Ferroptotic cardiomyocyte-derived exosomes promote cardiac macrophage M1 polarization during myocardial infarction

Mouse model of MI

SPF C57BL/6J male mice were purchased from the Junke Biological Co., LTD. (Nanjing, China). All mice were reared adaptively for one week at 20–24 °C with free access to food and water under natural light with day and night. After that, all mice were randomly divided into sham group (n = 5) and MI group (n = 5). To construct the MI model, mice were anesthetized with 5% isoflurane inhalation followed by thoracotomy and ligation of the left anterior descending (LAD) coronary artery. For the sham group, mice were underwent only thoracotomy without LAD ligation. At the end of the study, the mice were euthanized by 5% inhalation of isoflurane and cervical dislocation. The heart tissue of mice were isolated and frozen for subsequent experiments. All animal procedures were approved by the Animal Welfare and Ethics Group, Department of Experimental Animal Science, Fudan University (2020 Huashan Hospital JS-574).

Echocardiographic examination

The left ventricular systolic function of mice was detected by echocardiographic examination. Seven days after the operation, 3 mice in each group were randomly selected to observe their echocardiograms on the Mindray (Mindray Medical Co., Ltd.). Conventional parameters of echocardiography were collected including left ventricular end-diastolic diameter (LVEDD), left ventricular end-systolic diameter (LVESD), fractional shortening (FS), and ejection fraction (EF). EF (%) = (LVEDV-LVESV)/ LVESV ×100%; FS (%) = (LVEDD-LVESD)/LVESD ×100%.

Hematoxylin-Eosin (H&E) staining

Histopathological changes of the MI-induced heart tissue (3 mice per group) were observed by H&E staining. Seven days after the operation, all mice were euthanized by using 5% isoflurane inhalation followed by cervical dislocation, and then thoracotomy to removed heart tissue. Isolated heart tissue were underwent fixation using 4% paraformaldehyde, and paraffin embedding, and then cut into 5 µm-thick sections. Later, sections were xylene dewaxed and alcohol gradient dehydration followed by stained with haematoxylin and eosin. Finally, sections were imaged under an optical microscope.

Ferroptosis indicators measurement

In this study, malondialdehyde (MDA) content and total cellular iron (Fe²⁺) concentration were used to indicate the occurrence of ferroptosis. MDA content and total cellular concentration of Fe²⁺ in heart tissue and HL-1 cells were detected by using a MDA kit (S0131, Beyotime, China) and the colorimetric Iron Colorimetric Assay Kit (#E1042, APPLYGEN, China), respectively, according to the manufacturer’s instructions.

Western blotting

Total protein was extracted using RIPA lysis buffer (Thermo Fisher Scientific, Waltham, MA, USA) from heart tissue and HL-1 cells, and the concentration of the protein was detected by the BCA assay kit (Thermo Fisher Scientific, Waltham, MA, USA). Next, about 20 µg protein was subjected to 12% SDS-PAGE and then transferred on PVDF membrane (Sangon, Shanghai, China). The PVDF membranes were washed with water and stained with Ponceau S to confirm transfer, after that blocked with TBST solution contained 5% skimmed milk at 4 °C overnight. Later, PVDF membranes were incubated with primary antibodies at 4 °C overnight followed by incubated with secondary antibody Goat Anti-Mouse IgG H&L(HRP) (1:10000, ab205719, Abcam, Cambridge, UK) at 25 °C for 2 h. The protein bands were imaged by chemiluminescence apparatus (Shanghai Qinxiang, China) and quantified by Image J software. Primary antibodies including: Wnt1 (1:1000, sc514531, Santa Cruz), β-catenin (1:1000, 8480T, Cell Signaling Technology), IL-10 (1:1000, bs-0698R, Bioss), NOS2 (1:1000, sc7271, Santa Cruz), CD63 (ab68418, Abcam, Cambridge, UK), ACSL4 (1:10000, ab155282, Abcam, Cambridge, UK), TFRC (1:1000, ab214039, Abcam, Cambridge, UK), GAPDH (1:20000, 60004-1-Ig, Proteintech).

