The processed spatial multi-omic atlas data (sma) “All-snRNA-Spatial multi-omic map of human Myocardial infarction” was downloaded from the cellxgene database (https://cellxgene.cziscience.com/collections/8191c283-0816-424b-9b61-c3e1d6258a77), and was analyzed by the Seurat (v4.3.0) R package. Cardiomyocyte data was extracted and normalized, followed by principal component analysis (PCA) reduction, batch effect correction with the harmony package, and clustering using Seurat’s FindNeighbors and FindClusters function. The Unified Manifold Approximation and Projection (UMAP) was created via Seurat’s RunUMAP function. The weighted correlation network analysis (WGCNA) was taken by the hdwgcna R package (Morabito et al., 2021; Morabito et al., 2023). Pseudotime trajectory analysis was taken by the Monocle2 R package (Trapnell et al., 2014; Qiu et al., 2017a; Qiu et al., 2017b). The cell-cell communication was analyzed by the cellchat R package (Jin et al., 2021/02).

The ferroptosis-related genes in KEGG hsa04216 (https://www.kegg.jp/dbget-bin/www_bget?hsa04216) and wikipathways WP4313 (https://www.wikipathways.org/pathways/WP4313.html) were combined as a single geneset. We then calculated the z score (Amrute et al., 2023) of this ferroptosis pathway geneset across cardiac regions and cell types in the sma datasets. The heterogeneity distribution of the ferroptosis pathway geneset was observed.

Potential transcription factors of ATF3 were predicted by the PROMO database (http://alggen.lsi.upc.es/cgi-bin/promo_v3/promo/promoinit.cgi?dirDB=TF_8.3) with species restricted to humans. The promoter region of ATF3 was defined as the 2000 upstream bases and 100 downstream bases of the ATF3 gene sequence (hg38_knownGene_ENST00000341491.9, range = chr1:212606761-212620875) and acquired from the UCSC database (https://genome.ucsc.edu/) (Table 1).

The R language code used for single-cell data analysis and the ferroptosis pathway geneset were available in GitHub repositories (https://github.com/Xiao851213/SCFA_Ferroptosis_new/blob/main/20240329).

The human AC16 cardiomyocytes (cat. #C1360, WHELAB, China) and murine HL1 cardiomyocytes (cat. #C2173, WHELAB, China) were cultured in a humidified incubator (5% CO2, 37°C) with the Dulbecco’s modified Eagle’s (DMEM)/F12 1:1 medium (cat. #CB003, Shanghai Epizyme Biomedical Technology Co., Ltd, Shanghai, China) supplemented with 12.5% fetal bovine serum (FBS, cat. #S711-001S, the Lonsera) and penicillin/streptomycin (100 U/mL, 100 U/mL, cat. #CB010, Shanghai Epizyme Biomedical Technology Co., Ltd., Shanghai, China). Cells at 70%–80% confluence were used for subsequent experiments.

The hypoxia-reoxygenation (H/R) model was induced using the AnaeroPack™ (anaerobic cultivation set) with an airtight container (a 2.5 L rectangle jar, Mitsubishi gas chemical, Japanese) (Wen et al., 2021). In detail, cells cultured for 1 day were washed twice with phosphate‐buffered saline (PBS), cultured in sugar and serum‐free DMEM, and then placed into a sealed airtight container that contains an AnaeroPack, the oxygen concentration decreased to <0.1% within 1 hour, and the carbon dioxide concentration was maintained at about 5%. Hypoxia was continued for 10 h and terminated by removing the culture bottle from the airtight container and replacing it with a standard culture medium in a CO₂ incubator at 37°C for 6 h.

To assess the impact of SCFAs on the viability of cardiomyocytes in vitro, AC16 cells were seeded in 96‐well plates for 24 h with DMEM/F12 containing 12.5% FBS. The cells were then treated with either sodium acetate (NaAc, cat. #S116319, Aladdin, Shang, China), sodium butyrate (NaBu, cat. #S102954, Aladdin, Shanghai, China), sodium propionate (NaPr, cat. #S100121, Aladdin, Shanghai, China), or a SCFA mixture (NaAc: NaPr: NaBu ≈ 30:2:1) for 24 h.

