Construction and analysis for differentially expressed long non-coding RNAs and mRNAs in acute myocardial infarction.

Long noncoding RNAs (lncRNAs) are transcripts longer than 200 nucleotides. Some lncRNAs are related to acute myocardial infarction (AMI) and can serve as blood-based biomarkers for AMI detection. To identify whether new lncRNAs participate in AMI, the expression of lncRNAs and mRNAs was analysed by microarray analysis (Agilent human array) with the limma package in R in two series: five paired peripheral blood mononuclear cell (PBMC) samples and four paired plasma samples from different AMI patients. In PBMCs, a total of 2677 upregulated and 458 downregulated lncRNAs were significantly differentially expressed; additionally, 1168 mRNAs were upregulated and 1334 mRNAs were downregulated between the AMI patients and controls. In plasma, we found 41 upregulated and 51 downregulated lncRNAs that were differentially expressed, as well as 9 mRNAs that were upregulated and 9 mRNAs that were downregulated among the two groups. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were performed using the clusterProfiler package in R, and differentially expressed mRNAs were functionally annotated. The top differentially expressed mRNAs were associated with circadian rhythm, the NF-kB pathway, the p53 pathway and the metabolism pathway. We further performed target gene prediction and coexpression analysis and revealed the interrelationships among the significantly differentially expressed lncRNAs and mRNAs. The expression of four lncRNAs (uc002ddj.1, NR_047662, ENST00000581794.1 and ENST00000509938.1) was validated in the newly diagnosed AMI and control groups by quantitative real-time PCR (qRT-PCR). Our study demonstrated that the clustered expression of lncRNAs between PBMCs and plasma showed tremendous differences. The newly screened lncRNAs may play indispensable roles in the development of AMI, although their biological functions need to be further validated.

Patients and sample collection. The diagnostic criteria of AMI were made in accordance with the 2016 American College of Cardiology (ACC)/American Heart Association (AHA) Guideline. All AMI patients were recruited from the Department of Cardiology, the First Affiliated Hospital of Xinjiang Medical University, from December 2014 to September 2017, and chest pain began within 12 h before undergoing percutaneous coronary intervention (PCI). Sex-and age-matched controls were consecutively recruited from the Health Physical Examination Centre of our hospital, and they had no evidence of cardiovascular or other systemic disease based on the records of physical examination, ECG and blood test. For all the subjects, the exclusion criteria were as follows: (i) thrombolysis therapy and other heart diseases (such as congenital heart disease, cardiomyopathy or severe valvular abnormalities); (ii) active inflammation; and (iii) renal or hepatic dysfunction, autoimmune diseases and tumours. A cohort of 59 AMI patients and 59 controls was finally enrolled in this study. To profile lncR-NAs and mRNAs for microarray analysis in PBMCs, 5 AMI cases and 5 matched controls were randomly chosen and coded as PBMC-A1, -A2, -A3, -A4, and -A5 and PBMC-N1, -N2, -N3, -N4, and -N5. The other 4 AMI cases and 4 matched controls for chip analysis of plasma were randomly selected and coded as Plasma-A6, -A7, -A8, and -A9 and Plasma-N6, -N7, -N8, and -N9. The validation of dysregulated lncRNAs for microarray analysis was then analysed in the cohort of AMI patients (n = 50) and controls (n = 50). Blood samples (5 ml) collected from the peripheral veins of the patients were placed into test tubes containing EDTA before the administration of any anticoagulants.

Collection and purification of mononuclear cells. Blood samples (including PBMCs and plasma) were
collected within 2 hours after harvest, centrifuged at 3500 rpm for 10 min, and then composed of two layers. The upper plasma layer was directly used for plasma RNA extraction, and the lower layer was used for the lysis reaction. After mixing equal volumes of the lower blood cell layer and saline, we added Lymphocyte Separation Medium (TBD, Tianjin Biotechnology, Tianjin, China), which accounted for twice the total blood volume, to start the lysis reaction. Then, the layer was collected into a 1.5 ml EP tube and centrifuged at 10000 rpm for 5 min; thereafter, a white membrane was used as the white cell layer. After discarding the supernatant, the pellet remained at the bottom of the tubes.
in an Agilent Hybridization Oven, a Klenow enzyme labelling strategy 19 was used, and array hybridization was denatured.
