IL‐1‐dependent electrophysiological changes and cardiac neural remodeling in a mouse model of Kawasaki disease vasculitis

Summary

Kawasaki disease (KD) is the leading cause of acquired heart disease in children. In addition to coronary artery abnormalities, aneurysms and myocarditis, acute KD is also associated with echocardiogram (ECG) abnormalities in 40–80% of patients. Here, we show that these ECG changes are recapitulated in the Lactobacillus casei cell wall extract (LCWE)‐induced KD vasculitis mouse model. LCWE‐injected mice developed elevated heart rate and decreased R wave amplitude, with significant differences in prolonged ventricular repolarization. LCWE‐injected mice developed cardiac ganglion inflammation, that may affect the impulse‐conducting system in the myocardium. Furthermore, serum nerve growth factor (NGF) was significantly elevated in LCWE‐injected mice, similar to children with KD vasculitis, associated with increased neural remodeling of the myocardium. ECG abnormalities were prevented by blocking interleukin (IL)‐1 signaling with anakinra, and the increase in serum NGF and cardiac neural remodeling were similarly blocked in Il1r1^−/− mice and in wild‐type mice treated with anakinra. Thus, similar to clinical KD, the LCWE‐induced KD vasculitis mouse model also exhibits electrophysiological abnormalities and cardiac neuronal remodeling, and these changes can be prevented by blocking IL‐1 signaling. These data support the acceleration of anti‐IL‐1 therapy trials to benefit KD patients.

Using the mouse model of Kawasaki Disease, Abe et al observed ECG changes and ganglionitis as well as cardiac neural remodeling. These changes were dependent on IL‐1 signaling and are similar to the electrophysiological changes observed in Kawasaki Disease patients.

Introduction

Kawasaki disease (KD) is an acute febrile illness and systemic vasculitis that predominantly affects children under 5 years of age, and is the leading cause of acquired heart disease among children in the developed world, including the United States 1, 2, 3. Despite 40 years of investigation the etiology of KD remains unknown, but the current paradigm is that, in genetically susceptible children, KD is triggered by an infectious agent that elicits an autoinflammatory response directed to specific cardiovascular tissues 4, 5. Once believed to be a self‐limiting childhood disease, it is now known that acute KD vasculitis results in long‐term complications including ongoing vascular remodeling, coronary artery aneurism (CAA), ischemic heart disease and myocardial fibrosis extending into adulthood 6, 7.

Intravenous immunoglobulin G (IVIG) administered within the first 10 days of illness can reduce the incidence of CAA from 25 to 4%; however, nearly 30% of KD patients are IVIG‐resistant and at even higher risk for developing CAA 8, 9. A better understanding of the mechanisms of KD‐induced cardiac lesions is urgently needed to pave the way to new therapeutic targets. The limited tissue availability from KD patients has significantly impeded progress in understanding the etiology and the immunopathogenesis of this disease. However, the Lactobacillus casei cell wall extract (LCWE)‐induced KD vasculitis mouse model closely mimics the important histological and immune‐pathological features of the cardiovascular lesions observed in KD patients 10. This model has significantly increased the speed of novel discoveries in this disease, and has successfully been used to predict therapeutic efficacy in children with KD 11, 12.

Recent studies in human KD patients and with the LCWE mouse model have converged on the critical role of interleukin (IL)‐1 signaling in cardiovascular lesions of KD 12, 13, 14, 15, 16. Two clinical trials using the IL‐1 receptor (IL‐1R) antagonist anakinra are now completed in KD patients who do not respond to IVIG 17, and Phase III trials are being designed both in the United States and Europe.

While coronary artery abnormalities are seen in 25% of untreated KD patients, almost all children with acute KD present with myocarditis 18. Harada et al. reported that autopsies of 29 KD patients who died within 40 days of disease onset all displayed myocardial inflammation 19. However, electrocardiogram (ECG) abnormalities are not as well defined in KD patients. Clinically, impulse‐conducting system abnormalities such as arrhythmias, abnormal Q waves, prolonged PR and/or QT intervals, low‐voltage and non‐specific ST and T wave changes have occasionally been observed 20. Impulse‐conducting system abnormalities such as transient atrioventricular block, premature ventricular contraction, supraventricular tachycardia and ventricular tachycardia are seen in the early phase of KD 21, 22. Several studies have shown that there is an increased risk for ventricular arrhythmia during acute KD with increased dispersion of repolarization (QT dispersion) times as calculated from ECGs 23, 24, 25, 26. Children with KD have a significantly higher QTc dispersion in both 12‐ and 8‐lead analysis 27, and abnormal resting‐ and exercise‐induced repolarization is detectible long after KD onset, irrespective of the severity of coronary sequelae 28. The dispersion is indicative of ventricular repolarization and may represent an increased risk for developing ventricular arrhythmia in this population.

