Transcriptome Signatures Reveal Candidate Key Genes in the Peripheral Blood Mononuclear Cells of Patients With Coronary Artery Disease


 BackgroundCoronary artery disease (CAD) is one of the most common disorders in the cardiovascular system. This study aims to explore potential signaling pathways and important biomarkers that drive CAD development. MethodsThe CAD GEO Dataset GSE113079 was featured to screen differentially expressed genes (DEGs). The pathway and Gene Ontology (GO) enrichment analysis of DEGs were analyzed using the ToppGene. We screened hub and target genes from protein-protein interaction (PPI) networks, target gene - miRNA regulatory network and target gene - TF regulatory network, and Cytoscape software. Validations of hub genes were performed to evaluate their potential prognostic and diagnostic value for CAD. Results1,036 DEGs were captured according to screening criteria (525upregulated genes and 511downregulated genes). Pathway and Gene Ontology (GO) enrichment analysis of DEGs revealed that these up and down regulated genes are mainly enriched in thyronamine and iodothyronamine metabolism, cytokine-cytokine receptor interaction, nervous system process, cell cycle and nuclear membrane. Hub genes were validated to find out potential prognostic biomarkers, diagnostic biomarkers and novel therapeutic target for CAD. ConclusionsIn summary, our findings discovered pivotal gene expression signatures and signaling pathways in the progression of CAD. CAPN13, ACTBL2, ERBB3, GATA4, GNB4, NOTCH2, EXOSC10, RNF2, PSMA1 and PRKAA1 might contribute to the progression of CAD, which could have potential as biomarkers or therapeutic targets for CAD.


Introduction
Coronary artery disease (CAD) remains the one of leading healthy issues worldwide and 23.3 million people will die of CAD by 2030 [1]. The risk factors for CAD mainly smoking, high blood pressure, high blood cholesterol levels, diabetes, overweight or obesity, physical inactivity, high stress and unhealthy diet [2]. At present, surgery has been applied to improve survival of CAD patients [3]. However, the molecular pathogenesis of CAD advancement is still largely unknown.
As an inventive and high-throughput investigation facilitate the concurrent analysis of expression changes in thousands of genes in CAD samples and contributes to diagnosis, prognosis and new drug discovery [4]. In current years, there have been huge research on the molecular pathogenesis of CAD selected as targets for DEGs to construct differentially expressed miRNA. Target genes were arranged into the miRNA regulatory network separately to access each miRNA regulatory network which were visualized using Cytoscape (http://www.cytoscape.org/) [31]. DEGs (up and down regulated) interaction with more number of miRNAs consider as target genes.

Construction of target gene -TF regulatory network
Transcription factor gene data of the NetworkAnalyst (https://www.networkanalyst.ca/) [38] was used to identify the transcription factor regulatory networks linked with the target genes. The NetworkAnalyst describes transcription factor (TF) to genes from the perspective of ChEA database (http://amp.pharm.mssm.edu/lib/chea.jsp) [41] database resource. The NetworkAnalyst illustrate a more extensive transcription factor regulation network. Target genes were arranged into the TF regulatory network separately to access each transcription factor regulatory network which were visualized using Cytoscape (http://www.cytoscape.org/) [31]. DEGs (up and down regulated) interaction with more number of TFs consider as target genes.

Validations of hub genes
The human protein atlas database (HPA) (www.proteinatlas.org) [42] was used to analyze protein expression of hub genes in peripheral blood mononuclear cells in bone marrow. A receiver operating characteristic (ROC) curve was produce using the pROC package of the R software [43], and the area under the curve (AUC) was determined using generalized linear model (GLM) in machine learning algorithms to assess the predictive accuracy of hub genes.

Data preprocessing and identi cation of DEGs
The gene expression pro le with accession numbers GSE113079 was downloaded from GEO database. The results of before and after normalizing the microarray gene expression are shown in Fig. 1A and Fig.   1B. DEGs between peripheral blood mononuclear cells from CAD patients and peripheral blood mononuclear cells from healthy control were screened using limma package in R bioconductor (P-value <0.05, |logFC| > 0.97 for up regulated genes, and |logFC| < -0.963 for down regulated genes). In this study, 1,036 total DEGs (525 up regulated genes and 511 down regulated genes, respectively) in GSE113079 was screened. The total number of DEGs collected for each subject in the differential gene expression analysis is given in Table 1. A volcano diagram was constructed for the DEGs and is shown in Fig. 2. The DEGs (up and down regulated genes) are presented by a cluster heatmap in Fig. 3 and Fig. 4.

