Identification of macrophage driver genes in fibrosis caused by different heart diseases based on omics integration

Abstract

Background

Myocardial fibrosis, a hallmark of heart disease, is closely associated with macrophages, yet the genetic pathophysiology remains incompletely understood. In this study, we utilized integrated single-cell transcriptomics and bulk RNA-seq analysis to investigate the relationship between macrophages and myocardial fibrosis across omics integration.

Methods

We examined and curated existing single-cell data from dilated cardiomyopathy (DCM), ischemic cardiomyopathy (ICM), myocardial infarction (MI), and heart failure (HF), and analyzed the integrated data using cell communication, transcription factor identification, high dimensional weighted gene co-expression network analysis (hdWGCNA), and functional enrichment to elucidate the drivers of macrophage polarization and the macrophage-to-myofibroblast transition (MMT). Additionally, we assessed the accuracy of single-cell data from the perspective of driving factors, cell typing, anti-fibrosis performance of left ventricular assist device (LVAD). Candidate drugs were screened using L1000FWD.

Results

All four heart diseases exhibit myocardial fibrosis, with only MI showing an increase in macrophage proportions. Macrophages participate in myocardial fibrosis through various fibrogenic molecules, especially evident in DCM and MI. Abnormal RNA metabolism and dysregulated transcription are significant drivers of macrophage-mediated fibrosis. Furthermore, profibrotic macrophages exhibit M1 polarization and increased MMT. In HF patients, those responding to LVAD therapy showed a significant decrease in driver gene expression, M1 polarization, and MMT. Drug repurposing identified cinobufagin as a potential therapeutic agent.

Conclusion

Using integrated single-cell transcriptomics, we identified the drivers of macrophage-mediated myocardial fibrosis in four heart diseases and confirmed the therapeutic effect of LVAD on improving HF with single-cell accuracy, providing novel insights into the diagnosis and treatment of myocardial fibrosis.

Introduction

According to the World Health Organization report, cardiovascular diseases claim more lives annually than any other cause of death. In 2016, an estimated 17.9 million people succumbed to cardiovascular diseases, representing 31% of global deaths [1]. Among these, heart disease stands out as a leading cause of mortality. Clinical studies have identified vascular stenosis, myocardial necrosis, and arrhythmia as common contributors to heart disease [2]. Myocardial infarction (MI), often resulting from coronary stenosis, exemplifies a typical form of heart disease, while dilated cardiomyopathy (DCM) and ischemic cardiomyopathy (ICM) are characterized by myocardial necrosis. Heart failure (HF) represents a condition marked by abnormal heart pumping function, potentially stemming from coronary artery stenosis or myocardial necrosis. Given its high incidence, mortality, and disability rates, HF poses significant health risks [3]. Patients diagnosed with HF face a prognosis of recurrent hospitalizations and mortality, with approximately a 50% mortality rate within five years of diagnosis [4]. Hence, elucidating the pathological mechanisms underlying HF is crucial for the development and enhancement of treatment modalities aimed at reducing mortality risks.

HF encompasses ischemic HF and non-ischemic HF, with the former primarily attributed to coronary artery damage resulting in weakened myocardial contraction due to ischemic cell death and toxic metabolite accumulation. Conversely, non-ischemic HF is characterized by myocardial fibrosis (MF) stemming from excessive immune responses [5]. Myocardial fibrosis promotes myocardial cell hypertrophy and ventricular remodeling, which leads to the destruction of myocardial structure. Long-term heart injury makes it lose its elastic ability, resulting in insufficient venous blood discharge and insufficient arterial blood perfusion, which eventually leads to heart failure [6]. While unresolved inflammatory reactions are considered significant contributors to HF, the complexity of these reactions necessitates further elucidation of HF’s pathological mechanisms. Current understanding posits that inflammatory reactions precipitate MF, elevating HF risk. Excessive immune cell infiltration prompts collagen fiber accumulation, impairing myocardial contractility. Numerous studies have linked DCM, ICM, and MI to MF [2], heightening HF risk. MF primarily results from fibroblast-mediated extracellular matrix (ECM) release, with T cells and macrophages predominantly driving fibroblast activation [6,7,8]. Regulatory T cells are found to infiltrate damaged tissues, recruiting pro-inflammatory macrophages and exacerbating tissue injury [9]. Macrophage phenotype and polarization crucially influence tissue regeneration, with pro-inflammatory (M1) macrophages releasing interleukin and tumor necrosis factor (TNF) to accelerate inflammation, while anti-inflammatory (M2) macrophages secrete IL-10 to promote fibrosis [10]. Fibronectin 1 (FN1) and secreted phosphoprotein 1 (SPP1) are key profibrotic molecules in MF [11]. Recent research [12] indicates that macrophages with elevated FN1 and SPP1 expression coordinate fibroblast activation, contributing to MF and renal fibrosis, with potential reversal via CXCL4 inhibition. Thrombospondin-1 (THBS1), a multifunctional glycoprotein released by macrophages and adipocytes, mediates MF and HF promotion [13, 14]. However, further clarification of these molecular drivers of fibrosis is warranted.

