OR Oncology Reports 1021-335X 1791-2431 D.A. Spandidos 10.3892/or.2020.7526 or-43-05-1437 Articles Identification of lung adenocarcinoma biomarkers based on bioinformatic analysis and human samples DongSiyuan MenWanfu YangShize XuShun Department of Thoracic Surgery, First Hospital of China Medical University, Shenyang, Liaoning 110001, P.R. China Correspondence to: Dr Shun Xu, Department of Thoracic Surgery, First Hospital of China Medical University, 155 North Nanjing Street, Shenyang, Liaoning 110001, P.R. China, E-mail: xushun610539@sina.com 052020 28022020 43 5 1437 1450 02102018 23012020 Copyright: © Dong et al. 2020 This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

Lung adenocarcinoma is one of the most common malignant tumors worldwide. Although efforts have been made to clarify its pathology, the underlying molecular mechanisms of lung adenocarcinoma are still not clear. The microarray datasets GSE75037, GSE63459 and GSE32863 were downloaded from the Gene Expression Omnibus (GEO) database to identify biomarkers for effective lung adenocarcinoma diagnosis and therapy. The differentially expressed genes (DEGs) were identified by GEO2R, and function enrichment analyses were conducted using the Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO). The STRING database and Cytoscape software were used to construct and analyze the protein-protein interaction network (PPI). We identified 376 DEGs, consisting of 83 upregulated genes and 293 downregulated genes. Functional and pathway enrichment showed that the DEGs were mainly focused on regulation of cell proliferation, the transforming growth factor β receptor signaling pathway, cell adhesion, biological adhesion, and responses to hormone stimulus. Sixteen hub genes were identified and biological process analysis showed that these 16 hub genes were mainly involved in the M phase, cell cycle phases, the mitotic cell cycle, and nuclear division. We further confirmed the two genes with the highest node degree, DNA topoisomerase IIα (TOP2A) and aurora kinase A (AURKA), in lung adenocarcinoma cell lines and human samples. Both these genes were upregulated and associated with larger tumor size. Upregulation of AURKA in particular, was associated with lymphatic metastasis. In summary, identification of the DEGs and hub genes in our research enables us to elaborate the molecular mechanisms underlying the genesis and progression of lung adenocarcinoma and identify potential targets for the diagnosis and treatment of lung adenocarcinoma.

 lung adenocarcinoma TOP2A AURKA biomarker bioinformatics human samples Introduction

Non-small cell lung cancer (NSCLC) is the most common malignant tumor globally and is associated with an extremely high mortality rate (1). The incidence of NSCLC continues to surge globally. As with other tumors, lung adenocarcinoma which accounts for about 35–45% of all lung malignant tumors is a heterogeneous disease characterized by high rates of genetic mutation (2). Despite the emergence of diverse new approaches for the treatment of lung adenocarcinoma, such as targeted and immune therapy, long term survival is still poor (3,4). One of the main reasons for this is that most patients are diagnosed at an advanced stage. Thus, it is necessary to understand the molecular mechanisms behind lung adenocarcinoma genesis, growth and progression, and identify biomarkers that can be detected during the early stages of the disease.

Recently, high-throughput bioinformatic technologies such as microarrays have been widely used to screen for differentially expressed genes (DEGs) and identify the functional pathways involved in the genesis and development of lung adenocarcinoma. However, the reliable results are not easy to obtain due to the false-positive rates that may exist in every independent microarray analysis. Thus, we downloaded three original mRNA data sets (GSE32863, GSE63459 and GSE75037) from the Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/) and aimed to identify the DEGs between normal lung and lung adenocarcinoma tissues. Next, the Kyoto Encyclopedia of Genes and Genomes (KEGG; http://www.genome.jp/) and Gene Ontology (GO; http://www.geneontology.org) databases were used to identify biological processes enriched in DEGs, and integrated protein-protein interaction (PPI) network analysis was used to help us understand the molecular mechanisms underlying lung adenocarcinoma genesis and development. Sixteen hub genes and 376 DEGs were identified, which could be potential target genes and candidate biomarkers for lung adenocarcinoma. To minimize the false-positive rate of the microarray analysis, the results were then confirmed in cell lines and human sample tissues.

 Materials and methods <sec> <title>Microarray data

GEO is a public functional genomics data repository supporting MIAME-compliant data submissions. It accepts sequence based and array data. Tools are provided to help users query and download experiments and curated gene expression profiles (5). The GSE32863, GSE63459 and GSE75037 datasets produced by the Illumina HumanWG-6/Ref-8 v3.0 expression beadchip platform (Illumina Inc; http://www.illumina.com) were downloaded for further analysis. The GSE63459 dataset contains data from 33 lung adenocarcinoma tissue samples and 32 adjacent normal tissue samples (6). The GSE32863 dataset contains data from 58 lung adenocarcinoma tissue samples and 58 fresh frozen adjacent non-cancerous samples (7). Moreover, the GSE75037 dataset contains data from 84 lung adenocarcinoma and 84 adjacent non-cancerous lung tissue samples (8).

 Identification of DEGs

GEO2R is an online web tool (https://www.ncbi.nlm.nih.gov/geo/geo2r/) provided by the GEO for comparing GEO series to identify DEGs across experimental conditions. The cutoff criteria were set to P-value <0.05 and logFC (fold change)>1. We excluded probe sets without exact gene symbols, and genes with two or more probe sets were averaged.

 KEGG and GO enrichment analyses of DEGs

The functional annotation tools version 6.7 of the Database for Annotation, Integrated Discovery and Visualization (9) (DAVID; http://david.ncifcrf.gov) were used to extract biological information about our DEGs. KEGG is a public database used for understanding the functions and abilities of biological systems, such as cells, the organism and the ecosystem, from molecular-level information, especially large-scale molecular datasets acquired by genome sequencing (10). GO was also used to annotate genes and further analyze their biological functions. The DAVID online database was used to analyze the function and biological process of the screened DEGs. P<0.05 was considered to indicate statistical significance.

