Unraveling the relevance of SARS-Cov-2 infection and ferroptosis within the heart of COVID-19 patients

Abstract

Background

The coronavirus disease 2019 (COVID-19) was caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which led to a huge mortality rate and imposed significant costs on the health system, causing severe damage to the cells of different organs such as the heart. However, the exact details and mechanisms behind this damage are not clarified. Therefore, we aimed to identify the cell and molecular mechanism behind the heart damage caused by SARS-Cov-2 infection.

Methods

RNA-seq data for COVID-19 patients’ hearts was analyzed to obtain differentially expressed genes (DEGs) and differentially expressed ferroptosis-related genes (DEFRGs). Then, DEFRGs were used for analyzing GO and KEGG enrichment, and perdition of metabolites and drugs. we also constructed a PPI network and identified hub genes and functional modules for the DEFRGs. Subsequently, the hub genes were validated using two independent RNA-seq datasets. Finally, the miRNA-gene interaction networks were predicted in addition to a miRNA-TF co-regulatory network, and important miRNAs and transcription factors (TFs) were highlighted.

Findings

We found ferroptosis transcriptomic alterations within the hearts of COVID-19 patients. The enrichment analyses suggested the involvement of DEFRGs in the citrate cycle pathway, ferroptosis, carbon metabolism, amino acid biosynthesis, and response to oxidative stress. IL6, CDH1, AR, EGR1, SIRT3, GPT2, VDR, PCK2, VDR, and MUC1 were identified as the ferroptosis-related hub genes. The important miRNAs and TFs were miR-124-3P, miR-26b-5p, miR-183-5p, miR-34a-5p and miR-155-5p; EGR1, AR, IL6, HNF4A, SRC, EZH2, PPARA, and VDR.

Conclusion

These results provide a useful context and a cellular snapshot of how ferroptosis affects cardiomyocytes (CMs) in COVID-19 patients’ hearts. Besides, suppressing ferroptosis seems to be a beneficial therapeutic approach to mitigate heart damage in COVID-19.

1. Introduction

COVID-19 is caused by the positive-sense RNA virus SARS-CoV-2, which contains a large viral genome of 30 kb. COVID-19 was first reported in China and caused a huge economic and mortality toll [[1], [2], [3], [4]]. Globally, more than 5.42 million confirmed fatalities and 14.83 million estimated COVID-19 fatalities have been reported by World Health Organization (WHO) up to 2023 [5]. After SARS-CoV-2 enters the host cell, RNAs begin to multiply and as a result, the corresponding proteins are expressed and assembled into complete viral particles. The particles are eventually released along with the cell contents and ultimately lead to cell death [6]. The lung is usually the first target tissue of the SARS-CoV-2 and causes respiratory tract difficulties. However, In severe cases, it can cover and damage almost all the vital organs of the body [7]. The data presented in China shows exclusively that the heart is one of the most important organs affected by SARS-CoV-2 and 20–30 % of hospitalizations are due to heart damage with a 40 % mortality rate [8]. Importantly, cardiovascular diseases (CVDs) are a potential health complication in COVID-19 especially in elderly people and in patients with underlying diseases [9,10]. Moreover, SARS-CoV-2 makes the infected hearts prone to CVDs. The mechanisms underlying the development of COVID-19-induced cardiovascular injury are not entirely known. Postulated mechanisms include direct infection of the myocardium and the resultant viral myocarditis, endotheliitis as a direct consequence of viral infection of the endothelial cells, autonomic dysfunction, plaque rupture and coronary spasm or microthrombi due to systematic inflammation or cytokine storm [[11], [12], [13], [14]]. In SARS-CoV-2 infections, like other febrile illnesses, an exaggerated pro-inflammatory response, high blood viscosity, multisystem inflammatory syndrome, and increased risk of heart failure (HF) development that is caused in part by cytokine storms are seen [9,15,16]. Furthermore, a programmed cell death (PCD) pathway called ferroptosis is accompanied by the cytokine storms present in COVID-19 patient heart tissues [17].

Ferroptosis is a relatively new iron-dependent PCD characterized by lipid peroxidation and iron overload, which affects multiple cellular organelles such as mitochondria, lysosome, endoplasmic reticulum (ER), and plasma membrane (PM) [18,19]. This pathway is involved in many diseases like cancers, pneumonia, Alzheimer's disease, CVDs, and COVID-19 [20]. In viral infections, cell death can be a double-edged sword [21]. This is because it can lead to the death of virus-infected cells, thereby preventing virus spread along with uncontrolled cell death that causes tissue damage leading to CVD and adverse immune response [17,20,22]. The evidence to date suggests that the development of a variety of CVDs is driven by ferroptosis [23]. For instance, during HF progression, TLR4-NOX4 pathway has been shown to develop cell death through autophagy and ferroptosis [24]. Another study recently uncovered that the human umbilical cord blood derived miR-23a-3p expressing exosomes, inhibit ferroptosis by targeting DMT1 and subsequently reduce myocardial injury in acute myocardial infraction mice [25]. One of the potentially major complications of SARS-CoV-2 infection is cardiovascular damage [26], however the specific mechanisms underlying ferroptosis involvement in CVDs associated with COVID-19 remain indefinite [27]. Therefore, a better understanding of the molecular mechanisms behind this phenomenon is needed to improve our understanding and treatment options of SARS-CoV-2 infected hearts, as well as to prevent subsequent CVDs. Bioinformatics methods including RNA-seq analysis and its subsequent downstream processing methods have been successfully used to address issues related to diseases affected by ferroptosis and clarify the relevant cellular and molecular mechanisms behind the involvement of ferroptosis which can lead to improvements in treatment options and early detection [28,29].

