Identification of Hypoxia-Related Genes in Human Myocardial Infarction and Cancers

Abstract

Introduction

Myocardial infarction (MI) and cancer collectively account for over 50% of global mortality. Recent studies have revealed multiple associations between these two diseases, including chronic inflammation and oxidative stress, with particular focus on hypoxia-mediated signaling pathways. In ischemic myocardium, oxygen deprivation triggers apoptosis, fibrosis, and pathological tissue reorganization; in the tumor microenvironment (TME), hypoxia drives angiogenesis, metabolic reprogramming, and immune evasion. Thus, identifying differentially expressed genes related to hypoxia in MI may provide new targets for the treatment of MI and cancer.

Methods

The specimens from MI patients in this study were retrieved from the Gene Expression Omnibus (GEO) database. Using the “limma” package in R and weighted gene co-expression network analysis (WGCNA), a set of hypoxia-related differentially expressed genes was screened out. Subsequently, these hub genes were subjected to functional enrichment analysis, and their expression levels were verified in an independent dataset. Finally, transcriptional regulatory analysis and immune infiltration analysis were conducted for the hub genes, and their expression levels and prognostic values in various cancers were evaluated.

Results

In MI samples, nine genes, namely Immediate Early Response 3 (IER3), Heme Oxygenase 1 (HMOX1), Cyclin-Dependent Kinase Inhibitor 1A (CDKN1A), Plasminogen Activator Urokinase Receptor (PLAUR), MAF BZIP Transcription Factor F (MAFF), Solute Carrier Family 2 Member 3 (SLC2A3), Jun Proto-Oncogene (JUN), Transforming Growth Factor Beta Induced (TGFBI), and 6-Phosphofructo-2-Kinase/Fructose-2,6-Biphosphatase 3 (PFKFB3), were found to demonstrate significant dysregulation and to be closely associated with the occurrence of various cancers. Pan-cancer analysis further revealed the association of hub genes with cancer prognosis. Immune analysis also revealed their associations with resting CD4+ memory T cells and gamma delta T cells in TME.

Conclusion

IER3, HMOX1, CDKN1A, PLAUR, MAFF, SLC2A3, JUN, TGFBI, and PFKFB3 are potential biomarkers for MI and cancer. Research on hypoxia-related genes may provide new therapeutic targets for these two diseases.

Introduction

Myocardial infarction (MI) is a severe cardiovascular disease primarily caused by the rupture of unstable atherosclerotic plaques in the coronary arteries, which lead to thrombogenesis and acute occlusion of the coronary vessels. This eventually results in myocardial cell ischemia and hypoxia. The typical clinical manifestations of MI include chest pain, dyspnea, and arrhythmias, with severe cases potentially progressing to heart failure or even sudden death [1]. Globally, the incidence of this disease continues to rise. Despite the substantial reduction in acute mortality due to coronary artery bypass grafting, percutaneous coronary interventions, and pharmacological treatments, a considerable proportion of patients still experience poor long-term prognosis [2, 3]. Against this backdrop, novel therapeutic strategies based on gene regulation and protein regulation are expected to provide new avenues for improving clinical outcomes in MI patients [4].

In the ischemic and hypoxic microenvironment following MI, cardiomyocytes initiate a complex molecular regulatory network. Under hypoxic conditions, certain pivotal genes can be induced to exhibit differential expression patterns. For example, Hypoxia-inducible factor (HIF)-1α is rapidly activated, triggering downstream signaling pathways, which upregulate the expression of vascular endothelial growth factor to promote angiogenesis and improve myocardial perfusion [5]. The FMO2 gene is significantly upregulated under ischemic and hypoxic stress, and its encoded protein exerts antioxidant and cardioprotective effects through its protein disulfide isomerase activity [6]. Additionally, the expression of the antiapoptotic protein encoded by the TNFAIP3 gene is upregulated under hypoxic conditions, inhibiting cardiomyocyte apoptosis [7]. Therefore, exploring HRGs is beneficial for the treatment of MI.

Cancer is one of the diseases with a high mortality rate worldwide, and its essence lies in genetic mutations and abnormal cell proliferation. Its typical features include uncontrolled cell cycle progression, invasion, and metastasis, as well as the development of multidrug resistance. The development and evolution of cancer are not only modulated by intrinsic cellular factors but are also closely related to the TME [8, 9]. Hypoxia in the TME is considered a crucial factor in promoting cancer progression and metastasis. HIFs and their regulatory networks play key roles in cancer angiogenesis, metabolic reprogramming, and immune evasion [10]. For example, hypoxic conditions can activate the matrix metalloproteinase family, which decomposes the extracellular matrix (ECM) to facilitate cancer cell invasion and metastasis [11]. Additionally, hypoxia inhibits the antigen-presenting function of dendritic cells (DCs), reducing their ability to present cancer antigens and thereby weakening the immune response [12]. Moreover, hypoxic stress significantly enhances cancer cells’ resistance to apoptosis by activating key signaling axes such as PI3K/AKT-mTOR and MAPK-ERK [13]. Therefore, research on hypoxia-related genes (HRGs) is expected to provide insights into the development of new therapeutic strategies for cancer.

