NF-κB-Mediated lncRNA AC007271.3 as A ceRNA Promotes Carcinogenesis of Oral Squamous Cell Carcinoma by Regulating Slug


 Background: Oral squamous cell carcinoma (OSCC) is the most common oral cancer. Our previous studies confirmed that dysregulation function of long non-coding RNA (lncRNA) AC007271.3 was associated with a poor prognosis and overexpression of AC007271.3 promoted cell proliferation, migration, invasion and inhibited cell apoptosis in vitro, and promoted tumor growth in vivo. However, the underlying mechanisms of AC007271.3 dysregulation remained obscure.Methods: Bioinformatics databases were used to predicted the potential down-stream targeted of AC007271.3 and verified by dual luciferase reporter assay. Core promoter region of AC007271.3 was identified by luciferase activity assay and the potential transcription factor on it was verified by ChIP assay. Western blot and qRT-PCR were performed to detect the protein and messenger RNA (mRNA) levels, respectively. Animal experiments confirmed the metastatic ability in vivo.Results: AC007271.3 functioned as competing endogenous RNA (ceRNA) by binding to miR-125b-2-3p and upregulated the expression of Slug, which is a direct target of miR-125b-2-3p. AC007271.3 enhanced the expression of Slug and inhibited the expression of E-cadherin to promote the migration and invasion in OSCC cells. The expression of AC007271.3 was promoted by canonical nuclear factor-κB (NF-κB) pathway. Conclusion: Our study showed that the classical NF-κB pathway-activated AC007271.3 regulates EMT by miR-125b-2-3p / Slug / E-cadherin axis to promote the development of OSCC, implicating it as a novel potential target for therapeutic intervention in this disease.

micmic NC) were co-transfected into HEK 293T cells. The Luc-Pair Dual-Luciferase Assay Kit 2.0 (GeneCopoeia, MD, USA) was used to detect the re y and renilla luciferase activity. Fire y/renilla luciferase activity ratio represents the relative luciferase activities.
The product of the two scores was considered as the nal score. The nal scores were divided into two levels: 0-5 (Slug low expression) and more than 5 (Slug high expression).

Animal experiment
Lentivirus-NC and lentivirus-AC007271.3 were purchased from Obio Technology Corp., Ltd. (Shanghai, China) and transfected into SCC9 according to the operating instructions. Puromycin was used to select the AC007271.3 stably expressed cells (SCC9-AC007271.3) and its control cells (SCC9-NC). Five-weeksold female BALB/c nude mice were randomly divided into three groups (blank, SCC9-AC007271.3 and SCC9-NC). Approximately 1 × 10 6 SCC9-AC007271.3 cells or SCC9-NC cells were resuspended in 500 μL of PBS and then injected via tail vein. Simultaneously, injection of 500 μL of PBS without cells suspension was regarded as blank group. After 8 weeks, all mice were sacri ced, and the visible lung metastases were counted. Lung metastases were excised for HE or IHC staining. The expression of AC007271.3 and miR-125b-2-3p were detected by qRT-PCR and the expression of Slug and EMT markers were inspected by western blot. This animal experiment was approved by Southern Medical University Experimental Ethics Committee (L2018199), and all BALB/c nude mice were purchased from the Animal Care Unit of Southern Medical University.

AC007271.3 Promoter Region Cloning
The sequences of the 2000bp before the 5'UTR of AC007271.3 were regarded as promoter region and obtained from Ensemble database (http://asia.ensembl.org/index.html). According to the sequences, we designed several primers with speci c primers containing restriction enzyme protection sites of KpnI and Bgl II, respectively, for ampli cation the fragments (-1998/-2, -1508/-2, -1000/-2, -519/-2). Genomic DNA was used as a template to amplify by KOD-Plus-Neo (TOYOBO, Osaka, Japan) ampli cation enzyme. The reaction conditions were set following the instruction. 1% agarose gel electrophoresis was performed to select the bands in correct size and con rmed through sequencing (Sangon, Shanghai, China). The pGL3-Basic vector and correct fragments were digested with KpnI (TakaraBio, Dalian, China) and Bgl II (TakaraBio, Dalian, China) restriction enzymes. DNA ligation kit ver.2.1 (TakaraBio, Dalian, China) was used to connect the fragments and pGL3-Basic vector. DH5α competent cells were used for transformation. After the monoclonal colony sequence con rmation, the plasmids were extracted by using HiPure Plasmid EF Mini Kit (Magen, Guangzhou, China) for identi cation of core promoter of AC007271.3 by Dual luciferase reporter assay.
Chromatin Immunoprecipitation (ChIP) assay ChIP assay were carried out by Magna ChIP TM A / G Kits (Merck, Darmstadt, German) according to the manufacturer's protocols. Brie y, SCC9 cells were crosslinked with 1% formaldehyde, glycine quenching and lysed. The nuclear DNA was fragmented by sonication. NFKB1 (p50) speci c antibody (1:50, CST, MA, USA, #13586) was used for immunoprecipitation of cross-linked protein / DNA. Protein / DNA complexes were eluted and reverse cross-linked into free DNA. Puri ed DNA was used for qRT-PCR analysis to detect the enrich fragments.