The detectable physiological levels of SCFA are in the range of (acetate 0–410 μM; propionate 0–18.3 µM; butyrate 0–81 μM, including blood, cerebrospinal fluid (CSF), breast milk, and urine; Human Metabolome Database, http://www.hmdb.ca/) and the relative levels of the three SCFAs correspond to approximately 30:2:1 for acetate: propionate: butyrate (Yang et al., 2020). The physical and upper physiological levels of concentrations were adopted to obtain a dose-dependent curve. The concentrations of the SCFAs are presented in Table 2.

The ferroptosis model was induced by erastin (cat. #S7242, Selleck), a typical ferroptosis inducer (Yan et al., 2022). Erastin was diluted to a 10 mM working stock solution with dimethylsulfoxide (DMSO). AC16 cells were seeded in 96‐well plates for 24 h with DMEM/F12 containing 12.5% FBS, followed by exposure to 10 µM erastin for 24 h (Wu et al., 2023).

A Liperfluo probe (cat. #L248, Dojindo Molecular Technologies, Inc.) was used to evaluate cellular lipid peroxidation. Cells after the indicated treatments were washed with serum‐free DMEM and incubated with 5 μM Liperfluo for 30 min at 37°C (Nakamura et al., 2023). The intracellular Fe²⁺ was detected by the FerroOrange probe (cat. #F374, Dojindo Molecular Technologies, Inc.). Cells were incubated with 1 μM FerroOrange for 30 min at 37°C (Tian et al., 2021). Stained cells were observed using confocal scanning microscopy. The fluorescence of each group was evaluated using ImageJ software.

Cell viability was assessed using the cell counting kit-8 (CCK-8, cat. #CX001S, Shanghai Epizyme Biomedical Technology Co., Ltd, Shanghai, China). Briefly, cells were incubated with fresh medium (containing 10% CCK-8 reagent) for 2 h. The optical density at 450 nm (OD450) was determined by a microplate reader (BioTek, United States) and normalized to blank wells (cell-free medium with CCK-8 reagent).

Total RNA was acquired using a TRIzol reagent (cat. #R0016, Beyotime, Shanghai, China) and an RNA extraction kit (cat. #A2010A0402, BioTNT, Shanghai, China). The concentration was analyzed with a Nanodrop 8000 spectrophotometer (Thermo Fischer Scientific), with concentration at 50–120 ng/μL and A260/A280 of 1.8–2.1 for all samples. RNA was converted into cDNA using a reverse transcription kit (Wuhan servicebio Technology CO., LTD, Wuhan, China). Then, RT‐PCR was performed using SYBR Green qPCR Master Mix (Wuhan servicebio Technology CO., Ltd., Wuhan, China); expression was detected using a fast real‐time PCR system (CFX Connect, Bio-rad, CA, United States). Cycle counts for mRNA quantification were normalized to GAPDH. Relative expression (ΔCt) and quantification (RQ = 2^–ΔΔC) for each mRNA were calculated using the ΔΔCt method. All reactions were performed according to the manufacturer’s instructions. All primers were verified for producing a single specific PCR product via melting curve analysis. The primers used in the study are presented in Table 3.

