Diagnostic and predictive values of ferroptosis-related genes in heart failure

doi:10.21203/rs.3.rs-2624976/v1

Download PDF

Research Article

Diagnostic and predictive values of ferroptosis-related genes in heart failure

https://doi.org/10.21203/rs.3.rs-2624976/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background: Heart failure is a complex clinical syndrome, and there is growing evidence that ferroptosis is related to heart failure. This study sought to identify a new diagnostic model for ferroptosis-related genes in heart failure patients and analyze the signature genes associated with ferroptosis in heart failure.

Methods: The ferroptosis-related genes were found on the FerrDbwebsite, and the heart failure microarray datasets (GSE5406, GSE57338, GSE1145) were screened from the GEO database. The "limma" package in R software was then used to analyze the ferroptosis-related differentially expressed genes (DEGs), and functional enrichment analysis was carried out for ferroptosis-related DEGs. The differentially expressed ferroptosis-related genes were then screened using LASSO regression and SVM-RFE algorithms. The intersection was then used to get the signature genes. The signature genes served as the foundation for the diagnostic model. The diagnostic model was created using a nomogram and receiver operating characteristic curve (ROC), and the model's precision was assessed. The expression of the signature genes' signaling pathways was examined using GSEA. The CIBERSORT algorithm was then used to analyze immune cell infiltration and correlation analysis in the immune systems of heart failure patients. Finally, the testing set was used to evaluate the diagnostic and predictive value of signature genes in heart failure.

Results: The training set (GSE5406) was used to screen 127 ferroptosis-related differentially expressed genes, including 44 up-regulated and 83 down-regulated genes. Ferroptosis was significantly enriched for genes that were differentially expressed according to KEGG analysis, and oxidative stress was significantly enriched in genes according to GO-BP analysis. A diagnostic model and nomogram were successfully constructed based on the five differential genes with an area under the curve (AUC):0.952 (95% CI: 0.894-0.993), using the diagnostic model to differentiate between the normal control group and the heart failure group. Five ferroptosis-related differential genes (BECN1, SLC39A14, QSOX1, DAZAP1, TMSB4X) were screened and identified. Additionally, CD4-naive T cells were discovered to be related to heart failure patients. Finally, the diagnostic performance in the testing set (GSE57338, GSE1145) was confirmed, further demonstrating the accuracy and reliability of the study's findings.

Conclusion: A novel diagnostic model with significant value for heart failure was successfully established after five ferroptosis-related genes were screened and identified. Additionally, it might be beneficial for treating patients with heart failure and aid in understanding the part ferroptosis plays in the pathogenesis of the condition.

ferroptosis

heart failure

diagnosis

prediction

immune infiltration

The loss of cardiac homeostasis, which is dependent on the precise control of gene expression, leads to heart failure^[1]. Currently, BNP and NT-proBNP are the two most important biomarkers for the diagnosis and treatment of heart failure. These biomarkers can be used as guidelines for the clinical management of heart failure and as tools for population screening ^[2], but they are influenced by factors like race, age, sex, weight, renal function, arrhythmias, and medications. The primary imaging method for detecting late disease progression and assisting in risk assessment and treatment selection is echocardiography ^[3]. However, improving the management of individual heart failure has received greater focus lately ^[4]. In patients with acute or chronic heart failure, anemia is a recognized predictor of poor outcomes^[5]. Therefore, it's crucial for patients with heart failure to preserve ferroptosis homeostasis.

A controlled form of cell death known as ferroptosis is characterized by iron overload and results in the buildup of lipid hydroperoxides. Iron metabolism, amino acid metabolism, lipid metabolism, and other activities are principally involved in the complex regulatory mechanisms of ferroptosis ^[6]. Ferroptosis has been implicated in a number of diseases, including cardiomyopathy, myocardial infarction, ischemia/reperfusion injury, and heart failure. According to research, puerarin elevated GPX4 by suppressing NOX4 expression and prevented erastin or isoprenaline from inducing ferroptosis in cardiomyocytes. Puerarin therapy prevented heart failure in rats with pressure overload-induced heart failure by preventing ferroptosis and lessening mitochondrial atrophy ^[7]. Ferroptosis is a potential target in heart failure with preserved ejection fraction, according to a previous review ^[8]. It is unclear whether iron supplementation therapy is beneficial in patients with heart failure with preserved ejection fraction, and the precise mechanism needs to be further researched. Iron imbalance may be a common factor in many cardiovascular illnesses, according to an increasing body of research ^[9]. Since few research have examined the connection between ferroptosis-related genes and heart failure, it is important and necessary to learn whether ferroptosis may be used as a therapeutic target in heart failure and whether it can stop the progression of heart failure. Using bioinformatics analysis of expression data and ferroptosis-related genes in healthy and failing LV myocardial samples from the GEO database, this study sought to identify differentially expressed ferroptosis-related genes in heart failure and build a diagnostic model of ferroptosis-related genes in heart failure (Figure 1).

1.1 Data processing and gathering for microarray data sets

Microarray datasets from GEO were filtered. Heart failure was the search term, and "Homo sapiens" and expression analysis by the array were the inclusion criteria. After screening, GSE5406 (as a training set)^[10], GSE57338 (a test set)^[11], and GSE1145 (a test set) were used. GSE5406 contains 210 samples total, including 16 left ventricular myocardium without human failure and 194 left ventricular myocardium with human failure. GSE57338 consists of 313 samples total, including 136 left ventricular myocardium without human heart failure and 177 left ventricular myocardium with human heart failure. GSE1145 contains 79 left ventricular myocardium with heart failure and 11 healthy controls. Ninety samples altogether were collected (Table 1).

