Integrative analyses and validation of ferroptosis-related genes and mechanisms associated with cerebrovascular and cardiovascular ischemic diseases

Liao, Wei; Wen, Yuehui; Zeng, Chuan; Yang, Shaochun; Duan, Yanyu; He, Chunming; Liu, Ziyou

doi:10.1186/s12864-023-09829-w

Research
Open access
Published: 04 December 2023

Integrative analyses and validation of ferroptosis-related genes and mechanisms associated with cerebrovascular and cardiovascular ischemic diseases

Wei Liao^1,2,
Yuehui Wen³,
Chuan Zeng⁴,
Shaochun Yang^1,2,
Yanyu Duan^2,5,
Chunming He^1,2 &
…
Ziyou Liu^2,6,5

BMC Genomics volume 24, Article number: 731 (2023) Cite this article

991 Accesses
1 Altmetric
Metrics details

Abstract

Background

There has been a gradual increase in the occurrence of cardiovascular and cerebrovascular ischemic diseases, particularly as comorbidities. Yet, the mechanisms underlying these diseases remain unclear. Ferroptosis has emerged as a potential contributor to cardio-cerebral ischemic processes. Therefore, this study investigated the shared biological mechanisms between the two processes, as well as the role of ferroptosis genes in cardio-cerebral ischemic damage, by constructing co-expression modules for myocardial ischemia (MI) and ischemic stroke (IS) and a network of protein–protein interactions, mRNA-miRNA, mRNA-transcription factors (TFs), mRNA-RNA-binding proteins (RBPs), and mRNA-drug interactions.

Results

The study identified seven key genes, specifically ACSL1, TLR4, ADIPOR1, G0S2, PDK4, HP, PTGS2, and subjected them to functional enrichment analysis during ischemia. The predicted miRNAs were found to interact with 35 hub genes, and interactions were observed between 11 hub genes and 30 TF transcription factors. Additionally, 10 RBPs corresponding to 16 hub genes and 163 molecular compounds corresponding to 30 hub genes were identified. This study also clarified the levels of immune infiltration between MI and IS and different subtypes. Finally, we identified four hub genes, including TLR4, by using a diagnostic model constructed by Least Absolute Shrinkage and Selection Operator (LASSO) regression analysis; ADIPOR1, G0S2, and HP were shown to have diagnostic value for the co-pathogenesis of MI and cerebral ischemia by both validation test data and RT-qPCR assay.

Conclusions

To the best our knowledge, this study is the first to utilize multiple algorithms to comprehensively analyze the biological processes of MI and IS from various perspectives. The four hub genes, TLR4, ADIPOR1, G0S2, and HP, have proven valuable in offering insights for the investigation of shared injury pathways in cardio-cerebral injuries. Therefore, these genes may serve as diagnostic markers for cardio-cerebral ischemic diseases.

Peer Review reports

Introduction

The prevalence of cardio-cerebral comorbidities has been significantly increasing. Historically, most studies and therapies for cardio-cerebral vascular diseases have focused solely on a singular disease within the cardio-cerebral vascular system. This approach fails to provide a comprehensive and integrated diagnosis and treatment plan for patients with cardio-cerebral comorbidities, thereby impeding the precise management of cardio-cerebral vascular comorbidities. Research indicates that the mechanisms of injury in myocardial and cerebral ischemia are similar. Both conditions share a common pathological mechanism known as atherosclerosis [1]. Additionally, the pathological state following the onset of cardio-cerebral ischemic disease is closely associated with ischemia–reperfusion injury [2]. Furthermore, many molecular events that occur after neurological injury also occur after cardiac injury [3]. These findings suggest the presence of a shared injury mechanism in the heart and brain. Nevertheless, the precise cause of the injury remains uncertain and requires further investigation. Therefore, exploring shared therapeutic objectives for heart-brain injury, administering targeted treatments to safeguard the heart and brain, and adopting a brain-centered homeopathic approach, is imperative.

Our team has shown that glutamate receptors serve as prevalent indicators of damage in cardiac and cerebral ischemia-hypoxia-reperfusion injuries. Additionally, it can control the progression of cardiac and cerebral ischemia/reperfusion injury [4,5,6]. However, further studies are required to elucidate the underlying mechanisms. Research indicates that during cardiac transplantation and coronary artery occlusion, cardiomyocytes undergo ferroptosis, leading to inflammation and worsening of cardiac injury. The utilization of Fer-1, a ferroptosis inhibitor, has been shown to decrease the occurrence of ferroptosis in cardiomyocytes [7]. Additionally, in mice undergoing surgery for middle cerebral artery occlusion, ferroptosis intensifies surgically induced cerebral ischemia/reperfusion injury [8].

A recent biological information and validation study on ferroptosis-related genes (FRGs) during ischemic stroke (IS) screened hub genes related to ferroptosis and suggested a possible mechanism for dexmedetomidine-mediated inhibition of ferroptosis during IS [9], and during myocardial ischemia (MI), the same biosignature study identified genes related to ferroptosis and carried out immune infiltration analyses and confirmation of related hub genes [10, 11], these studies separately from the direction of MI and IS to explore the role of FRGs in MI and IS, but did not further study the common characteristics of the process of cardio-cerebral ischemia, and to further clarify the process of cardio-cerebral ischemia with a consistent trend of expression of the hub genes of ferroptosis, which is precisely the problem explored in this paper, the study of the FRGs in cardio-cerebral ischemia. The study of the common role of ferroptosis genes in ischemia is of great significance in the search for common pathways and mechanisms of ischemic injury.

This study aimed to investigate the role of ferroptosis in cardiac and cerebral ischemic injuries by exploring the common pathways of injury between the two organs. Using various machine algorithms, the optimal model of the most relevant FRGs in heart and brain injury was determined. Bioinformatics, clinical specimens, and cell model validation were used to study the hub genes and biological processes associated with ferroptosis during MI and IS. This analysis aimed to identify common injury mechanisms and targets for cardio-cerebral ischemia–reperfusion injury, which can aid in early intervention and treatment.

Materials and methods

Data acquisition

The MI related datasets GSE60993 [12], GSE66360 [13], and GSE48060obtained from the GEO database (https://www.ncbi.nlm.nih.gov/geo) [14]. The GSE60993 dataset was acquired from Homo sapiens, using the data platform being GPL6884. It consisted of 33 samples, including 17 MI, 7 normal, and 9 unstable angina pectoris samples. For the present analysis, only 17 MI samples and 7 normal samples from GSE60993 were included. TheGSE66360 dataset was obtained from Homo sapiens, with the data platform being GPL570. It comprises 99 samples, including 49 MI samples and 50 normal samples. All the samples were selected for analysis. Finally, the GSE48060 dataset, also from Homo sapiens, used the GPL570 data platform and contained 52 samples, including 31 MI and 31 normal samples. This analysis included 31 samples, including 21 normal samples. The training datasets used were GSE60993 and GSE66360, while GSE48060 was used as the validation dataset. Additionally, the GEO database provided IS associated datasets GSE22255 [15], GSE16561 [16], and GSE58294, which were downloaded. Datasets GSE22255, which consisted of 40 samples (20 IS cases and 20 normal cases), and GSE16561, which consisted of 63 samples (39 IS cases and 24 normal cases), were included in the analysis. The dataset GSE58294, which consisted of 92 samples (69 IS cases and 23 normal cases), was used as the validation dataset. All the samples were selected for inclusion in the analysis. The data platforms used were GPL570 for GSE22255 and GSE58294, and GPL6883 for GSE16561. Specific grouping information for the datasets can be found in the Supplementary material (Additional file 1: Table S1).

The GeneCards database (https://www.genecards.org) [17] was used to identify genes associated with ferroptosis. Supplementary Material (Additional file 2: Table S2) displays the specific names of the 698 genes acquired from the GeneCards database using the search term “Ferroptosis”.

Identification of differentially expressed genes (DEGs) associated with ferroptosis

To obtain the integrated MI dataset, we used the R package sva to de-batch the datasets GSE60993 and GSE66360, comprising 66 MI cases and 57 control (normal) samples. Additionally, we obtained an integrated IS dataset using datasets GSE22255 and GSE16561, which included 59 IS and 44 control (normal) samples. After normalizing the combined dataset using the limma package [18], the effectiveness of batch effect removal was confirmed by conducting Principal Component Analysis (PCA) on the expression matrix of the dataset both before and after the batch effect was eliminated. PCA is a data dimensionality reduction method that extracts the feature vectors (components) of high-dimensional data, converts them to low-dimensional data and displays these features in a two-dimensional graph.

The expression data of genes related to ferroptosis in the MI and IS were acquired by intersecting the gene expression data of the FRGs with the respective data for the MI and IS. To examine the impact of FRG gene expression levels in MI and IS, we conducted differential gene analysis on the combined dataset. This analysis utilized the R package limma [19] to identify significant differential genes. The threshold for differential genes was set at a fold change (FC) absolute value > 1.5 and p < 0.05. Additionally, genes with FC > 1.5 and p < 0.05 were considered up-regulated in expression. Differential genes were defined as those with FC > 1.5 and p < 0.05. Up-regulated differential genes were identified as those with FC > 1.5 and p < 0.05, while down-regulated differential genes were identified as those with FC < -1.5 and p < 0.05. The expression patterns of genes related to iron-induced cell death were illustrated using the ggplot2 package in R for volcano plotting and the pheatmap package in R for heatmap visualization.

Assessment of biological characteristics between disease and control samples

Gene Ontology (GO) enrichment analysis is a widely used approach to study the functional enrichment of genes on a large scale, across various dimensions and levels. This analysis is typically conducted on three levels: biological process (BP), molecular function (MF), and cellular component (CC) [18]. Kyoto Encyclopedia of Genes and Genomes (KEGG) is an extensively utilized repository for housing data on genomes, biological pathways, diseases, and medications [20]. Disease Ontology (DO) provides gene annotations from a disease standpoint. The R package clusterProfiler [21, 22] was used to perform GO functional annotation, KEGG pathway enrichment, and DO disease enrichment in order to identify significantly enriched biological processes. This analysis was conducted on DEGs related to ferroptosis between the disease and control samples from the integrated dataset of MI and IS. The significance threshold for enrichment analysis was set at p < 0.05.

Gene Set Enrichment Analysis (GSEA) is a computational technique used to assesswhether a pre-established group of genes exhibits significant variances between two biological conditions. It is frequently employed to evaluate alterations in pathways and biological process activity within expression dataset samples [23]. In order to examine the disparities in biological mechanisms between MI and IS disease samples and control samples, we utilized gene expression profiling datasets and referred to the gene sets 'c5.go.v7.4.entrez.gmt' and 'c2.cp.kegg’. v7.4. The datasets from entrez.gmt [24] were enriched and visualized using the GSEA method provided in the R package clusterProfiler. p < 0.05 was considered statistically significant.

