An in-silico approach leads to explore six genes as a molecular signatures of lung adenocarcinoma

Abstract

Due to heterogenetic-specific nature of the available biomarkers, the incidence of lung adenocarcinoma (LUAD) is on the rise worldwide. Previously reported LUAD-related hub genes were searched from the medical literature via literature mining and were processed to identify few top genes via degree method. Later, a comprehensive in silico methodology was applied on the selected real hub genes to identify their tumor driving, diagnostic, and prognostic roles in LUAD patients with divers clinicopathological variables. Out of total 145 extracted hub genes, six genes including CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1 were identified as real hub genes. The expression analysis showed that all these genes were significantly up-regulated across LUAD samples of different clinicopathological variables. In addition, a variety of unique correlations among the expression and of real hub genes and some other parameters including promoter methylation status, overall survival (OS), genetic changes, tumor purity, and immune cell infiltration have also been explored in the present study. Moreover, via TFS-miRNA-mRNA regulatory network, one important TF (E2F1) and one important miRNAs (hsa-mir-34a-5p) that targeted all the real hub genes were also identified. Finally, a variety of drugs also predicted to be very useful in treating LUAD. The discovery of the real hub genes, TFS-miRNA-mRNA network, and chemotherapeutic drugs associated with LUAD provides new insights into underlying mechanisms and treatment of LUAD overcoming heterogeneity barriers.

Introduction

Lung cancer is the second most abundantly diagnosed cancer worldwide [1]. It is estimated that more than 2.2 million new lung cancer cases and 1.7 million lung cancer-associated deaths occur per annum around the world [2]. In Pakistan, lung cancer is the third most prevalent cancer accounting for 5.9% new cases of all the cancers [2]. Although LUAD initial growth occurs in the early stage, the progression of this disease is much slower than other cancer subtypes. Moreover, the overall survival of LUAD is also reported to be lower than other cancer subtypes [2]. Recent studies highlighted that an exposure to asbestos, smoking cigarette and exposure to many other types of chemical can enhance the risk of LUAD development [3]. LUAD is a complicated and one of the most deadliest malignancies of the lung periphery where glandular cells secrete mucus and help in breathing [3]. This subtype accounts for more than 40% of lung cancer cases its rate is on the rise currently [5].

A significant improvement has been achieved in LUAD therapy recently, its prognosis is still very low with 18% of 5-year survival rates [6]. We all know that the accurate diagnosis of LUAD is a huge challenge, as it is often detected at the advanced stages [7]. Therefore, an accurate diagnosis of LUAD requires the discovery of new biomarkers that can detect LUAD precisely, increase survival rates, reduce the incidence of tumor invasions, and also enhance the chances of successful therapy [8].

During the last 10 years, researchers around the globe have used next-generation sequencing (NGS) as well as microarray methods to identify novel biomarkers and therapeutic targets in LUAD [9], small sample count however resulted in the significant inter-study inconsistency. In order to address this issue, Gene Expression Omnibus (GEO) database [10] provided the facility to researchers for archiving their expression datasets in GEO database to make it publically available for further integration with other similar datasets via in silico approaches to uncover molecular biomarkers more precisely. Previously, such GEO-based expression datasets of LUAD have been utilized by the earlier studies to discover biomarkers of LUAD [9], but as we know that biomarkers are highly specific biomolecules, and GEO-based LUAD expression datasets consist of cancer patients with different clinical variables, it is clinically impossible to use already identified GEO-based biomarkers in LUAD patients over heterogeneity-barrier.

In the current research, therefore, using a novel integrated approach, we utilized already identified biomarkers (hub genes) from GEO-based expression datasets of LUAD to prioritize a new system of six biomarkers over heterogeneity-barrier.

Materials and methods

Hub genes extraction

PubMed database was effectively searched in our study to find all studies dealing with GEO-expression datasets of LUAD for the identification of hub genes up to December 2021. Following keywords were used in a combination to search the relevant literature: “Hub genes AND Lung adenocarcinoma” or “Hub genes AND Lung neoplasia”. In total, 23 studies out of total 356 appeared studies were selected collectively analyzing 31 GEO expression datasets of LUAD to identify numerous hub genes. Later, all the selected studies were further subjected to extract and combined the reported hub genes for getting a consolidated pool.

