MiR-182-5p and its target HOXA9 in non-small cell lung cancer: a clinical and in-silico exploration with the combination of RT-qPCR, miRNA-seq and miRNA-chip

Background MiR-182-5p, a cancer-related microRNA (miRNA), modulates tumorigenesis and patient outcomes in various human malignances. This study interroted the clinicopathological significance and molecular mechanisms of miR-182-5p in non-small cell lung cancer (NSCLC). Methods The clinical significance of miR-182-5p in NSCLC subtypes was determined based on an analysis of 124 samples (lung adenocarcinomas [LUADs], n = 101; lung squamous cell carcinomas [LUSCs], n = 23) obtained from NSCLC patients and paired noncancer tissues and an analysis of data obtained from public miRNA-seq database, miRNA-chip database, and the scientific literature. The NSCLC samples (n = 124) were analyzed using the real-time quantitative polymerase chain reaction (RT-qPCR). Potential targets of miR-182-5p were identified using lists generated by miRWalk v.2.0, a comprehensive atlas of predicted and validated targets of miRNA-target interactions. Molecular events of miR-182-5p in NSCLC were unveiled based on a functional analysis of candidate targets. The association of miR-182-5p with one of the candidate target genes, homeobox A9 (HOXA9), was validated using in-house RT-qPCR and dual-luciferase reporter assays. Results The results of the in-house RT-qPCR assays analysis of data obtained from public miRNA-seq databases, miRNA-chip databases, and the scientific literature all supported upregulation of the expression level of miR-182-5p level in NSCLC. Moreover, the in-house RT-qPCR data supported the influence of upregulated miR-182-5p on malignant progression of NSCLC. In total, 774 prospective targets of miR-182-5p were identified. These targets were mainly clustered in pathways associated with biological processes, such as axonogenesis, axonal development, and Ras protein signal transduction, as well as pathways involved in axonal guidance, melanogenesis, and longevity regulation, in multiple species. Correlation analysis of the in-house RT-qPCR data and dual-luciferase reporter assays confirmed that HOXA9 was a direct target of miR-182-5p in NSCLC. Conclusions The miR-182-5p expression level was upregulated in NSCLC tissues. MiR-182-5p may exert oncogenic influence on NSCLC through regulating target genes such as HOXA9.


Background
According to data from the National Comprehensive Cancer Network, lung cancer (LC) is responsible for the majority of cancer-associated deaths worldwide [1]. There are two types of LC: non-small cell lung cancer (NSCLC) and small cell lung cancer [2]. Of these, NSLC is the most common and accounts for the majority cases of LC [1][2][3][4][5][6][7][8].
Although improvements in screening (i.e., diagnostic imaging and laboratory tests) and drug therapy have contributed greatly to NSCLC outcomes, the clinical outcome of NSCLC remains poor due to a lack of effective biomarkers for NSCLC [4]. At the time of diagnosis, most of patients have advanced stage disease because of atypical symptoms in the early stage of the disease [2]. Thus, NSCLC survival is poor, with 5-year survival lower than 20% [5]. Therefore, the development of novel screening and therapeutic strategies are of crucial importance for NSCLC patients.
The aim of the present study was to examine the clinicopathological value and molecular mechanisms of miR-182-5p in non-small cell lung cancer (NSCLC). With this aim in mind, we examined miR-182-5p overexpression patterns in lung adenocarcinomas (LUADs) and NSCLC.
We expect that the current study will facilitate understanding of the role of miR-182-5p in the pathogenesis of NSCLC and its potential value as a marker of NSCLC subtypes.