We have previously shown that LCWE‐induced KD vasculitis mouse model is associated with myocarditis and diminished ejection fraction, which was prevented by anakinra administration 12, 29, 30. In this study, we show that this mouse model also recapitulates electrophysiological and ECG abnormalities observed in KD patients. Treatment with anakinra effectively prevented these electrophysiological ECG abnormalities. Furthermore, LCWE‐treated mice exhibited inflammation in the atrial fat ganglia (ganglionitis) and increased circulating serum nerve growth factor (NGF) concentrations, which may affect the impulse‐conducting system in the myocardium leading to the ECG changes. Finally, we also observed cardiac neural remodeling in LCWE‐injected mice. The increased serum NGF and cardiac neural remodeling were both dependent on IL‐1 signaling, as these changes were not observed in Il1r1^−/− mice or in wild‐type (WT) mice treated with anakinra. These novel findings add to the overwhelming data supporting the notion that the LCWE‐induced KD vasculitis mouse model is currently the most valuable experimental model to study KD, and suggest that anti‐IL‐1 therapy trials should be accelerated to benefit KD patients, who almost always have myocarditis.

Materials and methods

Ethics statement

All animal experiments were performed according to the approved protocol [Institutional Animal Care and Use Committees (IACUC) protocol #5093] and guidelines of the Cedars‐Sinai Medical Center Institutional Animal Care and Use Committee. Cedars‐Sinai Medical Center is fully accredited by the Assessment and Accreditation of Laboratory Animal Care (AAALAC International) and abides by all applicable laws governing the use of laboratory animals. Laboratory animals are maintained in accordance with the applicable portions of the Animal Welfare Act and the guidelines prescribed by the Department of Health and Human Services (DHHS) publication, Guide for the Care and Use of Laboratory Animals.

Mice

WT CD45.1 and CD45.2 C57BL/6 and Il1r1^−/− mice were purchased from the Jackson Laboratory (Bar Harbor, ME, USA). All animals were housed under specific pathogen‐free conditions at the animal center of the Cedars‐Sinai Medical Center. Experiments were conducted under approved Institutional Animal Care and Use Committee protocols. The number of animals used in various experiments ranged from four to 10 in each group, as specified in the figure legends. Il1a^−/− mice were obtained from Dr Yoichiro Iwakura, Professor, Research Institute for Biomedical Science, Tokyo University of Science. Il1b^−/− mice obtained from Hal Hoffman, University of California, San Diego.

Reagents

Human IL‐1 receptor antagonist (IL1Ra, anakinra, 25 mg/kg; Swedish Orphan Biovitrum AB, Stockholm, Sweden) was given daily intraperitoneally (i.p.) starting 1 day before LCWE injection until day 5 post‐LCWE injection, as previously described 12, 29.

ECG recording in LCWE‐induced coronary arteritis model

L. casei (American Type Culture Collection 11578) cell wall extract was prepared as previously described 12, 31. Five‐week‐old C57BL/6 mice were injected i.p. with 500 μl LCWE in phosphate‐buffered saline (PBS) to induce KD. A control group of 5‐week‐old C57BL/6 mice received only PBS i.p. A treatment group of 5‐week‐old C57BL/6 mice received 500 μl LCWE to induce KD and were concurrently given 300 μg IL‐1Ra i.p. For ECG, animals were placed in isoflurane chamber for initial inhalation of 4% isoflurane for approximately 2 min (until visually determined to be asleep). The animal was then transferred to the ECG setup area, where the individual limbs were restrained, and the isoflurane mask applied. The inhaled isoflurane mask was set to 1–1·5%, depending on the mass of the animal, for the duration of the experiment. All animals were kept warm by providing a heating source (heat pad) under their bodies during the procedure and until they fully recovered from anesthesia. Anesthetized mice recovered in a bare cage on top of paper towels to prevent aspiration of bedding. The ECG recording provided data for the following: heart rate (HR), RR interval, P amplitude, P duration, PR interval, Q amplitude, R amplitude, S amplitude, QRS interval, ST height, T amplitude, T_peak‐T_end interval, QT interval and QTc interval. QT intervals were corrected by using Bazett’s formula (QTc interval). Waves were obtained from I lead. Heart rate was measured automatically by the software (Lab Chart 5; AD Instrument Inc., Colorado Springs CO, USA), and other segments were semi‐automatically measured. Good recording waves contained at least six time‐points. The consecutive six heartbeats were turned in the software, and the software summarized and averaged the wave. Averaged waves were measured manually by software tools. Segment measurements were performed as reported previously 32. All ECG recordings for all groups were performed at the same time of day.