Validation of lncRNAs by qRT-PCR. Then, 11 μl of purified RNA from plasma or 1 μg of purified RNA from PBMCs was used for reverse transcription (Qiagen, China). According to the instructions of the Power SYBR Green PCR Master Mix (Applied Biosystems, USA), PCR with a 20 μl mixture (consisting of 10 µl of SYBR Green, 1 μl of primers, 7 μl of DEPC and 2.0 μl of cDNA) was performed in a 7900HT Fast Real-Time PCR system (Applied Biosystems, USA). The expression levels of lncRNAs were calculated via the comparative Ct (ΔΔCt) method and compared with those of the controls. GAPDH served as the internal control. The specificity of the PCR products was estimated by melt curve analysis. The primers for the lncRNAs in qRT-PCR are listed in Table 1.
Mapping and identification of differentially expressed genes. CLUSTER 3.0 software with an adjusted data function was used to preprocess the data, for example, by performing log2 transformation, median centering and average linkage clustering. The lncRNA and mRNA array data were normalized and summarized by GeneSpring software V13.0 (Agilent, USA). Volcano plot filtering, scatter plot filtering and hierarchical clustering can screen out the different classifications of lncRNA and mRNA expression in the form of images. The implementation of tree visualization was performed with Java Treeview (Stanford University School of Medicine, Stanford, CA, USA). GO and KEGG pathway and disease analyses. Gene Ontology (GO) analysis (http://www.geneongoloty.org/) was performed to illustrate the unique biological significance of the differentially expressed genes 20 . The potential functions of the genes were divided into three categories: cellular components, molecular functions and biological processes. Based on the current Kyoto Encyclopedia of Genes and Genomes (KEGG) database, crucial pathways and related diseases were identified by KEGG pathway (http://www.genome.jp/kegg/) and disease (http://www.genome.jp/kegg/disease/) analysis. Fisher's exact test and the chi-square test were applied to identify the significant GO terms and KEGG data, and the standard of threshold values was formulated by the false discovery rate (FDR) and p-value.
Protein-protein interaction network construction. A protein-protein interaction (PPI) network could serve as a tool to assess the interactive relationships of the differentially expressed mRNAs, and one was constructed through the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING, http://string.embl. de/) 21 database with a combined score greater than 0.4 as the cut-off criterion. In addition, GO and KEGG enrichment analyses were performed by STRING. Visualization was carried out with the reliable online bioinformatics software Cytoscape v3.5.1 (Cytoscape Consortium, New York, NY). To acquire more meaningful mRNAs, the related analysis parameters used were as follows: species: Homo sapiens; active prediction methods: all; and degree centrality greater than one was set as a cut-off criterion.
Construction of the regulatory co-expression network. The network of coding-noncoding coexpressed genes was constructed with the biological functions, especially to recognize the novel and significant genes. Correlations between lncRNAs and their corresponding mRNAs were calculated with Pearson's correlation, and coefficients not less than 0.95 were used to draw the coexpression network through Cytoscape v 3.5.1 software. The degree centrality was used to determine the links, wherein a higher degree score indicates a more central location within the network.
Target gene prediction. Since the potential functions of most lncRNAs have not been well annotated, they were predicted on account of the related cis-and trans-mRNAs. Cis-regulated target genes were selected according to the distance of 10 kb near the coding gene location, while LncTar 22 was used to assume the trans-regulated target genes mainly in view of the different expression levels. Pearson correlation coefficient >0.99 and p < 0.05 were used for target gene prediction.  www.nature.com/scientificreports www.nature.com/scientificreports/ as the mean ± standard deviation (SD) and were compared using either the Mann-Whitney U test or the Kruskal-Wallis test as appropriate. The chi-square test or Fisher's test was used for categorical variables. The relative expression levels of lncRNAs between the two groups were calculated via independent-sample t-tests. GO and KEGG pathway analyses were evaluated using Fisher's exact test. For all statistical analyses, a two-tailed p < 0.05 was regarded as significant.

