Inhibition of Interleukin-1 Ameliorates Cardiac Dysfunction and Arrhythmias in a Murine Model of Kawasaki Disease

Abstract

BACKGROUND:

Kawasaki disease (KD) is an acute febrile illness and systemic vasculitis often associated with cardiac sequelae, including arrhythmias. Abundant evidence indicates a central role for interleukin (IL-1) and tumor necrosis factor-α (TNFα) signaling in the formation of arterial lesions in KD. We aimed to investigate the mechanisms underlying the development of electrophysiological abnormalities in a murine model of KD vasculitis.

METHODS:

Lactobacillus casei cell wall extract (LCWE)-induced KD vasculitis model was used to investigate the therapeutic efficacy of clinically-relevant IL-1 receptor antagonist (IL-1Ra) and TNFα neutralization. Echocardiography, in vivo electrophysiology, whole-heart optical mapping, and imaging were performed.

RESULTS:

KD vasculitis was associated with impaired ejection fraction, increased ventricular tachycardia, prolonged repolarization, and slowed conduction velocity. Since our transcriptomic analysis of human patients showed elevated levels of both IL-1β and TNFα, we asked whether either cytokine was linked to the development of myocardial dysfunction. Remarkably, only inhibition of IL-1 signaling by IL-1Ra but not TNFα neutralization was able to prevent changes in ejection fraction and arrhythmias, whereas both IL-1Ra and TNFα neutralization significantly improved vasculitis and heart vessel inflammation. The treatment of LCWE-injected mice with IL-1Ra also restored conduction velocity and improved the organization of connexin 43 at the intercalated disk. In contrast, in mice with gain-of-function of the IL-1 signaling pathway, LCWE induced spontaneous ventricular tachycardia and premature deaths.

CONCLUSIONS:

Our results characterize the electrophysiological abnormalities associated with LCWE-induced KD, and show that IL-1Ra is more effective in preventing KD-induced myocardial dysfunction and arrhythmias than anti-TNFα therapy. These findings support the advancement of clinical trials using IL-1Ra in KD patients.

Graphical Abstract

INTRODUCTION

Kawasaki disease (KD) is an acute febrile and systemic vasculitis that predominantly affects infants and young children; if untreated, 25% of patients develop coronary artery aneurysms (CAA) ¹. Although treatment with high doses of intravenous immunoglobulin (IVIG) improves the clinical outcome ^2,3, approximately 20% of KD patients are not responsive to this conventional intervention, and are at increased risk of developing long-term irreversible cardiovascular sequelae ^2,4. While the most important cardiovascular complications of KD are coronary artery lesions, such as CAA, and myocarditis, cardiac arrhythmia has also been recognized as a life-threatening manifestation of this disease ^5–7. Indeed, 40 to 80% of children who develop acute KD show electrocardiogram (ECG) abnormalities ^8–10, including ventricular arrhythmias, which are increasingly being recognized as a cause of death in children with KD ^7,11,12.

Among other cardiac manifestations, life-threatening cardiac arrhythmias, including ventricular tachycardia (VT) and atrioventricular block, have been reported, with particularly high predominance in KD patients with reduced left ventricular function ^5,6,13. However, the mechanisms underlying arrhythmias in KD have yet to be investigated.

Increased infiltration of immune cells ^14–16 and elevated circulating levels of proinflammatory cytokines, such as interleukin-1β (IL-1β) ^17,18 and tumor necrosis factor-α (TNF-α) ¹⁹ have been consistently reported in human biopsies and experimental murine models of KD ^20,21. Our group has contributed to the mechanistic understanding of the role of these cytokines in promoting vascular inflammation and the development of CAA ^20–25. The involvement of proinflammatory cytokines in the acute phase of KD suggests that blocking IL-1 signaling or neutralizing TNFα may have therapeutic efficacy in preventing vasculitis and heart vessel inflammation ^22,25,26. Recombinant human IL-1Ra, also known as anakinra, competitively inhibits the binding of IL-1α and IL-1β to IL-1R1. Blocking IL-1 signaling with IL-1Ra is beneficial to treat a broad spectrum of inflammatory conditions ^27,28, and a phase II open-label study indicates that anakinra is safe and has some efficiency in reducing KD clinical manifestations in IVIG-resistant KD patients ²⁹. Similarly, clinical efficacy, including shorter duration of fever, was observed when monoclonal antibodies against TNFα were used in KD patients for intensification of primary therapy or in those with an IVIG-refractory course ^30,31. Although accumulating evidence indicates that elevated cardiac proinflammatory cytokines predispose to arrhythmias in the context of heart failure and myocarditis ^32,33, the role of these factors in KD-associated arrhythmia is unknown.

Here, we addressed this question using the Lactobacillus casei cell wall extract (LCWE)-induced KD vasculitis mouse model, which closely phenocopies the immune and histopathological features of human KD vasculitis, such as coronary arteritis, aortitis, myocarditis, and aneurysms ^22,25,34. We have previously shown that aberrant IL-1 signaling drives ECG abnormalities in LCWE-injected mice developing vasculitis ³⁵. Here, we significantly build on that finding to demonstrate impairment of left ventricular ejection fraction and dilated cardiomyopathy in the LCWE-induced KD vasculitis murine model. We also sought to determine arrhythmic substrates and investigate the therapeutic potential of blocking IL-1 and TNF pathways on KD vasculitis-induced arrhythmogenesis. We found increased VT associated with prolonged repolarization and slowed conduction velocity in LCWE-injected mice. Underlying abnormalities included slowed electrical impulse propagation consistent with lateralization of connexin 43. Remarkably, we show that while inhibition of either IL-1 signaling or TNFα significantly improved vasculitis lesions, only inhibition of IL-1 signaling, but not TNFα neutralization, prevented the structural, functional, and electrophysiological remodeling in heart tissues of LCWE-injected mice, thereby resulting in lower VT incidence. Overall, our data emphasize the role of IL-1β in the arrhythmogenic and ventricular dysfunction in LCWE-induced KD vasculitis, add mechanistic insights and may further broaden the potential benefits of anti-IL-1 therapies for KD vasculitis patients.