Retro‐orbital blood collection

Blood samples were obtained retro‐orbitally. All mice were anesthetized with inhalant gas 100% isoflurane until visually sedated (slowed breathing). Approximately 0·5 ml of blood was collected just prior to euthanasia.

Immunohistological staining and analysis

For staining of the ganglion, heart tissues were collected 22 days after LCWE injection, embedded and frozen in optimal cutting temperature (OCT) compound (Sukura Finetek, Torrance, CA, USA). Hematoxylin and eosin (H&E)‐stained sections were used to assess ganglionitis and quantify cell infiltration of the cardiac ganglia. Briefly, the cardiac ganglia area was measured with Image J software on H&E‐stained heart sections from PBS and LCWE‐injected mice, cells infiltrating the cardia ganglia were enumerated and the data normalized to the surface area and expressed as cells per nm². Serial sections (7 μm) were fixed in cold acetone for 5 min and stained with the following antibodies: CD45.1‐phycoerythrin (PE) (eBiosciences, San Diego, CA, USA; #12‐0453‐81), CD3‐PE (BioLegend, San Diego, CA, USA; #100307), CD19‐fluorescein isothiocyanate (FITC) (eBiosciences; #11‐0193‐82), F4/80‐FITC (BioLegend; #123107), CD11c‐PE (eBiosciences; #12‐0114‐82), CD11b‐FITC (eBiosciences; #11‐0112‐41), Ly6G‐PE (BioLegend; #127607), rat‐IgG2a isotype‐control‐PE conjugated (eBiosciences; #12‐4321‐82) and PE‐Armenian hamster immunoglobulin (IgG) isotype control hamster‐IgG (BioLegend; #400907). For staining the myocardium, hearts were collected from WT and Il1r1^−/− mice 2 weeks after PBS or LCWE injection. The ventricle was cross‐sectioned from base to apex and tissue sections were collected at the papillary muscle level. Primary antibodies were anti‐growth‐associated protein 43 antibody [growth‐associated protein 43 (GAP43); EMD Millipore, Burlington MA, USA; #AB 5220] and anti‐tyrosine hydroxylase antibody (TH; EMD Millipore; #AB152). As the secondary antibody, we used the Alexa‐fluor 564 conjugated goat anti‐rabbit (ThermoFisher Scientific, Fremont, CA, USA; #A‐11011). Rabbit IgG (Invitrogen, Carlsbad, CA, USA; #02‐6102) was used as isotype control. Sections were then coverslipped in ProLong Gold anti‐fade mounting medium with 4′,6‐diamidino‐2‐phenylindole (DAPI) (ThermoFisher Scientific; P36931). The images were processed in the microscope (BZ‐X710; Keyence, Osaka, Japan). Imaging of the myocardium was performed at ×20 magnitude field on six different myocardium areas evenly distributed in the myocardium for each heart; respectively, septum, anterior, lateral and posterior wall in the middle‐ventricle. We did not include RV wall myocardium and atrial wall. Density of the positive signal on each image was assessed by using the software (BZ‐X analyser; Keyence) and the average of the density was calculated.

Serum NGF quantification

Serum NGF concentrations were measured using an NGF Sandwich enzyme‐linked immunosorbent assay (ELISA) kit (Millipore Sigma, Burlington, MA, USA; #CYT304), according to the manufacturer’s guidelines.

