Neutrophil extracellular traps and peptidylarginine deiminase 4–mediated inflammasome activation link diabetes to cardiorenal injury and heart failure

Abstract

Background and Aims

Diabetes is associated with increased risk of cardiovascular and renal disease. This study investigated the role of peptidylarginine deiminase 4 (PAD4), neutrophil extracellular traps (NETs), and inflammasome activation in diabetic cardiomyopathy (DCM) and kidney disease (DKD).

Methods

Endomyocardial biopsies (EMB) from heart failure (HF) patients (n = 20) with or without diabetes were stained for NETs. Wild-type (WT) and PAD4⁻/⁻ mice were subjected to streptozotocin (STZ)-induced diabetes and cardiac function, blood glucose, body weight, and exercise tolerance were assessed longitudinally. NETosis and ASC specks were evaluated in mouse and human neutrophils. Cardiac and renal fibrosis was assessed by Sirius Red/Fast Green staining. Confocal microscopy, ELISA, and flow cytometry were used to quantify NETs, IL-1β, von Willebrand factor (VWF), cytokine transforming growth factor beta-1 (TGF-β1), and neutrophil infiltration.

Results

Myocardial NET burden was increased in HF patients with diabetes. High glucose triggered inflammasome activation in human neutrophils. After STZ, PAD4⁻/⁻ and WT mice developed hyperglycaemia and weight loss, yet only WT neutrophils showed increased NETosis and ASC speck formation. Only diabetic WT mice exhibited elevated IL-1β and VWF levels, impaired cardiac function, reduced exercise tolerance, and pulmonary oedema; PAD4⁻/⁻ mice were protected. Wild-type diabetic hearts and kidneys showed greater fibrosis, neutrophil infiltration, NETs, and TGF-β1 levels. Kidney injury in WT mice was reflected by albuminuria and renal fibrosis, whereas PAD4⁻/⁻ mice preserved renal function.

Conclusions

Diabetes promotes neutrophil inflammasome activation and NETosis, driving cardiac and renal inflammation and fibrosis. Peptidylarginine deiminase 4 deficiency prevents heart failure and preserves kidney function in experimental diabetes.

Graphical Abstract

Hyperglycaemia activates the neutrophil inflammasome and PAD4-dependent NETosis, promoting IL-1β/TGF-β–driven inflammation and fibrosis in diabetic heart and kidney disease; PAD4 deficiency protects from these complications.

Introduction

Diabetes mellitus is a growing public health concern and a major driver of chronic kidney disease (CKD) and heart failure (HF) through metabolic, inflammatory, and neurohormonal pathways.^1–3 These conditions synergistically exacerbate morbidity and mortality in affected patients.^4–6 Despite efforts to manage hyperglycaemia, dyslipidaemia, and hypertension, many patients still progress to HF or CKD,⁷ underscoring the need for strategies that ensure cardiorenal protection.^8–10 However, due to the poorly understood molecular pathobiology of diabetes-related cardiorenal disease, targeted prevention methods are scarce. Emerging evidence suggests that diabetes-associated thromboinflammation plays a crucial role in the development and progression of these conditions.¹¹ Neutrophils are increasingly recognized as pivotal players in thromboinflammation.^12,13 Once activated, a subset of neutrophils can undergo a form of controlled cell death involving the release of unravelled nuclear DNA, termed neutrophil extracellular traps (NETs).¹³ We have previously shown that neutrophils from type 1 and 2 diabetic humans and mice are primed to release NETs in a process that heavily relies on the elevated expression and enzymatic activity of peptidylarginine deiminase 4 (PAD4).¹⁴ Decorated with histones, cytoplasmic and granular proteins, NETs provide large scaffolds propelling thrombosis, harm surrounding tissues, and present neoantigens, potentially triggering autoimmune disease.^13,15 This aligns well with our reports that PAD4-directed NETosis together with NLRP3 inflammasome assembly stimulate adverse myocardial remodelling in myocardial infarction,^{16 ,17} obesity,¹⁸ autoimmune disease,¹⁹ and advanced age.²⁰ Remarkably, the recent discovery of NETs in the myocardium of chronic HF patients marks a paradigm shift, highlighting the role of chronic neutrophil-mediated inflammation beyond acute cardiovascular events.^21,22 However, little is known about the pathobiology of NET-mediated injury and the effects of PAD4 in diabetic cardiorenal disease.

We recently demonstrated that PAD4 supports NLRP3 inflammasome assembly in vitro²³ and that NLRP3 activation promotes NET formation and adverse remodelling in myocardial infarction.¹⁷ NLRP3 components, including ASC, caspase-1, and IL-1β/IL-18, are closely associated with diabetes pathogenesis and complications.^24–27 Inhibiting these pathways can mitigate diabetic cardiomyopathy (DCM)^28–30 and nephropathy (DKD).³¹

To better understand how diabetes drives secondary organ damage, this study investigates the interplay between PAD4, NETs, inflammasome activation, and fibrosis in DCM and nephropathy. While PAD4-driven NETosis impairs wound healing in type 1 diabetes (T1D),¹⁴ its role in heart and kidney dysfunction remains poorly defined. We recently demonstrated that pharmacological PAD4 inhibition with the oral compound JBI-589 prevents inflammation-mediated HF in a murine model of rheumatoid arthritis,¹⁹ suggesting broader relevance in chronic inflammatory cardiac injury. Together with the detection of NETs in myocardium from chronic HF patients,^21,22 these findings support our hypothesis that PAD4 could promote maladaptive organ remodelling in diabetes more generally.

Methods

Human myocardial biopsies and neutrophil extracellular trap quantification

Endomyocardial biopsies (EMB) were obtained from the Cardiopathology, University Hospital Tübingen. This study was conducted from retrospective data collection and anonymized analysis of patients in accordance with the Declaration of Helsinki and in accordance with local government law without the requirement for informed consent under ethics approval no. 411/2021BO. Patients had HF of initially unclear aetiology; ischaemic heart disease, dilated cardiomyopathy, hypertrophic cardiomyopathy, restrictive cardiomyopathy, myocarditis, and amyloidosis were excluded during clinical work-up. For this analysis, 10 patients with diabetes and 10 without diabetes were included based on sufficient tissue for staining/quantification. Four-micrometre-thick paraffin sections underwent confocal microscopy for NETs, defined as extracellular structures positive for citrullinated histone 3 (H3Cit) and myeloperoxidase (MPO). The entire section of the EMB was systematically examined in all available high-power fields, and quantification was independently performed by two investigators blinded to the sample origin. Extended patient characteristics, staining protocols, and detailed quantification procedures are provided in the Supplementary data.

