Ripk1-dependent PANoptosis promotes EndMT and aggravates hypoglycemia-related cardiac fibrosis in diabetes

Abstract

Background

Hypoglycemia impairs cardiac function in patients with diabetes, yet its underlying mechanisms remain elusive. Exploring the mechanism of hypoglycemia-related cardiovascular endothelial damage in diabetes from a single-cell perspective and identifying specific intervention measures will assist with preventing adverse cardiovascular events in patients with diabetes.

Methods

This retrospective study includes participants who experienced ≥one hypoglycemia episode from Grade 3A hospitals to elucidate its effect on cardiac function in DM. Single-nucleus (sn) RNA-Seq has been used to determine lineage-specific gene expression, subpopulation composition, and intercellular communication. The findings were further validated through in vivo and in vitro biological assays.

Results

Hypoglycemia significantly reduced left ventricular ejection fraction (LVEF) and left ventricular fractional shortening (LVFS) in both diabetic patients and db/db mice. snRNA-seq of cardiac tissue under hypoglycemic conditions showed significant enrichment of key genes linked to cell death, inflammation, and fibrosis within endothelial cells, macrophages, fibroblasts, and cardiomyocytes, with endothelial cells demonstrating a particularly critical regulatory role. Hypoglycemia triggered Ripk1-dependent PANoptosis in endothelial cells, which was accompanied by EndMT and cardiac fibrosis. Ripk1 knockdown inhibits the activation of the COLLAGEN signaling pathway and attenuates PANoptosis, EndMT, and fibrosis in low glucose-treated Mouse Cardiac Microvascular Endothelial Cells (MCMECs). Correspondingly, endothelial cell-specific Ripk1 conditional knockout improved cardiac function in a diabetichypoglycemia mouse model and effectively mitigated cardiac PANoptosis, EndMT, and COLLAGEN pathway activation.

Conclusions

Hypoglycemia activates the COLLAGEN signaling pathway through Ripk1-dependent PANoptosis, thereby triggering EndMT and exacerbating cardiac fibrosis. Inhibition of Ripk1 effectively alleviates hypoglycemia-associated cardiac dysfunction.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-026-07935-3.

Introduction

As the common acute complication in patients with DM, hypoglycemia exerts adverse effects on cardiovascular system, significantly increasing the risk of myocardial infarction, malignant arrhythmias, and "dead in bed" syndrome [1, 2]. While some studies have explored cardiac electrophysiological explanations, the underlying mechanisms of hypoglycemia-induced adverse cardiac outcomes remain unclear [3, 4]. As the fundamental functional unit of the heart, endothelial cells play a critical role; their dysfunction may trigger a series of major adverse cardiovascular events such as impaired vasodilation, coronary artery spasm, and atherosclerotic plaque formation [5–7]. However, the physiological alterations of endothelial cells during hypoglycemia episodes in DM and the underlying injury mechanisms remain poorly understood.

PANoptosis represents an inflammatory programmed cell death (PCD) pathway that combines key characteristics of pyroptosis, apoptosis, and necroptosis and cannot be characterized by any one of them alone [8]. PANoptosis is regulated by multimeric protein complexes called PANoptosomes. Multiple upstream sensors of PANoptosis have been identified, including receptor-interacting protein kinase 1 (Ripk1), Z-DNA binding protein 1 (Zbp1), NOD-like receptors family pyrin domain containing 12 (Nlrp12), and absent in melanoma 2 (Aim2). These sensors can recognize specific stimuli and initiate PANoptosome assembly, leading to PANoptosis activation [9–11]. Only a few studies have proposed the link between PANoptosis and heart failure or cancer therapy-induced cardiotoxicity [12]. In diabetic hypoglycemia research, studies have shown that hypoglycemia-induced pyroptosis substantially to endothelial dysfunction in diabetic animal models [13, 14]. Furthermore, serum TNF-α (Tumor Necrosis Factor-alpha) levels—a key biomarker of necroptosis—were found to be significantly elevated during hypoglycemia episodes in DM patients, exhibiting strong correlation with endothelial dysfunction [15]. However, whether PANoptosis directly influences diabetic cardiac endothelial cells during hypoglycemia remains unclear.

This study aimed to investigate the impact of hypoglycemia on cardiac tissues in DM. The cross-sectional study was systematically conducted to evaluate the adverse impacts of hypoglycemia on cardiac function in DM patients. The snRNA-seq revealed the cell lineage-specific transcriptional changes, subpopulation dynamics, and intercellular communication patterns in diabetic cardiac tissues under hypoglycemic conditions. In this study, we specifically focused on the injury mechanisms and mediating roles of endothelial cells during hypoglycemia. These findings provide potential clinical intervention targets for hypoglycemia-related complications in DM.