 PPI network construction and analysis

The Search Tool for the Retrieval of Interacting Genes (STRING; http://string-db.org; version 10.0) was used to construct the PPI network from the DEGs (11). The sources for interactions are text mining, databases, experiments, neighborhood, co-expression, gene co-occurrence and fusion. We set the minimum required interaction score to 0.4. Cytoscape version 3.4.0 software (12) was used to visualize the molecular interaction networks of the DEGs. The APP plug-in, Molecular Complex Detection (MCODE) (13), was used to arrange the network topology to cluster densely connected genes. After the PPI networks was constructed, its key modules were searched by using the MCODE application. The parameter for inclusions are MCODE score >5, degree cutoff=2, node score cutoff=0.2, node density cutoff=0.1, k-score=2 and Max depth=100. Then, DAVID was used to perform the GO and KEGG analyses for these most significant modules.

 Hub gene screen and analysis

The criterion for being a hub gene selection was degree ≥10. Further analysis was performed using the cBioPortal online platform (http://www.cbioportal.org) to build the network of the DEGs and co-expressing genes (14). The mutation rates of the hub genes were also measured with the cBioPortal platform (15). Cytoscape's Biological Networks Gene Oncology tool (BiNGO) (version 3.0.3) was used for the biological process analysis and visualization (16). The University of California Santa Cruz (USCS) platform was used to analyze the hierarchical clustering of hub genes (17). Kaplan-Meier curves for overall survival and disease-free survival with these hub genes were obtained from cBioPortal. The expression profiles of DNA topoisomerase IIα (TOP2A) and aurora kinase A (AURKA) in 20 types of malignant tumors were analyzed and displayed using the Oncomine database (http://www.oncomine.com) (18).

 Analysis of TOP2A and AURKA expression in cell lines

To confirm our bioinformatics results, reverse transcription and quantitative real-time PCR (RT-qPCR) were conducted on lung adenocarcinoma (HCC827, A549 and H1975) cell lines and a human bronchial epithelial (HBE) cell line. A549, HCC827 and H1975 cells were purchased from the Shanghai Cell Bank (Shanghai, China) and were cultured using Roswell Park Memorial Institute (RPMI)-1640 medium (Gibco; Thermo Fisher Scientific, Inc.). The medium was supplemented with 100 U/ml penicillin and 100 µg/ml streptomycin (Gibco; Thermo Fisher Scientific, Inc.), and 10% fetal bovine serum (FBS) (Gibco; Thermo Fisher Scientific, Inc.) under a 5% CO2-containing humidified atmosphere at 37°C. Total RNA was extracted using TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc.). The isolated RNA was reverse-transcribed into cDNA using a reverse transcription kit (Takara, Dalian, China). RT-qPCR was performed as described in our previous research, 2 min at 50°C, 10 min at 95°C, 40 cycles at 95°C for 15 sec, and 60°C for 30 sec (19) and the results were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) levels. Primers were as follows: TOP2A (forward, 5-AGGATTCCGCAGTTACGTGG-3 and reverse, 5-CATGTCTGCCGCCCTTAGAA-3) (20) and AURKA (forward, 5-TTGGGTGGTCAGTACATGCTC-3 and reverse, 5-GTGAATTCAACCCGTGAT-3) (21) and GAPDH sense, 5′-CAATGACCCCTTCATTGACC-3′ and reverse, 5′-TGGAAGATGGTGATGGGATT-3′. The statistical analyses were conducted using SPSS version 21 (IBM Corp.). Results are displayed as mean ± SEM and differences between the HBE and cancerous cell lines were analyzed by one-way ANOVA. We further used the Tukey test to determine the significance between each cancer cell line and HBE. P-value <0.05 was considered to indicate statistical significance. Each experiment was repeated three times.

 Analysis of TOP2A and AURKA expression in human samples

The Ethics Committee of the First Hospital of the China Medical University (Shenyang, Liaoning, China) approved our research. Written informed consent was received from all participants. Seventy-two lung adenocarcinoma and paired non-cancerous tissues were obtained between February 2013 and June 2014 from 35 women and 37 men, ranging in age from 38 to 75, with a median age of 60. Patients who had received chemotherapy, target therapy and radiotherapy or had a history of malignant tumor were excluded. All of the diagnoses were confirmed by two experienced pathologists. The resected samples were preserved at −80°C until the mRNA of TOP2A and AURKA extraction were needed. Differences between cancerous and non-cancerous tissues were compared using the paired Student's t-test.

 Results <sec> <title>Identification of DEGs in lung adenocarcinoma

A total of 5,874 genes were found to be differentially expressed in non-cancerous and lung adenocarcinoma tissues (432 in GSE63459, 4,037 in GSE75037 and 1,405 in GSE32863) after standardizing the microarray data. A total of 376 DEGs were found in all three datasets (Venn diagram, Fig. 1A), consisting of 293 downregulated and 83 upregulated genes.

 GO and KEGG enrichment analyses of the DEGs

The DAVID online database was used to further analyze the biological classification, as well as functions and pathways enriched in DEGs. GO analysis showed that the biological processes (BP) of the DEGs were mainly involved in regulation of cell proliferation, the transforming growth factor β receptor signaling pathway, cell adhesion, biological adhesion and responses to hormone stimulus (Table I). Examination of their cell component showed that the DEGs were mainly located in the proteinaceous extracellular matrix, cell surface, cell-cell junction, and cell-substrate adherent junction. KEGG pathway analysis showed that the DEGs were mainly over represented in the TGF-β signaling pathway, cell adhesion molecules (CAMs), complement and coagulation cascades and ECM-receptor interaction (Table I).

 PPI network construction and module analysis

Cytoscape was used to build the DEG PPI network (Fig. 1B) and identify the most significant genes of the PPI network (Fig. 1C). Analysis of these genes with the DAVID platform found that they were mainly involved in M phases, cell cycle phase, the mitotic cell cycle and nuclear division (Table II).

 Hub gene selection and analysis

Sixteen genes were identified with a degree ≥10 and were defined as hub genes. The degree of each gene was calculated by the CytoScape software and represented the number of other genes with which it was connected. The hub gene symbol, full name, function and degree are listed in Table III. cBioPortal was then used to construct a network of the 16 hub genes and their co-expressed genes. The results of this and analysis of the BP are shown in Fig. 2A and B, respectively. oncoPrint analysis with the cBioPortal showed that TOP2A and AURKA have the highest genetic mutation rates of the hub genes in lung adenocarcinoma at 8 and 14%, respectively (Fig. 2C). Hierarchical clustering analysis revealed that these 16 hub genes could generally differentiate both primary and recurrent lung adenocarcinoma tissues from their adjacent non-cancerous lung tissues (Fig. 2D).