RNA sequencing

Heart tissue specimens were collected from male and female PBS‐injected (n = 10) and LCWE‐injected (n = 10) WT mice 7 days after LCWE injection. RNA was extracted using the RNeasy Mini kit (Qiagen, Valencia, CA, USA). RNA quality was further determined using a Bioanalyzer (Agilent, Santa Clara, CA, USA); only samples with an RNA integrity number > 7 were used. RNA sequencing and subsequent analysis were then performed as previously described 10, 15.

Statistical analysis

Results are reported as mean ± standard error of the mean (s.e.m.). A P‐value of  < 0·05 was considered statistically significant. The two‐tailed Student’s t‐test [at 95% confidence interval (CI)] was used to compare unpaired samples between experimental groups or one‐way analysis of variance (anova) with Bonferroni post‐test analysis. All data analyzed were normally distributed (*P < 0·05; **P < 0·01; ***P < 0·001).

Results

LCWE induces electrophysiological abnormalities in an IL‐1‐dependent manner

In order to understand if the LCWE‐induced KD vasculitis model recapitulates the electrophysiological abnormalities observed in KD children, we injected WT mice with LCWE and performed ECG. LCWE‐injected mice showed a significantly higher baseline HR (Fig. 1a) and decreased R‐wave amplitude (Fig. 1b) than the PBS control group 1 week post‐injection, which resolved by 2–3 weeks after injection (data not shown). At 1 week post‐LCWE injection we observed a trend toward increased QTc interval (Fig. 1c), which reached significance 2 weeks after injection (Fig. 1d).

Figure 1.

Lactobacillus cell wall extract (LCWE) induces electrophysiological abnormalities in an interleukin (IL)‐1‐dependent manner. Echocardiogram (ECG) parameters were measured in wild‐type (WT) mice following injection of phosphate‐buffered saline (PBS), LCWE or LCWE with anakinra. (a) Heart rate per min (HR) 1 week following injection. (b) R amplitude 1 week following injection. (c) QTc interval (QTc = QT/(RR)^1/2) 1 week following injection. (d) QTc interval measured 2 weeks post‐injection. **P < 0·01 by one‐way analysis of variance (anova) with Bonferroni post‐test analysis; n = 5, 4 and 5 for PBS, LCWE and LCWE + IL‐1Ra groups, respectively.

As we have previously shown that IL‐1 signaling is required for LCWE‐induced vasculitis and myocarditis, we next asked whether this pathway also underlies the electrophysiological abnormalities. Indeed, treatment with the IL‐1R antagonist anakinra prevented the LCWE‐induced HR increase (Fig. 1a), R‐wave amplitude decrease (Fig. 1b) and increased QT interval (Fig. 1c,d), suggesting that these electrophysiological changes appear to be IL‐1‐dependent.

NGF is induced in an IL‐1‐dependent manner in the LCWE KD model

Nerve growth factor (NGF) can be induced during acute inflammation, particularly in response to IL‐1β stimulation 33, 34, and has been associated with induction of myocardial electrophysiological alterations 35. Accordingly, the elevation in serum NGF observed in acute KD patients 36 may contribute to their development of ECG abnormalities. We therefore asked if LCWE‐induced KD vasculitis was associated with higher expression of NGF. Compared with heart tissues from PBS‐injected mice, NGF mRNA levels are significantly higher in heart tissues of LCWE‐injected mice (Fig. 2a). One week following LCWE injection, NGF concentrations in the serum of mice increased fivefold compared with PBS‐injected control mice (Fig. 2b). However, LCWE failed to increase serum levels of NGF in anakinra‐treated mice, as well as in IL‐1α‐ and IL‐1β‐deficient mice (Fig. 2b), indicating that IL‐1 signaling is required for NGF induction during LCWE‐induced KD vasculitis.

Figure 2.

Nerve growth factor (NGF) is induced by Lactobacillus cell wall extract (LCWE) and is (IL)‐1‐dependent. (a) RNA seq analysis of heart tissues. mRNA expression counts of NGF in heart tissues are shown from phosphate‐buffered saline (PBS) and LCWE‐injected mice 1 week post‐injection (n = 10/group). (b) Mice were injected with PBS or LCWE and serum was obtained 1 week later. NGF was measured by enzyme‐linked immunosorbent assay (ELISA). PBS (n = 6), LCWE (n = 10), LCWE + IL‐1Ra (n = 10), Il‐1α ^−/− + LCWE (n = 10) and Il‐1β ^−/− + LCWE (n = 10). *P ≤ 0·05; **P ≤ 0·01 by Student’s t‐test (a) or one‐way analysis of variance (anova) with Tukey’s post‐hoc test used for multiple comparison (b).