To investigate the drivers of macrophages releasing profibrotic molecules (FN1, SPP1 and THBS1), we utilized an integrated atlas of single-cell transcriptomes in DCM, ICM, MI and HF. Concurrently, we explored the dynamic evolution of profibrotic macrophages by integrating single-cell maps of MI models at different time points. We identified driving genes at the tissue level using three Bulk RNA-seq datasets. Finally, we confirmed the theoretical feasibility of left ventricular assist devices (LAVD) in treating HF at the single-cell level and identified potential drugs based on gene expression.

Methods

Data collection and standardization

Seven single-cell transcriptome datasets, including DCM, ICM, and HF, were retrieved from the GEO database. All datasets are summarized in Table S1. Integration of single-cell data was performed using ‘Harmony’ [15]. The steps for single-cell standardization processing were consistent with Seurat’s tutorial. Cluster biomarkers were defined as genes with adjusted P value < 0.01 and |log₂FC| > 0.5. For bulk RNA-seq data, the ‘sva’ function was employed to remove batch effects.

Differential expression gene analysis

The characteristic genes of standardized single-cell data were annotated using the ‘SingleR’ software package [16]. Furthermore, the ‘FindAllMarkers’ function was employed to analyze the differential expression of genes (DEGs) between normal and diseased states. For Bulk RNA-seq data, DEG analysis was conducted using the ‘limma’ function [17].

Construction of protein-protein interaction network and hub genes acquisition

The STRING database [18] was utilized to predict the protein-protein interactions (PPIs) of module genes. The results were imported into Cytoscape (v3.8.0), and the top 20 genes were determined using four algorithms, and the results obtained were intersected to identify hub genes.

GO and KEGG analysis

The DEGs was identified using the DAVID platform [19], and the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway were selected for functional annotation.

Construction of gene co-expression network and enrichment analysis

The co-expression network of hub genes was constructed, and functional enrichment analysis was performed using the GeneMania platform [20]. The ‘h.all.v2023.2’ dataset was utilized for Gene Set Enrichment Analysis (GSEA) [21].

Prediction of transcription factors in single cell level

SCENIC3 [22] was employed to predict TF. The specific steps are as follows: (a) the cisTarget database was utilized for initialization. (b) Genes matching the database were filtered and retained based on their read counts. (c) The Spearman correlation between TFs and their targets was then calculated.

High-dimensional weighted gene co-expression network analysis (hdWGCNA)

A Seurat object was constructed, and similar cell groups were identified using the K-nearest neighbors method [23]. Subsequently, the average expression of cells was calculated to generate a metacell expression matrix, and the optimal threshold is calculated and selected. Then the gene co-expression network is constructed for the expression matrix, and the characteristic genes and gene connectivity of the coordination module are calculated, and the genes with high connectivity are the characteristic genes of the module. Finally, the module genes were obtained using the ‘GetModules’ function. It is worth noting that modular genes are often the key genes that drive this kind of cell population.

Cell communication analysis between macrophages and fibroblasts

A CellChat object was constructed from the standardized single-cell data [24]. Subsequently, the ligand-receptor database was imported. Over-expressed genes were identified, and the relationship between ligands and receptors was established. This allows the inference of the cell communication network.