 Clinical significance of TOP2A and AURKA

Analysis of the association between these genes and disease-free survival and overall survival analysis was conducted using Kaplan-Meier curves in the cBioPortal platform. Lung adenocarcinoma patients with AURKA mutation had worse overall and disease-free survival and patients with ASPM (abnormal spindle microtubule assembly) mutation had worse disease-free survival (Fig. 3A and B). Moreover, AURKA and TOP2A had the highest node degrees at 22 and 31, respectively, implying that they may play significant roles in the genesis and development of lung adenocarcinoma. When analyzing the data from cBioPortal platform, we discovered that lung adenocarcinoma patients who had an AURKA mutation had reductions in overall survival (P=0.00192). However, this was not true for the TOP2A gene (P=0.775, Fig. 3A and B). The expression profile of AURKA and TOP2A in 20 types of human cancer tissues was displayed using the Oncomine database. TOP2A mRNA levels in bladder, brain, breast, colorectal, esophageal, kidney, gastric and sarcoma cancer tissues were higher than those in matched adjacent normal tissues (Fig. 4A). The AURKA mRNA levels in bladder, brain, breast, cervical, lung and liver cancer tissues were higher than those in adjacent matched normal tissues (Fig. 4B). When we analyzed six different datasets from the Oncomine database, we found that TOP2A and AURKA were significantly overexpressed in lung adenocarcinoma tissues compared with non-cancerous tissues (Fig. 4C and D) (7,22–28).

 Expression of TOP2A and AURKA in lung adenocarcinoma cell lines

To confirm the bioinformatics results, the expression of these two genes in lung adenocarcinoma cell lines and HBE cells were assessed by RT-qPCR. The mRNA levels of both TOP2A and AURKA were significantly higher in the lung adenocarcinoma cell lines than the HBE cells (Fig. 5A and B).

 Expression of TOP2A and AURKA in human lung adenocarcinoma and adjacent normal lung tissues

Table IV shows the clinicopathological characteristics of all the lung adenocarcinoma patients. TOP2A and AURKA were found to have 2.25- and 2.73-times higher expression levels in the cancer tissues than in the normal tissues, respectively (Fig. 5C and D). The 72 samples were then classified into two groups based on the relative expression levels of these two genes: The relatively high TOP2A and AURKA group (n=36, the first 36 cases having relative high TOP2A and AURKA were defined as the high-expression group; the remaining 36 cases were defined as the low-expression group) and the relatively low TOP2A and AURKA group (n=36). χ2 analysis was used to analyze the results. Analysis revealed that increased TOP2A expression was observed in tumors with a larger diameter. Increased AURKA expression was observed in tumors with a larger size and in lymphatic metastatic tumors (Tables IV and V).

 Discussion

Recent statistics on cancer globally revealed that lung cancer is the most commonly diagnosed malignant tumor and is the leading cause of cancer-related mortality accounting for 11.6% of all malignant tumors and 18.4% of cancer-related deaths. Lung adenocarcinoma is the most common subtype of malignant lung cancer, and its incidence is increasing rapidly (29). Air pollution and smoking are the two main etiological factors for lung adenocarcinoma (30,31). The microarray data of Sekine et al revealed that a human lung mucoepidermoid carcinoma cell line exposed to smoke with a charcoal filter had a total of 1,582 genetic mutations (32). Still, the molecular mechanisms underlying the genesis of lung adenocarcinoma remain unclear. Without an early diagnosis most patients are not candidates for curative therapies leading to the deeply unsatisfactory prognosis for the disease. Therefore, biological markers with satisfactory efficiency for early diagnosis and therapy are desperately needed. With the development of microarray technology, we are able to efficiently screen for changes in gene expression in lung adenocarcinoma, and this approach has been proven to be a very useful method for screening early stage biomarkers in both malignant and benign diseases (33–35).

In the present research, three mRNA microarray datasets were downloaded from the GEO database and subsequently analyzed to acquire differentially expressed genes (DEGs) between lung adenocarcinoma and adjacent non-cancerous lung tissues. In addition, a total of 376 DEGs were identified in the three datasets, consisting of 293 upregulated genes and 83 downregulated genes. The biological roles of the identified DEGs were then studied using KEGG and GO enrichment analyses. The downregulated genes were mainly overrepresented in cell division, nuclear division, DNA replication and second-messenger-mediated signaling, and the upregulated genes were mainly involved in cell cycle process, DNA metabolic process, the transforming growth factor (TGF)-β signaling pathway and angiogenesis. Previous articles have revealed that dysregulation of angiogenesis and activation of the TGF-β signaling pathway are associated with the carcinogenesis and progression of lung adenocarcinoma (36–38). In addition, DNA damage and metabolic process abnormalities often play a significant role in cell cycle regulation dysfunction and are associated with malignant tumors (39). In summary, our results are consistent with these previous reports and theories. KEGG enrichment analysis showed differences mainly in the TGF-β signaling pathway, complement and coagulation cascades, Cell adhesion molecules (CAMs), and ECM-receptor interaction, while changes identified by GO terms were mainly in regulation of cell adhesion, proliferation, the TGF-β receptor signaling pathway, biological adhesion and response to hormone stimulus.

Sixteen genes with degrees ≥10 were defined as hub genes (40). Two of these hub genes, TOP2A and AURKA, had the highest node degrees at 31 and 22, respectively. Gene mutation can promote tumorigenesis, thus the mutation rates of these 16 lung adenocarcinoma hub genes were screened with the cBioPortal platform. The four genes with the highest mutation rates were AURKA, TK1, CDC45 and TOP2A with mutation rates of 14, 13, 10 and 8% respectively, which indicates that these genes may play a significant role in tumorigenesis.

