Noise exposure does not make obvious changes in NRF2 target gene expression

Acoustic overexposure has been reported to cause ischemia-reperfusion and to increase the production of reactive oxygen species (ROS) and reactive nitrogen species (RNS), including hydrogen peroxide, superoxide and peroxynitrite. Because NRF2 has been shown to induce a battery of cytoprotective genes in response to ROS/RNS, we expected that NRF2 target genes would be activated after noise exposure.

We exposed wild-type and Nrf2^–/– mice to 96-dB SPL noise for 2hr and sacrificed them at 4hr post-exposure for the purification of total RNA from the whole cochlea. Contrary to our expectation, expression of the NRF2 target genes NAD(P)H:quinone oxidoreductase 1 (Nqo1), heme oxygenase 1 (Ho-1), glutamate-cysteine ligase, catalytic subunit (Gclc), glutamate-cysteine ligase, modifier subunit (Gclm) and thioredoxin reductase 1 (Txnrd1) did not differ before and after the noise exposure irrespective of genotype (Fig. 2). Based on previous reports suggesting that noise exposure enhances glutathione synthesis in a specific cell population in the cochlea, for instance the stria vascularis^19,20, it is probable that post-exposure induction of NRF2 target genes occurred in a limited compartment of the cochlea and was thereby undetectable in the whole cochlea. Alternatively, the noise exposure might be a modest inducer of NRF2.

Open in a separate window

Figure 2

Expression levels of NRF2 target genes in the whole cochlea with or without noise exposure.

Expression levels of Nrf2 and NRF2 target genes (Nqo1, Ho-1, Gclc, Gclm, and Txnrd1) in the whole cochlea at 4hr after noise exposure were examined by real-time RT-PCR. All of the samples were quantified against the same standard curve, and each expression level was normalized to Gapdh expression. **P<0.01. Unpaired two-tailed Student’s t-test was applied. The data represent the mean±s.d. (n=3).

CDDO-Im attenuates NIHL in an Nrf2-dependent manner

The clear susceptibility of Nrf2^–/– mice to acoustic hyperexposure combined with the absence of an obvious difference in the expression levels of NRF2 target genes led us to the idea that exogenously induced NRF2 activation would provide further protection from NIHL. We examined the effect of CDDO-Im, which is a triterpenoid possessing a high potency as an NRF2 inducer^16,17,18, in protecting the cochlea from the noise. The mice were administered CDDO-Im or vehicle solution and exposed to 96-dB noise for 2hr according to the regimens shown in Fig. 3a. The ABR thresholds were measured in the same protocol used in the experiments shown in Fig. 1.

Open in a separate window

Figure 3

CDDO-Im attenuates NIHL in an NRF2-dependent manner.

(a) Experimental design for acoustic overexposure and ABR test with three regimens of CDDO-Im. Noise exposure and ABR threshold measurements were conducted as described in the legend to Fig. 1a. The mice in the “pre” group received CDDO-Im 6hr before noise exposure. Those in the “3 times” group received CDDO-Im 6hr before, 1 day after, and 3 days after noise exposure. Those in the “post” group received CDDO-Im immediately after noise exposure. (b,c) ABR threshold shifts at 4hr (b) and 7 days (c) after noise exposure of mice treated with three different regimens of CDDO-Im. The differences between the vehicle-treated and CDDO-Im-treated “pre”/“3 times” groups were statistically significant at 7days (c). (d,e) ABR threshold shifts at 4hr (d) and 7days (e) after noise exposure of wild-type and Nrf2^–/– mice treated with the “pre” regimen of CDDO-Im. The differences between the wild-type and Nrf2^–/– mice treated with CDDO-Im were statistically significant at 7days (e). *P<0.05, **P<0.01. Analysis of variance followed by Bonferroni post hoc test was applied. The data represent the mean±s.e.m. (n=5–6). The data obtained from wild-type mice treated with vehicle and those with CDDO-Im pretreatment that are shown in panels b and c are duplicated in panels d and e for comparison.

We first optimized the conditions of CDDO-Im treatment (Fig. 3a). A group of mice were given a single dose of CDDO-Im before the noise exposure (CDDO-Im pre); another group of mice were given three doses, one before and two after the noise exposure (CDDO-Im 3 times); and still another group of mice were given a single dose right after the noise exposure (CDDO-Im post). The TTSs of the three CDDO-Im-treated groups were not different from that of the vehicle-treated group (Fig. 3b), whereas the PTS of the “CDDO-Im pre” group and the “CDDO-Im 3 times” group was significantly reduced compared with that of the vehicle-treated group (Fig. 3c). Of note, the PTS of the “CDDO-Im post” group was almost exactly the same as that of the vehicle-treated group, indicating that CDDO-Im treatment after noise exposure is ineffective for the suppression of NIHL development.

