FGF9 alleviates diabetic cardiomyopathy by activating Nrf2 via SQSTM1/p62-Keap1 in mice

Abstract

Fibroblast growth factor 9 (FGF9) plays a key role in development and cardioprotection, yet its function in diabetic cardiomyopathy (DCM) remains unclear. In a high-fat diet/streptozotocin (HFD/STZ)-induced DCM model, FGF9 attenuated cardiac hypertrophy, fibrosis, and systolic dysfunction, effects abolished in cardiomyocyte-specific Nrf2 knockout mice. Mechanistically, FGF9 restored AMPK activity and promoted autophagy, enhancing p62-mediated degradation of Keap1 and nuclear translocation of Nrf2. In neonatal rat cardiomyocytes (NRCMs), FGF9 reversed high glucose and palmitate (HG + PA)-induced suppression of AMPK phosphorylation, autophagic flux, and Nrf2 signaling. AAV9-mediated expression of wild-type AMPK (AMPK^WT) or a dominant-negative AMPK mutant (AMPK^T172A) confirmed that AMPK activation was essential for FGF9-induced Nrf2 activation. Functionally, FGF9 reduced lipid accumulation, preserved mitochondrial integrity, and alleviated oxidative stress. FGF9 ameliorates DCM via non-canonical autophagy-dependent activation of Nrf2, mediated by AMPK. These findings position FGF9 as a potential therapeutic target for diabetic myocardial injury.

FGF9 alleviates cardiac hypertrophy, fibrosis, and dysfunction in diabetic models by activating Nrf2 through AMPK-dependent, autophagy-mediated degradation of Keap1

Introduction

The global prevalence of diabetes mellitus, particularly type 2 diabetes mellitus (T2D), has been rising at an alarming rate over the past few decades^1,2. By 2045, the number of individuals affected by diabetes worldwide is projected to reach approximately 693 million. Cardiovascular disease (CVD) remains the leading cause of morbidity and mortality among diabetic patients, contributing to the development of heart failure in up to 80% of these individuals². As a result, diabetic cardiomyopathy (DCM), characterized by alterations in myocardial structure and function³, has become a well-established entity in various heart failure guidelines^4,5. However, the precise mechanisms driving DCM and the subsequent cardiovascular complications in diabetes remain inadequately understood.

Nuclear factor erythroid 2-related factor 2 (Nrf2) is a key transcription factor involved in cellular defense mechanisms, regulating the expression of several detoxifying and antioxidant genes. Under basal conditions, Nrf2 is maintained at low levels due to its ubiquitination and proteasomal degradation by the E3 ligase complex comprising Kelch-like ECH-associated protein 1 (Keap1) and Cullin 3 (Cul3)⁶. Upon oxidative stress, the Keap1-Cul3 complex dissociates, leading to Nrf2 stabilization and its translocation to the nucleus, where it activates the antioxidant response element (ARE) to induce the expression of antioxidant and detoxifying genes^7,8. In addition to this canonical pathway, p62 also interacts with the Keap1-Nrf2 complex, inhibiting Nrf2 degradation and promoting its nuclear accumulation, thereby enhancing the expression of genes involved in antioxidant defense and inflammation^9–11. Recent evidence highlights the protective role of Nrf2 in DCM, where it mitigates oxidative stress and inflammation-induced damage in the myocardium¹². Nrf2 activation has been shown to counteract high glucose-induced apoptosis and cellular damage in various tissues, including the heart, kidney, and vasculature¹³. However, impairments in autophagy, common in diabetic conditions, can inhibit Nrf2-mediated redox regulation and promote noncanonical activation of Nrf2 in cardiomyocytes¹⁴. Therefore, the Nrf2 signaling pathway represents a potential therapeutic target for addressing cardiovascular complications associated with obesity and diabetes.

Fibroblast growth factors (FGFs) comprise a large family of signaling molecules that mediate a wide range of biological and pathophysiological processes, including angiogenesis, wound healing, embryogenesis, and metabolic regulation^15,16. FGF9, a member of this family, selectively binds to fibroblast growth factor receptor 3 (FGFR3) and plays a pivotal role in the development of organs^17,18 such as the pancreas¹⁹, lungs^20–22, and osteogenesis and chondrogenesis²³. FGF9 has also been implicated in adipogenesis and systemic energy metabolism through its regulation of uncoupling protein 1 (UCP1)²⁴. While several studies have explored the role of FGF9 in non-cardiac conditions, such as in embryonic development and metabolic disorders, limited research has focused on its potential role in cardiovascular diseases. Notably, FGF9 has been shown to regulate myocardial differentiation and to reduce mortality following myocardial infarction in adult mice²⁵. Furthermore, FGF9 expression has been shown to decrease in diabetic kidney disease²⁶ and is associated with atrial fibrillation-related stroke²⁷. In our own analysis of the dataset, we found that FGF9 was significantly decreased in the myocardium in db/db mice. Additionally, GO analysis demonstrated that overexpressed FGF9 in db/db mice ameliorate oxidative stress response, tissue remodeling signaling pathways and positive regulation of protein metabolic process, accompanying by the increasing in response to fibroblast growth factor receptors. Despite these observations, the role of FGF9 in diabetic cardiomyopathy has not been explored.

In this study, we investigated the cardioprotective effects of FGF9 in diabetic cardiomyocytes. We found that FGF9 reduced lipid droplet accumulation, preserved mitochondrial integrity, and attenuated oxidative stress in diabetic hearts, thereby improving myocardial remodeling and cardiac systolic function. Additionally, we demonstrated that FGF9 promoted AMPK activity and activated a noncanonical Nrf2 signaling pathway through autophagy, ultimately contributing to the improvement of DCM. These findings provide a mechanistic framework for the observed clinical benefits of FGF9 and suggest that it may serve as both a therapeutic target and a functional biomarker in the chronic phase of diabetic cardiomyopathy.

Results

FGF9 is downregulated in the diabetic heart

HFD/STZ-treated and db/db mice are well-established rodent models for type 2 diabetes and its associated complications^28,29. To investigate the relationship between fibroblast growth factor (FGF) expression and diabetic cardiomyopathy (DCM), we conducted RNA sequencing analysis in db/db mice (Supplementary Fig. 1). This analysis revealed that the expression of FGF5, FGF7, FGF9, FGF11, FGF16, FGF18, and FGF21 was reduced in the myocardium of db/db mice (Fig. 1A). To further explore the role of FGF9, we delivered cardiotropic recombinant adeno-associated virus serotype 9 expressing FGF9 (rAAV9-cTnT-FGF9) into db/db mice. This approach not only significantly and specifically overexpressed FGF9 in cardiomyocytes but also led to slightly increased levels of FGF1, FGF2, FGF18, and FGF21 (Fig. 1A). Gene Ontology (GO) analysis indicated that FGF9 overexpression in db/db mice influenced biological processes associated with DCM, including lipid metabolic processes, long-chain fatty acid metabolism, and cardiac ventricle development (Fig. 1B). Additionally, Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed that FGF9 overexpression regulated lipid metabolism and cardiovascular diseases in db/db mice (Supplementary Fig. 1D). While the roles of other FGFs-such as FGF1³⁰, FGF5³¹, FGF16³², FGF18³³ and FGF21³⁴-in myocardial injury have been explored, the role of FGF9 in DCM remains poorly understood.

Fig. 1. Identification of the candidate FGFs-related genes in DCM.

A The corresponding FGFs genes correlated were visualized by the heatmap in db/db mice. The darker shade of red or blue represents the higher correlation level. B Gene ontology (GO) analysis of the influence of DCM genes in db/db mice with or without FGF9 overexpression. C Real-time quantitative PCR assays were performed to determine the mRNA levels of FGF9 in mice, n = 5. D Real-time quantitative PCR assays were performed to determine the mRNA levels of FGF9 in NRCMs, n = 5. E Heart lysate was analysed by western blotting with FGF9. n = 5. Quantification of FGF9 protein levels (right panel). F NRCMs lysate was analysed by western blotting with FGF9. n = 5. Quantification of FGF9 protein levels (right panel). All significance was assessed by two-tailed Students’ t test. Data are shown as means ± SD.

Notably, FGF9 gene expression is highly abundant in the lateral regions of the left and right ventricles, as well as near the apical interventricular groove, during embryonic development (from embryonic day 10.5 [E10.5] to E12.5). However, this expression decreases gradually after E14.5, with minimal expression observed in the ventricular muscle of the adult heart³⁵. Given these developmental changes, we further examined FGF9 expression in both db/db mice and NRCMs. Consistent with our clustering analysis (Fig. 1A), FGF9 expression was significantly downregulated in the myocardium of db/db mice (Fig. 1C) and in neonatal rat cardiomyocytes (NRCMs) exposed to high glucose (HG) and palmitic acid (PA) (Fig. 1D). Correspondingly, FGF9 protein levels were also markedly reduced in both db/db mice (Fig. 1E) and cultured NRCMs under diabetic conditions (Fig. 1F). Collectively, these findings suggest that FGF9 may play a critical role in the regulation of DCM.

FGF9 attenuates HG + PA induced ROS production and dysfunction in NRCMs

To investigate the potential therapeutic effects of FGF9 in DCM, we first assessed whether FGF9 could mitigate the cellular dysfunction induced by HG + PA treatment in NRCMs. Notably, FGF9 treatment significantly reduced HG + PA-induced cell hypertrophy, as evidenced by the reduction in cell size (Supplementary Fig. 2A, B) and the downregulation of hypertrophic and fibrosis-related marker genes, including atrial natriuretic peptide (ANP), brain natriuretic peptide (BNP), collagen I, and collagen III (Supplementary Fig. 2C). GO analysis further indicated that oxidative stress was a key pathogenic mechanism in DCM, with a marked reduction in the response to fibroblast growth factor receptors (Fig. 1B). We next explored the impact of FGF9 on oxidative stress in NRCMs under HG + PA conditions. HG + PA treatment resulted in a rapid generation of reactive oxygen species (ROS), as demonstrated by increased dihydroethidium (DHE) fluorescence intensity (Supplementary Fig. 2D, E) and the accumulation of hydrogen peroxide (Supplementary Fig. 2F). FGF9 treatment significantly attenuated apoptosis, as indicated by a decrease in TUNEL-positive cells (Supplementary Fig. 2G, H). Furthermore, we assessed the effect of FGF9 on inflammatory responses in NRCMs. HG + PA exposure led to an upregulation of pro-inflammatory cytokines, including IL-1β, IL-6, IL-8, and TNF-α, all of which were significantly inhibited by FGF9 treatment (Supplementary Fig. 2I). Given the central role of Nrf2 in regulating oxidative stress in diabetic complications and its protective effects in DCM¹³, we evaluated the expression of Nrf2. FGF9 treatment significantly restored Nrf2 protein levels in NRCMs exposed to HG + PA (Supplementary Fig. 2J, K). Notably, the expression of the Nrf2 target gene, heme oxygenase 1 (HO1), was markedly reduced by HG + PA treatment but was significantly upregulated following FGF9 treatment (Supplementary Fig. 2J, L). Moreover, FGF9 ameliorated HG + PA-induced collagen III overexpression, further supporting its role in mitigating myocardial fibrosis (Supplementary Fig. 2J, M). These results collectively indicate that FGF9 exerts a protective effect against DCM by modulating oxidative stress, inflammatory pathways, and fibrotic remodeling, highlighting its potential as a redox-sensitive therapeutic agent.

FGF9 ameliorates cardiac remodeling and dysfunction in the diabetic heart

To assess the role of FGF9 in diabetic cardiomyopathy, we specifically overexpressed FGF9 in cardiomyocytes by delivering cardiotropic recombinant adeno-associated virus serotype 9 (rAAV9) expressing FGF9 (rAAV9-cTnT-FGF9) into 6-week-old db/db mice via tail vein injection. Control animals were injected with rAAV9-null. Mice were monitored for 8 weeks and underwent echocardiographic assessment at 14 weeks of age (Supplementary Fig. 3A). Fasting blood glucose levels (Supplementary Table 1) and body weight (Supplementary Fig. 3B, C) were similar between the experimental and control groups, confirming that FGF9 treatment did not affect these parameters. Cardiac function was evaluated by echocardiography, revealing significant left ventricular (LV) systolic dysfunction in db/db mice. FGF9 overexpression improved LV systolic function, as evidenced by increased LV ejection fraction (LVEF) and LV fractional shortening (LVFS) (Fig. 2A–C; Supplementary Table 5). To further investigate the impact of FGF9 on cardiomyocyte function, we performed electron microscopy to examine cardiac ultrastructure in rAAV9-FGF9-treated and control db/db mice. Although diabetic hearts showed a slight reduction in mitochondrial density compared to nondiabetic controls, the difference was not statistically significant following FGF9 treatment, as shown by mitochondrial quantification (Supplementary Fig. 3D), indicating that FGF9 did not significantly affect mitochondrial biogenesis or turnover. However, significant alterations were observed in the size and morphology of individual mitochondria. Both the average cross-sectional area and the perimeter of mitochondria were significantly reduced in the diabetic heart, but these changes were reversed upon FGF9 overexpression (Fig. 2D–F). Furthermore, FGF9 treatment reduced the number of lipid droplets in the myocardium of db/db mice (Fig. 2G) and alleviated oxidative stress, as evidenced by decreased hydrogen peroxide accumulation (Fig. 2H) in cardiomyocytes exposed to high glucose and lipid conditions. These results collectively suggest that FGF9 improves both systolic and diastolic function in diabetic hearts.

Fig. 2. FGF9 ameliorates cardiac remodeling and dysfunction in the diabetic heart.