METHODS

The authors declare that all supporting data are available within the article and its online supplementary files.

Mice

C57BL/6J (stock 000664) and Il1rn^−/− (B6.129S-Il1^rntm1Dih/J, stock 004754) mice were obtained from the Jackson Laboratory. Only male animals at 5 weeks were used in this study as LCWE injection induces stronger and more consistent coronary vasculitis lesions and abdominal aorta aneurysms in male mice than in female mice ^23,25. WT mice were randomly assigned to each experimental group. Mice were housed under specific pathogen-free conditions and used according to the guidelines of the Cedars-Sinai Medical Center institutional committee. All animal studies were approved by the Institutional Animal Care and Use Committee at Cedars-Sinai Medical Center and performed following the Guide for the Care and Use of Laboratory Animals published by the U.S. National Institutes of Health.

Analysis of human gene expression datasets.

The publicly available gene expression datasets GSE68004 ³⁶, GSE73461 ³⁷, and GSE63881 ³⁸ were obtained from NCBI Gene Expression Omnibus. Transcriptomic data analysis was run by GEO2R software as part of the GEO database, and summary statistics were generated with the limma topTable function. The expression of IL1B and TNF was assessed in whole blood of acute KD patients and healthy control (HC) patients (n=37 HC and n=76 acute KD patients for GSE68004 ³⁶; n=55 HC and n=77 acute KD patients for GSE73461 ³⁷. The expression of IL1B and TNF was also analyzed in another dataset generated from whole blood of acute KD patients (n=146) and IVIG-treated convalescent KD patients (n=145) based on a Benjamini-Hochberg adjusted p-value <0.05 (GSE63881) ³⁸. Heatmaps showing the expression relative to the mean of selected genes were generated in R with the “ComplexHeatmap” package.

LCWE-induced KD vasculitis murine model

Lactobacillus casei (LCWE, ATCC #11578) was prepared as previously described ^20,22–24. Briefly, L. casei was grown in Lactobacillus de Man, Rogosa, and Sharpe broth (Difco) for 48 hours, harvested, and washed with PBS. The harvested bacteria were disrupted by 2 packed volumes of 4% SDS/PBS overnight. Cell wall fragments were washed 8 times with PBS to remove any residual SDS. The SDS-treated cell wall fragment was sonicated for 2 hours with a ¾-in horn and a garnet tip at maximum power. During sonication, the cell wall fragments were maintained by cooling in a dry ice/ethanol bath. After sonication, the cell wall fragments were spun for 20 minutes at 12 000 rpm and 4°C. The supernatant was centrifuged for 1 hour at 38 000 rpm and 4°C, and the pellet was discarded. The total rhamnose content of the cell wall extract was determined by a colorimetric phenol-sulfuric assay as described previously. 5-week-old male mice were injected i.p. with either 400 μg LCWE or 400μl PBS and euthanized two weeks post-injection. To inhibit IL-1 signaling, IL-1Ra (anakinra; 400μg per injection) was injected i.p. daily starting the day before LCWE injection until day 5 post-LCWE. To neutralize TNFα, a monoclonal antibody against TNFα (BioXcell; Clone XT3.11; 500 μg per injection) or isotype IgG control (BioXcell; BE0088; 500 μg per injection) was injected i.p. into WT mice every 3 days starting 1 day prior to LCWE injection until day 5 post-LCWE. Echocardiograms were performed on day 14 post-LCWE injection. Afterward, mice were sacrificed and heart tissues were harvested and embedded in Tissue-Tek Optimum Cutting Temperature (O.C.T.) compound (Sakura Finetek, catalog 4583). Serial cryosections (7μm) of heart tissues were stained with hematoxylin and eosin (H&E; MilliporeSigma, catalog MHS32). Heart tissue histopathological examinations of the aortic root for the severity of cardiovascular lesions (coronary arteritis and aortic root vasculitis/aortitis) were performed on H&E-stained tissue sections and given a heart vessel severity score by an expert pathologist blinded to the experimental groups, as we previously described ²². Briefly, coronary arteries and aorta (aortitis) lesions were each assessed for acute inflammation, and chronic inflammation with connective tissue proliferation using the following scoring system: 0 = no inflammation, 1 = rare inflammatory cells, 2 = scattered inflammatory cells, 3 = diffuse infiltrate of inflammatory cells, and 4 = dense clusters of inflammatory cells. All 4 scores (coronaries acute, coronaries chronic, aorta acute, aorta chronic) were combined to generate a severity score called “heart vessel inflammation score,” as previously published ²².

Transthoracic echocardiogram

Transthoracic echocardiography was performed using a VisualSonics Vevo 3100 system equipped with a MX550D 40MHz transducer (Visual Sonics) ^39–41. Mice were anesthetized with 4% isoflurane (induction) and maintained with 2–1% isoflurane on a body-temperature-controlled pad. Two-dimensional parasternal short-axis images of the LV were obtained using M-mode scans at the midventricular papillary muscle level. Images were digitally stored in cine loops consisting of 300 frames. Echocardiogram acquisition was initiated when the heart rate ranged between 420 to 500 beats per min. ECG monitoring was obtained using limb electrodes. Animals were tested in a randomized fashion. All the image analyses were conducted by an independent trained observer who was blinded to the experimental groups. LV dimensions and EF were obtained on 4 or more consecutive cardiac cycles by tracing the left ventricular free walls and averaged for each mouse. No data were excluded.