Human peripheral blood neutrophil isolation and ASC speck assay

Peripheral venous blood from healthy donors was collected after written informed consent (ethics application no. EK-Freiburg: 24-1057-S1). Neutrophils were isolated and incubated with either physiological glucose (5.5 mM) or high glucose (22 mM), unstimulated or stimulated with ionomycin (8 μM) for 2 h at 37°C in 5% CO₂. Inflammasome assembly was assessed by confocal microscopy imaging of ASC specks, and NET formation was assessed by MPO and extracellular DNA staining. Detailed staining protocols, antibody lists, and imaging parameters are provided in the Supplementary data.

Animal model and study design

All animal experiments were approved by the Institutional Animal Care and Use Committee of Boston Children’s Hospital (Protocol #20-01-4096R) and conducted according to NIH and ARRIVE guidelines. Male wild-type (WT) C57BL/6J and PAD4⁻/⁻ mice (on a C57BL/6J background) were housed under specific pathogen-free conditions. Type 1 diabetes was induced with streptozotocin (STZ; 50 mg/kg daily for 5 days). Mice were followed for 8 weeks with serial assessment of body weight, blood glucose, cardiac function by echocardiography (with a HF cut-off for left ventricular ejection fraction (LVEF) < 45%), and treadmill endurance testing. Details of histology, immunofluorescence, protein quantification, in vitro assays, and flow cytometry are provided in the Supplementary data.

Statistical analysis

Normality and homogeneity of variance were assessed before applying parametric or non-parametric tests. Comparisons between two independent groups used unpaired t-tests (Welch’s correction applied where variances were unequal) or Mann–Whitney U tests; ≥3 groups were analysed by one-/two-way ANOVA (Welch’s correction where appropriate; Tukey or Šidák post hoc) or Kruskal–Wallis (Dunn post hoc). For intradonor experiments with repeated conditions, two-way repeated-measures ANOVA with Bonferroni post hoc testing was applied; correlations used Spearman’s rank correlation. All analyses were two-sided with α= .05. Exact n, test type, and P values are reported in figure legends and Supplementary data online, Tables S1–S4.

A full description of experimental protocols, antibody lists, ELISA kits, and statistical procedures is provided in the Supplementary data.

Results

Myocardial neutrophil extracellular trap burden is increased in diabetic heart failure and high glucose promotes neutrophil inflammasome activation and NETosis

Endomyocardial biopsies were obtained from patients with HF of initially unclear aetiology (n = 20), in whom ischaemic heart disease, dilated cardiomyopathy, hypertrophic cardiomyopathy, restrictive cardiomyopathy, myocarditis, and amyloidosis were excluded. Among them, 10 patients had HF with diabetes and 10 had HF without diabetes. Age, sex, and HF phenotype distribution (HFpEF, HFmrEF, HFrEF) were comparable between groups (see Supplementary data online, Table S1 and Figure S1A). Confocal staining for H3Cit and MPO revealed NETs in all patient groups, but the fraction of NETing neutrophils (%H3Cit⁺ neutrophils) was higher in diabetic HF (Figure 1A and B). Across the cohort, NET burden inversely correlated with LVEF (Spearman r =−.57, 95% CI −.81 to −.15, P = .009), suggesting that greater NET burden is associated with more severe impairment of systolic function (Figure 1C). Both solitary neutrophils/NETs and focal ‘hotspots’ with dense infiltration and clustered NET deposits were observed within the same biopsies. These clusters likely represent localized sites of elevated inflammatory activity (Figure 1A).

Figure 1.

Myocardial neutrophil extracellular trap burden is increased in diabetic heart failure, and high glucose promotes neutrophil inflammasome activation and NETosis. (A) Representative confocal microscopy images (63×) of endomyocardial biopsy sections from patients with heart failure without and with diabetes, immunostained for myeloperoxidase (MPO), citrullinated histone 3 (H3Cit), and nuclei (DAPI). (B) Quantification of myocardial neutrophil infiltration (MPO⁺ cells mm⁻²) and NET burden (% H3Cit⁺ neutrophils) in heart failure patients without (n = 10) and with diabetes (n = 10). Data are presented as median (IQR) and compared using the Mann–Whitney U test. (C) Correlation between left ventricular ejection fraction and myocardial neutrophil extracellular trap burden across all patients. Correlation was assessed using Spearman’s rank correlation, and the regression line was fitted by simple linear regression using ordinary least squares; dashed curves indicate the 95% confidence band of the fit (Spearman r = −.57, 95% CI −.81 to −.15, p = .009). (D) Representative confocal microscopy images (63×) of peripheral blood neutrophils from healthy donors (n = 7) incubated for 2 h with physiological (5.5 mM) or high (22 mM) glucose, with or without ionomycin (8 μM), immunostained for MPO, ASC, and nuclei (DAPI). (E) Quantification of ASC speck formation (% of cells) and neutrophil extracellular trap formation (% H3Cit⁺ cells) under the indicated conditions. Data are shown as mean ± SD and analysed by repeated-measures two-way ANOVA [factors: glucose (5.5 mM vs 22 mM) and stimulation (±8 μM ionomycin)]. Individual donor trajectories can be found in Supplementary data online, Figure S2B. Exact P values are shown in the figure. ASC, apoptosis-associated speck-like protein containing a CARD; DAPI, 4′,6-diamidino-2-phenylindole; H3Cit, citrullinated histone 3; HF, heart failure; LVEF, left ventricular ejection fraction; MPO, myeloperoxidase; NET, neutrophil extracellular trap

Peripheral blood neutrophils from healthy donors (n = 7) were exposed to physiological (5.5 mM) or high (22 mM) glucose for 2 h, with or without stimulation. High glucose alone increased ASC speck formation in isolated neutrophils, an effect that became even more pronounced under ionomycin co-stimulation (Figure 1D and E). Consistent with our previous work linking hyperglycaemia to NET release,¹⁴ high glucose also enhanced NET formation, which strongly correlated with ASC speck formation (Spearman r = .89, 95% CI .77–.95, P < .0001) (see Supplementary data online, Figure S2B). Confocal images frequently showed ASC specks embedded within extracellular NET structures (Figure 1D; Supplementary data online, Figure S2A), supporting the close functional relationship between inflammasome activation and NETosis. These findings indicate elevated glucose is sufficient to directly prime and activate the neutrophil inflammasome and promote NETosis, providing a possible mechanistic bridge between hyperglycaemia and the heightened intracardiac NET burden observed in diabetic patients.

