This study was approved by the Ethics Committee of the Central China Branch of the National Center for Cardiovascular Diseases (approval number: FZX-IACUC-2024005). All animal manipulations were performed in accordance with procedures reviewed and approved by the Laboratory Animal Care Committee. Every effort was made to minimize animal suffering.

AC16 cells were cultured to reach 70%–80% confluence and treated with NanoAM at various concentrations (0, 3.125, 6.25, 12.5, 25, 75, 100, 150, 180, 200, and 300 μg/mL) for 24 h, 10 µL of CCK-8 (APExBIO, K1018) was added to each well for an additional hour. Absorbance was measured at 450 nm using a microplate reader (Full wavelength microplate reader, thermo).

Flow cytometry was utilized to evaluate the reactive oxygen species (ROS) levels in AC16 cells. The cells were divided into three groups: control group, H₂O₂-treated group, and H₂O₂+NanoAM-treated group. After pretreatment with NanoAM (15 mg/mL) for 6 h, H₂O₂ (300 nM) was added to induce for 24 h, we counted the fluorescence intensity of FITC.

AC16 cells were pretreated with 15 mg/kg NanoAM for 6 h and then induced with H₂O₂ (300 nM) for 24 h.10 μM DCFH-DA working solution (Beyotime, S0035S) was added and incubated at 37 °C for 30 min. Intracellular ROS levels were assessed by observing green fluorescence using a fluorescence microscope (Nikon, SAP18, excitation wavelength 485 nm, emission wavelength 530 nm)

AC16 cells were cultured and treated with NanoAM or H₂O₂ as described before. 10 μM JC-1 working solution (Biotime, C2006) diluted in serum-free DMEM) was added for 15 min. Green fluorescence (JC-1 monomers, indicating decreased mitochondrial membrane potential) and red fluorescence (JC-1 aggregates, indicating normal mitochondrial membrane potential) were observed using a fluorescence microscope (excitation wavelength 485 nm, emission wavelength 530 nm; excitation wavelength 585 nm, emission wavelength 590 nm). Changes in mitochondrial membrane potential were assessed by calculating the ratio of greeen to red fluorescence.

Six-weeks-old C57BL/6N mice were purchased from GemPharmatech (Jiangsu, China) and randomly divided into 3 groups: NC, HF(HFpEF), and HFi(HFpEF + NanoAM), (n = 6 per group), Mice in HF group was fed a high-fat diet (ResearchDiets, D12492, 60 kcal% fat) and 0.5 g/L L-NAME (N(ω)-nitro-L-arginine methyl ester hydrochloride) in drinking water for 12 weeks, while mice in NC group were fed with normal diet (Schiattarella et al., 2019). The mice in HFi group were given NanoAM (15 mg/kg) by tail vein injection twice a week based on the dietary strategy of the HF group. Then the mice underwent ultrasound examination (Fujifilm VisualSonics), blood pressure measurement (KENT Scientific, CODA4) and pressure-volume detection (AD Instruments, PL2604).

The targeting efficiency of NanoAM was evaluated using a small animal in vivo imaging system (IVIS Spectrum, United States). Briefly, C57BL/6N mice with an average weight of about 30 g were divided into a Cy5.5-NanoAM experimental group, a Cy5.5-NanoM non-targeted control group, and a solvent group. The dose was 15 mg/kg injected into the tail vein and the mice were tested 6 h later.

Mouse blood pressure was measured using a high-precision tail artery sphygmomanometer (KENT Scientific, CODA4). During the experiment, mice were placed on a temperature-controlled heating plate, covered with a black blanket to minimize stress caused by light stimulation, and then gently placed in a mouse restrainer to prevent injury. The sphygmomanometer, precisely attached to the mouse’s tail, monitored blood pressure changes in real time and calculated key parameters such as systolic, diastolic, and mean arterial pressure.

