Identification of key genes in sepsis-induced cardiomyopathy based on integrated bioinformatical analysis and experiments in vitro and in vivo

Data sources

Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) database (Clough & Barrett, 2016) was used to download the expression matrix associated with SIC. GSE171546 dataset contained 20 mice treated with cecal ligation and puncture at 0, 24, 48, and 72 h. RNA was extracted from the heart tissues, and mRNA expression was detected by Illumina NovaSeq 6000.

Data pre-processing

Firstly, the probe ID was replaced with gene symbols using the platform annotation information table. Genes with same symbol were incorporated. Expression matrixes samples were obtained according to ID. We use normalize between arrays in the limma (Smyth, 2005) package to standardize the quartile of the obtained chip expression data. Then, the ID of multiple expressions of the same gene was calculated as the average expression amount of the gene.

Timing analysis

Time series transcriptome data reflect the expression of genes at different moments. It can more accurately depict the gene expression levels of different biological processes and stages of the same biological process, which is more in line with the actual situation and is of great significance for studying the dynamics and diversity of biological processes. Mfuzz (Kumar & Futschik, 2007) clusters time series data based on a fuzzy clustering algorithm. Continuous up and down-regulation genes were selected for subsequent analysis according to the results of time series analysis.

Protein-protein interaction (PPI) network forecast

Hub genes in selected clusters were analyzed using String (Mering et al., 2003) online tool for PPI analysis. A combined score greater than 0.9 is chosen as the threshold for protein-protein interaction. Based on the adequate protein-protein interaction (PPI) relationship pairs, Cytoscape (Shannon et al., 2003) was then selected to perform the topology of the PPI relationship network. Most biological networks obey the attribute of a scale-free network. Therefore, the important node involved in the protein interaction relationship in the PPI network, namely hub protein, can be obtained using the connectivity degree analysis in network statistics. In this article, the nodes of the interaction networks were analyzed, and the scale-free nature of the interaction protein networks found the central proteins in the network.

GO and KEGG analysis

Functional enrichment analysis based on candidate genes was conducted using the Gene Ontology Database (Gene Ontology Consortium, 2004) and KEGG Pathway Database (Kanehisa, 2002). Fisher’s exact test is used to find out which specific functional items have the greatest correlation with a set of genes. Each item corresponds to a statistical p-value to indicate significance in the analysis results. The smaller the p-value, the higher probability of having a correlation between the item and the input gene. That is, most of the genes in this group have the description function corresponding to this item.

Prediction of mRNA target miRNA

Five miRNA databases, such as miRanda (Turner, 1985), miRDB (Chen & Wang, 2020), TargetScan (Edris, 2011), and miTarBase (Chou et al., 2018), were used to predict the relationships between mRNAs and miRNAs. The miRNA-mRNA relationships retrieved from at least two databases were selected for the subsequent construction of the TF-miRNA-mRNA regulatory network. According to literature-level TF-miRNA regulation data based on different species in the TransmiR v2.0 database (http://www.cuilab.cn/transmir), relationships between miRNA and TF were found. According to the relationship among miRNA-mRNA, miRNA-TF, and mRNA-TF, cytoscape was used to construct a TF-miRNA-mRNA regulatory network.

Animal model

C57BL/6J mice weighing 18–25 g were purchased from SLAC Laboratory Animal Co., Ltd. (Shanghai, China). These mice were housed in individually ventilated cages with ad libitum access to standard laboratory chow and water. The Ethical Committee of the Affiliated Hospital of Weifang Medical University granted ethical approvals for animal experiments (#2022WFM017). The Ethical Committee ratified all surgical procedures. Briefly, these mice were firstly anesthetized by intraperitoneal injection of pentobarbital (50 mg/kg), then they were fixed in the supine position, and the skin of the operation area was routinely disinfected. Finally, these mice were randomly divided into the Sham group and the treatment group (n = 6). As control, we performed cecal isolation and abdominal closure surgery on mice of the Sham group. In the treatment group, the distal cecum of mice was opened, lapped at 0.5 cm, and closed. After operation, all mice were reared in separate cages and fed freely. CO₂ inhalation was employed at the end of the experiment, following established guidelines.

Real-time quantitative PCR

TRIzol reagent (T9424; Sigma-Aldrich, Beijing, China) was used to extract RNA from heart tissues according to the manufacturer’s protocol. Then 1 μg RNA was used to synthesize cDNA, followed by gene expression analysis on ABI 7300 qPCR system. Relative mRNA levels were determined after normalization using GAPDH or U6 as an internal control.

