Introduction

Anthracyclines (ANTs) are widely utilized in the clinical treatment of human malignancies and are generated from Streptomyces peucetius var. caesius [1]. Doxorubicin (Dox), epirubicin (Epi), and idarubicin (Ida) are typical ANTs. However, anthracycline-induced cardiotoxicity (AIC) develops in a dose-dependent cumulative pattern and results a range of cardiovascular outcomes, rapidly attracting attention [2, 3]. In a prospective follow-up of 2625 adult patients receiving anthracyclines, Cardinal et al. reported a 9% incidence of cardiotoxicity. In 98% of cases, AIC was identified during the first year following the end of treatment [4].

Numerous types of research have clarified the fundamental mechanisms of AIC, with oxidative stress triggered by an imbalance between the generation and neutralization of reactive oxygen species (ROS) being the most frequently reported molecular mechanism [5, 6]. Higher ROS levels generated by ANTs’ direct interaction with the mitochondrial electron transport chain [7] exceed the antioxidant capacity and damage cardiac tissue susceptible to free radical damage [8]. However, the combination of ANTs with various antioxidants, such as manganese superoxide dismutase (Mn-SOD) and glutathione peroxidase 1 (GPx1), has inconsistent cardioprotective effects [9,17]. Ferroptosis highlighted by iron-dependent lipid peroxidation culminating in membrane rupture, is a newly defined nonapoptotic programmed cell death type [18,19,20]. ANTs accumulate in mitochondria by embedding into mtDNA in cardiomyocytes, disrupting iron utilization, and leading to iron overload; therefore, ANTs and iron accumulation synergistically trigger ferroptosis [20, 21]. Iron overload is typically linked to oxidative stress, which is defined by elevated levels of ROS and reduced levels of antioxidant enzymes [22], such as tetrahydrobiopterin (BH4), coenzyme Q10 and glutathione (GSH). GSH functions as the primary antioxidant in mammalian cells, essential for intracellular and extracellular equilibrium and ferroptosis [23]. In brief, reduced GSH levels lead to the synthesis of phospholipid hydroperoxides (PLOOH), which contribute to ROS accumulation and ferroptosis. Targeting ferroptosis, such as Fer-1, holds promise as a novel therapeutic strategy for cardioprotection [17, S1).

In vitro ORMF transcriptome profile clustering tree with samples split into two major groups and all control samples equally distributed along one branch (Fig. 2A).

Fig. 2
figure 2

Clustering of samples and detection of soft-thresholding power. (A) The hierarchical clustering tree was conducted to filter outliers. All samples are distributed in clusters, with controls scattered among the red branches of the cluster tree. (B) Analysis of the scale-free fit index and the mean connectivity for numerous soft-thresholding powers. (C) Dendrogram of all genes organized corresponding to a dissimilarity measure (1-TOM), where each branch represents a gene, and low-hanging clusters represent gatherings of genes (modules) with roughly comparable network attributes (branches) on the tree

Full size image

In order to uncover the responsive modules and hub genes in the adverse mechanisms of anthracycline cardiotoxicity, a weighted gene co-expression network was constructed using ORMFs. To construct the scale-free network, we selected β = 6 as the soft threshold (Fig. 2B). A total of 7 modules (Fig. 2C) were obtained based on the scale-free network. Among the modules, the blue module was strongly associated with in vivo drug exposure levels (Fig. 3). As a result, the blue module served as the hub module. The blue module contained 931 genes.

Fig. 3
figure 3

Heatmap of module-trait relationships displaying associations between module eigengenes and clinicopathological characteristics. A module eigengene is represented by each row, while clinicopathological characteristics are represented by each column. Numbers in the table refer to the correlation r and the P value in parentheses. The color legend illustrates the degree of correlation. Red and green colors indicate positive or negative relationships

Full size image

Selection of key genes and biological processes

As demonstrated by their high KME (eigengene connectivity) value, hub genes have the most connections in the network. Therefore, 17 genes with KME ≥ 0.95, including 1 ferroptosis-associated gene, 2 mitochondrial function-associated genes, and 16 oxidative stress-associated genes, were identified as hub genes.