Results

Clinical characteristics and the association between hypoglycemia and cardiac function

This study initially enrolled 30,676 patients diagnosed with DM at The First Affiliated Hospital of Chongqing Medical University from June 3, 2022, to May 31, 2024, ultimately including 7170 patients in the analysis. Baseline characteristics showed that patients who experienced hypoglycemia were older (70.0 vs. 67.0 years) and had a higher prevalence of abnormal left ventricular fractional shortening (LVFS) or LVEF (p < 0.01). Hypoglycemia was significantly associated with abnormal LVEF (OR: 1.49, 95% CI: 1.21–1.83) and abnormal LVFS (OR: 1.59, 95% CI: 1.25–2.01). Following multivariable adjustment, hypoglycemia remained independently associated with reduced LVEF (OR: 1.57, 95% CI: 1.27–1.94) and reduced LVFS (OR: 1.68, 95% CI: 1.31–2.13). Correspondingly, patients with abnormal cardiac function, defined by a reduced LVEF, experienced a significantly higher frequency of daily hypoglycemic events than those with normal cardiac function, further indicating a close association between hypoglycemic episodes and impaired cardiac function. Further details are provided in Supplementary Fig. S1 and Tables S1–S4. Further analysis of UK Biobank data showed that during follow-up, participants with T2DM who experienced hypoglycemia faced an elevated risk of developing cardiovascular disease (CVD). Kaplan-Meier survival curves further demonstrated that CVD onset occurred earlier among those with hypoglycemic events. The Supplementary Figs. S2–S3, and Tables S5–S6 provide detailed results.

Hypoglycemia aggravated cardiac dysfunction in diabetic mice

To investigate the effect of hypoglycemia on cardiac function in diabetic mice, we established the diabetic hypoglycemia (HDM) and diabetic model (DM), the modeling workflow has been drawn in Fig. 1A. All db/db mice met the diagnostic criteria for diabetes based on their fasting blood glucose (FBG) and random blood glucose (RBG) levels, as shown in Fig. 1B. The blood glucose levels of the mice in the experimental group were recorded during each insulin treatment to ensure successful induction of hypoglycemia in Fig. 2C. The LVEF and LVFS, the two most commonly used indices to evaluate left ventricular systolic function, were measured by echocardiography. As shown in Fig. 1D, LVEF and LVFS were futher decreased in the HDM group compared to the DM group.

Fig. 1.

Hypoglycemia impaired cardiac function in db/db mice. (A) Schematic representation of modeling of diabetic mice and mice with diabetes accompanied by recurrent hypoglycemia. (B) FBG and RBG of db/db mice. (C) Blood glucose records during recurrent hypoglycemia induction. (D) Representative transthoracic M-mode echocardiograms from each group and quantification results of LVEF and LVFS in each group (n = 8, *p < 0.05)

Fig. 2.

The changes in relative proportion for each cell type of cardiac tissues from DM and HDM. (A) Unbiased clustering of nuclei from all samples identifies 7 major cell types. (B) UMAP plot showing the expression of the established marker genes for each cell type. (C) Comparison of the nucleus densities in the UMAP space between the two conditions reveals remarkable changes in the relative proportion of cell types in HDM. Nuclei were randomly sampled in equal numbers for each group (n = 15,939). (D) Relative proportion of each cell type in each condition. +: expansion; –: contraction. (E) Heatmap showing the molecular signature of each lineage

snRNA-seq on cardiac tissues from db/db (DM) and diabetic hypoglycemia db/db mice (HDM)

The cardiac tissues from the DM and HDM groups were collected for snRNA-seq (each group n = 6). All samples were individually sequenced. After quality control, 55,122 nuclei (HDM: 39,183; DM: 15,939) were obtained. According to the expression of established markers [16] for each lineage, seven cell types were identified by joint clustering of the snRNA-seq data from both conditions: fibroblasts (FBs, marked by Gsn), endothelial cells (ECs, marked by Pecam1), macrophages (MACs, marked by Mrc1), cardiomyocytes (CMs, marked by Ryr2), pericytes (PCs, marked by Abcc9), glial cells (GCs, marked by Csmd1), and epicardial cells (EpiCs, marked by Muc16, Fig. 2A, B). Nuclei were randomly sampled in equal numbers for each group (n = 15,939), and comparison of nucleus densities in the UMAP space between the two conditions revealed remarkable changes in the relative proportion of cell types in the HDM group (Fig. 2C). Subsequent quantitative analysis (Fig. 2D) demonstrated mild expansion of endothelial cells and fibroblasts, accompanied by a slight reduction in macrophages and cardiomyocytes. The molecular signatures of each lineage were presented in Fig. 2E. Given that cardiomyocytes, endothelial cells, macrophages, and fibroblasts constitute the predominant cell populations in cardiac tissue [17, 18], our subsequent investigations primarily focused on the specific regulatory alterations of these cell types under hypoglycemic conditions.