Neutrophil inflammasome activation and NETosis are peptidylarginine deiminase 4–dependent in streptozotocin-induced type 1 diabetes

To investigate the mechanistic role of PAD4, NETs, and inflammasome in diabetic kidney and cardiovascular disease, a diabetes-like phenotype was induced in WT and PAD4^−/− mice through intraperitoneal injections of STZ for five consecutive days (Figure 2A).

Figure 2.

Neutrophil inflammasome activation and NETosis are peptidylarginine deiminase 4–dependent in streptozotocin-induced type 1 diabetes. (A) Timeline of experimental procedures: streptozotocin (50 mg/kg/day, i.p.) was injected for 5 consecutive days in wild-type and PAD4⁻/⁻ mice to induce type 1 diabetes mellitus. Blood glucose, echocardiography, and exercise tolerance were monitored every 2 weeks; analyses were performed 8 weeks post-induction. (B) Fed plasma glucose levels over time in WT + STZ (n = 10), PAD4⁻/⁻+ STZ (n = 10), and WT controls (n = 9). (C) Body weight (left) and relative body weight gain (right) over time in the same groups. Corresponding area under the curve analyses for glucose and body weight trajectories are provided in Supplementary data online, Figure S3. (D) Representative fluorescence microscopy images of neutrophils from streptozotocin-treated mice, unstimulated or treated with ionomycin (8 μM, 4 h); DNA (DAPI), ASC; red arrows indicate ASC speck oligomerization (representative of n = 5 experiments). (E) Quantification of ASC speck⁺ neutrophils (%) from WT + STZ, PAD4⁻/⁻+ STZ, and WT controls ± ionomycin (n = 5). (F) IL-1β in supernatants of neutrophils from WT + STZ and PAD4⁻/⁻+ STZ mice ± ionomycin (n = 5). (G) Plasma IL-1β and (H) plasma von Willebrand factor concentrations at 8 weeks post-streptozotocin (n = 8). (I) Neutrophil-to-lymphocyte ratio at 8 weeks in WT + STZ (n = 9), PAD4⁻/⁻+ STZ (n = 9), and WT controls (n = 8). Panels (B) and (C), (E) and (F), and (I) show data as mean ± SD and were analysed by ordinary one-way ANOVA; panel (G) shows median (IQR) and was analysed using the Mann–Whitney U test; panel (H) shows mean ± SD and was analysed using an unpaired t-test. Exact P values are shown in the figure. ASC, apoptosis-associated speck-like protein containing a CARD; DAPI, 4′,6-diamidino-2-phenylindole; IL-1β, interleukin-1β; NLR, neutrophil-to-lymphocyte ratio; PAD4, peptidylarginine deiminase 4; STZ, streptozotocin; T1D, type 1 diabetes mellitus; VWF, von Willebrand factor; WT, wild type

As expected,¹⁴ diabetes developed to the same degree in both PAD4^−/− and WT mice (Figure 2B and C). Both PAD4⁻/⁻ and WT diabetic mice displayed increased levels of blood glucose and decreased body weight gain when compared with age and sex matched healthy control animals. Interestingly, when assessing relative body weight changes per mouse, diabetic PAD4^−/− mice showed better maintenance of weight compared with diabetic WT mice (Figure 2C).

To investigate NETosis and inflammasome assembly in neutrophils under hyperglycaemic conditions, peripheral neutrophils from diabetic WT and PAD4^−/− mice were isolated and treated with ionomycin or vehicle for 4 h (Figure 2D and E). As expected,¹⁴ isolated neutrophils from diabetic WT mice were primed for NETosis compared with neutrophils isolated from healthy WT control mice (see Supplementary data online, Figure S3). This effect was completely obliterated by PAD4 deficiency with no NETosis in neutrophils from diabetic PAD4^−/− mice (see Supplementary data online, Figure S3). Shortly after stimulation, neutrophils of diabetic WT mice displayed markedly elevated ASC speck formation compared with healthy control animals (Figure 2D and E). In contrast, neutrophils from diabetic PAD4^−/− mice showed clearly reduced ASC speck formation upon stimulation compared with diabetic WT neutrophils, reaching levels even below those of age-matched healthy controls (Figure 2E). This shows that most ASC specks formed in stimulated WT neutrophils and indicates that PAD4 activity is required for the increased inflammasome assembly under diabetic conditions. To confirm that the observed ASC speck formation leads to functional inflammasome assembly, we measured downstream IL-1β secretion in the supernatant of 4 h cultivated WT and PAD4^−/− neutrophils from diabetic mice. Indeed, stimulated neutrophils from diabetic PAD4^−/− mice secreted lower amounts of IL-1β than stimulated neutrophils from diabetic WT mice (Figure 2F). In line with these in vitro findings, diabetic WT mice exhibited higher plasma levels of IL-1β and von Willebrand factor (VWF) compared with diabetic PAD4^−/− mice (Figure 2G and H).

We observed increased neutrophil-to-lymphocyte ratios (NLR), indicative of an enhanced inflammatory state in diabetic WT mice compared with diabetic PAD4^−/− mice (Figure 2I). Of note, NLR was higher in diabetic WT mice than in healthy controls, while NLR of diabetic PAD4^−/− mice was comparable with WT control mice (Figure 2I).