Protein-protein interaction (PPI) network, module, and enrichment analysis

For interpreting molecular mechanisms behind LUAD, we used STRING, was used STRING database [11] to construct the information of PPIs network with of not < 0.7. Then, the relationships between hub genes were explored via Cytoscape software (3.8.2) by calculating network properties including the distribution of network node degree, distribution of the shortest path, and proximity to the center [12]. Later, molecular Complex Detection (MCODE) analysis has helped us to find clusters of genes in the constructed PPI network with default cutoff criterions parameters such as “Degree cutoff = 2”, “node score cutoff = 0.2”, “k-core = 2” and “max. depth = 100”. Finally, the real hub genes were chosen via degree method and their enrichment analysis were performed using DAVID tool. A P-value of < 0.05 was selected to show the statistical differences.

Real hub genes expression profiling via UALCAN

The expression analysis of real hub genes were carried out via UALCAN (http://ualcan.path.uab.edu/) [13]. The UALCAN is The Cancer Genome Atlas (TCGA) data set analysis tool. In this study, for expression analysis of real hub genes we used TCGA LUAD dataset consisting of 515 cancerous and 59 normal samples. For statistics, UALCAN used a student t-test and normalized the obtained expression as transcript per million (TPM) reads.

Expression validation of real hub genes

The TIMER [14], GENT2 [15], GEPIA, DriverDBV3 [16], and UALCAN [13] are TCG multi-omics data analysis tools. In this study, we utilized these databases for the expression validation of the real hub genes using new independent LUAD patients’ cohorts.

DriverDBV3 analysis

DriverDBV3 [16] was conducted in this to analyze the promoter methylation levels of real hub genes across LAUD samples paired with normal controls using Pearson correlation analysis.

Prognostic potentials of real hub genes

Kaplan-Meier Plotter and GEPIA tools are widely used for prognostic potentials evaluation of any gene(s) of interest [17]. In the current study, via using these tool, we analyzed the effect of real hub genes expression on the Overall Survival (OS) of LUAD patients. A P-value < 0.05 was chosen as statistically significant.

Genetic alterations in real hub genes

Via cBioPortal, we used TCGA LUAD datasets for analyzing genetic alterations of real hub genes [18]. This online resource is a hub of cancer omics data which includes genetic mutations information, copy number variations, deep amplification, deep deletion, and mRNA expression level information. We used this database for conducting genetic alteration analysis in this study with default settings.

Tumor purity and immune cells analyses

TIMER [19] was conducted in this study to analyze the relationships between tumor purity, immune cells infiltration, and real hub genes expression across LUAD samples. A P-value < 0.05 was chosen as statistically significant.

Constructing TF-miRNA-mRNA network

ENCORI is an openly available public platform developed for the identification of more than 2.5 million TFS-miRNA-mRNA interactions [20,21]. All the target TFS and miRNAs of the real hub genes were screened in ENCORI using default settings. Finally, the TFS-miRNA-mRNA networks of real hub genes were visualized using Cytoscape (version 3.8.2).

Expression profiling of TFS and miRNAs

The expression profiling of real hub genes targeting TFS and miRNAs were also carried out via UALCAN [13] in a LUAD cohort taken from TCGA datasets using default setting.

CancerSEA-based analysis

CancerSEA is an online resource foe exploring relationships among gene(s) of interest and 14 different functional states at a single-cell level across several cancer types [22]. CancerSEA was conducted in this study to investigate relationships between real hub genes and these states at a single-cell level across LUAD. A P-value < 0.05 was chosen as statistically significant.

MuTarget-based analysis

MuTarget is an openly available platform that help the researcher to associate genetic mutations with gene expression across several human cancer types [23]. We conducted MuTarget analysis in this study to identify the mutant genes associated with gene expression alteration of real hub genes in LUAD. A P-value < 0.05 was chosen as statistically significant.

Comparative Toxicogenomics database

Comparative Toxicogenomics database (CTD) [24] was conducted to draw the real hub gene-drug interaction networks via Cytoscape, highlighting different potential drugs capable of decreasing or increasing the expression levels of real hub genes. In clinical application view point, the selected drugs will in the treatment of LUAD.

Results

Extraction of hub genes

In total, 23 studies were shortlisted for hub genes extraction. The selected studies were next subjected to hub genes extraction and finally, after normalizing overlapped genes, we were able to construct a get a brief pool of 145 hub genes from 31 GEO LUAD datasets comprising of 1980 LUAD and 1230 normal samples (Table 1). Without normalization, the original data can be seen in the Table S1.