Endothelial cell-specific regulatory alterations in HDM

Unbiased clustering partitioned endothelial cells into four subtypes: Endo Cell-1, Endo Cell-2, Endo Cell-3, and Endo Cell-4 (Fig. 3A). Hierarchical clustering revealed the relationships among different types of endothelial cells (Fig. 3B), with EC1 and EC2 representing two distinct subgroups. EC2 expressed high levels of atherosclerotic and inflammation markers, such as Cd36 (Platelet glycoprotein 4) [19] and the proinflammatory/fibrosis marker Hdac9 (Histone deacetylase 9) [20, 21], indicating a proinflammatory phenotype. EC1 demonstrated high expression of anti-aopototic genes such as Npr3 (encoding natriuretic peptide receptor 3) [22] and endothelial protective genes such as Cgnl1 (encoding cingulin-like 1) [23], suggesting its key role in maintaining endothelial homeostasis (Fig. 3D). In HDM group, the EC2 population expanded while the EC1 population decreased (Fig. 3C), suggesting that hypoglycemia prompted endothelial cells to shift towards to a less stable state. Immunofluorescence staining confirmed the presence of these two subtypes in the tissues of DM and HDM mice (Fig. 3E). In endothelial cells, 151 and 251 genes were significantly upregulated and downregulated, respectively (absolute of log2 fold change > 1, adjusted p-value < 0.05) [24]. The upregulated genes were predominantly enriched in pathways related to endothelial inflammation and fibrosis (e.g., “fluid shear stress and atherosclerosis,” “focal adhesion”, Fig. 3F), as well as immune responses (e.g. “antigen processing and presentation”).

Fig. 3.

Endothelial cell - specific regulatory changes in HDM. (A) Endothelial cell subpopulations are visualized in a UMAP plot. (B) These subpopulations were subjected to hierarchical clustering. (C) The relative proportion of each subgroup within the endothelial cell compartment is shown for each condition. (D) The heatmap depicts the molecular signature defining each subpopulation. (E) Immunofluorescence staining of cardiac tissues from DM and HDM mice, with CD31 marking endothelial cells, confirmed the presence of these subgroups; bar = 50 μm. (F) Representative enrichment terms for differentially expressed genes between the two groups are displayed, with significance determined by a hypergeometric test (adjusted p < 0.05)

Transcriptomic dynamics during endothelial cell state transition in HDM

To elucidate the transcriptomic dynamics of endothelial cells transition toward inflammatory and pro-death states in HDM mice, we reconstructed the trajectory through the pseudo-temporal ordering of the nuclei of endothelial cells using Slingshot [25] (Fig. 4A). And it revealed that the EC2 subtype occupied a relatively late-stage position in the pseudo-time trajectory (Fig. 4B). The pseudo-time distributions between the two conditions differed significantly (Fig. 4C; p < 2.2e × 10^{−1 6}, Kolmogorov–Smirnov test). Through tradeSeq [26] analysis, the genes exhibiting significantly different expression patterns along the trajectory between the two conditions were identified and clustered into six gene clusters (Fig. 4D; adjusted p-value < 0.05). Subsequently, potential key genes were prioritized based on the results of three independent analyses, including the difference in expression patterns along the trajectory (adjusted p-value < 0.05), the fold change of expression levels between conditions (absolute of log2 fold change > 1), and the centrality change in GRNs (DRN rank < 1000). Only genes encoding transcription factors (TFs), ligands, and receptors were considered. Ultimately, 15 candidate genes were identified (Fig. 4E), with Fig. 4F displaying the smoothed expression curves of representative candidate genes along the trajectory under both conditions. Notably, genes encoding RNA-binding proteins associated with neurodegenerative diseases, such as Fus (encoding fused in sarcoma) [27], and autoimmune disease-related gene Zfp366 (encoding Zfp36 ring finger protein) [28] were identified for the first time in cardiac endothelial tissues under hypoglycemic conditions. Futhermore, PCD pathway-related genes, such as Ripk1 [29] and Fas [30] (encoding Fas cell surface death receptor), whose activation is closely related to damage such as apoptosis, as well as genes involved in endothelial inflammatory regulation and anti-fibrotic process e.g., Klf2 [31] (encoding Kruppel-like factor 2), Klf4 [32], and Sema3f [33] (encoding semaphorin 3F), were initially detected in cardiac endothelial cells of a diabetic hypoglycemia model. Western blot, immunohistochemistry, and immunofluorescence staining of Fas and Klf2 in cardiac tissues from both groups of mice confirmed that compared to DM mice, the expression level of Fas was significantly increased, while the expression of the endothelium-protective Klf2 was decreased in the cardiac tissues of HDM mice (n = 6; p < 0.05, Wilcoxon rank-sum test; Fig. 4G–J).

Fig. 4.