1.2 Differentially expressed genes associated with ferroptosis and correlation analysis

The GEO data were annotated for correction using the R software "limma" package ^[12], and ferroptosis-related genes were collected and collated from the "FerrDb" database ^[13]. When multiple probes are mapped to the same gene symbol, the probes are converted to gene symbols in each dataset according to the annotation file of the platform, and the average value of the probes is selected as the gene expression value. Differentially expressed genes were analyzed between GEO database data and ferroptosis-related genes with adjusted p-values < 0.05, and heat maps of differentially expressed genes were plotted using the R software packages "pheatmap" and "corrplot," respectively, and correlation analysis.

1.3 Enrichment analysis of differentially expressed genes

Using the R program “clusterProfiler,” enrichment analysis of differentially expressed genes linked to ferroptosis was carried out ^[14]. This analysis included GO enrichment analysis and KEGG enrichment analysis. P values ＜0.05 were regarded as statistically significant in the GO analysis, which included three categories of biological processes (BP), cellular components (CC), and molecular functions (MF). KEGG analysis was utilized to look at potential signaling pathways.

1.4 Screening of characteristic genes by LASSO regression and SVM-RFE algorithm

Feature genes were screened using the least absolute shrinkage and selection operator (LASSO) and support vector machine recursive feature elimination (SVM-RFE) ^[16]. Differentially expressed genes were examined using the "glmnet, e1071" package in the R software ^[15]. The intersecting genes were then extracted from the intersection of the screened feature genes and visualized using Venn diagrams to create nomogram and high-precision diagnostic models. The "pROC" tool in R software was used to create the ROC curves for the signature genes and diagnostic models^[17]. The ROC curves were then used to assess the diagnostic models' accuracy and prediction precision.

1.5 GSEA analysis

The expression of the pertinent signaling pathways in the distinctive genes may be seen using the R software "GSEA" package ^[18]. Only the heart failure group samples were preserved after the samples from the normal control group were eliminated. The samples were then divided into groups based on the expression levels of the distinctive genes to create the groups with high and low expression. P ＜0.05 was used to filter the enrichment data, and only the top 6 routes with the highest significant enrichment were shown.

1.6 Immunoinfiltration analysis

The fraction of each sample's 22 immune cells was examined using the CIBERSORT method to evaluate gene expression ^[19]. The "limma" software was used to normalize the gene expression data, and the CIBERSOR method was used to translate the data to the expression of 22 immune cells. Perl (https://www.perl.org) was then used to filter the findings at P＜ 0.05 to provide the results for immune cell infiltration. The "vioplot" program was then used to create violin plots to compare the numbers and variations of immune cells between the two groups. The "ggplot2" package was used to illustrate the analysis after the "corrplot" tool was used to evaluate the correlation heat map between immune cell infiltration.

1.7 Validation of the diagnostic model

The diagnostic performance of the created diagnostic models was further evaluated in the testing set(GSE57338, GSE1145) by plotting Draw boxplot diagrams using the R software "ggplot2" package.

1.8 Statistical analysis

R software (version 4.2.1) and Perl tools were used for all statistical studies. The Wilcoxon test was employed for statistical comparisons between two groups, and the Kruskal-Wallis test was arbitrarily chosen when there were three or more groups. Statistical significance was defined as ＜P 0.05 (bilateral). The following values were significant: * P ＜0.05, **＜ P 0.01, and *** P＜ 0.001.

2.1 Analysis of differentially expressed genes for ferroptosis

Four hundred eighty-four ferroptosis-related genes were gathered and compiled from the "FerrDb" database. By combining the genes from the training set with those associated with ferroptosis, 127 ferroptosis-related differentially expressed genes were discovered, comprising 44 up-regulated genes and 83 down-regulated genes (Table 2). In order to identify the genes that differ between normal control and heart failure groups, differentially expressed genes were analyzed using R software and shown in a heat map (Figure 2). Samples were categorized, with normal control shown in blue and heart failure in red.

Table 1 Gene expression datasets used in this study

GEO accession	Platform	Year	Tissue
GSE5406	GPL96	2006	Left ventricular myocardium
GSE57338	GPL11532	2015	Heart left ventricle
GSE1145	GPL570	2004	Left Ventricle Biopsy