Ambulatory telemetry studies

Mice were implanted with an intraperitoneal telemetric ECG transmitter (ETA-F10, Data Sciences International) and electrodes were arranged subcutaneously in a lead II configuration ^40,41. After a postoperative recovery period of 7 days, 24 h continuous recording was performed before administration of LCWE (baseline). After the LCWE injection, animals were continuously monitored for 15 consecutive days. Telemetry data were recorded and analyzed using Ponemah software (Data Sciences International). VT was defined as (1) 6 or more consecutive broad QRS complexes of ventricular origin at a rate of >700 beats per minute, (2) absence of the intrinsic QRS complex, and (3) dissociated or indistinguishable atrial activity. For the analysis of PVC burden, the number of extrasystoles with characteristic wide QRS complexes of ventricular origin was counted during the day-night cycle.

Programmed electrical stimulation

Mice were anesthetized with an isoflurane and oxygen mixture and positioned on a temperature-regulated operating table. A 1.1F octapolar electrophysiology catheter (Millar, EPR-800) was used to deliver rectangular stimulus pulses of twice the pacing threshold using an electronic stimulator (PowerLab, AD Instruments). Ventricular tachycardia (VT) was induced by a train of 10 stimuli (S1, 100-ms interval) followed by a short coupled extra stimulation (S2) at a coupling interval from 70 ms with 2-ms decrements until the effective refractory period. Up to 3 extra stimuli (S2, S3, and S4) were used ⁴¹. VT was defined as more than 6 consecutive beats of ventricular origin.

High-resolution optical mapping

Mice were anesthetized with isoflurane and injected with heparin (100 U) before euthanasia. Hearts were rapidly excised and washed in oxygenated ice-cold Tyrode’s solution (in mmol/L): 135 NaCl, 5.4 KCl, 0.33 NaH₂PO₄, 1 MgCl₂.6H₂O, 1.3 CaCl2, 10 HEPES, and 10 glucose (pH 7.4 NaOH). The ascending aorta was cannulated and retrogradely perfused in a Langendorff apparatus with Tyrode solution (2–4 mL/min at 37^oC). Hearts were immobilized with blebbistatin to avoid motion artifacts (10 μmol/L; Sigma-Aldrich, #B0560). After 10 min, hearts were perfused with the voltage-sensitive dye RH237 (10 μmol/L; Invitrogen #S1109) for 10 min. Hearts were then washed in dye-free Tyrode solution and imaged using a MiCAM05 Ultima-L CMOS camera (100 × 100-pixel CMOS sensor, 10 × 10 mm) (SciMedia) with filters (excitation 520/35 nm, dichroic 560 nm, emission 715 nm long-pass). A bipolar electrode was positioned on the base of the left ventricular epicardium for pacing, which was performed at a basic cycle length of 140 ms using a 2 ms pulse at twice the diastolic threshold (PowerLab, AD Instruments) ⁴¹. Optical mapping data were analyzed using Rhythm for MATLAB software ⁴². Cardiac electrograms were recorded using three electrodes, immersed in the bath in close proximity to reduce electrical impedance.

Blood Pressure measurements

Blood pressure was assessed by a non-invasive tail-cuff system dedicated to rodents (BP-2000 series II, Visitech system). Animals were tested in a randomized fashion. A week before measurements, animals were acclimated to the tail-cuff plethysmography system to minimize the impact of the procedural stress. During experiments, all animals underwent a 15 min acclimation period on a heating pad before blood measurements. At least five successful measurements were averaged from each mouse. All recordings were made in a quiet room between 10 a.m. and 14 p.m. by an experienced investigator blinded to the experimental groups. No data were excluded.

Immunohistochemistry

Tissues were embedded in an OCT compound and frozen in 2-methylbutane precooled in liquid nitrogen, then stored at −80°C until sectioning. Serial sections of the heart were cut at the mid-papillary level in the transverse plane ⁴¹. All sections were cut to between 7 and 8 μm using a cryostat (CM3050S, Leica) and adhered to super frost microscope slides. Cryosections were fixed with 4% PFA solution (Fisher Scientific, AAJ19943K2) for 10 minutes, washed with PBS, and permeabilized with 0.1% Triton^™ X-100 (Sigma-Aldrich, T8787) in a serum-free protein block solution (#x090930–2, Dako) for 1 h at room temperature ⁴³. The primary antibodies included N-Cadherin (1:200, Abcam ab18203) and Connexin 43 (1:200, Abcam ab219493). After the overnight incubation, slides were washed three times with PBS and incubated with the appropriate Alexa Fluor–conjugated secondary antibody (1:250, Invitrogen) for 2 h at room temperature. After washes, slides were mounted with Fluoroshield with DAPI (Sigma-Aldrich) mounting medium. Slides were imaged using confocal fluorescence microscopy (Leica SP5) and quantified using ImageJ software. Pooled data were generated by averaging data from at least 4 fields per heart section.

Membrane permeability in whole heart preparations

Lucifer Yellow (LY) was used to quantify the entry of a small molecular weight dye into cells as described ⁴⁴. Hearts were Langendorff-perfused at a constant flow rate of 2 mL/min with a low-calcium solution containing (in mmol/L) 2 EGTA, 135 NaCl, 5.4 KCl, 1.0 MgSO4, 0.33 Na2HPO4, 10 HEPES, 5.5 Glucose, 30 Taurine and 10 nM of CaCl2 (pH 7.4 NaOH). The hearts were then perfused with a dye-added low-calcium solution containing (in mg/mL): 1 LY (Sigma-Aldrich, #L0259) and 0.04 Wheat Germ Agglutinin Alexa Fluor 633 (WGA; Thermo Fisher, #W21404) for 30 min at 37°C. Subsequently, the dye in the extracellular space was washed by perfusing hearts with Tyrode’s solution without EGTA and 1.8 mM of CaCl2 for 30 min at 37°C. The exposure to low Ca2+ was used to increase the probability of opening of Cx43 hemichannels, whereas the return to a Ca2+–containing solution was intended to prevent the exit of the dye from the intracellular space before fixation. Hearts were then perfused with 10 mL of 4% PFA containing PBS and kept in the same solution at 4°C overnight. The PFA-fixed heart was placed in contact with the glass bottom chamber and fluorescence images were taken using confocal fluorescence microscopy (Leica SP5, LY ex 405 nm em 470–580 nm and WGA ex 633 nm em 650–800 nm) and quantified using ImageJ software. Images were taken at several random sites in each tissue sample.