Transcriptomic dynamics during endothelial cell state transition in HDM. (A) Cellular trajectory reconstruction for the transition toward a failing endothelial cell state, performed using Slingshot. (B) Density curves illustrate the distribution of subpopulations along the trajectory. (C) Density curves showing the distribution of endothelial cells of the two groups along the trajectory. (D) A heatmap displayed the dynamic expression of genes with significantly different expression patterns along the trajectory between the two groups. (E) Key genes were prioritized by integrating three independent analyses: differential expression patterns, fold-change (FC) in expression levels, and centrality changes in gene regulatory networks (GRNs). The DRN rank denotes gene ordering based on GRN centrality changes from differential GRN analysis. log2FC represents the log2FC in endothelial cell expression. The wald stat is the natural logarithm of the statistic from the differential expression pattern analysis. Only genes encoding transcription factors, ligands, and receptors were considered. (F) Smoothed expression curves of representative candidate genes along the trajectory in two groups. (G) Western blotting and quantitative analysis of Klf2 and Fas protein expression in cardiac tissues from both groups (N = 6; *p < 0.05, Wilcoxon rank-sum test). (H–J) Representative immunohistochemical and immunofluorescence staining results for Klf2 and Fas in cardiac tissues from both mouse groups, with quantitative analysis (N = 6; *p < 0.05, Wilcoxon rank-sum test; scale bars = 100 μm or 25 μm)

Multi-cellular pathogenic shifts under hypoglycemia

Parallel to endothelial alterations, hypoglycemia induced maladaptive reprogramming across cardiac cell populations. Cardiomyocytes exhibited a shift from stable (CM2/Myh7⁺) to dysfunctional states (CM1/Pde4d⁺), with transcriptomic enrichment in fibrosis and arrhythmia pathways. Macrophages demonstrated M2-predominant polarization (Lyve1⁺/Egfr⁺) alongside pro-fibrotic gene signatures, while fibroblasts showed expansion of collagen-producing subsets (FB1) with Klf4↓/Zeb1↑ mediated pro-fibrotic switching. Critically, novel hypoglycemia-responsive regulators were identified across lineages (e.g., Bclaf1 in cardiomyocytes, Zbtb16 in macrophages, Zeb1 in fibroblasts), with protein validation confirming their dysregulation. These coordinated shifts reveal a multi-cellular pathological network promoting cardiac dysfunction under hypoglycemia. (Detailed cellular subtyping, trajectory analyses, and molecular validation are provided in Supplementary Figs. S4–S9).

Alterations in intercellular communication in HDM cardiac tissues inferred from snRNA-seq

Endothelial cells, macrophages, cardiomyocytes, and fibroblasts play the dominant role in cardiac tissues, coordinately regulating cardiac function. Based on snRNA-seq data, we applied CellChat [34] to infer ligand-receptor interactions among cell subtypes in DM and HDM cardiac tissues. The inferred total number (Fig. 5A) was significantly decreased, yet the strength of interactions (Fig. 5B) was significantly increased in HDM, suggesting enhanced intercellular communication under disease conditions. Notably, endothelial cells exhibited a substantial increase in both outgoing and incoming signaling interactions in terms of quantity and intensity (Fig. 5C), highlighting their central role in the pathophysiology of HDM. Of particular interest, the communication between CM1 and EC2 was significantly enhanced in HDM, as evidenced by the relative positioning of cardiac cells and endothelial cells in the two-dimensional signaling space when comparing DM and HDM (Fig. 5D, E). Subsequently, we compared the relative information flow of signaling pathways between the two groups (Fig. 5F). The results demonstrated significant activation of the VISFATIN, ESAM, SEMA6, and COLLAGEN pathways in HDM, along with the detection of an HDM-specifically activated EPHA pathway. Functional similarity clustering of the inferred communication networks via joint manifold learning (Fig. 5G) revealed the most pronounced Euclidean distance change in the COLLAGEN pathway (Fig. 5H), indicating its most significant functional alteration. Network centrality analysis demonstrated substantial shifts in the sender and receiver of the COLLAGEN pathway in HDM (Fig. 5I). Although the primary sender and receiver remained unchanged, the top mediator transitioned from EC1 to EC2, with the latter also emerging as the top influencer. Further analysis revealed that the Col4a1−(Itga9+Itgb1) ligand-receptor pair exhibited the highest contribution within the COLLAGEN signaling network of HDM (Fig. 5J). Western blot analysis confirmed the upregulation of these proteins in HDM (Fig. 5K). The spatial proximity and physical interaction between Col4a1, Itga9, and Itgb1 were validated using immunofluorescence staining and co-immunoprecipitation (co-IP), as shown in Supplementary Fig. S10, consistent with previous reports of Col4a1-integrin binding under pathological conditions [35]. Given that Col4a1, Itga9, and Itgb1 have been reported as extracellular matrix (ECM) [36–38], and excessive ECM accumulation can induce EndMT, we additionally assessed α-SMA, TGF-β, and CD31 to evaluate EndMT (Fig. 5L, N). The results confirmed EndMT activation in HDM. Furthermore, Masson’s staining revealed increased collagen deposition in HDM myocardial sections (Fig. 5M). Collectively, these results untangle the pivotal role of EndMT in the pathophysiology of cardiac dysfunction induced by hypoglycemia.

Fig. 5.

