Nanjing Jiancheng Bioengineering Institute) were used to determine the glucose consumption and lactate production according to the manufacturer's instructions, respectively. All data obtained were normalized to protein quantitative values.

| Subcellular fractionation
The nuclear RNA and cytoplasmic RNA were extracted by cell nucleus and cytoplasm RNA isolation kit (Beibei, Biotech, Co., Ltd) according to the manufacturer's instruction. Subsequently, qRT-PCR was utilized to determine the gene expression using TMEM161B-AS1, U6 and GAPDH specific primers.

| Fluorescence in situ hybridization (FISH)
TMEM161B-AS1 probe was synthesized and labelled using Cy3 by GenePharma Company. For FISH assay, Eca109 and KYSE30 cells were grown in 24-well plates with glass coverslips for 24 hours.

| Dual-luciferase reporter assay
The interaction of miR-23a-3p with TMEM161B-AS1 or HIF1AN was performed using the dual-luciferase reporter assay system in Eca109 was used to determine the luciferase activity 48h after transfection as described in the manufacturer's protocol.

| RNA immunoprecipitation (RIP)
RNA-binding protein immunoprecipitation kit was purchased from Millipore Company. RIP experiment was performed in ESCC cells according to the manufacturer's protocol. Briefly, RIP lysates were isolated from Eca109 and KYSE30 cells treated with miR-23a-3p mimic or NC mimic and then were applied to immunoprecipitation using either 5μl of anti-Ago2 antibody or 5μl of a normal mouse IgG using RNA-binding Protein Immunoprecipitation Kit. The TMEM161B-AS1 and miR-23a-3p enriched on beads was determined by qRT-PCR using corresponding specific primers.

| Western blot
Total proteins were obtained using RIPA lysis (Solarbio) from ESCC cells, and Bradford method was utilized to determine the protein concentration. The separation of the proteins was performed using SDS-PAGE and then was transferred to PVDF membrane (Millipore Corporation). After blocking with skimmed milk, the primary antibodies against HIF1AN (ab227550, 1:500), HIF-1α (ab51608,

| Statistical assay
GraphPad Prism v8.0 software was performed to examine all experimental data. All data were expressed as mean ±standard deviation (SD). Spearman was used to investigate the non-parametric data, and Pearson was performed to analyse the parametric data. Logrank test was used to determine the statistical difference of survival, and survival curves were drawn using Kaplan-Meier. For the F I G U R E 1 Reduced expressions of TMEM161B-AS1 and HIF1AN in ESCC tissues and their associations with prognosis of ESCC patients. A, StarBase v3.0 online tool was performed to investigate the TMEM161B-AS1 expression in 162 cases of ESCA samples and 11 normal samples. B, qRT-PCR assay for the TMEM161B-AS1 expression in 63 cases of ESCC samples and paired normal samples. C, The expression of TMEM161B-AS1 in ESCC samples with I + II stage and III + IV stage. D, The expression of TMEM161B-AS1 in ESCC samples with lymph node metastasis and without lymph node metastasis. E, Log-rank test was used to determine the association of TMEM161B-AS1 expression with the prognosis of ESCC patients. F, qRT-PCR assay for the HIF1AN expression in 63 cases of ESCC samples and paired normal samples. G, The expression of HIF1AN in ESCC samples with I + II stage and III + IV stage. H, The expression of HIF1AN in ESCC samples with lymph node metastasis and without lymph node metastasis. I, Log-rank test was used to determine the association of HIF1AN expression with the prognosis of ESCC patients. J, StarBase v3.0 online tool was carried out to examine the correlation of TMEM161B-AS1 expression with HIF1AN expression in 162 cases of ESCA samples. K, Pearson correlation assay was performed to detect the correlation of TMEM161B-AS1 expression with HIF1AN expression in 63 cases of ESCC tissues. **P < .01, ***P < .001 and ****P < .0001, indicating statistical significance Mann-Whitney test. The comparison between two groups was determined using a Student's t test, the comparison of ≥3 groups was determined using one-way ANOVA, and then, Bonferroni test was selected for further statistical assay when data sets contain >3 groups. A P value less than .05 was considered to indicate a statistically significant difference.