Given the increased propensity for neutrophil inflammasome assembly, NETosis, IL-1β, and VWF release, we hypothesized that PAD4 may contribute to diabetic organ damage and dysfunction.

Peptidylarginine deiminase 4 deficiency preserves cardiac function and prevents heart failure in streptozotocin-induced type 1 diabetes

To investigate the role and temporal dynamics of PAD4 in DCM and its impact on cardiac function, we performed serial echocardiography over the course of 8 weeks to assess structural and functional changes in diabetic WT and diabetic PAD4^−/− mice. We found that diabetic WT mice showed a robust decline in LVEF (43.8 ± 11.0%, 95% CI 35.3–52.3) (Figure 3A and B), consistent with literature on heart function in diabetic WT mice.³² By contrast, diabetic PAD4^−/− mice did not display any decline in LVEF (63.8 ± 5.5%, 95% CI 59.9–67.7), comparable with the LVEF of healthy control animals (Figure 3A and B). This was also true for fractional shortening (FS) (Figure 3C). In diastolic function, measured in pulse-wave and tissue Doppler images of the mitral valve flow velocities, a similar picture unfolded (Figure 3D–F; Supplementary data online, Figures S5 and S6 and Table S3). WT diabetic mice displayed a marked increase in E/e′ ratios, a significant increase in isovolumetric relaxation time (IVRT), a decline in E/A ratios and an increase in deceleration time (DT) when compared with diabetic PAD4^−/− mice (Figure 3D–F; Supplementary data online, Figures S5 and S6 and Table S3).

Figure 3.

Peptidylarginine deiminase 4 deficiency preserves cardiac function and limits adverse remodelling in streptozotocin-induced type 1 diabetes. (A) Representative M-mode images at end-systole and end-diastole 8 weeks after streptozotocin in wild-type and PAD4⁻/⁻ mice. (B and C) Systolic function over time: (B) ejection fraction and (C) fractional shortening at baseline, 2, 4, 6, and 8 weeks post-streptozotocin (n = 9–10 per genotype). (D) Representative pulsed-wave Doppler and tissue Doppler traces across the mitral valve at 8 weeks; E-wave, A-wave, and e′ indicated. (E and F) Diastolic function over time: (E) E/e′ ratio and (F) isovolumetric relaxation time (n = 9–10). (G) Representative left ventricular sections stained with Sirius Red/Fast Green at Week 8 (scale bar, 50 μm); arrows indicate fibrosis. (H) Quantification of fibrosis (% left ventricular area) at Week 8 (n = 6 per group). (I) Exercise capacity: distance run on a forced treadmill at baseline and 4 and 8 weeks (n = 9–10). (J) Plasma troponin-I at Week 8 (n = 8–9). (K) Lung wet/dry weight ratio at Week 8 in WT controls (n = 8) and streptozotocin-treated WT (n = 13) and PAD4⁻/⁻ mice (n = 12). (L) Heart weight/body weight at Week 8 in the same groups (n = 6–9). Area under the curve analyses of integrated echocardiographic parameters (B and C, E and F) and exercise capacity (I) are provided in Supplementary data online, Figures S6 and S7. Panels (B) and (C), (E) and (F), and (I) show data as mean ± SD; per-time-point group differences were analysed using unpaired t-test or Mann–Whitney U test. Panel (H) shows mean ± SD and was analysed using an unpaired t-test; panel (J) shows median (IQR) and was analysed by Mann–Whitney U test; panels (K) and (L) show mean ± SD and were analysed using ordinary one-way ANOVA. Exact P values are shown in the figure. Dia, diastole; EF, ejection fraction; E/e′, ratio of early mitral inflow velocity to early diastolic mitral annular velocity; FS, fractional shortening; IVRT, isovolumetric relaxation time; LV, left ventricle; PAD4, peptidylarginine deiminase 4; STZ, streptozotocin; Sys, systole; WT, wild type

We used Sirius Red/Fast Green staining to quantify collagen in heart tissue and assess interstitial cardiac fibrosis. Diabetic WT mice displayed increased cardiac fibrosis compared with diabetic PAD4^−/− mice (Figure 3G and H). Most interestingly, these changes resulted in the development of a drastic exercise intolerance in diabetic WT mice, whereas diabetic PAD4^−/− mice maintained robust exercise capacity throughout the entire 8-week period, as they were able to travel nearly four times farther than diabetic WT mice (Figure 3I). In line with the observed changes in systolic and diastolic heart function, diabetic WT mice showed increased plasma levels of Troponin I compared with diabetic PAD4^−/− mice (Figure 3J). The congestive HF not only resulted in earlier dropouts in exercise testing, but it also led to profound pulmonary oedema measured by increased lung wet-to-dry weight ratios (W/D ratio) (Figure 3K). Diabetic WT mice had severe pulmonary oedema compared with both diabetic PAD4^−/− mice and healthy controls, whereas PAD4 deficiency resulted in nearly normal W/D ratios (Figure 3K). This HF phenotype of diabetic WT mice was accompanied by cardiac hypertrophy with increased heart-to-body weight ratios compared with both diabetic PAD4^−/− mice and healthy controls (Figure 3L). Taking these results together, PAD4 deficiency strongly protected from DCM, diabetes-related decline in heart function, cardiac fibrosis and clinical signs of HF.