Methods
Tissue collection from NSCLC patients NSCLC tissue samples and paired noncancer tissue samples were obtained from 124 NSCLS patients The study was approved by the ethics committee of the First Affiliated Hospital of Guangxi Medical University, and written informed consent was obtained from all the patients.
All 124 NSCLC tissues were formalin fixed and paraffin embedded for subsequent experiments.
MiR-182-5p expression in NSCLC using miRNA-seq data Level 3 IlluminaHiSeq miRNA-seq data on miR-182-5p expression in NSCLC were obtained from recomputed and normalized the cancer genome atlas (TCGA) data in UCSC Xena (https://xena.ucsc.edu/). Alterations in the expression of miR-182-5p in LUADs, non-LUADs, NSCLC, and non-NSCLC, in addition to the distribution of miR-182-5p in groups of different clinical variables, were calculated to determine the clinical significance of miR-182-5p. The statistical analysis of the miRNA-seq data has been described in detail elsewhere [39]. The meta-analysis included all the in-house RT-qPCR, miRNA-seq, miRNA-chip, and literature data. Pooling of the standard mean difference (SMD) and creation of summary receiver operating characteristic (SROC) curves from all the included studies was done to determine the differential expression miR-182-5p and its potential utility in distinguishing NSCLC and noncancer cases. Details on the data processing and statistical analysis in the meta-analysis have been described in previous studies [40,41].

Functional annotation of candidate target genes of miR-182-5p and construction of a protein-protein interaction (PPI) network
Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were conducted using the ClusterProfiler package in R software v.3.5.2 to explore the enrichment of candidate target genes in biological process, cellular component, and molecular function pathways. Items with a p < 0.05 were considered statistically significant. The top 15 significant biological process, cellular component, and molecular function terms, as well as the top 10 significant KEGG pathway terms, were visualized as a bubble plot and chord plot using the GOplot package of R software v.3.5.2. A PPI network was subsequently built using the Search Tool for the Retrieval of Interacting Genes to illustrate the interactions between target genes.

Validation of miR-182-5p targeting of HOXA9
In a previous study, we detected the expression level of HOXA9 in the same cohort of NSCLC patients using the RT-qPCR [42]. The primers for HOXA9  validate the direct target binding between HOXA9 and miR-182-5p. The 3'UTR of HOXA9 (wild type or mutation type) comprising putative miR-182-5p binding sites was cloned into a psiCHECK-2 luciferase reporter vector (Promega, USA) to generate psiCHECK-HOXA9 3′-UTRs or psiCHECK-HOXA9-mut 3′UTRs. HEK-293 T cells were co-transfected with an miR-182-5p mimic, a negative mimic control, and a reporter vector of the psiCHECK-HOXA9 3′UTR or psiCHECK-HOXA9-mut 3′UTR. After incubation for 27 h, the luciferase activity was measured using dual luciferase assay (Promega, USA) according to the manufacturer's protocol. Luciferase activity was inferred based on the ratios of Renilla and firefly luciferase activity. Each experiment was repeated three times.

Results
Evaluation of the clinicopathological significance of miR-182-5p in NSCLC Analysis of RT-qPCR data The analysis of the RT-qPCR data demonstrated that miR-182-5p was significantly upregulated in LUAD tissues as compared with that in paired non-LUAD lung tissues (P < 0.001, Table 1, Additional file 1: Fig. S1). In general, miR-182-5p expression level was markedly higher in the majority of NSCLC tissues than in paired noncancer tissues (P < 0.001, Table 2, Additional file 2: Fig. S2). Overexpression of miR-182-5p in LUAD and NSCLC was strongly associated with clinical parameters including tumor size, TNM stage, and lymph node metastasis (P < 0.05, Tables 1 and 2). As shown by the ROC curve in Additional files 3 and 4: Fig. S3 and S4, miR-182-5p performed moderately well in differentiating LUAD from noncancer lung tissues and better in differentiating NSCLC tissues from noncancer lung tissues

Analysis of miRNA-seq data
In total, miRNA-seq data were obtained for 784 NSCLC tissues (LUAD, n = 448; LUSC, n = 336) and 89 noncancer tissues. The clinicopathological significance of miR-182-5p in LUAD and NSCLC is summarized in Table 3 and Table 4, respectively. In both LUAD and NSCLC tissues, miR-182-5p expression was markedly higher as compared with that in noncancer tissues (P < 0.001, Additional files 1 and 2: Fig. S1 and S2, Tables 3 and 4). As shown by the ROC curves in Additional files 3 and 4: Fig. S3 and S4, miR-182-5p expression appeared to distinguish LUAD and NSCLC from noncancer tissues (AUC = 0.98 and AUC = 0.96, respectively). The Kaplan-Meier curves revealed no significant relationship between miR-182-5p expression and survival of NSCLC patients and LUAD patients (data not shown). Thus, the prognostic role of miR-182-5p in NSCLC remains unclear and needs to be studied in future work.