LCWE induces cardiac ganglionitis

Cardiac rhythm abnormalities have been associated with inflammation of cardiac ganglia (ganglionitis) 37, so we next investigated whether this occurred in LCWE‐injected mice. We observed evidence of ganglionitis adjacent to the pulmonary vein–left atrium fat junction in LCWE‐injected mice that develop KD lesions, but not in control PBS‐injected mice (Fig. 3a,b). LCWE‐injected mice showed significantly higher numbers of infiltrating cells in the cardiac ganglia when compared to PBS control mice (Fig. 3c). Histological characterization of the cells infiltrating the cardiac nerves and ganglia (Fig. 4) of LCWE‐injected mice revealed that some of them were positive for CD45, a marker of immune cells. While none of the cells stained for CD19 (B cells) (Fig. 4c), some were CD3⁺ (T cells) (Fig. 4e). Furthermore, while none of the infiltrating cells were F4/80⁺ (macrophages) and Ly6G⁺ (Fig. 4e,i), both CD11c⁺ and CD11b⁺ cells were observed (Fig. 4g), suggesting that dendritic cells infiltrate the ganglia. In contrast, the coronary arteries adjacent to the aorta stained positively for all the immune cell markers tested (Fig. 4d,f,h,j). Thus, in addition to the inflammation observed in the coronaries and aortic root, our data now show that cardiac ganglionitis also occurs only in LCWE injected mice and not in PBS control mice (Fig. 4a–j), which may affect the impulse‐conducting system of the myocardium.

Figure 3.

Kawasaki disease (KD) vasculitis is associated with ganglionitis. Mice were injected with phosphate‐buffered saline (PBS) or Lactobacillus cell wall extract (LCWE) and killed 1 week after injection. (a,b) Representative haematoxylin and eosin (H&E)‐stained heart sections showing ganglia from PBS (a) or LCWE‐injected (b) wild‐type (WT) mice 1 week post‐LCWE injection. Images are representative of n = 5 per group and were taken at ×2 and ×20 magnification. Scale bar = 100 μm. (c) Quantification of infiltrating cells in the heart ganglia of PBS and LCWE‐injected mice 1 week post‐LCWE injection (n = 11–15). ***P ≤ 0·01 by Student’s t‐test. Ao = aorta.

Figure 4.

Immune cells infiltrate the cardiac ganglia during Lactobacillus cell wall extract (LCWE)‐induced Kawasaki disease (KD) vasculitis. Mice were injected with phosphate‐buffered saline (PBS) or LCWE and killed 1 week later. Immunofluorescence was performed on frozen sections to determine immune infiltrate constituents. (a) Isotype control image of ganglia. (b) Isotype control image of coronary lesion. (c,d) Ganglion (c) and coronary lesion staining (d) for CD45.1 (leukocyte common antigen, red arrows) and CD19 (B lymphocytes marker, green arrows). (e,f) Ganglion (e) and coronary lesion staining (f) for CD3 (T cell marker, red arrows) and F4/80 (macrophage/monocyte marker, green arrows). (g,h) Ganglion (g) and coronary lesion staining (h) CD11b‐ (myeloid cells, green arrows) and for CD11c (dendritic cell and macrophages, red arrows). Scale bars 10 µm for ganglia and 25 µm for coronary or magnification as indicated.

Cardiac neural remodeling in the LCWE‐induced KD vasculitis mouse model is IL‐1‐dependent

ECG changes could also be explained by neural remodeling in the myocardium. To determine if neuronal remodeling occurs in the myocardium of LCWE‐injected mice we stained the tissue for GAP43, a marker of axonal growth cones in dendrite processes of synaptic neurons. GAP43 staining was increased in LCWE‐injected mice compared with PBS‐injected mice (Fig. 5a,b). GAP43 is a marker for both sympathetic and parasympathetic nerves; to differentiate, we next stained for tyrosine hydroxylase (TH), a marker of sympathetic neurons. The density of TH⁺ neurons was increased in the LCWE‐injected mice compared with PBS‐injected mice (Fig. 5a,c). Il1r1^−/− mice were resistant to the LCWE‐induced increase in GAP43 and TH staining (Fig. 5a–c), indicating that IL‐1 signaling is also required for neural remodeling in this model.