Changes in intercellular communication in HDM cardiac tissue inferred from the scRNA-seq data. (A) Bar plot showing the total number of ligand-receptor interactions among the subpopulations of the cardiac tissues in two groups. (B) Bar plot showing the total interaction strength among the subpopulations of the cardiac tissues in both conditions. The total interaction strength was calculated by summing the communication probability of all inferred interactions. (C) Heatmap showing the differential number of interactions and differential interaction strength among subpopulations in two groups. (D, E) Bubble plot showing the incoming and outgoing interaction strength for each subpopulation in DM and HDM. (F) Relative information flow for each signaling pathway in two groups. (G) Joint manifold learning of the DM and HDM communication networks and grouping the signaling pathways based on functional similarity. (H) The Euclidean distance of each pathway in the learn joint manifold. A larger distance means a larger difference in functional similarity between two groups. (I) The major senders and receivers of the COLLAGEN signaling pathway inferred through network centrality analysis in DM (upper panel) and HDM (lower panel). (J) Relative contribution of each ligand-receptor pair to the overall signal of the COLLAGEN pathway in HDM. (K, L) Protein expression of Col4a1, Itga9, Itgb1, α-SMA and Tgf-β of cardiac tissues of two groups by Western blotting and quantitative analysis. N = 6, *p < 0.05, Wilcoxon rank-sum test. (M) Representative masson staining of heart tissues and quantitative analysis of collagen components in the two groups of mice. N = 6, *p < 0.05, Wilcoxon rank-sum test, bar = 200 μm. (N) Immunofluorescence staining showing the expression of Cd31 (red) and α-SMA (green) in DM and HDM heart tissues, bar = 10 μm

Pyroptosis, apoptosis, and necroptosis (PANoptosis) of diabetic cardiac endothelial cells are ignited by hypoglycemia

snRNA-seq revealed a significant upregulation of Ripk1 expression in endothelial cells under hypoglycemic conditions. Immunofluorescence co-localization staining for Ripk1 across cardiomyocytes, macrophages, endothelial cells, and fibroblasts further confirmed endothelial cells as the primary effector population (Fig. 6A). As a key regulator of cellular damage, Ripk1 has been demonstrated to modulate multiple cell damage pathways, including apoptosis, pyroptosis, and necroptosis [39]. These forms of cellular damage further promotes ECM accumulation [40], thereby driving EndMT and adversely affecting cardiac function. To elucidate the mechanism of endothelial cell injury under hypoglycemic conditions, we examined the expression and activity of core molecules in the aforementioned cell injury pathways in cardiac tissues of DM and HDM mice. The results showed that markers of pyroptosis (Caspase-1 and Gsdmd), apoptosis (Caspase-3 and Caspase-8), and necroptosis (Mlkl, Ripk1, and Rip3) were significantly upregulated, indicating that hypoglycemia activates PANoptosis in cardiac tissues. This conclusion was further supported by the increased expression of Nlrp3, a component of the PANoptosome complex (Fig. 6B). Given that PANoptosis involves multiple upstream sensor proteins, their expression was also assessed. As shown in Supplementary Fig. S11, no significant differences were observed in the expression of Aim2, Zbp1, and Nlrp12 in cardiac tissues from DM and HDM mice, further suggesting that Ripk1 may play a more dominant role in hypoglycemia-associated diabetic cardiac injury. In the in vitro low-glucose-induced MCMEC model, the PANoptosis pathway showed significant activation with repeated hypoglycemic exposure. Additionally, COLLAGEN pathway genes (Col4a1, Itga9, and Itgb1) and EndMT markers (α-SMA and TGF-β) were upregulated, while CD31 expression was markedly downregulated (Fig. 6C). Collectively, these in vivo and in vitro findings demonstrate that hypoglycemia induces concurrent PANoptosis and EndMT in both cardiac tissue and MCMECs.

Fig. 6.

Recurrent hypoglycemia induced apoptosis, necroptosis, and pyroptosis (PANoptosis) and EndMT in cardiac tissues of db/db mice and MCMECs. (A) Immunofluorescence staining showed the distribution of Ripk1 (green) in cardiomyocytes (marked by Ryr2, red), macrophages (marked by Mrc1, red), endothelial cells (marked by Cd31, red) and Fibroblasts (marked by Dcn, red), Bar = 100 μm. (B)Western blotting and quantitative analysis of protein expression of Nlrp3, Ripk1 (members of the PANoptosome), Casp 1, cleaved Casp 1, Gsdmd-f, Gsdmd-n(members of the pyroptosis), Casp 3, cleaved Casp 3, Casp 8 (members of the apoptosis), and Rip3, p-Rip3, t-Mlkl, p-Mlkl (members of the necroptosis) in cardiac tissues of two groups of mice. (C) Western blotting and quantitative analysis of protein expression of PANoptosis pathway and of COLLAGEN pathway proteins Col4a1, Itga9, Itgb1, and of EndMT marker proteins α-SMA and Tgf-β, and endothelial-specific marker CD31 in MCMECs. In the high glucose (HG) group, the glucose concentration was 35 mM. And cells in the low glucose (LG) group were subjected to high glucose treatment first, followed by one, two, or three cycles of low glucose medium treatment with a glucose concentration of 3 mM. Data are presented as mean ± sSEM, n = 6, Wilcoxon rank-sum test, *p < 0.05