Statistical analysis
All experimental assays were performed in triplicate. All data were presented as mean ± standard deviation (SD) of triplicate replicates and analyzed by SPSS statistics 20.0 (Chicago, USA). All statistical charts were manufactured in Graphpad Prism 7.0 (CA, USA). Student's t-test was performed to compare the differences of two groups. The correlation between miR-125b-2-3p (or Slug) and clinicopathological features was analyzed by using Fisher's exact tests. Pearson's correlation coe cient analysis was used to analyze the correlation between AC007271.3 and miR-125b-2-3p. Kaplan-Meier (K-M) curve was applied for detecting the comparison of overall survival in two groups.

Overexpression of miR-125b-2-3p inhibited migration and invasion in OSCC.
To investigate the roles of miR-125b-2-3p in OSCC carcinogenesis, rstly, the relative expression levels of miR-125b-2-3p in SCC9 and SCC15 cells were detected by qRT-PCR. The results showed that the expression levels of miR-125b-2-3p were signi cantly lower in OSCC cells than in HOK cells (Figure2 A). What's more, compared to the matched adjacent normal tissues, the expression levels of miR-125b-2-3p were down-regulated in 82 OSCC tissues (Figure2 B). Similar analysis result was also acquired from GEO database (Figure2 C). We further investigated the association between miR-125b-2-3p expression and clinicopathological features of 82 OSCC patients. The results indicated that the expression level of miR-125b-2-3p was negatively correlated with TNM classi cation (p=0.0429) and lymph node metastasis (p=0.0259) in OSCC patients (Supplemental Table S1).
Plenty of evidence showed that miRNAs could target the 3' untranslated region (UTR) of genes to suppress their expression. To investigate the molecular mechanism of miR-125b-2-3p suppressing the migration and invasion in OSCC cells, we predicted the target gene of miR-125b-2-3p based on the online TargetScan V7.2 software. Slug was selected as a candidate gene for its high binding score and 8mers seed sequences (Supplemental Data 1). Dual luciferase reporter assay results uncovered that obviously reduced luciferase activity was observed in Slug-wild type, while no remarkable change of luciferase activity in Slug-mutation type (Figure3 A). Furthermore, overexpression of miR-125b-2-3p decreased the protein level of Slug, while the result was just the opposite after knocking down miR-125b-2-3p (Figure3 B). However, Slug mRNA had no remarkable change no matter whether miR-125b-2-3p was overexpressed or low-expressed (Supplemental Data 2).