Table 2: Genes with differential expression

Gene	Control Mean	Treat Mean	pvalue	Type
PTGS2	3.735865938	3.860145361	0.027794258	Up
CS	10.626706	10.8006121	0.009957624	Up
EMC2	6.018840313	6.297643418	0.006439808	Up
ACSF2	5.820696063	6.022310149	0.000822308	Up
VDAC2	10.84736325	10.94575248	0.011049296	Up
NCOA4	9.868979313	10.08575503	0.006030582	Up
GABARAPL2	10.23240013	10.38750678	0.005536831	Up
PEBP1	9.85000675	10.08034807	0.003226382	Up
TGFBR1	5.507920188	5.590214753	0.03465182	Up
PEX12	4.819798313	4.993629665	0.002389567	Up
PEX3	4.748285	4.834721593	0.033924416	Up
SNCA	6.131134813	6.245663582	0.015494996	Up
SIRT3	7.307551063	7.4604355	0.023683465	Up
DLD	8.949755688	9.262399665	0.006607947	Up
PRKCA	5.906490188	6.002175459	0.002910267	Up
MICU1	7.571580875	7.733151072	0.000973354	Up
GSTZ1	6.554007813	6.752168397	0.001058303	Up
GJA1	8.94431375	9.629693392	7.94199E-06	Up
CIRBP	8.771418	9.06842033	0.000980565	Up
TRIM26	7.033414563	7.158416273	0.030978671	Up
FADS2	6.65823725	6.886957588	0.003782746	Up
EGR1	6.621337	6.912063237	0.032858933	Up
ADAM23	6.832706875	7.043762639	0.044239696	Up
MEG3	5.198592	5.325323912	0.012501913	Up
ENPP2	5.468563563	5.91218001	4.23552E-05	Up
ISCU	10.05952925	10.50558616	3.26565E-06	Up
CHMP5	6.620764375	6.861778526	0.001503779	Up
SLC16A1	6.5797635	6.848754711	0.008110024	Up
FZD7	5.624075813	6.231583784	1.91977E-05	Up
BRD3	7.4041135	7.649242608	0.001514856	Up
DECR1	9.244252	9.42292384	0.049936576	Up
PPP1R13L	7.823658563	8.036244665	0.00518761	Up
KDM3B	7.103550625	7.375644907	0.000263162	Up
PARP2	6.776521938	6.891308371	0.024488476	Up
PARP8	5.61751475	5.755076129	0.005726621	Up
PARP12	5.567965313	5.800322794	0.004390677	Up
BEX1	7.038020438	7.475995608	0.004822257	Up
TMSB4X	9.387276813	9.679762887	0.019317533	Up
KDM4A	6.97379375	7.080841325	0.012962154	Up
MPC1	9.885591688	10.14856652	0.003937913	Up
SIRT2	8.470080063	8.642663129	0.016623215	Up
MEF2C	6.712090188	6.91678732	0.008913374	Up
GSTM1	7.96118175	8.145704778	0.022518902	Up
RARRES2	5.613869313	5.910758706	0.010605693	Up
CHAC1	6.923034125	6.703196861	0.015133924	Down
HSPB1	11.85432056	11.39584508	0.004979217	Down
IREB2	3.612065813	3.501706	0.001074216	Down
G6PD	5.994550938	5.778824691	0.002339222	Down
PGD	7.340181625	6.970997706	0.000248243	Down
ACSL4	4.040328813	3.826882335	0.005432759	Down
LPCAT3	6.703436438	6.427789201	0.000809707	Down
NRAS	4.562735188	4.33994067	0.003077033	Down
SLC38A1	9.277158438	8.696509665	0.000995861	Down
SLC1A5	7.70871725	7.485839686	0.000803468	Down
ALOX5	5.241757688	5.140545964	0.009196584	Down
KEAP1	7.092501875	6.979204371	0.016721237	Down
ATG5	5.699945438	5.556223665	0.00040348	Down
PHKG2	6.805893563	6.672770665	0.003436858	Down
BECN1	7.743867	7.644581258	0.0458427	Down
SNX4	5.212448625	5.060937933	0.014955971	Down
SAT1	8.153058125	7.741631258	0.00210145	Down
MAPK1	7.0139955	6.75637768	0.001741221	Down
BID	5.789904688	5.573732144	0.000104356	Down
MAPK9	6.280735188	6.162290768	0.019210838	Down
MAPK14	6.173374	6.000712686	0.000809713	Down
PRKAA1	3.803406625	3.732561639	0.03114367	Down
ELAVL1	7.681876563	7.542367634	0.005153338	Down
EPAS1	8.7215115	8.527743062	0.011623874	Down
LPIN1	6.623586375	6.304288974	2.98129E-06	Down
TLR4	6.02606875	5.908545655	0.03950418	Down
ATF3	7.352771813	6.55603134	0.001296489	Down
MTDH	6.615996938	6.400821211	0.004048925	Down
POR	6.599822188	5.963002763	4.41598E-06	Down
ELOVL5	6.211679188	6.081705103	0.020225149	Down
PTEN	5.339030625	5.147868964	0.001983921	Down
TBK1	6.814853438	6.694369613	0.032683783	Down
IL6	5.22229475	4.920458366	0.030978616	Down
ATF4	9.910998188	9.320672938	1.18315E-06	Down
AQP3	6.6776735	6.184862644	3.68554E-05	Down
DCAF7	6.093050813	5.984369696	0.017315676	Down
MTCH1	10.42536638	10.12269746	6.04256E-05	Down
SLC39A14	9.363982063	8.920519273	2.09227E-05	Down
SLC25A28	7.386891188	7.200388412	0.002012683	Down
IFNA2	3.729314125	3.672419835	0.045146938	Down
PPARG	5.804893313	5.588686546	0.001003542	Down
QSOX1	8.118508313	7.770224696	2.83554E-05	Down
PIEZO1	6.497534375	6.294135531	0.014434041	Down
LIFR	3.974079125	3.859653294	0.019431835	Down
TIMP1	9.41335775	8.681354572	7.64864E-05	Down
CCDC6	4.801594188	4.594966144	0.023816177	Down
KMT2D	5.285662313	5.20385134	0.012884611	Down
AKR1C1	7.371374813	7.044923402	0.000944175	Down
AKR1C3	6.38689925	6.006244072	0.00919672	Down
NQO1	6.672161063	6.46768768	0.000803461	Down
SLC3A2	7.7229475	7.568800325	0.040540365	Down
MT1G	7.578630188	7.472363113	0.04846139	Down
SCD	7.07732775	6.557209325	0.040123468	Down
STAT3	9.485480063	8.79114951	6.70632E-07	Down
MTOR	5.905657625	5.843011948	0.025037898	Down
CDKN1A	9.301168875	8.543084103	3.04352E-06	Down
CBS	5.068929063	4.811010165	0.000440978	Down
BRD4	7.221245875	6.711181031	5.37517E-07	Down
PRDX6	9.670188938	9.269505299	0.000191083	Down
NF2	4.913303125	4.831663325	0.027341993	Down
ARNTL	5.864491938	5.61160816	0.000158411	Down
PLIN2	6.737046188	6.193575979	8.54368E-07	Down
ZFP36	8.994726063	8.675310959	0.03081177	Down
DAZAP1	6.689931563	6.291527067	5.5276E-06	Down
PIR	4.27115825	4.073019974	0.013762003	Down
FTL	11.07947188	10.76440668	0.012726159	Down
NT5DC2	6.920454938	6.532587041	0.001955557	Down
NCOA3	6.549206875	6.335434268	0.000489516	Down
ARF6	7.24059575	7.117578974	0.048706259	Down
AHCY	7.259544688	6.907764701	4.11952E-05	Down
SIAH2	6.72897925	6.554368912	0.00225734	Down
RELA	7.400368375	7.21615001	0.011207269	Down
PRDX1	9.817919	9.43888382	3.48088E-06	Down
MTF1	6.687935688	6.574762799	0.003013464	Down
COPZ1	7.69138125	7.587267887	0.012056201	Down
PARP6	7.715632438	7.59077383	0.013118598	Down
TXN	5.762842938	5.551780191	0.01409466	Down
CREB3	7.607109625	7.412534979	0.000254528	Down
MLST8	6.920732688	6.801852108	0.015495163	Down
VCP	8.008710813	7.785856964	0.001058196	Down
TRIB2	7.411926625	7.251504299	0.022519732	Down
PDK4	8.824693	7.619592098	0.009370584	Down
MAPKAP1	7.547510063	7.260061284	8.52003E-06	Down

