Long noncoding RNA LINC00968 inhibits lung adenocarcinoma progression through sponging miR-22-5p and further restoring CDC14A


 Background: Lung adenocarcinoma (LUAD) is a high aggressive human cancer which usually diagnosed at advanced stages. Accumulating evidences indicate that long noncoding RNAs (lncRNAs) are crucial participants in LUAD progression. Methods: The mRNA levels of LINC00968, miR-22-5p and cell division cycle 14A (CDC14A) were measured using quantitative real-time PCR. Cell proliferation was evaluated using cell counting kit-8 and flow cytometry. Cell migration and cell invasion were assessed by wound healing and transwell assay, respectively. The interactions between LINC00968 and miR-22-5p were validated by RNA immunoprecipitation, RNA pull down and luciferase reporter assay. Results: We found that lncRNA LINC00968 was significantly down-regulated in LUAD tissues and cell lines. LINC00968 level was positively correlated to survival rate, and negatively correlated to tumor node metastasis stage, tumor size and lymph node metastasis of LUAD patients. LINC00968 over-expression in LUAD cells inhibited cell proliferation and induced cell cycle arrest at G1 phase. LINC00968 over-expression also suppressed migration, invasion and epithelial mesenchymal transition (EMT) as evidenced by elevated E-cadherin, decreased N-cadherin, TWIST and SNAIL levels. We further validated that LINC00968 localized in cytoplasma and acted as an upstream of microRNA miR-22-5p, which was up-regulated in LUAD tissues and cell lines. Besides, elevated miR-22-5p expression abolished the effect of LINC00968 over-expression on LUAD progression including in vivo tumor growth. In addition, we first validated that cell division cycle 14A (CDC14A), which was down-regulated in LUAD tissues, was a downstream target of miR-22-5p. We over-expressed CDC14A in LUAD cells and miR-22-5p induced LUAD progression was partially reversed. Conclusion: our study demonstrated that LINC00968 inhibited proliferation, migration and invasion of LUAD by sponging miR-22-5p and further restoring CDC14A. This novel regulatory network might provide us with promising diagnostic and therapeutic target in LUAD treatment.


Background
Lung cancer remains one of the leading causes of cancer death in China and worldwide. Accounting for 21.9% of all cancer cases in 2018, lung cancer is the most common diagnosed cancer in male in China, which is mainly attribute to tobacco intake and absence of e cient early-stage diagnosis and treatment strategies [1,2]. Non-small-cell lung carcinoma (NSCLC) accounts for over 85% of all lung cancer cases, among which lung adenocarcinoma (LUAD) is the most common subtype [3]. LUAD is an aggressive human cancer which is resistant to conventional radiotherapy and chemotherapy. LUAD patients are often diagnosed at advanced stages, accompanying with disseminated metastasis and leaving them with limited treat options [4]. Smoking and exposure to radon, non-industrial air-pollution, asbestos and some chemicals are major risk factors for LUAD [5]. In spite of tremendous efforts we made in understanding the pathogenesis of LUAD in the past decades, the precise underlying mechanisms of LUAD progression continues to be enigmatic.
Long noncoding RNAs (lncRNAs) refer to non-protein coding RNA transcripts containing more than 200 nucleotides [6]. Although increasing evidences revealed that lncRNAs are involved in numerous human diseases, cancer included, full understanding of their exact molecular mechanisms is still far beyond completion [7]. Long intergenic non-protein coding RNA 968 (LINC00968) is reported to be dysregulated in several human cancers including breast cancer, osteosarcoma, and epithelial ovarian cancer [8][9][10].
Integrative microarray analysis showed that LINC00968 is signi cantly down-regulated in NSCLC tissues compared with normal lung tissues [11,12]. Yet the precise role that LINC00968 plays in LUAD progression remains unclear. MicroRNAs are another group of noncoding RNAs, approximately 22 nucleotides in length, which participate in a variety of physiological processes by inhibiting the expression of target genes [13]. MiR-22-5p was identi ed as a potential biomarker for acute myocardial infarction and gastric cardia adenocarcinoma [14][15][16]. Interestingly, Human MicroRNA Expression Database (HMED) shows that miR-22-5p is signi cantly up-regulated in lung cancer tissues. LINC00968 overexpression in human LUAD cell line A549 results in down-regulation of miR-22-5p level [17]. Besides, bioinformatics prediction on LncTar database (http://www.cuilab.cn/lnctar) demonstrated that LINC00968 could directly bind to miR-22-5p [18]. Consequently, we hypothesize that LINC00968 regulates LUAD progression through competitive binding to miR-22-5p. In addition, we found that cell division cycle 14A (CDC14A), a member of dual speci city protein tyrosine phosphatase family, was signi cantly downregulated in LUAD tissues (Gene Expression Pro ling Interactive Analysis, GEPIA) [19]. CDC14A also acts as a putative down-stream target of miR-22-5p as predicted on TargetScan database (http://www.targetscan.org/vert_72/). CDC14A was involved in the regulation of cell cycle and DNA repair, suggesting its crucial role in cancer progression [20]. Yet whether miR-22-5p regulates LUAD progression by targeting CDC14A remain elusive.
In the present study, we analyzed the correlation between LINC00968 level and clinicopathological characteristics of LUAD patients. Then we examined the expression levels of LINC00968, miR-22-5p and CDC14A in LUAD tissues. We also investigated the effect of LINC00968, miR-22-5p and CDC14A on proliferation, migration and invasion of LUAD cells. Finally, we validated the predicted correlation.

