Identification of Differentially Expressed Genes and Signaling Pathways in Acute Myocardial Infarction Based on Integrated Bioinformatics Analysis

Background Acute myocardial infarction (AMI) is a common disease with high morbidity and mortality around the world. The aim of this research was to determine the differentially expressed genes (DEGs), which may serve as potential therapeutic targets or new biomarkers in AMI. Methods From the Gene Expression Omnibus (GEO) database, three gene expression profiles (GSE775, GSE19322, and GSE97494) were downloaded. To identify the DEGs, integrated bioinformatics analysis and robust rank aggregation (RRA) method were applied. These DEGs were performed through Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses by using Clusterprofiler package. In order to explore the correlation between these DEGs, the interaction network of protein-protein internet (PPI) was constructed using the STRING database. Utilizing the MCODE plug-in of Cytoscape, the module analysis was performed. Utilizing the cytoHubba plug-in, the hub genes were screened out. Results 57 DEGs in total were identified, including 2 down- and 55 upregulated genes. These DEGs were mainly enriched in cytokine-cytokine receptor interaction, chemokine signaling pathway, TNF signaling pathway, and so on. The module analysis filtered out 18 key genes, including Cxcl5, Arg1, Cxcl1, Spp1, Selp, Ptx3, Tnfaip6, Mmp8, Serpine1, Ptgs2, Il6, Il1r2, Il1b, Ccl3, Ccr1, Hmox1, Cxcl2, and Ccl2. Ccr1 was the most fundamental gene in PPI network. 4 hub genes in total were identified, including Cxcl1, Cxcl2, Cxcl5, and Mmp8. Conclusion This study may provide credible molecular biomarkers in terms of screening, diagnosis, and prognosis for AMI. Meanwhile, it also serves as a basis for exploring new therapeutic target for AMI.


Introduction
Acute myocardial infarction (AMI), which represents the main public health issue around the world, is a common cardiac emergency with substantial morbidity and mortality. In the last two or three decades, although a downtrend of AMI has been observed because of the economic development and advances in medical science, its morbidity is still very high at about 44.57 in 100,000 people in China in 2013 [1]. Besides, the death rate of AMI was estimated to increase by 5.6 times from 1987 to 2014 [2]. Therefore, it is growing important for AMI to develop an early diagnosis and proper treatment strategy to prevent the occurrence of sudden mortality.
Fortunately, with the development of gene chip technique, more and more gene expression spectra were tested by gene chip technique in cardiovascular clinic and study. Microarray analysis was widely used in peripheral blood of patients with myocardial infarction [3] and the myocardium of mice [4]. Through microarray analysis, the potential genes associated with AMI will be obtained. For example, through the microarray analysis of GSE48060, Yuan Gao et al. [3] found that the MAX, BCL , NCOA , CCL , and GTF C might play 2 Cardiovascular Therapeutics a key role in AMI development, which provided valuable reference for future research. Many studies have found that early growth response factor 1 (EGR1) induces myocardial injury after AMI. Using bioinformatics analysis, Pan et al. [5] found that miR-146a can regulate the expression of EGR1. It offers help as treatment for AMI. Under many stringent states, including ischemia reperfusion, Heat shock proteins (Hsps) are produced. Novo G et al. [6] expounded the clinical significance and pathogenetic role of Hsp60 and HO-1 in AMI using bioinformatics analysis. Heart failure (HF) is a common complication after AMI. Qian C et al. [7] found that the DEGs, including FOS, THBS , CXCL , and ITGA B from the microarray data of GSE59867, may play a vital role in the occurrence and development of HF after AMI. In recent years, integrated bioinformatics analysis method is heavily used in cancer. For example, utilizing integrated bioinformatics analysis method, Guangwei et al. [8] reported the novel therapeutic targets for colorectal neoplasms. However, integrated bioinformatics analysis method is rarely employed in cardiovascular disease. In this research, three gene expression datasets, including GSE775, GSE19322, and GSE97494, were downloaded from the GEO database. These datasets were screened to identify the DEGs in each dataset. Next, using the RRA approach [9], a total of 57 DEGs, including 2 down-and 55 upregulated genes, were identified. Using Clusterprofiler [10], GO and KEGG analyses were performed, respectively. It was obviously shown that these DEGs were enriched in AMI-related functions and pathways. Then the PPI network was established by using the STRING database. The module analysis filtered out 18 key genes, including Cxcl , Arg , Cxcl , Spp , Selp, Ptx , Tnfaip , Mmp , Serpine , Ptgs , Il , Il r , Il b, Ccl , Ccr , Hmox , Cxcl , and Ccl . Ccr was the most fundamental gene in PPI network. 4 hub genes in total were identified, including Cxcl , Cxcl , Cxcl , and Mmp . Our result may provide a novel pathway for diagnosis and treatment of the AMI in the future.