Slug was overexpressed in OSCC and silencing Slug inhibited migration and invasion in OSCC cells.
Slug is an epithelial-mesenchymal transition (EMT) related transcription factor which could be involved in the invasion and distant metastasis of various tumor cells including OSCC [20,21] . To veri ed the potential biological functions of Slug in OSCC, Slug expression pro les and corresponding clinical data of 319 OSCC patients and 32 normal controls were downloaded from TCGA database. Bioinformatics analysis revealed that Slug expression was remarkably up-regulated (p<0.0001) (Figure3 C) and higher expression of Slug meant poor survival and prognosis (p=0.02847) (Figure3 D). To further validate these results, the Slug expression was detected in OSCC tumor tissues and matched ANTs with immunohistochemistry. Compared to the ANTs, Slug expression was signi cantly higher in OSCC tumor tissues (Figure3 E).
Clinicopathological characteristics analysis showed that high expression of Slug was positive correlated with TNM classi cation p=0.0301 , lymph node metastasis p=0.0306 and differentiation in OSCC patients p=0.0465 (Supplemental Table S2). Furthermore, we investigated the expression of Slug The biological function and molecular role of lncRNAs is closely associated its subcellular localization.
We previously identi ed that AC007271.3 was predominantly located in the cell cytoplasm by using RNA-FISH assay, which indicated that AC007271.3 might function as a ceRNA of miR-125b-2-3p and regulate Slug. To further determine the relationship among AC007271.3, miR-125b-2-3p and Slug, rstly, Pearson's correlation analysis was performed to uncover the association between AC007271.3 and miR-125b-2-3p expression in 82 OSCC tissues. The results indicated that the expression of miR-125b-2-3p was inversely injected into the tail vein of BALB/c nu/nu mice to establish a pulmonary metastasis model. 8 weeks later, the mice were sacri ced and lung tissues were photographed. We could clearly see more pulmonary metastatic nodules in SCC9-AC007271.3 group than in SCC9-NC group, which suggesting that AC007271.3 could signi cantly promote OSCC metastasis to lung tissues (Figure5 A-B). QRT-PCR analysis showed that the expression of AC007271.3 in pulmonary metastases was dramatically increased, while miR-125b-2-3p's RNA level was signi cantly decreased in the SCC9-AC007271.3 group in contrast to that in the SCC9-NC group (Figure5 C). Furthermore, H&E staining of lung sections indicated that overexpression of AC007271.3 increased the number and size of pulmonary metastases. The results of IHC staining and western blot also revealed a signi cant upregulation of Slug in pulmonary metastases of SCC9-AC007271.3 group compared with that in SCC9-NC group (Figure5 D-E). These results indicated that AC007271.3 contributed to OSCC metastasis in vivo.
In order to explore the upstream regulation mechanism of AC007271.3, we rstly consider the 2000bp before the 5'UTR of AC007271.3 to be the promoter region and obtained the promoter region sequences on Ensemble database. According to these sequences, we designed several primers (Supplemental Data 3) to constructed 4 fragment-by-fragment deletion pGL3 vectors of AC007271.3 promoter (-1998/-2, -1508/-2, -1000/-2, -519/-2). Luciferase activity assay was performed for the identi cation of core promoter region of AC007271.3 and the unidirectional deletion at -1000/-519 caused signi cant reduction in the luciferase activity (Figure6 A) so we preliminarily regarded -1000/-519 region as the core promoter of AC007271.3. Next, we predicted that NF-κB may be the potential transcription factor on the core promoter of AC007271.3 by using three different programs (Match -1.0, Patch 1.0 and AliBaba 2.1) on Gene-regulation online website. NF-κB is a dimer formed by different kinds of subunits. The prediction on JASPAR database indicated that NFKB1 may be the most critical subunit (Figure6 B, Supplemental Table  S3). To verify our conjecture, we designed 3 pairs of primers according to the predicted binding sites (AC007271.3-promoter Site 1: -953/-943, AC007271.3-promoter Site 2: -854/-842, AC007271.3-promoter Site 3: -581/-571) and carried out ChIP-qRT-PCR analysis, which suggested that NFKB1 enriched on the AC007271.3-promoter Site 2 (Figure6 C). NFKB1 is also known as p50, and p50/p65 is the most common type of NF-κB dimer, which is activated by TNF-α in classical NF-κB pathway and plays a positive regulatory role during gene transcription. Therefore, we measured the mRNA level of AC007271.3, miR-125b-2-3p and Slug after TNF-α (10ng/ml) treatment for the indicated times and the results showed that the mRNA level of AC007271.3 and Slug were stably activated after 0.5h and 2h, respectively. On the other hand, interestingly, the mRNA level of miR-125b-2-3p strongly increased after 0.5h but gradually decreased with the increasing time until 72h (Figure6 D). Western blot analysis result indicated that the protein level of Slug was also up-regulated after TNF-α treatment for 72h (Figure6 E), and at the same time, nuclear translocation of p65 and nuclear-phosphorylation of p50 and p65 were detected (Figure6 F).
These results suggested that the activation of canonical NF-κB pathway could positively regulate AC007271.3 and affect the expression of miR-125b-2-3p and Slug.