DNA topoisomerase II α (TOP2A) that encodes the DNA topoisomerase, an enzyme that alters and controls the topological state of DNA during transcription, has been shown to be correlated with an increased risk of developing brain metastases, drug resistance and an abnormal cell cycle (41–43). It is regarded as a target for several anticancer agents, such as etoposide and topotecan (44). In our research, the protein-protein interaction network (PPI) network revealed that TOP2A directly interacts with maternal embryonic leucine zipper kinase (MELK), CDC20, CCNB2, UBE2T, KIAA0101 and TK1, indicating that TOP2A plays a key role in lung adenocarcinoma. Two of these genes, CDC20 and MELK, are closely involved in tumorigenesis and the cell cycle. Cell division cycle 20 (CDC20) appears to function as a regulatory protein interacting with other proteins at several important phases in the cell cycle. CDC20 is activated in lung adenocarcinoma and overexpression is correlated with poor prognosis (45). MELK plays a key role in the proliferation and self-renewal of progenitor and tumor stem cells, and is overexpressed in lung adenocarcinoma, contributing to carcinogenesis. MELK is an effective target for kinase drugs (46). Moreover, the expression of TOP2A is upregulated in various tumors, such as colon and ovarian malignant tumors, and may be considered a sensitive biomarker for early detection and therapy of these tumors (47,48).

Aurora kinase A (AURKA) is a putative oncogene, associated with cell cycle-regulated kinase (49). GO annotations associated with this gene include protein tyrosine kinase activity, transferring phosphorus-containing groups and transferase activity. Overexpression of this gene is associated with several common features of malignant tumor cells, such as chromosomal instability, aneuploidy in mammalian cells and centrosomal duplication abnormalities (50,51). It has also been found to be overexpressed in several types of malignant tumors and has been associated with poor prognosis (52,53). The aurora kinase A inhibitor, alisertib, has been approved for therapy for solid tumors especially non-small cell lung cancer (NSCLC) and breast cancer, and has achieved satisfactory results (54). In addition, Chen et al revealed that AURKA antagonists can enhance the cytotoxicity of epidermal growth factor receptor-tyrosine kinase inhibitors (EGFR-TKIs) (55).

The relationships between TOP2A and AURKA expression and survival were further assessed with the cBioPortal platform. We found that changes in TOP2A expression were associated with a decrease in both disease-free and overall survival, although they were not statistically significant. While changes in AURKA expression were associated with significantly worse overall survival rates, but no significant change in disease-free survival rate. This may be explained by our study of 72 paired human samples which found that upregulation of AURKA was associated with lymphatic metastasis and that upregulation of TOP2A was only associated with tumor size. One of the underlying molecular reason may be that the overexpression of TOP2A arises from amplification and mutation, while the survival analysis using the cBioPortal platform was only based on the mutation of TOP2A. Thus, amplification but not mutation may result in overexpression of TOP2A, not related to changes in prognosis, although further study is needed to prove this hypothesis. Oncomine analysis revealed that higher AURKA mRNA levels were observed in colorectal cancer, breast cancer, lung cancer and sarcoma. Additionally, higher TOP2A mRNA levels were observed in colorectal cancer, breast cancer, lung cancer and brain cancer, indicating important roles of these two genes in the carcinogenesis and development of malignant tumors. However, these results also indicate that these two genes can only be used as broad-spectrum tumor markers as they cannot differentiate lung adenocarcinoma from other malignant tumors.

The UCSC (University of California Santa Cruz) cancer platform was used to hierarchically cluster the hub genes. Their expression levels in both the primary and recurrent tumors were upregulated compared with the normal tissues. The expression levels of these hub genes in recurrent tumors were higher than in the primary tumor. TOP2A in particular was found to have the highest expression in the recurrent tumors. We therefore infer that TOP2A and other hub genes may be regarded as early biomarkers for monitoring tumor recurrence.

To confirm our results, we detected the expression of TOP2A and AURKA in lung adenocarcinoma and human bronchial epithelial (HBE) cell lines. Both of these genes were downregulated in the lung adenocarcinoma cell lines. TOP2A and AURKA had their highest expression levels in H1975 cells with approximately 15 times higher expression than that in HBEs cells. Further experiments in human samples found that TOP2A and AURKA were both upregulated in lung adenocarcinoma tissues compared with non-cancerous tissues. The upregulation of TOP2A was found to be associated with larger tumor size, and AURKA was found to be associated with both larger tumor size and positive lymphatic metastasis. Thus, we demonstrated that TOP2A and AURKA are closely involved in lung adenocarcinoma using both bioinformatics and cell experiments. The cell lines we selected have varying levels of EMT. The H1975 cell line (56) has the highest relative level of EMT and also has higher levels of AURKA and TOP2A in our study. We also showed that higher AURKA expression is correlated with poor prognosis. Thus, the higher expression of AURKA may be correlated with higher levels of EMT, which result in metastasis and lead to poor prognosis.

In conclusion, our research aimed to identify DEGs, which may be involved in the genesis and development of lung adenocarcinoma. Two of the 16 hub genes were further studied in cell lines and human samples and may be regarded as biomarkers for the diagnosis of lung adenocarcinoma. Further research is needed to elucidate the mechanisms behind their changes in expression and their biological function in lung adenocarcinoma.

 Acknowledgements

Not applicable.

 Funding

This research was supported by a grant (no. 81702242) from the National Science Foundation of China and Doctoral Scientific Research Foundation of Liaoning Province (no. 20170520441).

 Availability of data and materials

The data used in this study are available from the corresponding author on reasonable request.

 Authors' contribution

SD and SX designed the research. WM carried out the data collection and analysis. SY conducted the experiments. SD wrote the manuscript. All authors read and approved the manuscript and agree to be accountable for all aspects of the research in ensuring that the accuracy or integrity of any part of the work are appropriately investigated and resolved.

 Ethics approval and consent to participate

The Ethics Committee of the First Hospital of the China Medical University (Shenyang, Liaoning, China) approved our research. Written informed consent was received from all of the included patients.

 Patient consent for publication

Not applicable.

 Competing interests

The authors declare that they have no competing interests.