Construction of ROC and analysis of gene expression

The expression and grouping information of the hub genes and TFs were imported into the “ROC curve” module on the bioinformatics platform. A single-gene receiver operating characteristic (ROC) curve were constructed with the false positive rate as the abscissa and the true positive rate as the ordinate. The closer the area under the curve (AUC) is to 1, the better the diagnostic effect.

Drug repurposing screening based on hub genes

L1000FWD was utilized to predict drugs targeting the hub genes, and the Schrödinger Maestro was utilized to conduct molecular docking between drugs and the hub genes. Molecular dynamics simulations were employed to assess the binding stability of drugs and proteins, thereby reflecting the dynamic binding process. The ‘System Builder’ module was utilized to construct an water-salt environment, while ‘Molecular Dynamics’ was employed to evaluate the stability of drug-protein binding, with an RMSD < 5 considered indicative of binding stability.

Result

Construction of single cell transcriptome atlas and cell annotation

The single-cell data from 9 normal and 20 DCM patients were integrated (Figure S1A, B), resulting in a total of 153,729 cells (Figure S2E), with 26 cell types identified through cell annotation. Integration of single-cell data from 2 normal and 6 ICM patients yielded 41,534 cells (Figure S1C, S1D, S2B), with 22 cell types identified (Figure S2F). Reanalysis of single-cell data from 14 normal and 13 HF patients yielded 138,367 cells (Figure S2C), with 15 cell types identified (Figure S2G). Reanalysis of sequencing data from 4 normal and 28 patients with MI resulted in 211,517 cells (Figure S2D), with 28 cell types identified (Figure S2H). Additionally, 36,654 cells were obtained from blood single-cell data including DCM and ICM, with five cell types identified after annotation (Figure S3A-C). In DCM patients, there was a significant increase in the proportions of cardiomyocytes, fibroblasts, T cells, and NK cells, while the proportions of monocytes and macrophages decreased (Fig. 1A, Table S2). In ICM patients, there was a slight increase in the ratio of cardiomyocytes, and a significant increase in the ratio of T cells to NK cells, with a significant decreased observed in monocytes and macrophages (Fig. 1B). Consistent with tissue-level findings, blood samples from DCM and ICM patients showed a significant increase in the ratio of T cells and NK cells, along with a significant decrease in the ratio of monocytes (Figure S3D). Compared to the normal group, the proportions of fibroblasts and NK cells did not increase in HF patients, but there was a significant increase in the ratio of monocytes (Fig. 1C). In MI patients, cardiomyocytes, fibroblasts, T cells, and NK cells did not proliferate, but there was a significant increase in the ratio of macrophages (Fig. 1D).

Fig. 1

Cell proportion of four single cell atlas. The cell ratio of DCM, ICM, HF and MI in normal and disease (A-D)

Phenotypes of fibroblasts and macrophages in different diseases and mechanism of macrophages promoting fibrosis

Compared to the normal group, the ECM receptors interaction and the ECM of fibroblasts from DCM, ICM, HF, and MI patients was significantly activated (Fig. 2A). Regarding macrophages, the IFN-γ signal, lipopolysaccharide response, and cytokine signaling were significantly activated in DCM and ICM patients (Fig. 2B). Cell-cell communication analysis reveals that in DCM, eight macrophage clusters release FN1 signals to 21 fibroblast clusters, while four macrophage clusters participate in the interaction with 22 fibroblast clusters through SPP1 signals. Additionally, six macrophage clusters release THBS signals to all fibroblast clusters (Figure S4D). In ICM, two clusters in macrophage release FN1 and SPP1 signals to different fibroblast clusters (Fig. 2D, Figure S4B). In HF patients, cluster 2 in macrophage send FN1 signals to different fibroblast clusters (Fig. 2E). Interestingly, similar to DCM, various macrophage clusters in MI patients can activate fibroblasts through three signals (Fig. 2F, S4C, S4E).