Clinical samples and ethics statement
All LUAD tissues and adjacent non-tumor lung tissues (n = 60) were collected with the consent of patients. All experiment regarding clinical samples were performed in accordance with the ethical guidelines of the Declaration of Helsinki and approved by the Ethics Committee of International Medical Center Hospital.
Quantitative real-time PCR LUAD and adjacent non-tumor tissues were smashed using a grinder. Patient samples and cell lines were lysed with Trizol reagent (ThermoFisher, USA) to extract total RNAs, which was further reversely transcribed into cDNA using SuperScript IV reverse transcriptase (ThermoFisher, USA). Then cDNA products were quanti ed by real-time PCR using SYBR Green (Solarbio, China). Housekeeping gene GAPDH was used as an internal control for LINC00968 and CDC14A, RNU6B was used as an internal control for miR-22-5p. Results from quantitative real-time was analyzed using g the 2 −△△CT method. All primers used were listed as below: Cell counting kit-8 (CCK-8)assay A549 and H1975 cells showed relative lower LINC00968 level among four LUAD cell lines, thus they were chosen for subsequent investigations. Cell proliferation was determined using cell counting kit-8 (MedChemExpress, USA) following the manufactures' instruction. A549 and H1975 cells were seeded into a 96-well plate (2 × 10 4 cells/well) and cultured for 24 hours. Cells were treated with LV-LINC00968/LV-NC, miR-22-5p agomir/NC-agomir or pcDNA3.1-CDC14A/vector, followed by incubation for 24 hours, 48 hours, 72 hours and 96 hours. Then cells were incubated with 10 µl CCK-8 solution for 3 hours and absorbance at 450 nm was measured using a microplate reader.
Cell cycle detection A549 and H1975 cells were harvested and re-suspended with PBS buffer after treatment. Cells were xed with 1 ml pre-cooled 70% ethanol for 2 hours and rinsed with PBS buffer. Then cells were incubated with 500 µl propidium iodide (50 µg/ml) for 37 ℃ for 30 minutes, followed by ow cytometry detection within 24 hours.

Wound healing and transwell assay
Wound healing assay was used to assess cell migration. A549 and H1975 cells were seeded into a 6-well plate and cultured to con uence, scratch on the cell momolayer was performed using a 200 µl piptte tip, followed by observation under a microscope 24 hours after scratch. Transwell assay was performed to evaluate cell invasive capacity. A549 and H1975 cells were seeded into the Matrigel pre-coated upper chamber and cultured with serum-free culture medium for 48 hours, and lower chamber with culture medium containing 20% FBS was used as an attractant. Cells invaded to the lower chamber were xed with 4% paraformaldehyde and stained with 1% crystal violet for 5 minutes, followed by cell counting under a microscope.
Western blotting A549 and H1975 cells were harvested and lysed with RIPA lysis buffer (Beyotime, China) after treatment, followed by quanti cation with commercial QuantiPro BCA assay kit (Sigma, USA). Then equivalent amount of total proteins were fractionated on SDS-polyacrylamide gel and then transferred onto a polyvinylidene uoride (PVDF) membrane (ThermoFisher, USA). The PVDF membrane was sealed with 5% skim milk and incubated with primary antibodies overnight at 4℃. After rinsing with PBS buffer for three times, the PVDF membrane were incubated with secondary antibodies for 60 minutes at room temperature. Then PVDF membrane was illuminated with ECL luminescence reagent (Sangon, China), and relative protein levels were analyzed using IPP6.0. Primary antibodies used in this study were Ecadherin antibody (1:1000, SinoBiological, China), N-cadherin antibody (1:1000, SinoBiological, China), SNAIL antibody (1:1000, A nity, USA), TWIST antibody (1:1000, A nity, USA) and CDC14A antibody (1:1000, A nity, USA). HRP-conjugated goat anti-rabbit antibody (1:5000, A nity, USA) was used as secondary antibody.