Current research indicates that there is a complex relationship between MI and cancer. On one hand, cancer treatments, including chemotherapy and radiation therapy, markedly increase cardiovascular risk. For instance, patients with breast cancer exhibit a significantly elevated rate of MI and heart failure following chemotherapy or radiotherapy [14]. On the other hand, MI itself may also elevate the risk of cancer [15]. Therefore, the HIF signaling pathway, as a key molecular hub, not only regulates angiogenic responses following myocardial ischemia but also drives TME remodeling, providing a theoretical basis for targeted therapies for both MI and cancer.

In summary, we identified 12 HRGs related to MI, 9 of which are closely related: IER3, HMOX1, CDKN1A, PLAUR, MAFF, SLC2A3, JUN, TGFBI, and PFKFB3. Notably, these genes not only correlate with MI but also exhibit close associations with various cancers.

Materials and Methods

Acquisition and Processing of RNA Sequencing Datasets

RNA sequencing datasets from patients with acute MI were retrieved from the GEO database (https://www.ncbi.nlm.nih.gov/geo). We selected datasets GSE19339 and GSE66360. The GSE19339 dataset includes thrombus samples from 4 MI patients and blood samples from 4 normal individuals. The GSE66360 dataset contains circulating endothelial cell samples from 49 MI patients and 50 normal individuals. For validating the gene expression levels, we used the GSE97320 dataset, which includes blood samples from 3 MI patients and 3 normal individuals. HRGs were selected from the hallmark gene set in the GSEA database (https://www.gsea-msigdb.org/gsea/index.jsp) [16]. The workflow of this study is shown in Figure 1.

Fig. 1.

Flowchart depicting the multistep screening strategy used to analyze the bioinformatics data.

Analysis of Variance

We divided the samples in the GSE19339 dataset into MI and control groups. Differential expression analysis was conducted using the “limma” package in R to identify DEGs. DEGs that fulfilled the conditions of |log₂FC| > 1 and p < 0.05 were considered to be significantly differentially expressed, and a volcano plot was constructed to present the results. Subsequently, the overlap of DEGs, key modules from WGCNA, and HRGs was illustrated using a Venn diagram. This process yielded a set of hypoxia-related DEGs.

Construction of Weighted Gene Co-Expression Network

Weighted gene co-expression network analysis (WGCNA) was performed in the GSE66360 dataset using the “WGCNA” R package. We utilized this approach to construct a co-expression network encompassing all genes. The “pickSoftThreshold” function was deployed to compute the soft-thresholding power (β), thereby quantifying the magnitude of co-expression similarity. The derived power value was applied to amplify co-expression similarity, followed by the calculation of the adjacency matrix. Subsequently, to assess the connectivity among genes within the network, the adjacency matrix was transformed into a topological overlap matrix, and the average linkage hierarchical clustering algorithm was employed to cluster genes with analogous expression patterns into the same module. Modules associated with clinical traits were identified via the computation of inter-module relationships and the correlations between gene modules and phenotypes. To identify key MI-related genes, based on the principle of WGCNA and combined with previous experience [17–19], we finally selected the top three modules with the highest correlation coefficients, and this modules satisfied R² > 0.3, p < 0.05.

Functional and Pathway Enrichment Analysis

Gene ontology (GO) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed to explore the functional impacts of gene expression. This process aims to annotate genes into a unified functional framework to elucidate their functional characteristics in biological processes.

Protein-Protein Interaction Network Analysis and Identification of Hub HRGs

The Search Tool for the Retrieval of Interacting Genes (STRING) was utilized to construct a functional protein association network and to evaluate the interactions among proteins [20]. The protein-protein interaction (PPI) network was visualized using Cytoscape software [21]. In the PPI network, hub genes were detected by employing the cytoHubba plugin and the MCODE algorithm within Cytoscape [22].

Construction of the Network for Predicted and Validated Core Genes, Transcription Factors, and MicroRNAs

We used the miRNet tool to build the network, and this platform integrates data from 11 different microRNA (miRNA) databases (www.mirnet.ca/) [23]. The network encompasses validated hub genes, transcription factor (TFs), and miRNAs, and we displayed it graphically using Cytoscape software. Using the Maximum Clique Centrality (MCC) method of the cytoHubba plugin in Cytoscape software, we calculate the top 10 miRNAs most significantly associated with MI [22]. Subsequently, the OncomiR database (https://oncomir.org/) was used to analyze the cancers associated with these top 10 miRNAs to identify their potential links [24].