Fig. 2

Enrichment analysis of DEGs and signal transduction promoting fibrosis. Functional changes of fibroblasts in four diseases (A), functional analysis of macrophages in four diseases (B), and FN1 signal release network of macrophages in four diseases (C-F)

Phenotype of DCM and driving gene of profibrotic macrophage

The findings revealed that in DCM, fibroblasts receiving profibrotic signals participate in the occurrence of myocardial fibrosis (Fig. 3A), while macrophages sending profibrotic signals showed extensive activation of Toll-like receptors, TNF, and ERK signaling pathways (Fig. 3B). Expression of various transcription factors was up-regulated in the profibrotic macrophages (Fig. 3C, Figure S4F). Genes identified by hdWGCNA (modules 1 and 5) were deemed as key clusters driving macrophage phenotypic transformation (Fig. 3D). Four algorithms were employed to identify hub gene of key clusters, resulting in a total of 20 genes (Fig. 3E). Expression level of all hub genes was up-regulated in profibrotic macrophages (Fig. 3F). Gene co-expression network and enrichment analysis demonstrated that highly expressed hub genes were localized in the endoplasmic reticulum and associated with mRNA catabolism, rRNA metabolism, and ubiquitination (Figure S5A). Given that macrophage-to-myofibroblast transition (MMT) can directly contribute to MF, it was observed that myofibroblast markers in profibrotic macrophages were significantly up-regulated (Fig. 3G), while M1 macrophage markers were significantly overexpressed (Fig. 3H).

Fig. 3

Driver gene identification of macrophages in DCM. Functional analysis of fibroblasts receiving profibrotic signals (A) and macrophages sending profibrotic signals (B); TFs in different macrophage clusters (C); Key module gene analysis (D), hub gene identification of module gene (E); The expression of hub gene in profibrotic macrophages (F); Expression of MMT markers (G) and polarization markers (H)

Driving gene of profibrotic macrophage in ICM

Similar to DCM, fibroblasts receiving signals exhibited abnormal the ECM and ECM receptors interactions, as well as impaired wound healing in ICM (Fig. 4A). Conversely, macrophages sending profibrotic signals displayed extensive activation of the ERK cascade, cytokine signaling, and Toll-like receptor signaling (Fig. 4B). Expression of CEBPD, EGR1, FOSB, and ATF3 was decreased in profibrotic macrophages (Fig. 4C, Figure S4G). 9 gene modules was identified as the key gene cluster of the 8th cluster of macrophages (Fig. 4D), and 19 hub genes were acquired using four algorithms (Fig. 4E). Gene expression analysis demonstrated that hub genes such as RPL3, RPL4, RPLP0, RPS2, and RPS9 were highly expressed in profibrotic macrophages, with associations to mRNA and rRNA metabolism (Fig. 4F, Figure S5B). Profibrotic macrophages exhibited higher expression of myofibroblast markers (Fig. 4G), as well as significantly increased expression of M1 markers (Fig. 4H).

Fig. 4

Driver gene identification of macrophages in ICM. The legend is the same as Fig. 3

Driving gene of profibrotic macrophage in HF

In HF patients, fibroblasts receiving profibrotic signals exhibited significantly increased integrin-mediated cell adhesion, cell-matrix adhesion, and ECM interaction, consistent with the overall fibroblast phenotype (Fig. 5A). Profibrotic macrophages showed activation of the PPAR signal, ECM, and interferon-γ signal (Fig. 5B). Expressions of FOXO1 and KLF6 was increased in profibrotic macrophages (Fig. 5C, Figure S4H). 13 module genes were obtained as characteristic gene clusters of the profibrotic macrophages (Fig. 5D), and 9 hub genes were identified using four algorithms (Fig. 5E). Gene expression analysis demonstrated that the expression of TLR4 and RUNX1 was decreased in profibrotic macrophages, while others increased, with involvement in the Toll-like receptor signaling pathway, monocyte proliferation, and interferon type I (Fig. 5F, Figure S5C). Additionally, the myofibroblast marker CD44 was highly expressed (Fig. 5G). Meanwhile, the proportion of M1-type markers in profibrotic macrophages was relatively high (Fig. 6H).