Figure 5.

Neural remodeling in myocardium after Lactobacillus cell wall extract (LCWE)‐induced myocarditis is (IL)‐1‐dependent. wild‐type (WT) and Il1r1^−/− mice were injected with LCWE and killed 2 weeks later. Immunofluorescence was performed on frozen section. (a) Immunofluorescence staining of the neuronal sprouting marker growth‐associated protein 43 (GAP43) and the sympathetic neuronal tyrosine hydrogenase (TH) in the myocardium. Scale bars, 100 µm. (b,c) Quantification of staining density of GAP43 (b) and TH (c) in the myocardium. n = 4, 5 and 7 for WT + phosphate‐buffered saline (PBS), WT + LCWE and Il1r^−/− + LCWE groups, respectively. P < 0·05 by one‐way analysis of variance (anova) with Bonferroni post‐test analysis.

Discussion

Between 40 and 80% of KD children show ECG abnormalities, which may lead to the development of fatal arrhythmias 38, 39, 40. Indeed, ventricular arrhythmias are increasingly being recognized as a cause of death in children in KD 41. Tachycardia has been reported as a potential risk indicator for coronary arterial lesion in KD 42, and inflammation with cell infiltration into the conduction system (atrioventricular conduction) has also been observed in autopsied KD hearts 43. In the current study we show that, in addition to aortitis, coronary arteritis, aneurysms and myocarditis, the LCWE‐induced KD vasculitis mouse model is also associated with acute ECG changes similar to those reported in KD children.

Among the ECG changes that have been reported in patients with KD, several studies have shown an increased risk for ventricular fibrillation (arrhythmia) with increased dispersion of repolarization times, as calculated from ECG 23, 24, 25, 26. A study of a North Indian cohort reported that children with KD had a significantly higher QTc dispersion 27. The mean QT interval is higher in KD patients compared with controls 44, and increased QT dispersion correlates with the severity of coronary lesions in KD 45. Similarly, QTc end dispersion is increased in patients with acute myocardial infarction 24 and significantly higher in KD patients with ventricular fibrillation 46. QT interval dispersion reflects the heterogeneity of ventricular repolarization, and specific changes in QT dispersion such as prolonged QTc have been shown to precede ventricular fibrillation 24. Here, we show evidence that prolonged QT interval is also observed in the LCWE‐induced KD mouse model. However, we were unable to measure QT dispersion in this experimental model because of a software limitation.

We also show for the first time, to our knowledge, that the LCWE‐induced KD murine model is associated with the development of cardiac ganglionitis. Even if some disparities are observed at the level of ganglion numbers and their distribution, the murine ganglionated plexus appears almost similar to the human plexus 47. The ganglia examined in this study belong to the cardiac neural plexus, which have been demonstrated as parasympathetic nerves 47, whereas the sympathetic nerves going to the heart are usually located in the cervical ganglia and supra‐thoracic ganglia 47.

Several studies suggest that KD vasculitis etiology maybe of viral origin 4; this hypothesis is supported by the seasonality of KD outbreaks and the age specificity of KD development, with children aged from 6 months to 5 years being at greatest risk 48, 49. Of interest, a wide range of viruses could potentially elicit immune responses resulting in cardiac ganglionitis development. Therefore, future studies should investigate if cardiac ganglionitis is also present in autopsy tissues obtained from KD patients.

A relationship between stellate ganglionitis and arrhythmia has previously been reported in the context of catecholaminergic polymorphic ventricular tachycardia and long QT syndrome 37. In that setting, ganglionitis is postulated to boost adrenergic activity, and thus trigger or enhance electric instability in patients who are already genetically predisposed to arrhythmias. Similarly, the ganglionitis observed in the LCWE‐induced KD vasculitis mouse may enhance electric instability of myocardium where immune cells have already infiltrated and caused myocarditis.

In this study, we did not observe evident signs of a permanent effect on the cardiac conduction system during LCWE‐induced KD vasculitis, as the electrophysiological abnormalities tend to disappear 3 weeks after LCWE injection. However, longitudinal cardiac screenings performed on Japanese students with or without KD history demonstrate not only findings of persistent myocardial changes and interstitial fibrosis resulting from acute myocarditis, but also long‐lasting neuronal remodeling and a permanent effect in the conduction system of the myocardium, as demonstrated by a higher proportion of abnormal ECG findings among adolescents with KD 50, 51.