Western blot

Proteins were isolated from left ventricular tissue using Bead Ruptor 12 homogenizer (Omni International) with ice-cold RIPA buffer enriched with halt^tm protease and phosphatase inhibitor cocktail (Thermo Fisher Scientific, #78442) and 5 mM EDTA. Extracts were then centrifuged at 12,000×g for 15 min at 4°C and the protein content was quantified by BCA Protein Assay (Thermo Fisher Scientific, #23227). Western blot was performed with the NuPAGE system (Thermo Fisher Scientific). 25 μg of proteins were resolved on 4–20% gradient gels, transferred onto a PVDF membrane, blocked at room temperature with 5% BSA solution, and incubated with the following primary antibodies: Cx43 (#C8093, 1:1000, Sigma), pCx43^ser368 (#ab194928, 1:1000, abcam), and GAPDH (#3683S, 1:3000, Cell signaling). All primary antibodies were incubated overnight with a blocking solution at 4°C. A secondary Alexa Fluor 488 antibody (#A32731, 1:1000, Invitrogen) was incubated for 60 minutes at room temperature. Immunoreactivity was detected by fluorescence or enhanced chemiluminescent ECL substrate and imaged using a ChemiDoc system (Bio-Rad Laboratories). The loading control was run on the same blot.

In vitro TNFα neutralization in bone marrow-derived macrophages (BMDMs)

Bone marrow samples were collected from the tibia and femurs of WT mice into RPMI containing 1% Penicillin/streptomycin (P/S) cocktail. After filtering through a cell strainer, cells were centrifuged at 500 g for 5 min and resuspended in RPMI enriched with 20% L929 cell-conditioned medium, 10% heat-inactivated fetal bovine serum (FBS), and 1% P/S cocktail, followed by growth on Petri dishes and differentiation to macrophages for 7 days. On day 7, cells were scraped, counted and replated on 48 well-plates (0.5×10⁶ cells/well) and treated with isotype IgG control (10 μg/mL) or monoclonal antibody against TNFα (10 μg/mL) for 2 hours followed by LCWE (30 μg/mL) or recombinant murine TNFα protein (20 ng/mL) for 24 hours, then media was collected for cytokine measurement.

In vivo TNFα neutralization in LCWE-injected WT mice

WT mice received i.p. injection of rat anti-mouse-TNFα mAb (500 μg single injection) or isotype IgG control (500 μg single injection) in 200μl PBS. 2 hours later, LCWE was injected i.p. (500 μg). Peritoneal lavage was collected 24 hours after LCWE injection, by washing the peritoneal cavity with PBS (1 ml) for cytokine measurement.

Cytokine measurements

Cytokine concentrations in culture supernatants or peritoneal lavage were quantified using ELISA for TNFα (eBioscience) and IL-6 (BD Biosciences) according to manufacturers’ protocols.

Statistical Analysis

Pooled data are expressed as mean ± SEM. Statistical comparisons were performed using GraphPad Prism 9 (San Diego, CA). Differences between groups were tested using a two-tailed unpaired Student’s t-test or one-way ANOVA followed by Tukey for multiple comparisons. When appropriate, repeated measures ANOVA followed by Bonferroni was used. Proportions were tested using Fisher’s exact test. Kaplan-Meier analysis was used for survival rates with the log-rank test. P < 0.05 was considered statistically significant.

RESULTS

LCWE-induced KD vasculitis is associated with cardiac mechanical, structural, and electrical abnormalities

Activation of pro-inflammatory cytokines such as IL-1 and TNF is known to promote cardiac arrhythmia ⁴⁵. To begin to investigate the role of these cytokines in KD, we used publicly available gene expression datasets generated from the whole blood of KD patients and healthy controls to analyze the expression of these genes (Figure 1A–B). Using two independent transcriptomic datasets (GSE68004 and GSE73461) ^36,37, we observed that IL1B and TNF genes were upregulated in acute KD patients when compared to controls (fold change > 1.5 and adjusted p-value<0.05; Figure 1B). Similar expression profiles were also observed in whole blood of acute KD compared to IVIG-treated convalescent KD patients ³⁸ (Figure 1C).

Figure 1. LCWE-induced KD vasculitis is associated with cardiac mechanical and electrical abnormalities.

A, Datasets were obtained from publicity available from whole blood transcriptomics from patients with acute Kawasaki disease (KD). B, Heatmaps showing increased expression of IL1B and TNF genes in patients with KD compared with healthy controls (HC). C, Expression of IL1B and TNF in whole blood of acute KD patients compared with convalescent IVIG-treated KD patients. D, Experimental design. E, Representative echocardiographic recordings, and quantification of ejection fraction (F) and LVID(d) (G). Representative electrocardiographic recordings (H) and quantification of QTc interval (I). Representative programmed electrical stimulation tracings and pacing protocol (J) and Incidence of pacing-induced ventricular tachycardia (K). Data are expressed as mean ± SEM. Unpaired Student’s t-test (F, G, and I). Fisher’s exact test (K). HC, healthy control; KD, Kawasaki disease; LCWE, Lactobacillus casei cell wall extract; LVID(d), left ventricular inner diastolic dimension; VT, ventricular tachycardia.