Fig. 5

Driver gene identification of macrophages in HF. The legend is the same as Fig. 3

Driving gene of profibrotic macrophage in MI

Similar to DCM, fibroblasts receiving profibrotic signals exhibited fibrosis in patients with MI. Interestingly, abnormal expression of the HIF-1 pathway appears to be associated with fibrosis (Fig. 6A). In addition to extensive activation of inflammatory signals, activation of the HIF-1 pathway is also linked to the promotion of fibrosis (Fig. 6B). Profibrotic molecules released by macrophage clusters may be correlated with high expression of transcription factor (Fig. 6C, Figure S3I). The module 6 identified by hdWGCNA was found to be the key gene cluster (Fig. 6D), with 16 hub genes identified using four algorithms (Fig. 6E). Gene expression analysis revealed that all hub genes were highly expressed in profibrotic macrophages, participating in endoplasmic reticulum stress and transcriptional regulation processes (Fig. 6F, Figure S5D). Furthermore, myofibroblast markers and M1-type marker of profibrotic macrophages were all highly expressed (Fig. 6G and H).

Fig. 6

Driver gene identification in MI macrophages. The legend is the same as Fig. 3

Analysis of common driving factors of profibrotic macrophages

The intersection of hub genes across the four diseases revealed that DCM and ICM share a common gene (Fig. 7A), whereas patients with MI appear to have a distinct set of hub genes. The high expression of RPS2, RPL3, and RPS9 collectively promoted the development of MF in DCM and ICM (Fig. 7C and D). Interestingly, the hub gene identified in DCM was found to be up-regulated in HF, potentially indicating a role in MF (Fig. 7C and E). Intersection of TFs revealed that FOS and FOSB are common transcriptional regulators of DCM and MI, while ATF3, EGR1, and CEBPD are common regulators of DCM and ICM (Fig. 7B). Gene expression analysis showed that FOS and CEBPD were lowly expressed in DCM, ICM and HF, while KLF6 was highly expressed in DCM, HF and MI (Figure S6A, S6C, S6D). The expression patterns of key TFs in ICM and HF were found to be similar (Figure S6B, S6C). Additionally, According to the TF prediction of hub gene, it is found that hub gene is regulated by many transcription factors, especially MI and DCM (Figure S6E-H).

Fig. 7

Analysis of gene intersection and expression of profibrotic macrophages. The intersection analysis of hub gene (A) and TFs (B). The expression of hub gene in profibrotic macrophages of DCM, ICM and HF (C-E)

Dynamic molecular mechanism of MF after MI model

The uniqueness of the driving factors of MI is evident. Hence, the study constructed a single-cell atlas (Figure S7A, S7B) comprising 139,825 cells from MI models across 10 datasets. Cell annotation revealed 16 cell types (Figure S8A), mirroring the cell composition observed in patients with MI. In terms of the ratio of cardiomyocytes and fibroblasts, the condition of MI patients resembles that of 3 days post-infarction, while that of DCM patients resembles that of 14 days post-infarction. A similar trend was observed for monocytes and macrophages (Figure S8B). Overall signaling dynamics experienced fluctuations between fibroblasts and macrophages at post-infarction (Figure S8C-E). However, unlike the overall signaling, THBS was initially released by macrophages, before being released by fibroblasts (Figure S8F-H). Macrophages release various profibrotic molecules at different times post-infarction (Figure S7D, S7F). Notably, fibroblasts are the key to secrete the FN1 at 14 days post-infarction (Figure S7E). Profibrotic macrophages were not regulated by transcription factors at 3 and 14 days post-infarction but were specifically regulated by Hif-1α at 7 days post-infarction (Figure S9A-C). At 3 days post-infarction, hub genes such as Ubc, Rack1, and Kras were highly expressed in Thbs1⁺ macrophages (Figure S9D, S10A), potentially unrelated to macrophage polarization (Figure S11A). Similarly to DCM, all hub genes of Thbs1⁺ macrophages and Fn1⁺ macrophages were highly expressed at 7 days post-infarction (Figure S9E, S10B, S10C), possibly associated with M1 macrophage polarization (Figure S11B, S11C). Rack1 and Rps27a of profibrotic macrophages were highly expressed at 14 days post-infarction (Figure S10D-F). The expression pattern of hub genes in Fn1⁺ macrophages and Spp1⁺ macrophages was consistent (Figure S10E, S10F), likely linked to M1 macrophage polarization (Figure S11D-F).