 References SiegelRLMillerKDJemalACancer statistics, 2018CA Cancer J Clin68730201810.3322/caac.2144229313949 LawrenceMSStojanovPPolakPKryukovGVCibulskisKSivachenkoACarterSLStewartCMermelCHRobertsSAMutational heterogeneity in cancer and the search for new cancer-associated genesNature499214218201310.1038/nature12213237705673919509 FerraraRMezquitaLTexierMLahmarJAudigier-ValetteCTessonnierLMazieresJZalcmanGBrosseauSLe MoulecSHyperprogressive disease in patients with advanced non-small cell lung cancer treated With PD-1/PD-L1 inhibitors or with single-agent chemotherapyJAMA Oncol415431552201810.1001/jamaoncol.2018.3676301932406248085 ParkCKChoHJChoiYDOhIJKimYCA phase II trial of osimertinib in the second-line treatment of non-small cell lung cancer with the EGFR T790M mutation, detected from circulating tumor DNA: LiquidLung-O-Cohort 2Cancer Res Treat51777787201910.4143/crt.2018.38730189719 BarrettTWilhiteSELedouxPEvangelistaCKimIFTomashevskyMMarshallKAPhillippyKHShermanPMHolkoMNCBI GEO: Archive for functional genomics data sets-updateNucleic Acids Res41(Database Issue)D991D995201323193258 RoblesAIAraiEMathéEAOkayamaHSchetterAJBrownDPetersenDBowmanEDNoroRWelshJAAn Integrated prognostic classifier for stage I lung adenocarcinoma based on mRNA, microRNA, and DNA methylation biomarkersJ Thorac Oncol1010371048201510.1097/JTO.0000000000000560261342234493931 SelamatSAChungBSGirardLZhangWZhangYCampanMSiegmundKDKossMNHagenJALamWLGenome-scale analysis of DNA methylation in lung adenocarcinoma and integration with mRNA expressionGenome Res2211971211201210.1101/gr.132662.111226138423396362 GirardLRodriguez-CanalesJBehrensCThompsonDMBotrosIWTangHXieYRekhtmanNTravisWDWistubaIIAn expression signature as an aid to the histologic classification of non-small cell lung cancerClin Cancer Res2248804889201610.1158/1078-0432.CCR-15-2900273544715492382 HuangDWShermanBTTanQCollinsJRAlvordWGRoayaeiJStephensRBaselerMWLaneHCLempickiRAThe DAVID Gene Functional Classification Tool: A novel biological module-centric algorithm to functionally analyze large gene listsGenome Biol8R183200710.1186/gb-2007-8-9-r183177849552375021 TanabeMKanehisaMUsing the KEGG database resourceCurr Protoc Bioinformatics Chapter 1Unit 1.12201210.1002/0471250953.bi0112s38 SzklarczykDMorrisJHCookHKuhnMWyderSSimonovicMSantosADonchevaNTRothABorkPThe STRING database in 2017: Quality-controlled protein-protein association networks, made broadly accessibleNucleic Acids Res45D362D368201710.1093/nar/gkw93727924014 ShannonPMarkielAOzierOBaligaNSWangJTRamageDAminNSchwikowskiBIdekerTCytoscape: A software environment for integrated models of biomolecular interaction networksGenome Res1324982504200310.1101/gr.123930314597658403769 KwonSKimHKimHSIdentification of pharmacologically tractable protein complexes in cancer using the R-based network clustering and visualization program MCODERBioMed Res Int20171016305201710.1155/2017/1016305286910135485287 DemchakBHullTReichMLiefeldTSmootMIdekerTMesirovJPCytoscape: The network visualization tool for GenomeSpace workflows. Version 2F1000Res3151201410.12688/f1000research.4492.2251655374133763 CeramiEGaoJDogrusozUGrossBESumerSOAksoyBAJacobsenAByrneCJHeuerMLLarssonEThe cBio cancer genomics portal: An open platform for exploring multidimensional cancer genomics dataCancer Discov2401404201210.1158/2159-8290.CD-12-0095225888773956037 MaereSHeymansKKuiperMBiNGO: A Cytoscape plugin to assess overrepresentation of gene ontology categories in biological networksBioinformatics2134483449200510.1093/bioinformatics/bti55115972284 CasperJZweigASVillarrealCTynerCSpeirMLRosenbloomKRRaneyBJLeeCMLeeBTKarolchikDThe UCSC Genome Browser database: 2018 updateNucleic Acids Res46D762D769201829106570 RhodesDRKalyana-SundaramSMahavisnoVVaramballyRYuJBriggsBBBarretteTRAnstetMJKincead-BealCKulkarniPOncomine 3.0: Genes, pathways, and networks in a collection of 18,000 cancer gene expression profilesNeoplasia9166180200710.1593/neo.07112173567131813932 DongSQuXLiWZhongXLiPYangSChenXShaoMZhangLThe long non-coding RNA, GAS5, enhances gefitinib-induced cell death in innate EGFR tyrosine kinase inhibitor-resistant lung adenocarcinoma cells with wide-type EGFR via downregulation of the IGF-1R expressionJ Hematol Oncol843201510.1186/s13045-015-0140-6259257414419450 LabbéDPSweeneyCJBrownMGalboPRosarioSWadoskyKMKuSYSjöströmMAlshalalfaMErhoNTOP2A and EZH2 provide early detection of an aggressive prostate cancer subgroupClin Cancer Res2370727083201710.1158/1078-0432.CCR-17-0413288999735690819 RouquierSPillaireMJCazauxCGiorgiDExpression of the microtubule-associated protein MAP9/ASAP and its partners AURKA and PLK1 in colorectal and breast cancersDis Markers2014798170201410.1155/2014/798170248766644022107 BeerDGKardiaSLHuangCCGiordanoTJLevinAMMisekDELinLChenGGharibTGThomasDGGene-expression profiles predict survival of patients with lung adenocarcinomaNat Med8816824200210.1038/nm73312118244 HouJAertsJden HamerBvan IjckenWden BakkerMRiegmanPvan der LeestCvan der SpekPFoekensJAHoogstedenHCGene expression-based classification of non-small cell lung carcinomas and survival predictionPLoS One5e10312201010.1371/journal.pone.0010312204219872858668 LandiMTDrachevaTRotunnoMFigueroaJDLiuHDasguptaAMannFEFukuokaJHamesMBergenAWGene expression signature of cigarette smoking and its role in lung adenocarcinoma development and survivalPLoS One3e1651200810.1371/journal.pone.0001651182971322249927 SuLJChangCWWuYCChenKCLinCJLiangSCLinCHWhang-PengJHsuSLChenCHHuangCYSelection of DDX5 as a novel internal control for Q-RT-PCR from microarray data using a block bootstrap re-sampling schemeBMC Genomics8140200710.1186/1471-2164-8-140175400401894975 YamagataNShyrYYanagisawaKEdgertonMDangTPGonzalezANadafSLarsenPRobertsJRNesbittJCA training-testing approach to the molecular classification of resected non-small cell lung cancerClin Cancer Res946954704200314581339 BhattacharjeeARichardsWGStauntonJLiCMontiSVasaPLaddCBeheshtiJBuenoRGilletteMClassification of human lung carcinomas by mRNA expression profiling reveals