Although these findings suggest that increased expression of IL-1β and TNF may participate in KD pathogenesis, there is no evidence of their contribution to myocardial dysfunction and arrhythmogenesis in KD. To address this question, we utilized the LCWE-induced KD vasculitis mouse model ^20,22–24. C57Bl6 mice received either PBS or LCWE and were subjected to initial screening by echocardiography on day 14 post-LCWE injection (Figure 1D). The endpoint was selected based on our previous studies using this model demonstrating the development of systemic and cardiac infiltration of inflammatory cells and characteristic formation of CAA ^20,22–24. As shown in Figure 1E–F, the left ventricular (LV) ejection fraction (EF) was reduced in LCWE-injected mice compared to PBS-injected controls. Structural LV remodeling was also observed in LCWE-injected mice, demonstrating mild dilated cardiomyopathy (Figure 1G and supplemental table 1). There were no differences in the heart weight to body weight ratio between PBS of LCWE-injected mice at day 14 (0.0052 ± 0.0005 vs 0.0049 ± 0.00029) respectively. There were also no differences in blood pressure measurements at day 7 and day 14 between PBS control and LCWE-injected mice (supplemental table 2). A surface ECG revealed prolonged QT and QTc in LCWE-injected mice compared to controls (Figure 1H–I and supplemental table 2), with no changes in other ECG parameters (supplemental table 2). The QTc prolongation prompted us to interrogate the susceptibility to ventricular arrhythmias in KD mice. In control animals, programmed electrical stimulation induced VT in only 1 of 8 mice (12.5%), while KD mice were much more susceptible, with 6 of 8 mice (75%) developing VT (6 of 8 mice, 75%) (Figure 1J–K).

Intrinsic cardiac electrophysiological remodeling in LCWE-induced KD mice

To gain insight into the mechanisms involved in the electrophysiological remodeling of heart tissues from LCWE-injected mice, we performed high-resolution optical mapping using an intact whole-heart preparation. Figure 2A shows representative action potential maps in hearts from PBS-injected control mice and LCWE-injected mice, paced at 140-millisecond cycle length. Although no changes were observed in the action potential duration (APD) at 20% of repolarization (APD20) (Figure 2B), cardiac APD at 50% and 80% of repolarization were significantly prolonged in LCWE-injected mice compared to controls (Figure 2C, D), suggesting changes in phases 2 and 3 of cardiac action potentials. The spatial heterogeneity of APD remained unchanged between groups (Figure 2E). Moreover, as shown in the representative isochronal maps (Figure 2F), the propagation of electrical impulses was significantly slower in LCWE-injected mice compared to controls (Figure 2G), without changes in the spatial heterogeneity of conduction velocity (Figure 2H). Together, these findings indicate an intrinsic cardiac electrophysiological remodeling in LCWE-induced KD mice.

Figure 2. Intrinsic cardiac electrophysiological remodeling during LCWE-induced KD vasculitis.

A, Representative images of action potential duration (APD) map at 140 ms pacing cycle length. Quantification of action potential duration at 20% (APD₂₀, B), 50% (APD₅₀, C), 80% (APD₈₀, D) repolarization, and action potential duration dispersion at 80% (E). F, Representative images of isochronal voltage map and quantification of conduction velocity (G) and conduction velocity dispersion (H). Data are expressed as mean ± SEM. Unpaired Student’s t-test (B-E and G-H). ADP: action potential duration; LCWE, Lactobacillus casei cell wall extract; ms, millisecond.

IL-1Ra is more effective at preventing electrophysiological remodeling and myocardial dysfunction in LCWE-induced KD mice than TNFα neutralization

Given the abundant evidence demonstrating the role of IL-1 and TNF-α in the development of CAA in KD ^21–25, we interrogated whether inhibition of these proinflammatory cytokines could alleviate cardiac remodeling in LCWE-injected mice. We treated mice with either IL-1Ra (anakinra) or TNFα monoclonal antibody (TNFα mAb), or isotype control (IgG) starting from 1 day prior to LCWE injection until day 5 post-LCWE (Figure 3A). The reduction of EF observed in LCWE-injected mice was prevented by treatment with IL-1Ra but remained unchanged by the treatment with TNFα mAb (Figure 3B–C). Moreover, only the treatment with IL-1Ra significantly prevented the dilation of the LV (Figure 3D). For completeness, see supplemental table 3.

Figure 3. Superior efficacy of IL-1R antagonist over TNFα neutralization in the prevention of cardiac dysfunction and arrhythmia in LCWE-injected mice.

A, Experimental design depicting the treatment of LCWE-injected mice with IgG isotype control, IL-1R antagonist, or TNF-α mAb. B, Representative echocardiographic recordings, and quantification of ejection fraction (C) and LVID(d) (D). E, Quantification of QTc interval. Representative programmed electrical stimulation tracings (F) and incidence of pacing-induced ventricular tachycardia (G). The dashed line in C, D, and E represents the mean value of healthy control mice. Data are expressed as mean ± SEM. One-way ANOVA with Tukey multiple comparison post hoc test (C, D, and E). Fisher’s exact test (G). LCWE, Lactobacillus casei cell wall extract; IgG, Immunoglobulin G; IL-1Ra, interleukin-1 receptor antagonist; TNFα, tumor necrosis factor; mAb, monoclonal antibody; LVID(d), left ventricular inner diastolic dimension; VT, ventricular tachycardia.

Similarly, the treatment with IL-1Ra reduced the LCWE-induced prolongation of the QTc interval, but this measure remained prolonged in TNFα mAb-treated KD mice (Figure 3E). For complete ECG analyses, see supplemental table 4. To determine whether the attenuation of electrophysiological changes prevented the development of ventricular arrhythmias in KD mice, we repeated the in vivo programmed electrical stimulation (Figure 3F). Indeed, the VT inducibility remained high in TNFα mAb-treated KD mice (5 of 8 mice, 62.5%), whereas it was significantly reduced in IL-1Ra-treated mice (3 of 10 mice, 30%) compared to IgG-treated LCWE-injected mice (Figure 3G).