Luciferase reporter assay
The interactions between LINC00968 and miR-22-5p and between miR-22-5p and CDC14A were validated using luciferase reporter assay. In brief, mutant type of LINC00968 and CDC14A, termed as LINC00968-MUT and CDC14A-MUT, absent miR-22-5p binding site were obtained using overlapping PCR. Wildtype and mutant type of LINC00968 and CDC14A were ampli ed and cloned into pmiRGLO vector. The recombinant pmiRGLO plasmids were co-transfected with miR-22-5p/NC mimics into A549 and H1975 cells. The binding activity was assessed by measuring absorbance at 560 nm using commercial Luciferase Reporter Assay Kit (ThermoFisher, USA).

Fluorescent in situ hybridation (FISH)
FISH assay was performed to detect the subcellular localization of LINC00968 in LUAD cells. In brief, A549 and H1975 cells were baked at 37 ℃ for two hours and dehydrated in 50%, 80% and 98% gradient ethanol for 3 minutes. Then A549 and H1975 cells were hybridized with cy3-labelled LINC00968 (RiboBio, China) probe or NC probe (RiboBio, China) for 16 hours at 42 ℃. After staining with DAPI avoiding light for 5 minutes, typical images were captured under a microscope.

RNA immunoprecipitation (RIP)
The enrichment of LINC00968 and miR-22-5p on Argonaute 2 (Ago2) protein was evaluated using RIP assay kit (MBL life science, Japan). A549 and H1975 cells were lysed using 200 µl RIP lysis buffer, cell lysate was centrifuged and 15 µl of supernatant was used as input. The rest of supernatant protein was incubated with anti-Ago2-coated agarose beads (MBL life science, Japan) overnight at 4℃, rabbit IgG was used as a control. The expression of Ago2 was detected by subsequent western blotting, and expression of LINC00968 and miR-22-5p was measured by quantitative real-time PCR.
RNA pull down A549 and H1975 cells were lysed using lysis buffer containing RNase inhibitor and proteinase inhibitor. Cell lysate was centrifuged and re-suspended with hybridation buffer. Then cell lysate was incubated with biotinylated miR-22-5p or NC probes for 5 hours and incubated with magnetic streptavidin beads overnight at 4℃. The enriched RNA was puri ed and analyzed using quantitative real-time PCR.
In vivo tumor growth Healthy 6 to 8-week old C57BL/6J mice were purchased from the experimental animal center of BIOCYTOGEN (China). A549 cells were cultured to exponential phase, infected with LV-LINC00968/LV-NC and miR-22-5p agomir. Then A549 cells (2 × 10 6 cells/mice) were subcutaneously injected to mice to establish tumor xenograft model (n = 6). Tumor volumes were measured every week for ve weeks, and mice were euthanized at fth week. Tumor weight was recorded and tumor tissues were xed with 4% paraformaldehyde for subsequent detection. All animal experiments were performed following the guidelines for the care and use of laboratory animals and approved by the Ethics Committee of International Medical Center Hospital.

Immunohistochemistry
Immunohistochemistry assay was performed to detect the expression of Ki67 in mouse tumor tissues. In brief, 5 µm of tissue sections were depara nized and boiled with 10 mM, pH6.0 citric acid buffer for 15 minutes for antigen retrieval. Then tissue sections were blocked with 5% goat serum for one hour and incubated with Ki67 antibody (1:200, SinoBiological, China) overnight at 4℃. After washing with TBS buffer (0.1% Tween 20 in PBS buffer), tissue sections were incubated with HRP-conjugated secondary antibody and Ki67 signal was developed using 3, 3'-diaminobenzidine ((Sangon, China)).