Discussion
Accumulating evidence indicated that lncRNA is closely related to tumorigenesis and development. Our recent studies con rmed that aberrant AC007271.3 levels in OSCC patients were signi cantly associated with clinical stage, especially in early-stage disease and serum AC007271.3 levels could also discriminate between OSCC and normal controls with high sensitivity and speci city, which suggesting that AC007271.3 could be a novel circulating biomarker for the determination of OSCC [11] . Furthermore, we found that AC007271.3 could promote cell proliferation, invasion and inhibit cell apoptosis of OSCC via the Wnt/β-catenin signaling pathway, which might provide a novel therapeutic approach for OSCC [12] . In this study, we found that overexpression of AC007271.3 could facilitate the metastasis of OSCC cells in vivo, which further demonstrated the tumorigenesis role of AC007271.3. However, the underlying mechanism of AC007271.3 in OSCC carcinogenesis keep unclear.
Plenty of evidence has indicated that lncRNAs can regulate target gene expression by functioning as a ceRNA for miRNA [15,22] . LncRNA functions are closely associated its subcellular localization [23] . We proved that AC007271.3 was mainly located in the cytoplasm. Thus, it indicated that AC007271.3 may function as an endogenous miRNA sponge to regulate the expression of target genes. In the present study, bioinformatics analysis and luciferase reporter assays revealed that miR-125b-2-3p is a target of AC007271.3. Moreover, AC007271.3 upregulation could decrease miR-125b-2-3p expression and deletion of miR-125b-2-3p could inhibit the expression of AC007271.3, which con rmed that AC007271.3 served as a ceRNA with miR-125b-2-3p. To our knowledge, this is the rst report to manifest that AC007271.3 can be regulated by miR-125b-2-3p in OSCC.
MiR-125b-2-3p is a member of human miR-125 family, which consists of three homologs (miR-125a, miR-125b-1, and miR-125b-2). The 125b-2-3p was derived from the 3 'arm of the precursor miRNA(pre-miRNA) of miR-125b-2 [24] . It had reported that miR-125b-2-3p was down-regulated in hepatocellular carcinoma [25] and small cell osteosarcoma [26] but upregulated in tumor stroma of colon cancer [27] . In OSCC, although miR-125b was found to inhibit the progression of OSCC [28] , the function of miR-125b-2-3p is still uncertain. In our study, we con rmed that the expression levels of miR-125b-2-3p were negatively correlated with TNM classi cation and lymph node metastasis in OSCC patients. Overexpression of miR-125b-2-3p could inhibit the migration and invasion of OSCC, while silencing of miR-125b had the opposite result. These results indicated that miR-125b-2-3p existed genetic heterogeneity in tumor. However, the underlying mechanisms of miR-125b-2-3p aberrant dysregulation in tumor are unknown. In hepatocellular carcinoma, low expression of miR-125b-2-3p could increase the expression of the target gene cyclin A2, further promoting the formation of CycA/CDK1 and then affect the G2/M phase of the cell cycle [25] .
Slug, also known as Snail Family Transcriptional Repressor 2 (Snai2), is a member of the Snail family (Snai1, Snai2, Snai3). It's an important transcription factor and plays a vital role in development and closely relates to the occurrence and development of various diseases including cancer [29] . Aberrant expression of Slug has been closely related to cancer stem cell formation, cell cycle regulation and apoptosis as well as invasion and metastasis [21] . As we predicted in the present study, Slug was a target gene of miR-125b-2-3p by bioinformatics analysis and then veri ed by luciferase reporter assays and western blot. The expression of Slug was remarkably up-regulated in OSCC tissues than in normal controls and meant unfavorable prognosis by bioinformatics analysis. We further found that Slug expression was signi cantly higher than in the adjacent normal tissues by immunohistochemistry and high expression of Slug was positive correlated with TNM classi cation, lymph node metastasis and differentiation in OSCC patients. In addition, knockdown of Slug expression remarkably impaired the ability of migration and invasion of OSCC cells. These results discovered that Slug might play an important role in OSCC carcinogenesis. In this study, we found that AC007271.3 positively regulated the expression of Slug. In the rescue experiment, over-expression of miR-125b-2-3p reversed the up-regulated Slug caused by over-expressed AC007271.3. Simultaneously, migration and invasion experiments unveiled that miR-125b-2-3p mimics and si-Slug can reversed the effects resulted from AC007271.3 overexpression. Take together, AC007271.3 could regulate Slug expression by sequestering endogenous miR-125b-2-3p.
Snai family members could directly bind to the promoter of E-cadherin by E-box region and negatively regulate its expression [30,31] . E-cadherin, N-cadherin and Vimentin were the frequent symbol markers for EMT [32] , which could enhance the invasion and migration of tumors. In this work, we found that the knockdown of Slug induced no signi cant changes on N-cadherin, Vimentin and β-catenin except a remarkably decrease of E-cadherin, which indicated that Slug could change the epithelial-mesenchymaltransition phenotype by inhibiting the expression of E-cadherin and then promoted the migration and invasion in OSCC cells. On the other hand, the E-cadherin / catenin complex, which bind to cytoskeletal components, is an important regulator to form a mature adherent junction [33] . The reduction in E-cadherin was related with the activating of β-catenin in colorectal cancer. [34] Combined with our results, we speculated that Slug may increase the dissociated β-catenin which could be activated and translocated into the nucleus by reducing the expression of E-cadherin. Moreover, recent researches implied that Wnt / β-catenin pathway participated in the regulation of Slug [35,36] , which may form a feed-back regulation between Slug/E-cadherin/β-catenin. However, the relation among Slug, E-cadherin and β-catenin in OSCC need to be further explored.
NF-κB is a regulator of expression of the κB light chain which rstly recognized in B cells. It is a homo-or hetero-dimer formed by any two of the ve subunits of RelA (p65), RelB, c-Rel, NFKB1 (p50), NFKB2 (p52) [37] . Plenty of evidence indicated that NF-κB pathway played an important role in in ammation and cancer [38,39] . In the canonical NF-κB pathway, tumor necrosis factor (TNF)-α stimulation induces IKKβ phosphorylation and then phosphorylates IκBα to promote its polyubiquitination and degradation, which lead to the release of p65/p50 heterodimer. P65/p50 dimers may be phosphorylated [40] and then translocate into nucleus and bind to the speci c DNA sequences to promote the transcription of target genes [41] . The NF-κB pathway has been fully studied in the past few decades. It was widely acknowledged that p50 / p65 dimer plays a direct role in the canonical NF-κB pathway to positively regulate the transcription of target genes [37][38][39]41] . Phosphorylation sites of p50 and p65 also regulate the function of this dimer [40,42] . Previous studies showed that p50 (Ser337) site phosphorylation could enhance the DNA binding ability of p50 [43] , while p65 (Ser536) phosphorylation could enhance transactivation potential of p65 [44] . In the research, after treatment of TNF-a in SCC9, the over-expressed AC007271.3 were detected, and the p50 (Ser337) and p65 (Ser536) were signi cantly up-regulated in nucleus. These results suggested that the activation of canonical NF-κB pathway could positively regulate the expression of AC007271.3. In addition, recent study showed that NF-κB pathway promoted the metastasis in head and neck squamous cell carcinoma cells by stabilizing Slug [45] . However, the underlying mechanism is not clear. Our results identi ed that classical NF-κB pathway probably regulated the migration and invasion of OSCC through the AC007271.3 / miR-125b-2-3p / Slug axis. To our knowledge, it's the rst time to propose this potential mechanism in OSCC.

