Bioinformatic screening for key miRNAs and genes associated with myocardial infarction

Despite significant advances in understanding of the causes of and treatment of myocardial infarction (MI) in recent years, morbidity and mortality is still high. The aim of this study was to identify miRNA and genes potentially associated with MI. mRNA and miRNA expression datasets were downloaded from the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/). Interactions between miRNA and the expression and function of target genes were analyzed, and a protein–protein interaction network was constructed. The diagnostic value of identified miRNA and genes was assessed. Quantitative RT‐PCR was applied to validate the results of the bioinformatics analysis. MiR‐27a, miR‐31*, miR‐1291, miR‐139‐5p, miR‐204, miR‐375, and target genes including CX3CR1,HSPA6, and TPM3 had potential diagnostic value. The genes TFEB,IRS2,GRB2,FASLG,LIMS1,CX3CR1,HSPA6,TPM3,LAT2,CEBPD,AQP9, and MAPKAPK2 were associated with recovery from MI. In conclusion, the identified miRNA and genes might be associated with the pathology of MI.

Despite significant advances in understanding of the causes of and treatment of myocardial infarction (MI) in recent years, morbidity and mortality is still high. The aim of this study was to identify miRNA and genes potentially associated with MI. mRNA and miRNA expression datasets were downloaded from the Gene Expression Omnibus database (http:// www.ncbi.nlm.nih.gov/geo/). Interactions between miRNA and the expression and function of target genes were analyzed, and a protein-protein interaction network was constructed. The diagnostic value of identified miRNA and genes was assessed. Quantitative RT-PCR was applied to validate the results of the bioinformatics analysis. MiR-27a, miR-31*, miR-1291, miR-139-5p, miR-204, miR-375, and target genes including CX3CR1, HSPA6, and TPM3 had potential diagnostic value. The genes TFEB, IRS2, GRB2, FASLG, LIMS1, CX3CR1, HSPA6, TPM3, LAT2, CEBPD, AQP9, and MAPKAPK2 were associated with recovery from MI. In conclusion, the identified miRNA and genes might be associated with the pathology of MI.
Myocardial infarction (MI) is the multifactorial injurious event, which involves all the components of the cardiac myocyte [1]. The partial or complete occlusion of a coronary artery is the cause of MI, which leads to apoptosis and necrosis in the myocardium. It is reported that smoking, hypertension, diabetes mellitus, hypercholesterolemia, or dyslipidemia are the main causal factors of MI [2]. Atherosclerosis also leads to MI [3]. Even though survival rates after MI have been remarkably improved by early revascularization therapy and drug treatment, a significant number of patients develop heart failure [4]. Despite significant advances in understanding of the causes of and treatment of MI in recent years, morbidity and mortality is still high [5].
miRNA are small, noncoding RNA that regulate the expression of target genes at the post-transcription level. miRNA can modulate important complex gene regulatory pathways involved in cardiovascular development [6][7][8]. More and more evidence reveals that signature expression pattern of miRNA plays a vital role in MI, cardiac arrhythmia, and pathological cardiac hypertrophy [9]. It is noted that some miRNA expressed in heart are remarkably deregulated in patients with acute MI compared with healthy controls [10,11]. It is found that several miRNA including miR-1, miR-21, miR-206, and miR-499-5p are deregulated in MI [12][13][14]. Clinically, myoglobin, cardiac troponins, N-terminal probrain natriuretic peptide, creatine kinases, and lactate dehydrogenase have been considered as diagnosis biomarkers of patients with acute MI [15][16][17][18]. It is worth mentioning that several miRNA including miR-1, miR-208a, miR-126, miR-122-5p, and miR-19a have been recognized as novel biomarkers for early diagnosis of acute MI [19][20][21][22]. Therefore, improving knowledge about the interaction between miRNA and target genes may be helpful in finding new pathological mechanism and markers for MI.
In this study, we aimed to find differentially expressed miRNA and genes in MI by integrated analysis. The miRNA-gene target analysis was subsequently performed. Then, functional enrichment analysis including Gene Ontology (GO) and Kyoto Encyclopedia of Genes Genomes (KEGG) was used to investigate the biological function of genes followed by construction of a protein-protein interaction (PPI) network of top 100 differentially expressed genes (DEGs; 50 up-regulated and 50 down-regulated). Receiver operating characteristic (ROC) analysis was applied to analyze the diagnostic usefulness of identified differentially expressed miRNA and genes. Quantitative RT-PCR (qRT-PCR) was used to validate the result of the bioinformatics analysis. GSE29532 and GSE48060 datasets were used for expression and recovery analysis of DEGs. Our study may be helpful in understanding the pathogenic mechanism and finding valuable diagnosis biomarkers for MI.