Methods
. . Affymetrix Macroarray Data. Utilizing the keywords "myocardial infarction," we screened the GEO database. Three GEO datasets were found, including GSE775 contributed by Schinke et al., GSE19322 contributed by Hunt et al., and GSE97494 contributed by Chikata et al. These gene expression profiles of AMI were downloaded based on GPL81 platform of Affymetrix Murine Genome U74A Version 2 Array, GPL339 platform of Affymetrix Mouse Expression 430A Array, and GPL6246 platform of Affymetrix Mouse Gene 1.0 ST Array, respectively. There were 18 samples that were from the region between the LAD artery and the apex of the mice, 9 mice within 24 hours after AMI and 9 shamoperated mice within 24 hours. Detailed information about the datasets is listed in Table 1. Through the R software package, the download files were handled.
. . Screening for DEGs. In order to find out DEGs of each GEO dataset, utilizing the R software and annotation package, the platform and series matrix file(s) were converted. These DEGs in AMI and sham operation group samples were analyzed by utilizing the limma package [11] in R. Log2(fold change) (log2FC) > 1 and a corrected value < 0.05 were used as the cut-off criteria of DEGs samples.
. . Integration of Microarray Data. Through limma packet analysis, we obtained the list of DEGs of the three microarray datasets. The list of down-and upregulated genes in the microarray data was saved. Subsequently, using the RRA approach, the comparison of multiple ranked gene lists was performed.
. . GO and KEGG Pathway Enrichment Analyses. Biological functions of the DEGs obtained from the integration of microarray data were explored with GO analysis using Clusterprofiler which is an R package utilized to compare the biological themes among gene clusters. Similarly, in order to identify the enrichment signaling pathways of DEGs, KEGG pathway analysis was performed by utilizing the Clusterprofiler package. A corrected p < 0.05 was the cut-off criterion.
. . PPI Network Integration, Modules Analysis, and Selection of Hub Genes. In order to identify the interaction between PPI, the PPI network was built using the STRING (version 11) online database. The highest confidence of the argument of interactions was set at >0.4. To draw an interaction of DEGs, the Cytoscape (version 3.6.1) software was used to visualize and analyse the PPI network. In order to find modules of the whole network, the Molecular Complex Detection (MCODE) plug-in of the Cytoscape software was applied. The hub genes were identified by using the plug-in cytoHubba [12] of the Cytoscape software, including Density of Maximum Neighborhood Component (DMNC) and Maximal Clique Centrality (MCC).