Cell culture and treatment

Mouse cardiac myocytes line of HL-1 was purchased from Shanghai Zhongqiao Xinzhou Biotech (ZQ0920; Shanghai, China) and mouse bone marrow derived macrophages line of RAW 264.7 was purchased from Procell (CP-M141, China). HL-1 cells and RAW 264.7 cells were cultured in DMEM culture medium contained 10% FBS and 1% penicillin/streptomycin (P/S; E607011, Sangon, China) at 37 °C in an atmosphere of 95% air and 5% CO₂.

To explore the effect of ferroptosis on exosome derived from cardiac myocytes, HL-1 cells were maintained at hypoxic conditions with 7% O₂ and treated with 50 µM ferrostatin-1 (Fer-1), a ferroptosis inhibitor, for 16 h. HL-1 cells treated with PBS as control. Later, culture supernatants were harvested for exosome isolation.

To explore the role of Wnt/β-catenin in exosome-mediated macrophage polarization, RAW 264.7 cells were treated with 10  µM IWR-1 (681669, Sigma-Aldrich), a Wnt/β-catenin signaling pathway inhibitor, for 48 h. Later, Wnt expression was detected by RT-qPCR and western blot.

Isolation and identification of exosome

Exosome derived from HL-1 cells was extracted using Exosome Isolation Kit (from cell culture media) (UR52121, Umibio, Shanghai, China) according to the manufacturer’s instructions. The morphology and size were observed by transmission electron microscopy (TEM). In briefly, 10 µL solution of exosome was placed onto the copper wire mesh and precipitate for 1 min. Then, washed with PBS and dropped 10 µL 2% phosphotungstic acid and precipitate for another 1 min. Finally, whole copper wire mesh was dried at room temperature for 2 min and subjected to image on a TEM (JEM-1200EX, Japan). The concentration and distribution of exosomes were tracked by nanoparticle tracking analysis (NTA) using Zetaview Particle Metrix and its corresponding software ZetaView 8.04.02.

Uptake of exosomes

The exosome derived from HL-1 cells-treated with erastin or PBS were labeled with DiI-red fluorescent kit (C1036, Beyotime, China), according to the manufacturer’s instructions. Next, DiI-labeled exosome were re-suspended in complete medium at a concentration of 5 µg/mL and then added to a dish for fluorescence microscopy in which RAW 264.7 cells were precultured for 24 h. After 12 h of coculture, the medium was removed by centrifugation and the RAW 264.7 cells were fixed with 4% paraformaldehyde and permeabilized with 0.2% Triton X-100, and stained with DAPI. Finally, RAW 264.7 cells was imaged on a fuorescence microscope. Quantification of exosome uptake was measured by ImageJ software according to mean fluorescent intensity.

Flow cytometry sorting of macrophages

After incubation of exosomes, M1-polarized and M2-polarized macrophages were separated by flow cytometry. In short, RAW 264.7 cells were centrifuged at 350g for 5 min to remove medium and re-suspended in PBS to incubated with NOS2-Alexa Fluor^® 488 (sc-7271, Santa Cruz) and Arginase-1 (D4E3M™) XP^® Rabbit (# 93668S, Cell Signaling Technology) at 25 °C for 30 min followed by a incubated with Cy3-labeled Goat Anti-Rabbit IgG(H+L) at 25 °C for 30 min in the dark. After centrifugation and resuspension, RAW 264.7 cells were sorted by CytoFLEX LX Flow Cytometer (Beckman, USA).

Phagocytosis of macrophages

Phagocytosis of the RAW 264.7 cells was investigated using Phagocytosis Assay Kit (Red Zymosan) (ab234054, Abcam, Cambridge, UK), according to the manufacturer’s instructions. In briefly, after incubation of exosomes (Macrophages without exosome treatment served as a blank control), RAW 264.7 cells were incubated with 5 µL of prelabeled zymosan particles for another 3 h and then washed with cold phagocytosis assay buffer. Finally, RAW 264.7 cells were observed by fluorescent microscopy. The phagocytic index is equal to the number of zymosan particles in the cell divided by the number of phagocytic cells multiplied by 100%.