Results
Study cohorts. This study recruited 59 AMI patients and 59 matched healthy controls. The expression profiles of lncRNAs and mRNAs from 9 AMI patients and 9 matched controls were detected using microarrays, and the remaining participants were used for validation. The detailed clinical and demographic characteristics of the 18 study participants are summarized in Table 2.

LncRNA and mRNA expression profiles in the PBMCs of AMI. Hierarchical clustering analysis
showed that the expression levels of lncRNAs and mRNAs were significantly different in AMI patients compared with the matched controls. Of the 41,000 lncRNAs detected on the microarray, 3135 lncRNAs were found to be differentially expressed in AMI, including 2677 upregulated and 458 downregulated lncRNAs. In the scatter plot and volcano plot, every dot represents a lncRNA, the red dots represent the upregulated lncRNAs (fold change (Fc)>2, p < 0.05), the green dots represent the downregulated lncRNAs (Fc < -2, p < 0.05), and the black dots represent the remaining lncRNAs (-2<Fc<2, p > 0.05) ( Differentially expressed lncRNAs and mRNAs in the plasma of AMI. Additionally, the locus-by-locus lncRNA probes from the plasma of AMI patients (n = 4) and matched controls (n = 4) were examined by microarrays, and all results are presented following the criteria of FDR-corrected p < 0.05 and absolute Fc>1.5. We found 41 upregulated and 51 downregulated lncRNAs after hierarchical clustering and scatter and volcano plot analysis ( Fig. 2A-C). The top 20 differentially expressed lncRNAs according to the Fc values are presented in Supplementary Table 3. We also found that 9 mRNAs were upregulated and 9 mRNAs were downregulated in AMI patients (Fig. 2D-F). The details of the 18 differentially expressed mRNAs are presented in Supplementary Table 4.
Basic characteristics of lncRNAs in the PBMCs of AMI. The basic characteristics of all differentially expressed lncRNAs and mRNAs in PBMCs, which are widely distributed in all chromosomes covering chromosomes X and Y, are shown in the Circos plot (Fig. 3A). The Venn diagram indicates the lncRNA counts and types according to the differentially regulated lncRNA categorizations. Next, we classified the distinctive PBMC lncRNAs into five categories 23 according to the genomic loci of their neighbouring genes (Fig. 3B). Although 590 lncRNAs were not successfully categorized, the well-annotated lncRNAs were classified into the following five categories: intergenic (40.42%), antisense (14.61%), intronic (8.61%), bidirectional (7.53%), and sense (0%). Moreover, the plasma lncRNAs were the most abundant in the intergenic (44.57%), sense (19.57%) and intronic (18.48%) categories versus the antisense (9.78%) and bidirectional (7.61%) categories (Fig. 3C). Functional analysis of differentially expressed genes in PBMCs in AMI. After GO analysis, we found that the most enriched GO terms were involved in endothelial cell activation, cardiolipin metabolic www.nature.com/scientificreports www.nature.com/scientificreports/ process, regulation of heart rate by cardiac conduction, cellular response to calcium ion and cardiac muscle cell action potential involved in contraction (Fig. 4A,B). KEGG pathway analysis showed that the differentially expressed mRNAs were mostly enriched in viral myocarditis, the mTOR signalling pathway 24 , circadian rhythm, arrhythmogenic right ventricular cardiomyopathy (ARVC), and the metabolism pathway 25 (Fig. 4C,D). In the disease analysis, the distinct mRNAs were enriched in several diseases, and the top 30 diseases are shown in Fig. 4E.