| Reduced TMEM161B-AS1 and HIF1AN expressions in ESCC tissues and their low expressions predict poor prognosis of ESCC patients
To investigate the expressions of TMEM161B-AS1 and HIF1AN in ESCC tissues and their clinic values, TCGA database and qRT-PCR were employed to examine their levels in ESCC tissues.
TCGA database revealed that TMEM161B-AS1 was significantly down-regulated in ESCA tissues (162 cases) compared with normal samples (11 cases) (P < .0001) ( Figure 1A). Further qRT-PCR assay demonstrated that TMEM161B-AS1 expression in 63 cases of ESCC samples was markedly lower than that in paired 63 cases of normal samples (P < .0001; Figure 1B). Furthermore, TMEM161B-AS1 expression in ESCC patients with III and IV stage as well as lymph node metastasis was dramatically lower than that in ESCC patients with I and II stage as well as without lymph node metastasis (P < .01; Figure 1C,D). Importantly, ESCC patients with high TMEM161B-AS1 expression displayed higher survival rate, compared with those with low TMEM161B-AS1 expression ( Figure 1E). Besides, HIF1AN displayed the similar expression pattern with TMEM161B-AS1 in ESCC tissues and ESCC tissues with different TNM stage and differential metastatic status ( Figure 1F-H). Similarly, ESCC patients harbouring high HIF1AN expression exhibited the higher survival rate than those with low HIF1AN expression ( Figure 1I). The data from TCGA and qRT-PCR revealed that TMEM161B-AS1 presented positive correlation with HIF1AN expression in ESCC samples ( Figure 1J,K). These data suggest that TMEM161B-AS1 and HIF1AN may participate in ESCC progression and may be a promising predictor for the prognosis of ESCC patients. cells ( Figure 2N,O). These findings suggest that TMEM161B-AS1 plays a key regulatory role in ESCC glycolysis.

| TMEM161B-AS1 functions as competitive endogenous RNA (ceRNA) by absorbing miR-23a-3p in ESCC cells
The coding probability of TMEM161B-AS1 was predicted using online tool CPAT (http://lilab.resea rch.bcm.edu/cpat/), and the data revealed that TMEM161B-AS1 had no protein-coding potential ( Figure 3A). Subsequently, the subcellular localization of TMEM161B-AS1 was determined using nuclear-cytoplasmic fractionation, and we found that TMEM161B-AS1 was appeared in both cell nuclear and cytoplasm of Eca109 and KYSE30, but mainly appearing in cytoplasm ( Figure 3B), which was further veri- ( Figure 3D). To validate this predictive result, we then carried out dual-luciferase reporter experiment to verify the possible binding of TMEM161B-AS1 and miR-23a-3p in ESCC cells. We found miR-23a-3p significantly reduced the luciferase activity in Eca109 and KYSE30 cells transfected with TMEM161B-AS1 WT vector, but not in TMEM161B-AS1 MUT vector, suggesting that miR-23a-3p is a direct target of TMEM161B-AS1 ( Figure 3E). Further RIP experiment revealed that relative level of TMEM161B-AS1 and miR-23a-3p in anti-Ago2 antibody group were obviously enhanced compared to IgG group ( Figure 3F,G). Furthermore, the relative enrichment of TMEM161B-AS1 in miR-23a-3p mimic group was markedly enhanced, compared with NC mimic group ( Figure 3H).

| Enhanced miR-23a-3p expression in ESCC tissues and its high expression predicts poor prognosis of ESCC patients
To further investigate the expression pattern of miR-23a-3p in ESCC tissues, TCGA database, GEO data set and qRT-PCR were employed to detect its expression and clinic value in ESCC. The data derived from TCGA and GEO data set GSE43732 revealed that ESCA tissues displayed higher miR-23a-3p level than normal samples ( Figure 4A,B), which was further validated by qRT-PCR in 63 cases of ESCC tissues and paired normal samples ( Figure 4C).

Further investigation revealed that the expression of miR-23a-3p
was not associated with ESCC patients' gender, age, invasion depth and differentiation degree ( Figure 4D-G), but tightly correlated with TNM stage and lymph node metastasis ( Figure 4H,I).
Importantly, high miR-23a-3p level predicted poor prognosis of ESCC patients ( Figure 4J). These data suggest that miR-23a-3p may be implicated in ESCC progression and may be new prognostic factor for ESCC patients.

| miR-23a-3p promotes the glycolysis of ESCC cells by suppressing HIF1AN
To unveil the possible molecular mechanisms of miR-23a-3p in the glycolysis of ESCC cells, we firstly examined the possible downstream target of miR-23a-3p using online tool TargetScan. We found that