Table 1.

List of the LUAD microarray expression datasets and the hub genes extracted selected studies

Dataset	C/N	Source of origin	Extracted hub genes	References
GSE118370	06/06	China	ADCY4, S1PR1, FPR2, PPBP, NMU, PF4, GCG, CCNA2, CCNB1, CDC20, CDCA5, CDCA8, FEN1, KIF2C, KPNA2, MCM6, NUSAP1, RACGAP1, RRM2, SPAG5, TOP2A, TPX2, CA4, PECAM1, DNAJB4, AGER, GIMAP6, C10orf54, DOCK4, LRRK2, EPAS1, LDB2, HOXA11-AS, PHACTR2, MSRB3, GHR, PLSCR4, EPB41L2, NPNT, FBXO32, IL6, MMP9, EDN1, FOS, CDK1, CDH1, BIRC5, VWF, UBE2C, CDKN3, CDKN2A, CD34, AURKA, CCNB2, EGR1, UBE2T, PBK, MELK, TNNC1, TPPX2, INS, LPL, HPGDS, DGAT1, UGT1A6, CYP2C9, KIAA0101, BUB1B, PRC1, CEACAM5, NQO1, LCN2, KRT8, EPCAM, ELF3, KRT19, DCN, SERPING1, GNG11, CXCL12, CAV1, DCY8, ADRB2, CALCA, GNGT1, NPSR1, CLDN5, COL1A1, SPP1, UBB, RAC1, ITGB1, SRC, C3, EGFR, TIMP1, GAS6, P4HB, CXCR4, FPR1, LYZ, OCIAD2, ETV4, COL10A, PROM2, MMP11, ABCC3, BAIAP2L1, FABP4, STX11, FHL1, TEK, FMO2, CRYAB, GRK5, TMEM100, DLGAP5, KIF11, RAD51AP1, CDC6, OIP5, NCAPG, CENPF, KIF4A, CDC25A, CDH5, BDNF, SELE, KIF23, PLK1, CBFA2T3, CR2, SEL1L3, TM6SF1, TSPAN32, ITGA6, MAPK11, RASA3, TLR6, AURKB, HMGA2, ASPM, CKS1B, CHRDL1, SPARCL1	[155-177]
GSE68465	442/0	USA
GSE68571	96/0	USA
GSE69405	52/150	South Korea
GSE40791	94/100	USA
GSE18842	72/86	Spain
GSE74706	18/18	Germany
GSE10072	135/135	USA
GSE29013	55/0	USA
GSE13213	117/0	Japan
GSE7670	66/59	Taiwan
GSE31547	30/20	USA
GSE136043	5/5	China
GSE140797	7/7	China
GSE63459	32/33	USA
GSE85841	8/8	China
GSE116959	57/11	France
GSE75037	83/83	USA
GSE32863	60/60	USA
GSE19188	78/78	Netherlands
GSE33532	80/20	Germany
GSE85716	06/06	China
GSE119004	23/0	Canada
GSE43458	84/26	USA
GSE6044	42/05	Germany
GSE62949	56/0	USA
GSE115002	48/56	China
GSE40419	88/76	USA
GSE66727	23/0	USA
GSE86337	10/0	Australia
GSE57148	7/182	South Korea
Total = 31	Total = 1980/1230		Total = 145

A PPI network, module, and enrichment analysis

STRING was conducted to get a PPI network of the pooled hub genes. The obtained PPI network was consist of 145 nodes and 2341 edges (Figure 1A). Then, by applying Cytoscape, the MCODE and Cytohubba applications has helped to identified one most significant module and 6 hub genes as the real hub genes (CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1) based on the degree scores (Figure 1B, 1C and Table 2). Later, the enrichment analysis has revealed that real hub are involved in different GO and KEGG terms including Mitotic nuclear division etc. BP terms, Microtubule cytoskeleton etc. CC terms, ATP binding etc. MF terms (Figure 2 and Table 3), and Cell cycle etc. KEGG terms (Figure 2D and Table 4).

Figure 1.

(A) A PPI network showing LUAD-related extracted hub genes from the selected studies. (B) A PPI network showing one significant module, (C) A PPI network of the hub genes identified in the significant module, and (D) Six real hub genes identified via degree method.

Table 2.

