A novel anti-peroxiredoxin autoantibody in patients with Kawasaki disease


Mikiya Fujieda, Department of Pediatrics, Kochi Medical School, Kochi University Kohasu, Oko-cho, Nankoku, Kochi 783–8505, Japan. Tel: +81 88 880 2355; Fax: +81 880 2356; email: fujiedam@kochi-u.ac.jp


Antibodies to the anti-oxidative peroxiredoxin (Prx) enzymes occur in both systemic autoimmune disease and vasculitis in adulthood. Because increased oxidative stress induces vasculitis in Kawasaki disease (KD), autoimmunity to Prxs in patients with KD was investigated. The presence of antibodies to Prx 1, 2 and 4 was analyzed by ELISA and Western blot. Of 30 patients with KD, 13 (43.3%) possessed antibodies to Prx 2, whereas these antibodies were present in only 1 of 10 patients (10.0%) with sepsis (4 with purulent meningitis and 6 with septicemia). In contrast, antibodies to Prx 1 and 4 were not detected in either group. There was no significant correlation among the titers of the three antibodies. Clinical parameters were compared between anti-Prx 2-positive and -negative patients. The presence of anti-Prx 2 antibodies correlated with a longer period of fever and poor response to high-dose γ-globulin therapy in patients with KD. Anti-Prx 2-positive patients had significantly greater excretion of urinary 8-isoprostaglandin than did anti-Prx 2-negative patients. These results provide the first evidence for an antibody to Prx 2 in patients with KD. They also suggest that this antibody might serve as a marker of disease severity and be involved in the pathophysiology of vasculitis in some patients with KD.

List of Abbreviations: 

anti-endothelial cell antibody


bacillus Calmette-Guérin


coronary artery lesions




C-reactive protein


8-isoprostaglandin F2α




human umbilical vein endothelial cells


immunoglobulin G


Kawasaki disease


optical density


PBS containing 0.1% Tween 20




rheumatoid arthritis


systemic lupus erythematosus


white blood cell

Kawasaki disease is a type of acute vasculitis of unknown etiology that occurs in childhood. We and others have reported that AECA is both found in KD and cytotoxic to endothelial cells in patients with KD (1–3). AECA induces complement-dependent cytotoxicity (1, 2), enhances expression of adhesion molecules, and causes monocyte adhesion to HUVECS (3). In addition, immune dysregulation in KD patients leads to abnormal apoptosis of neutrophils or mononuclear cells (4, 5), which may result in oxidative stress and autoimmunity. Prxs have peroxidase activity (ROOH + 2e→ ROH + H2O), which results in reduction of hydrogen peroxide, peroxynitrite and ROOH in various organs (6–10). Recently, we detected autoantibodies to Prxs 1 and 4 in sera from autoimmune individuals with systemic vasculitis, such as occurs in SLE or RA (6). Anti-Prx 2 antibody is also found more frequently in systemic vasculitis (7). Increased oxidative stress induces vasculitis in KD (8); therefore we speculated that antibodies to antioxidative enzymes might be involved in the pathogenesis of this disease. The present study was undertaken to determine whether antibodies to Prxs are present in sera from patients with KD, and whether these antibodies are predictive of the severity of vasculitis. We focused on Prx 2, which is located in both the cytosol and cell membrane and has high antioxidative activity in many types of cells, including the vascular endothelium (9).


Patients and materials

Serum samples were obtained from 30 patients who met the diagnostic guidelines for KD (KD group: 15 boys and 15 girls, age: 2.4 ± 2.1 years, range 0.1–8.0  years). Sera were obtained during the acute phase of illness, that is, within 5 days of onset. Three of 30 patients with KD had transient coronary artery dilatation detected by echocardiogram. All patients with KD received high-dose γ-Glb plus aspirin. Sera were also obtained before treatment from 10 patients with septic disease (sepsis group: three boys and seven girls, age: 2.4 ± 2.2 years, range 0.1–6.3 years). Four of these patients were diagnosed with purulent meningitis and six with septicemia. Control sera were obtained from 15 afebrile children (control group: 11 boys and 4 girls, age: 6.4 ± 5.7 years, range 0.7 – 15.0 years). Urine samples were obtained from some patients before treatment and from some afebrile children. The patients were treated at Kochi Medical School Hospital and Kochi Health Science Center Hospital. The study was approved by the Institutional Ethics Board and informed consent was obtained from the patients’ parents.