2.2 Correlation analysis of differentially expressed genes

R software was used to analyze differential genes, and correlations were obtained by correlation analysis of differential genes (Figure 3). The asterisk in the square of the graph denotes a significant correlation between the corresponding two genes, with the red asterisk grid denoting a positive regulatory relationship between genes and the blue asterisk grid denoting a negative regulatory relationship between two genes. The graph's colors, red for a positive correlation and blue for a negative correlation.

2.3 Enrichment analysis of differentially expressed genes

R software was used to analyze the enrichment analysis of differentially expressed genes (Figure 4), and P values less than 0.05 were used to filter the results. Hepatitis B, mastocytosis, Kaposi's sarcoma-associated herpesvirus infection, human cytomegalovirus infection, lipid and arteriosclerosis, AGE-RAGE signaling pathway in diabetic complications, FoxO signaling pathway, autophagy-animal, NOD-like receptor signaling pathway, and human T-cell leukemia virus one infection made up the top 10 enriched pathways in the KEGG analysis based on P value. The primary areas of cellular response to chemical stress, response to nutritional levels, response to oxidative stress, response to oxidative stress, response to external stimuli, etc., were enriched in GO-BP analysis through the GO medium. TORC2 complex, secretory granule lumen, TOR complex, cytoplasmic vesicle lumen, etc., are all considerably enriched in CC analysis. NAD+ ADP-ribosyltransferase activity, DNA-binding transcription factor binding, RNA polymerase II-specific DNA-binding transcription factor binding, NAD+-protein ADP-ribosyltransferase activity, and antioxidant activity were all enriched in GO-MF. According to the enrichment analysis's findings, heart failure may be significantly influenced by the activation of oxidative stress, ferroptosis, etc.

2.4 Screening for Ferroptosis-related signature genes in heart failure by machine learning

The R software "glmnet" package was used to perform LASSO regression analysis to clarify the diagnostic value of differentially expressed genes linked to ferroptosis (Figure 5A,5B). The number of genes corresponding to the point with the least cross-validation was discovered to be the number of signature genes found. Furthermore, the discovery of 17 genes using the LASSO regression method: BECN1, LPIN1, ATF4, AQP3, SLC39A14, QSOX1, STAT3, ENPP2, CBS, BRD4, PRDX6, DAZAP1, AHCY, PRDX1, CREB3, TMSB4X, and MAPKAP1.