Therefore, in this study, we aimed to unravel the ferroptosis-related cellular and molecular mechanisms affecting SARS-CoV-2 infected hearts. To achieve this objective a publicly available RNA-seq dataset was analyzed to obtain differentially expressed genes (DEGs). Differentially expressed ferroptosis-related genes (DEFRGs) were also found and subjected to downstream enrichment analyses for biological interpretation and biological context comprehension. Subsequently, hub genes and functional modules were predicted. Next, the regulatory network governing the DEFRGs’ PPI network was constructed. Finally, the compounds affecting the expression of DEFRGs and diseases relevant to the DEFRG list were predicted. Our results provide a potential mechanistic view of how ferroptosis-related genes impact the heart of COVID-19 patients, and most importantly how this subject can be further studied via different experimental approaches.

2. Methods

The workflow of the bioinformatics analyses used to explore the ferroptosis pathway in SARS-CoV-2 infected human heart tissue is shown in Fig. 1.

Fig. 1.

Flow chart of the study. GEO: Gene Expression Omnibus; DEGs: differentially expressed genes; FRGs: ferroptosis related genes; DEFRGs: differentially expressed ferroptosis-related genes; PPI: protein-protein interaction; GO: Gene Ontology; KEGG: Kyoto Encyclopedia of Genes, and Genomes.

2.1. Data acquisition and processing

Three RNA-Seq datasets (GSE169241, GSE151879, and GSE156754) were retrieved from the Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo) database. GSE169241; the training dataset used in this research, is based on GPL24676 Illumina NovaSeq 6000, and includes 23 samples, of which 3 COVID-19 patient heart autopsy samples, and 5 non-COVID-19 controls were involved in this study [30]. It should be added that although this dataset included some demographic features such as age, gender and duration of symptoms (days), they were not provided for each sample separately, thus it was not possible to evaluate the association of the ferroptosis pathway in SARS-CoV-2 infected human heart tissue to these features. Two external datasets GSE151879, and GSE156754, applying high throughput sequencing with Illumina NovaSeq 6000 (GPL246776), and Illumina NextSeq 500 (GPL18573) platforms for RNA sequence extraction, were used as external validation datasets [31,32]. From the 18 samples in GSE151879, three adult human CMs infected with SARS-CoV-2, and three uninfected human CMs were included in this research. Similarly, among the 30 samples of the GSE156754, three human induced pluripotent stem cells (iPSCs)-derived CMs infected with SARS-CoV-2 at a high multiplicity of infection (MOI), and three mock infection CMs were used in the validation phase.

The “DESeq2” (v 1.38.3) package in R software (v 4.2.2) was utilized to screen DEGs according to the screening criteria of adjusted P-value <0.05, and |logFC| ≥ 1 [33,34]. A total of 565 ferroptosis genes including drivers, suppressors, markers, and unclassified regulators were acquired from FerrDb (http://www.zhounan.org/ferrdb/current/). Subsequently, the DEGs were intersected with these ferroptosis genes to identify the differentially expressed ferroptosis-related genes (DEFRGs) between COVID-19 patients, and non-COVID samples, with a multiple list comparator tool (https://molbiotools.com/listcompare.php). “EnhancedVolcano” (v 1.16.0), and “pheatmap” (v1.0.12) packages of R software were used to generate the volcanoplot, and heatmap, respectively [35,36].

2.2. Functional enrichment analysis of DEFRGs

To explore the biological functions, and related pathways of the DEFRGs, Gene ontology (GO), and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed, using the Database for Annotation, Visualization, and Integrated Discovery (DAVID, https://david.ncifcrf.gov) [37]. GO analysis included Biological Process (BP), Cellular Component (CC), and Molecular Function (MF) analyses. Visualization of the top ten results of the analysis was carried out using SRplot, a free online platform for data visualization (https://www.bioinformatics.com.cn/en). P-value <0.05 was considered statistically significant.

2.3. Prediction of the DEFRGs-related metabolites and drugs

The Enrichr dataset-linked Human Metabolome Database (HMDB) was used to find DEFRGs-related metabolites (https://maayanlab.cloud/Enrichr/enrich) [38]. HMDB is a comprehensive metabolomics resource containing human metabolite information, as well as its associated enzymes, transporters, and disease-related properties [39]. Top ten metabolites related to the DEFRGs were selected and ranked according to P-value (P-value <0.05). Similarly, to find the top ten drugs/compounds that possibly target the DEFRGs, the Enrichr-linked Drug Signature Database (DSigDB) was used. DSigDB is a globally recognized gene set resource for translational research, and drug repurposing studies that link drugs/compounds to their target genes [40].

