Introduction

Diabetic cardiomyopathy (DbCM) is a specific type of cardiomyopathy, that is independent of hypertension, coronary artery disease and valvular heart disease. It is a primary lesion of the myocardium caused by chronic hyperglycemia. DbCM is initially characterized by diastolic dysfunction, increased atrial filling, prolonged isovolumic diastolic time, increased superior ventricular premature beat, increased left ventricular end-diastolic pressure, and decreased left ventricular end-diastolic volume, resulting in decreased left ventricular ejection fraction and cardiac output. As the disease progresses, patients experience systolic dysfunction, cardiac remodelling, myocardial fibrosis, myocardial hypertrophy, arrhythmia, and cardiac hypertrophy and finally progress to chronic heart failure [1,2,3]. There are still some difficulties in the clinical diagnosis of diabetic cardiomyopathy. Some researchers have proposed new biomarkers, such as glycosylated hemoglobin, glucose, low density lipoprotein, insulin-like growth factor binding protein 7 and transforming growth factor β1, which can help clinical diagnosis [4]. The pathogenesis of DbCM is very complicated and mainly involves disorders of glucose and lipid metabolism, oxidative stress, apoptosis and insulin resistance [2], but the key genes and molecular mechanisms have not been fully elucidated.

Disease models are essential for understanding cardiovascular pathogenesis and developing new therapeutic approaches. At present, research on DbCM focuses on cell and animal models, and human studies are lacking due to the particularity of heart samples. In recent years, human-induced pluripotent stem cell-derived cardiomyocytes (hiPSCs-CMs) have provided an excellent experimental model for cardiovascular diseases, such as arrhythmogenic cardiomyopathy [5] and hypertrophic cardiomyopathy [6]. Since these cells express cardiomyocyte-specific markers, they present typical electrophysiological characteristics of cardiomyocytes and respond to catecholamines and cholinergic drugs. Thus, hiPSCs-CMs have often been used to build cell models of cardiomyopathy. Some studies have reported that hiPSC-CMs develop insulin resistance under specific culture conditions, indicating that hiPSC-CMs have great potential in DbCM research [7].

In this study, differentially expressed genes (DEGs) between DbCM and control samples were first identified from NCBI Gene Expression Omnibus (GEO) datasets (GSE62203 and GSE197850), and the overlapping DEGs were reserved for downstream analysis. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed to explore the biological processes and molecular pathways of DEGs. Protein–protein interaction (PPI) networks were constructed to identify the hub genes among the DEGs. Overall, this study will provide new molecular targets and theoretical evidence for clinical treatment of DbCM.

Materials and methods

General information

hiPSCs-CMs datasets related to DbCM were collected from the NCBI GEO repository with accession numbers GSE62203 and GSE197850.

Differentially expressed genes (DEGs) identification

DEGs between the disease model and control groups from the GSE197850 dataset were directly downloaded, while the DEGs in GSE62203 were screened by using GEO2R. Genes with P < 0.05 and absolute fold change (FC) > 2 were considered significantly differentially expressed. The DEGs that were consistently up or downregulated in both datasets were used for further analysis.

Functional and pathway enrichment analysis

To investigate the role of the DEGs, GO and KEGG signalling pathway enrichment analyses were performed by using the DAVID online database. The GO analysis consisted of biological processes, cellular components and molecular functions. Terms with P < 0.05 were considered significantly enriched.

Protein–protein interaction (PPI) network construction

PPI networks of the common DEGs were constructed through STRING. The minimum needed interaction score was 0.4 and the retained PPI networks were visualized by Cytoscape.

Hub gene identification

The PPI networks from the STRING database were imported into Cytoscape, and the MNC algorithm was used to identify hub genes.

Evaluation of hub genes

ROC curves of hub genes were generated using GraphPad Prism 8.0. To estimate the specificity and sensitivity of hub genes for the diagnosis of DbCM, we performed receiver operating characteristic (ROC) analysis of each hub gene. The area under the curve (AUC) value was calculated to quantify the diagnostic performance. Genes with AUC values greater than 0.7 were considered as potential diagnostic indicators of DbCM. A two-sample t-test was used to compare the gene expression levels of DbCM and normal samples.