The accuracy graph (Figure 5C), where the horizontal coordinate indicates the number of genes and the vertical coordinate represents the cross-validation accuracy, can be created by applying the SVM-RFE method to screen for disease signature genes. The point corresponding to the lowest cross-validation error, i.e., the number of genes to be looked up, needs to be located on the graph of cross-validation error (Figure 5D), where the horizontal coordinate represents the number of genes, and the vertical coordinate represents the error of cross-validation. So far, 18 distinct genes have been identified, including SLC25A28, SLC39A14, MLST8, TMSB4X, KEAP1, CHAC1, EMC2, MEG3, PTGS2, QSOX1, MTDH, BECN1, KDM4A, SCD, AKR1C3, DAZAP1, MTOR, and ACSF2.

2.5 Venn diagram

The intersection of the genes discovered by the R software "VennDiagram" package was analyzed using the LASSO regression and SVM algorithms. As a result, five genes related to ferroptosis were identified for the diagnosis of heart failure (Figure 6), namely BECN1, SLC39A14, QSOX1, DAZAP1, and TMSB4X.

2.6 Construction of the diagnostic model

For each signature gene, a ROC curve can be generated using the R package (glmnet,pROC); the area under the curve ranges from 0.5 to 1, and the larger the area under the curve, the higher the accuracy of the gene as a disease diagnostic gene. As seen in the figure (Figure 7A), the DAZAP1 curve has the largest area under the curve, with an AUC value of 0.842, indicating higher accuracy in predicting the disease. The ROC curves of the distinctive genes were effectively generated (Figure 7B), and the AUC was evaluated, yielding an AUC value of 0.952 (95% CI: 0.894-0.993). The five differentially expressed genes of ferroptosis were successfully constructed as gene diagnostic models. The five differentially expressed genes for ferroptosis demonstrated good diagnostic value in differentiating between standard control and heart failure patients, demonstrating the diagnostic model's good diagnostic performance (Figure 7C). The box plot revealed significant differences in the levels of five genes between the standard control and heart failure groups. These five genes were dubbed heart failure diagnostic genes as a result.

The R software's "RMS" package created a nomogram (Figure 8). According to the nomogram, each gene's relative expression level is assigned a score. The total score, the sum of each score, can be used to identify heart failure and forecast its occurrence.

2.7 GSEA enrichment analysis

The expression of the signaling pathway in the signature gene can be seen by using the R software "GSEA" package to assess the up- and down-regulated signaling pathways for each diagnostic gene. If the peak of the curve is on the upper left side, this indicates that the signaling pathway is highly expressed in the signature gene. If it is on the lower right side, this indicates that the signaling route is lowly expressed in the diagnostic gene (Figure 9). Systemic lupus erythematosus, neuroactive ligand-receptor interaction, leishmania infection, cytokine receptor interaction, and cell adhesion molecules cams were expressed less frequently in the diagnostic gene BECN1. Systemic lupus erythematosus, ecm receptor interaction, cytokine receptor interaction, complement, and coagulation cascades were substantially elevated among the diagnostic genes QSOX1. However, Parkinson's disease and oxidative phosphorylation were expressed at low levels.

2.8 Immunocell infiltration analysis

The gene expression data for cell infiltration analysis was analyzed using the CIBERSORT method with 1000 simulations; the more simulations, the higher the accuracy gained; the findings were then filtered with P ＜0.05 to maintain the immune cell infiltration results. The results demonstrate that T cells CD4 naive differed between the two groups and if P＜ 0.05, this suggests a difference between the normal control group and the heart failure group. Twenty-two different infiltrating immune cells were seen (Figure 10A). Using the R software "corrplot" package, the correlation between the characteristic genes and immune cells was examined. A heat map of the correlation was produced (Figure 10B), with red denoting genes that are positively correlated and white denoting genes that are negatively correlated. It is clear that the genes for Macrophages M2 and TMSB4X are positively correlated.

2.9 Validation of diagnostic models using external datasets

To further validate the diagnostic model, the diagnostic performance of the characterized genes was further evaluated in the test set by assessing the expression differences in the normal control and heart failure groups. In the test set (GSE57388) (Figure 11A), the diagnostic genes BECN1, DAZAP1, SLC39A14, and QSOX1 were low expressed in the samples (P < 0.05), and TMSB4X was highly expressed in the samples (P < 0.05), indicating differences between the normal control and heart failure groups. In the test set (GSE1145) (Figure 11B), DAZAP1, SLC39A14, and QSOX1 were lowly expressed in the samples (P < 0.05), and TMSB4X were highly expressed in the samples (P < 0.05), however, the diagnostic gene BECN1 was lowly expressed in the samples (P > 0.05), with no significant differences.

Growing evidence suggests that ferroptosis, a non-apoptotic kind of programmed cell death lipid peroxidation-driven, has a substantial impact on the emergence of cardiovascular disease ^[20]. One hundred twenty-seven ferroptosis-related genes with differential expression were discovered in this study, of which 44 had elevated expression, and 83 had downregulated expression. In KEGG enrichment analysis, the pathways of neurodegeneration associated with multiple diseases, hepatitis B, Kaposi sarcoma-associated herpesvirus infection, human cytomegalovirus infection, lipid and atherosclerosis, human T-cell leukemia virus 1 infection, and oxidative stress are the major biological processes that are involved. The relationship between ferroptosis-related genes and heart failure, however, remains unknown^[21]. In this study, we investigated the relationship between genes associated with ferroptosis and heart failure, identified five signature genes related to ferroptosis, and developed a reliable model for diagnosing heart failure.