| D ISCUSS I ON
Oesophageal cancer is a complex disease that elicits a heavy society burden due to high mortality and poor prognosis. 28 The dissection of molecular mechanisms of ESCA pathogenesis and the discovery of new molecular targets are drastically critical for ESCA patients. Accumulating evidence has highlighted the pivotal regulatory roles of lncRNAs in ESCC progression, and many lncRNAs may be potential biomarkers for the prognosis of ESCC patients. [29][30][31][32][33] In this study, we highlighted the roles of TMEM161B-AS1 in Initially, we found that TMEM161B-AS1 and HIF1AN was presented as low expression in both ESCC tissues and cells.
TMEM161B-AS1 has been verified to be down-regulated in high aggressive endometrial cancer cell line, 27 implying its close association Accumulating evidence has demonstrated that lncRNA functions as ceRNA by sponging miRNAs to enhance the expressions of target genes, which will provide new insights into uncharacterized lncRNAs. 38 To further uncover the regulatory mechanisms transcript, and miR-23a-3p was verified as a direct target of TMEM161B-AS1 by dual-fluorescence reporter assay system and Ago2-RIP experiment. These data imply that TMEM161B-AS1 functions as ceRNA by manipulating miR-23a-3p level in ESCC cells.
To data, miR-23a-3p has been reported to be implicated in the development, progression and prognosis in multiple different tumour types. miR-23a-3p exhibited high expression in endoplasmic reticulum-stressed hepatocellular carcinoma (HCC)-derived exosomes, and its high expression predicted poor prognosis of HCC patients; mechanistically, miR-23a-3p regulated the expression of PD-L1 via PTEN-AKT signalling pathway. 40 In addition, miR-23a-3p functioned as tumour oncogene in renal cell carcinoma, and miR-23a-3p silence inhibited the proliferation and mobility in RCC cells by targeting PNRC2. 41 Another evidence revealed miR-23a-3p acted as tumour suppressor, its low expression predicted poor clinical outcome, and miR-23a-3p overexpression significantly inhibited the proliferation, invasion and tumorigenicity by targeting adenylate cyclase 1 (ADCY1). 42 Besides, miR-23a-3p was markedly decreased, and might be a potential prognostic indicator in oral squamous cell carcinoma. 43 The differential expression patterns of miR-23a-3p in a variety of tumours prompted us to further unveil its expression status in ESCC tissues and cells and possible molecular mechanisms.
Here, we revealed high miR-23a-3p expression in ESCC tissues and cells, and its high expression was tightly associated with TNM stage, lymph node metastasis and poor prognosis of ESCC patients.
Functionally, miR-23a-3p depletion suppressed the proliferation,  HIF1AN reversed the down-regulation of HIF1AN and up-regulation of glycolysis-related proteins HIF-1α, HK2, PFKM and LDHA regulated by TMEM161B-AS1 in Eca109 and KYSE30 cells at 48 h after transfection. *P < .05, **P < .01, ***P < .001 and ****P < .0001, indicating statistical significance F I G U R E 8 TMEM161B-AS1 suppresses cell proliferation, invasion and glycolysis by manipulating miR-23a-3p/ HIF1AN signal axis. TMEM161B-AS1 exhibits the low level in ESCC tissues and cells, and its overexpression inhibits the expression of miR-23a-3p expression in ESCC cells and further results in the up-regulation of HIF1AN expression in ESCC cells, which further triggers the suppression of ESCC glycolysis importantly, miR-23a-3p played an important role in ESCC glycolysis by targeting HIF1AN, a glycolysis-related regulatory protein, which was further implicated in ESCC progression.
On the basis of the information stated above, we hypothesized that TMEM161B-AS1 mediated the suppressions of cell proliferation, invasion and glycolysis was dependent on miR-23a-3p or HIF1AN level in ESCC cells. To this end, pcDNA3.1-TMEM161B-AS1 combined with miR-23a-3p mimic or HIF1AN siRNA as well as TMEM161B-AS1 siRNA along with miR-23a-3p inhibitor or pcDNA3.1-HIF1AN were applied to rescue experiment for further elucidation of TMEM161B-AS1 functions in ESCC cells. Our current data suggest that the suppression of cell proliferation, invasion and glycolysis triggered by TMEM161B-AS1 overexpression was partly reversed by miR-23a-3p mimic or HIF1AN siRNA, and opposite data were obtained when TMEM161B-AS1 siRNA was combined with miR-23a-3p inhibitor or pcDNA3.1-HIF1AN. These findings indicate that TMEM161B-AS1 plays a pivotal regulatory role in cell proliferation, invasion and glycolysis by manipulating miR-23a-3p/HIF1AN/ glycolysis-related enzyme pathway in ESCC.
In summary, our current data highlight the roles of The study was conducted in accordance with the International

Ethical Guidelines for Biomedical Research Involving Human
Subjects. All patients provided informed consent to participate in the study.

CO N FLI C T O F I NTE R E S T S
The authors declare 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 used to support the findings of this study are available from the corresponding author upon reasonable request.