Total cellular proteins were extracted using RIPA lysis buffer (cat. #FD008, HANGZHOU FUDE BIOLOGICAL TECHNOLOGY CO. LTD., China), ultrasonic lysis machine (cat. #VCX130, Sonics & Materials, INC. United States), and metal bath (cat. #HB120-S, DragonLab DWB, China). Proteins were separated via electrophoresis on a 4%–20% SDS gel (cat. #36250ES10, YEASEN, China) and transferred to PVDF membranes (cat. #IPVH00010, Millipore, Germany). After blocking with 5% bovine serum albumin (BSA, cat. #V908933, Merk, Germany) for 1 h, The PVDF membranes were incubated with primary antibodies, including anti-GAPDH (1:5000, cat. #A19056, ABclonal, China) (Bian et al., 2024), anti-ATF3 (1:1000, cat. #A13469, Abclonal, China) (Li et al., 2023; Liu et al., 2023), and anti-GPX4 (1:1000, cat. #CL488-67763, PTG, China) (Wang L. et al., 2022) antibodies at 4°C for >10 h. Subsequently, the membranes were incubated with the HRP-conjugated Goat anti-Rabbit/Mouse IgG (H + L) (cat. #AS014 & AS003, Abclonal, China) for 1 hour at room temperature. The protein bands were visualized with a Fdbio-Dural ECL Chemiluminescence Kit (cat. #FD8020, HANGZHOU FUDE BIOLOGICAL TECHNOLOGY CO. LTD., China) and imaged.

Our research delves into the crucial topic of the ferroptosis pathway genes and the pivotal role of ferroptosis regulator ATF3 in myocardial infarction. To create a comprehensive ferroptosis pathway geneset (FPG), we combined the genes in KEGG hsa04216 and wikipathways WP431. We then calculated the z score of this ferroptosis pathway geneset across cardiac regions and cell types in the published spatial multi-omic atlas dataset (the sma dataset) (Kuppe et al., 2022). This dataset provides an integrative high-resolution map of human cardiac remodeling after myocardial infarction using single-nucleus RNA sequencing (snRNA-seq), single-nucleus chromatin accessibility, and spatial transcriptomic profiling method. The dataset includes 31 samples from 23 individuals, including four non-transplanted donor hearts as controls (CTRL), and samples from tissues with necrotic areas (ischaemic zone (IZ) and border zone (BZ)) and the unaffected left ventricular myocardium (remote zone (RZ)) of patients with acute myocardial infarction. Nine human heart specimens at later stages after myocardial infarction that exhibited ischaemic heart disease were defined as fibrotic zone (FZ) samples. The snRNA-seq part of the sma dataset (191,795 cells included) was extracted for analysis in this study. Figure 1A illustrates the whole cells, identified cell types, and region sources of the sma snRNA-seq dataset. Potential regulators of ATF3 were predicted via the PROMO database (Supplementary Figure S1A) and combined as a geneset of potential regulators for ATF3 (GPRA).

In the sma snRNA-seq dataset, BZ samples have the largest proportion of cells with upregulated FPG (Figure 1B). When divided by cell types, FPG was enriched in adipocytes, myeloid, cardiomyocytes, and mast cells (Figure 1C). The FZ tissue has the most significant proportion of ATF3⁺ cells (Figure 1D). Elevated expression of ATF3 was also observed in IZ cells (Figure 1E). The relative enrichment of FPG, ATF3, and GPRA for different cell types differed among region groups (Supplementary Figure S2). In the BZ, there was the same elevation trend of FPG, ATF3, and GPRA in cardiomyocytes, adipocytes, and cycling cells [a cluster with enriched cell-cycle marker gene MKI67 and showed a high score of cell-cycle G2/M and S phases (Kuppe et al., 2022)]. This indicates the involvement of ATF3 and its potential transcription factors in the ferroptosis of cardiomyocytes, adipocytes, and cycling cells during post-MI cardiac remodeling.

In cardiomyocytes, FPG was upregulated in injury-related cells (BZ and FZ), while ATF3 and GPRA were mostly enriched in FZ (Figure 1F). The elevation of ATF3 was also found in IZ cardiomyocytes (Figure 1G). To identify the cardiomyocyte subpopulation that is critical to ferroptosis, we extracted cardiomyocyte data from the sma snRNA-seq dataset, corrected the batch effect (Supplementary Figure S3A, B), and re-clustered according to cell density on the Uniform manifold approximation and projection (UMAP) plot to recognize the subclusters of each region (Supplementary Figure S3C). Cardiomyocytes were clustered into 14 subpopulations, which are CM0-CM13 (Figure 1H; Supplementary Figure S4). The FPG was enriched in the FZ cluster CM09 (top marker genes: ABRA, DDIT3, and OTUD1) and the BZ cluster CM02 (top marker genes: UBASH3B, C4orf54, NRXN3). The co-enrichment of FPG, ATF3, and GPRA were also observed in cardiomyocytes’ CM09 and CM02 clusters (Figure 1I; Supplementary Figure S4B). In conclusion, ATF3 may involved but partially regulates ferroptosis pathway genes in myocardial infarction.