distinct adenocarcinoma subclassesProc Natl Acad Sci USA981379013795200110.1073/pnas.19150299811707567 GarberMETroyanskayaOGSchluensKPetersenSThaeslerZPacyna-GengelbachMvan de RijnMRosenGDPerouCMWhyteRIDiversity of gene expression in adenocarcinoma of the lungProc Natl Acad Sci USA981378413789200110.1073/pnas.24150079811707590 BrayFFerlayJSoerjomataramISiegelRLTorreLAJemalAGlobal cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countriesCA Cancer J Clin68394424201810.3322/caac.2149230207593 LiYJuhaszALMaLQCuiXInhalation bioaccessibility of PAHs in PM2.5: Implications for risk assessment and toxicity predictionSci Total Environ6505664201910.1016/j.scitotenv.2018.08.24630195131 PapadopoulosAGuidaFCénéeSCyrDSchmausARadoïLPaget-BaillySCartonMTarnaudCMenvielleGCigarette smoking and lung cancer in women: Results of the French ICARE case-control studyLung Cancer74369377201110.1016/j.lungcan.2011.04.01321620510 SekineTSakaguchiCFukanoYInvestigation by microarray analysis of effects of cigarette design characteristics on gene expression in human lung mucoepidermoid cancer cells NCI-H292 exposed to cigarette smokeExp Toxicol Pathol67143151201510.1016/j.etp.2014.11.00225497788 HuangZSuRQingCPengYLuoQLiJPlasma circular RNAs hsa_circ_0001953 and hsa_circ_0009024 as diagnostic biomarkers for active tuberculosisFront Microbiol92010201810.3389/fmicb.2018.02010302144346126419 TanTZRouanneMTanKTHuangRYThieryJPMolecular subtypes of urothelial bladder cancer: Results from a meta-cohort analysis of 2411 tumorsEur Urol75423432201910.1016/j.eururo.2018.08.02730213523 LiSYWuHCMaiHFZhenJXLiGSChenSJMicroarray-based analysis of whole-genome DNA methylation profiling in early detection of breast cancerJ Cell Biochem120658670201910.1002/jcb.2742330203578 MiaoSQiuTZhaoYWangHSunXWangYXuanYQinYJiaoWOverexpression of S100A13 protein is associated with tumor angiogenesis and poor survival in patients with early-stage non-small cell lung cancerThorac Cancer911361144201810.1111/1759-7714.12797300476266119616 EserPOJännePATGFβ pathway inhibition in the treatment of non-small cell lung cancerPharmacol Ther184112130201810.1016/j.pharmthera.2017.11.00429129643 BasuAKDNA damage, mutagenesis and cancerInt J Mol Sci19E970201810.3390/ijms1906176729570697 ParkCWBakYKimMJSrinivasraoGHwangJSungNKKimBYYuJHHongJTYoonDYThe novel small molecule STK899704 promotes senescence of the human A549 NSCLC cells by inducing DNA damage responses and cell cycle arrestFront Pharmacol9163201810.3389/fphar.2018.00163297132755912185 LiLLeiQZhangSKongLQinBScreening and identification of key biomarkers in hepatocellular carcinoma: Evidence from bioinformatic analysisOncol Rep3826072618201710.3892/or.2017.5946289014575780015 KnezLSodjaEKernIKošnikMCuferTPredictive value of multidrug resistance proteins, topoisomerases II and ERCC1 in small cell lung cancer: A systematic reviewLung Cancer72271279201110.1016/j.lungcan.2011.02.01421440950 HuangHLiuJMengQNiuGMultidrug resistance protein and topoisomerase 2 alpha expression in non-small cell lung cancer are related with brain metastasis postoperativelyInt J Clin Exp Pathol811537115422015266178874637703 SudanSRupasingheHPQuercetin-3-O-glucoside induces human DNA topoisomerase II inhibition, cell cycle arrest and apoptosis in hepatocellular carcinoma cellsAnticancer Res3416911699201424692698 KaurGReinhartRAMonksAEvansDMorrisJPolleyETeicherBABromodomain and hedgehog pathway targets in small cell lung cancerCancer Lett371225239201610.1016/j.canlet.2015.12.00126683772 ShiRSunQSunJWangXXiaWDongGWangAJiangFXuLCell division cycle 20 overexpression predicts poor prognosis for patients with lung adenocarcinomaTumour Biol391010428317692233201710.1177/101042831769223328349831 KlaegerSHeinzlmeirSWilhelmMPolzerHVickBKoenigPAReineckeMRuprechtBPetzoldtSMengCThe target landscape of clinical kinase drugsScience358eaan4368201710.1126/science.aan4368291918786542668 LazarisACKavantzasNGZorzosHSTsavarisNVDavarisPSMarkers of drug resistance in relapsing colon cancerJ Cancer Res Clin Oncol128114118200210.1007/s00432-001-0310-511862483 CostaMJHansenCLHoldenJAGuineeDJrTopoisomerase II alpha: Prognostic predictor and cell cycle marker in surface epithelial neoplasms of the ovary and peritoneumInt J Gynecol Pathol19248257200010.1097/00004347-200007000-0000910907174 TaylorNJBensenJTPooleCTroesterMAGammonMDLuoJMillikanRCOlshanAFGenetic variation in cell cycle regulatory gene AURKA and association with intrinsic breast cancer subtypeMol Carcinog5416681677201510.1002/mc.2223825328151 HoqueACarterJXiaWHungMCSahinAASenSLippmanSMLoss of aurora A/STK15/BTAK overexpression correlates with transition of in situ to invasive ductal carcinoma of the breastCancer Epidemiol Biomarkers Prev1215181522200314693746 ZhouHKuangJZhongLKuoWLGrayJWSahinABrinkleyBRSenSTumour amplified kinase STK15/BTAK induces centrosome amplification, aneuploidy and transformationNat Genet20189193199810.1038/24969771714 StaffSIsolaJJumppanenMTannerMAurora-A gene is frequently amplified in basal-like breast cancerOncol Rep23307312201020043089 LukasiewiczKBLingleWLAurora A, centrosome structure, and the centrosome cycleEnviron Mol Mutagen50602619200910.1002/em.2053319774610 MelicharBAdenisALockhartACBennounaJDeesECKayalehOObermannovaRDeMicheleAZatloukalPZhangBSafety and activity of alisertib, an investigational aurora kinase A inhibitor, in patients with breast cancer, small-cell lung cancer, non-small-cell lung cancer, head and neck squamous-cell carcinoma, and gastro-oesophageal adenocarcinoma: A five-arm phase 2 studyLancet Oncol16395405201510.1016/S1470-2045(15)70051-325728526 ChenJLuHZhouWYinHZhuLLiuCZhangPHuHYangYHanHAURKA upregulation plays a role in fibroblast-reduced gefitinib sensitivity in the NSCLC cell line HCC827Oncol Rep3318601866201510.3892/or.2015.3764256341134440218 DuanLYeLZhuangLZouXLiuSZhangYZhangLJinCHuangYVEGFC/VEGFR3 axis mediates TGFβ1-induced epithelial-to-mesenchymal transition in non-small cell lung cancer cellsPLoS One13e0200452201810.1371/journal.pone.0200452299959506040758