Beclin-1(BECN1), a gene associated with diagnostic models, is essential for autophagy and can accelerate iron death via conventional autophagic function and direct contact with SLC7A11. In vitro, BECN1 deletion prevents iron mortality brought on by SLC7A11 inhibitors like salazosulfapyridine and sorafenib but not by GPX4 inhibitors like RSL3 and FIN56^[22]. It has been discovered that BECN1 can control iron death responses in a regulatory-dependent manner and may have a function in promoting Ferroptosis that is both reliant on and independent of autophagy ^[23]. Beclin1 downregulation directly encourages cardiac remodeling in the setting of the cardiovascular system ^[24]. Activation of autophagy during ischemia is advantageous since it increases cell survival and cardiac performance. It has been demonstrated that autophagy is actively expressed in the heart in response to specific stimuli, such as nutritional deprivation, ischemia/reperfusion, and heart failure. The heart, however, appears to be harmed by prolonged excessive autophagy and significant induction of BECN1 during reperfusion ^[25]. Diseases linked with aging, such as cardiovascular failure and atherosclerosis, depend heavily on endothelial dysfunction and reduced autophagic activity. Despite the fact that autophagy is an essential cell-salvage mechanism for preserving cellular homeostasis, aberrant autophagy can result in cardiac defects and, ultimately, heart failure ^[26].

Sulfhydryl oxidase 1(QSOX1) is a secreted atypical disulfide catalyst that regulates the structure and functionality of the extracellular matrix. The levels of the RNAs QSOX1 and PLBD1 in peripheral blood have been demonstrated to be unique, independent predictors of left ventricular insufficiency following acute myocardial infarction ^[27]. For mice experiencing acute stress, QSOX1 might also be a brand-new target for cardioprotection. According to studies, QSOX1 is increased in the heart during acute heart failure. It is necessary for correct protein folding in the endosome or sarcoplasmic reticulum, which is required for abrogation and protective reactions during acute stress ^[28]. Additionally expressed in the human atherosclerotic core, quiescence sulfhydryl oxidase (QSOX) may be involved in apoptosis ^[29]. Through several redox pathways, quiescent in sulfhydryl oxidase 1b (QSOX1b) stimulates the migration and proliferation of vascular smooth muscle cells ^[30]. The novel biomarker QSOX1 is a possible biomarker for acute decompensated heart failure, and it is particularly effective when combined with BNP in identifying acute decompensated heart failure. Clinical and experimental research have provided proof that there is a connection between cardiovascular QSOX1 production and acute decompensated heart failure.

Thymosin beta-4 (TMSB4X) can influence Hand1-mediated vascular formation ^[31]. By regulating the proportion of G-actin or F-actin in platelets, TMSB4X is crucial for thrombosis ^[32]. TMSB4X has been discovered to be a growth and exercise factor for humans. Regardless of the level of insulin resistance or the type of exercise performed, TMSB4X is similarly markedly elevated in the plasma of exercising humans ^[33]. The extracellular signaling pathways in which TMSB4X boosts cell surface ATP levels via ATP synthase have been identified as extracellular regulators of angiogenesis, inflammation, wound healing, and development ^[34]. To promote Akt activation and improve heart function, TMSB4X may suppress NF-B activation and collagen-mediated fibrosis ^[35].

DAZAP1 (DAZ-associated protein 1) is an RNA-binding protein that, by selectively identifying certain cis-elements, controls cellular value addition and migration ^[36]. The expression of COX6C and the regulation of energy metabolism by DAZAP1, a negative regulator of pre-mRNA splicing, may control the amount of energy produced in mitochondria ^[37]. SLC39A14 (Metal cation symporter ZIP14) is crucial for tissue homeostasis, metabolism, development, and immunity as a plasma membrane electroneutral transporter protein. The zinc transporter protein SLC39A14 has been demonstrated to control inflammatory signals linked to obesity ^[38]. Additionally, it has been demonstrated that the majority of iron insufficiency in the world may be caused by an underlying zinc deficiency, which may control human iron absorption by regulating cellular zinc concentration^[39].

This study discovered a link between heart failure and CD4-naive T cells. It has been demonstrated that CD4T cells proliferate and activate in chronic heart failure, causing cardiac damage and remodeling, and that activated T lymphocytes play a significant role in pathological remodeling in heart failure ^[40]. As we age, the interaction between the immunological and circulatory systems causes T cells CD4 naive to increase myocardial inflammation, resulting in heart failure ^[41]. These results align with what the current study had expected, supporting the study's validity. The current study does, however, inevitably have certain limitations. First, we used the GEO database dataset to validate the diagnostic value of our diagnostic model, which could be further validated by collecting specimens from heart failure patients. Second, no in vivo and in vitro experiments were performed to validate the mechanisms of these diagnostic genes in heart failure, and there is a lack of further validation of the mechanisms associated with signaling pathways. Finally, it is necessary to include more newly identified ferroptosis-related genes in addition to the existing ferroptosis-related genes.

As a result, a novel diagnostic model for heart failure was created, which performs well in terms of diagnosis and could provide a promising therapeutic target. It also discovered five critical ferroptosis-related genes. In addition, it could assist us in understanding the molecular pathways of ferroptosis that are implicated in the etiology of heart failure.

Data availability statement

The datasets presented in this study can be found in online repositories. (GSE5406, GSE57338, GSE1145, and FerrDb)

Conflicts of interest

None of the authors have any conflicts of interest to disclose.