Cell culture

The human myocardial cell line AC16 was used in this study and was maintained in DMEM (Gibco), supplemented with 20% foetal bovine serum (Pricella), 100 U/ml penicillin (Solarbio) and 0.1 mg/ml streptomycin (Solarbio) at 37 ℃ in an atmosphere of 5% CO2. To establish a diabetic cardiomyopathy model, cells were treated with different conditions of (a) 25 mM glucose and 1% DMSO; (b) 30 mM glucose and 1 μM hydrocortisone (MedChemExpress) and cultured for 24 h.

RNA isolation and quantitative real-time polymerase chain reaction (qRT-PCR) analysis

Total RNA was purified using TRIzol reagent according to the manufacturer’s instructions (Invitrogen) and processed for cDNA synthesis using the HiScript Q RT SuperMix kit (Vazyme). cDNA was analysed by qRT-PCR with ChamQ Universal SYBR qPCR Master Mix (Vazyme) using the LightCycler 480 qPCR analysis system (Roche). The mRNA expression levels of target genes were normalized to GAPDH. The primers used in our study are listed in the Supplementary Table.

Statistical analysis

Using GAPDH as the housekeeping gene, we calculated the relative changes in the expression level of target genes in DbCM AC16 cells compared with the control group by 2-△△Ct, normalized the control group, repeated the experiment for six times, and detected the difference of components by unpaired t-test. P < 0.05 was considered to be statistically significant.

Results

The co-DEGs of the GSE62203 and GSE197850 datasets

A total of 364 DEGs, including 188 upregulated and 176 downregulated genes, were identified in the GSE62203 dataset. A total of 1400 DEGs, including 569 upregulated and 831 downregulated genes (Table 1), were directly obtained from GSE197850. A total of 78 DEGs were identified in these two gene expression profiles (Fig. 1a). Of these 73 genes were differentially expressed simultaneously (called co-DEGs), including 47 upregulated genes (64.38%) and 26 downregulated genes (35.62%), and were retained for downstream analysis (Fig. 1b). The expression levels of co-DEGs in GSE197850 and GSE62203 are shown in Fig. 1c and Fig. 1d, respectively.

Table 1 The number of upregulated and downregulated genes in GSE62203 and GSE197850
Full size table
Fig. 1
figure 1

DEGs in GSE62203 and GSE197850. (a) Venn diagram of the overlapping DEGs in GSE62203 and GSE197850, (b) Pie chart show the proportions of upregulated and downregulated genes shared in GSE62203 and GSE197850, (c, d) Heat map of the expression levels of 73 co-DEGs in the GSE197850 (c) and GSE62203 (d)

Full size image

GO and KEGG pathway enrichment analysis of co-DEGs

To explore the function of co-DEGs, GO and KEGG signalling pathway enrichment analyses were performed on 73 co-DEGs using the DAVID online database. Ninety-six terms were obtained from GO analysis. In the cellular components category, the DEGs were mainly enriched in cell surface, exosomes, mitochondria, endoplasmic reticulum and cytoplasm (Fig. 2a). In the biological processes category, the DEGs were involved in hypoxia response, glycolipid metabolism, cell proliferation and growth, apoptosis and myocardial contraction regulation (Fig. 2b); In the molecular functions category, the DEGs were mainly enriched in iron ion binding and hormone activity (Fig. 2c). KEGG analysis showed that the DEGs were mainly enriched in the HIF signalling pathway, amino acid anabolism and carbon metabolism (Fig. 2d).

Fig. 2
figure 2

GO and KEGG signalling pathway enrichment results. (a) Cellular components analysis of co-DEGs, (b) Biological processes analysis of co-DEGs, (c) Molecular functions analysis of co-DEGs and (d) KEGG signalling pathway analysis of co-DEGs

Full size image

PPI network construction

The online tool STRING was used to construct PPI networks of co-DEGs. In the network, there were 69 nodes and 176 edges. The average node degree was 5.1 and the local clustering coefficient was 0.507, which indicates that the PPI enrichment was statistically significant (Fig. 3).