Enzyme-linked immunosorbent assay for anti-peroxiredoxin antibodies

An ELISA for detection of anti-Prx antibodies was performed as described previously (6, 7). Each well of a multi-titer plate (Cook; Dynatech, Alexandria, VA, USA) was coated with 10 μg/mL of recombinant Prx 1, 2 or 4 (Lab Frontier, Seoul, Korea) in carbonate buffer, followed by blocking with 4-fold diluted Block Ace (Yukijirushi, Sapporo, Japan) for 1 hr at room temperature. Serum samples diluted at 1:200 were used for screening of autoantibodies to Prxs. After incubation with recombinant proteins for 2 hr, the wells were washed three times with PBST. The bound antibodies were reacted with horseradish peroxidase-conjugated goat anti-human IgG (Zymed Laboratories, San Francisco, CA, USA). Color was developed with o-phenylenediamine as a substrate and quantified using a microplate photometer at 490 nm. The anti-Prx antibody titer was calculated as the OD of the sample − the OD of the negative control (pooled sera from seven normal adults).

Western blot

Peroxiredoxin 2 protein was separated by 12.5% SDS PAGE and transferred to nitrocellulose membranes (Protran, Schleicher & Schuell, Dassel, Germany). The membranes were blocked with 1% Block Ace for 1 hr, washed in PBST for 30 min, and then incubated with serum samples diluted at 1:200 in 0.1% Block Ace for 1 hr. After washing three times in PBST, the bound antibodies were reacted with horseradish peroxidase-conjugated goat anti-human IgG (Zymed Laboratories) for 1 hr. Finally, the bound antibodies were visualized using diaminobenzidine.

Urinary 8-isoprostaglandin F2α

Oxidative stress was evaluated by measuring urinary excretion of 8-iso-PG, which is a reliable marker of lipid peroxidation. Urinary concentrations of 8-iso-PG were measured using a ELISA assay kit (Cell Biolabs, San Diego, CA, USA) according to the manual. The concentration was corrected by urinary Crn concentration.

Statistical analysis

Differences in the prevalence of antibodies to Prxs between the control and disease groups were compared by Mann-Whitney U test. The independence of results was tested by χ2 test or Fisher exact probability test. Differences in clinical parameters between anti-Prx 2-positive and -negative groups were compared by Student's t-test or χ2 test. Correlations among anti-Prxs antibody titers and between anti-Prx2 antibody titers and urinary concentrations of 8-iso-PG were evaluated by Spearman's correlation test. A level of P < 0.05 was considered to be statistically significant.


Incidence of autoantibodies to peroxiredoxins

Antibodies to Prxs were detected by ELISA. Sera with OD values greater than the mean + 3 SD of the control values (0.669, 0.454, and 0.830 for anti-Prx 1, anti-Prx 2, and anti-Prx 4 antibodies, respectively,) were considered positive.

The OD values for anti-Prx 2 antibodies were 0.456 ± 0.129, 0.351 ± 0.085 and 0.235 ± 0.073 in the KD, sepsis and control groups, respectively. Both disease groups had significantly higher mean OD values than did the control group (P= 0.00010 vs. KD group, P= 0.0055 vs. sepsis group), and the mean OD in the KD group was significantly higher (P= 0.039) than that in the sepsis group. Anti-Prx 2 antibodies were detected in 13 of 30 patients in the KD group (43.3%) and 1 of 10 patients in the sepsis group (10.0%), giving a significantly higher anti-Prx 2 antibody-positive rate in the KD group compared to the sepsis group (P= 0.045) (Fig. 1).

Figure 1.

Antibody titers to Prx 2 measured by ELISA. Serum samples diluted 1:200 were reacted with Prx 2. Numbers in parentheses indicate the percentages of anti-Prx 2-positive serum samples in each group. Numbers in square brackets indicate the ratio of anti-Prx 2-positive serum samples/tested serum samples in each group. The dotted line indicates the cut-off described in Materials and Methods. Open circles represent cases with CALs.