List of the real hub genes identified from a PPI network of the extracted 124 LUAD related hub genes

Sr. No	Name of the gene	MCODE Node Status	MCODE Score
1	CDC6	Clustered	38.90488
2	PBK	Clustered	38.90488
3	AURKA	Clustered	38.90488
4	KIF2C	Clustered	38.90488
5	OIP5	Clustered	38.90488
6	PRC1	Clustered	38.90488

Figure 2.

GO and KEGG enrichment analysis of real hub genes. (A) BF, (B) CC, (C) MF, and (D) KEGG analysis. A P-value <0.05 was consider as significant.

Table 3.

Details of the GO analysis of identified hub genes extracted from various GEO LUAD expression microarray datasets

GO Term	GO ID	Gene count	P-value	Gene name
BP
GO:0007067	Mitotic nuclear division	6	2.29554618712754E-7	PBK, OIP5, KIF2C, CDC6, AURKA
GO:0042787	Cell division	4	8.703225761335279E-5	OIP5, KIF2C, CDC6, AURKA
CC
GO:0015630	Microtubule cytoskeleton	4	5.527753221351057E-4	PRC1, KIF2C, AURKA
GO:0005634	Nucleus	5	0.0023131822778742635	PRC1, PBK, OIP5, KIF2C, CDC6, AURKA
GO:0051233	Spindle midzone	2	0.005202617432597425	CDC6, AURKA
GO:0005876	Spindle microtubule	2	0.012015152655811594	PRC1, AURKA
GO:0030496	Midbody	2	0.03489913615678372	CDK1, AURKA
MF
GO:0005524	ATP binding	4	0.006046362041106639	PBK, KIF2C, CDC6, AURKA
GO:0005515	Protein binding	2	0.03814879524822187	PRC1, PBK, OIP5, KIF2C, CDC6, AURKA

Table 4.

Details of the KEGG pathway analysis of identified hub genes extracted from various GEO LUAD expression microarray datasets

Pathway ID	Pathway Name	Gene count	P-value	Gene name
hsa04110	Cell cycle	2	1.616848175824832E-5	CDC6, AURKA
hsa04114	Oocyte meiosis	2	0.0018887038532489457	CDC6, AURKA

Expression analysis and cross validation of real hub genes expression

We initially measured real hub genes mRNA expression in 515 LUAD samples paired with 59 normal controls via UALCAN. Results of the analysis have shown the significant (P < 0.05) up-regulation of all the real hub genes (CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1) in LUAD samples of various clinical variables (cancer stage, race, gender, age, and nodal metastasis status) relative to normal controls (Figures 3 and 4). Later, TIMER, GENT2, GEPIA, and DriverDBV2 databases containing 515, 765, 483, and 710 LUAD samples paired with 59, 75, 59, and 111 normal controls, respectively were used to validate the mRNA expression of real hub genes on new independent cohorts. Our validation results showing significant (P < 0.05) up-regulation of all the real hub genes using these addition databases were also in agreement with the expression analysis results of ULACAN (Figure 5A-C). At last, we also validated real hub genes translational expression in 111 LUAD tissues paired with 111 controls via UALCAN. In view of our results, all real hub genes (CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1) were also found significantly (P < 0.05) up-regulated at protein level in LUAD patients relative to controls (Figure 5).

Figure 3.

mRNA expression analysis results of real hub genes in LUAD and normal controls via UALCAN. (A) CDC6, (B) PBK, (C) AURKA, (D), KIF2C, (E) OIP5, and (F) PRC1.

Figure 4.

mRNA and protein expression level validations of the real hub genes in LUAD patients paired with controls via different online expression databases. (A) mRNA expression level validation of real hub genes via TIMER, (B) mRNA expression level validation of real hub genes via GENT2, (C) mRNA expression level validation of real hub genes via GEPIA, (D) mRNA expression level validation of real hub genes via DriverDBV3, and (E) Protein expression level validation of real hub genes via UALCAN.

Figure 5.

Relative mRNA expression analysis results of real hub genes in LUAD patients stratified by different clinicopathological features. (A) Cancer stage (B) Patient’s race, (C) Patient’s gender, (D) Patient’s age, and (E) Nodal metastasis status.

Promoter methylation analysis

In this work, the levels of promoter methylation were assessed via DriverDBV3. In view of our results, all real hub genes (CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1) were found significantly (P < 0.05) hypomethylated relative to controls in LUAD patients (Figure 6).

Figure 6.