2.4. Construction of the protein-protein interaction (PPI) network and DEFRG module analysis

Interactions between the DEFRGs were predicted using the StringApp (v 2.0.1), which imports both physical interactions as well as functional associations among protein-protein pairs from STRING into Cytoscape (v 3.10.0) [41,42]. The cut-off criteria were set as a Confidence score ≥ o.4. The resulting PPI network was visualized with the Cytoscape application boundaryLayout (v 1.1); a Cytoscape app for visualizing force-directed networks within custom boundaries [43]. To derive the most probable cellular location of the DEFRGs, three databases, namely: COMPARTMENTS [44], UniProt [45], and Gene Ontology (GO) [46] were used.

The Molecular complex detection (MCODE) (v 2.0.2) plugin in Cytoscape was used to screen out the important functional modules in the PPI network of the DEFRGs [47]. The default parameters were set as follows: Degree Cutoff: 2, Node Score Cutoff: 0.2, K-Core: 2, and Max Depth: 100. Exploring the information about the genetic basis of human diseases is essential for drug discovery, and precision medicine [48]. DisGeNET is a comprehensive repository of genes, and variants associated with human diseases [48]. To illustrate the diseases and complications associated with the MCODE clusters of the DEFRGs PPI network, we used DisGeNET via NetworkAnalyst (https://www.networkanalyst.ca/) [49]. Degree filter >2 was considered as the cut-off value.

2.5. Hub DEFRG identification, and correlation analysis

The CytoHubba (v 0.1) plugin of Cytoscape was used to select the top 20 hub DEFRGs based on the Maximum Clique Centrality [50]. The correlation analysis of the 20 hub DEFRGs was carried out using the Spearman correlation in the “corrplot” (v 0.92) package of R software. P-value <0.05 was considered statistically significant [51].

2.6. miRNA-DEFRG regulatory network

The upstream miRNAs that potentially target the top 20 hub DEFRGs, were predicted using miRTarBase (v 8.0) via the NetworkAnalyst (v 3.0) platform [52]. Subsequently, the miRNA-DEFRG regulatory network was filtered with a degree value of greater than 2 and visualized utilizing Cytoscape.

2.7. Construction of the TF-miRNA co-regulatory network

The NetworkAnalyst-linked RegNetwork database was used to construct the TF-miRNA co-regulatory network, consisting of transcription factors (TFs), and miRNAs that potentially regulate the expression of the hub DEFRGs at the transcriptional, and the post-transcriptional level [53]. The TF-miRNA co-regulatory network was filtered with a degree value of greater than 4 and visualized utilizing Cytoscape.

2.8. Validation of the hub DEFRGs

The expression patterns of the hub DEFRGs were verified using two external datasets, GSE156754, and GSE151879. Box plots were generated, to visualize the expression levels of the hub DEFRGs between CMs infected with SARS-CoV-2, and uninfected CMs, using “tidyr” (v 1.3.0) [54], “ggpubr” (v 0.6.0) [55], and “rstatix” (v 0.7.0) [56] packages in R software. P-value <0.05 was considered significant.

2.9. Statistical analysis

Statistical analysis was performed using the R software (v 4.2.2). In all analysis, P-value <0.05 was considered significant.

3. Results

3.1. Identification of heart tissue DEFRGs between COVID-19 patients, and non-COVID-19 samples

The RNA-Seq data of GSE169241 was downloaded from GEO. A total of 4049 DEGs were obtained comparing COVID-19 patients, and non-COVID-19 samples and 136 DEFRGs were identified (Fig. 2A & Supplementary File 1). The volcano plot visualized the distribution of the DEFRGs among the DEGs (Fig. 2B). Heatmap analysis revealed the standardized expression pattern of these 136 DEFRGs (79 up-regulated, and 57 down-regulated) (Fig. 2C).

Fig. 2.

Differentially expressed ferroptosis related genes (DEFRGs) between COVID-19 infected heart tissues, and non-infected controls. (A) Venn diagram showing the intersection between the differentially expressed genes (DEGs) in GSE169241, and ferroptosis genes. (B) Volcano plot showing the distribution of DEFRGs among the DEGs in COVID-19 patients, and non-COVID controls. (C) Heatmap showing the expression pattern of the 136 DEFRGs.

3.2. Gene ontology, and KEGG pathway enrichment analyses of the DEFRGs

To explore the underlying mechanisms of ferroptosis in COVID-19 patients compared to non-COVID-19 samples, we used the DAVID database for GO, and KEGG pathway enrichment analyses of the 136 DEFRGs. The results showed that these genes were significantly enriched in the regulation of the circadian rhythm, protein ADP-ribosylation, cellular response to oxidative stress, negative regulation of fat cell differentiation, and negative regulation of apoptotic process under the BP category (Fig. 3A). The GO-CC enrichment analysis indicated that these genes were mostly distributed in different structures of cells including cytosol, nucleus, and the mitochondrion (Fig. 3B). In the MF category, the DEFRGs were mainly involved in NAD + ADP-ribosyltransferase activity, sequence-specific DNA binding, protein ADP-ribosylase activity, protein homodimerization activity, and zinc ion binding (Fig. 3C). Moreover, KEGG pathway analysis revealed that the DEFRGs were significantly enriched in oxocarboxylic acid metabolism, biosynthesis of amino acids, carbon metabolism, citrate cycle (TCA cycle), and ferroptosis (Fig. 3D).