MiRNA mimics and siRNA transfection

MiR-106b-3p mimics and mimics NC were synthesized by Sangon, China. HL-1 cells seed on six-well plates at a density of 2  × 10⁵ cells/well. When reaching about 80–90% confluence, cells were transfected with miR-106b-3p mimics or mimics NC using Lipofectamine™ 2000 (Invitrogen, Waltham, MA). MiR-106b-3p mimics, mimics NC, and Lipofectamine™ 2000 regent were diluted in 45 µL OPTI-MEM (CORNING). After that, diluted miR-106b-3p mimics was mixed with diluted Lipofectamine™ 2000 regent for 20 min, and then added into HL-1 cells. Cultured for another 24 h and transfection efficiency was measured using RT-qPCR.

RT-qPCR analysis

Total RNA was isolated using TRIzol reagent (Sigma-Aldrich, St. Louis, MO) and RNA quality was determined by using NanoDrop. To quantify the Wnt1 mRNA expression, RNA was reversed into cDNA using PrimeScript™ RT reagent Kit (Takara, China) and then the qPCR reaction was carried out using SYBR^® Premix Ex Taq II (Takara, China) according to the manufacturer’s protocol. For detecting the miRNAs expression, RNA was reversed into cDNA using PrimeScript RT reagent kit with gDNA Eraser (Takara, China) followed by qPCR reaction using TB Green Premix Ex Taq (Takara, China). GAPDH was used to normalize mRNA expression, and U6 was used to normalize miRNA expression. The 2−ΔΔCT method was used for calculating the relative expression level. All primers used in this study were shown in Table S1.

Statistical analysis

The statistical analysis of all data were presented as the mean ± SD of three independent experiments and processed by software of GraphPad Prism 9.0 (GraphPad Software, USA). Moreover, student’s t-test was used to compare with two groups and ANOVA followed by Tukey’s test was used to compare three groups. P <0.05 was considered statistically significant.

Cardiomyocyte ferroptosis accompanied by M1 macrophage infiltration during MI

We firstly established a mouse model of MI using LAD ligation and then examined cardiac function with echocardiography. Compared to the sham group, LVIDs (P = 0.0018) and LVIDd (P = 0.00015) in MI mice were significantly increased (Figs. 1A and 1B). In contrast, the EF (P = 0.017) and FS (P = 0.006) were substantial reduced post-MI, compared with sham group (Figs. 1A and 1B). Likewise, H&E staining further revealed that histomorphology of myocardial tissues in sham group was normal but badly damaged in MI group, as indicated by disordered cardiomyocyte and massive inflammatory cell infiltration (Fig. 1C). To determine whether myocardial tissue injury is related to ferroptosis, we measured two indicators of ferroptosis such as MDA content and the concentration of Fe²⁺. The results showed that, compared to the sham group, MDA content (P = 0.049; Fig. 1D) and the concentration of Fe²⁺ (P = 0.017; Fig. 1E) were significantly increased after MI. Moreover, MI induced a significant elevation in the expression of ACSL4 (P = 0.0067) and TFRC (P = 0.0021), molecular markers of ferroptosis (Fig. 1F), implicating that cardiomyocyte ferroptosis occurred during MI. Next, to explore if the MI altered macrophage polarization in myocardial tissue, we examined the change of polarization marker expression of M1 (NOS2) and M2 (IL-10). As described in Fig. 1G, relative to sham group, NOS2 expression was up-regulated post-MI (P = 0.0014), while IL-10 expression was the opposite (P = 0.0012), suggesting that MI favor the polarization of M1 macrophages in infarcted area. Taken together, these results suggested that cardiomyocyte ferroptosis accompanied by M1 macrophage infiltration during MI.