Bulk RNA-seq integration and gene expression analysis

The “sva” algorithm was employed to integrate six datasets, including Normal, DCM, and ICM (Figure S12). Patients with DCM and ICM exhibited activation of epithelial-mesenchymal transition (EMT) and inhibition of IL-6-JAK-STAT3 signaling and TNF-α signaling across three datasets (Figure S13). Compared with normal, hub genes CD74, RPS9, and EEF1A1 were up-regulated in DCM and ICM. RPS2 and RPL7 were down-regulated in DCM and ICM (Fig. 8A, C, E). Additionally, ATF3 and CEBPD were significantly down-regulated in both DCM and ICM, while EGR1 and FOSB were significantly up-regulated in ICM (Fig. 8B, D, F).

Fig. 8

Analysis of the expression of hub genes and transcription factor. Analysis of the expression of intersection hub genes in three Bulk RNAseq data sets (A, C, E), the expression of TFs in DCM patients (B) and the expression of common TFs in three data sets (B, D, F)

Diagnostic performance and gene expression correlation

The ROC demonstrated that CD74 exhibited better diagnostic efficiency (Figure S14A-C), while CEBPB emerged as the top TF (Figure S14D-F). Across three datasets, hub genes CD74, EEF1A1, and RPS2 showed a strong positive correlation with EMT genes in DCM (Figure S15A-C). Conversely, the expression of EGR1 and FOSB exhibited a strong correlation in ICM (Figure S15D-F).

Effect of LVAD therapy on driving gene expression

Single-cell data, encompassing HF patients undergoing LVAD therapy, were analyzed, resulting in a total of 185,881 cells representing 15 cell types (Fig. 9A and B). Cell proportion revealed that the proportion of cardiomyocytes and monocytes decreased in HF patients responding to LVAD treatment, while the proportion of T cells and macrophages increased (Fig. 9C). The response to LVAD treatment was reflected in the expression of driving genes (Fig. 9D and E). For instance, LYN and RUNX1 exhibited increased expression in HF patients, but markedly decreased in those responding to LVAD treatment, albeit less prominently in non-responders. LVAD treatment reversed the polarization state and MMT of macrophages in HF patients (Fig. 9F and G), providing a theoretical basis for HF treatment. However, in the treatment of DCM and ICM with LVAD, hub genes failed to discern the therapeutic effect of LVAD in myocardial tissue (Figure S16).

Fig. 9

LVAD regulates the gene expression of HF macrophages. cell grouping (A), cell annotation (B) and cell type ratio (C). Expression of hub genes (D), transcription factor (E), macrophage polarization marker (F) and MMT marker (G) in HF treated by LVAD

Drug repurposing analysis based on driving genes

Based on the driving gene expression pattern, five drugs were identified (Table S3). Molecular docking analyses revealed that five drugs combined with driver genes in the form of hydrogen bonds (Figure S17). Molecular dynamics simulations indicated that four drugs were stable in binding with EEF1A1, RPS2, and RPL7 (Fig. 10A, B and D), while three drugs were stable in binding with RPS9 (Fig. 10C), with cinobufagin showing stability in binding with all proteins.

Fig. 10

Analysis of drug-protein binding stability. RMSD values of five drugs binding with EEF1A1, RPS2, RPS9 and RPL7 (A-D)

Discussion

MF primarily involves the release of ECM and collagen by fibroblasts. In the early stages, innate immune cells become overactivated, creating an inflammatory environment that damages local tissues. Subsequently, there’s a shift towards inflammatory inhibition and activation of repair cells by anti-inflammatory immune cells. However, excessive repair can lead to poorly remodeled tissues, exacerbating HF [25]. The significant increase in DCM cardiomyocytes is a key factor contributing to decreased myocardial contractility. This study observed a notable rise in myocardial cell proportion in patients with cardiomyopathy (DCM and ICM), but not in HF and MI, suggesting a unique phenomenon in cardiomyopathy. Traditionally, fibroblast proliferation has been considered a hallmark of MF [26]. However, aside from cardiomyopathy, HF, and ischemic heart disease did not show an increased proportion of fibroblasts, despite consistent fibrosis. Hence, evaluating fibrosis solely based on fibroblast proliferation may be unreasonable [27]. T cells [28], comprising pro-inflammatory types (Th1, Th2, and Th17) and anti-inflammatory types (regulatory T cells), play diverse roles. Pro-inflammatory T cells create an inflammatory milieu that reduces collagen synthesis in fibroblasts, while anti-inflammatory T cells promote fibrosis by secreting IL-10. The increase in T cells observed in cardiomyopathy, particularly influencing MF, differs from ischemic heart disease. NK cells are implicated in renal fibrosis and liver fibrosis [29, 30], suggesting a potential avenue to combat fibrosis. Similar to T cells, NK cells showed an increase in cardiomyopathy, particularly impacting ICM. Conversely, in MI, macrophage proliferation may pose a potential risk. In summary, examining cell ratios reveals significant differences between cardiomyopathy and ischemic heart disease, influencing pathological progression. Personalized treatment approaches tailored to these differences may offer benefits to patients.