Expression and Survival Analysis

We utilized the SangerBox web server (http://sangerbox.com/Tool) to explore the expression differences and prognostic outcomes of validated hub genes across various cancers. The prognostic value of validated hub genes across different cancers was analyzed using the Kaplan-Meier survival analysis. We analyzed the correlation between the expression levels of validated hub genes and the overall survival (OS) of cancer patients. Univariate Cox regression analysis was used to calculate the hazard ratio (HR). HR >1 indicated that high expression of hub genes was a high-risk factor for poor prognosis in cancer patients. The log-rank test was used to assess statistical significance, and a result is considered statistically significant when p < 0.05.

Immune Infiltration Analysis

For the datasets GSE19339 and GSE66360, the CIBERSORT algorithm was applied to evaluate the proportions of different types of immune cells [25]. Spearman’s rank correlation method was utilized to examine the relationships between diverse immune cell types and the expression levels of hub genes. The relationship between immune cell infiltration levels in the TME and validated hub genes was analyzed using the TIMER2.0 database (http://timer.cistrome.org/) [26].

Statistical analyses were performed using R Project (V4.3.0). p < 0.05 was regarded as statistically significant.

Results

Analysis of Variance

In the GSE19339 dataset, 1,507 genes showed differential expression, among which 694 were downregulated and 909 were upregulated. The top 10 differentially expressed genes were visualized using heatmaps and volcano plots (Fig. 2a, b). We used GO analysis and KEGG enrichment analysis to further explore the biological functions of these DEGs. Functional enrichment analysis via GO indicated that these genes were significantly enriched in multiple biological processes, including positive regulation of cell adhesion, chemotaxis, collagen-containing ECM, cytokine receptor binding, and integrin binding (Fig. 2c). Additionally, KEGG pathway enrichment analysis revealed that these genes were enriched in pathways such as amebiasis, complement and coagulation cascades, interaction between viral proteins and cytokine/cytokine receptors, cytokine-cytokine receptor interaction, malaria, and fluid shear stress and atherosclerosis (Fig. 2d). After MI, reactive oxygen species caused by ischemia can induce the expression of chemokines and cytokines, promoting the stimulation of leukocyte integrins and the synthesis of adhesion molecules, thereby recruiting inflammatory cells to infiltrate the infarcted area [27]. Inflammatory cells can clear dead myocardial cells and ECM debris, and secrete cytokines, which in turn recruit and activate interstitial repair cells, playing a significant role in subsequent ECM repair. However, an excessive inflammatory response can also lead to myocardial remodeling. At this time, the complement cascade can inhibit the recruitment of leukocytes after MI, serving a protective role [28]. Currently, researchers are dedicated to exploring inflammatory signaling pathways with therapeutic potential, including complement cascade reactions, cell chemotaxis, selectins, cytokine-cytokine interactions, and leukocyte integrins, aiming to mitigate adverse reactions following MI [29]. Therefore, focusing on these MI-related biological processes may have significant implications for future treatment strategies.

Fig. 2.

Differential expression analysis and functional enrichment of the GSE19339 dataset. a Volcano plot of the top 10 most significantly differentially expressed genes. b Heatmap of the top 10 most significantly differentially expressed genes. c GO enrichment analysis. d DKEGG enrichment analysis.

Identification and Exploration of Key MI Modules Using WGCNA

In the GSE66360 dataset, a gene co-expression network was constructed using the “WGCNA” R package. First, we calculated the Pearson correlation coefficients between samples and performed average linkage hierarchical clustering analysis on all samples. Based on these results, we successfully built a scale-free network with a soft threshold of 19 (Fig. 3a). Dynamic hybrid cutting was utilized to perform hierarchical clustering and generate a dendrogram, where a leaf represents a gene and a branch represents a module that clusters genes exhibiting analogous expression patterns (Fig. 3b). Modules with similar functions were merged, and 11 modules were finally identified (Fig. 3c). We selected the top three modules most relevant to the disease: the pink, green, and magenta modules, and obtained 704 important genes. Subsequently, these genes were compared with the previously identified DEGs and HRGs, and a final set of 12 overlapping genes was determined (Fig. 3d): PLIN2, SDC2, IER3, HMOX1, FBP1, CDKN1A, PLAUR, MAFF, SLC2A3, JUN, TGFBI, and PFKFB3.

Fig. 3.

Gene co-expression network analysis and identification of hub genes in the GSE66360 dataset. a Selection of the soft threshold power (β) to construct a scale-free gene co-expression network. b Dendrogram showing the results of hierarchical clustering analysis, grouping genes with similar expression patterns into modules. c Identification of 11 modules by merging functionally equivalent gene clusters. d Intersection analysis between gene modules and previously identified differentially expressed genes and HRGs, leading to the identification of 12 hub genes.