The OD values for anti-Prx 1 and 4 antibodies were 0.200 ± 0.109 and 0.290 ± 0.156 in the KD group, 0.195 ± 0.080 and 0.308 ± 0.125 in the sepsis group, and 0.219 ± 0.150 and 0.347 ± 0.161 in the control group, respectively, with no significant differences among the three groups (Fig. 2). We detected no antibodies to Prx 1 or Prx 4 in patients in the KD and sepsis groups (Fig. 2). There was no significant correlation among anti-Prx 1, anti-Prx 2, and anti-Prx 4 antibody titers (data not shown).

Figure 2.

Antibody titers to Prx 1 and 4 measured by ELISA. Serum samples diluted 1:200 were reacted with Prx 1 or 4. The numbers in parentheses indicate the percentages of anti-Prx 1- or 4-positive serum samples in each group. The dotted line indicates the cut-off described in Materials and Methods. Open circles represent cases with CALs.

Western blot

Reactivity to Prx 2 was confirmed by Western blot using serum samples from three KD patients who were positive for anti-Prx 2 antibody and one KD patient who was negative for anti-Prx 2 antibody by ELISA. Representative results are shown in Fig. 3.

Figure 3.

Detection of anti-Prx 2 antibodies by Western blot. Recombinant Prx 2 proteins were separated by 12.5% SDS-PAGE and then transferred onto a nitrocellulose membrane and stained with Ponceau S or reacted with serum samples. Four representative cases of patients with KD are shown. Sera from patients 6, 12, and 27, who had CALs, were positive for anti-Prx 2 antibodies by ELISA assay (OD values: 0.484, 0.485, and 0.482, respectively), but serum from patient three, who did not have a CAL, was negative for anti-Prx 2 antibody by ELISA assay (OD value: 0.317).

Correlation between clinical variables and the presence of anti- peroxiredoxin 2 antibody in patients with Kawasaki disease

Three of the 30 acute KD patients subsequently developed CALs; all of these patients had anti-Prx 2 antibody. The mean anti-Prx 2 antibody titer in these sera (n= 3, OD value: 0.484 ± 0.002) was higher than that of patients with an uncomplicated course of KD (n= 27, OD value: 0.413 ± 0.145). However, there was no significant difference between titers in patients with CALs and those of other KD patients who were positive for anti-Prx 2 antibody but did not develop CALs (n= 10, OD value: 0.559 ± 0.094).

We compared clinical variables between patients who were positive and negative for anti-Prx 2 autoantibody (Table 1). Age at the time of sample collection and the period between onset and the start of therapy, including high-dose γ-Glb (4.8 ± 0.7 vs. 5.1 ± 0.9 days), did not differ significantly between the two subgroups. The duration of fever > 37.5°C was significantly longer in patients with anti-Prx 2 autoantibodies than in those without antibodies (P= 0.031). WBC counts, neutrophil counts, and concentrations of CRP in patients with anti-Prx 2 antibodies were slightly higher than in those without anti-Prx 2 antibodies. CALs occurred significantly more frequently in patients with anti-Prx 2 antibodies than in those without anti-Prx 2 antibodies (p= 0.037). A non-responder to high-dose γ-Glb therapy was defined as a patient in whom fever did not decrease within 24 to 48 hr of first γ-Glb therapy. The non-response rate was significantly greater in patients with anti-Prx 2 antibodies than in those without anti-Prx 2 antibodies (P= 0.0051). We found no significant correlation among CRP concentrations, WBC counts and anti-Prx 2 antibody titers (data not shown).

Table 1.  Comparison of clinical profiles in patients with KD with and without anti-Prx 2 antibody
VariablesAnti-Prx 2 antibody 
Positive (n= 13)Negative (n= 17)
  1. aperiod during which maximum body temperature was more than 37.5°C; bStudent's t-test; cχ2-test; γ-Glb, high dose γ-globulin therapy

Age (years)   2.9 ± 2.3   2.0 ± 1.9NSb
Fever (days)a   9.6 ± 4.5   6.2 ± 2.8P = 0.031b
WBC (/μL)13,614 ± 424712,959 ± 3151NSb
Neutrophil (/μL)   8480 ± 2707   7280 ± 3012NSb
CRP (mg/dL)  11.2 ± 3.9   9.6 ± 4.6NSb
Positive CAL3/13 (23.1%)0/17 (0%)P = 0.037c
Non-responder5/13(35.5%)0/17(0%)P = 0.0051c