Fig. 3.

Top 10 GO (Gene Ontology) terms, and Kyoto Encyclopedia of Genes, and Genomes (KEGG) pathways of the 136 DEFRGs between COVID-19 patients, and non-COVID controls. (A) Top 10 enriched terms of GO: biological process. (B) Top 10 enriched terms of GO: cellular component. (C) Top 10 enriched terms of GO: molecular function. (D) Top 10 enriched KEGG pathways (P-value <0.05).

3.3. Identification of DEFRGs-associated metabolites, and candidate drugs

TOP 10 drugs targeting the 136 DEFRGs were retrieved from Enrichr-linked DSigDB. These potential drugs include pyrvinium, astemizole, thioridazine, alexidine, trimipramine, Nickel sulfate, methylbenzethonium chloride, prenylamine, and niclosamide (Supplementary Table 1). Among the identified drug, Nickel sulfate, interacted with the highest number of DEFRGs.

Moreover, the analysis of the DEFRGs using the Enrichr-linked HMDB showed that they are linked to various metabolites. Top ten metabolites associated with the DEFRGs were Iron, Oxoglutaric acid, NAD, Pyridoxal 5′-phosphate, Famotidine, L-Cysteine, NADPH, L-glutamic acid, Niacinamide, and Coenzyme A (Supplementary Table 2).

3.4. PPI network construction, DEFRGs module analysis, and identification of gene-disease associations

To further explore the potential interactions among the DEFRGs, the PPI network was generated via the StringApp plugin in Cytoscape. Each DEFRG was assigned a subcellular location, to help assess the roles of the proteins in the network, and the PPI network was visualized using boundaryLayout in Cytoscape. After removing the isolated nodes, the constructed network contained 118 nodes, and 394 edges (Fig. 4). The DEFRGs were mostly categorized in the nucleus (26 nodes), cytosol (26 nodes), and mitochondrion (24 nodes). Moreover, the top DEFRGs with the highest degree of connectivity were IL6 (extracellular space), PPARA (nucleus), SIRT3 (mitochondrion), SRC (cytosol), CS (mitochondrion), and TXN (cytosol), respectively.

Fig. 4.

The protein-protein interaction network of the 136 DEFRGs. The DEFRGs are categorized into a cell tem plate, based on the assigned cellular location, using the Cytoscape plug-in boundaryLayout. The size of a node is proportional to the degree of the node.

Using the MCODE plugin in Cytoscape, six clusters were identified from the PPI network, comprising 42 genes (Fig. 5A). Through the analysis of gene-disease association by NetworkAnalyst, we found that the diseases: muscle hypotonia, autosomal recessive predisposition, and prostatic neoplasms have the highest correlations with the six identified clustered DEFRGs. Notably, the interaction network revealed that IL6, NOX4, and ADIPOQ are associated with heart failure. Moreover, IL6, EGR1, and ADIPOQ were linked to inflammation (Fig. 5B). These results show that IL6, and ADIPOQ are linked to both COVID-19, and heart failure.

Fig. 5.

(A) Six MCODE modules (clusters) extracted from the protein-protein interaction (PPI) network of the 136 DEFRGs. Darker nodes indicate higher MCODE score. (B) Gene-disease association network of the six DEFRG clusters (degree >2). (C) The PPI network of the top 20 hub DEFRGs. Darker nodes show higher MCC score. (D) Spearman correlation analysis of the 20 hub DEFRGs. Blue, and red represent positive, and negative correlation, respectively (P-value <0.05). Circles with a deeper color indicate a stronger correlation index. The numbers in the plot represent spearman correlation coefficient. Blank represents not significant coefficient (P-value <0.05).

3.5. Screening, correlation analysis, and validation of the hub DEFRGs

The MCC algorithm of CytoHubba plugin in Cytoscape was applied to screen out, and visualize the top 20 hub DEFRGs, including Interleukin 6 (IL6), Proto-oncogene tyrosine-protein kinase Src (SRC), Cadherin-1 (CDH1), Androgen receptor (AR), Hepatocyte nuclear factor 4 alpha (HNF4A), Enhancer of zeste homolog 2 (EZH2), Early growth response protein 1 (EGR1), glycogen synthase kinase-3 beta protein (GSK3B), Fumarate hydratase (FH), Mucin short variant S1 (MUC1), Glucose-6-phosphate dehydrogenase (G6PD), Citrate Synthase (CS), glutamic-oxaloacetic transaminase 1 (GOT1), Peroxisome Proliferator Activated Receptor Alpha (PPARA), Phosphoenolpyruvate Carboxykinase 2 (PCK2), Vitamin D receptor (VDR), Glutamic--pyruvic transaminase 2 (GPT2), Isocitrate dehydrogenase (IDH2), NAD-dependent deacetylase sirtuin-3 (SIRT3), and Asparagine synthetase (ASNS) (Fig. 5C).