After investigating the intersection of hub genes and the dynamic response genes, 8 genes related to AIC progression were obtained: CD36 Molecule (CD36), Cadherin 5 (CDH5), Chitinase 3 Like 1 (CHI3L1), Hemoglobin Subunit Alpha 2 (HBA2), Hydroxysteroid 11-Beta Dehydrogenase 1 (HSD11B1), Osteoglycin (OGN), Ribosomal Protein L8 (RPL8), and Von Willebrand Factor (VWF).

In order to identify biological functions and the pathways associated with the adverse mechanisms of AIC, we further conducted enrichment analysis for genes in the blue module. Gene Ontology (GO) enrichment analysis indicated that enriched biological functions and pathways related to oxidative stress and mitochondrial function were significantly enriched. Surprisingly, the results revealed that pathways pertaining to angiogenesis and the immune system, namely GO:0045766 positive regulation of angiogenesis and GO:0002683 negative regulation of immune system process, were both engaged and then chosen for more in-depth research (Fig. 4, Table S1).

Fig. 4
figure 4

The results of GO enrichment analysis based on biological processes (BP) of blue module genes

Full size image

Expression of the key genes and the biological processes

All 8 key genes were consistently differentially expressed across all ANT-treated conditions, and their expression levels varied over time. The expression of all the important genes, except RPL8, was reduced in the conditions treated with ANT in vitro. (Fig. 5). Notably, the protein level of Hemoglobin subunit alpha (HBA) encoded by HBA2 varied over time under conditions of Dox treatment in vitro (Fig. 6).

Fig. 5
figure 5

The dynamic alterations of the key genes characterized log2FC expression values at each time between anthracycline (ANT) and control samples. (A) CD36, (B) CDH5, (C) CHI3L1, (D) HBA2, (E) HSD11B1, (F) VWF, (G) RPL8, (H) OGN. The: therapeutic dose, Tox: toxic dose; 002, 008, 024, 072, 168, 240, 336 are corresponding exposure periods in hours; con: the control group; Dox: doxorubicin, Ida: idarubicin; Epi: epirubicin

Full size image
Fig. 6
figure 6

The dynamic alterations of Hemoglobin subunit alpha (HBA) characterized log2FC expression values at each time between doxorubicin (Dox) and control samples. The: therapeutic dose, Tox: toxic dose; 002, 008, 024, 072, 168, 240, 336 are corresponding exposure periods in hours; con: the control group; Dox: doxorubicin

Full size image

The quantitatively assessed angiogenesis score for each sample in the current study was calculated. Single-sample gene set enrichment analysis (ssGSEA) was also employed to assess the degree of immune cell infiltration in each sample. The dynamic characteristics of the angiogenesis score and the immune cells were also analyzed in the present study to identify dynamically altered pathways. The angiogenesis score and immune cell types, like activated CD4 T cells, immature B cells, memory B cells, natural killer cells, type 1 T helper cells, and Type 2 T helper cells, were down-regulated and showed a remarkable alteration across all ANT-treated conditions (Fig. 7).

Fig. 7
figure 7

The dynamic alterations of the angiogenesis score and the immune cells were characterized by comparing the expression levels of anthracycline (ANT) samples with control samples at each time point. (A) The angiogenesis score, (B) Activated CD4 T cell, (C) Immature B cell, (D) Memory B cell, (E) Natural killer cell, (F) Type 1 T helper cell, (G) Type 2 T helper cell. The: therapeutic dose, Tox: toxic dose; 002, 008, 024, 072, 168, 240, 336 are corresponding exposure periods in hours; con: the control group; Dox: doxorubicin, Ida: idarubicin; Epi: epirubicin

Full size image

In the human heart biopsy samples, there were no statistically significant differences in the expressions of key genes, immune cells, and angiogenesis score between cancer patients who received ANT and non-ANT treatment vs. non-cancer patients. In GSE59672, the expression of OGN (Fig. 8) and the immune cell types of immature B cells, natural killer cells and Type 2 T helper cells (Fig. 9) were dysregulated. In GSE2965, the expression level of RPL8 varied over time (Fig. 10). Notably, due to the limited number of probes, GSE2965 has only the key genes HSD11B1, CD36, RPL8 and HBA2, while pathway expression could not be analyzed. To explore the potential mechanism of RPL8 and CHI3L1 in AIC, Fig. 11 shows the pathways predicted by GSEA.