Identification of Hub HRGs and Network Construction

Twelve co-differentially expressed genes were input into the STRING database to further identify hub genes, resulting in the construction of a network containing 12 nodes and 43 edges (Fig. 4a). The MCC algorithm of the CytoHubba plugin in Cytoscape was used to calculate the top 10 most relevant hub genes in the PPI network. First, it identifies all the maximum cliques within a given biological network. Then, it calculates the MCC score for each node based on its participation in these maximum cliques. If a node is involved in multiple cliques or is closely connected to other nodes, it will have a higher MCC score, indicating that the node has a higher centrality and importance within the network [22]. In Figure 4b, the deeper the color, the greater the MCC score. We screened out the top 10 genes with the highest scores: PFKFB3, MAFF, IER3, PLAUR, CDKN1A, HMOX1, TGFBI, JUN, PLIN2, and SLC2A3. Additionally, the MCODE plugin in Cytoscape was used to identify significant gene clusters within the modules. The following criteria were used for screening: degree cutoff of 2, node score cutoff of 0.2, k-core of 2, and maximum depth of 100. Ultimately, a module comprising 9 nodes and 34 edges was identified as significant, and these 9 key nodes were displayed with yellow filling: JUN, IER3, HMOX1, CDKN1A, PLAUR, TGFBI, SLC2A3, MAFF, PFKFB3, PLIN2, and FBP1 (Fig. 4c). Integrating these results, a list of 9 hub HRGs was obtained, including IER3, HMOX1, CDKN1A, PLAUR, MAFF, SLC2A3, JUN, TGFBI, and PFKFB3.

Fig. 4.

Identification and network construction of hypoxia-related differentially expressed genes. a Protein-protein interaction (PPI) network generated using the STRING database. b Top 10 candidate hub genes identified from the PPI network using the MCC algorithm of the CytoHubba plugin. c Significant gene module identified using the MCODE plugin in Cytoscape.

Validation of Hub Genes

In the GSE97320 dataset, the expression profiles of the following 9 hub genes were validated in normal and MI tissues: IER3, HMOX1, CDKN1A, PLAUR, MAFF, SLC2A3, JUN, TGFBI, and PFKFB3. Figure 5 shows the expression differences of these genes in various samples.

Fig. 5.

Expression levels of hub genes in MI samples from the GSE97320 dataset.

Validation of Hub Genes, TFs, and miRNA Regulatory Networks

To construct the regulatory network of the validated hub genes, we used the miRNet database to identify TFs and miRNAs associated with these validated hub genes. A total of 151 TFs were identified, including AATF, ABL1, AR, ARID1A, ARNT, and ATF1, among others. Additionally, 876 miRNAs related to the validated hub genes were identified (online suppl. Table; for all online suppl. material, see https://doi.org/10.1159/000547896). A regulatory network related to the validated hub genes was constructed involving TFs and miRNAs, as shown in Figure 6. Using the MCC algorithm, the top 10 miRNAs most notably linked to MI were detected, including hsa-miR-92b-3p, hsa-miR-29c-3p, hsa-miR-27b-3p, hsa-let-7i-5p, hsa-miR-34a-5p, hsa-miR-103a-3p, hsa-miR-27a-3p, hsa-miR-16-5p, hsa-let-7f-5p, and hsa-let-7e-5p. The OncomiR database was used to analyze the cancers associated with these miRNAs. The results showed that the occurrence of 14 types of cancer was significantly associated with hsa-miR-92b-3p. Similarly, 16, 12, 11, 16, 12, 14, 13, 5, and 10 types of cancer were significantly associated with hsa-miR-29c-3p, hsa-miR-27b-3p, hsa-let-7i-5p, hsa-miR-34a-5p, hsa-miR-103a-3p, hsa-miR-27a-3p, hsa-miR-16-5p, hsa-let-7f-5p, and hsa-let-7e-5p (online suppl. Figure).

Fig. 6.

Illustrates the regulatory network of TFs and miRNAs associated with the validated hub genes. In the network, yellow squares represent hub genes, dark blue squares represent transcription factors, and light blue squares represent miRNAs.