Datasets
In this study, we searched datasets from the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih. gov/geo/). The study type was characterized as 'expression profiling by array'. All selected datasets were genome-wide expression data of mRNA/miRNA from MI group and normal group blood samples. And those standardized or primary datasets were included in this study. Finally, a total of three mRNA datasets (GSE34198, GSE48060, and GSE61145) and two miRNA datasets (GSE31568 and GSE61741) were screened, which was shown in Table 1.

Identification of differentially expressed genes and differentially expressed miRNA
With numbers of publicly available microarray databases, there will be an interest in combining data from different platforms. It is noted that metaMA package allows this integration of different platforms as it can handle missing data and eliminate the batch effects [23]. In this study, Limma and metaMA packages were used to identify the DEGs. The normal inverse method was used to combine the P value in metaMA. The false discovery rate (FDR) was performed for multiple testing corrections of raw P value through the Benjamin and Hochberg method [24,25]. The threshold of DEGs was set as FDR < 0.05. The clustering is constructed using the complete-linkage method together with the Euclidean distance. Each row represents a DEG and each column, a sample. DEGs clustering tree is shown on the right. The color scale illustrates the relative level of DEGs expression: purple, below the reference channel; green, higher than the reference.

miRNA-gene target analyses
Identifying the target genes of miRNA is a crucial step in exploring the function of miRNA in specific tissues and cells. Herein, six miRNA-target prediction tools (RNA22, miRanda, miRDB, miRWalk, PICTAR2, and Targetscan) were applied to predict the target DEGs of differentially expressed miRNA. The miRNA-targets that were predicted by more than four algorithms or verified by experiment in miRWalk database were screened out. Then, the miRNAtarget regulatory network was constructed, which was visualized using Cytoscape Software [26].
Functional annotation analyses of miRNA-target differentially expressed genes

Protein-protein interaction network construction
It is useful for understanding the molecule mechanism of MI by studying the interactions between proteins. To gain insights into the interaction between proteins encoded by DEGs and other proteins, the database of BioGRID (http://thebiogrid.org) was used to retrieve the predicted interactions between top 100 proteins encoded by DEGs (50 up-regulated and 50 down-regulated) and other proteins. Then, PPI network was visualized by the Cytoscape Software (http://cytoscape.org/). A node in the PPI network denotes protein, and the edge denotes the interactions.

Receiver operating characteristic analyses
Using pROC package in R language, we performed the receiver operating characteristic analyses to assess the diagnostic value of DEGs. The area under the curve (AUC) under binomial exact confidence interval was calculated, and the receiver operating characteristic curve was generated.

Validation of quantitative RT-PCR
In this study, five patients diagnosed as MI and five normal individuals were enrolled in this study. Both MI and corresponding normal blood samples were obtained and immediately frozen in liquid nitrogen. All participating individuals provided informed consent with the approval of the ethics committee of our hospital. Total RNA of fresh blood samples from MI patients and normal individuals was extracted using TRizol reagent (Invitrogen, Foster City, CA, USA) according to the manual instructions. SuperScript III Reverse Transcription Kit (Invitrogen) was used to synthesize the cDNA. qRT-PCR was performed using SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA) on Applied Biosystems 7500 (Applied Biosystems). GAPDH served as internal control for gene detection, and the relative expression of genes was calculated using the fold change equation.

Expression analyses in the early stage of myocardial infarction and recovery-related analysis of differentially expressed genes
To analyze the expression of DEGs in the early stage (different blood collection time points, time 1, time 2, time 3, time 4, time 5, and time 6) of MI in the dataset of GSE29532 and further investigate the association between DEGs and MI recovery in the dataset of GSE48060, 12 DEGs including transcription factor EB (TFEB), insulin receptor substrate 2 (IRS2), growth factor receptor bound protein 2 (GRB2), fas ligand (FASLG), LIM zinc finger domain containing 1 (LIMS1), C-X3-C motif chemokine receptor 1 (CX3CR1), heat shock protein family A (Hsp70) member 6 (HSPA6), tropomyosin 3 (TPM3), linker for activation of T cells family member 2 (LAT2), CCAAT/enhancer binding protein delta (CEBPD), aquaporin 9 (AQP9), and mitogen-activated protein kinase-activated protein kinase 2 (MAPKAPK2) were selected for analysis.