Gene Set Variation Analysis (GSVA) is an unsupervised analysis technique that is primarily employed to assess the gene expression matrix across various samples by converting it into a gene set expression matrix using a non-parametric approach. The enrichment results of the transcriptome gene sets were used to evaluate whether distinct metabolic pathways are enriched across various samples. To examine the variation in biological processes between disease and control samples of MI and IS in the combined dataset, we conducted gene set variation analysis on the gene expression profiling data of the disease and control samples of MI and IS. This analysis was performed using the R package” GSVA” [25], and the reference gene set was obtained from the MSigDB database [24]. The dataset's enrichment scores for each hallmark were calculated using the set 'h.all.v7.4.symbols.gmt', and the correlation between the dysregulated pathways in the patients was determined. p < 0.05 was considered statistically significant.

Construction of a diagnostic model for Ferroptosis

A risk score formula was established using the Least absolute shrinkage and selection operator (LASSO) algorithm analysis to reduce dimensionality and identify differentially expressed feature genes among the genes related to ferroptosis. The formula incorporates the individual normalized gene expression values weighted by the penalty coefficients of the feature genes.

$$\mathrm{riskScore}=\sum_{\mathrm i}\mathrm{Coefficient}\left({\mathrm{gene}}_{\mathrm i}\right)\ast\mathrm{mRNA}\;\mathrm{Expression}({\mathrm{gene}}_{\mathrm i})$$

Random Forest (RF) [26] is a combination of multiple algorithms that integrates multiple decision trees using the concept of integration learning. This belongs to the bagging (Bootstrap AGGregatING, Self-Service Sampling Integration) integration algorithm among other integration algorithms. The random forest is a frequently used technique for constructing models. This involves creating multiple decision trees and using the statistics of each tree to predict a specific sample. The final result was determined by selecting the most common prediction among the trees using a voting method. The RF package [27] was used for model construction using the expression matrix of DEGs related to ferroptosis in the integrated dataset. The parameters set.seed(234) and ntree = 1000 were employed.

$$I(X=xi)= -lo{g}_{2}p({x}_{i})$$

An Support Vector Machine (SVM) is a generalized linear classifier used in supervised learning to classify data into two categories. It determines the decision boundary as a hyperplane with a maximum margin, which is obtained by solving the learning samples. SVM calculate the practical risk by a hinge loss function and incorporates a regularization term into the solution system to optimize the structural risk. Kernel methods enable SVM to perform nonlinear classifications, making them sparse and robust classifiers. SVM employs a hinge loss function to calculate the empirical risk and incorporates a regularization term to optimize the structural risk, making it a classifier that is both sparse and resilient. The kernel method allows the SVM to perform nonlinear classification, making it a popular choice among kernel learning methods.

The Receiver Operating Characteristic (ROC) [28] is a graphical analysis tool that enables the selection of the most suitable model, elimination of the second-best model, or determination of the optimal threshold within a single model. The ROC curveis a combined measure that represents the sensitivity and specificity. This illustrated the interplay between sensitivity and specificity using a compositional approach. Typically, the area under the ROC curve ranges from 0.5—1. The diagnostic performance improved as the area under the curve (AUC) approaches 1. The AUC exhibits low accuracy ranging from 0.5 to 0.7, moderate accuracy from 0.7 to 0.9, and high accuracy at 0.9 and beyond.

To identify the diagnostic indicators linked to ferroptosis in MI and IS, the integrated datasets of MI and IS were subjected dimensionality reduction using there techniques: LASSO, RF, and SVM. Subsequently, ROC Curves were generated using the R package pROC [29], and the AUC was calculated to evaluate the precision of the diagnostic models and determine the most suitable model.

Weighted Gene Association Co-expression Network Analysis (WGCNA)

The WGCNA algorithm was used to examine the gene expression patterns in numerous samples to, facilitate gene clustering and module creation based on comparable gene expression patterns. Additionally, it allows for the analysis of connections between modules and their biological characteristics [30]. For this investigation, we employed the R software WGCNA [30] to construct gene co-expression networks pertaining to MI and IS. Initially, we utilized the ideal soft threshold β (8 for MI and 7 for IS) to create scale-free networks, individually. Next, the dissimilarity Topology Overlap Matrix (TOM)-based dissimilarity (dissTOM) was calculated and used to perform gene dendrograms and module identification with a maxClusterSize of 6000 and a minClusterSize of 30 to perform hierarchical clustering to identify co-expression modules. Subsequently, we computed Pearson’s correlation Coefficient (PCC) and its corresponding p value. These results were visualized using a heatmap to compare the module feature genes with clinical features. From this analysis, we identified disease-associated genes by selecting gene modules with an absolute correlation value greater than 0.3 and p < 0.05, which were considered clinically relevant.

Building a network of protein-protein interaction (PPI), mRNA-miRNA, mRNA-RBP, mRNA-TF and mRNA-drugs interactions

The PPI network [31] consists of proteins that interact with each other to engage in different aspects of life processes, including biological signaling, gene expression regulation, energy and material metabolism, and cell cycle control. The Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database [32] serves as a repository for exploring identified proteins and forecasting protein interactions. In this study, we utilized the STRING database to link the co-morbid genes of MI and IS with a PPI network (minimum required interaction score low confidence [0.150]. Subsequently, we exported the PPI and employed Cytoscape [33] software for additional analysis. The Cytohubba [34] plug-in consists of 12 algorithms (Betweennes, BottleNeck, Closeness, ClusteringCoefficien, Degre, DMN, EcCentricity, EPC, MCC, MNC, Radiality, and Stress), which calculate the top 30 nodes in each algorithm. Genes that appeared in at least 5 algorithms were defined as hub nodes, indicating their significant connectivity with other nodes. These hub nodes may play crucial roles in regulating the entire biological process and warrant further investigation.

MiRNAs, which are single-stranded RNA molecules encoded by endogenous genes, are a type of non-coding RNA with a length of approximately 19–25 nt. They play crucial regulatory roles in the evolution of biological development. In post-transcriptional gene regulation, miRNAs play crucial roles in tumorigenesis, biological development, organogenesis, and epigenetic regulation by effectively controlling target gene expression. Viral defense mechanisms play important regulatory roles. Typically, miRNAs possess an intricate regulatory system, in which a single miRNA can control numerous target genes; conversely, a target gene can be regulated by multiple miRNAs [27]. To examine the correlation between hub genes and miRNAs, we acquired miRNAs associated with hub genes from the Starbase database (http://starbase.sysu.edu.cn/). This database offers seven prediction programs (TargetScan, microT, miRmap, PicTar, RNA22, PITA, and miRanda) for predicting the results. To construct the mRNA-miRNA regulatory network, interactions between miRNAs and mRNAs were predicted using two or more programs,and Cytoscape software were used to visualize the mRNA-miRNA regulatory network.

Transcription factors (TFs) through gene expression control. To examine the regulatory function of central genes, the miRNet database was used to retrieve and construct the interaction network between TFs and central genes. Using the Cytoscape software, we visualized the network of interactions between hub genes and TFs.

Furthermore, the Encyclopedia of RNA Interactomes (ENCORI) database was used to predict the interactions between RBP and hub genes. We then filtered the mRNA-RBP interaction pairs based on the criteria of clipExpNum > = 5 and clipIDnum > 10, and subsequently constructed the mRNA-RBP interaction network.

The Comparative Toxicogenomics Database (CTD) (http://ctdbase.org/) is a digital platform that connects chemicals, genes, phenotypes, and diseases, and establishes toxicological data to enhance the comprehension of human health-related information. By applying the screening criterion of having at least three references and two organisms, the CTD database was used to identify potential medications or small molecule compounds that interact with hub genes. Furthermore, we employed Cytoscape to create a visual representation of the network of interactions between mRNA and drugs.

Molecular subtypes of Ferroptosis

The Uniform Manifold Approximation and Projection (UMAP) algorithm, a nonlinear dimensionality reduction technique, can partition or condense a group of individuals into a sequence of separate clusters using specified characteristics. By utilizing the umap package in R [35], the algorithm categorizes the integrated dataset of MI and IS patients into distinct subtypes, relying on characteristic genes.

Exploration of immune infiltration

The immune microenvironment is composed of a complex interconnected network primarily comprising immune cells, cells involved in inflammation, fibroblasts, and mesenchymal cells, as well as a variety of cytokines and chemokines. The examination of immune cell infiltration in samples plays a crucial role in guiding disease research and predicting treatment prognosis. Single sample gene set enrichment analysisss (GSEA) algorithm extends the GSEA method to determine the levels of` 28 immune cells in disease and control samples [32]. This information was presented as box plots, allowing visualization of the immune cell composition in patients with various MI and IS subtypes. Statistical significance was determined using the Wilcoxon test to calculate variations in the proportions of immune cells. A p-value < 0.05 was considered statistically significant.

CIBERSORx utilizes a machine learning technique that expands the algorithm framework to deduce gene expression profiles specific to cell types, eliminating the requirement for physically isolating cells. RNA-Seq data [36] is utilized to approximate the quantity of immune cells present in a specimen. Using the CIBERSORTx algorithm (https://cibersortx.stanford.edu/), we determined the prevalence of 22 immune cell types in various patient subtypes within the integrated MI and IS datasets. The R package corrplot [37] was used to create heat maps showing the correlation between the degree of immune cell infiltration.

ESTIMATE analysis is a computational method that measures the level of immune infiltration in tumor samples using gene expression data that indicates the presence of stromal and immune cell gene signatures. The ESTIMATE package for R [38] was used to estimate variations in immune scores between control samples and patients with MI and IS. This package calculates the correlation between hub gene expression levels and immune scores.

Modeling of cellular hypoxia-reoxygenation

Hypoxia-reoxygenation models were created using CL-0481 rat neuronal PC12 cells from ProCell Life Science&Technology Co., Ltd. (Wuhan, China) and iCell-r012 rat cardiomyocyte H9c2 cells from iCell Bioscience Co., Ltd. (Shanghai, China). The model was validated using a CCK-8 assay and flow cytometry., The detailed procedure has been described elsewhere [4], and the cells were gathered and preserved in a − 80℃ facility.

Clinical blood specimen collection

To further identify the presence of hub genes in brain and heart ischemia, we obtained nine blood samples from individuals undergoing clinical cardiopulmonary resuscitation, 12 blood samples from patients with acute IS, 12 blood pressure samples from patients with acute MI, and 12 blood pressure samples from individuals with a normal physical examination between March and October 2022. The collection time and specific treatments for all blood samples as described in our previous study [4].