Ferroptotic cardiomyocytes derived exosome identification

A study revealed that ferroptotic pancreatic cancer cells carry KRAS protein to macrophages, resulting in the M2 polarization of macrophages (Dai et al., 2020). Our above results proved that cardiomyocyte ferroptosis accompanied by M1 macrophage infiltration in MI, thus, we interrogate that ferroptotic cardiomyocytes derived exosome could crosstalk macrophage to induce M1 polarization. We constructed MI models of HL-1 cells induced by hypoxia, and treated with Fer-1 to mitigate the ferroptosis process. The results showed that Fer-1 treatment significantly alleviated the oxidative stress in hypoxia-induced HL-1 cells, which was manifested by the decrease of MDA content (P = 0.0081) and concentration of Fe²⁺ (P = 0.0022) (Figs. 2A and 2B). Compared with MI model of HL-1 cells, the protein expression of ACSL4 (P = 0.010) and TFRC (P = 0.027) also were significantly suppressed by Fer-1 treatment (Figs. 2C and 2D). These results suggested that MI cardiomyocyte undergo ferroptosis and this process can be hindered by Fer-1. Exosomes derived from MI model of ferroptotic HL-1 cells (MI-Exo) and from Fer-1-treated MI model of HL-1 cells (Fer-1-Exo) were isolated and observed by western blot. The expression of exosome marker CD63 was specific enriched in exosome groups relative to cell groups (Fig. 2E). According to TEM, the size of the harvested particles both of MI-Exo and Fer-1-Exo were around 50 nm to 200 nm with an elliptic sphere and central depression (Fig. 2F). NTA revealed that the size distribution of MI-Exo and Fer-1-Exo was 140.3 ± 66.7 nm and 140.3 ± 68.6 nm, respectively (Fig. 2G). These harvested particles have typical exosome morphology and size indicating that we have successfully isolated MI-Exo and Fer-1-Exo.

Ferroptotic cardiomyocytes derived exosome promote macrophage M1 polarization

To investigate the effect of ferroptotic cardiomyocytes derived exosome on macrophage M1 polarization, MI-Exo and Fer-1-Exo were used to incubate RAW 264.7 cells. The internalization results of MI-Exo and Fer-1-Exo by RAW 264.7 cells were displayed in Fig. 3A. DiI-labeled exosomes (red fluorescence) were uptaken by RAW 264.7 cells and showed co-localization with DAPI-labeled nuclei (blue fluorescence) in the merge images. The uptake of MI-Exo and Fer-Exo by macrophages was not statistically different (Fig. 3A). The results of western blot presented that compared with blank group, NOS2 expression (P = 0.013) was significantly increased in MI-Exo group and IL-10 expression (P = 0.0199) was on the opposite (Fig. 3B). This expression trend was consistent with the results of animal experiments. Furthermore, compared with MI-Exo group, Fer-1 treatment decreased NOS2 expression (P = 0.015) but elevated IL-10 expression (P = 0.035) in hypoxia-induced HL-1 cells. Arginase-1 antibody-labeled flow cytometry detected the proportion of M2 phenotype macrophages in each treatment group. Flow cytometry results also supported that MI-Exo caused a decrease (P = 0.046) in proportion of M2 phenotype macrophage and conversely, Fer-1 treatment promoted (P = 0.014) the proportion of M2 phenotype differentiation (Fig. 3C). Therefore, these results suggest that ferroptotic cardiomyocyte-derived exosomes promote macrophage M1 phenotype differentiation during MI and that ferroptosis inhibitor treatment favors macrophage differentiation to M2 phenotype.

Ferroptotic cardiomyocytes derived exosome repaired phagocytosis of RAW 264.7 cells

Macrophage polarization leads to functional alterations in their behavior. Next, we explored whether MI-Exo induced macrophage M1 polarization could change functional phagocytosis. Phagocytosis of zymosan particles by macrophage existed in all three group (Fig. 4A), and then phagocytic cells and phagocytic index were photometrically quantified based on the fluorescently labeled zymosan uptake. As shown in Fig. 4B, there was no difference in the number of phagocytic macrophages between blank group and MI-Exo group (P = 0.24), but significantly reduced in Fer-1-Exo group (P = 0.045). However, phagocytic index of macrophage was significantly down-regulated after co-cultured with MI-Exo (P < 0.0001), and then revised by Fer-1 treatment (P = 0.0002; Fig. 4C), suggesting that MI-Exo repaired phagocytosis of macrophage but this repairment can be partially eliminated by ferroptosis inhibitor.