Fig. 3
figure 3

PPI networks of co-DEGs

Full size image

Hub gene identification

The hub genes in the PPI networks were identified by CytoHubba (Fig. 4a). The top 10 hub genes including LDHA, PGK1, SLC2A1, ENO1, PFKFB3, EGLN1, MYC, PDK1, EGLN3 and BNIP3 were determined by the MNC algorithm (Fig. 4b and Table 2).

Fig. 4
figure 4

Protein interaction map of the top 10 hub genes. (a) A PPI network of co-DEGs constructed using Cytoscape and STRING, (b) The top 10 hub genes were identified by CytoHubba using MNC algorithm

Full size image
Table 2 The top 10 hub genes in DbCM
Full size table

Evaluation of hub genes

In the GSE62203 and GSE197850 datasets, except for the MYC gene, the mRNA expression levels of the hub genes were higher in the DbCM group compared to the control group.. To further explore the diagnostic value of these 10 genes in DbCM, the area under the curve (AUC) of receiver operating characteristic (ROC) was calculated to judge the diagnostic value of the 10 hub genes. The AUC of each hub gene was greater than 0.7, indicating that 10 hub genes might play a role in clinical diagnosis of DbCM (Fig. 5a and 6a).

Fig. 5
figure 5

Validation of hub genes in the GSE62203 dataset. ROC analysis of hub genes in DbCM

Full size image
Fig. 6
figure 6

Validation of hub genes in the GSE197850 dataset. ROC analysis of hub genes in DbCM

Full size image

Validation of the expression levels of hub genes in DbCM by qRT-PCR

To further verify the expression levels of the top 10 hub genes in DbCM, we established an in vitro model of DbCM by using the AC16 cell line and performed qRT-PCR to determine their mRNA expression levels. The results showed that the mRNA expression levels of the PFKFB3, MYC, EGLN3, EGLN1, BNIP3, and PGK1 genes were significantly increased in the DbCM group compared with the control group, which was consistent with the above results. In contrast, the mRNA expression levels of ENO1 was significantly decreased in the DbCM group. However, the mRNA expression levels of LDHA and SLC2A1 were slightly increased in the DbCM group compared with the control group, but this difference was no significant (Fig. 7a, b).

Fig. 7
figure 7

mRNA expression levels of the hub genes were quantified by qRT-PCR. (a) Agarose gel electrophoresis image of qRT-PCR product. Each band represented an individual sample. (b) Bar charts depicting the expression of hub genes in DbCM and Control group

Full size image

Discussion

Diabetic cardiomyopathy is one of the most common vascular complications in patients with diabetes. According to a survey of the 2021 International Diabetes Federation, cardiovascular complications of diabetes including DbCM are the most critical factors causing morbidity and mortality in patients with diabetes. The pathogenesis of DbCM mainly involves disorders of glucose and lipid metabolism, oxidative stress, endothelial cell dysfunction, inflammation and apoptosis. Thus, it is complex and has not been fully elucidated. Therefore, it is critical to find the key therapeutic target for DbCM. In this study, we analysed the DEGs in two DbCM datasets using hiPSCs-CMs to construct DbCM models, and performed GO and KEGG signalling pathway enrichment analyses. A total of 10 hub genes that may play important roles in DbCM were screened.

Establishing a cell model that can simulate the pathophysiological state of the human body is an important method in medical research. Although neonatal rat cardiomyocytes, the rat cardiomyocyte cell line H9c2, and the mouse cardiomyocyte cell line HL-1 are frequently employed in cardiovascular disease research [8], rat, mouse and human hearts differ greatly in terms of gene expression and energy metabolism during disease states [9]. Therefore, it is more reliable to construct a DbCM model using human-derived cardiomyocytes. Human induced pluripotent stem cells (hiPSCs) exhibit the properties of self-renewal and pluripotency. Moreover, cardiomyocytes differentiated from hiPSCs have great advantages in the field of cardiovascular diseases and have been widely used in the past decade. Therefore, the datasets of the diabetic cardiomyopathy model constructed by hiPSCs-CMs were selected for this study.