RIPK1 inhibition suppressed the activation of the COLLAGEN pathway and alleviated PANoptosis and EndMT in MCMECs under low glucose treatment

To determine whether MCMECs undergo PANoptosis, cells were pretreated with the apoptosis inhibitor Z-VAD-FMK, the necroptosis inhibitor Bardoxolone, and the pyroptosis inhibitor Disulfiram before low-glucose exposure. None of these inhibitors alone effectively suppressed the upregulation of multiple PANoptosis-related proteins (Fig. 7A), confirming the occurrence of PANoptosis. Since RIPK1 acts as an upstream sensor protein in the PANoptosis pathway, we used small interfering RNA (siRNA) targeting Ripk1 to clarify its regulatory role. Transfection of siRNA into MCMECs, followed by HG or LG treatment, showed that Ripk1 inhibition significantly attenuated low glucose-induced PANoptosis. The expression levels of PANoptosis pathway-related proteins were either lower than or not significantly different from those in the control group (Fig. 7B, C), contrasting with the effects observed without RIPK1 inhibition. Moreover, RIPK1 suppression alleviated the EndMT and COLLAGEN pathway (Fig. 7D). Pharmacological inhibition of Ripk1 using Nec1 also effectively mitigated MCMECs injury in the lowglucose model (as shown in Supplementary Fig. S12). These results demonstrate that low glucose-induced endothelial PANoptosis and EndMT are mediated by Ripk1, and that RIPK1 inhibition significantly suppresses endothelial injury.

Fig. 7.

Inhibition of Ripk1 expression in MCMECs could significantly suppress the PANoptosis and EndMT processes induced by recurrent low-glucose. (A) Western blotting and quantitative analysis of the protein expression of Gsdmd-f, Gsdmd-n, Asc, Ripk1, p-Ripk1, Nlrp3, Casp 8, Casp3, and cleaved Casp 3 in MCMECs. The concentrations of Bar, Dis and Z-vad were 100 nM, 5 μM and 50 μM, respectively. (B-D) Western blot analysis and quantitative results of protein expression related to PANoptosis, EndMT, and COLLAGEN pathways in MCMECs. The HG group was cultured in medium with a glucose concentration of 35 mM. Cells in the LG group were first subjected to HG treatment, followed by three cycles of treatment with medium containing a glucose concentration of 3 mM. Data are presented as mean ± SEM, n=6, wilcoxon rank-sum test, *p < 0.05

Endothelial cell-specific RIPK1 conditional knockout improves cardiac dysfunction in HDM

To investigate the cellular mechanism of Ripk1 in recurrent hypoglycemia-induced cardiac dysfunction, we generated endothelial cell-specific Ripk1 conditional knockout mice (Ripk1^△EC) and exposed them to the same diabetic-hypoglycemia model as their littermate controls (Ripk1^fl/fl). Echocardiography revealed that Ripk1^△EC mice exhibited significantly higher LVEF and LVFS than Ripk1^fl/fl controls (Fig. 8A). Western blot analysis showed markedly lower expression levels of proteins associated with PANoptosis, EndMT, and COLLAGEN pathways in the cardiac tissue of Ripk1^△EC mice (Fig. 8B-D). This reduction indicates that endothelial-specific deletion of Ripk1 effectively attenuates hypoglycemia-induced impairment of cardiac function in diabetes. Together, these results demonstrate that endothelial Ripk1 substantially contributes to diabetic hypoglycemia-induced cardiac dysfunction by regulating multiple pathological pathways.

Fig. 8.

Endothelial cell-specific RIPK1 conditional knockout improves cardiac dysfunction in HDM. (A) Representative transthoracic M-mode echocardiograms and quantitative analysis of LVEF and LVFS are shown for Ripk1^△EC and Ripk1^fl/fl mice after induction of the diabeteshypoglycemia model (n = 9, *p < 0.05). (B) Western blotting and quantitative analysis of COLLAGEN pathway proteins Col4a1, Itga9, Itgb1, and of EndMT marker proteins α-SMA and Tgf-β, and endothelial-specific marker CD31 in cardiac tissues of two groups of mice. (C, D) Western blotting and quantitative analysis of expression of PANoptosis pathway proteins. Data are presented as mean ± SEM, n = 6, Wilcoxon rank-sum test, *p < 0.05