Subsequently, spearman correlation analysis was performed to decipher the association between the expression levels of the 20 hub DEFRGs. The findings suggested that there was an interaction between the hub DEFRGs. Among them, PPARA had the highest significant positive correlation with CS, AR, GSK3B, and GOT1 (correlation coefficient = 0.98), while, FH and CDH1 had the highest significant negative correlation (correlation coefficient = - 0.95) (Fig. 5D).

3.6. Analysis of the miRNA-DEFRGs regulatory network

To investigate the post-transcriptional regulatory network between miRNAs, and the DEFRGs, the miRNA-gene interaction network was constructed, based on the 20 hub DEFRGs. The resulting network consisted of 36 nodes and 72 edges. Among the predicted miRNAs, miR-124-3P, miR-26b-5p, miR-183-5p, miR-34a-5p, and miR-155-5p had the highest number of interactions, with 7, 7, 5, 5, and 5 target DEFRGs, respectively (Supplementary Table 3). Moreover, GSK3B, EZH2, IL6, and AR were targeted by the highest number of miRNAs with 10, 9, 7, and 6 interactions, respectively (Fig. 6A).

Fig. 6.

(A) The miRNA-hub DEFRG interaction network (degree >2). The round nodes represent the hub DEFRGs, and the octagon nodes indicate the miRNAs. The size of the nodes is proportional to the degree of the node. DEFRGs with darker colors are regulated by higher number of miRNAs. (B) The TF-miRNA co-regulatory network of the hub DEFRGs (degree >4). The purple octagon nodes show the miRNAs, the blue diamond nods represent the TFs, and the rest of the nodes represent the hub DEFRGs. EGR1, AR, IL6, HNF4A, SRC, EZH2, PPARA, and VDR act both as hub DEFRGs, and TFs in this network. The size, and the shade of the nodes are proportional to the degree of the node.

3.7. Analysis of the TF-miRNA co-regulatory network

To explore the important regulatory TFs and miRNAs that modulate the expression of the hub DEFRGs, the miRNA-TF co-regulatory network was constructed. The regulatory network is comprised of 46 nodes and 157 edges. Interestingly, the results showed that EGR1, AR, IL6, HNF4A, SRC, EZH2, PPARA, and VDR act both as hub DEFRGs, and TFs in this network (Fig. 6B).

3.8. Validation of the hub DEFRGs

The expression levels of the 20 hub DEFRGs were eventually verified in two external datasets; GSE156754, and GSE151879 (P-value <0.05 was considered significant). Consistent with the results obtained from the training cohort, EGR1, and MUC1 expression levels were significantly up-regulated, whereas SIRT3 and AR expression levels were significantly down-regulated in CMs infected with SARS-CoV-2 compared with the mock-infection group in GSE156754 (Fig. 7A). In addition, PCK2, VDR, CDH1, GPT2, and IL6 expression levels were significantly up-regulated in adult human CMs infected with SARS-CoV-2 compared with the uninfected human CMs in GSE151879 (Fig. 7B).

Fig. 7.

Validation of the of the hub DEFRGs in two external datasets (GSE156754, and GSE 151879). (A) Box plots depicting expression levels of the hub DEFRGs in human induced pluripotent stem cells (iPSCs)-derived cardiomyocytes (CMs) infected with SARS-CoV-2 (high MOI, n = 3) compared with the mock infection CMs (n = 3) in GSE156754. (B) Expression levels of the hub DEFRGs in adult human CMs infected with SARS-CoV-2 (n = 3) compared with the uninfected human CMs (n = 3) in GSE151879 (P-value <0.05).

4. Discussion

The COVID-19 pandemic started in 2019 and had detrimental health effects along with high fatality numbers. This pandemic caused multi-organ complications including the heart. SARS-CoV-2 infection's impact on cellular and molecular mechanisms involved in PCD is not yet fully comprehended. The limited understanding of how this virus causes severe damage to various tissues and organs, leading to fatalities worldwide, underscores the urgent need for further research. Revealing the molecular mechanisms activated in the heart of COVID-19 patients can aid in finding suitable treatment strategies. Ferroptosis, a distinct form of PCD linking metabolism and redox biology has been gaining attention from scientists due to its involvement in virus-mediated diseases. Han et al. discovered that SARS-CoV-2 can infect pacemaker cells and trigger ferroptosis, in which the cells not only self-destruct, but also produce reactive oxygen molecules that can impact nearby cells and lead to cardiac arrhythmias [57]. Phosphatidylcholine; a ferroptosis signature, was observed in the myocardial autopsy of a patient with COVID-19-related lymphocytic myocarditis, proposing ferroptosis as a fatal factor in COVID-19-related cardiac damage [58]. In the present study, we showed how ferroptosis acts as an important molecular mechanism involved in damaging SARS-CoV-2 infected cells, which in turn might cause other CVDs and patient death. Here we used different bioinformatics approaches to analyze the ferroptosis-related genes that were differentially expressed in the heart of COVID-19 patients.