Results
Differentially expressed genes and differentially expressed miRNA analysis A total of 1007 DEGs were identified as the threshold of FDR < 0.05, consisting of 564 up-regulated and 443 down-regulated genes. Top 10 up-and down-regulated DEGs were presented in Table 2. The heat map of top 100 DEGs was shown in Fig. 1. All DEGs were listed in the  Table S1. In addition, a total of 38 differentially expressed miRNA including 14 up-regulated and 24 down-regulated miRNA. Top 10 up-and down-regulated differentially expressed miRNA was listed in Table 3. Figure 2 showed the heat map of all differentially expressed miRNA.

miRNA-target gene interactions
A total of 1186 miRNA-target pairs including 392 upregulated miRNA-down-regulated target pairs and 639 down-regulated miRNA-up-regulated target pairs were obtained. Among which, 132 up-regulated miRNAdown-regulated target pairs and 113 down-regulated miRNA-up-regulated target pairs have been confirmed by miRWalk. The MI-specific miRNA-target interaction network was shown in Fig. 3. Table 4 listed the target DEGs of differentially expressed miRNA.

Enrichment analyses of target genes of differentially expressed miRNA
In target genes analysis of differentially expressed miRNA, a total of 528 target DEGs were obtained.

Protein-protein interaction network
To obtain the interaction between the proteins encoded by DEGs and other proteins, PPI network was explored and visualize by Cytoscape.

Receiver operating characteristic curve analysis
We performed receiver operating characteristic curve analyses and calculated the AUC to assess the discriminatory ability of selected six miRNA (miR-27a, miR-31*, miR-139-5p, miR-204, miR-375, and miR-1291) and three DEGs (CX3CR1, HSPA6, and TPM3) from GEO dataset (Fig. 6). The AUC of all these DEGs and miRNA was > 0.7. MiR-27a, miR-31*, and miR-1291 had the largest AUC. For MI diagnosis, the specificity and sensitivity of miR-27a were 72.9% and 95%, respectively; the specificity and sensitivity of miR-31* were 78.6% and 85%, respectively; the specificity and sensitivity of miR-139-5p were 84.3% and 65%, respectively; the specificity and sensitivity of miR-204 were 77.1% and 75%, respectively; the specificity and sensitivity of miR-375 were 88.6% and 65%, respectively; the specificity and sensitivity of miR-1291 were 90% and 65%, respectively; the specificity and sensitivity of CX3CR1 were 96.8% and 52.4%, respectively; the specificity and sensitivity of HSPA6 were 67.7% and 81%, respectively; the specificity and sensitivity of TPM3 were 71% and 71.4%, respectively. In addition, the data of the receiver operating characteristic analysis including the C-statistic and 95% confidence interval and odds ratio and 95% confidence interval were listed in Table 6.