Discussion

The adverse impact of hypoglycemia on cardiac function in patients with DM have long been a focus of clinical concern. As diabetes progresses to advanced stages, the risk of hypoglycemia rises substantially. This increase is closely linked to declining physiological function, necessary adjustments to treatment regimens, and a higher burden of comorbidities during this period [41]. To achieve glycemic targets in advanced diabetes, clinical management often necessitates insulin replacement therapy or intensive glucose-lowering strategies. Insulin use has therefore become a principal trigger for hypoglycemia, especially among patients with advanced disease [42]. A retrospective study conducted in patients hospitalized for acute heart failure revealed that the coexistence of T2DM and hypoglycemia (blood glucose ≤ 3.9 mmol/L) at admission was a significant predictor of 3P-MACE (major adverse cardiovascular events). Notably, heart failure patients with both T2DM and hypoglycemia exhibited the highest all-cause mortality, significantly surpassing those with HF alone or HF combined with T2DM [43]. Even recurrent non-severe hypoglycemic episodes were associated with an elevated risk of subsequent severe hypoglycemia and major adverse cardiovascular events [44]. These findings collectively indicated that hypoglycemia exerts multifaceted detrimental effects on the cardiovascular system of T2DM patients. In this study, we observed a deterioration of cardiac function in diabetic patients and db/db mice, consistent with the conclusion of numerous prior investigations.

The impact of glycemic variability on endothelial function has been widely recognized. W L Jin et al. demonstrated that recurrent hypoglycemia exacerbated monocyte-endothelial cell adhesion in the aorta of GK rats and promoted nuclear translocation of NF-κB in endothelial cells [45]. In clinical observations, Keiichi Torimoto et al. also found that the reactive hyperemia index (RHI), a surrogate marker of endothelial function, was modulated by glycemic fluctuations, suggesting hypoglycemia as a key trigger for endothelial dysfunction [46]. In this study, snRNA-seq analysis of cardiac tissues from DM and HDM mice similarly revealed endothelial injury under hypoglycemia conditions. Among endothelial cells, the EC2 subtype with pro-inflammatory and pro-fibrotic properties was significantly upregulated in HDM. Pseudotime trajectory analysis further demonstrated the progressive dominance of EC2 during disease progression. Differentially expressed genes in endothelial cells were predominantly enriched in pathways such as “epithelial cell migration,” “fluid shear stress and atherosclerosis,” and “focal adhesion”, which are closely associated with endothelial inflammatory responses and fibrotic processes (including EndMT). Consistently, similar subtype shifts related to inflammation and fibrosis were observed in cardiac cells, macrophages, and fibroblasts. For instance, hypoglycemia activated the CM1 phenotype in cardiac cells, characterized by the upregulation of PDE4D—a marker gene previously linked to diabetic cardiac dysfunction. DEGs in cardiac cells were primarily enriched in pathways such as “myofibril assembly” and “contractile function”. In HDM mice, M1 macrophages, which maintain vascular homeostasis, were significantly reduced during disease progression, whereas the proportion of M2 macrophages with pro-fibrotic and apoptotic phenotypes increased. Fibroblast DEG enrichment analysis further highlighted pathways including “collagen-rich extracellular matrix” and “diabetic cardiomyopathy.” These findings provide additional evidence supporting the inflammatory and fibrotic alterations induced by hypoglycemia.

Subsequent analysis of intercellular communication among these four cell types revealed a significant enhancement of EC2–CM1 interactions in HDM, suggesting that endothelial-myocardial crosstalk may represent a pivotal pathological mechanism, with endothelial cells serving as the primary mediators of this process. In HDM, the COLLAGEN pathway was markedly activated, with Col4a1, Itga9, and Itgb1 emerging as the most significantly altered proteins within this pathway. These molecules have been previously reported as ECM that mediate cell-cell and cell-matrix adhesion and migration. Their excessive deposition can trigger endothelial-to-mesenchymal transition (EndMT), thereby disrupting normal myocardial architecture [36, 37, 47]. EndMT has been closely associated with endothelial cell injury. When endothelial cells are damaged, they release inflammatory cytokines that induce ECM deposition. Under the mediation of TGF-β, endothelial cells progressively lose their characteristic CD31 phenotype while acquiring mesenchymal cell markers (e.g., α-SMA). This phenotypic transition, accompanied by collagen fiber deposition, ultimately leads to tissue fibrosis [37]. In the present study, the activation of collagen pathways and EndMT was confirmed through Western blot analysis, Masson’s trichrome staining, and immunofluorescence assays.

Although the role of endothelial dysfunction in hypoglycemia-induced cardiac injury in DM has been recognized, the underlying pathogenic mechanisms remain incompletely elucidated. In this study, through phenotypic characterization of cardiac tissue and MCMECs under hypoglycemia conditions, it was observed that proteins associated with apoptosis, pyroptosis, and necroptosis pathways were significantly upregulated, indicating the occurrence of PANoptosis. In endothelial cells, the expression of RIPK1 was significantly upregulated. As one of the upstream sensor proteins of PANoptosis, RIPK1 has been reported to ameliorate myocardial fibrosis in T2DM rats induced by high glucose and high fat, primarily through the inhibition of RIPK1-mediated autophagy [48]. Liu Q et al. found that inhibition of the RIPK1/MTORC1 pathway could also improve apoptosis, inflammation, and oxidative stress in cardiac fibroblasts induced by hyperglycemia, thus inhibiting fibrotic changes [49]. Further examination of multiple PANoptosis upstream sensors revealed that only Ripk1 was significantly unregulated under recurrent hypoglycemia condition in mice cardiac tissues, suggesting its potentially dominant role in this process. However, the effect of RIPK1 on endothelial cells and EndMT remains uncharacterized, and its role in hypoglycemia-associated cardiac dysfunction in diabetes has seldom been investigated.