Peptidylarginine deiminase 4 drives cardiac fibrosis, NETosis, and thromboinflammation in diabetic cardiomyopathy

Cardiac inflammation, fibrosis, and apoptosis are hallmarks of STZ-induced DCM.⁹ Therefore, we set out to investigate the underlying structural tissue changes produced by diabetes to understand the cardioprotection observed in PAD4 deficiency. We stained LV sections of diabetic WT mice and diabetic PAD4^−/− mice for collagen type I (Col I) and platelet endothelial cell adhesion molecule 1 (PECAM-1/CD31) to evaluate extracellular matrix deposition and capillary density (Figure 4A). As expected, we could confirm fibrosis by increased Col I deposition in diabetic WT mice with large reticular strands of fibrotic tissue throughout the myocardium (Figure 4A), while diabetic PAD4^−/− mice displayed lower percentages of Col I-positive area in the LV (Figure 4A and B). To address whether the observed difference in fibrosis between diabetic WT and diabetic PAD4^−/− mice was associated with the extent of HF in the two genotypes, we correlated myocardial fibrosis and LVEF of all mice. Indeed, we observed a negative correlation (Pearson r =−.65, P = .03), between fibrosis and LVEF in DCM. Interestingly, we found lower capillary density in hearts of diabetic WT mice compared with diabetic PAD4^−/− mice (Figure 4A and C), indicative of vascular rarefaction in diabetes. We observed increased priming of circulating neutrophils in diabetic WT mice, with a heightened propensity to form ASC specks and NETs. This priming was absent in neutrophils from diabetic PAD4^−/− mice (Figure 2), which is why we measured myocardial neutrophil infiltration using flow cytometry (Figure 4D–F). Indeed, we found increased neutrophil infiltration into the myocardium of diabetic WT mice compared with diabetic PAD4^−/− mice (Figure 4E). Of note, overall fractions of myeloid cells were comparable between groups (Figure 4F).

Figure 4.

Peptidylarginine deiminase 4 deficiency reduces cardiac fibrosis, neutrophil infiltration, neutrophil extracellular trap burden, and inflammatory cytokines in streptozotocin-induced type 1 diabetes. (A) Representative immunofluorescence microscopy images of LV sections stained for collagen type I, vasculature (CD31), and nuclei (DAPI) at Week 8 post-streptozotocin (scale bar, 50 μm). (B) Quantification of collagen type I deposition (% left ventricular area) and (C) capillary density (% CD31⁺ of left ventricular area) in WT + STZ and PAD4⁻/⁻+ STZ mice (n = 6 per group). (D) Flow cytometry gating strategy for digested heart tissue: CD45⁺/CD11b⁺ double-positive events defined myeloid cells; Ly6G⁺ within this population defined neutrophils. (E) Quantification of cardiac neutrophils (Ly6G⁺ of CD45⁺/CD11b⁺) and (F) cardiac myeloid cells (CD11b⁺ of CD45⁺) at Week 8 post-streptozotocin (n = 6). (G) Representative immunofluorescence images (60×) of left ventricular sections from WT + STZ and PAD4⁻/⁻+STZ mice stained for Ly6G, H3Cit, and nuclei (DAPI); extracellular H3Cit adjacent to Ly6G⁺ cells was classified as neutrophil extracellular traps. (H) Quantification of total extracellular H3Cit⁺ area fraction (% left ventricular section) (n = 6). (I) IL-1β concentrations and (J) TGF-β1 concentrations in heart tissue at Week 8 post-streptozotocin, determined by ELISA in WT controls (n = 6), WT + STZ (n = 9), and PAD4⁻/⁻+ STZ (n = 6). Panels (B) and (C) show data as mean ± SD and were analysed using unpaired t-tests; panels (E), (F), (H), and (J) show data as median (IQR) and were analysed using the Mann–Whitney U test; panel (I) shows data as mean ± SD and was analysed using ordinary one-way ANOVA. Exact P values are shown in the figure. CD, cluster of differentiation; CD31, platelet endothelial cell adhesion molecule-1; DAPI, 4′,6-diamidino-2-phenylindole; H3Cit, citrullinated histone 3; IL-1β, interleukin-1β; LV, left ventricle; Ly6G, lymphocyte antigen 6 complex locus G6D; NET, neutrophil extracellular trap; PAD4, peptidylarginine deiminase 4; STZ, streptozotocin; TGF-β1, transforming growth factor-β1; WT, wild type

Since PAD4 is the major cellular regulator of NETosis and we observed increased myocardial NET burden in diabetic HF patients, we stained LV heart sections for H3Cit and Ly6G (Figure 4G). We found increased levels of extracellularly deposited H3Cit next to Ly6G⁺ cells (Figure 4H), indicative of increased intracardiac NET release in diabetic WT mice and—as expected—only traces of H3Cit staining in diabetic PAD4^−/− mice (Figure 4G and H). A major effector molecule of the proposed PAD4/inflammasome axis in neutrophils is the chemokine IL-1β. To examine whether the observed increase in neutrophil priming and infiltration also leads to PAD4/NLRP3 mediated inflammation in the myocardium, we measured tissue levels of IL-1β. Indeed, cardiac IL-1β levels were elevated in diabetic WT mice compared with healthy controls (Figure 4I). In contrast, IL-1β levels in the myocardium of diabetic PAD4^−/− mice remained low and were comparable with those of healthy age- and sex-matched controls (Figure 4I). This suggests that secreted, profibrotic IL-1β in the hearts of diabetic mice heavily relies on PAD4 activity. Neutrophil extracellular traps and IL-1β are well known to fuel thromboinflammation by activating the endothelium and promoting platelet adhesion. Here, immunofluorescence staining for VWF and platelet marker CD42b revealed increased platelet presence and VWF deposition in the hearts of diabetic WT mice compared with diabetic PAD4^−/− mice (see Supplementary data online, Figure S4). Our group has previously shown that PAD4 promotes age-related cardiac fibrosis,²⁰ potentially through NET-mediated recruitment of platelets and subsequent release of profibrotic mediators such as TGF-β.³³ Consistently, we observed lower levels of TGF-β1 in the hearts of diabetic PAD4^−/− mice when compared with diabetic WT mice (Figure 4J). This indicates that PAD4 is a significant regulator of TGF-β1 release in diabetic myocardium as well, providing valuable insights into the pathophysiology of fibrosis in DCM.