GO enrichment analyses revealed oxidative stress as a term. Today it is well-understood that a fine equilibrium between the oxidative stress and antioxidants is vital for the correct functions of cells. It has been shown that oxidative stress is an important factor for the initiation and progression of CVDs by inducing inflammation, endothelial dysfunction, and plaque formation in the arteries [59,60]. Recently, it has also been reported a role for ferroptosis in CVDs, which further validates ferroptosis effects on COVID-19 patients’ heart tissue [61,62]. Although Other studies have shown links between oxidative stress and ferroptosis in COVID-19 [17,63], the genes, TFs, and miRNAs that are potentially involved in damaging the heart tissue of COVID-19 patients have not been studied. This shows how ferroptosis might impart some of its effects on the heart and possibly promote CVD development.

DEFRGs cellular localization relayed some interesting information. Many DEFRGs were localized to the mitochondria that are a main source of ROS production, particularly during the process of oxidative phosphorylation. On the other hand, CMs have a distinctive high energy consumption rate and dependency on mitochondrial functions [64]. Furthermore, mitochondria are a ferroptosis promotion and suppression focal point [65]. This suggests that one possible reason behind heart damage severity in COVID-19 patients is how taxing it is on the heart tissue's mitochondria. The mTOR complex was also included in the GO results. The target of rapamycin (TOR) kinase functions with two separate multiprotein complexes including TORC1 and TORC2 [66]. TORC1 has been shown to repress or induce ferroptosis in different contexts and is a possible target for modulating tumorigenesis and ferroptosis [67]. In line with other studies, our findings suggest a correlation between ferroptosis and the mTOR complex, as well as highlighting the importance of studies focusing on modulation of the TOR complex to control the severity of COVID-19 disease.

Analysis of hub genes and functional modules identified several interesting and important factors. IL6, the most important hub gene, is a pro-inflammatory and anti-inflammatory cytokine that has been linked to different CVDs and proposed as a therapeutic target for a myriad of cardiac complications [68]. Additionally, patients with severe COVID-19 show higher levels of IL6 in their plasma and greatly benefit from IL6 antagonist therapy [69,70]. Interestingly, all of the hub genes identified in our study except FH interact with IL6. This adds validity to our results and suggests that ferroptosis-related hub genes simultaneously affect heart function, COVID-19 symptoms, and severity, as well as priming the heart for consequent CVDs. The interaction of IL6, SRC, and CDH1 is also seen in tumorigenesis and cell cycle regulation which is in line with the neoplasm-related terms seen in our results [71,72]. It is important to note that heart cells consider as non-dividing cells. Therefore, this interaction seems to have a another mission in the CMs. Knowing that CM hypertrophy in COVID-19 is associated with increased cardiac injury, we can hypothesize that another ferroptosis pathway mode of effect in the heart tissue of COVID-19 patients is the dysregulation of the cell cycle and cell size. This could lead to CM hypertrophy and the associated cardiac injuries [73].

AR is an inducer of CM hypertrophy [74]. Additionally, increased AR signaling activity is associated with increased COVID-19 severity [75]. Inhibition of AR expression has been shown to directly decline the levels of ACE2 and other Sars-CoV-2 entry co-receptors in CMs [75]. Another hub gene relevant to CM hypertrophy is HNF4A, which also regulates ACE2 levels. HNF4A has been associated with coronary heart disease and hypertrophic cardiomyopathy, and importantly it is present in the plasma and heart tissue of COVID-19 patients [76,77]. It is important to note that most of our identified ferroptosis-related hub genes are involved in CVD development, suggesting that COVID-19 patients are very succeptile for CVDs. For example, Heterochromatinization plays an important role in heart CVDs. EZH2 can regulate ACE2 expression by histone methylation of its promoter [78]. Genetic perturbation studies on mouse models have shown that EZH2 is involved in hypertrophic cardiomyopathy [79]. Furthermore, the role of EZH2 is dynamic and dependent on its interaction with different miRNAs, which will affect CVD-related genes and ferroptosis either positively or negatively [80,81]. Our findings suggest that EZH2 may induce heart damage in COVID-19 patients. Based on our predictions EZH2 interacts with important miRNAs in our study and shows increased expression during ferroptosis. This warrants further study into the nature of EZH2's interaction with the miRNAs in this study and how they can be used for therapeutic purposes. EGR1, another hub gene in our results, interacts with many nodes in the predicted regulatory network of DEFRGs including miRNAs and TFs. This makes EGR1 pivotal for future mechanistic studies on COVID-19's effects on the heart and the development of therapeutic strategies. Interestingly, EGR1 is also involved in different CVDs [82,83].

GSK3B, a kinase in the Wnt/β-catenin pathway, is a novel and important positive ferroptosis modulator [84]. GSK3B can be present as phosphorylated or dephosphorylated in vivo. The phosphorylated version can upregulate the expression of IL6 and TNF, which in turn decreases ACE2 expression [85]. Furthermore, GSK3B and MAFB control the pathogenic pulmonary macrophage transcriptome in COVID-19, and upregulate the expression of COVID-19 severity biomarkers. We can, therefore speculate that GSK3B enhances heart cell damage via the macrophage population within the heart tissue in addition to ferroptosis, suggesting connections between the ferroptosis pathway and pathogenic macrophages [86].