Analysis of miRNA-chip data
The initial search of the GEO database revealed 3204 studies. Of these, 248 studies were excluded Fig. 3 The distinguishing ability of miR-182-5p in NSCLC tissues based on data from miRNA-chips. a. SROC curves; b. Forest plot for sensitivity and specificity; c. Summary of positive likelihood ratio and negative likelihood ratio after screening the titles and abstracts. The final analysis included data of 25 eligible miRNA-chips on 1656 NSCLC samples (LUAD, n = 350) and 948 noncancer samples. Several of the datasets that contained information on miR-182-5p expression in LUSC (GSE29248, GSE47525, GSE19945, GSE51853, and GSE74190) have been mined in previous work [31]. The characteristics of all the included miRNAchip data are listed in Table 5. The differential expression of miR-182-5p and the discriminatory ability of miR-182-5p in distinguishing LUAD and NSCLC tissues from noncancer tissues are displayed in Additional file 1-6: Fig. S1-6. The forest plots in Figs. 1 and 2 support a noticeable increase in the miR-182-5p level in LUAD and NSCLC as compared with the level in noncancer lung samples (SMD = 0.81, 95% confidence interval [CI] = 0.59-1.04; SMD = 0.68, 95% CI = 0.59-0.77).
Due to obvious heterogeneity among individual studies (I 2 = 82.8%, P < 0.001; I 2 = 89.2%, P < 0.001), random-effect models were applied to merge the estimates. A subgroup analysis based on the source of the samples was employed to determine the origin of the heterogeneity. The pooled SMD of miR-182-5p expression in LUAD serum and tissue samples was 0.11 (− 0.20-0.43) and 1.55 (1.23-1.88), respectively (Fig. 1). For NSCLC, the pooled SMD of miR-182-5p expression in NSCLC serum and tissue samples was 0.29 (0.13-0.45) and 0.88 (0.77-0.99), respectively, suggesting that upregulation of miR- Fig. 4 The distinguishing value of miR-182-5p in NSCLC serum based on data from miRNA-chips. a. SROC curves; b. Forest plot for sensitivity and specificity; c. Summary of positive likelihood ratio and negative likelihood ratio 182-5p expression was more obvious in the tissue samples than in the serum samples (Fig. 2). A subsequent sensitivity analysis and test for publication bias reported no eccentric study and no publication bias (Figs. 1 and 2). SROC curves accompanied by forest plots of the sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio for NSCLC suggested that miR-182-5p expression in tissue better discriminated cancerous versus noncancerous tissue than miR-182-5p expression in serum (AUC = 0.93 and AUC = 0.69, respectively; Figs. 3 and 4).
Results of the meta-analysis incorporating in-house RT-qPCR data, miRNA-seq data and miRNA-chip data Based on the inclusion and exclusion criteria for the literature search, no studies were eligible for inclusion in the meta-analysis. Thus, the meta-analysis included 2564 NSCLC samples (LUAD, n = 899) and 1161 noncancer samples obtained from the in-house RT-qPCR, miRNA-seq, and miRNA-chip data analyses. The results of this comprehensive meta-analysis were consistent with those of the GEO meta-analysis, which confirmed upregulation of miR-182-5p in NSCLC tissues (SMD = 0.89 (0.55-1.22) Figs. 6). Upregulation of miR-182-5p expression was more apparent in the tissue samples than in the serum samples, and miR-182-5p expression in the tissue samples had stronger discriminating power in terms of cancer versus noncancer than miR-182-5p expression did in serum samples (Figs. 3, 4, 5, 6 and 7).
Molecular mechanism of miR-182-5p in NSCLC Functional annotation of candidate target genes in a PPI network In total, 774 genes were identified as candidate target genes in eight of the 12 prediction platforms (Additional file 7). As shown in Fig. 8 and Table 6, these candidate target genes were significantly enriched in biological processes, such as axonogenesis, axonal development, and Ras protein signal transduction. According to the chord plot in Fig. 8, these target genes appeared to mainly participate in pathways involved in axonal guidance, melanogenesis, and longevity regulation in multiple species. The complicated interactions between the candidate target genes were illustrated in a PPI network (Fig. 9).