Corroborating our in vivo findings, optical mapping from explanted hearts demonstrated a significant APD80 shortening in IL-1Ra-treated LCWE-injected mice compared to IgG-treated mice (Figure 4A–B). Notably, although APD80 was shortened by IL-1Ra treatment, the values remained significantly longer relative to PBS-injected control mice (healthy hearts; Figure 2A–D, dashed line Figure 4B). No changes were observed in LCWE-injected mice treated with TNFα mAb (Figure 4A–B). Cardiac APD80 dispersion was unchanged between groups (Figure 4C). Along with these findings, cardiac conduction velocity was significantly faster in the hearts of mice treated with IL-1Ra compared to IgG-treated and TNFα mAb-treated mice (Figure 4D–E). No changes in conduction velocity heterogeneity were noted (Figure 4F). Although the conduction velocity was reduced in LCWE-injected mice compared to PBS-injected control mice (Figure 2G) but fully restored by IL-1Ra treatment, this cannot be explained by changes in fibrosis, which were not present in this acute model of experimental KD vasculitis (data not shown).

Figure 4. IL-1R antagonist prevents cardiac electrophysiological remodeling in LCWE-induced KD vasculitis.

A, Representative images of action potential duration (APD) map at 140 ms pacing cycle length. Quantification of action potential duration at 80% (APD₈₀, B) repolarization, and action potential duration dispersion at 80% (C). D, Representative images of isochronal voltage map and quantification of conduction velocity (E) and conduction velocity dispersion (F). The dashed line in B-C and E-F represents the mean value of healthy control mice. Data are expressed as mean ± SEM. One-way ANOVA with Tukey multiple comparison post hoc test (B-C and E-F). ADP: action potential duration; LCWE, Lactobacillus casei cell wall extract; IgG, Immunoglobulin G; IL-1Ra, interleukin-1 receptor antagonist; TNFα, tumor necrosis factor; mAb, monoclonal antibody; ms, millisecond.

To ensure that the murine anti-TNFα mAb dose and regimen we used were indeed effective in inhibiting TNFα in our model, we first tested its effects on bone marrow-derived macrophages (BMDMs) in-vitro and on LCWE-induced responses in-vivo. The murine anti-TNFα mAb that we used significantly blocked LCWE-induced TNFα and recombinant murine TNFα-induced IL-6 release from BMDMs (Supplemental Fig 1A and B). In vivo, the anti-TNFα mAb blocked LCWE-induced TNFα and IL-6 production in the peritoneal fluid of mice at the 24-hour time point (Supplemental Fig 1C and D). Finally, in parallel experiments, we investigated the effect of the anti-TNFα mAb on blocking the development of vasculitis by measuring heart vessel inflammation score. We observed that the same dose and regimen of the murine anti-TNFα mAb that we used in our electrophysiology experiments significantly inhibited heart vessel inflammation, performing as well as the IL-1R antagonist (anakinra) (Supplemental Fig 1E and F). Together, these results unequivocally demonstrate that the murine anti-TNFα mAb at the dose and regimen we used was effective in suppressing heart vessel inflammation in the LCWE-induced KD vasculitis model, and its inability to disrupt the development of arrhythmias or myocardial dysfunction was not due to inadequate dosing.

Considering that gap junction channels play an important role in cardiac electrical conduction, and connexin43 (Cx43) is the main ventricular gap junction⁴⁶, we next investigated the abundance of Cx43 and its localization pattern in cardiomyocytes in the setting of LCWE-induced KD and IL-1Ra treatment. We did not observe any differences in the total protein levels and phosphorylation of Cx43 in LV lysates between groups (Figure 5A–C). However, immunostaining revealed significant lateralization of Cx43 in the myocardium of LCWE-injected mice compared to healthy controls such that it co-localized with N-cadherin at the intercalated disk. These changes in Cx43 localization were prevented by treatment with IL-1Ra (Figure 5D–E).

Figure 5. IL-1R antagonist prevents the lateralization of Cx43 and membrane permeability in LCWE-induced KD vasculitis.

A, Representative western blot images. B, Quantification of Cx43 protein expression. C, Quantification of phosphorylated Cx43^ser368 protein expression. D, Representative images of Cx43 immunostaining in left ventricular sections. E, Quantification of Cx43 and N-cadherin colocalization. Scale Bar = 50 μm. F, Representative images of Lucifer yellow dye transfer assay in whole heart preparations. G, Average Lucifer yellow intensity measured from cells in the following groups: PBS (208 cells from 3 mice), LCWE + IgG (372 cells from 3 mice), and LCWE + IL-1Ra (267 cells from 3 mice). Scale Bar = 25 μm. Data are expressed as mean ± SEM. One-way ANOVA with Tukey multiple comparison post hoc test. LCWE: Lactobacillus casei cell wall extract; IL-1Ra, interleukin-1 receptor antagonist; Cx43, connexin 43; LY, Lucifer yellow; WGA, wheat germ agglutinin.

The loss of Cx43 gap junction at the intercalated disk led us to devise the lucifer yellow dye transfer assay in Langendorff-perfused whole hearts to assess whether lateralized Cx43-based hemichannels would alter the membrane permeability (see methods). We confirmed that mislocalization of Cx43 in LCWE-induced KD vasculitis was associated with increased membrane permeability, and the restoration of intercalated Cx43 localization by IL-1Ra treatment restored cardiomyocyte membrane permeability (Figure 5F–G).