Correlation of promoter methylation with mRNA expression of real hub genes in LUAD paired with controls. (A) CDC6, (B) PBK, (C) AURKA, (D), KIF2C, (E) OIP5, and (F) PRC1.

Real hub genes expression and prognosis in LUAD

We next evaluated the prognostic values (OS duration) of real hub genes via Kaplan-Meier Plotter and GEPIA tool. In view of our results, the higher levels of CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1 were found significantly (P < 0.05) linked to the poor OS of the LUAD patients, hence, we speculate these six real hub genes as the good prognostic biomarkers for predicting OS of LUAD patients (Figure 7A, 7B).

Figure 7.

Association between OS and real hub genes expression in LUAD patients. (A) via KM plotter and (B) via GEPIA tool.

Genetic alterations information of real hub genes

Then, for enquiring genetic alterations in real hub genes, we conducted cBioPortal analysis. As shown in Figure 8A, PBK gene showed highest frequency (6%) of genomic alterations, while AURKA and KIF2C showed the second highest genomic alteration rate of 3%. Meanwhile other genes, including OIP5, PRC1, and CDC6 have shown a genetic alteration frequencies of 2%, 2%, and, 1.6%, respectively, in the analyzed LUAD samples. Furthermore, in PBK, and PRC1, the most frequently observed genetic alterations were deep deletion and missense mutations (Figure 8A), while other remaining real hub genes were enriched in deep amplification alterations only. In addition, we have also observed that most frequent mutation in CDC6 gene (M263I) was lied in AAA_22 domain (Figure 8B), and in PBK and KIF2C, the most frequent mutations including E180Q, and R510L/S were present in their respective Pkinase, and Kinesin domains (Figure 8B), while no mutation was detected in AURKA. However, on the other hand, most frequently seen mutations (S107F and T370I) in OIP5 and PRC1 were found in Yipee-MIs18 and MAP65-A3E1 functional domains, respectively (Figure 8B).

Figure 8.

Frequency and distribution of genomic alterations associated with real hub genes in LUAD. (A) Frequency of genomic alteration, (B) distribution of mutations in protein domains of real hub genes, and (C) types of CNVs in LUAD samples.

Tumor purity and immune cells infiltration analysis of real hub genes

The correlations among CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1 mRNA expressions and tumor purity, CD8+ T, and CD4+ T immune cells infiltration across LUAD have been analyzed via TIMER database. Our results have revealed the significant positive correlations between the expression of the real hub genes and tumor purity (Figure 9). While notable negative correlations between real hub genes’ expression and CD4+ T and CD8+ T immune cells infiltration levels.

Figure 9.

Correlation analysis of real hub genes expression with tumor purity, CD4+ T, and CD8+ T cells in LUAD. (A) CDC6, (B) PBK, (C) AURKA, (D) KIF2C, (E) OIP5, and (F) PRC1.

TFS-miRNA-mRNA network

As shown in Figure 10, the TFS-miRNA-mRNA interaction network of real hub genes constructed via ENCORI and Cytoscape is consists of 10 TFS, 28 miRNAs, and 6 mRNAs. Via degree method, one TF (E2F1), and one miRNA (hsa-mir-34a-5p) were identified as top TF and miRNA targeting all 6 real hub genes. Previously, has-mir-34a-5p in axis with LEF1 gene found to be involved in the pathogenesis of CESC [25], while, the role of has-mir-34a-5p miRNA is unclear in LUAD. Moreover, a previous study also revealed that E2F1 promotes EMT by regulating ZEB2 as a transcription factor in lung cancer [26]. The TFS-miRNA-mRNA co-regulatory network here in the current research has highlighted that E2F1 and hsa-mir-34a-5p can also act as the potential inducer of LUAD by dysregulating the identified real hub genes as an E2F1-hsa-mir-124-3p/CDC6/PBK/AURKA/KIF2C/OIP5/PRC1 axis. To further confirm the participation of identified TF and miRNA in LUAD development via dysregulating the real hub genes, we further checked the expression of E2F1 and has-mir-34a-5p in LUAD patients via UALCAN. As a result, a significant up-regulation of E2F1 and has-mir-34a-5p was also observed in LUAD samples than normal samples (Figure 10).

Figure 10.

TF-miRNA-mRNA network analysis of real hub genes in LUAD. (A) miRNAs targeting real hub genes, (B) has-mir-34a-5p miRNA targeting real hub genes, (C) relative expression of has-mir-34a-5p in LUAD and normal controls, (D) TFS targeting real hub genes, (E) E2F1 targeting real hub genes, and (F) relative expression analysis of E2F1 in LUAD samples paired with controls. The pink and grey nodes represent the miRNA, red nodes represent the hub gene, while green node represent the TFS.