Western blot

Proteins were extracted from heart tissue with RIPA lysis buffer (Beyotime Biotechnology). Protein samples (60 μg) were separated by SDS-PAGE electrophoresis and transferred to PVDF membranes (Millipore). After blocking, protein on the membrane was incubated with primary antibodies such as MMP-9 (NBP2-13173; Novus Biologicals, Centennial, CO, USA), TPT1 (PA5-34503) and FTH1 (NBP1-31944, Novus Biologicals) at 4 °C overnight. Next day they wereincubated with HRP-conjugated secondary antibody. FluorChemE imager (Alpha) was used for visualization, and the expression level of specific protein was normalized to GAPDH level.

Verifying the involvement of TPT1 in the development of SIC

Hl-1 myocardial cell line of mice purchased from the American Type Culture Collection was used for further verification. In brief, cells were cultured in DMEM medium containing 15% newborn bovine serum at the density of 3 × 10⁶ cells per bottle. Cells were firstly inoculated in culture bottles with a base area of 75 cm², then the experiment was carried out after 24 h, in this step cells should culture in DMEM medium without newborn bovine serum. Lipopolysacchride (LPS, 25 mg·L⁻¹) was used to induced myocardial cell injury model for 24 h. The protein expression levels of TPT1 and GAPDH were detected by western blot at 6, 12 and 24 h after LPS treatment.

Another experiment was performed to evaluate the role of TPT1 in development of SIC. The small interfering RNA for TPT1, 5′-AAGGTACCGAAAGCACAGTAA-3′ (siRNA1), or 5′-AACCATCACCTGCAGGAAACA-3′ (siRNA2) were synthesized by Jima Pharmaceutical Technology Co., Ltd. China and siRNA duplex 5′-AACCATCACTTACAAGAAACC-3′ was used as control. Hl-1 cell line was cultured in DMEM medium for 24 h and treated differently for 24 h; LPS group: Hl-1 myocardial cell line was treated with LPS (25 mg·L⁻¹); si-TPT1 group: Hl-1 cell line with TPT1 silencing was treated with LPS (25 mg·L⁻¹); LPS+NC group: Hl-1 cell line was cultured in DMEM medium with LPS (25 mg·L⁻¹). Then, the apoptotic cells were detected using TdT-mediated dUTP Nick-End Labeling (TUNEL). Flow cytometry was used to detect the cell cycle characteristics from the four groups.

Statistical analysis

Statistical analysis was performed using a Mann-Whitney tests, unpaired two-tailed Student’s t-test or one-way ANOVA. Statistical significance is denoted as p value smaller than 0.05. In all graphs standard error of the mean (SEM) is calculated and error bars are plotted according to mean ± SEM.

Continuous up-regulation and down-regulation gene screening

Mfuzz was used to cluster time-series data based on a fuzzy clustering algorithm. K-nearest neighbor weight algorithm (KNNW) was selected for the algorithm. The standard deviation of gene screening was 0.25, and the neighborhood coefficient membership was 0.5. Finally, six gene clusters were obtained, and genes in cluster1 and cluster3 were selected for further analysis (Fig. 1). In cluster1, 221 genes showed a continuous down-regulation expression trend. Among them, the first five genes in coefficient membership were Pm20d2, Ybx2, Hacd1, Clasp1, and Ecrg4 (Table 1). As for cluster3, 342 up-regulation genes were obtained, which contained Mmp8, Rps13, Rps15a, Plp2, and Vps51.

Functional analysis of the genes in cluster1 and cluster3

A total of 563 genes of the continuous up-regulated and down-regulated clusters were selected to perform the GO and KEGG analysis. Figure 2A and Table 2 revealed that the genes mainly enriched in several biological processes (BP) terms such as Translation, Peptide biosynthetic process, Peptide metabolic process, Amide biosynthetic process and Gene expression. As for cellular component (CC) terms, the cytosolic ribosome, ribosome, cytosolic large ribosomal subunit, large ribosomal subunit and cytosolic small ribosomal subunit were shown in Fig. 2B. Figure 3C showed the significant enriched molecular function (MF) terms, such as rRNA binding, RNA binding, 5S_rRNA binding, ubiquitin-protein transferase regulator activity and translation factor activity RNA binding.