Pathway enrichment analysesof DEGs
Pathway enrichment analyses were performed using ToppGene, analyzing the pathway classi cation of DEGs (up and down regulated genes). Pathways of up regulated were mainly enriched in thyronamine and iodothyronamine metabolism, trehalose degradation, cytokine-cytokine receptor interaction, Olfactory disease, epithelial cell differentiation, calcium signaling pathway, E2F transcription factor network, notchmediated HES/HEY network, receptor regulator activity, cation transport and Jak-STAT signaling pathway. The result of PPI network of down regulated was illustrated in Fig 7. A total of 5135 nodes with 10628 edges were re ected in this well-established network system. The statistical results and scatter plot for node degree distribution, betweenness centrality, stress centrality, closeness centrality and clustring coe cient are shown in Fig. 8A -8E and indicated that FYN, PAK2, CUL3, RPS6, NOTCH2, PDE4D,   SPATA21, MYBL1, SMURF1, PDGFRB, DLG3, ADHFE1, NMB, SLC25A36, MLLT1 and RNF2 were the hub genes with high node degree distribution, betweenness centrality, stress centrality, closeness centrality and low clustring coe cient in the network are listed in Table 6. The top hub genes in this PPI network were selected for further pathway and GO enrichment analyses using the ToppGene database. The results indicated that the hub genes were mainly enriched in natural killer cell mediated cytotoxicity, Fcepsilon receptor I signaling in mast cells, signaling by interleukins, mTORsignaling pathway, notch signaling pathway, purine metabolism, gene expression, endocytosis, cytokine-cytokine receptor interaction, regulation of hydrolase activity, signaling receptor binding, organelle envelope, nuclear chromatin and post-translational protein modi cation.
Four signi cant modules were selected for each up and down regulated genes using the PEWCC1E plugin. The top four modules for up regulated genes were selected (module 13, 105 nodes and 235 edges; module 20, 77 nodes and 97 edges; module 21, 73 nodes and 81 edges; module 34, 53 nodes and 58 edges) are shown in Fig. 9. The results revealed that hub genes (ACTG2, GATA4, EGFR, TP73, ACTBL2, FOXJ1, BMP7 and CDK5R2) in these signi cant modules were mostly enriched in the muscle contraction, notch-mediated HES/HEY network, cytokine-cytokine receptor interaction, E2F transcription factor network, actin cytoskeleton, epithelial cell differentiation, biological adhesion and neuron projection. Similarly, top four modules for down regulated genes were selected (module 1, 92 nodes and 186 edges; module 2, 56 nodes and 187 edges; module 5, 49 nodes and 144 edges; module 11, 29 nodes and 57 edges) are shown in Fig. 10. The results revealed that hub genes (RPS6, PAK2, PODN, LMNA, EIF1AX, RPS27, HSPA8, FYN and LMNB1) in these signi cant modules were mostly enriched in the mTOR signaling pathway, Fc-epsilon receptor I signaling in mast cells, ensemble of genes encoding extracellular matrix and extracellular matrix-associated proteins, caspase cascade in apoptosis, postsynapse, cell cycle, regulation of immune system process and positive regulation of signal transduction.

Construction of target gene -miRNA regulatory network
NetworkAnalyst was applied to screen the miRNAs of the up and down regulated genes. The miRNAs predicted by at least two databases (among the following databases: DIANA-TarBase and miRTarBase) were diagnosed as the miRNAs of the target genes. Then, Cytoscape software was used to draw the target gene -miRNA regulatory network. The target gene -miRNA regulatory network for up regulated genes included 1867 nodes and 3735 edges (Fig. 11). As shown in Table 7, TRIM72 regulates 123 miRNAs (ex,hsa-mir-4537), TET3 regulates 105 miRNAs (ex,hsa-mir-3148), NFIB regulates 89 miRNAs (ex,hsa-mir-4517), SLC19A3 regulates 80 miRNAs (ex,hsa-mir-4500) and SMOC1 regulates 123 miRNAs (ex,hsa-mir-6133) were considered as target gene. We performed pathway and GO enrichment analysis of these predicted target genes, which mainly enriched in muscle contraction, FOXA1 transcription factor network, intrinsic component of plasma membrane and biological adhesion. The target gene -miRNA regulatory network for down regulated genes included 2529 nodes and 10243 edges (Fig.12). As shown in Table 7, PPP1R15B regulates 168 miRNAs (ex, hsa-mir-7150), WEE1 regulates 167 miRNAs (ex,hsa-mir-3926), RPRD2 regulates 152 miRNAs (ex,hsa-mir-4452), LCOR regulates 146 miRNAs (ex,hsa-mir-4310) and SAR1A regulates 145 miRNAs (ex,hsa-mir-5698) were considered as target gene. We performed pathway and GO enrichment analysis of these predicted target genes, which mainly enriched in regulation of hydrolase activity, cell cycle, gene expression, nuclear chromatin and protein processing in endoplasmic reticulum.