To further validate our hypothesis that IL-1 signaling plays a central role in the development of arrhythmogenesis in the KD vasculitis mouse model, we used IL-1 receptor antagonist (IL-1Ra) deficient mice (Il1rn^−/−), which exhibit enhanced responses to IL-1 stimulation ⁴⁷. Il1rn^−/− mice and WT littermates were implanted a telemetric device enabling continuous monitoring of spontaneous arrhythmias in conscious animals, and then injected with LCWE (Figure 6A). Surprisingly, beginning 7 days after LWCE administration, Il1rn^−/− mice began experiencing sudden deaths, whereas no mortality was recorded in WT mice that received LCWE (Figure 6B). Due to the mortality, we analyzed the incidence of spontaneous arrhythmia until the 7^th day post LCWE injection. As shown in Figure 6C, the morphology of the ECG waveform was changed in both groups upon LCWE injection, but the depression of the T wave in Il1rn^−/− mice was more profound than that seen in WT mice. Figure 6D shows that Il1rn^−/− mice are more susceptible to develop spontaneous VT than WT mice, although among the animals that developed VT, there was no statistical difference in the duration of VT between groups (Figure 6E). Moreover, the administration of LCWE equally increased the number of premature ventricular contractions in both groups (Figure 6F). Ambulatory ECG recordings showed that all Il1rn^−/− mice injected with LCWE developed advanced atrioventricular (AV) conduction block, which led to ventricular arrest (Figure 6G).

Figure 6. Gain-of-function of IL-1 signaling facilitates arrhythmias and sudden death in LCWE-induced KD vasculitis.

A, Experimental design. B, Survival analysis up to 15 days post LCWE injection. C, Representative electrocardiographic recordings before (baseline) and 7 days after LCWE injection. Asterisks highlight the depression of the T wave in knockout mice for IL1-Ra. D, Incidence of spontaneous ventricular tachycardia; E, Duration of ventricular tachycardia; F, Incidence of spontaneous PVCs per hour (n=8). G, Representative electrocardiographic recordings. Data are expressed as mean ± SEM. Log-rank test (B). Fisher’s exact test (D). RM-ANOVA followed by Bonferroni post hoc test (F). LCWE: Lactobacillus casei cell wall extract; WT, wild type; IL-1Ra, interleukin-1 receptor antagonist; Il1rn^−/−, knockout mice for IL-1Ra; VT, ventricular tachycardia; PVCs, premature ventricular contractions; NSVT, non-sustained ventricular tachycardia; AV block, atrioventricular block.

DISCUSSION

Although most KD cases are associated with typical vasculitis manifestation and CAA remodeling, adverse cardiac events are often observed ^1,20,48. Indeed, approximately 40–80% of KD children develop cardiac electrical abnormalities during acute KD, which may lead to the development of fatal arrhythmias ^8–10. Our study provides the first mechanistic evidence demonstrating that prolonged ventricular repolarization and slowed electrical conduction underlie cardiac arrhythmias in LCWE-injected mice developing KD vasculitis. In addition, despite the potential of both neutralization of TNFα and inhibition of IL-1β signaling to prevent vascular lesions in KD ^{18,30,49–51}, our findings indicate that blocking IL-1 signaling with IL-1Ra (anakinra) is more effective than TNFα neutralization at improving myocardial dysfunction and preventing cardiac electrical remodeling and arrhythmias in the LCWE murine model of acute KD vasculitis. Notably, although the LCWE-induced KD mouse model of vasculitis is associated with myocarditis^22,26, myocardial dysfunction, reduced ejection fraction^26,52, and increased levels of NT-proBNP²⁶, these symptoms were relatively mild to moderate and were not associated with blood pressure changes in the animals, similar to findings in KD patients¹.

Our recent findings using spatial transcriptomics in heart tissue from LCWE-injected mice revealed profound infiltrations of immune cells, such as macrophages, monocytes, T cells, and dendritic cells ²¹. We then took advantage of publicly available human transcriptomic datasets of whole blood from healthy controls, acute and IVIG-treated convalescent KD patients to validate the systemic upregulation of IL1B and TNF in acute KD patients. Similar to human KD, LCWE-injected mice develop acute myocarditis followed by long-term sequelae ⁵². While the precise mechanisms of how immunocompetent cells participate in arrhythmogenesis ⁴⁵ in diseases such as KD vasculitis remains uncertain, acute myocarditis in this experimental KD model and in KD patients may contribute to ECG changes and arrhythmia, but it is also widely accepted that released pro-inflammatory cytokines facilitate rhythm disorders and may contribute to cardiac arrhythmia ^32,33,53,54.

We and others have previously shown that activation of NLRP3 inflammasome and IL-1 signaling pathways play a central role in mediating systemic inflammation and CAA development in the LCWE model of KD and in KD patients ^{21–23,25,55}. We have previously reported that LCWE-injected mice also develop acute myocarditis which can be prevented by IL-1Ra treatment but not with human anti-TNFα mAb ^22,26. This observation that IL-1Ra (anakinra) can reduce both coronary arterial and myocardial inflammation in the LCWE model received considerable attention in the field ⁵⁶. Furthermore, we have previously shown that LCWE-injected mice recapitulate ECG abnormalities reported in acute KD patients, and also develop cardiac ganglionitis in an IL-1-dependent manner ³⁵. Here, we significantly built on those pilot findings to provide the first comprehensive analysis of the electrophysiological and ventricular function of hearts from LCWE-injected mice. We show that LCWE induces APD prolongation, CV slowing, and impairment of left ventricular ejection fraction, which has rarely been reported in children with Kawasaki disease ^57,58. Although the treatment of LCWE-injected mice with IL-1Ra partially restored the repolarization, the most striking finding was the complete restoration of electrical conduction. Moreover, our findings provide evidence that blockade of the IL-1 pathway, but not the TNFα pathway, prevents adverse contractile and electrical remodeling in KD hearts. Accumulating evidence demonstrates that IL-1 impairs the myocardial contractility, favoring adverse ventricular remodeling and heart failure ^59,60, and our findings substantiate the critical role of IL-1 signaling in both systolic impairment and arrhythmia incidence. Our findings are also in accordance with previous evidence that IL-1Ra significantly reduces overall myocardial inflammation, while TNF inhibition did not significantly reduce the incidence of myocarditis in this murine model of KD vasculitis ^22,56. The involvement of the NLRP3 inflammasome activation in the vasculature and myocardium in this murine model of KD vasculitis ^21,34, provides a logic explanation for the efficacy of IL-1Ra that we observed and expands these findings to other inflammasome-mediated injury processes, as well as non-infection auto-inflammatory processes.