The expression levels of 10 DEGs in the two datasets were consistent, and the AUC was greater than 0.7, indicating high diagnostic value. LDHA is a subtype of lactate dehydrogenase, which is one of the key enzymes in glycolysis, and can catalyse pyruvate to lactate. Studies have shown that increased LDHA activity can promote lactate levels and lactate-pyruvate conversion in patients with diabetes [10]. However, the expression levels of LDHA were reduced in the hearts of STZ-induced diabetic mice, possibly due to the inhibition of glycolysis [11]. Interestingly, we verified changes in the mRNA levels of LDHA by qRT-PCR and found no statistically significant difference between the DbCM and control groups. This may be due to the different maturation of the cell lines. SLC2A1, also known as GLUT1, is involved in the regulation of glycometabolism, and its expression levels were increased in hiPSCs-CMs stimulated by high glucose, which is consistent with the results of our study. A previous study revealed that empagliflozin induced cardiac dysfunction by upregulating the expression of SLC2A1 [12]. Fructose-2-kinase 6-phosphate/fructose-2, 6-diphosphatase 3 (PFKFB3), regulates the rate of glycolysis by controlling the levels of fructose-2, 6-diphosphate (F-2,6-BP), which is believed to be an effective allosteric activator of fructose-1-kinase 6-phosphate (PFK1). PFK1 is a key rate-limiting enzyme in glycolysis that catalyses the conversion of F-6-P to F-1, 6-BP and is considered to be a major rate-limiting step in glycolytic flux [13]. In astrocytes, PFKFB3 is maintained a low level through proteasome-mediated ubiquitination degradation, ensuring the predominance of the pentose phosphate pathway in glucose metabolism. Therefore, the level of NADPH is increased, which contributes to oxidative damage repair [14, 15]. The mRNA expression level of PFKFB3 is upregulated, the rate of glycolysis is increased, and the inflammatory response is enhanced in endothelial cells in a high glucose environment [16, 17]. Consistent with previous studies, we found that the mRNA expression level of PFKFB3 was elevated in the DbCM group, which was consistent with previous studies. Therefore, we speculate that PFKFB3 may play a role in the pathogenesis of DbCM. PGK1 and ENO1 are also key enzymes for glycolysis. PGK1 is an oncogene in cancer occurrence and progression. It can promote cell growth and proliferation [18]. Additionally, PGK1 is a key metabolic enzyme in the glycolysis process. It functions to catalyze the conversion of 1, 3-diphosphoglyceric acid (1,3-BPG) to 3-phosphoglyceric acid (3-PG) and produces the first ATP in the glycolysis pathway [19]. Our study first identified elevated PGK1 expression levels in DbCM. This may provide a new molecular target for DbCM research. Similarly, ENO1 is also involved in the processes of division, proliferation, metastasis and apoptosis of tumour cells [20]. The results of bioinformatics analysis showed that the expression level of ENO1 in DbCM was increased, but in vitro experiments showed that the expression level of ENO1 in the DbCM model was decreased. The role of ENO1 in DbCM may need further clarification in the future. Therefore, LDHA, SLC2A1, PFKFB3, PGK1 and ENO1 may be potential therapeutic targets for DbCM, especially PGK1 and ENO1.