Fig. 8
figure 8

The expression of OGN was dysregulated in GSE59672

Full size image
Fig. 9
figure 9

The immune cell types of immature B cells, natural killer cells and Type 2 T helper cells were dysregulated in GSE59672. (A) immature B cells (B) natural killer cells (C) Type 2 T helper cells

Full size image
Fig. 10
figure 10

The expression level of RPL8 varied over time in GSE2965

Full size image
Fig. 11
figure 11

The pathways correlated with CHI3L1 and RPL8. (A) CHI3L1 (B) RPL8

Full size image

Molecular docking of anthracyclines with the key genes

Molecular docking was used to to characterize the interaction of anthracyclines with key genes. The protein product of RPL8 and CHI3L1 genes showed significant binding affinity for anthracyclines (binding affinity < -4.8). Notably, GSE2965 did not detect CHI3L1 expression due to the limited number of probes, whereas GSE59672 had a small number of samples, and the expression of RPL8 and CHI3L1 was different but not statistically significant (Figure S2). The amino acid residues of RPL8 and CHI3L1 interacting with doxorubicin are shown in Fig. 12. Gene names, protein database IDs and binding affinities are shown in Table S2.

Fig. 12
figure 12

The 3D diagrams of Dox and its binding to RPL8 and CHI3L1. (A) RPL8 (B) CHI3L1

Full size image

Discussion

AIC is a multifactorial disorder involving oxidative stress, mitochondrial malfunction, and ferroptosis. A single regulator cannot adequately figure out these factors. ROS generation and release are frequently associated with ferroptosis, which may result in mitochondrial dysfunction. The effect of ANTs on ORMFs was investigated in an in vitro cardiac microtissue experiment and heart failure biopsies from patients treated with ANTs in this study. The dynamic alterations of these genes and their cellular mechanisms were appraised. Our study is the first to employ ORMFs to describe the time course of ANTs-induced cardiotoxicity and guide effective therapy.

These key genes could be biological indicators for cardiotoxicity and provide a novel understanding of the dynamic alterations caused by ANTs. One of the key genes was VWF, an acceptable biomarker of endothelial dysfunction in vascular diseases [30]. VWF, elevated throughout chemotherapy, particularly in individuals with vascular disorders [31], could be used as a diagnostic strategy for early diagnosing vascular diseases in chemotherapy-treated cancer patients [32]. CD36 (fatty acid translocase), coordinating myocardial lipid metabolism by modulating lipid signaling and promoting the metabolism of long-chain fatty acids on cardiomyocytes, is upregulated in models for advanced cancers. In the therapy of AIC, CD36 is a significant target. Our results are inconsistent with the reported [31, 32], with decreased VWF and CD36 expression relative to controls. VWF were assessed before and until 1 year after chemotherapy in metastatic testicular cancer patients cured with bleomycin, etoposide and cisplatin. VWF increased during chemotherapy, especially in patients with vascular events. We hypothesize that the reason for the inconsistency are as follows: Cardiotoxicity due to ANTs is dose-dependent and might manifest immediately or take time to manifest. Variable doses of ANTs and study endpoints across studies, which interfere with the quantitative assessment of genes. Additionally, many animal and human studies focus on whole blood RNA specimens, whereas our study is based on cardiac RAN, and cardiac-based RNA is particularly scarce in human studies.