Using specific knockout mice and specific monoclonal antibodies, we have previously shown that both IL-1α and IL-1β play important and nonredundant roles in the development of LCWE-induced vasculitis²⁵, which explains our choice to use IL-1Ra (anakinra), which blocks both isoforms. However, future studies are warranted to determine the specific role(s) of IL-1α and IL-1β in the myocardial dysfunction and arrhythmias that we observed in the current study.

The propagation of cardiac electric impulses is determined by multiple factors, including extracellular remodeling and cell-to-cell coupling via gap junctions. However, we did not observe increased myocardial fibrosis in the acute phase of this experimental KD model, thus ruling out a role for extracellular remodeling in the slowed conduction in KD mice. Fibrotic remodeling of the working myocardium has only been reported in cardiac biopsies obtained from patients with at least 3 years elapsed since the acute KD phase ^61,62. Here, at the early stage, we observed lateralization of the gap junction protein Cx43, without changes in its total expression level. Downregulation of Cx43 expression has been shown in different forms of cardiomyopathy ^41,63,64, and future investigation is needed to determine its abundance at the late stages of KD. One possible explanation for the mislocalization of Cx43 in LCWE-injected KD mice might be secondary to the altered metabolic conditions of ischemic substrates ^65,66. We previously demonstrated that the later stages of the LCWE-induced KD vasculitis are associated with a reduced microvascular capillary network in the myocardium ⁵², which limits the oxygen supply, predisposing cardiomyocytes to ischemic cell death. Nevertheless, the loss of Cx43 at the cell junction combined with increased membrane permeability through Cx43 hemichannels might eventually result in electrical and mechanical disruption in this acute model of KD ^44,46,67,68. Moreover, IL-1β has been shown to alter ion channels ⁶⁹; therefore, future studies are needed to establish the potential involvement of ionic remodeling in the arrhythmogenesis of KD.

Although we did not observe sudden death in the wild-type mice receiving LCWE, our findings indicate the presence of cardiac electrophysiological changes in the acute phase, which could lead to more severe events at later stages. The increased incidence of spontaneous VT and sudden death in Il1rn^−/− mice, which have excess IL-1 signaling, serves as an unequivocal validation of the central role of this pathway in KD pathogenesis. LCWE-treated Il1rn^−/− mice showed ECG changes including prolonged ST depression followed by complete AV block and death (full AV block), however, we were not able to investigate the exact cause of mortality and presence of severe systemic inflammation. Interestingly, clinical cases of heart block and bradyarrhythmias have been reported during the acute phase of Kawasaki disease ^70–75. Thus, the study of KD pathophysiology at an early stage provides invaluable mechanistic insights into the critical molecular players for the progression and development of cardiovascular lesions as well as electrophysiologic changes.

STUDY LIMITATIONS

The LCWE-induced KD vasculitis and coronary arteritis model is a well-accepted experimental model of KD vasculitis with striking similarities to the immune-histopathological lesions of human KD ²⁰. The arrhythmia, myocarditis and myocardial dysfunction that we observed in the LCWE-induced KD mouse model recapitulate this more severe phenotype of the acute stage disease, and provide us with a unique opportunity to generate novel insights that could then be tested in children with KD. While we acknowledge that no single animal model of any disease can truly replicate the exact clinical findings of human disease, this experimental model of KD has proved to be an important tool to investigate the immunopathological aspects of KD, and lead to significant mechanistic insights²⁰ that can then be validated in human disease. Our current findings in the LCWE-induced experimental KD vasculitis model follow the same path. Finally, whether the superior therapeutic efficacy of IL-1Ra over anti-TNF-α treatment to prevent myocardial dysfunction, structural, and electrophysiological remodeling in this experimental mouse model of KD also translates to children with KD will need to be investigated in head-to-head clinical studies.

CONCLUSIONS

In summary, we demonstrated superior therapeutic efficacy of IL-1Ra over anti-TNF-α treatment to prevent cardiac function, structural, and electrophysiological remodeling in LCWE-injected mice developing KD vasculitis. Moreover, the reduction of ventricular arrhythmia in LCWE-injected mice treated with IL-1Ra was linked with the restoration of normal electrical impulse propagation and restoration of connexin 43 localization. These novel findings add to accumulating data supporting the use of anti-IL-1 therapy to prevent KD-induced cardiovascular inflammation, myocarditis, as well as changes in the arrhythmogenic substrate in not only IVIG refractory but all KD patients, as first suggested almost a decade ago ⁵⁶.

Highlights.

• Electrophysiological abnormalities, arrhythmias and myocardial dysfunction occur in an experimental murine model of acute Kawasaki disease vasculitis in an IL-1-dependent manner.

• Inhibition of IL-1 signaling by IL-1 Receptor antagonist (IL-1Ra), but not TNF neutralization, improves ejection fraction, myocardial dysfunction, and arrhythmias in the LCWE-induced experimental Kawasaki disease model.

• Reduction in ventricular arrhythmias by IL-1 Receptor antagonist (IL-1Ra) in the experimental murine model of Kawasaki disease is linked to restoration of normal electrical impulse propagation and restoration of connexin 43 localization.

Non-standard Abbreviations and Acronyms

APD

action potential duration

AV

atrioventricular

BMDM

bone marrow-derived macrophage

CAA

coronary artery aneurysm

Cx43

connexin43

ECG

electrocardiogram

EF

ejection fraction

IgG

Immunoglobulin G

IL-1

interleukin 1

IL-1Ra

IL-1 receptor antagonist

IVIG

intravenous immunoglobulin

KD

Kawasaki disease

LCWE

Lactobacillus casei cell wall extract

LV

left ventricle

PVC

premature ventricular contraction

TNF-α

tumor necrosis factor-α

VT

ventricular tachycardia

WT

wild type