To intuitively clarify the links of the expressed lncRNAs with functions and genes, we then visualized the hierarchical clustering results, as shown in Fig. 4F, which presents a clearer image to view the relationships www.nature.com/scientificreports www.nature.com/scientificreports/ between the top 10 upregulated and 10 downregulated lncRNAs and pathways through the enrichment analysis of the coexpressed mRNAs. A total of 20 lncRNAs were related to the occurrence and development process of heart diseases, such as cardiac muscle and vascular smooth muscle contraction and platelet activation. The most enriched pathways corresponding to the dysregulation of lncRNAs related to heart diseases were the peroxisome proliferator-activated receptor (PPAR) signalling pathway, NF-kappa B signalling pathway, p53 signalling pathway and Janus kinase signal transducer 2 and activator of transcription (JAK/STAT) signalling pathway. www.nature.com/scientificreports www.nature.com/scientificreports/ PPI network analysis of mRNAs in the PBMCs of AMI. According to the GO and KEGG enrichment analyses of mRNAs associated with cardiovascular development and progression, almost 200 mRNAs were screened out through the STRING database and a PPI network was constructed. Visualization was performed with Cytoscape v3.5.1. As shown in Fig. 5A, several PPI nodes had high connectivity degrees: EGFR (degree=37), TGFB1 (degree=24), RAC2 (degree=23), AKT2 (degree=25), HDAC2 (degree=24), CXCL12 (degree=22), and STAT1 (degree=22).

Differences in lncRNA expression between PBMCs and plasma.
To express the level of nucleic acids more intuitively and conveniently in blood, we not only evaluated the microarray chips with PBMCs but also with www.nature.com/scientificreports www.nature.com/scientificreports/ plasma. Undoubtedly, the difference is obvious between PBMCs and plasma in terms of the quality of RNA. The abundance of RNA in PBMCs was higher than that in plasma (p < 0.05) (Fig. 6B). There were 14,130 overlapping www.nature.com/scientificreports www.nature.com/scientificreports/ lncRNAs between the two chips (Fig. 6C). Among the 14,130 total overlapping genes, 1174 lncRNAs were found to be differentially expressed in PBMCs with Fc>2 and p < 0.05, and 1085 lncRNAs were expressed in plasma.  www.nature.com/scientificreports www.nature.com/scientificreports/ Among the two levels, we found 60 lncRNAs not only expressed in plasma but also significantly differentially expressed in PBMCs. The 60 lncRNAs in AMI patients are presented in Supplementary Table 5.

Discussion
There is no doubt that lncRNAs have become important factors in biology; however, the majority of lncRNAs have not yet been researched. In this study, we described the lncRNA and mRNA expression profiles of clinical samples from 9 AMI patients and 9 controls via microarray analysis. To the best of our knowledge, this is the first study to identify differentially expressed lncRNAs in two sample types: PBMCs and plasma. Using a comprehensive analysis, we demonstrated that in AMI, not only mRNA expression but also lncRNA expression can be dysregulated. Functional coexpression analysis indicated that some of the differentially expressed lncRNAs and www.nature.com/scientificreports www.nature.com/scientificreports/ mRNAs might play crucial regulatory roles in several mechanisms previously implicated in the pathogenesis of AMI. Importantly, many of these altered lncRNAs are linked to bioprocesses associated with the inflammatory response, oxidative stress and apoptosis. Thus, we suggest that these altered lncRNAs might play indispensable roles as AMI biomarkers.
In PBMCs, a total of 2677 upregulated and 458 downregulated lncRNAs were identified to be significantly and differentially expressed; moreover, 1168 mRNAs were upregulated and 1334 mRNAs were downregulated between AMI patients and controls. In plasma, we found 41 upregulated and 51 downregulated lncRNAs that were differentially expressed, as well as 9 mRNAs that were upregulated and 9 mRNAs that were downregulated among the two groups. After analysing the data from the two chips, we obtained 14,130 overlapping lncRNAs. Among them, we found that 60 lncRNAs were differentially expressed in both PBMCs and plasma. The details of the plasma lncRNAs varied among coronary heart disease (CHD) patients, not only in our research but also in another report. Yang 14 detected 174 lncRNAs that were upregulated, whereas 91 were downregulated according to Fc>2 and p < 0.01 using the Human LncRNA Array V2.0 (8660 K, Arraystar). Similarly, several studies have investigated the relationship between lncRNAs and AMI through expression profiling in myocardial infarction (MI) mice, but the findings of novel heart-specific lncRNAs were also diverse. For example, Ounzain 26 found 143 upregulated and 291 downregulated lncRNAs; however, 301 upregulated and 341 downregulated lncRNAs were reported in another study 27 . One might wonder why this phenomenon exists, and there are some possible reasons that can explain it. In addition to the diversity of the clinical and demographic characteristics of the enrolled samples, the amount of RNA expression in plasma, PBMCs and tissue is tremendously diverse, even when the RNA quantity accords the standard of the microarray chip. Another reason is that the nucleic acid source in plasma, PBMCs and tissue differs. Unlike in plasma and tissue, nucleic acids released through ruptured cardiomyocytes are the main origin in plasma.
To date, most lncRNA functions have not been well annotated; therefore, GO and KEGG pathway and disease analyses have played indispensable roles in forecasting the functional relations among the coexpressed mRNAs and diseases. The GO analyses divided mRNAs into several functional modules, which are related to cardiac conduction, cellular response to calcium ion and cardiac muscle cell action potential involved in contraction, indicating the validity of the microarray. The KEGG analyses revealed that the mTOR signalling pathway, circadian rhythm, ARVC, the metabolism pathway, mitral valve diseases and QT interval were important pathways. All of these KEGG pathways participate in the process of AMI, providing evidence for the precision of the microarray. Then, we identified the top 20 upregulated and downregulated lncRNAs and their functional links. Corresponding to the dysregulation of lncRNAs in AMI, the most enriched pathways related to AMI were the PPAR signalling pathway, NF-kB signalling pathway, p53 signalling pathway and JAK/STAT signalling pathway. According to previous studies, it is important in cytokine regulation for PPAR 25,[28][29][30][31] to inhibit atherosclerosis and regulate glucose metabolism and lipid metabolism. Zhou et al. 31 reported that myocardial apoptosis is associated with PPAR-γ following AMI in a rat model. The NF-kappa B signalling pathway 32,33 , p53 signalling pathway 32,34 and JAK2/STAT3 signalling pathway 29,35-37 play essential roles in CHD and also have significant effects on the inflammatory response, oxidative stress and apoptosis. During ischaemic injury, Leung et al. 38 reported that the presence of CD4-expressing T cells, TNF-α and IL-10 were all in response to neoantigen inflammation. Regarding the role of the NF-kB signalling pathway, we hypothesized that its participation in the pathogenesis of AMI may also be related to CD4-expressing T cells, TNF-α and IL-10 in myocardial ischaemia and infarction. Next, we analysed the coexpression network between altered lncRNAs and mRNAs. We specifically chose the 4 lncRNAs, namely, uc002ddj.1, NR_047662.2, ENST00000509938.1, and ENST00000581794.1, based on their coexpressed mRNAs, which are related to the occurrence and development of AMI. For example, NEXN is known as an actin-binding protein that is associated with dilated cardiomyopathy 39 , hypertrophic cardiomyopathy 40 , CAD 41 and atherosclerosis 42 . León et al. 43 reported that KANSL1 microduplication, in combination with the 22q11.2 deletion, is associated with an increased risk of CHD. Several studies revealed that PKD1 is not only associated with vascular endothelial growth factor (VEGF) 44 but also exhibited differential expression in the myocardial tissue of a rat model 45 . Moreover, TAGAP 46 , AQP4 47,48 , EPHX2 49 and GNB5 50,51 all provided new insights into the molecular mechanisms through the inflammatory response and apoptosis pathways underlying cardiovascular disease regulation. However, this hypothesis should be further investigated. The STRING database adopts public data and combined with our microarray data, it was used to construct a PPI network that facilitates the creation of several subnetworks through the degree method. A PPI network was built to identify key correlated functional mRNAs by utilizing the coexpression network and GO annotations. As a result, we found 7 key mRNAs with connectivity degrees greater than 20, including EGFR, TGFB1, RAC2, AKT2, HDAC2, CXCL12 and STAT1. Most mRNAs are closely associated with positive regulation of blood vessel endothelial cell migration, mononuclear cell proliferation and chemotaxis, apoptosis, response to oxidative stress, mTOR signalling pathway and JAK/STAT signalling pathway. Similarly, the results need further research for hypothesis formulation and validation.