The function of ORMFs implies the pathways of ANTs, especially the immunological infiltration and the angiogenesis signaling pathway, to be critical issues in the ongoing studies of ANTs-induced cardiotoxicity. Immunity contributed significantly to the occurrence and progression of AIC [33]. Proteins in the blood represent drug-induced immune-mediated reactions and may potentially be sensitive biomarkers of drug-induced cardiotoxicity. High baseline IgE levels at cardiotoxic chemotherapy exposure are related to a lower risk of AIC in breast cancer patients receiving doxorubicin and trastuzumab therapies [34]. One of the hypothesized protective effects of high IgE is that it enhances the induction of T cell-mediated cytotoxicity by stimulating myeloid antigen-presenting cells [35]. Another study identified that breast cancer patients showed elevated plasma levels of cytokines CCL27 and macrophage migration inhibitory factor (MIF) following two cycles of Dox [36, 37]. Dox has been demonstrated to disrupt the immune system through innate immune cells, leading to progressive cardiac damage. Neutrophil infiltration in the heart plays a significant role in cardiac toxicity caused by doxorubicin. In breast cancer patients with Dox-induced cardiotoxicity, a high plasma neutrophil extracellular trap level was seen [38]. Another study confirmed that neutrophil elevation in cardiac tissue reached its highest point seven days following a single bolus of Dox administration [39]. Mechanisms of neutrophil-induced acute interstitial cystitis (AIC) include disrupting vascular structures like pericytes and endothelial cells and increasing collagen deposition [40]. Macrophage undergoing marked phenotypic and functional changes has been associated with AIC evolution. Dox-induced cardiotoxicity affects the balance of circulating M1 pro-inflammatory macrophages recruited from monocytes and M2 anti-inflammatory macrophages engaged in tissue repair [41]. The M1 macrophage population regulating the development of cardiac injury was elevated [42, 43], while the M2 macrophage population was suppressed in Dox-induced acute cardiotoxicity mice models [42, 44, 45]. Immunomodulatory strategies to reverse cardiac M1/M2 macrophage imbalance due to Dox-induced transfer of mitochondria, gut microbiota modulation, inflammation, oxidative stress, apoptosis, and autophagy, may prevent the doxorubicin-induced cardiotoxicity [46,47,48,49]. Curiously, our study did not screen for the above immune cells, only showing dynamic changes of Activated CD4 T cell, Immature B cell, Memory B cell, Natural killer cell, Type 1 T helper cell, and Type 2 T helper cell over time and there are few reports about these cells. These novel findings suggest the immune system is a prospective regulator of AIC. However, the role of the immune system in AIC requires a more detailed study.

Regarding the toxic mechanisms of AIC, vascular homeostasis, and angiogenesis have received little attention [5, 50, 51]. Numerous cells, such as endothelial cells (EC), vascular smooth muscle cells, and fibroblasts, as well as a variety of vasoactive chemicals, such as nitric oxide (NO) and vascular endothelial growth factor (VEGF), and their downstream intracellular signaling pathways, are involved vascular homeostasis and angiogenesis. Vascular endothelial growth factor-B (VEGF-B) regulates the proliferation, sprouting, and migration of ECs and can potentially induce coronary vessel growth and cardiac hypertrophy [52]. Dox therapy, both in vivo and in vitro, demonstrated a considerable effect on reducing the expression of VEGF-B [53]. VEGF-B gene therapy prevented Dox-induced DNA damage and mitochondrial dysfunction, increased left ventricular volume, and did not stimulate tumor growth In Dox-treated mice [51]. VEGF-B may be a suitable and powerful treatment for AIC, in line with previous studies demonstrating its benefits in various heart disease models [54,55,56]. Dox-induced cardiotoxicity was further demonstrated to trigger endothelial damage by reducing the release and activity of critical endothelial factors leading to endothelial cell death, disrupting the endothelial barrier function, and increasing vascular permeability. This can precipitate the onset and progression of cardiomyopathy [57, 58]. Therefore, the biochemical and molecular properties of vascular homeostasis undergoing AIC therapy must be considered when develo** therapeutic drugs to comprehensively treat the entire cardiovascular system.

This study has several limitations that need to be addressed. Firstly, there was no statistical difference in key gene expression, immune cells, and angiogenesis scores between cancer patients and non-cancer patients receiving ANT and non-ANT treatment, which could be attributed to the limited quantity of human samples, and the specific reasons need to be further elucidated by external data sets. Secondly, quantitative studies at the microtissue cellular level, such as key gene expression and immune cell infiltration, are inconsistent with the reported results of animal and human specimens, and further research is needed. Thirdly, although our study shows that AIC is related to angiogenesis, further research is required to elucidate the specific cells and pathways involved in this mechanism.

Conclusion

Increasing evidence shows that ORMFs have an indispensable role in AIC. Our study found that the dynamic alternation of oxidative stress - mitochondrial dysfunction - ferroptosis genes may impact cardiac immune functions and angiogenesis. These findings implicate an angle that warrants thorough investigation.