Discussion
Myocardial infarction is an important clinical problem because of its large contribution to mortality. Therefore, it is urgent to elucidate MI carcinogenesis mechanism for developing novel diagnose that will be highly specific to malignant cells, with minimal or no risk of adverse effect. In this study, we found several differentially expressed miRNA and genes, which may play an important role in the development of MI. Linker for activation of T cells family member 2 is involved in the process of calcium mobilization, which is associated with coronary artery calcification in atherosclerosis [27]. Herein, we found that LAT2 was regulated by both miR-767-5p and miR-1245 in the blood of MI. MiR-767-5p has been found differentially expressed in the heart tissue of patients with MI [11]. However, there are not any reports about miR-1245 in the development  of heart. Further research is needed to study the function of miR-1245.
CCAAT/enhancer binding protein delta (also called CELF) is a transcription factor important in activating the expression of inflammatory genes in cardiac myocytes [28]. In animal models, the expression of CEBPD altered in skeletal and muscular functions and overexpression of the dominant negative CEBPD protein results in fibrosis, cardiac hypertrophy, and dilated cardiomyopathy [29][30][31]. In our study, CEBPD was regulated by miR-455-3p in the blood of MI. It is noted that the expression of miR-455-3p was deregulated early during acute MI [32].
C-X3-C motif chemokine receptor 1 is associated with the prevalence of coronary heart disease or MI [33]. In this study, we found that CX3CR1 was under the regulation of miR-27a. Romaine et al. [34] found that miR-27a had a high specificity in predicting the occurrence of left ventricular failure 6 months after acute MI. It is worth mentioning that miR-27a is the prognostic indicator for acute MI [35]. In this study, we found that both miR-27a and target gene CX3CR1 had a great diagnostic value for MI. Aquaporin 9 is a gap junction network gene that is vital to heart function. It is reported that AQP9 is also related to acute MI [36]. Kuiper et al. [37] found that TFEB was expressed in the myocardium of the adult. In this study, we found that AQP9 and TFEB were under the regulation of miR-330-3p in blood of MI. It is noted that miR-330-3p is up-regulated in heart but down-regulated in the plasma of patients with chronic heart failure [38].
In addition, we also found several DEGs (such as IRS2, GRB2, FASLG, and LIMS1) with a high degree in the PPI network. IRS2 plays a key role in cardiac homeostasis regulation [39]. Zawada et al. [40] also found that GRB2 plays an important role in the signaling pathway for cardiac hypertrophy. Herein, we found both IRS2 and GRB2 were regulated by miR-200a. It is demonstrated that miR-200a is involved in the cardiovascular differentiation [41].
Fas ligand (also called TNFSF6) is a member of the TNF family and the main activator of the extrinsic apoptotic pathway that binds the TNF receptor to induce apoptosis during MI [42]. LIMS1 (also called PINCH1) has been suggested to be associated with left-sided congenital heart disease [43]. In this study, FASLG and LIMS1 were under the regulation of miR-520c-3p. It is found that treatment with miR-520c will increase MMP-9 expression, which regulates remodeling of the left ventricle after MI and is tightly linked to the inflammatory response [44].
It is reported that MAPK is a key signal pathway in MI [45]. According to KEGG pathway enrichment analysis in MI, MAPK signal pathway was found covered the most DEGs, such as HSPA6 and MAPKAPK2. HSPA6 is found to be a regulated protein in planned MI patient samples [46]. MAP-KAPK2 is the substrate for p38-MAPK and less abundant in failing heart [47,48]. In the present study, we found that HSPA6 and MAPKAPK2 were regulated by miR-31* and miR-1291, respectively. It is worth mentioning that HSPA6 and miR-31* had a great diagnose value for MI. MiR-1291 has been identified as a potential diagnosis biomarker for acute MI [49]. Clinically, HCM is defined in the presence of left ventricular hypertrophy in the absence of hypertension and valve disease. In this study, HCM was found to be the most enriched signal pathway, which involved several genes such as TPM3. TPM3 is found to be a regulated protein in planned MI patient samples [46]. Herein, we found that TPM3 was regulated by miR-139-5p, and both miR-139-5p and TPM3 had a great diagnose value for MI. In human autopsy samples, miR-139-5p is down-regulated earlier, within just 7 days following MI [11].
Beside miR-27a, miR-31*, and miR-139-5p, we also found that miR-204 and miR-375 had the diagnose value for MI. MiR-204 is an autophagy-modulating miRNA that was related to cardiovascular disease [50]. It is reported that the expression of miR-375 is remarkably up-regulated in heart tissue of MI and circulating miR-375 is a potential diagnostic biomarker for MI [51,52]. In this study, we found both miR-204 and miR-375 had a great diagnose value for MI.
To analyze the expression of TFEB, IRS2, GRB2, FASLG, LIMS1, CX3CR1, HSPA6, TPM3, LAT2, CEBPD, AQP9, and MAPKAPK2 in the early stage of MI and further study the association between these genes with MI recovery, the datasets of GSE29532 and GSE48060 were used for analysis. Our results showed that all these genes were differentially expressed in different blood collection points. Moreover, these genes were associated with the recovery of MI. Therefore, we inferred that these genes may be considered as biomarkers for early stages of MI, as well as for monitoring early MI recovery.
In summary, we found several differentially expressed miRNA and genes in the blood of MI. MiR-27a, miR-31*, miR-1291, miR-139-5p, miR-204, miR-375, and target genes including CX3CR1, HSPA6, and TPM3 had a great diagnose value for MI. Additionally, MAPK and HCM were important signal pathways in the development of MI. TFEB, IRS2, GRB2, FASLG, LIMS1, CX3CR1, HSPA6, TPM3, LAT2, CEBPD, AQP9, and MAPKAPK2 may regard as biomarkers in MI early stage and recovery. Our study may be helpful in understanding the pathology mechanism of MI and could provide the clues in clinical diagnose and drug design of MI. There are limitations to our study. Firstly, sample size in the qRT-PCR was small. Larger numbers of blood samples are needed for further research. Secondly, some animal models and cell culture experiments are needed to validate and explore the potential function of identified differentially expressed miRNA and genes in MI.