Current research indicates that myocardial fibrosis is an inevitable pathological process following MI, and hsa-miR-29c-3p is a significant regulator of myocardial fibrosis. It may affect the process of myocardial fibrosis after MI by modulating the synthesis and degradation of the ECM, thereby influencing myocardial survival and function [30, 31]. In cancer, hsa-miR-29c-3p exerts a significant antitumor effect by targeting specific genes. For example, hsa-miR-29c-3p can target SPARC and TIAM1, respectively, to inhibit the proliferation and metastasis of colorectal cancer cells and nasopharyngeal carcinoma cells and can also affect the prognosis of pancreatic cancer by targeting MAPK1 [32–34]. Hsa-let-7i-5p plays an important role in the cell cycle: Overexpression of hsa-let-7i-5p during MI suppresses cardiomyocyte proliferation, which is detrimental to the repair of myocardial cells. Conversely, when the level of hsa-let-7i-5p is low, the expression levels of E2F2 and CCND2 are elevated, thereby facilitating cardiomyocyte proliferation. This might offer a potential therapeutic strategy for myocardial cell repair post-MI [35]. Hsa-miR-34a-5p reduces SIRT1 activity by targeting it, thereby promoting cell apoptosis after MI [36]. Additionally, it can target antiapoptotic genes such as Bcl-2, thereby exacerbating myocardial injury, increasing infarct size, and reducing left ventricular function [37]. In various cancers, hsa-miR-34a-5p primarily exerts a cancer-suppressive role. It targets multiple oncogenes to inhibit the proliferation, migration, and invasion of cancer cells. It can also suppress the expression of inflammatory factors, reduce the infiltration of inflammatory cells, and thus inhibit cancer progression [38]. Moreover, hsa-miR-34a-5p can enhance the body’s immune surveillance of cancers by regulating the expression of immune checkpoint molecules [39]. These findings indicate that these key miRNAs are of significant importance in MI and various types of cancer.

Expression of Validated Genes in Various Cancer Types

In order to delve deeper into the potential roles of IER3, HMOX1, CDKN1A, PLAUR, MAFF, SLC2A3, JUN, TGFBI, and PFKFB3 in cancer development, we explored their expression patterns in different types of cancer. We utilized the SangerBox database to conduct differential expression analysis on the validated hub genes, in order to explore their unique expression levels in different cancers. The upregulation and downregulation of hub genes in various types of cancer are shown in Figure 7. Light blue indicates downregulation of the gene in that particular cancer, while red indicates upregulation. The white color indicates that the gene’s expression profile shows no significant variations in that cancer. Additionally, the expression profiles of each gene across cancers are provided in online supplementary figure. These discoveries uncover the notable divergent expression patterns of these genes throughout diverse cancer types, indicating that they may play important roles in the initiation of cancer.

Fig. 7.

Upregulation and downregulation of hub genes in various types of cancer.

Prognostic Value of Validated Genes in Pan-Cancer

Survival analysis is a core methodology for exploring and quantifying disease prognosis, as well as evaluating the influence of potential prognostic factors on clinical endpoints. We performed a comprehensive evaluation to delineate the relationship between validated hub gene expression and OS. Figure 8a–i display forest plots summarizing the HRs linking hub gene expression to OS across cancer cohorts. All results from the forest plots are summarized in Table 1. Overall, the expression levels of IER3, HMOX1, CDKN1A, PLAUR, MAFF, SLC2A3, JUN, TGFBI, and PFKFB3 have shown significant prognostic correlations in various cancers. These findings indicate that these genes could represent key candidate pan-cancer prognostic biomarkers.

Fig. 8.

a–i Association between the expression levels of hub genes and OS in patients. Forest plot of HRs for IER3, HMOX1, CDKN1A, PLAUR, MAFF, SLC2A3, JUN, TGFBI, and PFKFB3 in human pan-cancer.

Table 1.

The prognostic relationship between hub genes and pan-cancer

Gene	High expression leads to a poor prognosis	Low expression leads to a poor prognosis
IER3	GBMLGG, CESC, LUAD, KIPAN, HNSC, GBM, KIRC, LIHC, PAAD	SKCM, SKCM-M
HMOX1	GBMLGG, KIPAN, THYM, LIHC, LGG, UVM, LAML, ALL, ALL-R	​
CDKN1A	GBMLGG, LUSC, MESO, LGG, UVM, PAAD, DLBC	KIRP, KIPAN, KIRC, ALL-R
PLAUR	GBMLGG, LGG, LAML, CESC, ACC, LUAD, KIPAN, HNSC, GBM, KIRC, MESO, LIHC, UVM, PAAD, TGCT	​
MAFF	GBMLGG, LUAD, HNSC, LGG, LIHC, PAAD, ACC	BRCA, ALL, ALL-R
SLC2A3	GBMLGG, LGG, CESC, ESCA, STES, KIPAN, STAD, HNSC, GBM, BLCA, THCA, MESO, UVM, PAAD, ACC	ALL
JUN	GBMLGG, LGG, CESC, ESCA, STES, THYM, BLCA, MESO	LAML, KIRC, neuroblastoma, ALL-R
TGFBI	GBMLGG, LGG, CESC, LUAD, KIPAN, HNSC, GBM, KIRC, BLCA, UVM, PAAD	SKCM-M
PFKFB3	CESC, KIRP, KIPAN, LIHC, ACC	SKCM, ALL

Immune Infiltration Analysis in MI Samples

A growing body of evidence indicates that MI is closely related to the immune microenvironment. Figure 9a and b show the proportions of different immune cells in MI samples evaluated by the CIBERSORT algorithm. Only the proportions of gamma delta T cells and CD4+ memory T cells demonstrated notable discrepancies when comparing the two datasets (p < 0.05), so we will focus on these two types of immune cells in subsequent analyses. As illustrated in Figures 9c and d, in the GSE19339 dataset, SLC2A3 and PFKFB3 are inversely correlated with resting CD4+ memory T cells, while IER3, PLAUR, SLC2A3, and PFKFB3 are inversely correlated with gamma delta T cells. However, in the GSE66360 dataset, IER3, HMOX1, CDKN1A, PLAUR, MAFF, and TGFBI are inversely correlated with resting CD4+ memory T cells, while IER3, HMOX1, CDKN1A, PLAUR, MAFF are inversely correlated with gamma delta T cells.

Fig. 9.

Immune infiltration landscape in myocardial samples. a, b Bar charts showing the proportions of different immune cell types in samples from the GSE66360 and GSE19339 datasets (* indicates p < 0.05). c, d Correlation analysis between hub genes and immune cells.

Correlation between Validated Genes and Resting CD4+ Memory T Cells, as well as Gamma Delta T Cells in the TME

CD4+ memory T cells can maintain long-term activity in the TME, rapidly initiate immune responses during tumor recurrence, and exert a significant effect in the TME. For instance, CD4+ memory T cells can secrete cytotoxic factors (such as granzyme B and perforin) to directly destroy cancer cells. Alternatively, they can secrete cytokines (such as IFN-γ and TNF-α) to activate other immune cells (such as macrophages and DCs), thereby enhancing the body’s immune response capability [40, 41]. Gamma delta T cells also have similar functions, such as secreting IFN-γ and TNF-α, directly killing tumor cells, or enhancing the antigen-presenting capacity of DCs to augment the cytotoxic T cells’ ability to kill tumor cells [42, 43]. Even in the absence of major histocompatibility complex (MHC), gamma delta T cells can recognize stress-induced antigens on the surface of tumor cells and mount an attack against them. This endows gamma delta T cells with unique advantages in treating tumors with low mutation burden or MHC deficiency. Currently, researchers have harnessed this special potential, enabling gamma delta T cells to serve a critical function in treating patients with advanced solid tumors and tuberculosis [44–46]. Figure 10a and b show the results of the correlation between hub genes and CD4+ memory T cells, as well as gamma delta T cells, analyzed by the CIBERSORT algorithm in TIMER2.0. There is a significant association between gene expression and the abundance of immune cells in various cancers. Specifically, in 16 different cancers, the expression of IER3 is significantly associated with the level of resting CD4+ memory T cells. Similarly, in 16, 10, 19, 12, 16, 15, 20, and 18 different cancer types, the expression of HMOX1, CDKN1A, PLAUR, MAFF, SLC2A3, JUN, TGFBI, and PFKFB3 is significantly correlated with the level of resting CD4+ memory T cells. Among the 8, 6, 10, 7, 6, 6, 8, 5, and 8 different cancer types, the expression of IER3, HMOX1, CDKN1A, PLAUR, MAFF, SLC2A3, JUN, TGFBI, and PFKFB3 were significantly associated with the abundance of gamma delta T cells. These findings demonstrate the significant role of this validated hub gene set within the TME.

Fig. 10.

a, b Correlation between hub genes and resting CD4+ memory T cells and gamma delta T cells in TME.

Discussion

MI and malignant cancers together account for over 50% of global mortality, imposing a significant socioeconomic burden [47, 48]. MI originates from acute ischemic injury of myocardial tissue, while cancer is characterized by uncontrolled proliferation driven by genetic and microenvironmental alterations. Recent epidemiological studies have revealed a strong association between the two conditions: the cardiovascular disease mortality rate increases among cancer patients, and the cancer incidence rate rises among patients with cardiovascular diseases [49–51]. This bidirectional relationship may stem from common risk factors (such as chronic inflammation and oxidative stress) and molecular interactions, particularly hypoxia-mediated signaling pathways. In ischemic myocardium, oxygen deprivation triggers apoptosis, fibrosis, and pathological remodeling; in the TME, hypoxia drives angiogenesis, metabolic reprogramming, and immune evasion [52, 53]. First, we identified 1,507 DEGs between MI patients and controls. Then, we constructed a key module using WGCNA and intersected it with HRGs, obtaining a total of 12 hub genes. Additionally, through bioinformatics analysis, we validated nine key genes: IER3, HMOX1, CDKN1A, PLAUR, MAFF, SLC2A3, JUN, TGFBI, and PFKFB3. Finally, we conducted a pan-cancer analysis to study the expression profiles, prognostic value, and immune infiltration of these nine genes across various cancer types. These results suggest that research on genes related to hypoxia may offer novel ideas for the treatment of MI and cancer.

IER3, also designated IEX-1, is a stress-inducible gene that is rapidly upregulated under stress conditions [54]. In the rat ischemia model, its high expression may have a cardioprotective role in the apoptosis and necrosis of cardiomyocytes by promoting the phosphorylation and particle translocation of PKCε and reducing the accumulation of intracellular reactive oxygen species [55]. Additionally, studies have shown that IEX-1 can inhibit local responses after vascular injury, such as suppressing endothelial regeneration, which may serve as a protective effect after MI [56]. In cancer, IER3 exhibits dual roles. In cancers such as cervical cancer, breast cancer, and lung cancer, it functions as an oncogene by interacting with RNA-binding protein HnRNPK and engaging in molecular cross-talk with its antisense RNA partner IER3-AS1 [57]. However, in neuroblastoma, IER3 acts as a cancer suppressor by regulating ADAM19 to inhibit cancer cell dissemination [58].

The HO-1 protein encoded by the HMOX1 gene is a stress-responsive protein that functions as an antioxidant, anti-inflammatory, and cytoprotective mediator [59]. Upon reperfusion following MI, HO-1 attenuates oxidative and inflammatory injury by catalyzing heme degradation to generate cytoprotective metabolites such as carbon monoxide and bilirubin, thereby limiting myocardial damage. In the MI model, overexpression of HO-1 can significantly reduce infarct size and improve cardiac function. Mice lacking the HO-1 protein showed more severe myocardial damage after ischemia-reperfusion [60]. In addition, HO-1 not only alleviates myocardial damage caused by acute ischemic events, but also counteracts cardiac insufficiency caused by chronic heart failure. In the cardiomyocytes of Bach1 gene deficient mice, HO-1 expression was enhanced, showing anti-hypertrophy and anti-remodeling effects, effectively preventing left ventricular remodeling and hypertrophy caused by coronary artery ligation [61]. In cancer, HMOX1 plays an important role by promoting mechanisms such as angiogenesis, regulating inflammatory response, and immune escape [62, 63].

The CDKN1A encodes p21 and plays a pivotal role in cell-cycle regulation by modulating the activity of cyclin-dependent kinase 2 or cyclin-dependent kinase 4 complexes. Its expression is strictly regulated by the tumor suppressor protein p53 and is a key factor in p53-mediated G1 phase arrest [64, 65]. Additionally, p21 can interact with proliferating cell nuclear antigen to govern DNA replication and repair. Additional studies have shown that elevated p21 expression can induce matrix metalloproteinase 9 upregulation, thereby augmenting the disseminative capacity of cancer cells [66].

The PLAUR gene encodes the urokinase-type plasminogen activator receptor, whose primary role is to catalyze the conversion of plasminogen to plasmin, thereby initiating a cascade of proteolytic reactions. These reactions can degrade the ECM, promoting cell migration, invasion, and angiogenesis [67]. PLAUR is upregulated in multiple cancers and correlates with poor prognosis [68]. During MI, high expression of PLAUR promotes the degradation and remodeling of the ECM, which may play an important role in tissue repair and inflammatory response after MI [69, 70].

MAFF is a newly identified central regulator of the atherosclerosis-associated hepatic network, capable of eliciting gene-specific expression patterns that modulate coronary artery disease risk, and it serves as a critical mediator in cell proliferation, differentiation, antioxidant activity, and inflammatory pathways [71, 72]. Additionally, MAFF may enhance the antioxidant capacity of arterial endothelial cells by activating antioxidant response element-related genes in these cells, reducing cellular damage and thus potentially exerting a protective effect during MI [73]. In the TME, hypoxic conditions activate HIF-1, which in turn upregulates the expression of MAFF. Studies have shown that MAFF mainly promotes cancer progression by enhancing cancer cell invasion and metastasis. For example, in breast cancer, MAFF expression is significantly higher in lymph node metastases than in the primary cancer and is positively correlated with hypoxia markers [74, 75].

SLC2A3 is primarily involved in the transport of glucose across the cell membrane. During MI, it may affect cellular energy supply and survival through the metabolic state of cardiomyocytes [76]. Cancer cells mainly obtain nutrients through glycolysis, and SLC2A3 promotes cancer progression and metastasis by enhancing glucose uptake and metabolism to provide energy for cancer cells. For example, in colorectal cancer, SLC2A3 plays a key effect in perineural invasion by regulating epithelial-mesenchymal transition. Overexpression of SLC2A3 in triple-negative breast cancer can activate inflammatory M1 tumor-associated macrophages, leading to aggressive cancer cells [77–79].

The JUN gene encodes the c-Jun protein. Knockout of c-Jun in embryonic stem cells promotes the differentiation and development of cardiomyocytes. After MI, c-Jun may affect the cardiac repair process by inhibiting cell reprogramming and promoting cardiomyocyte apoptosis [80]. In cancer, c-Jun can enhance the activity of neutrophils, amplifying their pro-angiogenic and pro-inflammatory effects, thereby promoting cancer growth and metastasis. JUN is upregulated in a broad spectrum of cancers, including head and neck squamous cell carcinoma, colorectal cancer, and breast cancer [81].

The TGFBI gene encodes a TGF-β-induced ECM protein that is essential for cell adhesion, embryonic development, bone formation, and the pathogenesis of diverse diseases [82]. After MI, tissue injury induces the activation of the TGF-β signaling pathway, which in turn upregulates the expression of TGFBI. TGFBI binds to collagen types I, II, and IV and participates in cell-cell, collagen, and matrix interactions, thereby promoting myocardial fibrosis and leading to further myocardial damage [83, 84]. TGFBI is also one of the markers of epithelial-mesenchymal transition, which is intimately linked to cancer cell invasion and metastasis. In the TME, TGFBI fosters tumor immune evasion by dampening immune cell function. Its overexpression is associated with lymph node metastasis and poor prognosis in various cancers, such as renal cancer, colorectal cancer, and so on [85].

The PFKFB3 gene encodes a key glycolytic enzyme that regulates the glycolysis pathway. PFKFB3 expression is markedly upregulated under hypoxic and inflammatory conditions. Under hypoxic conditions, high expression of PFKFB3 enhances glycolytic metabolism and promotes the secretion of inflammatory factors, thereby causing myocardial injury [86]. PFKFB3 is overexpressed in various cancers and promotes rapid proliferation of cancer cells by enhancing glycolytic metabolism. Due to its key role in cancer metabolism, PFKFB3 inhibitors have become a research hotspot in cancer therapy. For instance, small-molecule PFKFB3 inhibitors have achieved significant breakthroughs in inhibiting cancer invasion [87].

Overall, HRGs primarily reduce cardiomyocyte necrosis, promote tissue repair, and alleviate myocardial fibrosis via antioxidant and anti-inflammatory effects as well as by regulating endothelial cell function. In cancers, they exert a dual role: in some cancers, they drive progression by enhancing cancer cell proliferation, promoting angiogenesis, and suppressing immune cell function; in others, they exert antitumor effects by inhibiting cancer cell migration and augmenting antioxidant capacity.

Conclusion

This study has identified IER3, HMOX1, CDKN1A, PLAUR, MAFF, SLC2A3, JUN, TGFBI, and PFKFB3 as promising biomarkers in the hypoxic microenvironments of MI and cancer. These findings may provide important evidence for subsequent new treatments for MI and cancer. Although our study offers novel insights into the pivotal roles of HRGs in both MI and cancer, it should be noted that our data are derived exclusively from bioinformatics analysis. Therefore, further in vivo and in vitro experiments are necessary to substantiate our findings. Future efforts should concentrate on elucidating the specific effector mechanisms of the relevant molecules and systematically evaluating the therapeutic efficacy of gene-targeted interventions.

Acknowledgments

We are very grateful to the staff of the “GEO” database and the TCGA database for their support of the data used in this study.

Statement of Ethics

An ethics statement was not required for this study type as it is based exclusively on data extracted from GEO database (https://www.ncbi.nlm.nih.gov/geo). No human studies were carried out by the authors for this article. No animal studies were carried out by the authors for this article.

Conflict of Interest Statement

The authors declare no competing interests.

Funding Sources

This work received funding from Changzhou Applied Basic Research Program (CJ20220095), Changzhou Key Medical Discipline (CZXK202202), Changzhou Sci&Tech Program (CJ20235085) and Changzhou “Longcheng Talent Program” (No. CQ20220127).

Author Contributions

As the guarantor, Xiaoyu Yang and Can Hou conceived the study. Si Yan collected the receipts and drafted the manuscript. Zhichao Fang and Xin Xue revised the manuscript.

Funding Statement

This work received funding from Changzhou Applied Basic Research Program (CJ20220095), Changzhou Key Medical Discipline (CZXK202202), Changzhou Sci&Tech Program (CJ20235085) and Changzhou “Longcheng Talent Program” (No. CQ20220127).

Data Availability Statement

All the raw data of this study can be publicly accessible from GEO database (https://www.ncbi.nlm.nih.gov/geo).