Peptidylarginine deiminase 4 promotes albuminuria, renal fibrosis, and NETosis in kidneys of diabetic mice

Diabetic kidney disease is the leading cause of CKD worldwide.³⁴ Chronic kidney disease, in turn, is an independent risk factor for cardiovascular disease, HF, and all-cause mortality.^34,35 Hence, we set out to delve into the pathobiological changes in the kidneys of our diabetic mice. To assess renal injury, we measured urine albumin in diabetic WT and PAD4^−/− mice (Figure 5A). Albuminuria was elevated in diabetic WT mice compared with PAD4⁻/⁻ mice (Figure 5A), pointing towards glomerular damage and suggesting the protection provided by PAD4 deficiency in the heart may extend to the kidneys as well. In addition, diabetic WT mice exhibited an increased kidney weight to body weight ratio compared with PAD4⁻/⁻ mice, consistent with renal hypertrophy (Figure 5B). Since fibrosis was a key feature of the observed myocardial changes and renal fibrosis is a major hallmark of DKD,⁶ we stained kidney sections for Sirius Red/Fast Green to quantify renal collagen deposition (Figure 5C). Accordingly, diabetic WT mice exhibited higher interstitial renal fibrosis compared with diabetic PAD4^−/− mice (Figure 5C and D). Interestingly, the increase in renal fibrosis was associated with elevated CD68⁺ immune cell infiltration, indicative of enhanced monocyte/macrophage recruitment (Figure 5E and F), as well as increased expression of the profibrotic cytokine TGF-β1 in diabetic WT compared with diabetic PAD4^−/− mice (Figure 5G). In addition, kidney sections stained for H3Cit revealed pronounced NET deposition in diabetic WT kidneys, whereas only minimal signal was detectable in diabetic PAD4⁻/⁻ kidneys (Figure 5H and I). Collectively, these findings offer evidence that the anti-fibrotic and anti-inflammatory effects of PAD4 deficiency in the diabetic heart also manifest in the kidney, supporting a PAD4-mediated cardiorenal axis of injury in T1D.

Figure 5.

Peptidylarginine deiminase 4 deficiency reduces renal injury, fibrosis, inflammation, and neutrophil extracellular trap burden in streptozotocin-induced type 1 diabetes. (A) Urinary albumin concentrations (μg/mL) at Week 8 post-streptozotocin, measured by ELISA, in WT + STZ (n = 9) and PAD4⁻/⁻+ STZ mice (n = 8). (B) Kidney weight/body weight (mg/g) at Week 8 in WT controls (n = 6), WT + STZ (n = 9), and PAD4⁻/⁻+ STZ mice (n = 8). (C) Representative kidney sections stained with Sirius Red/Fast Green at Week 8 (scale bar, 50 μm); arrows indicate fibrotic tissue. (D) Quantification of renal fibrosis (% Sirius Red⁺ area) (n = 5 per group). (E) Representative kidney sections immunostained for CD68, TGF-β, and nuclei (DAPI) at Week 8 (scale bar, 40 μm). (F) Quantification of CD68⁺ cells (mm⁻²) in kidney sections (n = 5). (G) TGF-β1 concentrations in kidney tissue at Week 8, measured by ELISA. (H) Representative immunofluorescence microscopy images (60×) of kidney sections immunostained for Ly6G, H3Cit, and nuclei (DAPI); extracellular H3Cit deposits were considered neutrophil extracellular traps. (I) Quantification of total extracellular H3Cit⁺ area fraction (% of kidney section) (n = 5). Panels (A) and (I) show data as median (IQR) and were analysed using the Mann–Whitney U test; panel (B) shows data as mean ± SD and was analysed using Welch’s ANOVA (unequal variances); panels (D), (F), and (G) show data as mean ± SD and were analysed using unpaired t-tests. Exact P values are shown in the figure. CD, cluster of differentiation; CD68, cluster of differentiation 68; DAPI, 4′,6-diamidino-2-phenylindole; H3Cit, citrullinated histone 3; IL-1β, interleukin-1β; Ly6G, lymphocyte antigen 6 complex locus G6D; NET, neutrophil extracellular trap; PAD4, peptidylarginine deiminase 4; STZ, streptozotocin; TGF-β1, transforming growth factor-β1; WT, wild type

Discussion

The global prevalence of diabetes was estimated to exceed 500 million individuals worldwide in 2021 and is expected to rise to more than 750 million by 2045.¹ Common diabetes complications are kidney disease, observed in ~40% of patients, HF affecting approximately 30% of patients, altered wound healing and many more—all drastically increasing morbidity and mortality.^1–5,14 The underlying immunomodulatory mechanisms driving diabetic cardiorenal disease are ill defined and therapeutic targets scarce, underscoring the urgent need for further research in this field.^8–10 In this study, we addressed the PAD4–inflammasome axis in the development of cardiorenal disease in murine T1D. We identify at least five functionally relevant pathways depending on PAD4 activity: neutrophil priming for NLRP3 inflammasome assembly, neutrophil IL-1β secretion, increased myocardial and renal NET deposition, increased VWF levels indicative of thromboinflammation, and elevated TGF-β1 with consecutive organ fibrosis. In parallel, we could extend these observations to patients with HF by demonstrating an increased intracardiac NET burden in diabetic versus nondiabetic HF and by linking hyperglycaemia directly to neutrophil inflammasome activation and NETosis in humans.

Diabetic WT mice developed an expected diabetes-related decline in heart-function over the course of the study.³² Intriguingly, diabetic PAD4^−/− mice maintained their heart function over the study period comparable with healthy control mice both for systolic (LVEF) as well as diastolic (E/e′ ratio) parameters. This is in keeping with our previous findings that show PAD4 prevents the development of hypertrophic cardiomyopathy and HF in collagen-induced arthritis and obesity.^18,19 Together, these data suggest that PAD4 activation is critical for signalling pathways involved in pathological myocardial remodelling. Among PAD isoforms, PAD4 is unique in being predominantly expressed in granulocytes^36,37 and localized to the nucleus due to its nuclear localization signal.³⁸ This allows PAD4 to directly modify nuclear chromatin structure, enabling the large-scale decondensation required for NET release.³⁹

We have described that PAD4 levels are elevated in diabetes¹⁴ and more recently that it supports NLRP3 Inflammasome assembly in neutrophils.²³ While the upregulation of the NLRP3 inflammasome in monocytes and macrophages of diabetic patients and mouse models is well established, its role and assembly within neutrophils under diabetic conditions remain unexplored.^26,27,40 We hypothesized that overactivation of PAD4 in diabetes would lead to increased inflammasome assembly in neutrophils. Indeed, our results show that neutrophils from diabetic mice are primed for inflammasome assembly and activity, as indicated by increased polymerization of the adaptor protein ASC (speck formation) and IL-1β secretion following stimulation. Consistently, exposure of human neutrophils to high glucose increased ASC speck formation and NETosis alike, with released ASC specks frequently embedded within NET structures, linking hyperglycaemia directly to inflammasome-driven NET release. This excessive release of polymerized ASC with NETs represents a self-amplifying loop,²³ as inflammasome components can be transferred to other cells in a prion-like manner to propagate inflammation.^41,42

