Main

Kawasaki disease (KD) is a systemic vasculitis syndrome which preferentially affects infants and children (1,2). Past large epidemics in Japan, peaked incidence at 9–11 mo of age, which coincides with waning of maternal immunity and symptoms partly similar to other infectious disorders suggest that some microorganisms may trigger the disease. However, despite large efforts, the cause of this mysterious disease still remains unknown. Contrary to their initial expectation of a benign illness, coronary artery lesions (CALs) developed in 15–25% of untreated patients has made KD a leading cause of acquired heart disease of children in the developed countries (3). Intravenous gamma globulin (IVIG) therapy dramatically reduced occurrence of CALs, however, about 15% of the patients poorly respond to the treatment and are at higher risk for CALs. Following epidemiologic findings have suggested that genetic predisposition might underlie the etiology of KD. First, KD is much common in East Asian countries. In Japan, incidence of KD is continuously increasing and in recent years, more than 180 per 100,000 children younger than 4 y are affected annually (4).

The incidence is 10–20 times higher than that of Western countries. The same level of incidence in Japanese ancestries living in Hawaii (5) is indicating that the predilection for oriental populations might not be due to geographical reasons. Second, KD has familial accumulation. Relative risk for sibs (λs) is about 10 and recent study revealed that two-generation KD patients exist more than expected (6,7). Thus, KD can be considered as a complex multifactorial disease (Fig. 1). Identification of genetic factors related to individual susceptibility or different incidence among ethnicity might provide a clue to unravel the enigma of KD. Recent advances in molecular genetics have greatly accelerated identification of susceptibility genes for complex diseases. Currently, there are two mainstreams of strategy for identification of disease genes. One is candidate gene approach and the other is genome-wide approach.

Figure 1
figure 1

Multiple factors that could be linked to Kawasaki disease.

Full size image

Candidate Gene Approach

Most genetic studies of KD have focused on candidate genes. This approach is derived from “functional cloning” used in searching for genes responsible for monogenic disorders. Genes for analyses were selected based on the information of their known function or role in the disease pathophysiology.1

HLA

Human leukocyte antigens (HLA) are located on the chromosome 6p21.3 region and their polymorphisms are associated with various diseases directly or indirectly as a consequence of linkage disequilibrium (LD) between polymorphisms of other neighboring genes. There are large differences in distribution of HLA alleles among races or ethnicity. Initial genetic studies of KD were focused on HLA class I antigens. Matsuda et al. (8) and Kato et al. (9) reported that Japanese specific variant of HLA-Bw22 (now called as Bw54) was predominant in Japanese KD patients. In the studies of white (10,11) and Jewish population (12), association between HLA-Bw51 and KD was found. However, neither trend of association was replicated in the Southern Chinese population (13) and in the Korean population (14). One of the reasons that KD has been regarded as infection-triggered disorder is past epidemic in Japan and in the United States. HLA-B44 was predominant in the epidemic KD cases in Boston (11). This finding was supported by the report by Kaslow et al. (15) in which HLA-A2, B44 and Cw5 antigen combination was associated with KD patients involved in an outbreak in Maryland. On the other hand, Harada et al. (16) analyzed sharing of HLA-A-B-C-DR haplotypes between 23 affected sib pairs of Japanese KD patients and failed to find evidence of association. Several groups have studied relationship between KD and HLA class II antigens, however, no significant association has been observed until now (17–19). Thus, a single allele or haplotype of HLA that confers KD susceptibility has not been identified yet. Previous studies about association between HLA and KD were summarized in Table 1. Considering that no evidence of linkage was observed near 6p region in the genome-wide affected sib-pair (ASP) study of KD (20), the involvement of HLA in KD pathogenesis might be limited or more complex than expected.

Table 1 Association studies between HLA and KD
Full size table

Non-HLA Genes

Tumor necrosis factor-α (TNF-α) is a proinflammatory cytokine mainly produced by macrophages. The possible involvement of this cytokine in vascular damage and elevated serum level in the acute phase KD patients who developed CALs suggests its key role in the disease pathogenesis (21,22). Recent case series suggesting successful treatment using MAb against TNF-α (infliximab) on IVIG resistant patients (23–25) back up this idea. Association of single nucleotide polymorphisms (SNPs) that may alter TNF-α expression with various inflammatory diseases has been extensively studied. Kamizono et al. (26) conducted association study about five SNPs (−1031 T>C, −863 C>A, −857 C>T, −308 G>A and −238 G>A) located at 5′ flanking region of TNF-α gene in the Japanese population. In this study, no association was observed between these polymorphisms and KD occurrence or CALs formation. Quasney et al. (27) investigated the distribution of + 250 A>G of lymphotoxin-α gene in addition to −308 G>A of TNF-α and found positive association of TNF-α −308 A/G genotype with CALs in white KD patients. Chien et al. (28) and Ahn et al. (29) reported negative association between polymorphisms in TNF-α gene and susceptibility to KD or CALs in the Taiwanese and the Korean populations, respectively. In a recent report by Cheung et al. (30), TNF-α −308 A allele was associated with KD in the Chinese population. Cheung also showed that this SNP was associated with intima-media thickness of carotid arteries.