Validation of miR-182-5p targeting of HOXA9
Among the candidate target genes, we selected HOXA9 and studied the relationship between it and miR-182-5p.

Discussion
MiRNAs are important in the occurrence and development of LC [43][44][45][46][47][48]. Recent studies reported that dysregulation of the expression of multiple miRNAs, including miR-182-5p, was significantly correlated with tumorigenesis of LC [34]. Although several studies have demonstrated the oncogenic effect of miR-182-5p in NSCLC [49][50][51][52], interactions between miR-182-5p and target genes in NSCLC remained unclear. In particular, the molecular mechanism of miR-182-5p in NSCLC was unclear. We previously demonstrated the oncogenic consequences of miR-182-5p in LUSCs through a combinatory analysis of data from RT-qPCR assays of 23 Fig. 6 The comprehensive meta-analysis for miR-182-5p expression in NSCLC. a. Forest plot for overall SMD; b. Subgroup analysis; c. Funnel plot of publication bias; d. Sensitivity analysis samples obtained from LUSC patients treated in our hospital, miRNA-seq data, and miRNA-chip data. We hypothesized that the expression pattern of miR-182-5p was similar in all the subtypes of NSCLC. In the present study, miR-182-5p was overexpressed in LUADs according to data from RT-qPCR assays of 124 samples obtained from NSCLC patients treated in our hospital, miRNA-seq data, and miRNA-chip data. Thus, we investigated the clinicopathological significance of miR-182-5p in NSCLC using in-house RT-qPCR data, miRNA-seq data, miRNA-chip data, and data in the scientific literature to explore the underlying molecular mechanism via a functional analysis of target genes.
The aforementioned data supported marked upregulation of miR-182-5p in NSCLC. Furthermore, the results of the RT-qPCR assays supported the influence of upregulated miR-182-5p on malignant clinical progression of NSCLC, which was consistent with the findings of previous studies [49][50][51][52]. It should be noted that there were some contradictions between the results of the RT-qPCR assays and those of the miRNA-seq data analysis. The discord might stem from different sources of patient cohorts and methods for calculating miR-182-5p expression. The expression of miR-182-5p in the NSCLC samples analyzed using the in-house RT-qPCR was calculated based on the 2-Δcq algorithm In contrast, miR-182-5p expression in the miRNA-seq data was log2 (total_ reads per million + 1) transformed in IlluminaHiSeq_miRNASeq platform. Nevertheless, miR-182-5p was upregulated in both the LUAD and NSCLC cohorts according to the miRNA-seq data, and miR-182-5p exhibited a trend toward elevated expression in samples from patients with malignant clinical progression of NSCLC, which was in agreement with the overall results.  The SROC curves generated from all the datasets suggested that miR-182-5p could differentiate between LUAD or NSCLC and noncancer lung tissues. We believe that the large number of LC samples (NSCLC, N = 2564; non cancer, N = 1161) included in the present study support the findings.
To yield a deeper understanding of the molecular basis of the role of miR-182-5p in the carcinogenesis of NSCLC, we carried out functional annotations for candidate target genes and created a PPI network. The results indicated that miR-182-5p may exert an oncogenic influence on NSCLC via involvement in various biological processes, such as axonogenesis, axonal development, and Ras protein signal transduction, as well as in pathways including axonal guidance, melanogenesis, and longevity regulation in multiple species. The intricate regulatory network between the candidate target genes in the PPI network indicated that cooperation or antagonism between target genes may constitute an important link in the course of NSCLC.
Among the candidate target genes, HOXA9, a member of the HOX gene family, encodes a series of transcription factors with critical roles in cancer [53]. Previous research showed that HOXA9 had oncogenic functions in hematologic cancers and anticancer effects in breast cancer and NSCLC [54,55]. In this study, to shed light on the regulatory relationship between miR-182-5p and HOXA9, we studied the expression level of HOXA9 in NSCLC and verified the relationships between miR-182-5p and HOXA9 through a correlation analysis and dual-luciferase reporter assay. The results showed that upregulation of miR-182-5p in LUAD or NSCLC was significantly correlated with downregulation of HOXA9 in LUAD or NSCLC. The direct regulatory association between miR-182-5p and HOXA9 was confirmed by the dual-luciferase reporter assay. Based on these findings, we conclude that miR-182-5p may affect the initiation and development of NSCLC by targeting HOXA9 to diminish the tumor-inhibitory effect of HOXA9 on NSCLC.
Several limitations of this study should be acknowledged. First, we did not validate the oncogenic effect of miR-182-5p on biological events of NSCLC through in vitro or in vivo experiments. Second, this study focused on the clinicopathological significance of miR-182-5p and the miR-182-5p-centered molecular mechanism in NSCLC. Alterations in the expression of various genes, such as EGFR, ALK, ROS1, KRAS, and BRAF, play essential roles in NSCLC, and these genes serve as targets of chemotherapy [56]. We did not explore the interactions between miR-182-5p and these genes in NSCLC. Third, the diagnostic value of miR-182-5p in serum was not verified in a large clinical NSCLC sample. Exosomal miRNAs have potential as diagnostic biomarkers for cancers because of their stability, nondegradability, and ease of detection [57]. The diagnostic value of exosomal miR-182-5p in NSCLC was not studied in current work.