Authors’ contributions

All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication.

Consent for Publication

Publication consent has been obtained from all participants

Ethics approval and consent to participate

Not applicable.

Acknowledgements

We thanked the authors who built up these datasets (GSE5406, GSE57338, and GSE1145). Their work provided convenience for this article greatly.

Funding information

This study was supported by the First Affiliated Hospital, School of Medicine, Shihezi University, the Corps Financial Technology Plan Project of China (No. 2020AB023).

Gomes CPC, Schroen B, Kuster GM, Robinson EL, Ford K, Squire IB, Heymans S, Martelli F, Emanueli C, Devaux Y. EU-CardioRNA COST Action (CA17129). Regulatory RNAs in Heart Failure. Circulation. 2020 Jan;28(4):313–28.
Castiglione V, Aimo A, Vergaro G, Saccaro L, Passino C, Emdin M. Biomarkers for the diagnosis and management of heart failure. Heart Fail Rev. 2022 Mar;27(2):625–43.
Melillo E, Masarone D, Oh JK, Verrengia M, Valente F, Vastarella R, Ammendola E, Pacileo R, Pacileo G. Echocardiography in Advanced Heart Failure for Diagnosis, Management, and Prognosis. Heart Fail Clin. 2021 Oct;17(4):547–60.
Agewall S. Optimizing individual heart failure treatment.Eur Heart J Cardiovasc Pharmacother. 2022 Feb16;8(2):105–107.
Anand I, McMurray JJV, Whitmore J, et al. Anemia and its relationship to clinical outcome in heart failure. Circulation. 2004;110:149–54.
Wu X, Li Y, Zhang S, Zhou X. Ferroptosis as a novel therapeutic target for cardiovascular disease. Theranostics. 2021 Jan 1;11(7):3052–3059.
Liu B, Zhao C, Li H, Chen X, Ding Y, Xu S. Puerarin protects against heart failure induced by pressure overload through mitigation of ferroptosis. Biochem Biophys Res Commun. 2018;497:233–40.
Li Q, Zhao Z, Zhou X, Yan Y, Shi L, Chen J, Fu B, Mao J. Ferroptosis: The Potential Target in Heart Failure with Preserved Ejection Fraction. Cells. 2022 Sep 12;11(18):2842.
Fang X, Ardehali H, Min J, Wang F. The molecular and metabolic landscape of iron and ferroptosis in cardiovascular disease.Nat Rev Cardiol. 2022 Jul4:1–17.
Hannenhalli S, Putt ME, Gilmore JM, Wang J et al. Transcriptional genomics associates FOX transcription factors with human heart failure.Circulation 2006 Sep19;114(12):1269–76.
Liu Y, Morley M, Brandimarto J, Hannenhalli S, et al. RNA-Seq identifies novel myocardial gene expression signatures of heart failure. Genomics. 2015 Feb;105(2):83–9.
Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015 Apr;20(7):e47.
Zhou N, Bao J, FerrDb:. A Manually Curated Resource for Regulators and Markers of Ferroptosis and Ferroptosis-Disease Associations. Database (Oxford) (2020) 2020:baaa021.
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. Innov (NY). 2021;2(3):100141.
Biau G, Scornet E. A random forest guided tour. TEST-SPAIN. 2015;25:1–31. 10.1007/s11749-016-0481-7.
Huang ML, Hung YH, Lee WM, Li RK, Jiang BR. SVM-RFE based feature selection and taguchi parameters optimization for multiclass SVM classifier. ScientificWorldJournal. 2014;2014:795624.
Kamarudin AN, Cox T, Kolamunnage-Dona R, Time-Dependent ROC. Curve Analysis in Medical Research: Current Methods and Applications. BMC Med Res Methodol. 2017;17(1):53.
Peng J, Han L, Shang X. A novel method for predicting cell abundance based on single-cell RNA-seq data. BMC Bioinf. 2021;22:281.
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;37:773–82.
Chen X, Yu C, Kang R, Kroemer G, Tang D. Cellular degradation systems in ferroptosis. Cell Death Differ. 2021 Apr;28(4):1135–48.
Li N, Jiang W, Wang W, Xiong R, Wu X, Geng Q. Ferroptosis and its emerging roles in cardiovascular diseases. Pharmacol Res. 2021 Apr;166:105466.
Song X, Zhu S, Chen P, Hou W, Wen Q, Liu J, et al. AMPK-mediated BECN1 phosphorylation promotes ferroptosis by directly blocking system Xc(–) activity. Curr Biol. 2018;28:2388–99. e2385.
Lee H, Zandkarimi F, Zhang Y, Meena JK, Kim J, Zhuang L, et al. Energy-stress-mediated AMPK activation inhibits ferroptosis. Nat Cell Biol. 2020;22:225–34.
Wang L, Ma H, Huang P, Xie Y, Near D, Wang H, Xu J, Yang Y, Xu Y, Garbutt T, Zhou Y, Liu Z, Yin C, Bressan M, Taylor JM, Liu J, Qian L. Down-regulation of Beclin1 promotes direct cardiac reprogramming. Sci Transl Med. 2020 Oct 21;12(566):eaay7856.