Time courses of anti-Prx 2 antibody titers in the acute and convalescent phases (2 to 4 weeks after intravenous administration of a total of 2 – 4 g/kg of γ-Glb) were analyzed in eight patients with KD (five with anti-Prx 2 antibody, two of five patients with CALs and three without antibody). The titers of anti-Prx 2 antibody in these two phases were 0.455 ± 0.123 and 0.393 ± 0.132, respectively. The titer decreased with clinical improvement, but this change was not significant. The ratio of anti-Prx 2 antibody titer to serum IgG (g/L) significantly decreased (P= 0.00059) between pre-γ-Glb therapy (0.681 ± 0.196 [g/L], serum IgG concentrations: 699 ± 250 mg/dL) and post-γ-Glb therapy (0.210 ± 0.106 [g/L, serum IgG concentrations: 1795 ± 550 mg/dL) in eight patients with KD.

Oxidative stress and anti-peroxiredoxin 2 antibody

Urinary 8-iso-PG concentrations were significantly increased in 16 KD patients (790 ± 89 pg/mg Crn) compared with seven sepsis patients (622 ± 130 pg/mg Crn) and ten controls (401 ± 36 pg/mg Crn) (P= 0.011 vs. sepsis group, P= 0.00025 vs. control group). There was also a significant difference between the KD and sepsis groups (P= 0.00074). In the KD group, urinary 8-iso-PG concentrations in nine patients with anti-Prx 2 antibody (838 ± 77 pg/mg Crn) was significantly greater (P= 0.0094) than in seven patients without anti-Prx 2 antibody (720 ± 77 pg/mg Crn). However, urinary 8-iso-PG concentrations in three patients with CALs (840 ± 27 pg/mg Crn) were equal to those of another six KD patients who were positive for anti-Prx 2 antibody but did not develop CALs (837 ± 95 pg/mg Crn) (Fig. 4). There was a significant correlation between anti-Prx 2 antibody titers and urinary concentrations of 8-iso-PG (P= 0.045) in 16 KD patients (Fig. 5).

Figure 4.

Urinary concentrations of 8-iso-PG measured by ELISA. Open circles represent cases who were positive for anti-peroxiredoxin 2 antibody. Urinary concentrations of 8-iso-PG in three KD patients with CALs (cases 6, 12, 27) were 830, 870, and 829 pg/mg Crn, respectively.

Figure 5.

Correlation between anti-Prx 2 antibody titers and urinary concentrations of 8-iso-PG in 16 patients with KD. Sample number was 16, r = 0.517318. There is a significant correlation between the two variables (P= 0.045).


Peroxiredoxin are a recently discovered and characterized family of thiol-specific antioxidant proteins (9). Six Prxs have been reported in humans. Prxs are found in various subcellular locations, including Prx 1 in the cytosol and nucleus; Prx 2 in the cytosol and cell membrane; and Prx 4 in the cytosol, Golgi and secretory fluid. Reversible modifications of Prxs from HUVECs based on changes in isoelectric points after exposure to peroxide have been described (10, 11). Prx 2 is a non-secreted cytosolic protein that has the most powerful peroxidase activity of the six Prxs (12–14). Prx oxidation status has been shown to reflect oxidative stress in the vasculature and to correlate with the extent of atherosclerotic lesions in an animal model (15). Since oxidative stress is known to cause inflammation, including vasculitis (16) and KD (8, 17), impaired function of Prx 2 triggered by anti-Prx antibodies may initiate vasculitis and KD.

We detected anti-Prx 2 antibodies in 43.3% of patients with KD, this rate being significantly higher than that for patients with sepsis or for controls. Thus, anti-Prx 2 antibody may be involved in the pathophysiology of vasculitis in some patients with KD. A specific role of anti-Prx 2 antibody in KD was further suggested by correlations between the presence of autoantibody and both disease severity and poorer response to high-dose γ-Glb therapy. We examined two γ-Glb preparations, both with final concentrations corresponding to 500 mg/dL (IgG), and obtained OD values of 0.303 and 0.289. We evaluated these values as negative for anti-Prx 2 antibody. In this study, mean IgG concentrations in diluted serum samples (× 200) from eight patients with KD were very low (3.5 mg/dL for pre-γ-Glb therapy and 9.0 mg/dL for post-γ-Glb therapy). We found that titers of anti-Prx2 antibody decreased 2 to 4 weeks after administration of high-dose γ-Glb therapy. In addition, the ratio of anti-Prx 2 antibody titers to serum immunoglobulin G concentrations (g/L) decreased significantly between pre-γ-Glb and post-γ-Glb therapy in eight patients with KD. Therefore, it is unlikely that the γ-Glb preparation has a major influence on anti-Prx 2 antibody titer. Although anti-Prx 2 antibody activity and function decrease with treatment, it is also possible that this antibody is spontaneously eliminated slowly from sera of patients with KD. Further studies are required to examine serial long-term changes in anti-Prx 2 antibody titer, especially in patients where γ-Glb ceases to be effective or the KD flares up.