Burns et al. (31) applied family-based association study to KD for the first time. Ninety-five SNPs in 58 candidate genes for cardiovascular diseases and inflammation were selected and their transmission patterns in 269 trios of KD patients and their parents were investigated. They found a SNP in the promoter region of interleukin (IL)-4 gene was asymmetrically transmitted. IL-4 is located in the cytokine gene cluster on 5q31 region where positive linkage signal was observed in a sib-pair linkage study of KD (20). However, replication study performed by two Taiwanese groups failed to validate the finding (32,33). IL-4 is known to promote differentiation of naive T cells into Th2 cells. Elevated serum IL-4 in the acute phase of KD was reported, and many aspects of immune response in KD pathophysiology are associated with Th2 reaction, hence the cytokine could play a role in the pathogenesis of KD. Further investigation is needed to understand the involvement of the SNPs of IL-4 gene in individual susceptibility to KD. Association studies focused on SNPs or copy number variations of genes for other cytokines, as well as chemokines and their receptors have also been conducted (32,34–40; Table 2).

Table 2 Candidate genes previously tested for association with KD
Full size table

Genes related to vasoactive or angiogenic molecules also can be considered as candidates for KD susceptibility or severity. Vascular endothelial growth factor (VEGF) is expressed in various types of cells including leukocytes and vascular smooth muscle cells. Binding of VEGF to its receptor (VEGFR-1 and VEGFR-2) expressed on endothelial cells induces cell proliferation, survival, migration, and angiogenesis. Its ability to induce vascular hyperpermeability and chemotaxis of bone marrow-derived cells suggest significant roles of VEGF in inflammation (41). VEGF is up-regulated in the acute phase of KD and the serum level of this protein is associated with formation of CALs in one study (42). Kariyazono et al. (43) reported a SNP in the 5′ untranslated region (UTR) of VEGF (rs2010963) was associated with CALs in the Japanese population. In the study of Dutch KD patients, 2 SNPs other than rs2010963 showed association with KD (44). With regard to association of rs2010963 with KD or CALs, two inconsistent results were also reported by Taiwanese groups (45,46).

Association of the insertion/deletion (I/D) polymorphism of a gene for angiotensin I converting enzyme with cardiovascular diseases and hypertension has been extensively examined. This polymorphism is located in intron 16 and is in strong LD with other SNPs in the 5′ and 3′ regions of the gene. Takeuchi et al. (47) reported association of I/I genotype with CALs in the Japanese. Taiwanese and Korean groups reported I allele or I/I genotype was associated with KD susceptibility but not with risk for CALs (48,49). Fukazawa et al. (50) reported that the D allele in concert with the C allele of the SNP in the 3′ UTR (+1166 A/C) of angiotensin II type I receptor gene increased risk for coronary artery stenosis in another Japanese cohort.

Matrix metalloproteinases (MMPs), produced by a variety of cell types, play important roles in various physiologic and pathologic processes by degrading extracellular matrices. The activity of MMPs is controlled by their endogenous inhibitor, the tissue inhibitors of metalloproteinases (TIMPs). Imbalances between MMPs and TIMPs are related to pathologic conditions such as arthritis, tumor metastasis, and aortic aneurysms. MMPs and TIMPs are elevated in the serum or vascular tissue of acute KD patients and association of increased MMP9/TIMP2 and MMP3/TIMP1 ratios with risk for CALs formation was reported (51). Furuno et al. (52) identified that promoter polymorphisms of TIMP2 gene was associated with increased risk of CALs. 6A allele of an insertion/deletion polymorphism (5A/6A) of MMP-3 gene promoter was more predominant in the Korean KD patients with CAL (53). Although not significant, the 5A allele showed a trend of association with CALs in the Japanese population (54). Another promoter SNP of MMP-3 was not associated with KD or CALs in the Korean population (55). Ikeda et al. (54) conducted association study of five functional polymorphisms of MMP-2, 3, 9, 12, and 13 genes and observed significant association between a promoter SNP or a haplotype of MMP-13 and CALs.

