In this study, we used HCAECs, as a mimic of KD, to study the possible mechanisms responsible for the clinical benefits of adding a corticosteroid to standard IVIG therapy for patients with severe KD. We first examined the effects of high-dose IgG and DEX, a synthetic corticosteroid, on cellular damage to HCAECs caused by inflammatory stimuli. The degree of cellular damage was evaluated by the level of HMGB1 protein released by HCAECs in response to stimulation with three inflammatory cytokines, TNF-α, IL-1α and IL-1β. We found that DEX, but not IgG, significantly inhibited the release of HMGB1 by HCAECs (Fig. 1a). Furthermore, there was no significant change in HMGB1 mRNA expression levels in HCAECs under any of the test conditions (Fig. 1b). This suggested that the elevated HMGB1 protein in the culture supernatants was not newly-synthesized protein, but passively-released protein due to cellular damage caused by the inflammatory cytokine stimulation. In fact, the TNF-α-induced HMGB1 release from HCAECs was not inhibited by treatment with 1 µM monensin A (Golgi-Stop reagent; data not shown). Consistent with previous reports using human umbilical vein endothelial cells (HUVECs) [22, 23], DEX effectively inhibited endothelial cell apoptosis by reducing caspase 3/7 activities in HCAECs (Fig. 1c).
Interestingly, a Korean research group recently reported that HMGB1 single nucleotide polymorphisms (SNPs) were significantly associated with both IVIG resistance and CAL formation in Korean KD patients, but not with KD susceptibility . Those findings suggest that the amount of HMGB1 released from damaged endothelial cells might be related to the severity and complications in KD patients, but not their susceptibility to KD. As far as we examined, HMGB1 failed to directly induce an inflammatory response by HCAECs (data not shown). However, once released into the extracellular milieu, HMGB1 reportedly activated monocytes/macrophages to produce multiple proinflammatory cytokines [24, 25] and exerted several inhibitory effects on regulatory T cell activities [26, 27]. Thus, extracellular HMGB1 might act on various types of leukocytes, perhaps leading to KD aggravation. Blood HMGB1 levels may reflect the degree of coronary vascular endothelial cell damage in KD patients. Accordingly, stratifying patients by adding the blood HMGB1 level to the existing risk score(s) for predicting IVIG resistance may increase the probability of success of combination therapy consisting of IVIG plus a corticosteroid.
Unlike HMGB1, IL-1α—another DAMP—was significantly induced in HCAECs at the mRNA expression level by inflammatory stimuli (Fig. 2b vs. Fig. 1b). Furthermore, consistent with the results of qPCR, DEX effectively inhibited cytokine-induced intracellular IL-1α protein (Fig. 2c). Similar to HMGB1, TNF-α-induced IL-1α release by HCAECs was not inhibited by treatment with 1 µM monensin A (Golgi-Stop reagent; data not shown), indicating that intracellularly accumulated IL-1α protein was passively released from damaged HCAECs. Therefore, we speculate that the decrease in IL-1α protein seen with DEX (Fig. 2a) was due to the combination of DEX’s suppression of IL-1α mRNA expression (Fig. 2b) and its anti-cytotoxic effect on HCAECs (Fig. 1). Notably, IL-1α can induce a strong inflammatory response (IL-6 and G-CSF production) comparable to that seen with IL-1β, even at lower concentrations compared to TNF-α (Additional file 1: Fig. S1). In order to compare and evaluate the efficacy of corticosteroid and IgG under conditions with similar levels of inflammation and cytotoxicity, 100 ng/ml of TNF-α and 10 ng/ml of IL-1s were used as inflammatory stimuli in this study.
Although we previously reported that IVIG treatment hardly inhibited IL-1β-induced IL-6 and G-CSF production , it should be noted that IL-1α stimulation resulted in IVIG resistance (Fig. 3a, lower graphs). Several recent studies reported an association between IL-1s and IVIG resistance in KD patients. In a microarray study using whole-blood RNA, IL-1-associated signaling pathways were upregulated in IVIG-resistant KD patients compared to IVIG-responsive patients . Two previous case reports suggested a beneficial effect of anakinra (an IL-1R antagonist that blocks the activity of both IL-1α and IL-1β) on IVIG-resistant KD [29, 30]. Based on those findings, clinical trials of IL-1 blockade for IVIG-resistant KD patients are being conducted in Western Europe and the USA .
Like IL-1s, TNF-α has been reported to be involved in the pathogenesis of KD. Serum levels of TNF-α were significantly elevated and correlated with the incidence of CALs in acute KD patients [32, 33]. Furthermore, TNF-α blockade effectively prevented the development of coronary vasculitis in murine models of KD [34, 35]. In fact, a clinical trial of an anti-TNF monoclonal antibody (mAb) showed clinical effectiveness, including reduced fever duration and CAL formation . Thus, although both TNF-α and IL-1s have been clearly implicated as key pathogenic cytokines in KD, there is currently limited understanding of whether their roles are distinct or overlapping. Stock et al. recently addressed this issue in a murine model of KD and provided evidence that TNF-α and IL-1s play temporally distinct and non-redundant roles in driving cardiac inflammation . Specifically, TNF-α, but not IL-1s, was essential for the development of acute-phase myocarditis, whereas IL-1s were indispensable for the subsequent development of coronary vasculitis . These findings suggest the possibility that TNF-α is more critical for the onset of KD, whereas IL-1s may be more crucially involved in the progression and prognosis of KD than TNF-α. Taken together with our present findings, administration of a corticosteroid as early as possible may contribute to suppression of KD progression by inhibiting the expression and/or release of IL-1s.
Corticosteroids are widely used as potent anti-inflammatory drugs to treat various inflammatory diseases. As a preliminary experiment, we examined for concentration-dependency of DEX’s inhibitory effects on IL-6 and G-CSF production and IL-1α release induced by inflammatory stimuli (Additional file 2: Fig. S2). We found that the inhibitory effects of DEX were indeed concentration-dependent, but they almost reached a plateau at 100 nM to 10000 nM of DEX. When the blood concentration of corticosteroid used in the RAISE study  is converted to DEX on the basis of the titer, it is about 10000 nM. However, sufficient effects were observed even at 100 nM and 1000 nM DEX in our in vitro experiment (Additional file 2: Fig. S2), and for that reason we used 1000 nM DEX in this study.
Corticosteroids are known to suppress nuclear factor kappa B (NF-κB), which promotes transcriptional activation of various inflammatory genes, including IL6 [38-40]. We previously demonstrated that IVIG did not inhibit activation of NF-κB, whereas it significantly inhibited activation of another transcription factor, CCAAT/enhancer-binding protein delta (C/EBPδ) , as well as C/EBPβ . Therefore, synergistic effects between a corticosteroid and IgG seem likely because their anti-inflammatory mechanisms apparently involve non-overlapping pathways. Indeed, these drugs were more effective in suppressing IL-6 production and IL-1α release when added immediately after the inflammatory stimulation (Fig. 4). Thus, adding a corticosteroid to standard IVIG therapy at an early stage of inflammation in KD patients may have a better anti-inflammatory effect by inhibiting both KD-related cytokine production and release of IVIG-refractory factors, including HMGB1 and IL-1α.