When performing cardiac ultrasound examination of mice, it was used to induce and maintain mice. The mice were anesthetized with isoflurane (2% concentration) and fixed on a heating platform in a supine position to maintain the body temperature, and the chest hair was removed with depilatory cream to facilitate the contact of the ultrasound probe. Echocardiography was performed by vevo3100 (Fujifilm VisualSonics, Canada). M-mode ultrasound and Doppler ultrasound modes were used to evaluate left ventricular systolic and diastolic function, including ejection fraction (LVEF), Early diastolic transmitral flow velocity/Atrial contraction-induced transmitral flow velocity (E/A) ratio, early mitral inflow velocity/Early diastolic mitral annular velocity (E/E′) ratio, and Isovolumic Relaxation Time (IVRT). All data were analyzed using Vevo Lab software (Fujifilm VisualSonics, Canada).

Mice were anesthetized with intraperitoneal injection of tribromoethyl aicolaol and then fixed on the operating table (AD Instruments,PL2604). The trachea was exposed and connected to a ventilator to maintain breathing. The left carotid artery was then isolated and a Millar catheter was inserted into the left ventricle. Different fluids were injected and the left ventricular pressure-volume curve was recorded. Finally, parameters such as left ventricular pressure and heart rate were analyzed to assess cardiac function.

According to the standard protocol, the cardiac tissue paraffin sections were stained with the Sirius Red dye (Servicebio, G1018). Subsequently, the sections were observed using a confocal microscope (Nikon) to record the distribution and morphology of the collagen fibers.

According to the standard protocol, the cardiac sections were stained with dihydroethidium (DHE, Biotime, S0063) and DAPI (1 μg/mL) and then observed and photographed using a confocal microscope.

The experiment began after C57BL/6N mice were fasted for 12 h (water was not allowed). Blood was collected from the tail vein to measure basal blood glucose (0 min). Mice were then intraperitoneally injected with 20% glucose solution (2 g/kg). Blood was collected from the tail vein 15, 30, 60, 90, and 120 min after injection. Blood glucose levels were measured using blood glucose test strips (ACCU-CHEK, 06454011). The area under the curve (AUC) was calculated to assess glucose tolerance in the mice.

To maintain the consistency of the fasting condition in mice. C57BL/6N mice were fasted for 12 h (water was not allowed) and then given a small amount of chow. The experiment began after another 6 h of fasting. Blood was collected from the tail vein to measure basal blood glucose (0 min). Immediately afterward, 0.5 U/kg insulin solution (HTBT, G202505047) was intraperitoneally injected. Blood was collected from the tail vein 15, 30, 60, and 120 min after injection. The area under the curve (AUC) was calculated to assess the mice’s insulin sensitivity.

The qPCR was performed according to the standard protocol with CFX Connect™ Fluorescence quantitative PCR detection system (BIO-RAD). The primers used as follows: Mouse SOCS3-F: ATG​GTC​ACC​CAC​AGC​AAG​TTT; Mouse SOCS3-R: TCC​AGT​AGA​ATC​CGC​TCT​CCT; Mouse SOCS3-F: TGC​AGG​AAG​AAA​CCG​TTG​GAG; Mouse SOCS3-R: CTC​GTT​TTA​GGA​CTG​GAC​ACT​TG; Mouse IL-1β-F:TGCCACCTTTTGACAGTGATG; Mouse IL-1β-R:TGATGTGCTGCTGCGAGATT; Mouse IL-6-F:TAGTCCTTCCTACCCCAATTTCC; Mouse IL-6-R:TTGGTCCTTAGCCACTCCTTC; Mouse TNF-α-F:CCCTCACACTCAGATCAT CTTCT; Mouse TNF-α-R:GCTACGACGTGGGCTACAG; Mouse GAPDH-F:TGCGACTTCAACAGCAACTC; Mouse GAPDH-R:ATGTAGGCAATGAGGTCCAC. The results were exported in excel format. Then, “internal control correction” was performed on the target gene, followed by “control correction” for the experimental group. The difference between the two steps is ^△△CT. Finally, the CT difference was linearly transformed to the expression fold by 2^(^-△△CT).

Western blotting was used to assess cardiac insulin resistance in mice using standard protocols. Antibodies against IRS1-Ser307 (Proteintech, 85238-1-RR, dilution 1:5000), P-AKT1-T308+AKT2-T309+AKT3-T305 (ABclonal, AP1266, dilution 1:800), SOCS3 (Proteintech, 66797-1-Ig, dilution 1:20,000), and GAPDH (Servicebio, GB15002-100, dilution 1:8000) were used.

After dewaxing and rehydration with gradient ethanol of the mouse heart paraffin sections (5 µM), they were treated with 0.1% Triton X-100 for 10 min and 1% BSA for 30 min for blocking. The sections were incubated with GLUT4 antibody (Proteintech, 66846-1-Ig) at 4 °C overnight; washed with PBS 3 × 5 min, incubated at room temperature in the dark with fluorescent secondary antibody for 1 h. Washed with PBS 3 × 5 min again, then added fluorescent WGA (Thermo W11261) in the dark for 30 min, sealed with DAPI and photographed under a fluorescence microscope.

The bulk RNA-seq of heart tissue was performed in accordance with established protocols (Shi et al., 2022). In brief, heart tissues were collected after perfusion with old PBS. Atrial and connective tissue were promptly excised, and the remaining ventricular tissues were rapidly frozen in liquid nitrogen and stored at −80 °C until further processing. The samples were then sent to OE Biotech Co., Ltd. (Shanghai, China) for paired-end sequencing (PE150) on an Illumian Novaseq™ 6000. After sequencing and quality control, the clean reads were aligned to the mm10 genome using HISAT2 (https://daehwankimlab.github.io/hisat2/), and quantification was performed with HTSeq-count (Anders et al., 2015). Differential gene expression analysis was executed using DEseq2, and matrix normalized expression matrix was generated using the variance stabilizing transformation (VST) function of DEseq2 (Love et al., 2014).

The Mfuzz package (version 2.64.0), which employs fuzzy c-means clustering, was utilized to perform cluster analysis on the matrix of normalized valuse derived from bulk RNA-seq data (Gao et al., 2017). This analysis aimed to categorize genes into distinct clusters. Post-cluster analysis, clusters that did not demonstrate a significant opposing trend between HF_vs._NC and HFi_vs._HF were manually excluded. The genes within the retained clusters underwent further analysis using the Mfuzz package. Ultimately, after three iterations of Mfuzz analysis, the results were obtained.

Data were analysed by SPSS (Version 26.0, IBM SPSS Statistics). Normality was tested using the Shapiro-Wilk method. For differences between two independent groups, parametric tests (independent t-test) and nonparametric tests (Mann-Whitney U test) were used based on the normal distribution of the data. When statistical analysis involved three or more groups, one-way analysis of variance (ANOVA) Tukey’s post hoc test was used for normally distributed data, and the Kruskal–Wallis test was used for non-normally distributed data. All statistical analyses were performed using SPSS. Data are showed as mean ± standard deviation, and P < 0.05 was considered statistically significant. Statistical analysis and drawing were performed using the GraphPad Prism 8 (GraphPad, San Diego, California, United States). Analysis of correlation by simple linear regression model. Significant among groups was analyzed by One-way ANOVA. Unpaired t-test was used for comparing depending on whether the samples are paired. A two-sided P value <0.05 was considered statistically significant.

Mn–ZIF nanozymes were synthesized through a coordination–self-assembly–in situ metal doping strategy (Figure 1A). Briefly, dopamine, Zn(NO₃)₂·6H₂O, and Mn (NO₃)₂·6H₂O were dissolved in deionized water to form solution A, while 2-methylimidazole was dissolved in 9 mL of deionized water to form solution B. Subsequently, solution A was rapidly poured into solution B under magnetic stirring at room temperature and stirred for 12 h to complete the coordination and self-assembly process. The resulting suspension was centrifuged and washed three times with deionized water, followed by freeze-drying to obtain brown Mn–ZIF powders.

Transmission electron microscopy (TEM) revealed that the obtained Mn–ZIF particles exhibited a uniform polygonal morphology with good monodispersity (Figure 1B). X-ray diffraction (XRD) showed characteristic peaks identical to those of ZIF-8, with no additional diffraction peaks, confirming that the crystalline topology of ZIF-8 was preserved after Mn doping (Figure 1C). Dynamic light scattering (DLS, Figures 1D,E) analysis indicated a hydrodynamic diameter of 80 ± 20 nm with a low polydispersity index (PDI <0.30) (Figures 1D,E). The zeta potential was measured as −20.11 ± 5.36 mV, suggesting excellent colloidal stability (Figure 1F). X-ray photoelectron spectroscopy (XPS) revealed the coexistence of Mn 2p3/2 (641.5 eV) and Mn 2p1/2 (643.2 eV) peaks, as well as Zn 2p3/2 (1021.5 eV) and Zn 2p1/2 (1044.6 eV) peaks, verifying the successful coordination of Mn²⁺ and Zn²⁺ ions within the ZIF framework without structural disruption (Figures 1G–I). Catalytic activity evaluation demonstrated that Mn–ZIF generated 9 mg/L dissolved oxygen within 10 min in 200 mM H₂O₂ (Figure 1J), inhibited 50% of superoxide anion radicals at a concentration of 0.0009601 μg/mL (Figure 1K), and exhibited strong ABTS radical scavenging activity across various concentrations (6.25, 12.5, 25, and 50 μg/mL) (Figure 1L). These results indicate that Mn–ZIF possesses integrated superoxide dismutase (SOD)-, catalase (CAT)-, and radical scavenging–like activities, providing a robust structural and functional basis for subsequent in vivo antioxidant applications.

To achieve the cardiac targeting ability of Mn–ZIF nanozyme, atrial natriuretic peptide (ANP) was modified on the surface of Mn–ZIF to enhance its cardiac targeting property. As shown in Supplementary Figure S1, the bare Mn-ZIF nanozyme exhibited a zeta potential of approximately −23 mV, consistent with its inherent surface properties. Upon coating with a neutral lipid layer (without ANP) to form NanoM, the surface potential shifted to about −8 mV. This significant change confirms the successful formation of a lipid shell, which effectively masks the original surface charge of the core. Following the subsequent incorporation of ANP to yield the final NanoAM (Mn–ZIF nanozyme is modified by ANP), the zeta potential showed a further, distinct shift to approximately −10 mV. The sequential and consistent changes in surface potential strongly support the successful integration of ANP into the nanoparticle structure.

To evaluate the biocompatibility of the material, CCK8 assay was conducted and the results showed that cell viability remained above 70% across all NanoAM -treated groups (0–300 μg/mL), indicating that the material has ideal safety (Figure 2A). Flow cytometry analysis revealed a rightward shift in the fluorescence peak in the H₂O₂-treated group compared to the Control, whereas the peak in the H₂O₂ + NanoAM group closely resembled that of the Control (Figures 2B,C). The DCFH-DA assay further supported these findings, with elevated fluorescence intensity in the H₂O₂ group (∼1.75 fold) that was markedly attenuated by NanoAM treatment (Figures 2D,E). The JC-1 assay showed a rise in JC-1 monomer fluorescence and a reduction in JC-1 aggregates upon H₂O₂ exposure, both of which were mitigated by NanoAM (Figures 2F,G). Collectively, these results demonstrate the potent ROS scavenging capacity of NanoAM.

To confirm that NanoAM could accumulate effectively in the heart without inducing systemic toxicity, we first assessed its hemocompatibility. Hemolysis assays revealed that the hemolysis rate remained within a safe range (<5%) at concentrations up to 200 μg/mL (Figure 3A). Subsequently, mice were administered 15 mg/kg NanoAM via tail vein injection twice a week for 12 weeks. Histological examination by H&E staining showed no obvious pathological changes in the heart, liver, spleen, lung, or kidney (Figure 3B). Furthermore, no significant differences were observed in key biochemical markers of liver and kidney function between NanoAM-treated and untreated mice (Figures 3C–F), indicating minimal adverse effects on organ function. These results demonstrate the good biocompatibility of NanoAM.

We next investigated the cardiac targeting capability of NanoAM using a small-animal in vivo imaging system. 15 mg/kg NanoAM was injected into the tail vein twice a week for 2 weeks. NanoAM showed significant accumulation in the heart, whereas control particles without NanoAM did not exhibit cardiac targeting (Figure 3G). This confirms that NanoAM possesses specific affinity for cardiac tissue.

To evaluate the therapeutic potential of NanoAM in HFpEF, 18 mice were randomly divided into three groups: normal control (NC, normal chow and water), heart failure (HF, high-fat diet [HFD; 60% kcal fat, D12492] and 0.5 g/L L-NAME in drinking water for 12 weeks), and HF with NanoAM intervention (HFi, HFD + L-NAME + intravenous NanoAM at 15 mg/kg, twice weekly) (Figure 4A). Non-invasive tail blood pressure measurement system showed that systolic blood pressure increased from 95.2 mmHg (NC group) to 115.2 mmHg in HF group (p < 0.001), and decreased to 104.6 mmHg after NanoAM treatment (p < 0.05). Diastolic blood pressure increased from 69.7 mmHg (NC group) to 88 mmHg in HF group (p < 0.01), and was reduced to 77 mmHg following NanoAM administration (p < 0.05) (Figure 4B). After 12 weeks of HFD + L-NAME feeding, mice developed a clinical HFpEF-like phenotype: echocardiography revealed no significant differences in left ventricular ejection fraction (EF) and fractional shortening (FS) among the NC, HFpEF, and HFi groups (Figures 4C–E). And the E/A ratio decreased from 1.4 (NC) to 1.15 in HF mice (p < 0.001), and recovered to 1.5 with NanoAM treatment (Figure 4F). The IVRT increased from 20 ms (NC) to 25.2 ms in the HF group (p < 0.01), and decreased to 18.7 ms after intervention (p < 0.05) (Figure 4G). The E/E′ ratio also increased from 33 (NC) to 47.2 in the HF group, and declined to 36.7 post-treatment (p < 0.01) (Figure 4H). Pressure-volume (PV) loops analysis indicated preserved systolic function but impaired diastolic function in the HF group, evident as an upward shift of the PV loop (Figure 4I). While end-systolic PV relationship (ESPVR) slope Ees did not differ among groups (Figure 4J), the end-diastolic PV relationship (EDPVR) slope β increased from 0.1 mmHg/μL (NC) to 0.7 mmHg/μL, and returned to 0.11 mmHg/μL after NanoAM administration (p < 0.001) (Figure 4K). The serum NT-proBNP level in the NanoAM group was significantly lower than that in the HF group (3383.7 pg/ml vs 4777.3 pg/ml) (Figure 4L). Sirius red staining revealed increased collagen deposition (red) in HF mice (2.6%) compared to NC (0.4%), which was ameliorated by NanoAM (0.3%) (p < 0.001) (Figures 4M,N). DHE staining showed elevated ROS level in HF heart (∼5.5 fold) compared to NC (p < 0.001), which was mitigated by NanoAM (∼1.12 fold) (Figures 4O,P). In addition, although there was no significant difference in body weight between the HFi group and the HF group, the heart weight/tibia length ratio (HW/TL) was statistically significantly decreased in the HFi group (Supplementary Figures S2A, B). In sum, NanoAM alleviates heart dysfunction and hypertension in HFpEF mice.

To further investigate the molecular mechanisms underlying the effect of NanoAM on HFpEF, we performed bulk RNA-seq on cardiac tissues from normal control (NC), HFpEF (HF), and NanoAM-treated HFpEF (HFi) groups. Principal component analysis (PCA) (Figure 5A) and sample correlation analysis (Figure 5B) demonstrated good intra-group consistency, reflecting high data quality. To identify genes potentially reversed by NanoAM treatment, we focused on genes that were differentially expressed in HF vs. NC and showed opposite expression trends in HFi vs. HF. A total of 1165 differentially expressed genes (DEGs) were identified between HF and NC, including 663 upregulated and 502 downregulated genes (Supplementary Table S1). Gene co-expression analysis was performed using the Mfuzz packages in R studio. After multiple rounds of filtering, six gene clusters with coherent expression patterns were identified (Figure 5C; Supplementary Figure S3) (Supplementary Table S2). Cluster 1, 4, 5, and 6 comprised genes upregulated in HF and downregulated in HFi, while cluster 2 and 3 contained genes downregulated in HF and upregulated in HFi. KEGG pathway analysis for each cluster revealed significant enrichment of metabolism-related pathways, such as PPAR signaling, glycerolipid metabolism, and insulin resistance in cluster 2 and 5. The PI3K-Akt signaling pathway and the inflammation-related TNF signaling pathway were markedly enriched in cluster 3 (Figure 5D) which are known to be involved in cardiac dysfunction. Integrated KEGG analysis of all co-expressed genes further confirmed significant enrichment of the insulin resistance, PI3K-Akt signaling, and TNF signaling pathways, with insulin resistance being the most prominently enriched (Figure 5E).

We further investigated which specific genes were reversed by NanoAM. A total of 130 DEGs were identified between HFi and HF, including 102 upregulated and 28 downregulated genes (Supplementary Table S3). By integrating these with the DEGs from the HF vs. NC comparison (Figures 6A,B), we identified 48 genes, of which 31 genes that were downregulated in HF vs. NC and upregulated in HFi vs. HF; and 17 genes that were upregulated in HF vs. NC and downregulated in HFi vs. HF (Figures 6C,D). Among these, only Socs3 overlapped with the insulin resistance pathway genes identified in Figure 5E (Gfpt2, Slc27a3, Socs3, Trib3, Rps6kal, Tnfrsfla) (Figure 6E). Socs3 was upregulated in HF compared to NC and downregulated after NanoAM treatment.

Socs3, a key suppressor of cytokine signaling, is induced by inflammatory cytokines such as IL-6 and IL-10 (Gao et al., 2018) and inhibits insulin signaling by targeting IRS1 (Rui et al., 2002). Western blot analysis revealed that NanoAM downregulated SOCS3 protein expression, reduced phosphorylation of IRS1 at Ser307, enhanced AKT2 phosphorylation, and promoted downstream insulin signaling (Figures 7A,B). We further evaluated Socs3 expressions among NC, HF and HFi groups using qPCR. The qPCR results further showed that Socs3 mRNA levels were significantly increased in HFpEF hearts (10.7-fold vs. NC, p < 0.001) and decreased after NanoAM treatment (2.3-fold) (Figure 7C). Immunofluorescence staining of heart sections showed that NanoAM promoted GLUT4 translocation from the cytoplasm to the membrane, indicating improved glucose uptake (Figures 7D,E). Then we evaluated inflammatory factors (IL-6, TNF-α, IL-1β, CRP) expression among NC, HF and HFi groups using qPCR. Results showed that these target genes’ mRNA levels were significantly increased in HFpEF hearts and decreased after NanoAM treatment (Figures 7F–I). Meanwhile, serum IL-6 and high-sensitivity C-reactive protein (hs-CRP) levels were significantly elevated in HFpEF mice and effectively alleviated after NanoAM treatment (Figures 7J,K).