We found that IL6 also interacts with Mucin short variant S1 (MUC1) which is the membrane-bound mucin protein. Mucin proteins play vital roles in innate immunity, immunomodulatory pathways and infectious disease progression. They can also hinder or promote pathogen entry into the cell via steric hindrance [87]. MUC1 can increase the phosphorylation of GSK3B and also inhibit ferroptosis [87]. In addition, MUC1 has been demonstrated to have value for COVID-19 severity diagnosis in the airway mucus [88].

FH was a hub gene that does not interact with IL6 and is not present in the functional module of IL6. FH is a ferroptosis and complement activation inhibitor. A sharp drop in its levels along with ApoE in the blood is a predictor of patient death [89,90]. From the stated information about the hub genes, we can hypothesize that these ferroptosis genes provide a novel mechanistic insight into how ferroptosis affects the heart of patients and priming for subsequent CVDs. Furthermore, many of the hub genes stated probably affect the heart tissue via multiple non-ferroptotic pathways including CM hypertrophy, regulation of ACE2 levels, heterochromatinization, pathogenic macrophage-induced cell damage, and steric hindrance of virus entry. We can also infer that ferroptosis activation in COVID-19 patients’ hearts paves the road for CVDs. The analysis of gene-disease associations could be used to elucidate the molecular mechanisms of ferroptosis in COVID-19-associated CVDs, and subsequently design therapeutic strategies for COVID-19 cardiovascular complications.

The regulatory network governing the hub genes is comprised of miRNAs and TFs. Our results suggest that miR-124-3P affects the maximum number of genes compared to other miRNAs. miR-124-3P has been used to inhibit ferroptosis and ischemia–reperfusion-related injuries in rat models and holds promise for targeting this study's hub genes [91]. Interestingly, miR-124-3P also seems to target DEGs in COVID-19 patients with hypertrophic cardiomyopathy [92]. Other bioinformatics studies analyzing the transcriptome of COVID-19 patients for severity biomarkers found miR-26b-5p downregulated in the moderate symptom group, and have linked miR-26b-5p with thrombosis as well [93,94]. These findings add validity to the possible control of hub genes via miR-26b-5p. These miRNAs have also been detected in other SARS-CoV-2 infected cells. For example, miR-34a-5p is associated with endothelial and inflammatory signaling regulation and viral diseases in lung biopsies of COVID-19 patients [95]. Therefore, we can hypothesize that the important miRNAs mentioned in this study seem to be bolstered by other studies and warrant further experimental analysis.

Prominent drugs that target the DEFRG or affect them have been predicted in this study. The top 3 predicted drugs are pyrvinium, astemizole, and thioridazine. Pyrvinium is a Wnt/β-catenin pathway inhibitor that exerts cardioprotection in an isoproterenol (ISO) induced cardiotoxicity rat model [96]. These models showed symptoms including oxidative stress, inflammatory cytokine release, ATP level reduction, and fibrotic protein over-expression. Elevated levels of fibrotic proteins result in cardiac hypertrophy, myocardial necrosis, and functional and histological changes in the heart. In this case, Pyrvinium showed antioxidant, anti-inflammatory, and anti-fibrotic properties [96]. Astemizole is a non-sedative histamine H1 receptor blocker known to cause cardiac arrhythmias. Astemizole can induce rat CM hypertrophy and cytotoxicity and might warrant future testing as an inducer of COVID-19 CVDs in model animals [97]. Additionally, fingolimod hydrochloride has been shown to inhibit astemizole-induced hypertrophy and cytotoxicity and can be considered for further testing as a therapeutic agent [97]. Lastly, thioridazine has shown promising antiviral activity against RNA viruses besides SARS-CoV-2 [98]. The identified drugs in our study might reduce COVID-19 severity. They also might significantly reduce the odds of hospitalization, as well as prevent CVD development in COVID-19 patients.

Our study showed Iron, Oxoglutaric acid, NAD, Pyridoxal 5′-phosphate, Famotidine, L-Cysteine, NADPH and L-glutamic acid as important metabolites that interact with the DEFRGs. In this context, we can hypothesize that the presence of iron adds validity to the occurrence of ferroptosis in parallel to the other mechanisms of heart damage induction by the DEFRGs mentioned. Furthermore, the importance of iron levels in COVID-19 patients includes their anemia symptoms and the role of iron in viral replication, which are probably linked to the iron imbalance caused by ferroptosis. Oxoglutaric acid is a metabolite that links carbon and amino acid metabolism which are both important pathways in our gene enrichment analyses [99]. Interestingly, lowered oxoglutaric acid serum levels have been seen in metabolomics studies of COVID-19 patients [100]. On the contrary, elevated serum 2-oxoglutaric acid was reported as a diagnostic metabolic biomarker of HF [101]. However, to our knowledge, a detailed explanation of the exact role of oxoglutaric acid in the context of heart diseases in COVID-19 does not exist. Therefore, this also holds promise for future mechanistic studies. Pyridoxal 5′-phosphate (PLP) is the biologically active form of vitamine B6, which is shown to alleviate lipopolysaccharide-induced ferroptosis and apoptosis leading to myocardial injury through Nuclear factor erythroid 2- related factor 2 (Nrf2) activation [102]. In light of COVID, PLP has been reported to mitigate immune dysregulation and cardiomyophathy [103,104]. Famotidine; a selective blocker of the histamine H2 receptor, is widely used to control the acid in the stomach. Clinical data suggest that famotidine may mitigate COVID-19 disease, but its mechanism of action remain unkown [105]. Development of COVID-19 is partially mediated by dysfunctional mast cell activation and histamine release. A possible explanation of the evident famotidine activity as a COVID-19 therapy is that, this drug acts through its antagonism or inverse_agonism of histamine signaling and its arrestin biased activation [106]. Moreover, a lower peak ferritin value detected among patients given famotidine supports the hypothesis that it may decrease cytokine release in the setting of SARS-CoV-2 infection [107]. On the other hand, the cardioprotective effect of famotidine has been reported to be mediated by reduction of oxidative stress, and therefore maintanence of the healthy cardiac stem cell population [108]. Further comprehensive studies of these metabolites will deepen our understanding of the role of ferroptosis in COVID-19-related cardiac complicatios, which may provide promising candidate biomarkers and therapeutic targets.