KEGG analysis was also analyzed and the results were shown in Fig. 2D. Ribosome, lysosome, glycolysis/gluconeogenesis, valine leucine and isoleucine degradation, and carbon metabolism pathways were focused on.

PPI network analysis

PPI network analysis utilizing the string database and a combined score threshold of 0.9 was used to identify SIC hub genes. As Fig. 3 shown, Rps9, Rps15a, Rps27, Rps3, Rps5 and Rps6 had the highest connectivity with 66 and 65 other proteins, respectively. In addition, complex interactions among Rps13, Rps14, Rps23 and Rps25 were also found, indicating potential functional groups for SIC. The list of the top 20 genes of the PPI network is shown in Table 3.

Statistics of the target miRNAs of the selected genes

The continuously up-regulated and down-regulated genes were selected to predict the target miRNAs using five miRNA databases. As Fig. 4A shown, 155 miRNA-mRNA pairs were found in miTareBase and miRanda databases, and 1,340 miRNA-mRNA pairs were in miRDB and TargetScan. A total of 2,900 miRNA-mRNA pairs verified at least two datasets were selected for further analysis. Next, the miRNAs were used to predict the TF-miRNA regulated pairs using TransmiR v2.0 database and 2,763 TF-miRNA pairs were obtained. Furthermore, the Trrust database was used to predict the TF-mRNA pairs, generating a total of 109 TF-mRNA pairs.

Regulatory network of TF-miRNA-mRNA

By predicting the regulatory relationship of mRNA and miRNA, a network with 143 regulatory relation pairs was conducted (Fig. 5). Among the down-regulated genes, Ptp4a1, Pth and Nt5e owned the most pairs of regulatory relationships, which were 26, 10 and 10, respectively. As for up-regulated genes, Fam126b, Ebf1, St3gal6 and Hmgn3 interacted with 16, nine, seven and seven miRNAs, respectively. According to the statistical data, miR-294 regulated three genes simultaneously, such as Ddhd1, Cxadr, and Ptp4a1. According to the relationship among miRNA-mRNA, TF-miRNA and TF-mRNA above, the TF-miRNA-mRNA regulatory network, which contained three subnetworks, was conducted, as Fig. 4B shown. The left subnetworks indicated that the TF Rela directly regulated the expression of Fth1 and indirectly regulated Fth1 by miR-150-3p. In the middle subnetwork, TF Smad3 promoted the expression of Mmp9 and Tpt1 or accomplished it indirectly by regulating miR-323-5p and miR-654-5p, respectively. Gene Bcl2l1 was related to miR-342-3p. miR-342-5p and TF Stat6.

In addition, the TF-mRNA regulatory pairs were also selected for PPI network analysis. As shown in Fig. 5A, the TF Tpt1 promoted the expression of 69 genes, mainly belonging to the Rps and Rpl gene families. The up-regulated expression of Mmp8, Ctsd and Ctss was regulated by Mmp9 and Fth1 (Fig. 5B).

Validation of the hub genes by qPCR and western blot

To validate the accuracy of the omics data and the hub genes in PPI networks, 12 adult mice were treated with laparotomy (Sham group, n = 6) and cecal ligation and puncture (T group, n = 6). Firstly, qPCR of three hub genes like Tpt1, Mmp9 and Fth1 were performed between SIC and healthy samples. As shown in Fig. 6A, the high expression of Tpt1, Mmp9 and Fth1 in SIC samples were detected, consistent with our transcriptome data. The high protein expression level of TPT1, MMP9 and FTH1 had been verified again in the western blot experiment, which was consistent with the qPCR results (Fig. 6B).

TPT1 is involved in the development of SIC

Hl-1 myocardial cell line of mice was induced by LPS (25 mg·L⁻¹) for 24 h to detect the protein expression. Western blot results (Fig. 7A) showed that the expression of TPT1 protein increased significantly in the process of LPS-induced SIC, indicating that TPT1 may be positively correlated with the development of SIC. Cell cycle flow cytometry showed that LPS significantly inhibited cells in G2/M phase, which was reversed by TPT1 silencing (Fig. 7B).

In order to detect the effect of TPT1 on cell apoptosis, TUNEL was performed based on the HL-1 cells induced by LPS. Results showed that the LPS group significantly increased apoptosis, while the si-TPT1 group reversed apoptosis (Fig. 8).