Validation using real-time fluorescence quantitative PCR (RT-qPCR)

RT-qPCR was used to determine the mRNA levels of hub genes. RNA was extracted using the CWBIO Ultrapure RNA Extraction Kit (CW0581M). The cDNA was synthesized using a reverse transcription kit (R223-01; Vazyme). The procedure involves three steps. Each reaction was a total volume of 20 μL, consisting of 10 μL ChamQ Universal SYBR qPCR Master Mix (Q711-02, Vazyme China) (2X), 4 μL of both forward and reverse primers, 1 μL cDNA, and the necessary amount of nuclease-free water. Each sample was analyzed in triplicate. The particular cycling conditions were as follows: initial denaturation at a temperature of 95 °C for 10 min, followed by denaturation at 95 °C for 10 s, annealing at 58 °C for 30 s, and extension at 72 °C for 30 s. A melting curve (final dissociation curve) was generated. The 2^−ΔΔCt method was used to calculate the relative expression of the gene, with β-actin chosen as the endogenous reference. Supplementary Material (Additional file 3: Table S3 and Additional file 4: Table S4) displays the primers used.

Statistical analysis

R software (version 4.1.1) was used for all data processing and analyses. The statistical significance of variables that followed a normal distribution was assessed using an independent t-test to compare the two groups of continuous variables. Differences between variables that did not follow a normal distribution were analyzed using the Wilcoxon rank-sum test to compare the two groups of independent variables. Pearson’s correlation analysis was used to calculate correlation coefficients between various genes. ROC curves were generated using the R package pROC to evaluate the precision of the diagnostic model, and the AUC was computed. The statistical significance of all p-values was assessed using a two-sided test, and p < 0.05 was considered to indicate statistical significance. The results of RT-qPCR are reported as means ± standard deviation. Comparison of two groups of data that were normally distributed was conducted using the Student’s t-test. Groups were compared using one-way analysis of variance (ANOVA). Unless otherwise specified, p value of < 0.05 was considered statistically significant.

Results

Analysis of DEGs associated with ferroptosis

According to the flowchart (Fig. 1), the MI datasets GSE60993 and GSE66360 were initially combined (Additional file 5: Figure S1A), revealing a noticeable batch effect in the merged data (Additional file 5: Figure S1C). Subsequently, the batch effect was eliminated from the datasets (Additional file 5: Figure S1B) to obtain gene expression profile data with consistent expression levels (Additional file 5: Figure S1D). The integrated data comprised 66 MI and 57 normal samples. Subsequently, the IS datasets GSE22255 and GSE16561 were combined (Additional file 5: Figure S1E), which revealed a noticeable discrepancy between them (Additional file 5: Figure S1G). This discrepancy was eliminated from the datasets (Additional file 5: Figure S1F), resulting in gene expression profiling data that exhibited uniform expression levels (Additional file 5: Figure S1H). The integrated data comprised of data from 59 patients with IS and 44 healthy controls.

To identify the DEGs associated with ferroptosis, we obtained the expression data of FRGs in MI and IS by intersecting the combined datasets of MI and IS with the FRGs. The analysis resulted in 22 DEGs, with 20 upregulated (FC > 1.5 and p < 0.05) and 2 downregulated genes (FC < − 1.5 and p < 0.05) (Fig. 2A). The DEGs were visualized using the R package pheatmap (Fig. 2C), which revealed significant differences in expression among the different subgroups of the combined MI dataset. The analysis of IS and control samples resulted in the identification of 66 differentially expressed genes: 35 upregulated and 31 downregulated genes (Fig. 2B). Heat maps created using the R package pheatmap (Fig. 2D) clearly indicate the 66 differentially expressed genes in various subgroups of the combined IS dataset. Box plots of expression levels revealed that all genes related to ferroptosis that were differentially expressed in MI showed significant differential expression in both disease and normal samples (Fig. 2E). Similarly, all genes related to ferroptosis, which were differentially expressed in IS, also showed significant differential expression in both the disease and normal samples (Fig. 2F).

To identify the FRGs that showed co-differential expression in both MI and IS, we intersected 22 differentially expressed FRGs from the merged MI dataset with 65 differentially expressed FRGs from the merged IS dataset. This resulted in the identification of seven FRGs (ACSL1, TLR4, ADIPOR1, G0S2, PDK4, HP, and PTGS2) that were differentially expressed in both MI and IS. Box plots of expression levels revealed that the seven genes related to ferroptosis, which were co-differentially expressed in MI and IS, exhibited significant differential expression between diseased and normal samples in both the MI-merged dataset (Fig. 3A) and the IS-merged dataset (Fig. 3B). Next, we conducted ROC validation on the combined MI and IS datasets to assess the diagnostic impact of the seven FRGs co-differentially expressed in MI and IS. These seven genes have a discernible diagnostic effect on both MI and normal samples (Fig. 3C) and a specific diagnostic effect on both IS and normal samples (Fig. 3D).

A total of 22 genes related to ferroptosis showed differential expression between MI and normal samples. We examined the correlation between the expression levels of these 22 genes in the normal and MI samples. The findings indicate that the majority of differentially expressed FRGs exhibited a positive correlation in MI samples (Fig. 4A), and a negative correlation in normal samples (Fig. 4B).

A total of 65 genes related to ferroptosis were differentially expressed between the IS and normal samples. Among these genes, 35 were upregulated and 31 were downregulated. Next, we examined the association between the expression levels of 35 distinct FRGs that showed increased expression, and 31 FRGs that showed decreased expression in both normal and MI samples. The findings indicated that the majority of FRGs that exhibited increased expression displayed favorable associations in both IS samples (Fig. 4C) and normal samples (Fig. 4D). FRGs expressed in a distinct manner, exhibiting reduced expression, exhibited positive correlations in nearly all of the IS samples (Fig. 4E) and in almost all of the normal samples (Fig. 4F).

The GO functional annotation of MI DEGs in the merged datasets revealed that these genes were primarily enriched in various biological processes, including tissue and temperature homeostasis, and smooth muscle cell proliferation (Fig. 5A, Additional file 6: Table S5). Moreover, in terms of cellular components, they were associated with transcription regulator complexes, tertiary granule lumens, and tertiary granules (Fig. 5C). Furthermore, these genes exhibited molecular functions, such as tetrapyrrole binding, STAT family protein binding, and protein heterodimerization activity (Fig. 5E). KEGG Pathways were enriched in Yersinia infection, toll-like receptor signaling pathway, TNF signaling pathway, and other pathways (Fig. 5G, Additional file 7: Table S6). Finally, they were also associated with diseases such as varicose veins, trypanosomiasis, and retinal diseases (Fig. 5I).

Functional annotation of GO revealed that the DEGs in the combined dataset were primarily enriched in biological processes such as temperature regulation, oxidative stress response, and nutrient-level response (Fig. 5B, Additional file 8: Table S7). They were also associated with cellular components such as the vesicle lumen, U1 snRNP, and the spliceosomal snRNP complex (Fig. 5D). Additionally, these genes exhibited molecular functions such as ubiquitin protein ligase binding, ubiquitin-like protein ligase binding, and ubiquitin-like protein-conjugating enzyme binding (Fig. 5F). Furthermore, they were enriched in KEGG pathways such as Spliceosome, Shigellosis, and Peroxisome (Fig. 5H, Additional file 9: Table S8). Finally, these genes were linked to diseases, such as stomach carcinoma, polycystic ovary syndrome, and peripheral nervous system neoplasms (Fig. 5J).

GSEA and GSVA enrichment

GSEA was performed on the merged MI and IS datasets to investigate the correlation between expressed and involved biological processes, affected cellular components, and molecular functions in disease and control samples. Promoted biological processes in MI disease samples (Fig. 6A, Additional file 10: Table S9) included igg binding (Fig. 6C), rage receptor binding (Fig. 6D), positive regulation of membrane protein ectodomain proteolysis, immunoglobulin binding, and regulation of dendritic cell differentiation. Conversely, inhibition of biological processes in MI disease samples (Fig. 6B) comprised negative regulation of multicellular organism growth (Fig. 6E), pseudouridine synthase activity (Fig. 6F), pseudouridine synthesis, small nucleolar ribonucleoprotein complex, and positive regulation of RNA polymerase ii transcription pre-initiation complex assembly. Promoted biological processes in the IS disease samples (Fig. 6G, Additional file 11: Table S10) included oxygen transport (Fig. 6I), platelet alpha granule membrane (Fig. 6J), innate immune response in the mucosa, pattern recognition receptor activity, and positive regulation of vascular endothelial growth factor production. Conversely, the inhibition of biological processes in IS disease samples (Fig. 6H) involved nucleotide salvage (Fig. 6K), RNA methylation (Fig. 6L), positive regulation of mitochondrial translation, mitochondrial small ribosomal subunit, and RNA modification, among others.

To investigate the functional disparities between the disease and control samples in the merged MI and IS datasets, we conducted GSVA. The results revealed that MI samples exhibited significant activation of hallmark angiogenesis, hallmark apical junctions, hallmark apical surfaces, hallmark apoptosis, and other biological processes (Fig. 7A, Additional file 12: Table S11). Additionally, IS samples showed significant activation of hallmark hedgehog signaling, hallmark heme metabolism, and hallmark hypoxia (Fig. 7B, Additional file 13: Table S12). Furthermore, we computed the connections between differentially expressed genes in relation to ferroptosis and hallmark biological processes in 22 MI cases. The findings indicate that secreted protein acidic and rich in cysteine (SPARC) demonstrated a significantly positive correlation with hallmark glycolysis, hallmark 16 JAK/STAT3 signaling, hallmark inflammatory response, hallmark interferon alpha response, and hallmark interferon gamma response. Additionally, SPARC exhibited a significant negative correlation with hallmark pancreatic beta cells and spermatogenesis (Fig. 7C). Correlation analysis between 65 IS genes associated with ferroptosis and hallmark biological processes revealed that ACSL1 exhibited a significant positive correlation with upregulated hallmark KRAS signaling, downregulated hallmark uv response dn, and ADIPOR1. In contrast, the hallmark e2f targets, hallmark interferon-alpha response, hallmark oxidative phosphorylation, and hallmark unfolded protein response exhibited a significant negative correlation (Fig. 7D).

Co-expression modules for MI and IS

WGCNA was applied to the integrated MI datasets. To ensure a topology without scaling, the recommended soft threshold for scale-free R was 0.80 (Additional file 14: Figure S2A). The WGCNA identified five modules in the integrated MI dataset. Each gene module is indicated in a different color. To evaluate the correlation between each module and disease, heat maps were generated using Spearman’s correlation coefficients, illustrating the module-feature relationships (Fig. 8A). The modules named “gray” and “brown” strongly correlated with MI and were chosen as MI-associated modules (gray module, R = 0.3, P = 8e-04;’ brown module, R = 0.55, P = 7e-11). (Fig. 8C). Furthermore, we graphed the gene importance compared to the module affiliation for the “gray” (Additional file 14: Figure S2B) and “yellow” (Additional file 14: Figure S2C) modules individually. Scatter plot of gene significance vs. module membership. The total number of MI-associated genes was 3407.

WGCNA was used to identify gene modules linked to IS using integrated IS datasets. To ensure a scale-free topology (Additional file 14: Figure. S2D), the ideal soft threshold for scale-free R was determined to be 0.80. WGCNA identified nine modules in the integrated IS dataset, wherein each color corresponded to a distinct gene module. To evaluate the correlation between each module and disease, heat maps illustrating the module-feature relationships were generated using Spearman correlation coefficients (Fig. 8B). The IS-associated modules, namely the pink module (R = 0.31, P = 0.002) and blue module (R = 0.38, P = 7e-05), exhibited a strong correlation with IS (Fig. 8D). Furthermore, we graphed scatter plots displaying the correlation between gene significance and module membership for the “pink” (Additional file 14: Figure S2E) and "blue" (Additional file 14: Figure S2F) modules, respectively. We obtained a total of 3412 genes related to MI.

Shared gene and network analysis of MI and IS

To identify shared genes between MI and IS, we extracted co-expressed genes associated with diseases in MI and IS from the combined datasets. Subsequently, we identified 3137 and 3806 genes that were differentially expressed in MI and IS, respectively. By intersecting the set of genes co-expressed with diseases and the set of genes differentially expressed in MI and IS, we identified 168 genes shared between MI and IS (Fig. 9A).

To examine the interactions among the DEGs, we built a PPI network using 168 common genes associated with MI and IS. The PPI network comprised 653 pairs of interactions and 163 shared genes. This network was visualized using Cytoscape. WDR59, ARG1, and MPO had the highest number of gene interactions (29, 29, and 28, respectively) (Fig. 9B).

Hub genes were extracted by calculating the top 30 nodes in each of the 12 algorithms in CytoHubba and selecting 35 genes that appeared in at least 5 algorithms (Fig. 9C). Using the R package “GOSemSim” [39], we calculated the semantic similarity of hub genes in Go. These findings indicated that CTSG, ORM1, and MPO exhibited functional correlations with multiple genes (Fig. 9D).

We predicted the miRNAs that interacted with 35 hub genes. Subsequently, we utilized Cytoscape to visually represent the network of mRNA-miRNA interactions (Fig. 9E). miRNAs are represented as sky-blue oval blocks in the network of mRNA-miRNA interactions, whereas mRNAs are depicted as red dots. The mRNA-miRNA interaction network reveals that our network comprised 11 hub genes and 71 miRNA molecules, forming a total of 84 mRNA-miRNA interaction pairs.

Using Cytoscape (Fig. 9F), we visualized the interaction relationship data of 11 hub genes and 30 TFs in the mRNA-TF network that we constructed. The mRNA-TF interaction network contained 39 pairs of mRNA-TF interactions.

A network of hub genes was built and 10 RNA-binding proteins (RBPs) corresponding to 16 hub genes were identified (Fig. 9G). The two RBPs, CSTF2T and FMR1, simultaneously target 11 crucial genes. The mRNA-RBP interaction network contained 81 pairs of mRNA-RBP interactions.

In the mRNA-drug interaction network (Fig. 9H), we built a network of hub genes associated with mRNA drugs and discovered 163 potential drugs or molecular compounds linked to 30 hub genes. We discovered that C006780 simultaneously targeted 20 hub genes. The network of interactions between the mRNA and drugs consisted of a grand total of 349 pairs.

Construction of a diagnostic model related to Ferroptosis

The diagnostic value of the seven FRGs codifferentially expressed in MI and IS in the integrated MI datasets was assessed by analyzing their expression using RF (Fig. 10A). MeanDecreaseGini indicates the decrease in the Gini coefficient of a node. We filtered the results of specific analyses using MeanDecreaseGini > 0 as a criterion because nodes with reduced Gini coefficients were more valuable. From the results presented in (Fig. 10B), we identified six diagnostic markers of MI derived from seven genes associated with ferroptosis that were co-differentially expressed in both MI and IS. The markers were determined using the RF algorithm. In addition, ROC curves were generated for the combined MI datasets using the established RF model (Fig. 10C), which yielded an AUC value of 0.617. Subsequently, we employed a SVM to screen the selected genes and construct an MI diagnostic model. The radial kernel function was utilized, and the optimal number of feature genes obtained from the seven FRGs codifferentially expressed in MI and IS was five. Among these, two were randomly selected to represent the support vectors in the data (Fig. 10D). Furthermore, we predicted and plotted ROC curves based on the established model using the integrated MI datasets (Fig. 10E), resulting in an AUC value of 0.887. In conclusion, a diagnostic model for MI was developed using LASSO regression analysis (Fig. 10F). This model identified four MI diagnostic markers from a set of seven FRGs that were differentially expressed in both MI and IS. Furthermore, we visualized the results of the LASSO regression and generated trajectory plots for the LASSO variables (Fig. 10G). ROC curves were predicted for the integrated MI datasets using the established LASSO regression model (Fig. 10H) and achieved an AUC value of 0.896.

To assess the diagnostic significance of the seven MI and IS co-DEGs related to ferroptosis in the combined IS dataset, we examined the expression of these genes using the RF algorithm (Fig. 10I). MeanDecreaseGini indicates a decrease in the Gini coefficient of a node. The more Gini coefficients are reduced by the nodes, the more valuable they are. We filtered the specific analysis results using MeanDecreaseGini > 6 as the criterion. The findings indicated (Fig. 10J) that we identified six diagnostic indicators for IS from seven FRGs that were codifferentially expressed by MI and IS using the RF algorithm. Subsequently, we used the established RF model to predict and visualize the ROC curves on the integrated IS datasets (Fig. 10K), achieving an AUC value of 0.816. Next, we employed an SVM to screen the selected genes and construct the IS diagnostic model, utilizing the radial as the kernel function. Among the seven FRGs co-differentially expressed by MI and IS, we identified five optimal feature genes, two of which were randomly chosen to demonstrate the support vectors in the data (Fig. 10L). Furthermore, we predicted and plotted ROC curves based on the model constructed using the integrated IS dataset (Fig. 10M), which yielded an AUC value of 0. 841.In conclusion, a diagnostic model for IS was developed using the LASSO regression analysis (Fig. 10N). All seven FRGs, which were codifferentially expressed in MI and IS, served as diagnostic markers for MI. Furthermore, we visualized the results of the LASSO regression and generated trajectory plots for the LASSO variables (Fig. 10O). Additionally, we predicted and plotted ROC curves for the integrated IS datasets using the established LASSO regression model (Fig. 10P) and achieved an AUC value of 0.892.

After evaluating the AUC values of the diagnostic models created using RF, SVM, and LASSO, LASSO regression analysis was chosen as the ultimate method for constructing the diagnostic models for both MI and IS. The MI diagnostic model comprised 4 hub genes (TLR4, ADIPOR1, G0S2, and HP), whereas the IS diagnostic model consisted of 7 hub genes (ACSL1, TLR4, ADIPOR1, G0S2, PDK4, HP, and PTGS2). These models were selected as the final models.

To confirm whether these hub genes exhibited the same differential expression and expression patterns in other datasets, we generated box plots of the hub genes in the GSE48060 dataset by comparing the MI and normal samples (Fig. 11A). Of the four hub genes, three displayed significant differences in expression between the MI and normal samples, and their expression patterns aligned with the combined datasets. Additionally, we used the established LASSO regression model, which included the four hub genes, to predict the GSE48060 dataset and plotted ROC curves (Fig. 11B), resulting in an AUC value of 0.624. Furthermore, we independently verified the ROC curve of the hub genes in the model. The figure clearly illustrates that HP (Fig. 11E) and ADIPOR1 (Fig. 11F) exhibited a significant difference in expression between the MI and normal samples, demonstrating specific diagnostic implications.

We created box plots comparing the hub genes in the GSE58294 dataset between IS and normal samples (Fig. 11C). Of the seven hub genes, five exhibited notable variations in expression between IS and normal samples. Expression patterns were aligned with the combined datasets. Using the GSE58294 dataset, a LASSO regression model was constructed with seven hub genes, allowing expression prediction. Subsequently, an ROC curve (Fig. 11D) was generated with an AUC value of 0.706. Furthermore, we conducted ROC validation and observed that G0S2 (Fig. 11G) and ACSL1 (Fig. 11H) exhibited specific diagnostic impacts on both the IS and Normal samples.

Immune infiltration analysis of combined datasets

The ssGSEA was used to evaluate the extent of immune cell infiltration in the combined datasets by comparing MI and IS samples to normal samples. The results indicate that the levels of immune cells, such as activated CD4 T cells, activated dendritic cells, central memory CD8 T cells, neutrophils, and T follicular helper cells, were higher than those in the normal samples. The levels of eosinophils, macrophages, natural killer cells, natural killer T cells, and regulatory T cells were notably elevated in MI samples (Fig. 12A). Compared to normal samples, IS samples exhibited significantly higher levels of these immune cells (Fig. 12B). The correlation between the expression levels of the identified genes and immune cells in MI and IS samples was calculated. The findings indicated a significant correlation between the expression of the identified genes, TLR4 and G0S2, and the presence of various immune cells in the MI samples (Fig. 12C). Similarly, the expression of the identified genes, TLR4 and HP, was significantly correlated with the content of multiple immune cells in IS samples (Fig. 12D). By separately calculating the correlation between the immune cell composition in MI and IS samples, we observed that monocytes, central memory CD8 + T cells, immature dendritic cells, regulatory T cells, gamma delta T cells in IS samples, and macrophages exhibited a positive correlation (Fig. 12E). In IS samples, there was a positive correlation between Type 2 T helper cells and macrophages, monocytes, natural killer cells, neutrophils, mast cells, regulatory T cells, T follicular helper cells, natural killer T cells, activated dendritic cells, central memory CD8 T cells, MDSC, plasmacytoid dendritic cells, immature dendritic cells, and gamma delta T cells. However, memory B cells were negatively correlated with most immune cells (Fig. 12F).

Identification of patient subtypes by characterized genes

Using the UMAP method (Fig. 13A), two subtypes of patients, cluster1 and cluster2, were identified based on four hub genes. Cluster1 contained 23 samples whereas cluster2 contained 43 samples. Cluster analysis revealed variations in the identified genes between the two sample groups (Fig. 13C). This indicated that the four crucial genes exhibited significant differential expression in distinct subgroups of the MI combined dataset. The differences in the expression of the four hub genes between the two MI disease subtypes (cluster1 and cluster2) were examined, and the results are illustrated in a group comparison plot (Fig. 13E). Additionally, the featured genes exhibited differential expression between the two patient subtypes (p < 0.05).

Using the UMAP method (Fig. 13B), two subtypes of patients, namely cluster1 and cluster2, were identified based on the seven hub genes. There were 25 samples in Cluster1 and 34 samples in cluster2.The the clustering outcomes revealed variations in the identified genes between the two sample groups (Fig. 13D). This indicated that the seven hub genes exhibited differential expression in distinct subgroups of the combined IS dataset. The differences in expression of the seven hub genes between the two IS disease subtypes (cluster1 and cluster2) were examined, and the findings of the expression difference analysis are illustrated using a group comparison plot (Fig. 13F). Additionally, a significant proportion of the highlighted genes exhibited differential expression between the two patient subtypes (p < 0.05).

Variations in immune traits among subcategories

To evaluate the extent of immune cell infiltration in the two distinct patient subtypes, the CIBERSORTx algorithm was employed. According to the findings, a notable distinction was observed in the correlation of immune cells between cluster1 and cluster2 patient groups (Fig. 14A and B) in individuals with MI. The correlation between the four hub genes and immune cell content in patients with the MI subtype was calculated separately. These findings indicate that the expression levels of multiple hub genes in patients in the cluster1 group were significantly positively correlated with NK cells at rest, macrophages M2, and neutrophils. Additionally, the expression levels of T CD4 memory cells at rest and in activated dendritic cells demonstrated a significant negative correlation with the expression levels of several hub genes (Fig. 14C). In the cluster2 patient group, expression of the central HP gene showed a significant positive correlation with the presence of various immune cells (Fig. 14D).

The correlation between immune cells in patients in cluster1 group (Fig. 14E) differed significantly from that in patients in cluster2 group (Fig. 14F). Associations between the seven hub genes and immune cell composition in patients with IS were also individually computed. These findings indicated a significant positive correlation between the expression level of the hub gene PDK4 and the presence of various immune cells in patients belonging to the CLUSTER1 group (Fig. 14G). In patients in cluster2 group, there was a strong positive correlation (Fig. 14H) between the abundance of various immune cells and the expression level of the central gene, TLR4.

RT-qPCR validation of hub genes expression in cellular models and clinical disease blood specimens

Compared to the controls, TLR4 mRNA expression was significantly increased in both cell models and blood specimens of clinical diseases (Fig. 15A and G). In the PC12 cell model, ADIPOR1 mRNA expression did not differ significantly from that in normal PC12 cells but was significantly higher in the H9c2 cell model and blood specimens from patients with clinical diseases (Fig. 15B and H). GOS2 mRNA expression was significantly higher in both cell models and blood specimens of patients with clinical diseases than in the control group (Fig. 15C and I). Similarly, HP mRNA showed a significant increase in both the cell model and blood specimens of patients with clinical diseases compared to the control group (Fig. 15D and J). ACSL1 mRNA expression was significantly elevated in the IS and MI groups compared to the control group, whereas there was no significant change in expression in the cardiopulmonary resuscitation group compared to the control group, and in the cellular model, where the expression of ACSL1 mRNA was significantly elevated (Fig. 15E and K), and the PDK4 mRNA expression was significantly higher in the IS group compared with the control group, whereas there was no significant change in expression in the MI and cardiopulmonary resuscitation(CPR) groups compared with the control group, and PDK4 mRNA expression was significantly higher in the PC12 cell model, whereas there was no significant change in expression in the H9c2 cell model (Figs. 15F and L). PTGS2 mRNA expression was significantly higher in MI group compared to controls, but no significant change in IS and CRR groups (Fig. 15M), In the cellular model, we confirmed by repeated verification of specimen RNA, primers and experimental conditions that PTGS2 mRNA expression was low in rats, and RT-qPCR was unable to detect the relevant expression of PTGS2 mRNA.

Discussion

Effective preventative and curative techniques for cardio-cerebral damage resulting from cardiac arrest and complex cardio-cerebral vascular conditions are limited, leading to unsatisfactory treatment outcomes due to the limited timeframe and irreversible damage to heart muscle cells and neurons [40,41,42]. Despite functional and structural disparities, research indicates a remarkable resemblance in injury mechanisms between the heart and brain [1, 43]. However, the precise underlying mechanism remains unclear and requires further investigation. The involvement of ferroptosis in the progression of various diseases, including tumors and neurodegenerative diseases, has been previously reported [44]. An experiment conducted in mice with myocardial ischemia–reperfusion injury demonstrated an increase in non-heme iron content in the myocardium. Ferroptosis inhibitors alleviate ventricular remodeling and injury, indicating ferroptosis in the cardiomyocytes of mice with myocardial ischemia–reperfusion injury [45]. Following myocardial reperfusion, ACSL4 and iron levels increased, whereas GPX4 levels decreased [46]. Li et al. found that in reperfusion injury due to cardiac transplantation or cardiac coronary artery occlusion, cardiomyocytes undergo ferroptosis and trigger inflammation that exacerbates cardiac injury, and the use of an inhibitor of ferroptosis reduces cardiomyocyte ferroptosis [7]. Tuo et al. discovered that the effects of surgery on cerebral ischemia–reperfusion injury caused by middle cerebral artery occlusion (MCAO) were worsened by ferroptosis [47]. This study demonstrated that the suppression of ferroptosis reduced the damage caused by cerebral ischemia–reperfusion injury in the hippocampal neurons of mice in the MCAO model. This was achieved by enhancing GPX4 expression as described in a previous study [48]. These reports uncover the involvement of ferroptosis in cardiovascular disorders; however, additional investigations are required to fully understand its precise molecular mechanisms. This study aimed to establish a diagnostic model for the FRGs in patients with MI and IS. Multiple machine algorithms have been used to achieve this goal. Therefore, the best diagnostic model was selected for this study. In the future, bioinformatics methods will be used to analyze the biological processes of the hub genes, and further clinical specimens and cellular modeling experiments will be conducted to verify these findings. This study identified the hub genes related to ferroptosis in the process of cardiac ischemia–reperfusion injury and explored the biological mechanism of FRGs in this process.

In this study, we analyzed the expression of seven hub genes associated with ferroptosis in MI, IS, CRP, and physical examination populations, as well as in the H9c2 and PC12 cellular models, based on the aims of our study and the results of our validation dataset. We discovered that TLR4, ADIPOR1, G0S2, and HP play significant roles in the pathogenesis of MI and IS. The test set also confirmed that these hub genes have a diagnostic value for understanding the pathogenesis of MI and IS.

TLR4 (Toll-like receptor 4 (TLR4) plays a significant role in initiating the innate immune response and transmitting signals via a cascade of molecules [49]. The inflammatory response plays a crucial role following cardiac and cerebral ischemia/reperfusion injury [50, 51]. Recently, the association between TLR4 and inflammation was investigated. TLR4 exhibits the greatest expression among cardiac TLRs, Research has indicated that blocking the TLR4 signaling pathway can decrease the inflammatory response in heart muscle and potentially protect against further harm to the already injured myocardium [52]. Preclinical research has utilized genetically modified animal models to investigate the involvement of TLR4 in promoting inflammation and development of cardiac fibrosis and dysfunction [53]. Additionally, investigations have shown that during stroke, molecules typically contained within cells are released into the extracellular space owing to uncontrolled cell death. TLR4 has been identified as a potential target for stroke treatment because of its ability to bind circulating immune cells and activate an inflammatory response [54]. Additionally, the TLR4-NF-κB signaling pathway has been shown to alleviate inflammation in acute kidney injury [55]. TLR4 is a central gene involved in ferroptosis associated with ischemic stroke [56]. Recent findings also suggest that TLR4 plays a role in cardio-cerebral ischemic disorders, as evidenced by decreased expression of TLR4 in the cortex and hippocampus after cardiopulmonary resuscitation in rat models [57]. Taken together, these studies and our report further suggest an important role for TLR4 in cardiac ischemia–reperfusion injury; therefore, it is worthwhile to investigate the mechanisms of iron death and the role of TLR4 in this process.

ADIPOR1 (adiponectin receptor 1) plays a role in controlling fatty acid breakdown and glucose concentration. Interaction between ADIPOR1 and APPL1 diminishes the cardioprotective effects of adiponectin on myocardial ischemia/reperfusion injury in type 2 diabetic mice [58]. Autophagy-mediated ADIPOR1-AMPK signaling can suppress apoptosis in brain cells induced by cardiac arrest or cardiopulmonary resuscitation [59]. Additionally, the ADIPOR1-AMPK signaling pathway plays a crucial role in IS injury and can hinder the progression of cerebral injury after IS through lipocalin [60]. These findings align with our research and provide a basis for further investigation of shared pathways in cardiac and cerebral ischemic injuries.

G0S2 (G0/G1 switch 2) is primarily found in mitochondria and plays a role in promoting the extrinsic apoptotic signaling pathway. A retrospective clinical study indicated that low levels of G0S2 expression in peripheral blood may be a marker for MI [61]. Overexpression of G0S2 attenuates the decline in cardiomyocyte ATP and increases mitochondrial ATP production under hypoxia, whereas the knockdown of G0S2 expression leads to an increase in cardiomyocyte ATP decline [62]. No studies have investigated the relationship between G0S2 and IS have been published. However, in conjunction with the results of our study, we observed a significant increase in G0S2 expression in the peripheral blood of IS patients who underwent cardiopulmonary resuscitation. Further studies are required to elucidate the precise mechanisms.

ADIPOR1 and G0S2, as important findings in our study, provide a strong basis for us to study the common pathways of heart-brain ischemic injury in the direction of pre-glutamate metabolism and lipid metabolism. The aberrant expression of these two genes may play a key role in future clinical diagnosis and treatment, but further studies are still needed to verify their biological functions and potential clinical application value.

HP (haptoglobin) encodes a preprotein that binds to tetramers to produce binding beads. Bound adhesins can overutilize the available hemoglobin, thereby preventing oxidative damage caused by iron, inflammation, atherosclerosis, and cerebrovascular disease [63, 64]. The binding bead protein genotype is a consistent marker of coronary heart disease risk in individuals with elevated glycosylated hemoglobin levels [65]. In diabetic patients, the binding bead protein genotype is a major determinant of cardiovascular disease risk and can be used to predict CVD risk of cardiovascular disease in the presence of diabetes [66]. This suggests that HP is closely linked to the development of cardiovascular diseases, particularly ischemic diseases. Additionally, different HP genotypes are closely associated with atherosclerosis in patients with ischemic stroke, and the HP2-2 genotype serves as a genetic biomarker for precision medicine and personalized healthcare in stroke patients [67]. The MCAO model also showed that proteins attached to the binding beads had an effect on macrophage/microglia-induced inflammation and were able to protect the brain from ischemic injury. This attachment improves survival and motor function and reduces brain damage in rats. Therefore, HP may be an effective treatment for cerebral ischemia [68]. Along with the corresponding analysis of biological functions, this provides a direction for future investigations into the shared pathways of ischemic injury.

Using GO functional analysis, we examined the DEGs in the combined datasets and discovered that these genes shared numerous common functions in MI and IS, including oxidative stress and apoptosis. KEGG pathway enrichment analysis indicated that the DEGs were primarily enriched in MI and IS within pathways such as the Toll-like signaling pathway, IL-17 signaling pathway, Ferroptosis, and Adipocytokine signaling pathway. Various studies have demonstrated that resveratrol effectively improves acute MI through mechanisms associated with reduced oxidative stress and inflammation, potentially by affecting TLR4 expression [69]. Additionally, the IL-17 signaling pathway can influence the connection between multiple sclerosis and acute MI, making it a potential target for pharmacological intervention in acute MI [70]. Similarly, in acute IS, the upregulated expression of lncRNA ENSG00000226482 has been identified as a diagnostic and therapeutic biomarker. This effect is mainly achieved through activation of the adipocytokine signaling pathway [71]. The findings of these studies align with ours and provide direction for further investigation of the functions of upstream and downstream hub genes in ischemic injury.

We used ssGSEA to evaluate the levels of immune infiltration in the MI and IS datasets. Additionally, we investigated the features of distinct ferroptosis-related hub genes in various subtypes of MI and IS along with their immune profiles. The analysis revealed significant differential expression of most hub genes among the subtypes and substantial variation in immune cell correlations across different disease subtypes. During MI and IS, the infiltration of various immune cell types is strongly associated with the progression of cardio-cerebral vascular diseases [72,73,74,75]. Therefore, it is imperative to further investigate associated immune infiltration.

In the field of computational biology, the continuous progress of interactions prediction research has provided us with insights into the molecular mechanisms of MI and IS as well as potential biomarkers. In particular, mRNA-TF, miRNA-ncRNA, and mRNA-miRNA interaction prediction has become an area of great interest, which is crucial for revealing the interactions of miRNAs, non-coding RNAs, etc. with MI and IS. In future studies: we plan to further improve and develop more accurate interactions prediction models to better identify the effects of mRNA-miRNA-ncRNA interactions on MI and IS. In the future, we will explore more multi-omics data, including transcriptomics, proteomics, metabolomics data and single-cell multi-omics data [76], to further comprehensively understand the pathophysiological processes of MI and IS by using autocoders, non-negative matrix decomposition, conditional random field graph convolutional networks, network distance analysis and asymmetric autocoders for the framework of single-cell data analysis [77,78,79,80], and eventually We will provide better diagnostic and therapeutic methods for cardiac and cerebral ischemia injuries through functional validation and clinical practice.

In a previous study, we examined the biological mechanisms of the genes associated with glutamate receptors in MI and IS. Additionally, we identified glutamate receptor-related hub genes using LASSO regression analysis. However, we did not comprehensively analyze the biological processes involved in MI and IS [4]. Ferroptosis is a specific glutamate excitotoxic process caused by ROS aggregation [81]. This study utilized more relevant datasets and multiple machine algorithms(RF, SVM, LASSO, CIBERSORX, ESTIMATE analysis, and nonlinear dimensionality reduction algorithm, etc.) to further explore the role of ferroptosis in cardio-cerebral ischemia and the related biological processes during cardio-cerebral ischemia, and screened out ferroptosis -related hub genes by using the test datasets as well as the validation of the clinical specimens and the cellular models, which provided an effective guideline to further explore the common pathway of cardio-cerebral injury, and provided theoretical support for the subsequent exploration of common pathways of cardio-cerebral ischemia injury based on the processes of glutamate signaling and ferroptosis. Nevertheless, the current investigation has some shortcomings, and our confirmation of the hub genes was limited to the initial sample and cellular verification, necessitating further exploration of the precise mechanism in the forthcoming research. Owing to the limited data available in the current datasets, it was not possible to integrate the hub genes with clinical information to analyze their correlation with onset time, prognosis, and other factors. Therefore, to thoroughly investigate the mechanism of ischemic injury, we collected pertinent blood samples and clinical data, and prepared them for additional sequencing.

Conclusions

Four ferroptosis-related hub genes, TLR4, ADIPOR1, G0S2, and HP, can be used as diagnostic markers of cardio-cerebral ischemic diseases and provide an effective guideline for further exploration of the common injury pathways of cardio-cerebral injuries.

Availability of data and materials

This study contains all data and materials that can be verified and accessed, as they have been deposited in online repositories. Detailed information about these repositories is available in the article or Additional Materia.

Abbreviations

ROC:: Receiver operating characteristic
MI:: Myocardial ischemia
IS:: Ischemic stroke
PPIs:: Protein-protein interactions
TFs:: Transcription factors
RBPs:: RNA-binding proteins
ssGSEA:: Single sample gene set enrichment analysis
UMAP:: Uniform Manifold Approximation and Projection
FRGs:: Ferroptosis-related genes
GEO:: Gene Expression Omnibus
DEGs:: Differentially expressed genes
PCA:: Principal Component Analysis
GO:: Gene Ontology
BP:: Biological process
MF:: Molecular function
CC:: Cellular component
TF :: Transcription Factor
KEGG:: Kyoto Encyclopedia of Genes and Genomes
DO:: Disease Ontology
GSEA:: Gene Set Enrichment Analysis
GSVA :: Gene Set Variation Analysis
LASSO:: Least absolute shrinkage and selection operator
RF:: Random Forest
SVM:: Support Vector Machine
WGCNA:: Weighted Gene Co-expression Network Analysis
AUC:: Area under the curve
STRING:: Search Tool for the Retrieval of Interacting Genes/Proteins
ENCORI:: Encyclopedia of RNA Interactomes
CTD:: Comparative Toxicogenomics Database
RT-qPCR:: Real-time fluorescence quantitative PCR
TLR4:: Toll-like receptor 4
ADIPOR1:: Adiponectin receptor 1
GOS2:: G0/G1 switch 2
HP:: Haptoglobin

References

Markousis-Mavrogenis G. Neurocardiology and Atherosclerosis: the effect of ethnic differences on heart-brain interaction. J Neurosci Rural Pract. 2017;8(4):504–5.
Article PubMed PubMed Central Google Scholar
Min J, Young G, Umar A, Kampfschulte A, Ahrar A, Miller M, et al. Neurogenic cardiac outcome in patients after acute ischemic stroke: the brain and heart connection. J Stroke Cerebrovasc Dis. 2022;31(12):106859.
Article PubMed Google Scholar
Mooe T, Eriksson P, Stegmayr B. Ischemic stroke after acute myocardial infarction a population-based study. Stroke. 1997;28(4):762–7.
Article CAS PubMed Google Scholar
Liao W, He C, Yang S, Zhou M, Zeng C, Luo M, et al. Bioinformatics and experimental analyses of glutamate receptor and its targets genes in myocardial and cerebral ischemia. BMC Genomics. 2023;24(1):300.
Article CAS PubMed PubMed Central Google Scholar
Liu ZY, Hu S, Zhong QW, Tian CN, Ma HM, Yu JJ. N-Methyl-D-aspartate receptor-driven calcium influx potentiates the adverse effects of myocardial ischemia-reperfusion injury ex vivo. J Cardiovasc Pharmacol. 2017;70(5):329–38.
Article CAS PubMed PubMed Central Google Scholar
Liu ZY, Zhong QW, Tian CN, Ma HM, Yu JJ, Hu S. NMDA receptor-driven calcium influx promotes ischemic human cardiomyocyte apoptosis through a p38 MAPK-mediated mechanism. J Cell Biochem. 2019;120(4):4872–82.
Article CAS PubMed Google Scholar
Li W, Feng G, Gauthier JM, Lokshina I, Higashikubo R, Evans S, et al. Ferroptotic cell death and TLR4/Trif signaling initiate neutrophil recruitment after heart transplantation. J Clin Investig. 2019;129(6):2293–304.
Article PubMed PubMed Central Google Scholar
Liu J, Zhou Y, Xie C, Li C, Ma L, Zhang Y. Anti-ferroptotic effects of bone marrow mesenchymal stem cell-derived extracellular vesicles loaded with ferrostatin-1 in cerebral ischemia-reperfusion injury associate with the GPX4/COX-2 Axis. Neurochem Res. 2023;48(2):502–18.
Article CAS PubMed Google Scholar
Liu C, Li Z, Xi H. Bioinformatics analysis and in vivo validation of ferroptosis-related genes in ischemic stroke. Front Pharmacol. 2022;13:940260.
Article CAS PubMed PubMed Central Google Scholar
Wang Z, He Z, Xuan Q, Zhang Y, Xu J, Lin J, et al. Analysis of the potential ferroptosis mechanism and multitemporal expression change of central ferroptosis-related genes in cardiac ischemia-reperfusion injury. Front Physiol. 2022;13:934901.
Article PubMed PubMed Central Google Scholar
Miao M, Cao S, Tian Y, Liu D, Chen L, Chai Q, et al. Potential diagnostic biomarkers: 6 cuproptosis- and ferroptosis-related genes linking immune infiltration in acute myocardial infarction. Genes Immun. 2023;24(4):159–70.
Article CAS PubMed PubMed Central Google Scholar
Park HJ, Noh JH, Eun JW, Koh YS, Seo SM, Park WS, et al. Assessment and diagnostic relevance of novel serum biomarkers for early decision of ST-elevation myocardial infarction. Oncotarget. 2015;6(15):12970–83.
Article PubMed PubMed Central Google Scholar
Muse ED, Kramer ER, Wang H, Barrett P, Parviz F, Novotny MA, et al. A whole blood molecular signature for acute myocardial infarction. Sci Rep. 2017;7(1):12268.
Article PubMed PubMed Central Google Scholar
Barrett T, Wilhite SE, Ledoux P, Evangelista C, Kim IF, Tomashevsky M, et al. NCBI GEO: archive for functional genomics data sets–update. Nucleic Acids Res. 2013;41(Database issue):D991-95.
CAS PubMed Google Scholar
Krug T, Gabriel JP, Taipa R, Fonseca BV, Domingues-Montanari S, Fernandez-Cadenas I, et al. TTC7B emerges as a novel risk factor for ischemic stroke through the convergence of several genome-wide approaches. J Cereb Blood Flow Metab. 2012;32(6):1061–72.
Article CAS PubMed PubMed Central Google Scholar
Barr TL, Conley Y, Ding J, Dillman A, Warach S, Singleton A, et al. Genomic biomarkers and cellular pathways of ischemic stroke by RNA gene expression profiling. Neurology. 2010;75(11):1009–14.
Article CAS PubMed PubMed Central Google Scholar
Stelzer G, Rosen N, Plaschkes I, Zimmerman S, Twik M, Fishilevich S, et al. The GeneCards Suite: from gene data mining to disease genome sequence analyses. Curr Protoc Bioinformatics. 2016;54:1.30.1-1.30.33.
Article PubMed Google Scholar
Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene ontology: tool for the unification of biology. Nat Genet. 2000;25(1):25–9.
Article CAS PubMed PubMed Central Google Scholar
Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47.
Article PubMed PubMed Central Google Scholar
Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 2000;28(1):27–30.
Article CAS PubMed PubMed Central Google Scholar
Yu G, Wang LG, Han Y, He QY. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16(5):284–7.
Article CAS PubMed PubMed Central Google Scholar
Wu T, Hu E, Xu S, Chen M, Guo P, Dai Z, et al. clusterProfiler 4.0: a universal enrichment tool for interpreting omics data. Innovation (N Y). 2021;2(3):100141.
CAS Google Scholar
Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 2005;102(43):15545–50.
Article CAS PubMed PubMed Central Google Scholar
Liberzon A, Birger C, Thorvaldsdottir H, Ghandi M, Mesirov JP, Tamayo P. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst. 2015;1(6):417–25.
Article CAS PubMed PubMed Central Google Scholar
Hanzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. 2013;14:7.
Article PubMed PubMed Central Google Scholar
Lee K, Jeong HO, Lee S, Jeong WK. CPEM: Accurate cancer type classification based on somatic alterations using an ensemble of a random forest and a deep neural network. Sci Rep. 2019;9(1):16927.
Article PubMed PubMed Central Google Scholar
Lu TX, Rothenberg ME. MicroRNA. J Allergy Clin Immunol. 2018;141(4):1202–7.
Article CAS PubMed Google Scholar
Mandrekar JN. Receiver operating characteristic curve in diagnostic test assessment. J Thorac Oncol. 2010;5(9):1315–6.
Article PubMed Google Scholar
Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77.
Article PubMed PubMed Central Google Scholar
Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559.
Article PubMed PubMed Central Google Scholar
Zhou J, Xiong W, Wang Y, Guan J. Protein function prediction based on PPI networks: network reconstruction vs edge enrichment. Front Genet. 2021;12:758131.
Article CAS PubMed PubMed Central Google Scholar
Barbie DA, Tamayo P, Boehm JS, Kim SY, Moody SE, Dunn IF, et al. Systematic RNA interference reveals that oncogenic KRAS-driven cancers require TBK1. Nature. 2009;462(7269):108–12.
Article CAS PubMed PubMed Central Google Scholar
Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13(11):2498–504.
Article CAS PubMed PubMed Central Google Scholar
Chin CH, Chen SH, Wu HH, Ho CW, Ko MT, Lin CY. cytoHubba: identifying hub objects and sub-networks from complex interactome. BMC Sys Biol. 2014;8 Suppl 4(Suppl 4):S11.
Article Google Scholar
Milošević D, Medeiros AS, Stojković Piperac M, Cvijanović D, Soininen J, Milosavljević A, et al. The application of Uniform Manifold Approximation and Projection (UMAP) for unconstrained ordination and classification of biological indicators in aquatic ecology. Sci Total Environ. 2022;815:152365.
Article PubMed Google Scholar
Newman AM, Steen CB, Liu CL, Gentles AJ, Chaudhuri AA, Scherer F, et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat Biotechnol. 2019;37(7):773–82.
Article CAS PubMed PubMed Central Google Scholar
Pesenti C, Navone SE, Guarnaccia L, Terrasi A, Costanza J, Silipigni R, et al. The genetic landscape of human glioblastoma and matched primary cancer stem cells reveals intratumour similarity and intertumour heterogeneity. Stem cells international. 2019;2019:2617030.
Article PubMed PubMed Central Google Scholar
Yoshihara K, Shahmoradgoli M, Martínez E, Vegesna R, Kim H, Torres-Garcia W, et al. Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun. 2013;4:2612. https://doi.org/10.1038/ncomms3612.
Article CAS PubMed Google Scholar
Yu G. Gene ontology semantic similarity analysis using GOSemSim. Methods in molecular biology (Clifton, NJ). 2020;2117:207–15.
Article CAS Google Scholar
Elmelliti H, Vahedian-Azimi A, Albazoon F, Alqahwachi H, Akbar A, Shehatta AL, et al. Outcomes of patients with in- and out-of-hospital cardiac arrest on extracorporeal cardiopulmonary resuscitation: a single-center retrospective cohort study. Curr Probl Cardiol. 2023;48(5):101578.
Article PubMed Google Scholar
Belohlavek J, Smalcova J, Rob D, Franek O, Smid O, Pokorna M, et al. Effect of intra-arrest transport, extracorporeal cardiopulmonary resuscitation, and immediate invasive assessment and treatment on functional neurologic outcome in refractory out-of-hospital cardiac arrest: a randomized clinical trial. JAMA. 2022;327(8):737–47.
Article PubMed PubMed Central Google Scholar
Zhang Q, Jia M, Wang Y, Wang Q, Wu J. Cell death mechanisms in cerebral ischemia-reperfusion injury. Neurochem Res. 2022;47(12):3525–42.
Article CAS PubMed Google Scholar
Sposato LA, Lam M, Allen B, Richard L, Shariff SZ, Saposnik G. First-ever ischemic stroke and increased risk of incident heart disease in older adults. Neurology. 2020;94(15):e1559–70.
Article PubMed Google Scholar
Jiang X, Stockwell BR, Conrad M. Ferroptosis: mechanisms, biology and role in disease. Nat Rev Mol Cell Biol. 2021;22(4):266–82.
Article PubMed PubMed Central Google Scholar
Fang X, Wang H, Han D, Xie E, Yang X, Wei J, et al. Ferroptosis as a target for protection against cardiomyopathy. Proc Natl Acad Sci USA. 2019;116(7):2672–80.
Article CAS PubMed PubMed Central Google Scholar
Tang LJ, Luo XJ, Tu H, Chen H, Xiong XM, Li NS, et al. Ferroptosis occurs in phase of reperfusion but not ischemia in rat heart following ischemia or ischemia/reperfusion. Naunyn Schmiedebergs Arch Pharmacol. 2021;394(2):401–10.
Article CAS PubMed Google Scholar
Tuo QZ, Lei P, Jackman KA, Li XL, Xiong H, Li XL, et al. Tau-mediated iron export prevents ferroptotic damage after ischemic stroke. Mol Psychiatry. 2017;22(11):1520–30.
Article CAS PubMed Google Scholar
Guan X, Li X, Yang X, Yan J, Shi P, Ba L, et al. The neuroprotective effects of carvacrol on ischemia/reperfusion-induced hippocampal neuronal impairment by ferroptosis mitigation. Life Sci. 2019;235:116795.
Article CAS PubMed Google Scholar
Huang G, Yao D, Yan X, Zheng M, Yan P, Chen X, et al. Emerging role of toll-like receptors signaling and its regulators in preterm birth: a narrative review. Arch Gynecol Obstet. 2023;308(2):319–39.
Article CAS PubMed Google Scholar
Wang H, Zhang S, Xie L, Zhong Z, Yan F. Neuroinflammation and peripheral immunity: Focus on ischemic stroke. Int Immunopharmacol. 2023;120:110332.
Article CAS PubMed Google Scholar
Liu Y, Li L, Wang Z, Zhang J, Zhou Z. Myocardial ischemia-reperfusion injury; molecular mechanisms and prevention. Microvasc Res. 2023;149:104565.
Article PubMed Google Scholar
Vaez H, Soraya H, Garjani A, Gholikhani T. Toll-Like Receptor 4 (TLR4) and AMPK Relevance in cardiovascular disease. Adv Pharm Bull. 2023;13(1):36–47.
CAS PubMed Google Scholar
Gawali B, Sridharan V, Krager KJ, Boerma M, Pawar SA. TLR4-a pertinent player in radiation-induced heart disease? Genes. 2023;14(5):1002.
Article CAS PubMed PubMed Central Google Scholar
Durán-Laforet V, Peña-Martínez C, García-Culebras A, Alzamora L, Moro MA, Lizasoain I. Pathophysiological and pharmacological relevance of TLR4 in peripheral immune cells after stroke. Pharmacol Ther. 2021;228:107933.
Article PubMed Google Scholar
Lv B, Shen N, Cheng Z, Chen Y, Ding H, Yuan J, et al. Strategies for Biomaterial-Based Spinal Cord Injury Repair via the TLR4-NF-κB Signaling Pathway. Front Bioeng Biotechnol. 2021;9:813169.
Article PubMed Google Scholar
Chen G, Li L, Tao H. bioinformatics identification of ferroptosis-related biomarkers and therapeutic compounds in ischemic stroke. Front Neurol. 2021;12: 745240.
Article PubMed PubMed Central Google Scholar
Gao Y, Liu H, Zhou J, Guo M, Sun J, Duan M. The protective effect OF C23 in a rat model of cardiac arrest and resuscitation. Shock (Augusta, Ga). 2023;59(6):892–901.
Article CAS PubMed Google Scholar
Ren C, Yi W, Jiang B, Gao E, Liang J, Zhang B, et al. Diminished AdipoR1/appl1 interaction mediates reduced cardioprotective actions of adiponectin against myocardial ischemia/reperfusion injury in Type-2 diabetic mice. Stem Cells Int. 2023;2023:7441367.
Article PubMed PubMed Central Google Scholar
He Y, Liu B, Yao P, Shao Y, Cheng Y, Zhao J, et al. Adiponectin inhibits cardiac arrest/cardiopulmonary resuscitation-induced apoptosis in brain by increasing autophagy involved in AdipoR1-AMPK signaling. Mol Med Rep. 2020;22(2):870–8.
Article CAS PubMed PubMed Central Google Scholar
Liu B, Liu J, Wang J, Sun F, Jiang S, Hu F, et al. Adiponectin Protects Against Cerebral Ischemic Injury Through AdipoR1/AMPK Pathways. Front Pharmacol. 2019;10:597.
Article CAS PubMed PubMed Central Google Scholar
Wang X, Meng H, Ruan J, Chen W, Meng F. Low G0S2 gene expression levels in peripheral blood may be a genetic marker of acute myocardial infarction in patients with stable coronary atherosclerotic disease: a retrospective clinical study. Medicine. 2021;100(3):e23468.
Article CAS PubMed PubMed Central Google Scholar
Kamikubo K, Kato H, Kioka H, Yamazaki S, Tsukamoto O, Nishida Y, et al. A molecular triage process mediated by RING finger protein 126 and BCL2-associated athanogene 6 regulates degradation of G(0)/G(1) switch gene 2. J Biol Chem. 2019;294(40):14562–73.
Article CAS PubMed PubMed Central Google Scholar
Zhang T, Xiang L. Elevated plasma haptoglobin level as a potential marker for poor prognosis in acute cerebral infarction. Eur Neurol. 2018;79(3–4):154–60.
Article PubMed Google Scholar
Brea D, Sobrino T, Blanco M, Fraga M, Agulla J, Rodríguez-Yáñez M, et al. Usefulness of haptoglobin and serum amyloid A proteins as biomarkers for atherothrombotic ischemic stroke diagnosis confirmation. Atherosclerosis. 2009;205(2):561–7.
Article CAS PubMed Google Scholar
Cahill LE, Levy AP, Chiuve SE, Jensen MK, Wang H, Shara NM, et al. Haptoglobin genotype is a consistent marker of coronary heart disease risk among individuals with elevated glycosylated hemoglobin. J Am Coll Cardiol. 2013;61(7):728–37.
Article CAS PubMed PubMed Central Google Scholar
Somer S, Levy AP. The role of haptoglobin polymorphism in cardiovascular disease in the setting of diabetes. Int J Mol Sci. 2020;22(1):287.
Article PubMed PubMed Central Google Scholar
Merkler A, Sertić J, Bazina Martinović A, Križ T, Miličić I, Šimić M, et al. Haptoglobin genotype 2–2 associated with atherosclerosis in patients with ischemic stroke. Gene. 2020;752:144786.
Article CAS PubMed Google Scholar
Morimoto M, Nakano T, Egashira S, Irie K, Matsuyama K, Wada M, et al. Haptoglobin regulates macrophage/microglia-induced inflammation and prevents ischemic brain damage via binding to HMGB1. J Am Heart Assoc. 2022;11(6):e024424.
Article CAS PubMed PubMed Central Google Scholar
Xing Z, He Q, Xiong Y, Zeng X. Systematic Pharmacology Reveals the Antioxidative Stress and Anti-Inflammatory Mechanisms of Resveratrol Intervention in Myocardial Ischemia-Reperfusion Injury. Evidence-based complementary and alternative medicine : eCAM. 2021;2021:5515396.
Article PubMed Google Scholar
Jain S. Role of interleukin-17 signaling pathway in the interaction between multiple sclerosis and acute myocardial infarction. Multiple sclerosis and related disorders. 2022;58:103515.
Article CAS PubMed Google Scholar
Li J, Hao M, Yang B, Shi T, Zhang Y, Feng J, et al. Long non-coding RNAs expression profile and functional analysis of acute ischemic stroke. Medicine. 2020;99(50):e22964.
Article CAS PubMed PubMed Central Google Scholar
Kawano T, Shimamura M, Nakagami H, Kanki H, Sasaki T, Mochizuki H. Temporal and spatial profile of polymorphonuclear myeloid-derived suppressor cells (PMN-MDSCs) in ischemic stroke in mice. PLoS ONE. 2019;14(5):e0215482.
Article PubMed PubMed Central Google Scholar
He B, Quan LP, Cai CY, Yu DY, Yan W, Wei QJ, et al. Dysregulation and imbalance of innate and adaptive immunity are involved in the cardiomyopathy progression. Frontiers in cardiovascular medicine. 2022;9:973279.
Article CAS PubMed PubMed Central Google Scholar
Hofmann U, Frantz S. Role of lymphocytes in myocardial injury, healing, and remodeling after myocardial infarction. Circ Res. 2015;116(2):354–67.
Article CAS PubMed Google Scholar
He YJ, Cong L, Liang SL, Ma X, Tian JN, Li H, et al. Discovery and validation of Ferroptosis-related molecular patterns and immune characteristics in Alzheimer’s disease. Front Aging Neurosci. 2022;14:1056312.
Article CAS PubMed PubMed Central Google Scholar
Hu H, Feng Z, Lin H, Cheng J, Lyu J, Zhang Y, et al. Gene function and cell surface protein association analysis based on single-cell multiomics data. Comput Biol Med. 2023;157:106733.
Article CAS PubMed Google Scholar
Gao H, Sun J, Wang Y, Lu Y, Liu L, Zhao Q, et al. Predicting metabolite-disease associations based on auto-encoder and non-negative matrix factorization. Brief Bioinformatics. 2023;24(5):bbad259.
Article PubMed Google Scholar
Wang W, Zhang L, Sun J, Zhao Q, Shuai J. Predicting the potential human lncRNA-miRNA interactions based on graph convolution network with conditional random field. Brief Bioinformatics. 2022;23(6):bbac463.
Article PubMed Google Scholar
Zhang L, Yang P, Feng H, Zhao Q, Liu H. Using Network Distance Analysis to Predict lncRNA-miRNA Interactions. Interdiscip Sci. 2021;13(3):535–45.
Article CAS PubMed Google Scholar
Meng R, Yin S, Sun J, Hu H, Zhao Q. scAAGA: Single cell data analysis framework using asymmetric autoencoder with gene attention. Comput Biol Med. 2023;165:107414.
Article CAS PubMed Google Scholar
Ou M, Jiang Y, Ji Y, Zhou Q, Du Z, Zhu H, et al. Role and mechanism of ferroptosis in neurological diseases. Mol Metab. 2022;61:101502.
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

In this study, we would like to thank all the colleagues who took part in the study.

Funding

This study includes funding from the National Natural Science Foundation of China(No. 81960326), the Leading Talents Project of Jiangxi Provincial Science and the Technology Department's Training Plan for Major Disciplines and Technical Leaders(No. 20213BCJ0001), Jiangxi Human Audio Signal Intelligent Medical Engineering Research Center, the First Affiliated Hospital of Gannan Medical College Doctoral Initiation Program(QD202318), Science and Technology Projects of Jiangxi Provincial Education Department(GJ2201436).

Author information

Authors and Affiliations

Department of Neurosurgery, First Affiliated of Gannan Medical University, Ganzhou, Jiangxi, China
Wei Liao, Shaochun Yang & Chunming He
Key Laboratory of Prevention and Treatment of Cardiovascular and Cerebrovascular Diseases, Ministry of Education, Gannan Medical University, Ganzhou, Jiangxi, China
Wei Liao, Shaochun Yang, Yanyu Duan, Chunming He & Ziyou Liu
Shanghai Jiao Tong University School of Medicine, Shanghai, China
Yuehui Wen
Gannan Medical University, Ganzhou, Jiangxi, China
Chuan Zeng
Heart Medical Centre, First Affiliated of Gannan Medical University, Ganzhou, Jiangxi, China
Yanyu Duan & Ziyou Liu
Department of Cardiac Surgery, First Affiliated of Gannan Medical University, Ganzhou, Jiangxi, China
Ziyou Liu

Authors

Wei Liao
View author publications
You can also search for this author in PubMed Google Scholar
Yuehui Wen
View author publications
You can also search for this author in PubMed Google Scholar
Chuan Zeng
View author publications
You can also search for this author in PubMed Google Scholar
Shaochun Yang
View author publications
You can also search for this author in PubMed Google Scholar
Yanyu Duan
View author publications
You can also search for this author in PubMed Google Scholar
Chunming He
View author publications
You can also search for this author in PubMed Google Scholar
Ziyou Liu
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

ZL and CH designed the study. WL and YW collected research data. WL and YW performed data analysis. YW and CZ performed experimental validation. WL wrote the manuscript. SY and YD revised the manuscript. ZL and WL provided funding acquisition, all authors read and approved the final manuscript.

Corresponding authors

Correspondence to Chunming He or Ziyou Liu.

Ethics declarations

Ethics approval and consent to participate

The Gannan Medical University's Ethics Committee and Institutional Review Board approved this study with informed consent from all participants. During the course of the study, all methods were performed by the ethical standards of the institutional and/or national research committees, as well as with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1:

Table S1. GEO Dataset Information list.

Additional file 2:

Table S2. Ferroptosis-related genes.

Additional file 3:

Table S3. Primer sequences for cell experiments.

Additional file 4:

Table S4. Primer sequences for blood specimens.

Additional file 5:

Fig. S1. Data processing for merged data sets. (A) Merging of the MI datasets GSE60993 and GSE66360. (B) Removing batch effects between MI datasets. (C)A significant batch effect between the MI combined data. (D) Gene expression profiling data with consistent expression levels after treatment. (E) Merging of the IS datasets GSE22255和GSE16561. (F) Removing batch effects between IS datasets. (G)A significant batch effect between the IS combined data. (H) Gene expression profiling data with consistent expression levels after treatment. MI: Myocardial Infarction. IS: Ischemic Stroke.

Additional file 6:

Table S5. GO enrichment results of MI.

Additional file 7:

Table S6. KEGG enrichment results of MI.

Additional file 8:

Table S7. GO enrichment results of IS.

Additional file 9:

Table S8. KEGG enrichment results of IS.

Additional file 10:

Table S9. GSEA-MI.

Additional file 11:

Table S10. GSEA-IS.

Additional file 12:

Table S11. GSVA-MI.

Additional file 13:

Table S12. GSVA-IS.

Additional file 14:

Fig. S2. Construction of co-expression modules for MI and IS. (A) The optimal soft threshold for MI gene module scale-free R is 0.80 to ensure a non-scaling topology. (B) Plotting the scatter plot of gray module gene significance vs. module membership for MI. (C) Plotting the scatter plot of yellow module gene significance vs. module membership for MI. (D)The optimal soft threshold for IS gene module scale-free R is 0.80 to ensure a non-scaling topology. (E) Plotting the scatter plot of pink module gene significance vs. module membership for IS. (F) Plotting the scatter plot of blue module gene significance vs. module membership for IS. MI: Myocardial Infarction. IS: Ischemic Stroke.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article

Liao, W., Wen, Y., Zeng, C. et al. Integrative analyses and validation of ferroptosis-related genes and mechanisms associated with cerebrovascular and cardiovascular ischemic diseases. BMC Genomics 24, 731 (2023). https://doi.org/10.1186/s12864-023-09829-w

Download citation

Received: 08 August 2023
Accepted: 22 November 2023
Published: 04 December 2023
DOI: https://doi.org/10.1186/s12864-023-09829-w

Integrative analyses and validation of ferroptosis-related genes and mechanisms associated with cerebrovascular and cardiovascular ischemic diseases

Abstract

Background

Results

Conclusions

Introduction

Materials and methods

Data acquisition

Identification of differentially expressed genes (DEGs) associated with ferroptosis

Assessment of biological characteristics between disease and control samples

Construction of a diagnostic model for Ferroptosis

Weighted Gene Association Co-expression Network Analysis (WGCNA)

Building a network of protein-protein interaction (PPI), mRNA-miRNA, mRNA-RBP, mRNA-TF and mRNA-drugs interactions

Molecular subtypes of Ferroptosis

Exploration of immune infiltration

Modeling of cellular hypoxia-reoxygenation

Clinical blood specimen collection

Validation using real-time fluorescence quantitative PCR (RT-qPCR)

Statistical analysis

Results

Analysis of DEGs associated with ferroptosis

GSEA and GSVA enrichment

Co-expression modules for MI and IS

Shared gene and network analysis of MI and IS

Construction of a diagnostic model related to Ferroptosis

Immune infiltration analysis of combined datasets

Identification of patient subtypes by characterized genes

Variations in immune traits among subcategories

RT-qPCR validation of hub genes expression in cellular models and clinical disease blood specimens

Discussion

Conclusions

Availability of data and materials

Abbreviations

References

Acknowledgements

Funding

Author information

Authors and Affiliations

Contributions

Corresponding authors

Ethics declarations

Ethics approval and consent to participate

Consent for publication

Competing interests

Additional information

Publisher’s Note

Supplementary Information

Rights and permissions

About this article

Cite this article

Share this article

Keywords

BMC Genomics

Contact us