Fibroblasts exhibit abnormal metabolism of the ECM and collagen post-MF, yet this phenomenon varies across different diseases. For instance, DCM and MI display significant changes during tissue repair, suggesting that MF plays a crucial role in specific diseases. Moreover, this study highlights that macrophages in cardiomyopathy exhibit M1 polarization, characterized by reactions to lipopolysaccharide and interferon-γ signals [31], whereas HF and MI show the opposite trend. This discrepancy may be attributed to the high heterogeneity of immune cells. Notably, macrophages play a significant role in promoting fibrosis, as evidenced by their substantial recruitment in lung fibrosis induced by coronavirus [32]. Macrophages participate in fibroblast activation through pathways involving FN1, THBS1 and SPP1, primarily concentrated in DCM and MI, aligning with the sedimentary facies of fibroblast ECM. Interestingly, in all four diseases, profibrotic macrophages undergo transformation into myofibroblasts and exhibit an M1-type macrophage phenotype. This common profibrotic expression underscores the pivotal role of macrophages in fibrosis. However, the drivers of this phenomenon vary across diseases. Both types of cardiomyopathy display abnormal RNA metabolism, yet the up-and-down regulation of hub genes differs. Nonetheless, DCM and MI exhibit striking consistency in the expression patterns of driver genes, reflected in the extensive activation of their TFs. Notably, profibrotic macrophages in DCM and MI are regulated by FOSB, which plays a role in cell differentiation regulation. Intriguingly, FOSB is also implicated in the high expression of hub genes in DCM and MI. However, this up-regulation of FOSB is not identified in bulk RNA-seq. Conversely, LAVD treatment of HF significantly reverses the expression pattern of macrophages. Additionally, ATF3, an adaptive response gene, is highly expressed in DCM fibroblasts and participates in regulating hub gene expression. However, ATF3 is not recognizable at the tissue level, indicating its role as a cell-specific transcriptional regulator. These findings underscore a strong internal correlation between cardiomyopathy and ischemic heart disease.

MI is a prototypical ischemic heart disease, and the study highlights significant differences between MI and cardiomyopathy in terms of hub gene expression. Additionally, profibrotic macrophages in MI display distinct pathological phenomena, characterized by increased hypoxia. The study observed a shift in the proportion of fibroblasts from decreasing to increasing after MI, alongside a trajectory of macrophage proliferation followed by a decrease, mirroring the transition from MI to cardiomyopathy observed in patients. Analysis of hub gene expression patterns revealed a transition from extensive activation in at 7 days post-infarction to high expression of RNA metabolism-related genes at 14 days post-infarction, consistent with observations in patients with cardiomyopathy. Furthermore, LVAD was used in HF treatment demonstrated promising outcomes in terms of hub gene expression, MMT and macrophage polarization, offering a new scientific basis for LVAD therapy. Cinobufagin, known for its positive inotropic effect similar to digitalis glycosides, has been found to alleviate HF induced by acute ischemia [33]. Drug repurposing studies identified cinobufagin as a candidate drug for HF treatment, demonstrating stable binding to several hub genes. This suggests its potential therapeutic efficacy in HF management. This comprehensive approach sheds light on the intricate mechanisms underlying ischemic heart diseases and offers novel insights into potential therapeutic strategies.

Overall, this research provides a comprehensive understanding of the molecular mechanisms underlying cardiomyopathy, ischemic heart disease, and HF. By shedding light on the pathophysiological processes and therapeutic interventions, it opens new avenues for the development of diagnostic biomarkers and targeted therapies, ultimately improving patient outcomes in HF management.