To further elucidate the regulatory role of RIPK1 in endothelial cells under hypoglycemia conditions, we inhibited RIPK1 expression in MCMECs using siRNA and the specific inhibitor Nec-1. When cells were treated with high- and low-glucose, the expression levels of multiple proteins in the PANoptosis pathway exhibited varying degrees of reduction, with some even significantly lower than those observed in the low-glucose-treated control group. However, inhibition of apoptosis, necrosis or pyroptosis pathways alone failed to achieve the therapeutic effects comparable to RIPK1 inhibition. Moreover, following RIPK1 inhibition, the expression levels of COLLAGEN pathway proteins and EndMT marker proteins in low-glucose treated group were significantly downregulated, whereas CD31 expression was upregulated. These results indicated that RIPK1 inhibition effectively reduces ECM deposition by surpressing PANoptosis in endothelial cells, thereby suppressing EndMT progression and COLLAGEN pathway. These results indicate that inhibition of RIPK1 can effectively reduce extracellular matrix (ECM) deposition by suppressing PANoptosis in endothelial cells, thereby inhibiting the EndMT process and collagen pathway activation. To validate this underlying mechanism, endothelial cellspecific Ripk1 conditional knockout mice (Ripk1^△EC) were generated, and following induction by diabetesrecurrent hypoglycemia, cardiac function in Ripk1^△EC mice was significantly improved compared with control Ripk1^fl/fl mice, as evidenced by increased LVEF and LVFS. Western blot analysis further demonstrated that the expression of proteins associated with PANoptosis, EndMT damage, and the COLLAGEN pathway in cardiac tissues of Ripk1^△EC mice was markedly reduced relative to control mice. In conclusion, our study demonstrated that RIPK1-mediated PANoptosis exacerbates cardiac fibrosis by activating EndMT in diabetic hypoglycemia injury. Inhibition of RIPK1 effectively suppresses PANoptosis and EndMT, consequently ameliorating cardiac functional impairment.

Study limitations

We did not examine the long-term effects of Ripk1 inhibition therapy or potential compensatory pathway activation in this study, leaving its sustained efficacy and safety profile incompletely defined. Furthermore, the research was conducted exclusively in the db/db mouse model and was not validated in other diabetic animal models, limiting the generalizability of the findings. Although the ex vivo model incorporated multiple time points to observe the temporal progression of endothelial cell injury under low-glucose conditions, a corresponding multi-timepoint design was absent from the in vivo studies, and this omission represents a critical direction for subsequent investigation.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Shuting Chang: conceptualization, methodology, formal analysis, investigation, writing - Original Draft, funding acquisition; Yuce Peng: sftware, data curation; Minghao Luo: writing - review & editing, funding acquisition, supervision; Dingyi Lv: writing - review & editing, supervision; Na Li: methodology, funding acquisition, formal analysis; Guanzhao Zhang: methodology, formal analysis; Yi Jiang: methodology, formal analysis; Dan MA: methodology, investigation; Yanyao Huang: methodology, investigation; Xunjia Li: investigation, writing - Original Draft, funding acquisition; Deyu Zuo: investigation, writing - Original Draft, funding acquisition; Suxin Luo: project administration; visualization; An He: conceptualization, visualization, formal analysis, funding acquisition, investigation. An He was the only one who handled all processes related to the UK Biobank data and drew the graphical abstract. All authors have read and approved the final manuscript.

Funding

National Natural Science Foundation of China (NO. 82300477), Chongqing Natural Science Foundation (General Project) (NO. CSTB2022NSCQ-MSX0081), Chongqing Postdoctoral Innovative Talent Support Program (NO. CQBX202213). The National Natural Science Foundation of China (NO. 82305006 and 82574619), The Chongqing Natural Science Foundation Innovation and Development Joint Fund (Chongqing Education Commission) (NO. CSTB2024NSCQ- LZX0079), The Chongqing Medical Young Talents Program (No. YXQN202415). Doctoral Candidate Scientific Research Innovation Project of The First Affiliated Hospital of Chongqing Medical University (CYYY-BSYJSKYCXXM202402). Scientific and Technological Research Program of Chongqing Municipal Education Commission (KJQN202400434).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

All animal experiments were conducted in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Animal Ethics Committee of The First Affiliated Hospital of Chongqing Medical University.

Consent for publication

Written informed consent was obtained from all individual participants for the publication of any potentially identifiable data or images included in this article.

Competing interests

The authors declare that they have no competing interests.