Venn diagram, protein-protein interaction network and the most significant components of DEGs. (A) DEGs were selected with P-value <0.05 and a fold change >2 among the mRNA expression profiling sets GSE75037, GSE63459 and GSE32863. These three datasets manifested an overlap of 376 genes. (B) The protein-protein interaction network of DEGs was constructed using Cytoscape software. (C) The most significant components of the DEGs were obtained from protein-protein interaction network with 16 nodes. DEGs, differentially expressed genes.

Interaction network of the hub genes and the biological process analysis. (A) cBioPortal platform were used to analyze the hub genes and the co-expression genes. The hub genes are marked with a bold outline. Co-expression genes are marked with a thin outline. (B) The plugin of Cytoscape, BiNGO, was adopted to conduct the analysis of biological process. P-value of the ontologies are represented by different color shade. The yellow node indicates higher functional enrichment than white. The numbers of the genes involved in the ontologies are represented by the different size of the node. P<0.05 was considered statistically significant. Interaction network of the hub genes and the biological process analysis. (C) The hub gene alteration rates in lung adenocarcinoma were screened from cBioPortal platform; the red-colored bars represent the upregulation of the gene. (D) The UCSC (University of California Santa Cruz) cancer platform was used to construct the hub gene hierarchical clustering. The primary tumor, recurrent tumor and normal tissue are represented in different colors.

The relationship between alteration of the hub genes and overall survival (A) and disease free survival (B) were explored in the cBioPortal platform. P<0.05 was considered statistically significant. To note, the full names of all the gene symbols used in the figures are listed in Table III.

Expression profiles of (A) TOP2A and (B) AURKA in 20 malignant tumor types are represented using the Oncomine database. The number represent the cases meeting the threshold for TOP2A and AURKA. Heat maps of (C) TOP2A and (D) AURKA gene expression in lung carcinoma samples vs. adjacent lung tissues in the Oncomine database. (C) 1. Lung carcinoma vs. normal lung, Beer et al (2002) (22). 2. Lung carcinoma vs. normal lung, Hou et al (2010) (23). 3. Lung carcinoma vs. normal lung, Landi et al (2008) (24). 4. Lung carcinoma vs. normal lung, Selamat et al (2012) (7). 5. Lung carcinoma vs. normal lung, Su et al (2007) (25). 6. Lung carcinoma vs. normal lung, Yamagata et al (2003) (26). (D) 1. Lung carcinoma vs. normal lung, Bhattacharjee et al (2001) (27). 2. Lung carcinoma vs. normal lung, Garber et al (2001) (28). 3. Lung carcinoma vs. normal lung, Hou et al (2010) (23). 4. Lung carcinoma vs. normal lung, Landi et al (2008) (24). 5. Lung carcinoma vs. normal lung, Selamat et al (2012) (7). 6. Lung carcinoma vs. normal lung, Su et al (2007) (25). The P-value for a gene is its P-value for the median-ranked analysis. The fold change represents the relative expression of the tumor tissue compared with the normal tissue. AURKA, aurora kinase A; TOP2A, DNA topoisomerase II α.

RT-qPCR to confirm the upregulation of TOP2A and AURKA in lung adenocarcinoma and HBE cell lines and human samples. (A) TOP2A expression in lung adenocarcinoma cell lines A549, HCC827 and H1975 and a human bronchial epithelial (HBE) cell line. (B) AURKA expression in the lung adenocarcinoma cell lines. (C) TOP2A expression in lung adenocarcinoma tissue and non-cancerous lung tissue samples. (D) AURKA expression in lung adenocarcinoma tissue and non-cancerous lung tissue samples. AURKA, aurora kinase A; TOP2A, DNA topoisomerase II α. *P<0.05.

KEGG and GO pathway enrichment analysis of DEGs in the lung adenocarcinoma samples.