Conclusion
In conclusion, our research con rmed that the classical NF-κB pathway-regulated AC007271.3 played as a ceRNA of miR-125b-2-3p to regulate Slug gene and promote the migration and invasion in OSCC cells, which suggested that AC007271.3 may be a diagnostic molecule and therapeutic target for OSCC. Availability of data and materials: The datasets used during the current study are available from the corresponding author on reasonable request.

Figure 2
MiR-125b-2-3p regulated migration and invasion in OSCC cells. a. Lower expression of miR-125b-2-3p was detected in two OSCC cells compared to HOK cells. b. Differential expression of miR-125b-2-3p in 82  and SCC9-NC group. c. The expression of AC007271.3 and miR-125b-2-3p in pulmonary metastases were detected by qRT-PCR. d. Representative images of HE and IHC staining. Higher level of Slug was detected in SCC9-AC007271.3 group than that in SCC9-NC group. e. Western blot con rmed the up-regulated of Slug and the down-regulated of E-cadherin in SCC9-AC007271.3 group. *p < 0.05, **p < 0.01, ***p<0.001. Figure 6 NFKB1 was enriched on the core promoter region of AC007271.3 and canonical NF-κB pathway upregulated the expression of AC007271.3. a. After transfecting fragment-by-fragment deletion pGL3