Prolyl hydroxylase (EGLN) oxygen-dependent hydroxylation of specific proline residues in hypoxia-inducible factor subunits participates in the oxygen-sensing pathway. EGLN1 and EGLN3 are two important members of the EGLN protein family that have been reported to play an important role in the pathogenesis and progression of DbCM. EGLN1, also known as PHD2, was significantly increased in the heart tissue of db/db mice. Overexpression of angiotensin-1 may rescue the impairment of angiogenesis in diabetes by inhibiting PHD2 expression [21]. The expression levels of EGLN3, also known as PHD3, were significantly increased in DbCM rats. Knockdown of EGLN3 can improve cardiac function and inhibit fibrosis and apoptosis in DbCM rats. The expression levels of EGLN3 were significantly increased when H9c2 cells were exposed to high glucose conditions. The decrease in EGLN3 levels could inhibit the increase in apoptosis caused by high glucose conditions [22]. Similarly, the results of GO and KEGG analyses of DEGs in our study showed that EGLN1 and EGLN3 play important roles in DbCM by participating in the cellular hypoxia response and HIF-1 signalling pathway. Therefore, our study confirmed the increased expression of EGLN1 and EGLN3 in DbCM, supporting the conclusions of previous studies.

Apoptosis is one of the key molecular mechanisms in the pathogenesis and progression of DbCM. Bcl-2 interacting protein 3 (BNIP3) is a pro-apoptotic protein, and the levels of BNIP3 were significantly increased in DbCM rat hearts. This funding is consistent with the conclusions of our study. The increased expression levels of BNIP3 in H9c2 cells induced by high glucose stimulation can lead to increased apoptosis. This indicates that BNIP3, one of the important molecules of apoptosis, is involved in the progression of DbCM [23]. MYC is a proto-oncogene that expresses a nuclear phosphorylated protein and plays an important role in the cell cycle and apoptosis. Studies have shown that the expression level of c-MYC in the heart of T2D rats was increased [24]. Interestingly, after high glucose stimulation in H9c2 cardiomyocytes, c-myc protein expression levels were significantly decreased and the cell proliferation cycle was arrested [25]. In our study, the mRNA expression levels of MYC were also significantly increased, possibly due to its involvement in apoptosis via the Wnt/ beta-catenin pathway [26]. These seemingly contradictory results may be mainly due to different mouse strains and different disease stages simulated by the two disease models. The C-myc protein is involved in various cellular pathophysiological processes and can promote the pathogenesis of DbCM. Thus, it may be an important molecular target for further study in the future. PDK1, a serine/threonine protein kinase, inhibits apoptosis by increasing the phosphorylation of AKT. Hou and Lin et al. discovered that the phosphorylation of PDK was decreased in DM rats and mice [15, 27]. However, the low expression level of PDK1 in AC16 makes it difficult to detect PDK1 mRNA by qRT-PCR methods.

There are some limitations of this study. First, we used AC16 cells to verify the expression of hub genes in our study. Although AC16 is a human cardiac cell line and is widely used in human cardiology research, AC16 still differs from hiPSCs-CMs in terms of maturity and gene expression. Second, only 16 samples from both datasets were used in our study for analysis. In the future, we will expand the sample size for further analysis. Third, PDK1 expression was low in AC16 cells, and thus, its role in DbCM cannot be verified. In the future, PDK1 overexpressing AC16 cells could be used to confirm the expression changes of PDK1 in the diabetic state.

Conclusions

In this study, 73 genes with altered expression in DbCM were screened and were mainly found to be enriched in glucose and lipid metabolism disorders, hypoxia response, cell apoptosis, cell proliferation regulation, and cytoplasmic and HIF signalling pathways. Furthermore, we identified 10 hub genes including LDHA, PGK1, SLC2A1, ENO1, PFKFB3, EGLN1, MYC, PDK1, EGLN3 and BNIP3 from the DEGs. In vitro studies, we found that the mRNA expression levels of PGK1, PFKFB3, MYC, EGLN1, EGLN3 and BNIP3 were upregulated, LDHA and SLC2A1 were unchanged, and ENO1 was downregulated in the DbCM group. Our study first found PGK1 and ENO1 were involved in DbCM, and the effect of PDK1 and MYC in DbCM needs to be verified and clarified. Our study provided new molecular targets and theoretical basis for the clinical diagnosis of DbCM, and offered a new perspective for future DbCM research.