Construction of target gene -TF regulatory network
NetworkAnalyst was applied to screen the TFs of the up and down regulated genes. The TFs predicted by database (ChEA database) was diagnosed as the TFs of the target genes. Then, Cytoscape software was used to draw the target gene -TF regulatory network. The target gene -TF regulatory network for up regulated genes included 539 nodes and 5790 edges (Fig. 13). As shown in Table 8, ACTL8 regulates 145 TFs (ex, EGR1), LHFPL3 regulates 132 TFs (ex,SOX2), CXCL12 regulates 119 TFs (ex,SUZ12), GLI2 regulates 117 TFs (ex, AR) and C7 regulates 114 TFs (ex, TP53) were considered as target gene. We performed pathway and GO enrichment analysis of these predicted target genes, which mainly enriched in epithelial cell differentiation, cytokine-cytokine receptor interaction, pathways in cancer and innate immune system. The target gene -TF regulatory network for down regulated genes included 608 nodes and 10262 edges (Fig. 14). As shown in Table 8, PRIM2 regulates 218 TFs (ex, SOX2), regulates 211 TFs (ex, MYC), GMDS regulates 210 TFs (ex, SPI1), C5ORF58 regulates 190 TFs (ex, RUNX1) and C10orf88 regulates 180 TFs (ex, FLI1) were considered as target gene. We performed pathway and GO enrichment analysis of these predicted target genes, which mainly enriched in metabolic pathways, gene expression, asparagine N-linked glycosylation and cell cycle.
In the target gene -TF regulatory network, 5 up regulated genes and 5 down regulated genes with a high node degree was chosen as target gene. GLI2 was linked with progression of obesity [169], but this gene may be responsible for advancement of CAD. Our study found that LHFPL3 is up regulated in CAD and has potential as a novel diagnostic and prognostic biomarker, similarly, our study found that EXOSC10, GDP-mannose 4,6-dehydratase (GMDS), C5ORF58 and C10orf88 are down regulated in CAD and has potential as a novel diagnostic and prognostic biomarker, and therapeutic target.
We used immune histochemical (IHC) analysis, receiver operating characteristic (ROC) curve and RT-PCR to analyze the association of 5 up and 5 down regulated hub genes expression in CAD. The results showed that only 5 up (CAPN13, ACTBL2, ERBB3, GATA4 and GNB4) and 5 down (NOTCH2, EXOSC10, RNF2, PSMA1 and PRKAA1) regulated hub genes were related to the CAD. We then evaluated the prognostic value of these only 5 up (CAPN13, ACTBL2, ERBB3, GATA4 and GNB4) and 5 down (NOTCH2, EXOSC10, RNF2, PSMA1 and PRKAA1) regulated hub genes using the ROC curve, indicating that they have potential diagnostic value for CAD.
In conclusion, 1,036 DEGs (525 up rand 511 down regulated gene) were screened out from the GSE113079 dataset, which may contain hub genes contributing to the pathogenesis of CAD. Through our bioinformatics analysis, hub genes including CAPN13, ACTBL2, ERBB3, GATA4, GNB4, NOTCH2, EXOSC10, RNF2, PSMA1 and PRKAA1 might contribute to the progression of CAD, which could serve as novel diagnostic and prognostic biomarkers and therapeutic targets for CAD. Declarations Acknowledgement I thank Lin Li, Department of Epidemiology, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Diseases, Xicheng District, Beijing, China, very much, the author who deposited their microarray dataset, GSE113079, into the public GEO database.

Con ict of interest
The authors declare that they have no con ict of interest.

Ethical approval
This article does not contain any studies with human participants or animals performed by any of the authors.

Informed consent
No informed consent because this study does not contain human or animals participants.

Availability of data and materials
The datasets supporting the conclusions of this article are available in the GEO (Gene Expression Omnibus) (https://www.ncbi.nlm.nih.gov/geo/) repository.   Volcano plot of differentially expressed genes. Genes with a signi cant change of more than two-fold were selected. Protein-protein interaction network of up regulated genes. Green nodes denotes up regulated genes. Protein-protein interaction network of down regulated genes. Red nodes denotes down regulated genes. Modules in PPI network. The green nodes denote the up regulated genes Page 33/36

Figure 10
Modules in PPI network. The red nodes denote the down regulated genes.

Figure 11
The network of up regulated genes and their related miRNAs. The green circles nodes are the up regulated genes, and chocolate color diamond nodes are the miRNAs Figure 12 The network of down regulated genes and their related miRNAs. The red circles nodes are the down regulated genes, and blue diamond nodes are the miRNAs Immune histochemical analyses of hub genes were produced using the human protein atlas (HPA) online platform.