Single-cell functional analysis

CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1 genes further involvement in LUAD at single cell level was explored via CancerSEA database. As a result, the CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1 were revealed to be linked (positively or negatively) with fourteen different states at single cell level in LUAD (Figure 11A). However, real hub genes expression was notably negatively correlated with “DNA repair state, while positively correlated with Cell cycle, DNA Damage, Proliferation, Stemness Invasion, EMT, and Apoptosis states” (Figure 11B).

Figure 11.

Real hub genes and different divers states association in LUAD. (A) Correlation analysis of real hub genes expression with 14 different states in LUAD, and (B) Correlation analysis of real hub genes expression only significant states in LUAD.

Real hub genes and their correlated different other mutant genes

MuTarget database with default settings “(FC > 1.4 and P < 0.05)” has helped us in this study to select us top 3 mutant genes for each real hub gene. As shown in Figure 12, top 3 mutant genes which positively correlated with the expression of each real hub gene are TP53, TTN, and OR4C15 with CDC6, TP53, KEAP1, and COL6A6 with PBK, TP53, CSMD3, and SORCS1 with AURKA, TP53, CSMD3, and XIR2P with KIF2C, TP53, SMARCA4, and SCN1A with OIP5, and TP53, TTN, and HERC2 with PRC1.

Figure 12.

CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1 genes positively correlated mutant genes in LUAD from MuTarget. (A) with CDC6, (B) with PBK, (C) with AURKA, (D) with KIF2C, (E) with OIP5, and (F) with PRC1.

Drug-gene interaction analysis

Through CTD database, several drugs associated with 6 real hub genes were selected, and the relationships between them were visualized via Cytoscape (Figure 13). Finally, the drawn drug-gene interaction network reveled that all real hub genes including CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1 can potentially be regulated by several chemotherapeutic drugs. For instance, aflatoxin B1, and estradiol can elevate CDC6 expression while bisphenol A and cannabidiol can reduce CDC20 expression level (Figure 13).

Figure 13.

Screening of real hub genes-associated chemotherapeutic drugs. (A-F) indicates chemotherapeutic drugs that can decrease or increase the expression of the real hub genes. (A) CDC6-associated chemotherapeutic drugs, (B) PBK-associated chemotherapeutic drugs, (C) AURKA-associated chemotherapeutic drugs, (D) KIF2C-associated chemotherapeutic drugs, (E) OIP5-associated chemotherapeutic drugs, and (F) PRC1-associated chemotherapeutic drugs. Red arrows: drugs that increase the real hub genes expression, Green arrows: drug that decrease the real hub genes expression while the numbers of arrows represent the supported numbers of studies in literature.

Discussion

Previously, many studies have been carried out so far to explore molecular biomarkers and mechanisms behind LUAD for its accurate detection and treatment, the incidence of LUAD and mortality rate of LUAD patients is still increasing worldwide due to the heterogenetic-specific nature of the available biomarkers. In this study, we explored 6 real hub genes including CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1 that can overcome the heterogenetic-specific barrier and can apply as biomarkers in LUAD patients of different clinical variables. Moreover, GO and KEGG enrichment analysis revealed that identified real hub genes were significantly involved in diverse GO and KEGG terms (Figure 2).

CDC6 gene is present on chromosome 17 and encodes the CDC6 protein, which is a member of AAA+ ATPases (ATPases associated with cellular activities) family and play an important role in the initiation of DNA replication [27]. CDC6 usually found in the nucleus of cell, but is translocated to the cytoplasm during S-phase after its phosphorylation by CDKs [28]. When cell division begins, CDC6 mediates the assembly of the pre-replicative complex and maintains it for the loading of minichromosomal maintenance (MCM) proteins on DNA [29]. After the formation of MCM-chromatin complex, CDC6 is no longer needed and undergoes proteasome degradation [30]. Thus, the abnormal expression of CDC6 affects the replication mechanism [31]. Previously, an aberrant expression of CDC6 is significantly associated with breast cancer [32], colorectal cancer progression [33], invasiveness of cervical cancer [34], prostate cancer metastasis [35,36], pulmonary cancer [37], inhibition of apoptotic caspases 3 and 9 in pancreatic cancer [38], hepatocellular carcinoma [39,40], proliferation of esophageal squamous cell carcinoma [41], renal cell carcinoma [42], gastric cancer [43], bladder cancer [44], and osteosarcoma [45].