Term Description Count in gene set P-value
GO:0048545 Response to steroid hormone stimulus 27 9.53E-14
GO:0009725 Response to hormone stimulus 33 2.60E-11
GO:0009719 Response to endogenous stimulus 33 3.31E-10
GO:0042127 Regulation of cell proliferation 45 9.09E-09
GO:0001501 Skeletal system development 26 3.68E-08
GO:0010033 Response to organic substance 41 5.71E-08
GO:0007179 TGF-β receptor signaling pathway 12 6.25E-08
GO:0043627 Response to estrogen stimulus 15 6.80E-08
GO:0007155 Cell adhesion 40 7.66E-08
GO:0022610 Biological adhesion 40 7.78E-08
hsa04512 ECM-receptor interaction 11 5.33E-05
hsa04670 Leukocyte transendothelial migration 10 0.003333312
hsa04514 Cell adhesion molecules (CAMs) 10 0.006973812
hsa04510 Focal adhesion 12 0.014731626
hsa04350 TGF- β signaling pathway 7 0.024809421
hsa04610 Complement and coagulation cascades 6 0.032656397
hsa03320 PPAR signaling pathway 6 0.032656397

KEGG, Kyoto Encyclopedia of Genes and Genomes; GO, Gene Ontology; DEGs, differentially expressed genes.

KEGG and GO pathway enrichment analysis of DEGs in the most significant module.

Pathway ID Pathway description Count in gene set FDR
GO:0000279 M phase 9 3.72E-07
GO:0000278 Mitotic cell cycle 9 9.46E-07
GO:0000280 Nuclear division 8 1.07E-06
GO:0007067 Mitosis 8 1.07E-06
GO:0000087 M phase of mitotic cell cycle 8 1.21E-06
GO:0048285 Organelle fission 8 1.41E-06
GO:0022403 Cell cycle phase 9 2.30E-06
GO:0051301 Cell division 8 8.24E-06
GO:0007049 Cell cycle 10 1.24E-05
GO:0022402 Cell cycle process 9 2.65E-05
GO:0007059 Chromosome segregation 4 0.08986
GO:0030261 Chromosome condensation 3 0.367997
GO:0000226 Microtubule cytoskeleton organization 4 0.523613
GO:0007051 Spindle organization 3 1.195525
GO:0007017 Microtubule-based process 4 2.500246
hsa04114 Oocyte meiosis 4 0.096833
hsa04110 Cell cycle 4 0.141584

KEGG, Kyoto Encyclopedia of Genes and Genomes; GO, Gene Ontology; DEGs, differentially expressed genes; FDR, false discovery rate.

Summary of the hub gene functions.

No. Gene symbol Full name Function Degree
  1 TOP2A DNA topoisomerase II α TOP2A functions as the target for various anticancer agents and mutations in it are associated with drug resistance 31
  2 AURKA Aurora kinase A AUEKA plays a role in tumor development and progression 22
  3 UBE2C Ubiquitin conjugating enzyme E2 C UBE2C is required for the destruction of mitotic cyclins and for cell cycle progression, and is involved in cancer progression 20
  4 KIAA0101 (PCLAF) PCNA clamp associated factor PCNA-binding protein acts as a regulator of DNA repair during DNA replication 20
  5 CDC20 Cell division cycle 20 CDC20 is a regulatory protein interacting with several other proteins at multiple points in the cell cycle 19
  6 CCNB2 Cyclin B2 CCNB2 is one of the essential components of the cell cycle regulatory machinery 18
  7 TK1 Thymidine kinase 1 High level of TK1 is used as a biomarker for diagnosing and categorizing many types of cancers 17
  8 PTTG1 Pituitary tumor-transforming 1 PTTG1 product has transforming activity in vitro and tumori genic activity in vivo, and it is highly expressed in various tumors 17
  9 MELK Maternal embryonic leucine zipper kinase Diseases associated with MELK include uterine corpus endometrial carcinoma. Among its related pathways are Neuroscience 17
10 NUSAP1 Nucleolar and spindle associ ated protein 1 NUSAP1 is a nucleolar-spindle-associated protein that plays a role in spindle microtubule organization 16
11 CDC45 Cell division cycle 45 The protein encoded by CDC45 is an essential protein required for the initiation of DNA replication 16
12 ASPM Abnormal spindle microtubule assembly ASPM is essential for normal mitotic spindle function in embryonic neuroblasts 16
13 UBE2T Ubiquitin conjugating enzyme E2 T The protein encoded by UBE2T catalyzes the covalent attachment of ubiquitin to protein substrates. Defects in UBE2T are associated with Fanconi anemia of complementation group T 15
14 CDCA5 Cell division cycle associated 5 Among its related pathways are Cell cycle, Mitotic and MicroRNAs in cancer 15
15 PRC1 Protein regulator of cytokinesis 1 PRV1 encodes a protein that is involved in cytokinesis which has been shown to be a substrate of several cyclin-dependent kinases 14
16 CDCA7 Cell division cycle associated 7 CDCA7 was identified as a c-Myc responsive gene. Overexpression of this gene enhances the transformation of lymphoblastoid cells 13

Clinicopathologic associations of TOP2A expression in lung adenocarcinoma (N=72).

Relative TOP2A expression
Clinical parameters No. of cases Low High P-value
Age (years) 0.237
  >60 39 22 17
  ≤60 33 14 19
Sex 0.099
  Male 37 15 22
  Female 35 21 14
Smoking 1
  Smoker 42 21 21
  Non-smoker 30 15 15
Maximum diameter (cm) 0.003
  <3 26 19 7
  ≥3 46 17 29
Lymphatic metastasis 0.812
  Positive 41 21 20
  Negative 31 15 16
Metastasis 0.607
  M0 68 33 35
  M1 4 3 1

TOP2A, DNA topoisomerase II α.

Clinicopathologic associations of AURKA expression in lung adenocarcinoma (N=72).

Relative AURKA expression
Clinical parameters No. of cases Low High P-value
Age (years) 0.813
  >60 39 20 19
  ≤60 33 16 17
Sex 0.099
  Male 37 22 15
  Female 35 14 21
Smoking 0.633
  Smoker 42 20 22
  Non-smoker 20 16 14
Maximum diameter (cm) 0.003
  <3 26 19 7
  ≥3 46 17 29
Lymphatic metastasis 0.002
  Positive 41 14 27
  Negative 31 22 9
Metastasis 1
  M0 68 34 34
  M1 4 2 2

AURKA, aurora kinase A.