There has been growing interest in involvement of innate immune system in the pathophysiology of inflammatory diseases. Genes for pattern recognition receptors (PRRs) which recognize pathogen-associated molecular patterns and downstream signaling peptides which cause cytokine production through activating nuclear factor kappa-B (NF-kB) and/or caspase-1 have become an active area of genetic research. As for KD, positive association of the SNPs in the promoter of CD14, a PRR for LPS expressed on monocytes/macrophages and neutrophils with CALs was reported (56). Mannose binding lectin (MBL) and C-reactive protein (CRP), known as acute phase proteins produced in liver, also have roles in innate immune reaction. MBL can initiate complement activation directly by binding to mannose and n-acetyl glucosamine residues on the surface of microorganisms. Ligands of CRP are phosphocholine on microorganisms or damaged cell membranes. CRP also initiates complement activation by binding to C1q. Association between functional polymorphisms of MBL gene and CALs in KD patients younger than 1 y was described (57,58). Cheung et al. (30,59) reported that polymorphisms in MBL and CRP were associated with arterial stiffness or intima-media thickness. Ouchi et al. (60) reported association between a GT repeat polymorphism in 5′ region of solute carrier family 11 (proton-coupled divalent metal ion transporters), member 1 (SLC11A1) gene, and KD. SLC11A1 is a divalent cation (Fe2+, Zn2+, and Mn2+) transporter on the endosomes and regulates macrophage activation. Association of polymorphisms in genes for 5,10-methylenetetrahydrofolate reductase (61), Fc gamma receptors (62,63), UDP glucuronosyltransferase (64), heme oxygenase (64), MHC class I polypeptide-related sequence A (65), platelet-activating factor acetylhydrolase (66), NO syntheses (67), and CD40 ligand (68,69) with KD or phenotypes related to KD have also been studied. Previous association studies of candidate genes are summarized in Table 2.

Overview of Previous Candidate Gene Studies

Several candidate genes have been examined in independent cohorts of the same or different ethnicity, however most of the results were conflicting (Table 2). Most of the previous studies have been carried out with a single small cohort of KD patients and findings were reported without validation in additional case-control sets. This step of confirmation is crucial to rule out type I error (false positive). Population stratification in the first association report also can result in irreproducible results. Correction for multiple comparisons, which is essential for genome-wide approach described in the following paragraph, has to be conducted also in candidate gene studies. Recruiting sufficient number of the patients is indispensable not only to reduce false-negative conclusions but also to obtain association, which remains significant after correction. Compared with the other common diseases in the adults, there are difficulties in collecting DNA samples from young patients. Methods to extract genomic DNA from specimens other than blood (mouthwash, saliva, hairs, and nail clippings) and a multicenter collaboration that allows us to increase size of the cohorts and to validate the association in multiple sample sets may be beneficial.

Sometimes association is not replicated in different ethnicity even if the initial observation was definitely validated in the cohorts of the same ethnic group. One of the reasons is different relative importance of particular genetic variants in susceptibility of complex diseases between ethnicity. Difference in LD or haplotype structures among ethnicity may also lead conflicting results. Most of the replication studies tend to focus on only one or several variations significant in previous studies. When the initial positive association was indirect observation due to LD between the markers tested and a particular gene or variation, the association would not be detected in a different ethnic group in which the responsible variation is not linked with the same markers. Therefore, researchers of both discovery and replication studies should be careful in selecting variations to test.

Genome-wide Approach

In contrast to candidate gene approach, which is based on assumption, a strategy for searching disease causing mutations or variations from the whole genome relying only on the positional information is called genome-wide approach. Currently, we and another group are searching for susceptibility genes for KD by this approach predominantly studying Japanese and white KD patients, respectively. In this section, I introduce our recent accomplishments in genome-wide linkage study using microsatellite markers (20) and linkage disequilibrium mapping using SNPs (70).

Linkage Study

In linkage studies, genetic markers linked to diseases are searched by analyzing transmission patterns of the markers within families of multiple patients. Owing to the short history of KD, there is no large kindred with KD available for linkage analysis. Therefore, we applied ASP method, a kind of nonparametric linkage study suitable for investigating genetic factors of complex diseases. ASP method can be conducted only by analyzing DNA samples of siblings affected (and their healthy parents if possible). We have recruited more than 80 families including sibling cases all over Japan and genotyped about 400 polymorphic microsatellite markers. The lod scores were calculated by estimating numbers of shared alleles identical by descent for each marker loci. As a result, in 10 chromosomal loci (4q35, 5q34, 6q27, 7p15, 8q24, 12q24, 18q23, 19q13.2, Xp22, and Xq27) positive linkage signals (maximum lod score > 1.0) were observed (20). Among these candidate loci, 12q24 region showed the most significant evidence of linkage (maximum lod score = 2.69).

Linkage Disequilibrium Mapping

Because of limited number of mitosis analyzed in ASP method, the resolution of mapping achieved was relatively low and the responsible genes stand in the crowd of a hundred and several tens of genes around each linkage peak. We tried to narrow down the candidate loci and identify susceptibility genes by LD mapping using SNPs. SNPs distributed in 10–25 Mbp area surrounding the linkage peaks were selected from the database and 94 KD cases and 564 controls were genotyped. To increase the power of screening, patients with a positive family history of KD was chosen by priority. In the systematic SNP screening by case-control association study, we identified an LD block spanning 150 kb on 19q13.2 region containing three SNPs significantly associated with KD. Subsequent validation study using different case-control sample sets confirmed the association of the SNPs. Resequencing of the block and subsequent case-control study identified additional six SNPs significantly associated with KD. These candidate SNPs were further screened by transmission disequilibrium test conducted on American multiethnic KD patients–parents trios. Because of difference in LD structure among ethnicity, of the nine significant SNPs, only four centromeric SNPs showed the same trend of association (Fig. 2).

Figure 2
figure 2

Genes and significant SNPs on the LD block identified in the chromosome 19q13.2 region. Genes oriented q terminus to p terminus are in upper row, with genes in the opposite orientation shown below. Arrowheads indicate the position of SNPs significantly associated with KD. *SNPs associated with KD in Japanese. **SNPs associated with KD both in Japanese and Americans. †SNPs associated with KD in Japanese but not in Americans. Adapted from Onouchi et al. (70) with permission.

Full size image

Identification of Inositol 1,4,5-trisphosphate 3-kinase C as a Susceptibility Gene for KD

Among the four genes in which the significant four SNPs were located, we focused on the known function of ITPKC as a kinase of inositol 1,4,5-trsiphosphate (IP3). IP3 is a second messenger molecule in various types of cells including T cells, macrophages, and neutrophils involved in the pathogenesis of KD. IP3 transduces signals from cell surface receptors (71) and, in T cells, plays an important role in signal transduction of Ca2+/nuclear factor of activated T-cells (NFAT) pathway (72–74). Although three isoenzymes of ITPK proteins (ITPKA, ITPKB, and ITPKC) have been known, the relative importance of ITPKC in the immune system was unclear. We investigated mRNA expression of ITPKC in various normal tissues and revealed that ITPKC was greatly induced in PBMCs when activated. Compared with the other isoenzymes, ITPKC was most abundantly expressed in PBMCs and leukemic cell lines and most significantly induced in response to cell stimulation (70). Therefore, we speculated that ITPKC might be related to inflammation. Previous in silico study (75), which has predicted NF-kB binding sequence in the promoter of ITPKC, supported the idea that ITPKC is a potent immune gene.

Negative Regulatory Role of ITPKC in T-cell Receptor Signal Transduction

Overexpression of ITPKC in Jurkat cells resulted in reduced NFAT activation and IL-2 expression. Conversely, knock down of ITPKC by short hairpin RNA enhances NFAT activity and IL-2 expression (70). These data indicate that ITPKC acts as a negative regulator of Ca2+/NFAT pathway in T cells by modulating the amount of IP3. Through functional analyses of significant SNPs, we clarified that itpkc_3 G/C (rs28493229), a SNP located in intron 1 of ITPKC, reduces mRNA expression of ITPKC in PBMCs down to 70% by altering splicing efficiency (70). Increased stability of secondary structure of pre-mRNA and reduced binding affinity of some RNA binding proteins related to splicing by the nucleotide change were though to underlie these observations. Reduced ITPKC activity associated with C allele of itpkc_3 may lead enhanced T-cell activation in the pathophysiology of KD. The proposed role of ITPKC and itpkc_3 in Ca2+/NFAT pathway in T cell was demonstrated in Fig. 3. In mice, itpkB (not itpkC) and IP4 play important roles in development of thymocytes and selection and activation of B cells (76–78). Investigation of the role of human ITPKC and biologic significance of itpkc_3 in other immune cells (e.g., macrophages, B cells, and neutrophils) or nonimmune cells (e.g., endothelial cells and cardiac myocytes) may lead further understanding of the pathogenesis of KD.