Conclusions
In conclusion, the oncogenic role of miR-182-5p in NSCLC was confirmed by comprehensively analyzing data obtained from RT-qPCR assays, miRNA-seq and miRNA-chip database. Multiple target genes, including HOXA9, may play a role in the molecular mechanism of HOXA9 in NSCLC.
Additional file 1: Figure S1. Differential expression of miR-182-5p in LUAD and noncancer lung tissues based on data from in-house RT-qPCR, miRNA-seq and miRNA-chips. The distribution of miR-182-5p in LUAD and noncancer lung tissues was illustrated in the color of blue and red,  Table S1. Predicted target genes of hsa-miR-182-5p from miRWalk database. Putative target genes of hsa-miR-182-5p were predicted by 12 algorithms within mRNA selected regions. LG analyzed all experiment results, prepared for tables and figures, and revised the manuscript. SBY contributed to data analysis and paper writing. JY contributed to the conception, modification of the study and instruction of experiment and revision work. JLK contributed to the conception, modification of the study and instruction of experiment and revision work. KS contributed to the conception and modification of the study, and writing of the paper draft. FCM contributed to the conception and modification of the study, and writing of the paper draft. LZH helped to perform the RT-qPCR and analyze the RT-qPCR data. JL contributed to analyze the target genes by GO, KEGG pathway and PPI network and corrected the paper. SYY analyzed the data from RT-qPCR, GEO and TCGA database and wrote the results. RQH performed RT-qPCR, collected and analyzed the data from TCGA and GEO database and corrected the part of the results. XHH contributed to the design of the study, guided the study method, and corrected the paper. GC contributed to the design of the study, supervised all experiments and corrected the paper. All authors read and approved the final manuscript.

Funding
The study was supported by funds from National Natural Science
Ethics approval and consent to participate The study was approved by the Ethics Committee of the First Affiliated Hospital of Guangxi Medical University. Written informed consent was obtained from all of the patients.