. . Identification of DEGs in GSE , GSE
, and GSE . Three expression microarray datasets, including GSE775, GSE19322, and GSE97494, were used to perform  background correction and quartile data normalization by the limma package. Meanwhile, using the limma package (log2FC >1, corrected p <0.05), the GSE775 dataset was screened and 2149 DEGs were obtained, including 23 downand 2126 upregulated genes. Using the same methodology, 597 DEGs were obtained from the GSE19322 dataset, including 446 down-and 151 upregulated genes, and 4534 DEGs were confirmed from the GSE97494 dataset, including 3879 down-and 655 upregulated genes. Many DEGs in two sets of sample data of each microarray, three microarrays in total, are shown in Figures 1(a)-1(c), also known as the volcano plots of DEGs. In order to evaluate the biological repeatability, we drew an association diagram, which indicated that the biological repeatability of the sample was well, as shown in . . Identification of DEGs in AMI Utilizing Integrated Bioinformatics Analysis. Using the RRA method according to Log2FC >1 and a corrected p <0.05, the list of DEGs of the three microarray datasets were analyzed. A total of 57 DEGs were determined by rank analysis, including 2 down-and 55 upregulated genes, as shown in Table 2. The heatmap of the 57 DEGs was drawn by heatmap package, which is shown in Figure 3.
. . GO Analysis of DEGs. Using Clusterprofiler package, biological annotation of the DEGs obtained by RRA approach Figure 2: Hierarchical clustering heatmap of DEGs, which was screened on the basis of log2FC >1.0 and a corrected p <0.05. Notes: (a) GSE775 data, (b) GSE19322 data, and (c) GSE97494 data. Red represents that the expression of genes is relatively upregulated. Blue represents that the expression of genes is relatively downregulated. Gray represents the expression of genes without significant changes. Abbreviation: DEGs, differentially expressed genes; FC, fold change. was performed. The down-and upregulated genes with value <0.05 were obtained from GO functional enrichment. From GO functional enrichment analysis, we identified that these DEGs were mainly enriched in the following functional categories, including receptor ligand activity, cytokine activity, cytokine receptor binding, G-protein coupled receptor binding, carbohydrate binding, chemokine activity, and chemokine receptor binding. GO analyses are shown in Figure 4. Meaningful results of the GO analysis of DEGs in AMI are listed in Table 3.
. . KEGG Pathway Analysis of DEGs. Top 20 KEGG pathway analyses of DEGs are shown in Table 4 and Figure 5. Table 4 shows that these DEGs were primarily enriched in the cytokine-cytokine receptor interaction, Chemokine signaling pathway, TNF signaling pathway, and so on.
. . Establishing the PPI Network, Conducting Modules Analysis, and Selection of Hub Genes. In order to ulteriorly explore the biological characteristics of these DEGs, a PPI network was created using the STRING database. There were 56 nodes and 240 edges in this network, including 2 Cardiovascular Therapeutics 5 Cxcl2 S100a8 Arg1 Figure 3: The heatmap of differentially expressed genes. Notes: each column and row represents one dataset and one gene, respectively. Red and green represent logFC >0 and logFC <0, respectively. The logFC values are shown in each rectangle. The gradual color ranged from green to red represents the changing process from downregulation to upregulation. Abbreviation: FC, fold change. down-and 54 upregulated genes (see the supplementary document (available here)), as shown in Figure 6(a). Subsequently, a vital module was confirmed from the whole network, a total of 18 nodes and 117 edges in this module, as shown in Figure 6(b). 18 key genes in total were identified, including Cxcl , Arg , Cxcl , Spp , Selp, Ptx , Tnfaip , Mmp , Serpine , Ptgs , Il , Il r , Il b, Ccl , Ccr , Hmox , Cxcl , and Ccl . Ccr was the most key gene in PPI network. These genes in the module were mainly enriched in the cytokinecytokine receptor interaction, TNF signaling pathway, Tolllike receptor signaling pathway, and chemokine signaling pathway, as shown in Table 4. Utilizing the cytoHubba plugin, Cxcl , Cxcl , Cxcl , and Mmp hub genes were screened out, as shown in Figure 6(c).