Statistical analysis
All results in our study were presented as means ± SD and all data analysis was conducted using Graphpad Prism 8.0. Mean values between two groups were analyzed by Students' t-test. Survival rate was analyzed using Log-rank (Mantel-Cox) test. Data from Cell counting kit-8 assay, tumor growth and luciferase reporter assay were analyzed by Two-way ANOVA. The correlation between LINC00968 level and clinicpathological characteristics of LUAD patients were analyzed using Pearson χ2 test. The rest of results were analyzed using One-way ANOVA. All experiments were repeated for more than three times and p values less than 0.05 were considered as statistically signi cant.

Results
LINC00968 was signi cantly down-regulated in LUAD tissues and cell lines TCGA database showed that LINC00968 was signi cantly down-regulated in LUAD tissues (Fig. 1A). We rst examined the LINC00968 level in 60 LUAD patients. LUAD tissues showed signi cantly lower LINC00968 level compared with adjacent non-tumor tissues (Fig. 1B). Further follow-up visit showed that LUAD patients with relative high LINC00968 level have signi cantly higher survival rate in contrast with that with relative low LINC00968 level (Fig. 1C). Then we analyzed the correlation between LINC00968 level and clinicpathological characteristics of LUAD patients. Tumor node metastasis (TNM) stage, tumor size and Lymph node metastasis were negatively correlated to LINC00968 level (Table.1). Besides, we detected the LINC00968 level in normal lung epithelial cell line BEAS2B and four LUAD cell lines, LUAD cells showed signi cantly lower LINC00968 level compared with BEAS2B cells, especially in A549 and H1975 cells (Fig. 1D). These results implied that LINC00968 may serve as a crucial regulator of LUAD progression.

LINC00968 over-expression inhibited proliferation, migration and invasion of LUAD cells
We over-expressed LINC00968 in A549 and H1975 cells using lentivirus vector, and LINC00968 level was elevated for more than twice in LV-LINC00968 group ( Fig. 2A). We detected cell proliferation using CCK-8 assay, LV-LINC00968 signi cantly inhibited proliferation in both A549 and H1975 cells 48 hours after infection compared with LV-NC (Fig. 2B). Cell cycle detection by ow cytometry showed that cells at G1 phase were increased, while cells at S phase and G2 phase were reduced in both A549 and H1975 cells after LV-LINC00968 infection (Fig. 2C). Wound healing assay revealed that LINC00968 over-expression remarkably suppressed the migratory ability of in A549 and H1975 cells (Fig. 2D). Further transwell assay showed similar results, the number of LUAD cells invaded to lower chamber in LV-LINC00968 group was conspicuously fewer than that in LV-NC group (Fig. 2E). Besides, LINC00968 over-expression increased the E-cadherin protein level, while reduced the expression of N-cadherin, TWIST and SNAIL in both A549 and H1975 cells (Fig. 2F). These results suggested that LINC00968 over-expression inhibited the malignance of LUAD cells.
LINC00968 was a putative regulator of miR-22-5p The mRNA level of miR-22-5p was signi cantly higher in LUAD patients compared with that in Normal control (Fig. 3A). Bioinformatics prediction indicated that LINC00968 might competitively bind to miR-22-5p (Fig. 3B), which inspired us to explore the interaction between LINC00968 and miR-22-5p. Further luciferase reporter assay showed that miR-22-5p mimics signi cantly inhibited the luciferase activity in LINC00968-WT group compared with NC mimics, whereas no obvious effect of miR-22-5p mimics on luciferase activity in LINC00968-MUT group was detected (Fig. 3C). Moreover, LV-LINC00968 infection signi cantly reduced the miR-22-5p level compared to LV-NC in both A549 and H1975 cells (Fig. 3D). FISH assay was performed to determine the subcellular localization of LINC00968, and LINC00968 was mainly localized in cytoplasm (Fig. 3E). RIP assay showed that Ago2 was successfully precipitated, accompanied with enrichment of LINC00968 and miR-22-5p in Ago2 complex (Fig. 3F). Besides, biotinylated miR-22-5p probe successfully pulled down LINC00968 compared with NC probe in both A549 and H1975 cells (Fig. 3G). These results revealed that LINC00968 acted as a putative regulator of miR-22-5p.

Discussion
Large-scale cancer genomics projects revealed that numerous lncRNAs are aberrantly expressed in a variety of human cancers [21]. LncRNAs exert their diverse function in the tumorgenesis and tumor development via acting as oncogenes, tumor suppressors or regulatory RNAs [22]. However, to date only about 1% of all 3,000 human lncRNAs has been characterized [7]. LINC00968 was rst identi ed by Hui et al in 2015, and the following researches demonstrated its diverse role in different human cancers. LINC00968 inhibited the progression of breast cancer yet acted as an oncogene in epithelial ovarian cancer and osteosarcoma [8][9][10]12]. Here in our study, we investigated the exact role of LINC00968 in the progression of LUAD. Tissue-speci c expression analysis showed that LINC00968 was speci cally expressed in lung [23]. We rst detected the expression level of LINC00968 in LUAD tissues, and the down-regulation of LINC00968 was highly in accordance with previous researches [12,24]. Lower LINC00968 level accompanied with lower survival rate, more advanced TNM stage, larger tumor size as well as more positive lymphnode metatstasis. All these results implied that LINC00968 are involved in LUAD progression and may serve as a novel diagnostic target. Then we cloned LINC00968 into lentivirus vector for over-expression in two LUAD cell lines which displayed lowest LINC00968 level. Uncontrolled malignant proliferation and cell cycle are critical characteristics of human cancers [25]. LINC00968 overexpression signi cantly inhibited the proliferation of LUAD cells and induced cell cycle arrest at G1 phase, further tumor xenograft assay showed that LINC00968 inhibited tumor growth in vivo, indicating that LINC00968 is a putative LUAD suppressor. Distant metastasis is responsible for majority of cancer death.
Migration, invasion and epithelial mesenchymal transition (EMT) are most crucial procedures involving in tumor metastasis, which confer cancer cells with enhanced resistance to apoptotic stimuli [26]. Here we found that LINC00968 over-expression inhibited the migration and invasion of LUAD cells. EMT was also suppressed as evidenced by elevated E-cadherin and reduced EMT markers N-cadherin, SNAIL and TWIST [27]. Our ndings demonstrated that LINC00968 over-expression inhibited LUAD progression and LINC00968 is a promising diagnostic and therapeutic target for LUAD treatment.
The "competing endogenous RNA" (ceRNA) hypothesized that lncRNAs could competitively bind to microRNAs and forms a new regulatory network across the transcriptome [28]. To date, the role of miR-22-5p in LUAD progression is poorly understood. In the present study, we rst found that miR-22-5p was signi cantly up-regulated in LUAD tissues and cell lines, indicating that miR-22-5p might serve as a potential early-stage biomarker for LUAD diagnosis. We also validated the interaction between LINC00968 and miR-22-5p in our study, which was highly consistent with previous miRNA microarray analysis [17]. Besides, LINC00968 over-expression-induced inhibited proliferation, migration and invasion of LUAD cells were signi cantly reversed by miR-22-5p supplement, indicating that LINC00968 may function as LUAD suppressor through sponging miR-22-5p and this novel regulatory network is a potential therapeutic target for LUAD management.
Cell cycle progression is mainly promoted by cyclins and cyclin-dependent kinases (CDKs), which makes them promising targets for cell cycle control and hence cancer therapy [29]. CDC14 phosphatase family is highly conserved among yeast and human. Active CDC14A reverses CDK phosphorylation and further regulates the mitotic entry via regulating CDC25A and CDC25B [30, 31]. Paulsen et al validated that CDC14A interacts with the CDK1/cyclin B complex, suggesting that CDC14A inhibits the activity of CDK1/cyclin B complex during interphase and therefore induces cell cycle arrest. Besides, CDC14A overexpression may activate p53 signaling and further induce cell apoptosis or cell cycle arrest [20]. In the present study, CDC14A was signi cantly down-regulated in both LUAD tissues and cell lines, which was in line with previous study [20,32]. In addition, we rst validated that CDC14A was a downstream target of miR-22-5p, and LUAD progression facilitated by miR-22-5p was reversed by CDC14A over-expression, indicating that CDC14A was a putative tumor suppressor [32].

Conclusion
In conclusion, we validated a novel regulatory network in LUAD progression: LINC00968 inhibits proliferation, migration and invasion of LUAD through sponging miR-22-5p and further restoring CDC14A. LINC00968 and CDC14A were putative tumor suppressors and the novel regulatory network may offer us new therapeutic targets and strategies in LUAD management.  The correlation between LINC00968 RNA level and clinicopathological characteristics of lung adenocarcinoma patients was analyzed using chi-square tests.