We and others have previously demonstrated that enhanced IL‐1 signaling plays a critical role in the development of coronary arteritis, aortic aneurysms and myocarditis lesions in murine models of KD vasculitis 12, 15, 16, 29. Here, we demonstrate that activity of this central cytokine pathway also underlies the development of ganglionitis, neural remodeling and cardiac rhythm abnormalities. A role for IL‐1 signaling in arrhythmia is in agreement with studies in patients with connective tissue diseases (CTDs), who exhibit elevations in IL‐1β in concert with prolonged QTc intervals 52. Furthermore, experiments on pig and mouse ventricular cells have demonstrated the ability of cytokines to prolong action potential duration, possibly by enhancing ionized calcium 53. Similarly, neural remodeling is associated with prolonged action potential duration and longer QTc intervals 54.

Serum levels of NGF are strikingly increased in the acute phase of KD in patients compared with controls 36. Furthermore, NGF was found to be up‐regulated in KD coronary vessels and was detected in inflammatory cells infiltrating the adventitia and myocardium in KD tissues 55. In addition to its neurotropic function, NGF has been recognized as an important mediator of inflammation 56, 57, 58, 59. IL‐1β is known to induce evidence of neuroimmune cross‐talk NGF release in various immune cells 34, 60, 61. We found that IL‐1 signaling was necessary for the increase in serum NGF in the LCWE‐mediated KD model, as NGF was not elevated in mice deficient in IL‐1α or IL‐1β, or in those treated with IL‐1Ra (anakinra).

While no animal model can fully mimic human disease, the LCWE‐induced KD mouse model has been widely accepted in the Kawasaki research community as the most reliable experimental model of KD 10, 12, 15, 16, 62. Overwhelming evidence already suggests that this model mimics the histopathological changes observed in human KD pathology, including the epicardial proximal coronary involvement and the presence of luminal myofibroblast proliferation 10, despite one published opinion arguing against it 63. Moreover, as in children with KD, coronary arteritis in LCWE‐treated mice responds to therapy with IVIG 11 and is suppressed by treatment with anti‐tumor necrosis factor (TNF)‐α antibody 12. Like human KD, we have shown that this model is also associated with systemic inflammation, increased body temperature and pyrogens such as IL‐1β and prostaglandin E₂ (PGE₂) 12. The model recapitulates key features of the human disease, including the role of dendritic cells (DCs) 16 and Toll‐like receptor (TLR)‐2/myeloid differentiation primary response 88 (MyD88) signaling 62, as well as IL‐1α and IL‐1β 12, 15, 16. Importantly, the demonstrated beneficial effect of IL‐1R antagonist (anakinra) in the LCWE mouse model 12 led to the two anakinra clinical trials in KD patients (NCT02179853) that recently ended with plans to start Phase III soon.

We have previously shown that anakinra prevents myocarditis, coronary arteritis and aneurysm formation in the LCWE‐induced KD vasculitis murine model 12, 29. Here, we show that blocking the IL‐1 signaling pathway with anakinra also prevents and improves KD‐induced electrophysiological and ECG changes that occur during the acute stages of KD. We also demonstrate that immune cells infiltrate the cardiac ganglia and nerves. These findings are compelling, as they add to our understanding of cardiac dysfunction during KD vasculitis and may potentially explain the observations of arrhythmia in this condition.

In summary, our current observations show that the LCWE‐induced KD mouse model is associated with electrophysiological and ECG abnormalities similar to those reported in patients with KD, and that these are dependent on the elevation of IL‐1 signaling characteristic of the disease. These findings reinforce the value of the LCWE‐induced KD vasculitis mouse model, and broaden the potential therapeutic benefits of IL‐1 blocking agents for KD patients.

Disclosures

None to declare.

Author contributions

M. A., D. D. R, A. C. G., E. C., Y. L. and M. C. F. performed the studies. M. A., D. D. R., T. J. A. L., K. S., S. C., T. R. C., M. N. R. and M. Arditi designed the study. M. A., D. R., P. S., T. R. C., M. N. R., and M. Arditi wrote the manuscript.