Discussion
AMI is one of the common kinds of coronary heart disease with high morbidity and mortality all over the world. In recent years, the number of patients with AMI is increasing annually. Controlling the number of patients with AMI and exploring the molecular mechanism of AMI are urgent to be solved.
In the study, using integrated bioinformatics and RRA analysis method, a total of 57 DEGs, including 2 downand 55 upregulated genes, were identified from the GSE775, GSE19322, and GSE97494 database. From GO functional enrichment analysis, we identified that these DEGs were mainly enriched in the following functional categories, including receptor ligand activity, cytokine activity, cytokine Most of these genes in AMI have been reported, which indicated that the results of integrated bioinformatics analysis were reliable. Chemokine (C-C motif) receptor 1 (Ccr ), the highest score, was identified from the module. Ccr is inflammationassociated gene, which may be a novel biomarker for the diagnosis and prognosis of AMI [13]. It exerts an important role in controlling inflammation [14]. Significantly, during the pathogenesis of AMI, inflammation of the coronary artery is the key process [15,16]. We found that Ccr mainly enriched in cytokine-cytokine receptor interaction and chemokine signaling pathway from the KEGG pathway analysis, which may be a direction of future research for diagnosis and treatment of AMI. In mice, chemokine (C-X-C motif) ligand 2 (Cxcl ) plays a kind of the potent neutrophil chemoattractants [17]. Using pharmacologic inhibition of circulating Cxcl , researchers found neutrophil recruitment reduced at the site of myocardial infarction and injury within the infarcted myocardium alleviated [17]. Expression of Cxcl and Cxcl in AMI was elevated, which aggravated acute inflammation after myocardial injury and promoted cardiac rupture [18,19]. From the KEGG pathway analysis, we found that Cxcl mainly enriched in cytokine-cytokine receptor interaction, TNF signaling pathway, and chemokine signaling pathway. Thus, Cxcl may play a key role in regulating cardiac remodeling following myocardial infarction (MI). Chemokine (C-C motif) ligand 3 (Ccl ) is also an important circulating chemokine. Tineke et al. [20] showed that CCL is highly upregulated in patients with AMI. Vandervelde et al. clearly   showed that the Ccl mRNA expression was upregulated in ischemic myocardium [21]. These evidences indicated that Ccl is closely associated with myocardial ischemia. Our study found that Ccl was primarily enriched in cytokinecytokine receptor interaction, Toll-like receptor signaling pathway, and chemokine signaling pathway. In experimental models of AMI, the innate immune response was induced through activation of Toll-like receptor (TLR)2 and TLR4 on circulating blood cells, which increases infarct size and influences ventricular remodeling [22,23]. In the model of myocardial infarction, pharmacological inhibition of TLR2 or TRL4 can decrease monocyte inflow into the infarcted region, decrease the infarct area, and enhance myocardial remodeling [24][25][26]. From the above evidence, we identified the importance of cytokine-cytokine receptor interaction and chemokine signaling pathway in the occurrence and development of AMI. Prostaglandin-endoperoxide synthase 2 (PTGS , also named as COX-2), which can increase the neoplastic process by promoting proliferation, suppressing apoptosis, and angiogenesis, is an enzyme during conversion of arachidonic acid to prostaglandins [27]. PTGS has high expression in every kind of tumor, which was usually induced by cancer promoters, oncogenes, and cytokines [28]. PTGS gene associated with the decreasing risk of stroke and MI has been demonstrated [29]. Therefore, it plays a crucial role in treatment of MI. From KEGG pathway analysis, we found that PTGS was enriched in TNF signaling pathway. It has been reported that TNF signaling pathway was associated with cardiac remodeling following MI [30]. So we speculate that PTGS exerts an important role on regulating cardiac remodeling following MI through TNF signaling pathway. We look forward to the result being confirmed by future experiments.
Among these genes, a novel gene Tnfaip (tumor necrosis factor-stimulated gene-6) was obviously differentially expressed in AMI. Interestingly, this gene was mainly reported in inflammatory bowel disease [31]. According to integrated bioinformatics analysis, we speculated that Tnfaip may play an important role in AMI, which could be a novel target for the treatment of AMI. Thus, further studies are needed in order to verify it.
Wei Gong et al. [32] have found that trimetazidine can prevent cardiac rupture in mice with AMI through inhibiting the expression of Mmp and Mmp , which indicates that the MMP family may be associated with cardiac remodeling after AMI. Matrix metalloproteinase-8 (Mmp ), a member of the MMP family, has gained growing attention in recent years. Previous research had only identified that types I, II, and III collagens are the substrates of Mmp . However, in recent years, an increasing number of other proteins were detected as the substrates of Mmp , including chemokine (C-X-C motif) ligand 5 (CXCL ) [33], macrophage inflammatory protein-1 [34], chemokine (C-X-C motif) ligand 11 (CXCL ) [35], and angiotensin-1 [36]. Research indicated that Mmp can regulate the function and behavior of multiple cell types, including stem/progenitor cells [37], endothelial cells [38], smooth muscle cells [39], and neutrophils [34]. Study showed that gingival crevicular fluid Mmp concentrations significantly increase in patients with AMI [40]. Bioinformatics analysis indicates that Mmp may be associated with prognosis of AMI. Nevertheless, little is known about the relation between Mmp and cardiac remodeling. Therefore, more experiments were needed to verify it in the future.
It is noticeable that there have been papers researching the differentially expressed genes in AMI. However, the results of those papers were somewhat different from ours. The following reasons may account for this phenomenon: (1) some studies [3,13], which have been reported, are peripheral blood microarray analysis of patients with AMI. Nevertheless, our study is microarray analysis of LV myocardium of mouse with AMI. Because the sample origin and the timing of specimen collection are different [4], which leads to somewhat different results, (2) different batches of microarray analysis, to some extent, also have somewhat different results; (3) compared with other studies of AMI, our study provides an integrated bioinformatics analysis of DEGs of AMI by means of statistical methods. We may provide credible results. Of course, it is important that the results are validated in follow-up experiments.

Conclusion
In conclusion, our study provides an integrated bioinformatics analysis of DEGs of AMI. This research provides numerous genes associated with AMI. This study may provide credible molecular biomarkers in terms of screening, diagnosis, and prognosis for AMI. Meanwhile, it also serves as a basis for exploring new therapeutic target for AMI. Compared with other studies of AMI, innovation point and merit of our current study was that the RRA method was utilized for the first time in exploring DEGs in AMI study. This study also has certain limitations. In this study, 18 microarrays were only screened, which is not enough. The limited sample size may easily lead to false positive results. Therefore, to verify the current findings, it is necessary to perform more experiments.

Data Availability
The data used to support the findings of this study are included within the supplementary information file.

Conflicts of Interest
The authors declare that they have no conflicts of interest.