Non-small cell lung cancer (NSCLC) is one of the leading causes of cancer mortality. MicroRNAs (miRNAs), small noncoding RNAs, regulate the expression of genes that play roles in human cancer via posttranscriptional inhibition.
To identify the potential miRNA biomarkers in NSCLC, we downloaded the miRNA expression profile (ID: GSE29248) of NSCLC from the Gene Expression Omnibus (GEO) database and analyzed the differentially expressed miRNAs in NSCLC tissue compared with normal control tissue. Then the targets of these differentially expressed miRNAs were screened and used in network construction and functional enrichment analysis.
We identified a total of 17 miRNAs that showed a significantly differential expression in NSCLC tissue. We found that miR-34b and miR-520h might play important roles in the regulation of NSCLC, miR-22 might be a novel biomarker as an oncogene, and miR-448 might promote, while miR-654-3p prevents, NSCLC progression.
Our study may provide the groundwork for further clinical molecular target therapy experiments in NSCLC.
Tumori 2015; 101(1): 104 - 110
Article Type: ORIGINAL RESEARCH ARTICLE
AuthorsChi Xu, Yisheng Zheng, Duohuang Lian, Shixin Ye, Jinrong Yang, Zhiyong Zeng
- • Available online on 16/02/2015
- • Published in print on 20/03/2015
- • Accepted on 30/06/2015
This article is available as full text PDF.
Lung cancer is the most common cause of cancer-related mortality in patients worldwide (1), of which non-small cell lung cancer (NSCLC) accounts for 85% of all pulmonary carcinomas. Despite the recent advances in therapies for NSCLC, the high mortality rate of NSCLC patients has not decreased markedly over the years (2). It is thus worth exploring more effective and safe treatment strategies.
MicroRNAs (miRNAs), a class of endogenous short noncoding RNAs, negatively regulate gene expression by binding to their targets in both coding and untranslated regions via base-pairing with complementary sequences with the mRNA molecules (3-4-5). Over the years, an increasing number of miRNAs have emerged as involved in lung cancer pathogenesis, including in the development and metastasis process. MiRNAs have been used as biomarkers for early diagnosis, targeted therapy and clinical prognosis, which provides a new direction for diagnosis and treatment (6). In recent years, exploration of the targeted therapy of miRNAs has attracted much attention and has made great progress (2, 7-8-9).
Targeting miRNA expression in cancer is mainly classified into 2 strategies, including the direct strategy that either blocks the expression of oncogenic miRNAs by the use of oligonucleotides or substitutes for the loss of expression of a tumor suppressor miRNA via virus-based constructs, and the indirect strategy that modulates miRNA expression by targeting its transcription and processing, through drug therapies (9, 10). The let-7 microRNA directly represses cancer growth in the lung of the mouse (7). Fontana et al demonstrated that antagomir-17-5p treatment can efficiently inhibit tumor growth in vivo (8). Anti-miR cotransfections of miR-34c, miR-145 and miR-142-5p significantly repressed lung cancer cell growth (11).
MiRNA expression proﬁling is becoming an increasingly important tool to reveal gene activity in cancer, and many studies indicate that miRNAs can act as oncogenes and/or tumor suppressor genes (2, 9, 12). The primary aim of our study was to identify potential miRNA biomarkers in NSCLC via miRNA expression profiles.
We extracted miRNA expression profile (ID: GSE29248) from the National Center of Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database, which had been deposited by Ma et al (13). This microarray study was carried out in the tissue bank of Zhoushan Hospital (China) and included 12 chips from 6 NSCLC tissues (3 adenocarcinomas and 3 squamous cell carcinomas of the lung) and 6 matched normal controls from adjacent tissues. These tissues were all obtained from male patients. Among them, 4 patients were older than 60 years and 2 patients younger than 60 years. The tumor stages were clearly defined as stage I in 1 patient, while the other 5 were in stage II, III and IV. Additionally, regarding T classifications based on the 7th edition of the Union for International Cancer Control (UICC), 2 patients had T1-T2 and 4 patients had T3-T4 cancer. The platform used was the Illumina Human v2 MicroRNA expression beadchip. We downloaded the raw data and the probe annotation file.
The downloaded microarray data was normalized using robust multiarray average (RMA) (14) algorithm in R, Student’s
Target genes for differentially expressed miRNAs
Target genes of the 6 most significantly differentially expressed miRNAs were retrieved from the miRecords and miRTarBase databases. MiRecords, is an integrated resource for experimentally validated animal miRNA–target interactions (MTIs), component of 1,135 records of validated MTIs between 301 miRNAs and 902 target genes in 7 animal species (15). As a database updated frequently, miRTarBase has accumulated 3,576 experimentally verified MTIs between 657 miRNAs and 2,297 target genes among 17 species by manually surveying the pertinent literature related to functional studies of miRNAs (16). In our study, we integrated target genes from the 2 databases and extracted target genes validated experimentally. Those target genes shared by both databases were deemed to be more reliable.