Karasawa et al. examined the subcellular localization of Prx 2, the binding affinity of anti-Prx 2 antibodies to live endothelial cells and the effects of anti-Prx antibodies on secretion of cytokines and chemokines from endothelial cells (7). Prx 2 has been detected in the membrane and organelle fractions of endothelial cells lines, as well as in whole cell lysates. Anti-Prx 2 antibody bound to the surface of live HUVECs and stimulation of these cells with rabbit anti-Prx 2 antibody upregulates secretion of interleukin-6, interleukin-8 and granulocyte colony-stimulating factor by more than 2-fold compared to stimulation with rabbit IgG (7). Suzuki et al. showed that KD patients with CALs had significantly higher concentrations of serum granulocyte colony-stimulating factor than did patients without CALs (18). However, we could not confirm whether the function and binding sites of human anti-Prx 2 antibody are the same as for rabbit anti-Prx 2 antibody. In addition, the concentrations of anti-Prx 2 antibodies in patients with CALs were not extremely high, compared to the other KD patients with anti-Prx 2 antibodies. Therefore, anti-Prx 2 antibody might be a predictive marker for refractoriness to high dose γ-Glb therapy.

A homolog of Prx 2 is conserved in many microorganisms and is expressed in BCG at a higher level than in Mycobacterium tuberculosis (19). Thus, immune responses triggered by a microorganism with cross-reactive antigens might be associated with the development of KD. In a mouse model of KD, Nakamura et al. showed that intravenous injection with anti-Prx 2 antibodies resulted in coronary arteritis, but only after prior inoculation with BCG (20). In the present study, the titers of anti-Prx 2 antibody in the three patients with CALs were not higher than those in the other ten patients in the anti-Prx 2 antibody-positive group who did not have CALs. However, the three patients with CALs had a significantly longer duration of fever (10.3 ± 2.7 vs. 9.2 ± 4.0 days, P= 0.045), higher CRP concentrations and increased WBC counts compared to the ten patients without CALs. The rates of BCG vaccination and local swelling were the same in the antibody-positive and -negative subgroups (data not shown). These findings suggest that coronary arteritis in KD is not induced by anti-Prx 2 antibodies alone, and that other microorganisms besides BCG, as well as autoantibodies against other endothelial antigens (especially human coronary artery endothelial antigens), might also be required to induce coronary arteritis.

The reason for the higher prevalence of the anti-Prx 2 antibody compared to the anti-Prx 1 and 4 antibodies is unclear. The amino acid sequence identities are 78% between Prx 2 and Prx 1, and 70% between Prx 2 and Prx 4 (6, 21). However, we found no significant correlations among the occurrence and titers for anti-Prx 1, 2, and 4 antibodies. Thus, generation of these three antibodies appears to be independent; therefore we cannot assume cross-reactivity among them.

On the other hand, this study revealed that oxidative stress evaluated by urinary excretion of 8-iso PG contributes to the acute phase of KD, a finding which is supported by previous report (8). In addition, the present study showed a positive correlation between anti-Prx 2 antibody titers and urinary 8-iso-PG concentrations. Therefore, the presence of anti-Prx 2 antibody might reflect the production of oxidative stress.

In conclusion, we identified a novel antigen, Prx 2, as an AECA target. Anti-Prx 2 antibody may be a predictor of disease severity and may also be involved in the pathophysiology of vasculitis in some patients with KD. However, further studies are needed to examine the diagnostic value of the autoantibody and to elucidate the mechanism of its possible association with development of CALs on a large scale, since the patients number in the present study was small and only three patients with KD developed CALs.


This study was supported in part by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan; the Ministry of Health, Labor and Welfare of Japan; and the Japan Rheumatism Foundation.


None of the authors has a conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and we affirm that this report is consistent with those guidelines.