Zhai P, Sadoshima J. Glycogen synthase kinase-3β controls autophagy during myocardial ischemia and reperfusion. Autophagy. 2012 Jan;8(1):138–9.
Kishore R, Krishnamurthy P, Garikipati VN, Benedict C, Nickoloff E, Khan M, Johnson J, Gumpert AM, Koch WJ, Verma SK. Interleukin-10 inhibits chronic angiotensin II-induced pathological autophagy.J Mol Cell Cardiol. 2015 Dec;89(Pt B):203 – 13.
Vanhaverbeke M, Vausort M, Veltman D, Zhang L, Wu M, Laenen G, Gillijns H, Moreau Y, Van De Bartunek J, Devaux Y, Janssens S, Sinnaeve PR. EU-CardioRNA COST Action CA17129. Peripheral Blood RNA Levels of QSOX1 and PLBD1 Are New Independent Predictors of Left Ventricular Dysfunction After Acute Myocardial Infarction. Circ Genom Precis Med. 2019 Dec;12(12):e002656.
Caillard A, Sadoune M, Cescau A, Meddour M, Gandon M, Polidano E, Delcayre C, Da Silva K, Manivet P, Gomez AM, Cohen-Solal A, Vodovar N, Li Z, Mebazaa A, Samuel JL. QSOX1, a novel actor of cardiac protection upon acute stress in mice. J Mol Cell Cardiol. 2018 Jun;119:75–86.
de Andrade CR, Stolf BS, Debbas V, Rosa DS, Kalil J, Coelho V, Laurindo FR. Quiescin sulfhydryl oxidase (QSOX) is expressed in the human atheroma core: possible role in apoptosis. In Vitro Cell Dev Biol Anim. 2011 Dec;47(10):716–27.
França KC, Martinez PA, Prado ML, Lo SM, Borges BE, Zanata SM, San Martin A, Nakao LS. Quiescin/sulfhydryl oxidase 1b (QSOX1b) induces migration and proliferation of vascular smooth muscle cells by distinct redox pathways. Arch Biochem Biophys. 2020 Jan;15:679:108220.
Smart N, Dubé KN, Riley PR. Identification of Thymosin β4 as an effector of Hand1-mediated vascular development.Nat Commun. 2010 Jul27;1(4):46.
Scheller I, Beck S, Göb V, Gross C, Neagoe RAI, Aurbach K, Bender M, Stegner D, Nagy Z, Nieswandt B. Thymosin β4 is essential for thrombus formation by controlling the G-actin/F-actin equilibrium in platelets.Haematologica. 2021 Aug 5.
Gonzalez-Franquesa A, Stocks B, Borg ML, Kuefner M, Dalbram E, Nielsen TS, Agrawal A, Pankratova S, Chibalin AV, Karlsson HKR, Gheibi S, Björnholm M, Jørgensen NR, Clemmensen C, Hostrup M, Treebak JT, Krook A, Zierath JR, Deshmukh AS. Discovery of thymosin β4 as a human exerkine and growth factor. Am J Physiol Cell Physiol. 2021 Nov 1;321(5):C770-C778.
Freeman KW, Bowman BR, Zetter BR. Regenerative protein thymosin beta-4 is a novel regulator of purinergic signaling. FASEB J. 2011 Mar;25(3):907–15.
Sopko N, Qin Y, Finan A, Dadabayev A, Chigurupati S, Qin J, Penn MS, Gupta S. Significance of thymosin β4 and implication of PINCH-1-ILK-α-parvin (PIP) complex in human dilated cardiomyopathy. PLoS ONE. 2011;6(5):e20184.
Choudhury R, Roy SG, Tsai YS, Tripathy A, Graves LM, Wang Z. The splicing activator DAZAP1 integrates splicing control into MEK/Erk-regulated cell proliferation and migration. Nat Commun. 2014;5:3078.
Sasaki K, Ono M, Takabe K, Suzuki A, Kurihara Y. Specific intron-dependent loading of DAZAP1 onto the cox6c transcript suppresses pre-mRNA splicing efficacy and induces cell growth retardation.Gene. 2018 May30;657:1–8.
Troche C, Aydemir TB, Cousins RJ. Zinc transporter Slc39a14 regulates inflammatory signaling associated with hypertrophic adiposity.Am J Physiol Endocrinol Metab. 2016 Feb15;310(4):E258-68.
Knez M, Graham RD, Welch RM, Stangoulis JC. New perspectives on the regulation of iron absorption via cellular zinc concentrations in humans. Crit Rev Food Sci Nutr. 2017 Jul;3(10):2128–43.
Bansal SS, Ismahil MA, Goel M, Patel B, Hamid T, Rokosh G, Prabhu SD. Activated T Lymphocytes are Essential Drivers of Pathological Remodeling in Ischemic Heart Failure. Circ Heart Fail. 2017 Mar;10(3):e003688.
Delgobo M, Heinrichs M, Hapke N, Ashour D, Appel M, Srivastava M, Heckel T, Spyridopoulos I, Hofmann U, Frantz S, Ramos GC. Terminally Differentiated CD4⁺ T Cells Promote Myocardial Inflammaging. Front Immunol. 2021 Feb;19:12:584538.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Diagnostic and predictive values of ferroptosis-related genes in heart failure

Status:

Version 1

Abstract

Figures

Background

Materials And Methods

Results

Discussion

Conclusion

Declarations

References

Additional Declarations

Status:

Version 1