5. Conclusions

In summary, we identified the potentially important DEFRGs activated in the heart of COVID-19 patients. We comprehensively analyzed the identified DEFRGs to find potential cellular and molecular mechanisms involved in heart damage occurrence in SARS-CoV-2 infected cells. Additionally, CMs seem especially susceptible to the unique interplay of the DEFRGs due to the importance of mitochondria, oxidative stress-relevant organelles, and loss of irreplaceable cells as can be inferred from the results of this study. The predicted regulatory miRNAs and TFs also supported the hypotheses above. Lastly, the predicted drugs, metabolites, miRNAs, and TFs mentioned within this study might warrant further examination for the development of therapeutic approaches. The important genes mentioned in this study require further wet-lab scrutiny such as qRT-PCR and protein localization using microscopic imaging and fluorescent labeled antibodies to establish molecular and cellular mechanisms with certainty. and Importantly, many of the ferroptosis-related genes, pathways, metabolites, and miRNAs are involved in CVD development, suggesting that SARS-CoV-2 lays the foundations for CVDs via ferroptosis induction.

Data availability

The RNA-Seq data used in this study are openly available at Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/). The GEO accession numbers for the microarray data are GSE169241, GSE151879, and GSE156754.

CRediT authorship contribution statement

Amin Alizadeh Saghati: Writing – original draft, Investigation, Formal analysis. Zahra Sharifi: Writing – review & editing, Writing – original draft, Formal analysis, Data curation. Mehdi Hatamikhah: Writing – original draft. Marieh Salimi: Writing – original draft, Methodology. Mahmood Talkhabi: Supervision.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Abbreviation List

Coronavirus Disease 2019

(COVID-19)

Severe Acute Respiratory Syndrome Coronavirus 2

(SARS-CoV-2)

Differentially Expressed Genes

(DEGs)

Differentially Expressed Ferroptosis-Related Genes

(DEFRGs)

Cardiomyocytes

(CMs)

Cardiovascular Diseases

(CVDs)

World Health Organization

(WHO)

Heart Failure

(HF)

Programmed Cell Death

(PCD)

Endoplasmic Reticulum

(ER)

Plasma Membrane

(PM)

Gene Expression Omnibus

(GEO)

Induced Pluripotent Stem Cells

(iPSCs)

Multiplicity Of Infection

(MOI)

Gene ontology

(GO)

Kyoto Encyclopedia of Genes and Genomes

(KEGG)

Database for Annotation, Visualization, and Integrated Discovery

(DAVID)

Biological Process

(BP)

Cellular Component

(CC)

Molecular Function

(MF)

Drug Signature Database

(DSigDB)

Protein-Protein Interaction

(PPI)

The Molecular complex detection

(MCODE)

Transcription Factors

(TFs)

Interleukin 6

(IL6)

Proto-oncogene tyrosine-protein kinase Src

(SRC)

Cadherin-1

(CDH1)

Androgen receptor

(AR)

Hepatocyte Nuclear Factor 4 Alpha

(HNF4A)

Enhancer of Zeste Homolog 2

(EZH2)

Early Growth Response Protein 1

(EGR1)

Glycogen Synthase Kinase-3 Beta Protein

(GSK3B)

Fumarate Hydratase

(FH)

Mucin Short Variant S1

(MUC1)

Glucose-6-Phosphate Dehydrogenase

(G6PD)

Citrate Synthase

(CS)

Glutamic-Oxaloacetic Transaminase 1

(GOT1)

Peroxisome Proliferator Activated Receptor Alpha

(PPARA)

Phosphoenolpyruvate Carboxykinase 2

(PCK2)

Vitamin D Receptor

(VDR)

Glutamic--Pyruvic Transaminase 2

(GPT2)

Isocitrate Dehydrogenase

(IDH2)

NAD-Dependent Deacetylase Sirtuin-3

(SIRT3)

Asparagine Synthetase

(ASNS)

Target of Rapamycin Kinase

(TOR kinase)

Isoproterenol

(ISO)

Pyridoxal 5′-phosphate

(PLP)

Nuclear factor erythroid 2- related factor 2

(Nrf2)

Appendix A. Supplementary data

The following are the Supplementary data to this article: