Circular RNA hsa_circ_0002483 promotes growth and invasion of lung adenocarcinoma by sponging miR-125a-3p

Background Increasing evidence indicates that the aberrant expression of circular RNAs (circRNAs) is involved in the pathogenesis and progression of lung adenocarcinoma (LUAC). However, the function and molecular mechanisms of hsa_circ_0002483 (circ_0002483) in LUAC remain unclear. Methods The association between circ_0002483 expression and clinicopathological characteristics and prognosis in patients with LUAC was analyzed by fluorescence in situ hybridization. The functional experiments such as CCK-8, colony formation and Transwell assays and a subcutaneous tumor model were conducted to determine the role of circ_0002483 in LUAC cells. The specific binding between circ_0002483 and miR-125a-3p was validated by RNA immunoprecipitation, luciferase gene report and qRT-PCR assays. The effects of circ_0002483 on miR-125a-3p-mediated C-C motif chemokine ligand 4 (CCL4)-CCR5 axis were assessed by Western blot analysis. Results We found that circ_0002483 was upregulated in LUAC tissue samples and associated with Tumor Node Metastasis (TNM) stage and poor survival in patients with LUAC. Knockdown of circ_0002483 inhibited proliferation, colony formation and invasion of A549 and PC9 cells in vitro, whereas overexpression of circ_0002483 harbored the opposite effects. Furthermore, circ_0002483 sponged miR-125a-3p and negatively regulated its expression. CCL4 was identified as a direct target of miR-125a-3p. The rescue experiments showed that miR-125a-3p mimics reversed the tumor-promoting effects of circ_0002483 by targeting CCL4-CCR5 axis in A549 and PC9 cells. In addition, the in vivo experiment further validated that knockdown of circ_0002483 repressed tumor growth. Conclusions Our findings demonstrated that circ_0002483 could act as a sponge of miR-125a-3p to upregulate CCL4-CCR5 axis, contributing to the tumorigenesis of LUAC, and represent a potential therapeutic target for LUAC. Supplementary Information The online version contains supplementary material available at 10.1186/s12935-021-02241-y.


Introduction
Lung cancer is one of the most malignant tumors and its incidence and mortality are increasing rapidly, threatening to the public health and human life [1]. With the development of treatment methods, lung adenocarcinoma (LUAC) as a subclass of non-small cell lung cancer (NSCLC) has been well-treated, but the advanced cases still harbors a poor prognosis duo to its distant metastasis [2]. Accumulating data display that the aberrant expression of noncoding RNAs is associated with the prognosis and progression of LUAC [3][4][5]. Therefore, identi cation of cancer-related noncoding RNAs may provide potential biomarkers for the early detection of LUAC.
Circular RNA (circRNA) as a subclass of noncoding RNAs is characterized by covalently closed loop structures and RNA stability owing to resistance to RNase R [6]. Increasing data demonstrate that circRNAs are aberrantly expressed and associated with the prognosis and progression of LUAC [7][8][9].
MicroRNAs (miRNAs) as another substyle of noncoding RNAs can target mRNAs by post-transcriptional levels and serve as promising biomarkers for early diagnosis and prognosis of LUAC [17]. Some studies show that miR-125a-3p and miR-125a-5p, downregulated in NSCLC, possess the inverse effects on migration and invasion of lung cancer cells [18], and downregulation of miR-125a-3p is associated with tumorigenesis and poor prognosis in patients with NSCLC [19]. Moreover, miR-125a-3p represses lung cancer growth and invasion by regulating the mouse double minute 2 homolog/p53 signaling [20], and NSCLC proliferation and migration by targeting metastasis-associated gene 1 [21]. miR-125a-3p is also sponged by lncRNA MALAT1 to regulate FOXM1 expression in hepatocellular carcinoma [22], and sponged by circ-MAPK4 in gliomas, circ_0012919 in systemic lupus erythematous and circLMF1 in aortic smooth muscle cells [23][24][25].
In the present study, we identi ed a differentially-expressed hsa_circ_0002483 between LUAC and normal tissue samples, and found that upregulation of circ_0002483 was associated with TNM stage and poor survival in patients with LUAC. Further in vitro and in vivo experiments veri ed that circ_0002483 promoted the growth and invasion of LUAC cells by sponging miR-125a-3p and upregulating CCL4-CCR5 axis, thereby providing a novel target for LUAC.

Bioinformatic analysis
The circRNA pro ling was used to identify the differentially-expressed circRNAs between LUAC and normal tissues and downloaded from the Gene Expression Omnibus (GEO) dataset (https://www.gcbi.com.cn/gclib/html/index).
Cell Culture and Transfection LUAC cell lines (A549 and PC9) used in these studies were provided by Cell bank of Shanghai Chinese Academy of Sciences (Shanghai, China), and cultured in DMEM medium supplemented with 10% heatinactivated FBS in a humidi ed atmosphere containing 5% CO2 at 37°C. Lentivirus mediated si-circ_0002483 (5'-AACAGAATATGACAGATACCTdTdT-3'), its negative control (si-NC), circ_0002483 overexpression plasmids, miR-125a-3p mimics and inhibitors were provided by GenePharma (Shanghai, China) and used for transfection into A549 and PC9 cells. Master Mix II (Thermo Fisher Scienti c, Runcorn, UK) were used to examine miR-125a-3p levels. U6 or βactin was used as an internal control. The data were quanti ed using 2 -∆∆CT equation in triplicate. The primer sequences used were shown in Table S1.
Western Blot Analysis A549 and PC9 cell lines were harvested and protein was extracted using RIPA lysis. Primary antibodies against anti-CCL4 (ab25129, abcam), anti-CCR5 (ab65850, abcam) and anti-GAPDH (ab8245, abcam) were diluted (1:1000) and incubated overnight at 4°C. After rinsing, the polyvinylidene uoride (PVDF) membrane of the antibodies was transferred onto the system. Captured signal was quanti ed by Image Lab Software 3.0 (Bio-Rad), and GAPDH was used as an internal parameter.

MTT, Colony formation and Transwell Assays
Cell viability, colony formation and invasive capabilities were conducted by MTT and Transwell assays according to the previous report [26].

Actinomycin D and RNase R Treatment
Transcription was prevented by the addition of 2 mg/ml Actinomycin D and DMSO (Sigma-Aldrich, St. Louis, MO, USA) was used as the control group. Total RNA was incubated for 30 min at 37℃with 3 U/μg of RNase R (Epicentre Technologies, Madison, WI, USA).
Dual-Luciferase Reporter Assay A549 and PC9 cells were seeded into 24-well plates, and PRL-TK-Luc report vectors containing WT or Mut 3'UTR of circ_0002483 and CCL4 were co-transfected with miR-125a-3p mimic or inhibitor into A549 and PC9 cell lines. After the transfection for 48 h, luciferase activities were detected with a dual-luciferase reporter system (Promega, Madison, WI).

RNA Immunoprecipitation
RNA immunoprecipitation (RIP) assay was conducted using a Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, Billerica, MA) according to the manufacturer's instructions.

Animal Experiments
Six-week-old female immune-de cient nude mice (BALB/c-nu) were injected subcutaneously with 5 × 10 7 A549 cells stably transfected with si-circ_0002483 or si-NC. Mice were monitored daily and developed a subcutaneous tumor. The tumor volume was detected every other day by using a formula: volume = length x width 2 /2. This study was approved by Animal Ethics Committee of The Shenzhen People's Hospital.

Immunohistochemistry (IHC)
Ki-67 levels in xenograft tumor tissues were assessed using IHC assay. The detailed description of IHC was conducted as previously reported [4].

Statistical Analysis
Statistical analyses were performed by SPSS 20.0 (IBM, SPSS, Chicago, IL, USA) and GraphPad Prism.
Student's t-test or Chi-square test was used to estimate the statistical signi cance for comparisons of two groups. Pearson correlation analysis was used to analyze the correlations. Overall survival curve was drawn with the Kaplan-Meier and log-rank test. P < 0.05 was considered statistical signi cance.

Upregulation of circ_0002483 was associated with poor survival in patients with LUAC
The GEO dataset (GSE101684) was used to screen the differentially-expressed circRNAs between LUAC and non-cancerous tissue samples and top 10 upregulated or downregulated circRNAs were identi ed according to the FC > 2 and P < 0.01, of which hsa_circ_0002483 harbored a remarkable elevation in LUAC ( Figure 1A). Then, FISH analysis validated that circ_0002483 expression levels were increased in pair-matched LUAC tissue samples as compared with the non-cancerous tissues ( Figure 1B; n = 80, P = 0.01). The similar result was shown in LUAC patients with stage III-IV (n = 42) as compared with those with stage I-II (n = 38) ( Figure 1C; P = 0.02).
In addition, we analyzed the association of circ_002483 with the clinicopathological characteristics in patients with LUAC and found that elevated expression of circ_002483 was related with TNM stage (Table 1; P = 0.048) rather than other parameters (P > 0.05) in LUAC. On the basis of the cutoff value of circ_002483, we divided the cases into circ_002483-high (n = 43) and circ_002483-low groups (n = 37). Kaplan-Meier analysis uncovered that the patients with circ_002483-high group possessed a poorer survival as compared with those with circ_002483-low expression ( Figure 1D; P = 0.014). Univariate and Multivariate analysis unveiled that TNM stage and pathological stage rather than circ_002483 expression are independent prognostic factors of poor survival in patients with LUAC (Tables S2). Identi cation of a novel circ_0002483 and its cellular localization It is indicated that hsa_circ_0002483 (chr8:141874410-141900868) is originated from exon 2, 6 regions within protein tyrosine kinase 2 (PTK2) locus, and the spliced sequence is about 482 bp (Figure 2A). The back-spliced junction of circ_0002483 was ampli ed by divergent primer and circ_0002483 could be ampli ed from only cDNA but not gDNA in A549 cells ( Figure 2B). Relative to linear PTK2, circ_0002483 produced a resistance to RNase R treatment in A549 and PC9 cell lines, and circ_0002483 harbored a loop structure in LUAC cells ( Figure 2C). After A549 cells were treated by Actinomycin D, qRT-PCR displayed that circ_0002483 was more stable than PTK2 ( Figure 2D). qRT-PCR and FISH analysis revealed that circ_0002483 was mainly localized in the cytoplasm of LUAC cells and tissues ( Figure 2E, F).

Circ_0002483 facilitated proliferation, colony formation and invasion in vitro
Upregulation of circ_0002483 in LUAC suggested that it might be an oncogenic factor. To test this hypothesis, we evaluated the function of circ_0002483 in A549 and PC9 cells and constructed the overexpression vectors and the siRNA against circ_0002483. As indicated in Figure 3A, the overexpression vectors and the siRNA of circ_0002483 could markedly increase and decrease the expression of circ_0002483 in A549 and PC9 cell lines, respectively. MTT assay was conducted to chart the growth curve which indicated that upregulation of circ_0002483 obviously increased the proliferation viability of A549 and PC9 cells, while downregulation of circ_0002483 repressed cell growth ( Figure 3B). Colony formation assay further veri ed that the cell colony number of A549 and PC9 was remarkably elevated by upregulation of circ_0002483 and reduced by downregulation of circ_0002483 ( Figure 3C, D). In addition, the effects of circ_0002483 on LUAC cell invasion were estimated by Transwell assay, which indicated that overexpression of circ_0002483 increased invasive capabilities of A549 and PC9 cells, while knockdown of circ_0002483 harbored the opposite effects ( Figure 3E, F).
Circ_0002483 was negatively associated with miR-125a-3p expression in LUAC tissue samples To demonstrate the underlying mechanisms of circ_0002483 in LUAC cells, we identi ed the binding sites of 5 miRNAs with circ_0002483 3'UTR by Circular RNA pro ling and Interactome ( Figure 4A). TCGA cohort showed that, relative to the other 4 miRNAs, only miR-125a-3p possessed a decreased expression in paired (n = 39) and unpaired LUAC tissue samples (n = 448, Figure 4B). The downregulation and cytoplastic location of miR-125a-3p in pair-matched LUAC tissues were further con rmed by FISH analysis (n = 80, P = 0.025; Figure 4C-E). Pearson correlation analysis demonstrated that circ_0002483 harbored a negative correlation with miR-125a-3p expression in LUAC tissue samples (P = 0.036; Figure  4E).
Then, we analyzed the association of miR-125a-3p with clinicopathological characteristics in patients with LUAC (Table S3) and found that miR-125a-3p expression was related with gender (P = 0.012) and lymph node metastasis (P = 0.013) in patients with LUAC. However, the patients with miR-125a-3p-low expression showed no difference in poor survival and tumor recurrence as compared with those with miR-378a-3p-high expression ( Figure S1).

Circ_0002483 acted as a sponge of miR-125a-3p
The speci c binding sites of miR-125a-3p with WT or Mut circ_0002483 3'UTR were demonstrated in Figure 5A. To determine whether circ_0002483 could bind with miR-125a-3p, we co-transfected A549 and PC9 cells with WT or Mut circ_0002483 3'UTR reporter vectors and miR-125a-3p mimics or inhibitor, and found that miR-125a-3p mimics decreased the luciferase activities of WT circ_0002483 3'UTR in A549 and PC9 cells, while miR-125a-3p inhibitor enhanced such an effect ( Figure 5B). But, miR-125a-3p exhibited no effects on those of Mut circ_0002483 3'UTR. Further qRT-PCR analysis indicated that upregulation of circ_0002483 reduced the expression of miR-125a-3p, and downregulation of circ_0002483 increased its expression ( Figure 5C, D), whereas miR-125a-3p exhibited no effects on circ_0002483 expression in A549 and PC9 cells ( Figure S2). Moreover, RIP assay was executed for Ago2 protein in A549 and PC9 cells, and the endogenous expression of circ_0002483 and miR-125a-3p pulled down from Ago2-expressed A549 and PC9 cells, indicated by qRT-PCR analysis was mainly enriched in Ago2 pellet relative to the input control ( Figure 5E, F). After co-transfection with circ_0002483 and miR-125a-3p mimics or si-circ_0002483 and miR-125a-3p inhibitors into A549 and PC9 cells for 5 days, we found that miR-125a-3p suppressed the proliferation viability and attenuated circ_0002483-caused cell proliferation, while miR-125a-3p inhibitors displayed the opposite effects ( Figure 5G).

MiR-125a-3p reversed circ_0002483-caused upregulation of CCL4-CCR5 axis
The targets of miR-125a-3p were identi ed by Targetscan7.1 and mirPathv.3 and the binding sites of miR-125a-3p with 3'UTR of CCL4 were demonstrated in Figure 6A. To determine whether miR-125a-3p could bind with 3'UTR of CCL4, we co-transfected A549 and PC9 cells with WT or Mut CCL4 3'UTR reporter vectors and miR-125a-3p mimics, and found that miR-125a-3p mimics decreased the luciferase activities of WT CCL4 3'UTR in A549 and PC9 cells, however, miR-125a-3p exhibited no effects on those of Mut CCL4 3'UTR ( Figure 6B). TCGA cohort indicated that CCL4 and CCR5 were upregulated in 515 LUAC tissue samples (Figure 6C, D; P < 0.0001) and miR-125a-3p harbored a negative correlation with both of them in LUAC ( Figure 6E; P < 0.05) . In addition, we found that high expression of CCL4 harbored no association with clinicopathological features in patients with LUAC with (Table S4), and the patients with CCL4-high expression possessed no difference in poor survival and tumor recurrence relative to those with CCL4-low expression ( Figure S3). Moreover, qRT-PCR and Western blot analyses demonstrated that miR-125a-3p inhibited the expression of CCL4 and CCR5 and reversed circ_0002483-caused upregulation of CCL4-CCR5 axis in A549 cells ( Figure 6F, G).

Knockdown of circ_0002483 repressed cell growth in vivo
To ascertain the effects of circ_0002483 on LUAC tumor growth in vivo, stably transfected A549 cells infected with si-NC or si-circ_0002484 were constructed and subcutaneously injected into 6-week old female nude mice ( Figure 7A). Moreover, the tumor volume and weight in the si-circ_0002384 group were smaller and lighter than the si-NC group ( Figure 7B, C). Hematoxylin and eosin (HE) staining and IHC of Ki-67 indicated that knockdown of circ_0002483 inhibited cell proliferation as compared with the si-NC group ( Figure 7D).

Discussion
A sea of studies have shown that the aberrant expression of circRNA is related with the prognosis, pathogenesis and progression in multiple cancers including LUAC [26-28]. hsa_circ_0000792 and circ_0013958 are identi ed as potential biomarkers of LUAD [7,9] and hsa_circ_0000190 is linked to tumor progression and poor prognosis in advanced lung cancer [8]. The tissue and plasmid levels of hsa_circ_0003221 (circPTK2) are correlated with poor differentiation and lymph node metastasis in bladder cancer [29]. Herein, we identi ed a novel differentially-expressed hsa_circ_0002483 (circPTK2) in LUAD tissue samples and found that, elevated expression of circ_0002483 was associated with TNM stage and poor survival in patients with LUAD and might provide a potential prognostic factor for LUAD.
It has been shown that miR-125a-3p is downregulated and inhibits cell growth and invasion of LUAC [18][19][20][21]. In accordance, we found that miR-125a-3p expression was decreased in LUAC and associated with lymph node metastasis in patients with LUAC. miR-125a-3p harbored a negative correlation with circ_0002483 expression, repressed cell proliferation and reversed circ_0002483-induced tumor-promoting effects. In addition, CCL4-CCR5 axis contributes to breast cancer metastasis [36] and targeting CCR5 inhibits colorectal cancer liver metastasis [37]. We herein identi ed that CCL4 was upregulated in LUAC tissues and regarded as a direct target of miR-125a-3p. miR-125a-3p downregulated CCL4-CCR5 axis and reversed circ_0002483-caused upregulation of this axis. Our results suggested that circ_0002483 might act as a sponge of miR-125a-3p to upregulate CCL4-CCR5 axis, contributing to the tumorigenesis of LUAC.
Taken together, elevated expression of circ_0002483 is associated with TNM stage and poor survival in patients with LUAC. circ_0002483 facilitates the tumorigenesis and invasion of LUAC by sponging miR-125a-3p and upregulating CCL4-CCR5 axis. Our ndings might offer a potential therapeutic biomarker for LUAC.