Figure 3
figure 3

Functional significance of itpkc_3 on ITPKC mRNA and Ca2+/NFAT pathway. (A) Effect of itpkc_3 C allele on splicing of ITPKC pre mRNA. The C allele of itpkc_3 reduces splicing efficiency of IPTKC premRNA. mRNAs harboring unspliced intron 1 cannot be translated properly and will be degraded early by nonsense-mediated decay mechanism. (B) Proposed role of ITPKC as a negative regulator of Ca2+/NFAT pathway. When the T-cell receptor (TCR) is bound by antigen/MHC complex on antigen presenting cells (APCs), adaptor molecules and kinases are recruited and phospholipse C-γ1 (PLC-γ1) is activated by phosphorylation of its tyrosine residue. IP3 and diacylglycerol (DAG), another second messenger molecule, are generated by hydrolysis of phosphatidylinositol 3,4-bisphosphate (PIP2) by activated PLC-γ1. IP3 binds to its receptor expressed on endoplasmic reticulum (ER) membrane and causes the release of Ca2+ into the cytoplasm. Then depletion of Ca2+ store in ER evokes a process termed as store operated Ca2+ entry in which extracellular Ca2+ enters through calcium release-activated Ca2+ channels on the plasma membrane. Recent advances in research identified the role of stromal interaction molecule (STIM) as a sensor of Ca2+ in ER and ORAI as a calcium release-activated Ca2+ channel. Cytoplasmic Ca2+ binds calmodulin, which in turn activates calcineurin, a calmodulin-dependent phosphatase. Activated calcineurin dephosphorylate NFAT in the cytoplasm and lead nuclear translocation of NFAT. NFAT in the nucleus drives transcription of genes important in T cell activation as a homodimer or heterodimer with other transcription factors. AP1 is one of the transcription partners of NFAT, which is activated by a signal from TCR mediated by DAG (72–74). Reactions and amounts of molecules increased by the effect of itpkc_3 C alleles were represented by red characters and arrows and those reduced by blue, respectively. [Ca2+]i: intracellular free Ca2+ concentration.

Full size image

Association of itkpc_3 with CAL Formation and IVIG Responsiveness

As 15% of KD patients poorly respond to IVIG and these patients are at higher risk for developing CALs, how to predict patients' risk for CALs and responsiveness to IVIG and start additional or alternative therapy before they are destined to develop CALs is the most pressing issue. The C allele of itpkc_3 was more predominant in patients with CALs and those refractory to IVIG (Table 3, Ref. 70). Considering that these phenotypes somewhat depend on severity of inflammation, our observations may be reasonable. In Japan, experience of treating IVIG resistant KD cases successfully with Cyclosporin A (CsA) is accumulating (unpublished observations). CsA is an immunosuppressant targeting calcineurin that dephosphorylate and lead nuclear translocation of NFAT. Proof of effectiveness of CsA in the treatment of KD would provide additional support for the importance of Ca2+/NFAT pathway activation in the pathogenesis of KD.

Table 3 Association between itpkc_3 and KD
Full size table

There should be genetic variations which directly affects vascular elasticity or anti-inflammatory mechanism of IVIG and increase individual risk for a more severe course when affected with KD. Large-scale association study between KD patients with CAL and those without CAL, as well as between IVIG responders and nonresponders will identify a set of genetic markers efficiently and will contribute in future evidence-based and personalized medicine.

Conclusion

Completion of human genome project and information of frequencies and LD of the SNPs provided by International HapMap project has made genome-wide approach of investigating genes for complex diseases quite feasible. Genome-wide association study (GWAS) with platforms by which 500 thousands to 1 million SNPs can be genotyped at a time has become a mainstream. During the next few years, multiple susceptibility genes may be identified from such large-scale studies. We can also hope that new insight in the pathophysiology of KD may be derived from the findings in hypothesis-free genetic research. Abovementioned genome resources may also contribute to candidate gene approach. Selecting tag SNPs that represent haplotypes by the information of LD may facilitate an efficient and quality screening of candidate genes. Pathways and molecular networks in which susceptibility genes identified in genome-wide approach are involved will expand the search range of candidate genes. It is suggestive that association of the SNPs in some downstream genes of NFAT (TNF-α, CD40L, and IL-4) with KD and/or outcome has been reported.

Although much remains to be done, it is hoped that both genetic approaches will complement one another in clarifying the genetic background of KD, open the door to elucidation of the etiology, and allow for establishment of new therapeutic and preventive strategies.