Interaction network analysis
To annotate functional interactions between those target genes, we constructed interaction networks to discern how they were associated with other genes in the networks. The Osprey software platform provides visualization and manipulation of complex interaction networks, using the General Repository for Interaction Datasets (GRID) and the Biomolecular Interaction Network Database (BIND) as databases (17). We used the Osprey platform to annotate functional interactions between differentially expressed miRNAs and target genes and protein–protein interaction networks (18).
Gene ontology enrichment analysis
Gene ontology (GO) analysis is increasingly used for functional studies of microarray data (19). The Database for Annotation, Visualization and Integrated Discovery (DAVID), a database for annotation, visualization and integrated discovery, provides functional annotation for large-scale genomic or transcriptomic data (20). To investigate the functions of miRNAs in the development of NSCLC, we used DAVID to identify overrepresented GO categories in biological processes, based on a hypergeometric distribution algorithm. The p value was adjusted by Henjamin-Hochberg method based on the multest package to produce the false discovery rate (FDR). The FDR <0.05 was chosen as the threshold (21).
Screening of different expressed miRNAs
The microarray data that downloaded from GEO database was normalized using the RMA algorithm. As shown in
Microarray gene expression profile. The pink samples are non-small cell lung cancer (NSCLC) tissue, and the blue samples are adjacent normal tissue.
MiRNAs differentially expressed in NSCLC tissue compared with normal tissue
|MiRNAs in italics, represent the top 3 down-regulated and 3 up-regulated miRNAs, respectively.|
|MiRNA = microRNA; NSCLC = non-small cell lung cancer.|
|*the significant differentially expressed miRNA.|
Screening of target mRNAs
As miRNAs negatively regulate gene expression by targeting mRNAs at the posttranscriptional level, it was necessary to identify putative target genes for a better understanding of their function. The total target genes for the selected miRNAs in this study were predicted from the miRecords and miRTarBase database. Finally, 31 target genes corresponding to the 6 miRNAs were obtained (
Target genes of differentially expressed miRNAs from miRecords and miRTarBase databases
|miRNA||Target in miRTarBase||Target in miRecords|
|MiRNA = microRNA.|
|*The predicted target genes of down-regulated miRNAs.|
|†The predicted target genes of up-regulated miRNAs.|
|hsa-miR-520h||ABCG2, CDKN1A, ID1, ID3, VEGFA||ABCG2|
|hsa-miR-34b||CDK4, CDK6, MET, MYC, ZAP70, VEGFA,||MYC, NOTCH4, NOTCH2, HMGA2, Notch1, Bcl-2|
|hsa-miR-22||ACVR1C, BDNF, BMP7, ESR1, HDAC4, HTR2C, MAOA, MYCBP, TFRC, PPARA, RGS2||MAX, PTEN, BMP7, ESR1, PPARA|
Interaction network construction of differentially expressed miRNA target genes
We used the Osprey software for all interactions between the 6 differentially expressed miRNAs and their target genes, along with the target gene interaction network (
Interaction network constructed between differentially expressed (microRNAs (miRNAs) and their interactive genes.