PBK gene is found on chromosome 8 and encodes a protein PBK of 322 amino acids. The encoded protein is a member of mitogen-activated serine/threonine-protein kinase (MAPKK) family [46,47] and is hard to detect in normal somatic tissues but is found frequently in cancerous tissues promoting cell survival, proliferation, and metastasis [48,49]. Abnormal expression of PBK is found to be associated with different cancers including thyroid carcinoma [50], prostate cancer progression [51,52], ovarian cancer [53,54], cervical cancer by ERK/c-Myc signaling [55], pathogenesis of breast cancer by EMT up-regulation [56] and invasion by TGFβ1-induced NFkB-dependent Snail/Slug [57], glioblastoma [58,59], oxaliplatin resistance in liver cancer [60,61], pancreatic cancer [62], NSCLC proliferation by paclitaxel resistance and inhibition of autophagic cell death [63,64], T-cell leukemia or lymphoma [65], colorectal cancer [66,67], oral squamous cell carcinoma [68], nasopharyngeal carcinoma [69], human endometrial cancer [70], urinary bladder transitional cell cancer [62], multiple myeloma [71], esophageal squamous cell carcinoma [72,73], and gastric cancer [74,75].

AURKA aids in controlling the cell cycle [76]. Various types of cancers have been linked with AURKA overexpression, such as prostate cancer [77], breast cancer [78] colon tumorigenesis [79], gastric cancer [80], head and neck cancer [81], liver metastasis [82], hepatocarcinogenesis [83], bladder cancer [84], NSCLC progression [85], ovarian cells [86], and esophageal cancer [87]. After that, many subsequent studies have been carried out to evaluate the AURKA expression level in LUAD, still, its role is poorly understood.

KIF2C gene encodes for a protein of 792 amino acids which belongs to a kinesin superfamily proteins (KIFs). This protein mainly found in the cell bodies and dendrites of the adult neurons in both the peripheral and central nervous systems [88]. KIF2C gene is consider very crucial for mitotic spindle dynamics, chromosomal segregation during anaphase, and separation of sister chromatids [89]. Dysregulation of KIF2C protein is reported to results in chromosome instability [90] and is also found to be significantly associated with bladder cancer invasion [91], Breast cancer [92,93], mediation of Wnt/β-catenin/mTORC1 signaling [94,95] and MEK/ERK signaling in hepatocellular carcinoma [96], thyroid cancer metastasis via TGF-β1/Smad pathway [97], non-small cell lung cancer (NSCLC) [98,99], colorectal cancer [100,101], gastric cancer [102], endometrial carcinoma [103], regulation of AKT/mTOR pathways in nasopharyngeal carcinoma [104], testicular carcinoma [105], esophageal squamous cell carcinoma [72], platinum-resistant in ovarian cancer [106], and regulation of Wnt/β-catenin pathway in cervical cancer [107].

OIP5 gene is located on chromosome 15 and encodes for a 25 kDa protein belonging to cancer-testis antigens (CTA) family [108]. The members of this family are crucial for the structure and function of kinetochore and centromeric region [109]. In the medical literature, an abnormal expression of OIP5 was linked with proliferation of colorectal and gastric cancer cells [110,111], clear cell and peripheral renal cell carcinoma (CCRCC, PRCC) progression [112,113], acute myeloid leukemia [114], proliferation of lung and esophageal cancers by RAF1 interaction [115], proliferation of thyroid cancer cell by Wnt/β-catenin signaling [116], pancreatic cancer via activation of AGR2/AKT/ERK [117] and miR-186-5p/NGFR signaling pathway [118], breast cancer by regulating GLO1 expression [119] and miR-139-5p/Notch1 pathway [120], NSCLC metastasis by mTOR signaling pathway [121] and miR-140-5p/HDAC7/VEGFA signaling [122], cervical cancer by regulating ROCK1 expression [123], ovarian cancer by mediating miR-128-3p/CCNG1 pathway [124,125], progression of osteosarcoma via miR-137-3p/PTN axis [126], multiple myeloma progression [127], prostate cancer progression via miR-128-3p/SLC7A11 signaling [128], gallbladder cancer [129], head and neck squamous cell carcinoma [130], endometrial cancer by PTEN/AKT pathway [131], Warburg effect in cervical cancer by miR-124-5p/IDH2/HIF-1α signaling [132], proliferation of bladder cancer via miR-217/MTDH pathway [133,134], NPC progression by modulating JAK2/STAT3 [135], and metastasis of glioblastoma [136].