Currently, AMI exerts substantial burdens on society, and the diagnostic and prognostic cornerstones of the clinical assessment of AMI often have uncertainty 52,53 . It is urgent to develop more easily accessible and highly accurate blood-based biomarkers for the early detection of AMI. PBMCs mainly consist of monocytes and lymphocytes, which are important groups of cells in the innate immune system. A hallmark of the onset and development of CAD is the continuous and increasing monocyte adhesion to the endothelium. Previous studies have shown a pivotal role for PBMCs in the pathogenesis of CAD and atherosclerotic plaque progression 54 . In our study, it is also worth mentioning that the expression levels of the lncRNAs uc002ddj.1 and NR_047662 were decreased and those of ENST00000509938.1 and ENST00000581794.1 were increased in the setting of AMI, which identified previously non-reported novel lncRNAs with altered expression. Through further analysis, we found that the changes in the Fc direction of lncRNA expression were maintained but less than those in the microarray data. We considered that the reasons for the difference were as follows. First, clinical and individual (2020) 10:6989 | https://doi.org/10.1038/s41598-020-63840-9 www.nature.com/scientificreports www.nature.com/scientificreports/ differences must be considered. Second, the methods were slightly different. The microarray was performed according to the Agilent Gene Expression Analysis protocol, and qPCR validation was performed using the SYBR Green PCR Master Mix method. Last, the trend of changes was consistent, which was in accordance with the findings of other reports 23 . Increasingly, studies have proven that the stability of plasma miRNAs is primarily attributed to their packaging in microvesicles or the formation of complexes with miRNA-binding proteins 55,56 . It is not well known whether lncRNAs have similar properties to these described miRNAs. Nevertheless, increasing emphasis has been placed on lncRNAs that can serve as biomarkers in plasma, e.g., lncRNA H19 57 for gastric cancer, lncRNA HULC 58 for HCC, lncRNA CoroMarker for AMI 14 and lncRNA LIPCAR 8 for heart failure after MI. The differential lncRNAs identified in plasma still need to be further validated by large samples.
We must address some limitations of our study. First, data reliability is limited due to the small sample size for the microarrays. Additionally, we focused on predicting differentially expressed lncRNA and mRNA functions, not determining the molecular mechanism of these lncRNAs. Further experiments are needed. Finally, all participants in this study were from one hospital, and divergences between different regions and nationalities are inevitable. Therefore, validation should be carried out in more prospective and larger clinical trials.

conclusion
Our study analysed and elaborated the expression profiles of lncRNAs and mRNAs that previously have unreported bioinformation in PBMCs and plasma between AMI and matched controls using microarrays. The dysregulated lncRNAs and mRNAs were defined and screened out with the development of high-throughput microarrays, and the application of bioinformatics provided new insights into researching the underlying molecular mechanisms in AMI. In the future, on one aspect, it is necessary to validate the expression of the identified key lncRNAs and mRNAs in a multi-centre and larger-scale study. On the other hand, to understand the pathological mechanism of AMI, we need to further investigate the biological significance of AMI in animal or cell line models. As such, special lncRNAs and mRNAs can serve as diagnostic, prognostic and therapeutic biomarkers in AMI.

Data availability
All data relevant to the study are included in the article or uploaded as supplementary information. No additional data are available.