It has been reported that the Wnt signaling pathway is responsible for macrophage polarization (Abaricia et al., 2020). Therefore, we detected classical molecules of the Wnt signaling pathway, Wnt1 and β-catenin, to clarify the activation of RAW 264.7 cells polarization signals during MI. As expected, compared with blank group, the expression of Wnt1 (P = 0.0007) and β-catenin (P = 0.001) were up-regulated in MI-Exo group, suggesting that Wnt signaling pathway was activated and involved in macrophage M1 polarization during MI progression, whereas this activation of Wnt signaling pathway was suppressed by Fer-1 treatment (Wnt1: P = 0.0054; β-catenin: P = 0.0048; Fig. 4D). To further confirm above results, the specific Wnt signaling pathway inhibitor IWR-1 was applied. As shown in Fig. 4E, MI-Exo-induced expression of M1 polarization marker NOS2 was significantly inhibited by IWR-1 (MI-Exo vs MI-Exo+ IWR-1, P = 0.0003), and the synergy of IWR-1 and Fer-1 could further reduce the expression of NOS2 (Fer-1-Exo vs Fer-1-Exo+IWR-1, P = 0.016). The effect of IWR-1 on the M2 polarization marker IL-10 was completely opposite (Fig. 4E). These results indicated that the Wnt signaling pathway is involved in ferroptotic cardiomyocytes derived exosome induced macrophage polarization during MI.

Ferroptotic cardiomyocytes derived exosome promote macrophage M1 polarization via miR-106b-3p

We further explored which signal moleculars mainly carried by ferroptotic cardiomyocytes derived exosome to regulate Wnt signaling pathway and ultimately affect macrophage polarization. We first analyzed the mRNAs that can be encapsulated in exosomes through the ExoCarta database (http://exocarta.org/download) and found that there are only 5 exosomal mRNAs (Actb, Pdcd6ip, C3, RAB14, and F5) in common between humans and mice, but these mRNAs were not associated with the Wnt signaling pathway. Existing studies point towards exosomes may affect macrophage polarization by delivering miRNAs (Ma et al., 2022), these reminded us to pay attention to the change in miRNA profile in MI-Exo. As shown in Fig. 5A, there were 54 overlaps between human exosome and mouse exosome. In addition, the results of miRWalk database (http://mirwalk.umm.uni-heidelberg.de/) analysis showed that 148 miRNAs that can targeted regulate Wnt1 are shared by human and mice (Fig. 5B). Next, a total of 21 potential miRNAs were obtained from the intersection of the aforementioned two overlapping results (Fig. 5C). The presence of these 21 miRNAs in MI-Exo and Fer-1-Exo was verified thereafter using RT-qPCR. A heatmap displayed the expression results of 21 miRNAs and revealed that only 7 miRNAs were significantly up-regulated in Fer-1-Exo compared with MI-Exo, of which miR-106b-3p had the highest fold up-regulation (P = 0.00092; Fig. 5D). Thus, we focused on miR-106b-3p. We subsequently examined the miR-106b-3p expression on macrophages after incubation with MI-Exo or Fer-1-Exo. As expected, Fer-1-Exo incubated-macrophages had higher miR-106b-3p expression than MI-Exo group (P < 0.0001; Fig. 5E). We thereafter determined whether the miR-106b-3p carried by MI-Exo regulates macrophage polarization by Wnt1. The overexpression efficiency of miR-106b-3p mimics was good both in HL-1 cells (P = 0.00031; Fig. 5F) and HL-1 cells-derived exosome (P = 0.00023; Fig. 5G). Then, the exosomes overexpressing miR-106b-3p were co-cultured with macrophages. The results showed that Exo-miR-106b-3p mimics significantly suppressed the Wnt1 expression at mRNA (P = 0.024; Fig. 5H) and protein level (P = 0.00014; Fig. 5I), suggesting that miR-106b-3p targeted regulate Wnt1 in macrophages. Collectively, in MI model, ferroptosis leads to a decrease in the content of exosomal miR-106b-3p released by cardiomyocytes, resulting in the activation of Wnt1, which ultimately promotes M1 polarization in macrophages, whereas Fer-1 can block ferroptosis and restore the content of exosomal miR-106b-3p in cardiomyocyte thereby Fer-1 has a therapeutic effect.