PRC1 gene is mapped on chromosome 15 and encodes for PRC1 protein of 620 amino acids [137]. This gene highly express at early mitosis during S-G2/M phases and reduce when the cell enters into the G1 phase [137]. Expression variations in PRC1 is earlier associated with different human cancers including lung adenocarcinoma [138,139], gastric carcinoma [140], ovarian cancer [141], cholangiocarcinoma [142], liver carcinoma [143,144], prostate cancer [145], breast cancer [93,146], bladder cancer [147], colon cancer [148], osteosarcoma progression [149], human endometrial cancer [150], and nasopharyngeal carcinoma [151]. In this study, we found its significant (P < 0.05) up-regulation in LUAD patients of different clinical characteristics as compared to the normal controls. Taken together the expression profiling of the real hub genes including CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1, we have suggested that up-regulation of these six genes may serve as a potential biomarker in LUAD patients regardless of different clinical characteristics relative to controls.

Moreover, the survival analysis further suggested the application of identified hub genes as the potential prognostic biomarkers in LUAD patients. Next, we observed that real hub genes genetically altered in a small number of LUAD samples and their promoter regions were hypomethylation.

Next, we observed that real hub genes genetically altered in a small number of LUAD samples. Further analysis also revealed that mutations in the real hub genes (CDC6, PBK, KIF2C, OIP5, and PRC1) can change amino acids at distinct positions in the encoded proteins. The correlation analysis among real hub genes expression and promoter methylation levels has shown the expected significant (P < 0.05) negative correlations in LUAD. That is why, we speculated that promoter hypomethylation might be a key factor involved in the up-regulation of real hub genes across LUAD samples.

In the current study, we observed that two immune cells infiltration levels in LUAD could induce differential expression of the CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1 genes. Via TIMER, we found that real hub genes expression are significantly correlated with CD4+ T, CD8+ T immune cells, and tumor purity in LUAD. Some studies have earlier investigated the tumor-associated roles of T cells across LUAD. These studies explored that activated CD8+ T cells in LUAD often elicited type I immune responses, which indicate a favorable prognosis [152,153], however, on the other hand, Th2, and Th17 cells were found to be linked with tumor progression and unfavorable prognosis [154]. The intriguing relationships between tumour purity, CD4+ T, CD8+ T immune cells, and real hub gene expression levels found in this study may inspire fresh approaches to the treatment of LUAD.

We further elucidated that E2F1 TF and hsa-mir-124-3p miRNA target all six real hub genes (CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1). To our knowledge, this study is the first to describe the role of E2F1 and hsa-mir-124-3p in tumorigenesis when combined with CDC6, PBK, AURKA, KIF2C, OIP5, and PRC1 in LUAD. This important knowledge can be applied to the treatment of LUAD to control how real hub genes are expressed.

It was also noticed in the current study that real hub genes’ expressions were positively associated with “Cell cycle, DNA Damage, Proliferation, Stemness Invasion, EMT, and Apoptosis” in LUAD. The function of the identified hub genes in LUAD development is being collectively investigated for the first time in this study.

To find mutant genes that change the expression of real hub genes, we expanded the network of real hub genes in this study using the muTarget platform. TP53, TTN, and OR4C15 mutant genes were correlated with CDC6, TP53, KEAP1, and COL6A6 mutant genes were correlated with PBK, TP53, CSMD3, and SORCS1 mutant genes were correlated with AURKA, TP53, CSMD3, and XIR2P mutant genes were correlated with KIF2C, TP53, SMARCA4, and SCN1A mutant genes were correlated with OIP5, and TP53, TTN, and HERC2 mutant genes were correlated with PRC1. This knowledge could be useful in developing multi-gene and individualized therapeutic strategies for LUAD patients.

Conclusion

Through integrated bioinformatics approach, our study has revealed the 6 real hub genes (hub genes of hub genes) which might play pathogenic roles in LUAD development and can also use as a novel diagnostic and prognostic biomarker for the LUAD patients regardless of heterogeneity barriers. However, more in-depth experimental studies are needed before clinical applications.

Disclosure of conflict of interest

None.