FGF9 overexpression vector (AAV9-cTnT-FGF9) and control vector (AAV9-LacZ) were injected intravenously into tail veins of 6 weeks old male db/db and db/dm mice, respectively, noninvasive echocardiographic assessments were performed to monitor cardiac function after the 8 weeks injection. A Echocardiographic measurement of left ventricular ejection fraction (B) and left fractional ventricular shortening (C). D Representative electron microscopic (EM) images of heart sections. Scale bar, 2 µm and then zoom in 3 times. Quantification of average mitochondria cross-sectional area (E), perimeter (F), lipid droplets (D, yellow arrow) in cardiomyocytes and quantification of the number of lipid droplets per 100 μm² of hearts (G). More than 10 mitochondria in 5 mice for each group were measured. H Hydrogen peroxide levels (by Amplex Red assay) quantified in different groups; n = 5. I, J Representative images of H&E-stained (n = 5. Scale bar = 0.8 mm) and WGA-stained sections (scale bar = 25 μm for lower WGA staining), and quantification of cardiomyocyte cross-sectional area based on WGA staining (K). L Picrosirius red (PSR) staining in the interstitial (top, Scale bar = 30 μm) and peripheral area (bottom, Scale bar = 30 μm) and quantification of the LV collagen volume (M). n = 5. N Real-time quantitative PCR assays were performed to determine the mRNA levels of ANP, BNP, collagen I and collagen Ⅲ. n = 5. All significance was assessed by two-way ANOVA. Data are shown as means ± SD.

To evaluate the effects of FGF9 on cardiac structural remodeling, we examined myocardial hypertrophy in db/db mice. FGF9 overexpression attenuated myocardial hypertrophy, as demonstrated by a reduction in heart size (Fig. 2I). Furthermore, wheat germ agglutinin (WGA) staining revealed that the cross-sectional area of cardiomyocytes from FGF9-overexpressing db/db mice was significantly smaller than that of control db/db mice (Fig. 2J, K). FGF9 treatment also mitigated LV collagen deposition, as indicated by reduced collagen volume in FGF9-treated hearts (Fig. 2L, M). This effect was corroborated by real-time PCR, which showed a significant reduction in the mRNA levels of hypertrophic and fibrotic markers, including BNP, ANP, collagen I, and collagen III (Fig. 2N). At the molecular level, FGF9 treatment significantly restored the expression of Nrf2 protein in db/db hearts and upregulated the Nrf2 target gene, heme oxygenase 1 (HO1), which was downregulated in untreated db/db mice (Supplementary Fig. 3E–G). Additionally, FGF9 treatment reduced collagen III accumulation (Supplementary Fig. 3E, H) and hydroxyproline content (Supplementary Fig. 3I) in hearts exposed to high glucose and lipid conditions. Together, these data demonstrate that FGF9 exerts beneficial effects on cardiac remodeling and dysfunction in the diabetic heart, likely through the modulation of oxidative stress, fibrosis, and hypertrophy, and by restoring Nrf2-mediated protective signaling pathways.

Cardiomyocyte-specific Nrf2 deficiency hindered the protective effect of FGF9 on DCM

To further elucidate the role of Nrf2 in FGF9-mediated protection against diabetic cardiomyopathy, neonatal rat cardiomyocytes (NRCMs) were transduced with Nrf2-targeting shRNA or a scrambled control sequence under high glucose (HG) and palmitic acid (PA) treatment conditions (Supplementary Fig. 4H, I). Depletion of Nrf2 significantly attenuated the protective effects of FGF9, as evidenced by increased cardiomyocyte size (Supplementary Fig. 4A, B), heightened oxidative stress (Supplementary Fig. 4C–E), and enhanced apoptosis (Supplementary Fig. 4F, G). Notably, si-Nrf2 treatment abolished FGF9-induced upregulation of antioxidant response elements, reflected in decreased levels of heme oxygenase 1 (HO-1) at both mRNA and protein levels (Supplementary Fig. 4H, J, K). Furthermore, Nrf2 knockdown counteracted FGF9-mediated inhibition of collagen III expression (Supplementary Fig. 4H, L) and enhanced the expression of ANP and BNP at the mRNA level (Supplementary Fig. 4M). These findings underscore the critical role of Nrf2 in mediating the protective effects of FGF9 in diabetic cardiomyopathy.

To further investigate the functional significance of Nrf2 in cardiomyocytes, we generated cardiomyocyte-specific inducible Nrf2 knockout mice (Myh6-CreEsr1, Nrf2^f/f; Nrf2-CKO) (Supplementary Fig. 5A). Cre recombination and Nrf2^f/f allele excision were confirmed by PCR (Supplementary Fig. 5B), and immunoblot analysis demonstrated a 60% reduction in Nrf2 protein expression in whole cardiac tissue (Supplementary Fig. 5C, D). In vivo studies revealed no significant phenotypic differences between wild-type (WT; Nrf2^f/f), heterozygous (Myh6-CreEsr1; Nrf2^f/+), and homozygous (Myh6-CreEsr1; Nrf2^f/f) mice (Supplementary Fig. 5E). Then, we specifically overexpressed FGF9 in cardiomyocytes by delivering cardiotropic rAAV9 expressing FGF9 (rAAV9-cTnT-FGF9) into T2DM mice established by HFD and STZ. Control animals were injected with rAAV9-null. Mice were monitored for 8 weeks and underwent echocardiographic assessment at 18 weeks of age (Supplementary Fig. 5F, G, L, M). Fasting blood glucose levels were similar between the experimental and control groups before or after inducible T2D (Supplementary Table 2). As shown in Fig. 3, FGF9 treatment significantly reduced cardiomyocyte hypertrophy (Fig. 3A, B), myocardial fibrosis (Fig. 3C, D), and cardiac systolic and diastolic dysfunction (Fig. 3E–G; Supplementary Table 6) in db/db mice. However, these protective effects were significantly diminished in cardiomyocyte-specific Nrf2-deficient mice. In parallel, Nrf2 depletion reversed the protective effects of FGF9 on mitochondrial integrity (Fig. 3H–J) and lipid accumulation (Fig. 3K), with a marked attenuation of FGF9’s ability to reduce hydrogen peroxide levels in Nrf2-CKO mice (Fig. 3L). At the molecular level, the absence of Nrf2 in the cardiomyocytes abolished FGF9’s regulation of Nrf2 and HO-1 protein (Supplementary Fig. 5G–I) and promoted the expression of collagen III (Supplementary Fig. 5G, J). Additionally, Nrf2-CKO did not block the promotion of FGF9 on autophagy (Supplementary Fig. 5G, K). Moreover, Nrf2-CKO mice failed to effectively suppress the mRNA levels of hypertrophic and fibrotic marker genes following FGF9 treatment (Fig. 3M). These results highlight the pivotal role of Nrf2 in mediating the protective effects of FGF9 against diabetic cardiomyopathy.

Fig. 3. Cardiomyocyte-specific Nrf2 deficiency hindered the protective effect of FGF9 on DCM.

Nrf2-WT and Nrf2-CKO mice with the T2DM background were established by feeding HFD and by intraperitoneal STZ (Supplementary Fig. 5F). A Representative images of H&E-stained (n = 5. Scale bar = 0.8 mm) and WGA-stained sections (scale bar = 25 μm for lower WGA staining), and quantification of cardiomyocyte cross-sectional area based on WGA staining (B). C Picrosirius red (PSR) staining in the interstitial (top, Scale bar = 30 μm) and peripheral area (bottom, Scale bar = 30 μm) and quantification of the LV collagen volume (D). E Echocardiographic measurement of left ventricular ejection fraction (F) and left fractional ventricular shortening (G). H Representative electron microscopic (EM) images of heart sections. Scale bar, 2 µm and then zoom in 3 times. Quantification of average mitochondria cross-sectional area (I), perimeter (J), lipid droplets (K, yellow arrow) in cardiomyocytes and quantification of the number of lipid droplets per 100 μm² of hearts (K). More than 10 mitochondria in 5 mice for each group were measured. L Hydrogen peroxide levels (by Amplex Red assay) quantified in different groups; n = 5. M Real-time quantitative PCR assays were performed to determine the mRNA levels of ANP, BNP, collagen I and collagen Ⅲ. n = 5. All significance was assessed by two-way ANOVA. Data are shown as means ± SD.

FGF9 activates Nrf2 via the p62-Keap1 pathway in NRCMs

FGF9 has been shown to induce left ventricular concentric hypertrophy with microvascular dilatation, promote angiogenesis, and alleviate fibrosis following myocardial infarction without inducing pathological hypertrophy or interstitial fibrosis²⁵. Diabetes, however, induces autophagy deficiency, which disrupts Nrf2-mediated defense mechanisms while activating a pathological Nrf2-driven program that promotes ferroptosis in cardiomyocytes, exacerbating diabetic cardiomyopathy¹⁴. In the present study, we hypothesized that FGF9 might influence Nrf2 activity via the p62/Keap1/Nrf2 signaling pathway, as it has been previously reported that activation of this pathway liberates Nrf2 from Keap1, enabling its nuclear translocation³⁶. In agreement with this, we found that FGF9 treatment restored Nrf2 protein expression in NRCMs exposed to HG + PA. Furthermore, treatment with the proteasome inhibitor MG132 enhanced the FGF9-induced Nrf2 protein accumulation (Fig. 4A, C). Notably, FGF9 treatment reduced Keap1 protein levels under HG + PA conditions, but co-treatment with MG132 did not further affect Keap1 protein levels, suggesting that FGF9 modulates Keap1 stability through a non-proteasomal degradation mechanism (Fig. 4A, D). Co-immunoprecipitation analysis revealed that HG + PA treatment promoted the interaction between Nrf2 and Keap1 (Fig. 4A, B), leading to Nrf2 ubiquitination and subsequent degradation (Supplementary Fig. 6A–C). FGF9 treatment disrupted this interaction, thereby preventing Nrf2 degradation. However, FGF9 did not significantly increase Nrf2 binding to Keap1 in MG132-treated cells, further supporting the idea that FGF9 regulates Keap1 degradation through a proteasome-independent mechanism (Fig. 4A–D).

Fig. 4. FGF9 activates Nrf2 via the p62-Keap1 pathway in NRCMs.

NRCMs were cultured either in MAN (5.5 mM) and HG + PA (33 mM HG + 200 μM PA) medium in the presence or absence of FGF9 (20 ng/mL). A–D NRCMs were cultured with MG132 (5 μM) for 48 h. The cell lysates were subjected to immunoprecipitation with the Keap1 antibody, followed by immunoblotting with the indicated antibodies. The quantitative analysis of immunoprecipitation (B) and immunoblots (C, D). E–H, NRCMs were cultured with Baf-A1 (20 nM) for 48 h. The cell lysates were subjected to immunoprecipitation with the Keap1 antibody, followed by immunoblotting with the indicated antibodies. The quantitative analysis of immunoprecipitation (F, G) and immunoblots (H). I–L NRCMs were cultured with Compound C (10 μM) for 48 h. The level of relative protein was evaluated by Western blot. The quantitative analysis of immunoblots (J–L). All significance was assessed by one-way ANOVA. Data shown in the graphs represent the means ± SD of independent experiments.

Given that Nrf2 accumulation is associated with autophagic degradation^9,36–38, we next explored whether FGF9 influences Nrf2 expression via autophagy regulation. It is well established that Keap1 interacts with p62, an autophagy adapter, to release Nrf2 for nuclear translocation³⁸. We found that HG + PA treatment increased the interaction between Keap1 and p62 in NRCMs, and this interaction returned to baseline levels following FGF9 treatment (Fig. 4E, F). Interestingly, co-treatment with the autophagy inhibitor Baf-A1 enhanced the binding of Keap1 to Nrf2, even in the presence of FGF9 (Fig. 4E, G), suggesting that Baf-A1 inhibits FGF9-induced autophagic degradation of Keap1 and p62 (Fig. 4E, H). Additionally, immunofluorescence analysis further demonstrated that Keap1 undergoes basal autophagic degradation mediated by LC3 under physiological conditions. However, exposure to high glucose and high-lipid environments markedly suppressed autophagy, accompanied by a reduction in the LC3-Keap1 interaction. Notably, treatment with FGF9 restored autophagic flux and reestablished LC3-mediated Keap1 degradation. Upon co-treatment with Baf-A1, both Keap1 and LC3 accumulated in the cytoplasm, indicating a blockade of autophagosome-lysosome fusion (Supplementary Fig. 6G). Consistently, under metabolic stress, FGF9 enhanced the autophagic degradation of Keap1 through increased P62-Keap1 interaction, whereas Baf-A1 promoted the formation of P62-Keap1 complexes without facilitating their degradation (Supplementary Fig. 6H). There results demonstrated that the accumulation of Keap1 protein increased the binding probability of Keap1 to Nrf2 (Fig. 4E, G, Supplementary Fig. 6I), thereby promoting Nrf2 ubiquitination and degradation (Supplementary Fig. 6D–F).

In view of the fact that FGF9 is involved in the regulation of brown adipose formation and regulates the whole body energy metabolism²⁴, we further investigate the involvement of autophagy in FGF9-mediated Nrf2 activation, and hypothesized that FGF9 may regulate autophagy through the energy-sensing kinase AMPK. Consistent with this, we found that treatment with Compound C, a selective AMPK inhibitor, blocked the effects of FGF9 on AMPK phosphorylation, autophagy induction, and nuclear translocation of Nrf2 (Fig. 4I–L). These data suggest that FGF9 ameliorates HG + PA-induced cardiomyocyte dysfunction by promoting Keap1 degradation, potentially through an AMPK/p62-dependent autophagy pathway that facilitates Nrf2 nuclear translocation.

FGF9 activates Nrf2 via the p62-Keap1 pathway in db/db mice

Consistent with our hypothesis, myocardial overexpression of FGF9 restored Nrf2 protein levels in db/db mice, an effect further enhanced by proteasome inhibition with MG132 (Supplementary Fig. 7A, C). Notably, Keap1 protein levels were significantly reduced following FGF9 overexpression, whereas MG132 co-treatment did not produce additional changes in Keap1 abundance, suggesting that FGF9 promotes Keap1 degradation through a proteasome-independent mechanism (Supplementary Fig. 7A, D). Co-immunoprecipitation analysis revealed enhanced Keap1-Nrf2 interaction in db/db hearts (Supplementary Fig. 7A, B), leading to increased Nrf2 ubiquitination and degradation. Given the established role of autophagy in Keap1 turnover, we next investigated whether FGF9 modulates myocardial autophagic flux. FGF9 overexpression reversed the diabetes-induced increase in Keap1-p62 interaction and restored p62 homeostasis in db/db hearts (Supplementary Fig. 7E, F). Inhibition of autophagy with Baf-A1 markedly enhanced Keap1-Nrf2 binding and prevented Keap1 degradation, even in the presence of FGF9 (Supplementary Fig. 7E, G), indicating that autophagy is required for FGF9-mediated Keap1 clearance (Supplementary Fig. 7E, H). These findings suggest that pathological accumulation of Keap1 in diabetic myocardium promotes excessive sequestration and ubiquitination of Nrf2, thereby impairing its transcriptional activity. To further elucidate the upstream regulatory mechanism, we assessed the role of AMPK in FGF9-mediated autophagic activation. Pharmacological inhibition of AMPK with Compound C abrogated FGF9-induced AMPK phosphorylation (Supplementary Fig. 7J), suppressed autophagy activation (Supplementary Fig. 7J, K), and inhibited Nrf2 nuclear accumulation in db/db hearts (Supplementary Fig. 7J, L). Collectively, these results indicate that FGF9 restores Nrf2 signaling in diabetic myocardium by promoting Keap1 degradation via an AMPK/p62-dependent autophagy pathway.

FGF9 promotes AMPK activity and attenuates ROS production and dysfunction in NRCMs

AMPKα2 activity is significantly downregulated in STZ-induced diabetic models compared to AMPKα1, and reduced AMPK activity is central to the development of diabetic cardiomyopathy, contributing to impaired cardiac autophagy³⁹. To further investigate the mechanisms by which FGF9 modulates Nrf2 signaling and provides cardiac protection, NRCMs were transfected with plasmids encoding LacZ (control), wild-type AMPKα2 (AMPKα2^WT), or a dominant-negative mutant AMPKα2 (AMPKα2^T172A)⁴⁰. As expected, FGF9 treatment alleviated HG + PA-induced cardiomyocyte enlargement (Fig. 5A, B), reactive oxygen species (ROS) production (Fig. 5C-E), and the proportion of TUNEL-positive cells (Fig. 5F, G) in AMPKα2^WT-overexpressing NRCMs. However, these effects were largely abolished in AMPKα2^T172A-overexpressing NRCMs. Consistent with previous findings, the AMPKα2^T172A mutation abolished FGF9-mediated phosphorylation of AMPK (Fig. 5H, I). Furthermore, AMPKα2^T172A mutation resulted in a decrease in autophagic activity (Fig. 5H, J, K), leading to an accumulation of Keap1 (Fig. 5H, L), which binds Nrf2 for ubiquitination and degradation, thereby reducing Nrf2 nuclear translocation (Fig. 5H, M) and the protein levels of the downstream antioxidant HO-1 (Fig. 5H, N). In contrast, these effects were not observed in AMPKα2^WT-expressing NRCMs (Fig. 5H–N). These findings indicate that the AMPKα2^T172A mutation abrogates the effects of FGF9 on Nrf2 signaling, suggesting that FGF9 activates Nrf2 signaling through a noncanonical pathway via AMPKα activation to promote autophagy.

Fig. 5. FGF9 promotes AMPK activity and attenuates ROS production and dysfunction in NRCMs.

NRCMs were infected with a plasmid encoding LacZ (control), wild-type AMPKα2 (AMPKα2^WT), or dominant negative mutation (AMPKα2^T172A). After transfection, NRCMs were cultured with HG + PA (33 mM HG + 200 μM PA) medium in the presence or absence of FGF9 (20 ng/mL). A Representative images of NRCMs (blue: DAPI; red: cTnT; scale bar = 20 μm). Quantification of cell surface area as shown in right (B), n = 5. C Fluorescent images and quantitation of superoxide levels (right, D). Scale bars = 150 μm. n = 5. E Hydrogen peroxide levels (by Amplex Red assay) quantified in different groups, n = 5. F The apoptotic cells (green, TUNEL⁺ cells, blue: DAPI). Scale bars = 60 μm. The quantitative analysis of TUNEL⁺ cells (right, G). n = 5. H NRCMs lysate was analysed by western blotting with indicated antibodies. n = 5. Quantification of relative protein levels (I–N). All significance was assessed by one-way ANOVA. Data shown in the graphs represent the means ± SD of independent experiments.

FGF9 promotes Nrf2 expression through AMPK-induced autophagy in DCM

To further explore the role of cardiac AMPK in the pathogenesis of diabetic cardiomyopathy (DCM), we administered adeno-associated virus serotype 9 (AAV9) vectors carrying LacZ (control), wild-type AMPKα2 (AMPKα2^WT), or AMPKα2^T172A mutant under the control of the murine cardiac troponin T (cTnT) core promoter (AAV9-cTNT-AMPKα^WT; AAV9-cTNT-AMPKα2^T172A) to 6-week-old db/db mice via tail vein injection (Supplementary Fig. 8A). No significant differences in fasting blood glucose levels or body weight were observed between db/db mice with or without FGF9 treatment (Supplementary Table 3 and Supplementary Fig. 8B). Echocardiographic analysis revealed that AMPKα^WT overexpression significantly improved systolic dysfunction in db/db mice upon FGF9 treatment, whereas AMPKα2^T172A mutation prevented the beneficial effects of FGF9 on systolic function (Fig. 6A–C and Supplementary Table 7). Mitochondrial dynamics, including mitochondrial mass, size, and lipid accumulation, were also assessed. Transmission electron microscopy demonstrated that FGF9 treatment in AMPKα^WT-overexpressing db/db mice increased mitochondrial area (Fig. 6D–F), reduced lipid deposition (Fig. 6D, G), and decreased the number of damaged mitochondria (Fig. 6D–F). However, these protective effects were abolished in AMPKα2^T172A mutant db/db mice (Fig. 6D–G). Additionally, FGF9 treatment in AMPKα2^T172A mutant db/db mice failed to improve oxidative stress (Supplementary Fig. 8C). Histological analysis showed that FGF9 ameliorated left ventricular structural remodeling in diabetic mice, as indicated by a reduction in cardiomyocyte cross-sectional area and left ventricular collagen volume. However, these improvements were significantly impaired in AMPKα2^T172A mutant db/db mice (Fig. 6H–M). Immunoblot analysis further confirmed that AMPKα2^T172A mutation led to decreased AMPK phosphorylation (Supplementary Fig. 8D, E), reduced autophagic activity (Supplementary Fig. 8D, F, G), and accumulation of Keap1 (Supplementary Fig. 8D–H), which bound Nrf2 for ubiquitination and degradation. Consequently, nuclear localization of Nrf2 (Supplementary Fig. 8D, I) and the expression of its downstream target HO-1 (Supplementary Fig. 8D, J) were significantly reduced. In contrast, these effects were not observed in AMPKα^WT-overexpressing mice. The AMPKα2^T172A mutation thus abolished FGF9-induced Nrf2 signaling, further supporting the notion that FGF9 activates Nrf2 via AMPKα-mediated autophagy. These results suggest that FGF9 alleviates diabetic heart failure, cardiac remodeling, lipid overload, and mitochondrial dysfunction in a heart-specific, AMPK-dependent manner. By activating AMPK and promoting autophagy, FGF9 facilitates the noncanonical activation of Nrf2 signaling, which plays a critical role in the protection of cardiomyocytes from oxidative stress and metabolic damage. These findings provide novel insights into the molecular mechanisms underlying the cardioprotective effects of FGF9 in diabetic cardiomyopathy.

Fig. 6. FGF9 promotes Nrf2 expression through AMPK-induced autophagy in DCM.

FGF9 overexpression vector (AAV9-cTnT-FGF9) and control vector (AAV9-LacZ), Prkaa2 (AAV9-cTnT-AMPKα2^WT) and mutant Prkaa2 (AAV9-cTnT-AMPKα2^T172A) were injected intravenously into tail veins of 6 weeks old male db/db and db/dm mice, respectively. Noninvasive echocardiographic assessments were performed to monitor cardiac function after the 8 weeks injection. A Echocardiographic measurement of left ventricular ejection fraction (B) and left fractional ventricular shortening (C). D Representative electron microscopic (EM) images of heart sections. Scale bar, 2 µm and then zoom in 3 times. Quantification of average mitochondria cross-sectional area (E), perimeter (F), lipid droplets (D, yellow arrow) in cardiomyocytes and quantification of the number of lipid droplets per 100 μm² of hearts (G). More than 10 mitochondria in 5 mice for each group were measured. H, I Representative images of H&E-stained (n = 5. Scale bar = 0.8 mm) and WGA-stained sections (scale bar = 25 μm for lower WGA staining), and quantification of cardiomyocyte cross-sectional area based on WGA staining (J). K Picrosirius red (PSR) staining in the interstitial (top, Scale bar = 30 μm) and peripheral area (bottom, Scale bar = 30 μm) and quantification of the LV collagen volume (L). M Left ventricular collagen quantification by hydroxyproline assay (μg/mg). n = 5. All significance was assessed by one-way ANOVA. Data are shown as means ± SD.

Discussion

Epidemiological evidence strongly associates heart failure with type 2 diabetes (T2D), underscoring the overlap between these two conditions^41,42. In 1972, Rubler introduced the concept of diabetic cardiomyopathy (DCM) to distinguish it from other forms of cardiomyopathy⁴³, with DCM affecting approximately 12% of patients with diabetes and contributing significantly to heart failure and mortality⁴⁴. Over the past decades, the identification of molecular targets for DCM treatment has been an area of intensive research^45,46. In this study, we used a combination of high glucose (HG) and palmitic acid (PA) to replicate the in vivo conditions of hyperglycemia and hyperlipidemia in T2D, as previously reported in various in vitro models^47,48.

Hyperglycemia-induced cardiomyocyte death triggers a cascade of events leading to inflammation, myocardial hypertrophy, and fibrosis—a process collectively known as cardiac remodeling. Numerous therapeutic strategies targeting cardiac remodeling, oxidative stress, inflammation, and metabolic dysfunction have shown promise in preclinical studies and clinical trials⁴⁹. However, these molecular targeting approaches have yet to yield successful clinical outcomes, and their translation into practical therapies for DCM remains limited^50,51. It is well established that cell death is an early event in the pathogenesis of DCM, preceding the structural changes that lead to ventricular dysfunction^52,53. Therefore, targeting pathways involved in cardiomyocyte death at the initial stages of diabetic cardiac dysfunction may offer a translational strategy for therapeutic intervention.

The potential role of fibroblast growth factors (FGFs) in cardiovascular diseases has gained increasing attention in recent years⁵⁴. Paracrine factors such as FGF3, FGF8, FGF9, FGF10, FGF15/19, and FGF16 are crucial for cardiac development⁵⁵. In addition, FGF2⁵⁶, FGF5³¹, FGF9⁵⁷, FGF10⁵⁸, FGF16³², and FGF21⁵⁹ have been shown to exert cardioprotective effects under pathological conditions, while FGF12⁶⁰, FGF13⁶¹, and FGF23⁶² are implicated in adaptive hypertrophic signaling, promoting cardiac hypertrophy. Our previous work demonstrated that FGF18 alleviates isoproterenol (ISO) and transverse aortic constriction (TAC)-induced hypertrophy in male mice through the FYN/NOX4 signaling axis³³. These studies highlight the pivotal role of FGFs in cardiovascular disease pathogenesis and suggest their potential as biomarkers for cardiac injury.

FGF9, a paracrine factor that selectively binds to FGFR3, has been primarily studied in the context of bone²³ and lung biology^20–22. However, its relationship with cardiovascular diseases remains poorly understood, with limited studies examining its role in myocardial infarction²⁵, diabetic kidney disease (DKD)²⁶, and atrial fibrillation-related stroke²⁷. The biological functions of FGF9 in the myocardium and its underlying mechanisms require further investigation. In this study, we observed that FGF9 expression was significantly decreased in the hearts of db/db mice. Gene ontology (GO) analysis of transcriptomic data revealed that FGF9 overexpression in db/db mice improved oxidative stress responses, tissue remodeling pathways, and protein metabolism processes, accompanied by an increase in fibroblast growth factor receptor (FGFR) signaling. Given that FGF9 is known to modulate systemic energy metabolism²⁴, we observed that, despite no differences in fasting blood glucose or body weight between control and FGF9-overexpressing db/db mice, FGF9 reduced lipid droplet accumulation in the myocardium. Furthermore, FGF9 alleviated oxidative stress and hydrogen peroxide buildup in cardiomyocytes exposed to high glucose and fatty acids. These findings suggest that FGF9 plays a critical role in maintaining myocardial homeostasis.

Autophagy is a key cellular catabolic process that responds to energy stress and is essential for maintaining myocardial homeostasis, particularly under altered metabolic conditions. Recent research has highlighted the close relationship between autophagy and oxidative stress through complex signaling pathways^9,37,38. Our study found that FGF9 alleviated HG + PA-induced injury in NRCMs by enhancing the autophagy-related p62-Keap1-Nrf2 signaling pathway, leading to increased Nrf2 activation, reduced ROS production, and decreased pro-inflammatory cytokine expression. Nrf2, a master regulator of cellular defense against oxidative stress, also modulates key metabolic pathways, including lipid, carbohydrate, and amino acid metabolism, and iron homeostasis⁶³. Interestingly, studies have shown that Nrf2-deficient mice exhibit elevated hepatic FGF21 expression and improved glucose tolerance when fed a high-fat diet^64,65. In our study, cardiomyocyte-specific inducible Nrf2 knockout (Nrf2-CKO) mice exhibited little effect on lipid and glucose metabolism in T2D. However, Nrf2 deletion abrogated the protective effects of FGF9 on ROS production, myocardial hypertrophy, and functional injury, both in vivo and in vitro. These findings emphasize that Nrf2 is a key mediator of the beneficial effects of FGF9 in ameliorating DCM.

Under normal conditions, Nrf2 maintains redox and metabolic homeostasis by regulating the expression of genes involved in iron and lipid metabolism, as well as redox balance in the heart. However, conflicting studies have suggested that cardiac-specific autophagy deficiency can activate Nrf2, leading to myocardial damage in pressure-overloaded hearts⁶⁶, and that inhibiting autophagy in type 1 diabetic cardiomyocytes promotes Nrf2-mediated ferroptosis⁶⁷. These discrepancies are likely due to differences in the context and timing of Nrf2 activation, with acute or controlled activation being protective, whereas chronic activation, particularly in an autophagy-deficient environment, can be harmful^68,69. In our study, FGF9 enhanced AMPK phosphorylation and activated the p62-Keap1-Nrf2 pathway in the presence of high glucose and fatty acids, thereby reducing oxidative stress, inflammation, and cardiac remodeling in diabetic hearts. However, the AMPKα2^T172A mutant abrogated the beneficial effects of FGF9 on autophagy, cardiac hypertrophy, and fibrosis, suggesting that restoring autophagic flux is critical for sustaining Nrf2 activation and its cardioprotective effects in DCM.

Mitochondrial dysfunction plays a central role in ROS production in diabetic hearts and is a major contributor to pathological remodeling in DCM⁷⁰. Structural abnormalities and disorganization of mitochondria are common in several diabetic models and are closely associated with impaired ATP production, reduced ejection fraction, and decreased fractional shortening³⁴. Elevated mitochondrial ROS levels have been observed in cardiac biopsies from T2D patients⁷¹. In our study, diabetic hearts exhibited significant mitochondrial dysfunction, including structural abnormalities, excessive ROS production, and lipid accumulation. These effects were reversed by cardiomyocyte-specific FGF9 overexpression in db/db mice but were not observed in AMPKα2^T172A mutant mice, further underscoring the role of FGF9 in mitochondrial function and diabetic cardiac health. These findings provide valuable insights into the pathogenesis of DCM and suggest that FGF9 may serve as a promising therapeutic strategy for maintaining mitochondrial integrity and improving cardiac function in diabetic hearts.

There are several limitations to our study. First, although we observed that FGF9 ameliorated mitochondrial dysfunction and lipid accumulation in the heart, the specific mechanisms by which FGF9 regulates mitochondrial dynamics remain unclear and warrant further investigation. Second, while we demonstrated reduced FGF9 expression in db/db and Nrf2-CKO mice with DCM, clinical data are still lacking to validate whether FGF9 could serve as a molecular marker for DCM therapy. Third, our study focused primarily on T2D-induced DCM, and further research is needed to explore the role of FGF9 in type 1 diabetes-related cardiac injury. Finally, we did not assess whether long-term overexpression of FGF9 in animal models consistently plays a role in primarily activating AMPK, because the cardiomyocyte-specific overexpression of FGF9 may have slightly affected the expression of other FGFs, which have also been reported to activate AMPK^72,73.

In conclusion, our study provides the first evidence that FGF9 reduces lipid droplet accumulation in diabetic cardiomyocytes, stabilizes mitochondrial size, and reduces oxidative stress, thereby improving myocardial remodeling and systolic function in diabetic hearts. We also demonstrate that FGF9 enhances AMPK activity and activates the autophagy-induced p62-Keap1-Nrf2 noncanonical pathway, improving diabetic cardiomyopathy (Fig. 7). These findings offer a mechanistic basis for clinical observations and suggest that FGF9 could serve as a potential therapeutic factor or functional biomarker for chronic DCM.

Fig. 7. Schematic figure illustrates that FGF9 ameliorates DCM and dysfunction by maintaining redox homeostasis by AMPK-p62-Keap1 signal axis.

FGF9 enhances AMPK activity and activates the autophagy-induced p62-Keap1-Nrf2 noncanonical pathway, improving diabetic cardiomyopathy. In conclusion, the present study identifies a previously unknown biological function of FGF9 in DCM. Created in BioRender. Yunjie, C. (2025) https://BioRender.com/xgb23yb.

Materials and methods

Animal procedures

Diabetic db/db mice and their control littermates (db/dm) were purchased from GemPharmatech Co., Ltd. Adeno-associated virus serotype 9 (AAV9)^74,75 carrying mouse fibroblast growth factor 9 (FGF9) (AAV9-ctnt-FGF9) or a control vector (AAV9-ctnt-LacZ), as well as AAV9 vectors encoding mouse Prkaa2 (AAV9-ctnt-AMPKα2^WT) or mutant Prkaa2 (AAV9-ctnt-AMPKα2^T172A), were obtained from OBiO Tech, Inc. The viruses were intravenously administered into the tail veins of 6-week-old male db/db or db/dm mice, respectively.

For experiments involving wild-type (WT, Nrf2^f/f) and Nrf2 conditional knockout (CKO) mice (Nrf2^f/f crossed with Myh6-CreEsr1), which were purchased from Cyagen Biosciences, tamoxifen (Santa Cruz Biotechnology, sc-208414) was administered at a dose of 45 mg/kg/day for 5 consecutive days via intraperitoneal (i.p.) injection. Mice were subsequently allowed a 14-day post-injection period to achieve efficient gene knockout. Genotyping of the transgenic mice was performed by polymerase chain reaction (PCR) analysis using tail DNA and the following primer sequences:

Myh6-CreEsr1

Forward: 5′-TCTATTGCACACAGCAATCCA-3′

Reverse: 5′-CCAGCATTGTGAGAACAAGG-3′

Nrf2 ^f/f

Forward: 5′-GGTCGTAATGGCTTACATCTTCAA-3′

Reverse: 5′-TGCAAGCTCCTATACGTTGTGAT-3′

Nrf2-WT and Nrf2-CKO mice (6–18 weeks old) with a Type 2 diabetes mellitus (T2DM) background were established by feeding a high-fat diet (HFD, Medicience Ltd., 21.9 kJ/g, 60% of energy derived from fat; MD12033) and by intraperitoneal injection of streptozotocin (STZ, Sigma-Aldrich, V900890). After 1 month on the HFD, mice in the HFD group were subjected to three consecutive days of overnight fasting followed by a single i.p. injection of STZ (35 mg/kg), while control mice were injected with citrate buffer. AAV9 vectors encoding mouse FGF9 (AAV9-ctnt-FGF9) or control vector (AAV9-ctnt-LacZ) were intravenously administered at the start of HFD feeding in Nrf2-WT or Nrf2-CKO mice (Supplementary Fig. 5F).

Before the commencement of the study, all mice were acclimatized to their new environment for 4–6 days. Mice were housed under standard conditions (temperature 21 ± 2 °C, humidity 50 ± 15%, with a 12-h light-dark cycle). Fasting blood glucose levels were measured at baseline and at the conclusion of the study (Supplementary Tables 1–3). All animal procedures were conducted in compliance with the institution’s ethical guidelines and were approved by the Institutional Animal Care and Use Committee (IACUC) of Ningbo University, China. We have complied with all relevant ethical regulations for animal use.

Isolation and culture of neonatal rat cardiomyocytes (NRCMs)

NRCMs were isolated from the ventricles of Sprague-Dawley rats at postnatal day 2, as previously described^76,77. Transient transfection of Nrf2-targeting shRNA (TL711582, ORIGENE) or control shRNA was performed in NRCMs using Lipofectamine RNAiMAX (Invitrogen) following the manufacturer’s protocol. NRCMs were subsequently treated with mannitol (33 mM: 5.5 mM glucose + 27.5 mM D-mannitol; M4125, Sigma-Aldrich) or high glucose plus palmitic acid (HG + PA; 33 mM glucose + 200 μM PA) for 48 h^47,48, with or without FGF9 (20 ng/mL, 7399-F9, Novus Biologicals). For the analysis of the signaling pathways, MG132 (5 μM for 48 h, S2619), bafilomycin A1 (Baf-A1, 20 nM for 48 h, S1413), or Compound C (10 μM for 48 h, S7306) were used.

Immunoblotting analysis

Proteins were separated by SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes, which were incubated overnight with the appropriate primary antibodies followed by 2 h of incubation with secondary antibodies at room temperature. Immunoreactive bands were detected using Pierce ECL Western blotting substrate (WBKLS0500, Millipore). Primary antibodies used included those against p62 (sc-48402, 1:1000), LC3 (sc-271625, 1:1000), FGF9 (sc-8413 1:1000), p-AMPK (2531S, 1: 1000), AMPK (2532S, 1:1000), Collagen III (db14567, 1:1000), Nrf2 (A1244, 1:1000), Keap1 (10503-2, 1:1000), Ubiquitin (sc-8017, 1:1000), β-actin (20536-1, 1:1000), and Lamin B (66095-1, 1:1000).

RNA isolation and quantitative real-time-PCR (qRT-PCR)

Total RNA was extracted from heart tissue and cardiomyocytes using TRIzol reagent (Takara Bio, 9108), and 1 ng of RNA was reverse-transcribed to cDNA using the HiScript® III Reverse Transcriptase kit (Vazyme, R223-01). qRT-PCR was performed using a QuantStudio™ 3 Real-Time PCR Detection System and ChamQ Universal SYBR qPCR Master Mix (Vazyme, Q711-02) with specific primers listed in Supplementary Table 4. Relative expression levels were determined by the 2^{−∆∆C.T.} method and normalized to endogenous β-actin.

Echocardiography

Echocardiographic measurements were obtained using a Vevo 1100 Ultrasound System (VisualSonics, Toronto, Canada) with a high-frequency (30 MHz) linear array transducer. Systolic and diastolic ventricular diameters, wall thickness, left ventricular fractional shortening (LVFS), and left ventricular ejection fraction (LVEF) were calculated to assess systolic function (Supplementary Tables 5–7). All measurements were performed by an experienced technician who was blinded to the group allocation.

Tissue collection and histological analysis

Hearts were harvested, fixed in 4% paraformaldehyde, and processed for paraffin embedding. Sections (5 µm) were deparaffinized and stained with hematoxylin and eosin (HE) or picrosirius red (Sigma-Aldrich). Images were acquired using a Nikon Eclipse Ni light microscope, and histopathological analysis was performed by a blinded pathologist.

Oxidative stress measurement

Myocardial superoxide generation was measured via in situ dihydroethidium (DHE, Thermo Fisher Scientific, D11347) staining. Briefly, NRCMs were incubated with DHE (5 μM) in a humidified chamber at 37 °C, followed by PBS washing. The staining (red) images were captured with the microscope (EVOS, Thermo Fisher Scientific, MA, United States). The fluorescence intensity of DHE staining was measured using the ImageJ software (version 1.8.0), and the results were expressed as fold change against the corresponding controls.

Hydrogen peroxide levels in mouse hearts and NRCMs were measured using the Amplex Red assay (Invitrogen), following the manufacturer’s protocol. Fluorescence was detected at an excitation wavelength of 530 nm and an emission wavelength of 590 nm using a Mithras LB 940 reader (Berthold Technologies). A standard curve generated with fresh H₂O₂ was used for quantification.

Immunofluorescent staining

Deparaffinized heart sections (5 µm) were stained with WGA-Alexa/488 (Thermo Fisher Scientific, W11261). Cardiomyocytes were incubated with primary antibody against cardiac troponin T (A4914, 1:200), p62 (sc-48402, 1:50), LC3 (sc-271625, 1:50), Nrf2 (A1244, 1:200), Keap1 (60027-1, 1:200) at 4 °C overnight. After incubation and wash, cells were blocked with Alexa fluor 647-conjugated anti-rabbit IgG secondary antibody (711-605-152, Jackson) and Alexa fluor 488-conjugated anti-mouse IgG secondary antibody (115-545-166, Jackson) for 1 h. After incubation and PBS washes, nuclei were counterstained with DAPI for 15 min. Images were captured with a Leica TCS SP8 confocal microscope, and the total tissue-positive areas were quantified using ImageJ software. Data were expressed as a percentage of the positive staining area relative to the total area.

Transmission electron microscopy

After the indicated treatments, tissue blocks (approximately 3 mm³) were fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer and processed for routine electron microscopy. Samples were embedded in EPON resin and examined with an HT-7800 Transmission Electron Microscope.

TUNEL staining

Cardiomyocytes were subjected to TUNEL staining using the DeadEnd™ Fluorometric TUNEL System (Promega, G3250), according to the manufacturer’s instructions. Nuclei were stained with DAPI, and images were captured using a Leica TCS SP8 confocal microscope. TUNEL-positive cardiomyocytes were counted in three randomly selected fields.

Hydroxyproline assay

Total collagen content was assessed using a hydroxyproline assay kit (Solarbio, BC0255). Snap-frozen left ventricular free walls were hydrolyzed in 6 M HCl at 120 °C for 16 h, neutralized with 6 M NaOH, and the final neutralized hydrolysate was used for hydroxyproline measurement. The hydroxyproline concentration was determined using a microplate reader and normalized to protein concentration.

Statistical analysis

All analyses were performed with the experimenter blinded to the mice and cultured cells groups. Statistical comparisons were made with a two-tailed Students’ t test for two experimental groups or ANOVA with Tukey for multiple groups. Statistical analyses were done using GraphPad Prism 10.2 (GraphPad Software). n = X biologically independent samples/animals/independent experiments.

Author contributions

Gen Chen and Ning An conceived, designed and supervised the study and analyzed the data. Gen Chen, Yunjie Chen and Ning An performed experiments. Gen Chen and Ning An wrote the manuscript. Xuan Zhou, Bingyu Wang and Lin Li revised the manuscript. All authors contributed to reviewing and editing the manuscript and approved its final version.

Peer review

Peer review information

Communications Biology thanks Ting Zhao and the other anonymous reviewers for their contribution to the peer review of this work. Primary handling editors: Jesmond Dalli and Chris Hine.

Data availability

All data supporting the findings of this study are available within the paper and its Supplementary Information. RNA sequencing data are provided in the Supplementary Data 1 and CRA031152 dataset (https://ngdc.cncb.ac.cn/gsa/). All numerical data values are provided in Supplementary Data 2. Data supporting the findings of this study are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-025-09077-6.