Correspondingly, increased levels of NLRP3 inflammasome and NET components have also been observed in diabetic patients.^{15,27,43–45} Both NETs and activated neutrophils have been described to drive pathological myocardial fibrosis in mice and humans.^46–48 Indeed, excessive deposition of interstitial collagen I was absent in diabetic PAD4^−/− mice in contrast to the diabetic WT mice. The extent to which PAD4 deficiency kept not only diastolic but also systolic LV function normal in our model may suggest the involvement of additional mechanisms than averted cardiac fibrosis. Recent studies highlight hyperglycaemia-induced cardiomyocyte pyroptosis as an important cellular process in DCM, a pathway the NLRP3 inflammasome is profoundly involved in through its pore-forming downstream effector enzyme gasdermin D.⁴⁹ Considering our results showing an effect of PAD4 on inflammasome activation and reduced levels of cardiac cell damage markers in PAD4 deficient diabetic mice, it is reasonable to assume that the PAD4–inflammasome axis is a key factor in the pathobiology of cardiac cell injury in T1D. While PAD4 is most abundantly expressed in neutrophils,^36,37 its relatively low expression in monocytes and macrophages (where other PAD isoenzymes like PAD2 predominate⁵⁰) may increase in the diseased myocardium, potentially also contributing to the observed upregulation of inflammatory pathways.⁵¹

Our echocardiographic findings align with recent evidence in T1D patients, indicating that LV diastolic dysfunction already occurs at an early stage of DCM development.^52–54 Increasing evidence suggests that the metabolic signatures of DCM evolve in a time- and stage-dependent manner.⁵⁵ Interestingly, PAD4^−/− mice displayed time-dependent changes in LV-stiffness marker IVRT giving rise to the idea of potential metabolic and functional adaptations of cardiomyocytes to hyperglycaemia in the absence of PAD4.

Furthermore, the normal systolic and diastolic LV function observed in diabetic PAD4^−/− mice was accompanied by significantly improved exercise tolerance as early as 4 weeks post-diabetes induction, along with reduced pulmonary congestion. In the kidney, PAD4 deficiency was likewise protective: albuminuria increased in diabetic WT but not in PAD4^−/− mice, and histology demonstrated reduced interstitial fibrosis, lower TGF-β1 expression, and markedly less NET deposition in PAD4^−/− kidneys. These renal changes mirror the myocardial phenotype (absent NET deposition, and decreased fibrosis) and together support a PAD4-dependent cardiorenal axis of injury. Excessive interstitial fibrosis is a key feature of DKD and DCM, driving loss of organ function^6,56 and commonly accompanied by reduced capillary density,^57–59 which we observed in hearts of diabetic WT mice compared with diabetic PAD4^−/− mice.

Our results reveal excessive NET deposition in the hearts and kidneys of diabetic WT mice, likely contributing to the observed organ dysfunction and adverse remodelling similar to previous reports in MI.^16,17,60 Importantly, NETs have recently been identified in the myocardium of chronic HF and dilated cardiomyopathy patients,^21,22 shifting the paradigm towards chronic NET-driven myocardial inflammation. In alignment, our analysis of EMB confirms the presence of NETs across different HF phenotypes, with a higher fraction of NETting neutrophils in diabetic compared with nondiabetic HF in humans. Neutrophil extracellular trap burden correlated inversely with LVEF, and focal ‘hotspots’ of NET accumulation suggest localized inflammatory niches within the failing myocardium. Diabetes emerges as an amplifier of this process, consistent with our observation that high glucose not only augments NETosis,¹⁴ but also inflammasome assembly.

While historically associated with autoimmune kidney diseases such as ANCA vasculitis and lupus nephritis,^61–63 NETs have now also been implicated in DKD, where they promote glomerular endothelial injury and pyroptosis, as recently shown in both patients and diabetic mouse models.⁶⁴ Consistent with these reports, we observed abundant NET deposition and increased immune cell infiltration in diabetic WT kidneys—both of which were profoundly reduced in PAD4-deficient mice—suggesting that PAD4-mediated NET release actively contributes to DKD pathogenesis. Mechanistically, NETs and their components—such as MPO and histones—are known to activate fibroblasts and drive their differentiation into myofibroblasts.⁶⁵ Moreover, NET-associated proteins retain extracellular activity capable of trapping, cleaving, and activating latent TGF-β, a central mediator of fibrosis.^66,67 Notably, in the present study, TGF-β1 levels were significantly elevated in both heart and kidney tissues of diabetic WT mice compared with PAD4-deficient mice, suggesting that NETs may promote cardiac and renal fibrosis through TGF-β1–dependent pathways. In parallel, activated platelets, efficiently recruited by NETs and thus abundant at sites of NET deposition, could serve as the major reservoir of secreted TGF-β1, further amplifying this profibrotic signalling in the presence of NETs and PAD4 activity.⁶⁸

The NLRP3-inflammasome has also been demonstrated to be a key player in pathogenic fibrosis by facilitating the maturation of its profibrotic effector enzyme IL-1β.⁶⁹ Indeed, we found reduced levels of IL-1β in heart tissues of diabetic PAD4^−/− mice, when compared with diabetic WT mice. IL-1β is known not only to promote fibrosis, but also to enhance leukocyte recruitment by upregulating adhesion molecules such as ICAM-1 on endothelial cells.⁷⁰ Neutrophils could well be the primary source of IL-1β in this context, as we observed increased ASC speck formation and IL-1β release from isolated neutrophils of diabetic WT mice, but not PAD4^−/− mice, similar to our previous report of neutrophil-derived IL-1β in early MI.¹⁷ The data we present in this study reinforce the functional interplay of PAD4 and the inflammasome in promoting IL-1β release and potentially tissue fibrosis via TGF-β1.