GO enrichment analysis
The biological processes of the total selected target genes and miRNAs from the constructed network were enriched using the GO database. In this study, 24 and 12 enriched terms were found, respectively, corresponding to the 3 down-regulated and the 3 up-regulated miRNAs with our cutoff criterion (
The enriched GO terms in differentially expressed miRNA target interaction network
|GO ID||Corr p value||x||Description|
|Corr = Corrected; GO = gene ontology; miRNA = microRNA; X = number of genes in the enrichment.|
|*Enriched GO terms in the interaction network of the down-regulated miRNA.|
|†Enriched GO terms in the interaction network of the up-regulated miRNA.|
|48523||2.68E-06||12||Negative regulation of cellular process|
|10468||2.82E-06||14||Regulation of gene expression|
|48522||5.00E-06||12||Positive regulation of cellular process|
|48519||5.31E-06||12||Negative regulation of biological process|
|19222||7.19E-06||15||Regulation of metabolic process|
|48856||8.35E-06||13||Anatomical structure development|
|48869||9.68E-06||11||Cellular developmental process|
|48518||1.07E-05||12||Positive regulation of biological process|
|60255||1.07E-05||14||Regulation of macromolecule metabolic process|
|32501||2.01E-05||15||Multicellular organismal process|
|50896||2.03E-05||14||Response to stimulus|
|10556||8.38E-05||12||Regulation of macromolecule biosynthetic process|
|31326||1.23E-04||12||Regulation of cellular biosynthetic process|
|50794||1.23E-04||16||Regulation of cellular process|
|9889||1.29E-04||12||Regulation of biosynthetic process|
|31323||1.38E-04||13||Regulation of cellular metabolic process|
|50789||2.15E-04||16||Regulation of biological process|
|80090||4.40E-04||12||Regulation of primary metabolic process|
|42221||2.98E-06||11||Response to chemical stimulus|
|48519||2.53E-04||10||Negative regulation of biological process|
|32501||2.53E-04||13||Multicellular organismal process|
|50896||3.32E-04||12||Response to stimulus|
|50794||1.62E-02||12||Regulation of cellular process|
|44238||1.78E-02||11||Primary metabolic process|
|50789||2.10E-02||12||Regulation of biological process|
|44237||2.70E-02||10||Cellular metabolic process|
Recently, dysregulation of miRNA expression was found to contribute to the initiation and progression of cancer. Depending on their target genes (22) miRNAs can act either as oncogenes or tumor suppressor genes (23).
In this study, we analyzed the miRNA profiling of NSCLC from the GEO database to identify new miRNA targets for therapeutic intervention. A total of 17 miRNAs displayed a significantly differential expression in NSCLC tissue compared with adjacent nonmalignant tissue. Furthermore, we performed interaction network analysis of the top 3 up-regulated (hsa-miR-22, hsa-miR-518b and hsa-miR-654-3p) and top 3 down-regulated differentially expressed miRNAs (hsa-miR-520h, hsa-miR-448 and hsa-miR-34b). Since miRNAs play important roles in posttranscriptional regression by targeting mRNAs, we performed interaction network analysis of these miRNAs and their target genes.
NOTCH1 and VEGFA were the identified hubs of miR-34b and miR-520h in the network of down-regulated miRNAs and targets (
Meanwhile, our findings showed that down-regulated miR-448 interacted with special AT-rich sequence-binding protein 1 (SATB1) in the network (
In addition, up-regulated miR-22 and miR-654-3p are believed to be crucial in the regulation of NSCLC, as their target genes are
Additionally, 5 of our selected miRNAs (except for miR-224) were not consistent with the results in the study by Ma et al (13), due to the different methods used in our studies. We used Student’s t-test in the multitest package to compare all of the individual expression values of miRNAs, and we selected differentially expressed miRNAs with a p value <0.05 and |log2 FC| >1 as threshold, while Ma et al calculated the mean expression values of miRNAs and then selected differentially expressed miRNAs using the SAM method, with a FDR <0.1 as the cutoff criteria. Besides, our analysis results were verified using the Limma package in R and the GEO2R software in the GEO database. Thus the selected miRNAs were different from those in the study by Ma et al Therefore, further experimental validation is required.
In conclusion, we identified the differentially expressed miRNAs in NSCLC tissue compared with noncancerous tissue. Our analysis identified several differentially expressed miRNA target genes, which might play crucial roles in NSCLC, including Notch1, VEGFA, SATB1, CDKN1A and PTEN. Further, we speculated that miR-34b and miR-520h might play important roles in the regulation of NSCLC by targeting genes in the notch signing. MiR-22 might be useful as a novel therapeutic agent as an oncogene. MiR-448 might function in promoting NSCLC cell metastasis via targeting SATB1, while miR-654-3p might play a role in preventing NSCLC by interacting with CDKN1A. Our study provides the groundwork for further clinical molecular target therapy experiments.
- Xu, Chi [PubMed] [Google Scholar] 1
- Zheng, Yisheng [PubMed] [Google Scholar] 2
- Lian, Duohuang [PubMed] [Google Scholar] 1
- Ye, Shixin [PubMed] [Google Scholar] 1
- Yang, Jinrong [PubMed] [Google Scholar] 1
- Zeng, Zhiyong [PubMed] [Google Scholar] 1, * Corresponding Author (firstname.lastname@example.org)
Department of Thoracic and Cardiovascular Surgery, Fuzhou General Hospital, Fujian Medical University, Fuzhou, Fujian - PR China
Department of Respiratory and Critical Care Medicine, Fuzong Clinical College of Fujian Medical University, Fuzhou General Hospital, Fuzhou, Fujian - PR China
C. Xu and Y. Zheng contributed equally to this work and should be considered co–first authors