We then adopted human (AC16 cell line) and murine (HL1 cell line) cardiomyocytes to study the effect of SCFAs on ferroptosis and ATF3 expression at physiological conditions, 1 h-hypoxia exposure, hypoxia-reoxygenation (H/R) model, and erastin-induced ferroptosis model. A dose-dependent toxic experiment was run to detect the suitable dose for SCFA treatment. The concentration of acetate, propionate, butyrate, and SCFA mixture was divided into seven levels (Table 2). The 24-h treatment of SCFAs with the C6 and C7 concentrations decreased cell viability (Supplementary Figure S5). In the 1-h hypoxia model, SCFAs (sodium propionate (NaPr), sodium butyrate (NaBu), and mixture) significantly and consistently decrease cardiomyocyte viability (Supplementary Figure S5B). The C5 level concentration (the maximum dose that does not reduce cell viability) was adopted for the following experiments.

SCFA treatment can promote the mRNA expression level of ATF3, either in mice (normal and 1h-hypoxia model, Figure 2A) or human cardiomyocyte cell line (normal, 1h-hypoxia exposure), H/R exposure (10-h hypoxia plus 6-h re-oxygen), and erastin-induced ferroptosis model, Figure 2B). The potential promoters of ATF3 (VDR, EGR3, PAX5, and SP1) were also affected by 24-h SCFA exposure. In the murine cardiomyocyte cell line, NaPr and NaBu upregulated VDR under both normal and hypoxic conditions, while hypoxia attenuated the VDR upregulation by NaPr. EGR3 was upregulated by either a single or a mixture of SCFA. The effect of sodium acetate (NaAc) and the SCFA mixture on EGR3 was attenuated by 1-h hypoxia exposure (Figure 2A). In the human cardiomyocyte cell line, VDR and SP1 were upregulated by SCFA mixture in 1-h hypoxia exposure and H/R model (Figure 2B, upper & middle panel). SCFA mixture promotes the expression of PAX5 in the H/R model but not the 1-h hypoxia model (Figure 2B, upper & middle panel). In the erastin-induced ferroptosis model, VDR and EGR3 were increased under SCFA mixture stimulation (Figure 2B, lower panel).

We also detected the effect of SCFA on the ATF3 protein level. The results were totally opposite to those of the mRNA level. SCFA decreased the ATF3 protein in nearly all cases except the 1 h-hypoxia model of the human cardiomyocyte AC16 cell line (Figure 2C). The protein expression of anti-ferroptosis markers, glutathione peroxidase 4 (GPX4), was inhibited by the SCFA mixture except in the 1h-hypoxia model of the mouse cardiomyocyte HL1 cell line.

SCFA treatment can increase the cell viability of AC16 cardiomyocytes in H/R exposure (p < 0.0001) but not the erastin-induced ferroptosis model (p = 0.9984) (Figure 2D). To confirm the occurrence of ferroptosis, we performed Liperfluo staining and FerroOrange staining in the H/R model and erastin-induced ferroptosis model with or without SCFA rescue (Figures 2E, F). Liperfluo staining showed obvious lipid peroxidation in response to H/R or erastin stimulation. This effect in the erastin stimulation model was rescued by SCFA treatment. In the H/R model, a decrease in lipid peroxides was observed in the SCFA-managed group but without statistical significance. The fluorescence intensity of FerroOrange, a Fe²⁺-specific probe, increased sharply upon erastin stimulation. SCFA treatment can reverse erastin-induced Fe²⁺ increase but may strengthen the Fe²⁺ in the H/R model.