Endothelial activation with elevated VWF release has been described as another major player in the inception and progression of fibrosis through promotion of immune cell recruitment.⁷¹ Excitingly, VWF has recently been described to exacerbate HF in human patients⁷² and mice¹⁶ alike through the formation and retainment of NETs. Treatment with NET inhibitors or rhADAMTS13 (the enzyme that cleaves VWF) was able to improve cardiac outcomes as conceptualized.⁷² We have recently established that the neutrophil NLRP3 inflammasome participates in VWF release in myocardial infarction.¹⁷ Here, we consequently show reduced levels of VWF in plasma and myocardium of PAD4 deficient diabetic mice.

Consistent with our previous findings that PAD4 promotes NLRP3 inflammasome assembly²³ and that NLRP3 directs neutrophil chemotaxis,⁷³ we show reduced neutrophil infiltration in cardiac tissues of diabetic PAD4^−/− mice compared with diabetic WT mice. However, reduced immune cell infiltration in tissue seems not to be limited to neutrophils alone, since we also show lower CD68 signal in kidney sections of diabetic PAD4^−/− mice, reflective of reduced monocyte/macrophage infiltration. In both patients and animal models of DKD, early monocyte recruitment is a hallmark of disease progression, driven by upregulation of adhesion molecules such as ICAM-1 and chemokines like MCP-1 in response to hyperglycaemia.⁷⁴ Neutrophils can amplify this recruitment by releasing IL-1β and NETs, which enhances endothelial adhesion molecule expression and facilitates monocyte infiltration.⁷⁵

Taken together, our study demonstrates that PAD4 drives a multifaceted inflammatory programme in diabetes that integrates neutrophil priming, inflammasome activation, NET formation, thromboinflammatory signalling with elevated VWF, and TGF-β1–mediated fibrosis across cardiac and renal tissues. The convergence of mouse and human data—higher intracardiac NET burden in diabetic HF, localized NET ‘hotspots’, and glucose-primed inflammasome/NETosis in human neutrophils—provides a translational framework linking metabolic stress to chronic tissue inflammation and fibrosis. These results highlight PAD4 not only as a central regulator of immune-mediated injury but as a promising therapeutic target for preventing chronic organ dysfunction in diabetic cardiorenal disease.

Conclusion

In summary, we provide compelling evidence that the PAD4–inflammasome axis drives profibrotic remodelling in the heart and kidney in diabetes-related cardiorenal disease. Our findings reveal an increased propensity of neutrophils in diabetes to assemble the inflammasome and release NETs and IL-1β. In our T1D model, PAD4 deficiency reduced myocardial and renal fibrosis, tissue IL-1β and TGF-β1 levels, NET deposition, and the endothelial activation marker VWF. Importantly, human HF patients with diabetes displayed increased myocardial NET burden. These data position the neutrophil PAD4–inflammasome axis as a promising therapeutic target to mitigate chronic inflammation and adverse remodelling in diabetic heart and kidney disease.

Limitations

Through the various crosstalk mechanisms between pathways that promote NETosis, IL-1β, and NLRP3 inflammasome assembly, it cannot be differentiated whether the observed effects of PAD4 deficiency are due to impaired NET release, post-translational protein modification, transcriptional changes, or altered metabolic inflammation. Moreover, emerging evidence suggests that PAD4 knockout may elicit beneficial compensatory mechanisms that promote cardiac regeneration, underscoring the need for further mechanistic studies to disentangle these pathways.

Specifically, the observed changes in hearts and kidneys presented at an early stage of murine T1D and we cannot rule out diverging patterns of heart and kidney function as the disease progresses further. Future studies are needed to address later stages of murine DCM and DKD. Peptidylarginine deiminase 4 is also involved in chromatin remodelling and gene regulation beyond neutrophils, and although neutrophils are the predominant source, relatively low PAD4 expression in other cells may contribute under inflammatory conditions. This underscores the need for long-term and cell type–specific studies to evaluate the safety, specificity, and potential off-target effects of PAD4-targeted therapies in clinical settings. Importantly, in ageing mice, lifelong absence of PAD4 preserves cardiac function without evidence of adverse effects.

Finally, potential confounding cannot be fully excluded. Differences in glycaemic control, systemic inflammation, or HF severity (particularly in human biopsies and diabetic mice) may have influenced some associations. Nonetheless, the consistency of results across models and endpoints makes major confounding unlikely.

Supplementary Material

Supplementary data are available at European Heart Journal online.

Key Question.

What role do neutrophil extracellular traps (NETs) and peptidylarginine deiminase 4 (PAD4) play in driving neutrophil-mediated inflammation, fibrosis, and organ dysfunction in diabetes, and can their inhibition prevent heart and kidney complications?

Key Finding.

Diabetes increased myocardial NET burden in heart failure (HF) patients and high glucose triggered neutrophil inflammasome activation. In diabetic mice, PAD4 promoted neutrophil inflammasome activation and NETosis, driving IL-1β and TGF-β–mediated inflammation, fibrosis, and organ dysfunction. PAD4 deficiency prevented HF and preserved kidney function despite persistent hyperglycaemia.

Take Home Message.

NETs and PAD4 connect hyperglycemia to neutrophil-driven inflammation and fibrosis in diabetic heart and kidney disease. Its inhibition may provide a novel therapeutic strategy for preventing cardiorenal complications in diabetes.

Translational perspective.

Diabetic cardiomyopathy and kidney disease remain major causes of morbidity in patients with diabetes. Myocardial NET deposition was elevated in diabetic heart failure patients and correlated with systolic impairment. Our study identifies PAD4 as a central regulator of neutrophil inflammasome activation and NETosis, linking systemic inflammation in diabetes to fibrosis and organ dysfunction. Peptidylarginine deiminase 4 deficiency preserved cardiac and renal structure and function in a murine model of type 1 diabetes. These findings highlight PAD4 as a promising therapeutic target to prevent heart and kidney failure in patients with diabetes.

Data Availability

The data supporting the findings of this study are available from the corresponding authors upon reasonable request.