To verify the NRF2 dependency of the protective effect of CDDO-Im, we treated wild-type and Nrf2^–/– mice with CDDO-Im according to the “CDDO-Im pre” protocol. The TTS did not significantly differ between the two genotypes (Fig. 3d), while the PTS was significantly elevated in the Nrf2^–/– mice irrespective of the CDDO-Im treatment applied (Fig. 3e), indicating that CDDO-Im is ineffective in Nrf2^–/– mice even if it is administered before the noise exposure. Thus, NRF2 is required for the efficacy of CDDO-Im, and pretreatment with CDDO-Im before the noise exposure is critical for the prevention of NIHL.

NRF2 protects cochlea from noise-induced injury

To clarify the morphological basis of the CDDO-Im-mediated protection, we performed histological analysis of hair cells in the organ of Corti. In surface preparation images, we found the occasional loss of hair cells in the vehicle-treated wild-type mice, whereas missing hair cells were hardly found in the CDDO-Im-treated wild-type mice (Fig. 4a, upper two panels). In contrast, substantial numbers of hair cells were lost in the Nrf2^–/– mice regardless of the CDDO-Im treatment applied (Fig. 4a, lower two panels). To quantify the histological changes, we counted the damaged outer hair cells in each of four frequency-specific regions (8.0, 11.3, 16.0, and 32.0kHz) of the cochlea (Fig. 4b). Irrespective of the CDDO-Im treatment, fewer damaged hair cells were observed in the wild-type mice compared with the Nrf2^–/– mice in all the frequency-specific regions except for the 8.0kHz region. CDDO-Im treatment tended to reduce the number of damaged hair cells in the wild-type mice, although it did not reach statistical significance. Because the NIHL in the wild-type mice was rather mild irrespective of the CDDO-Im treatment, it was expected that the hair cell loss would not be obvious, based on a previous report that a noise-induced PTS of less than 40dB is not necessarily accompanied by hair cell loss but results from stereocilia damage alone^21,22. In the Nrf2^–/– mice, CDDO-Im treatment did not have any effect on the cell number decrease. These results indicate that NRF2 prevents the loss of hair cells in the cochlea.

Open in a separate window

Figure 4

CDDO-Im protects cochlear hair cells from noise-induced damage.

(a) Surface preparation images of the organ of Corti from wild-type and Nrf2^–/– mice, which were sacrificed after 7days of noise exposure. CDDO-Im was administered 6hr before the exposure. Damaged outer hair cells are indicated with arrowheads. All of the experiments were performed at least three times. OHCs, outer hair cells; IHCs, inner hair cells. The scale bar corresponds to 40μm. (b) Quantitative analysis of the damaged outer hair cells in each 400-μm length of the four frequency-specific regions (8.0, 11.3, 16.0, and 32.0kHz) of the cochlea. *P<0.05, **P<0.01. Analysis of variance followed by Bonferroni post hoc test was applied. The data represent the mean±s.e.m. (n=4).

CDDO-Im upregulates NRF2 target genes in the cochlea

To confirm the effect of CDDO-Im, which was intraperitoneally administered, on cochlear protection, we examined the induction of NRF2 target genes in the cochlea. Expression levels of NRF2 target genes (Nqo1, Ho-1, Gclc, Gclm, and Txnrd1) were examined in the whole cochlea at 6hr after CDDO-Im administration. CDDO-Im significantly upregulated the NRF2 target genes in the wild-type mice but not in the Nrf2^–/– mice (Fig. 5a). To verify the accumulation of NRF2, we conducted immunoblot analysis of NRF2 using the cochlear lysates (Fig. 5b). Consistent with the expression levels of NRF2 target genes (Figs 2 and ​and5a),5a), the protein levels of NRF2 were not elevated by the simple noise exposure but were elevated in the cochlea after the CDDO-Im administration. Thus, intraperitoneally administered CDDO-Im was effective in activating NRF2 to increase the expression of NRF2 target genes in the cochlea.

Open in a separate window

Figure 5

CDDO-Im upregulates the expression of NRF2 target genes in the whole cochlea in an NRF2-dependent manner.

(a) Expression levels of Nrf2 and NRF2 target genes (Nqo1, Ho-1, Gclc, Gclm, and Txnrd1) in the whole cochlea 6hr after CDDO-Im administration. All of the samples were quantified against the same standard curve, and each expression level was normalized to Gapdh expression. *P<0.05. Analysis of variance followed by Bonferroni post hoc test was applied. The data represent the mean±s.d. (n=3). (b) Immunoblot analysis of NRF2 in the whole cochlea at 4hr after noise exposure or 6hr after CDDO-Im administration. Lamin B was detected as a loading control. Full-length blots are presented in Supplementary Figure S1.

CDDO-Im reduces post-exposure oxidative stress in the cochlea

To evaluate an impact of CDDO-Im treatment on the level of oxidative stress induced by the noise exposure, we exploited 4-hydroxy-2-nonenal (4HNE) as an indicator of oxidative stress and detected 4HNE adducts in the organ of Corti by immunofluorescence (Fig. 6a). Mice were pretreated with or without CDDO-Im at 6hr before the start of the noise exposure and sacrificed 4hr after the end of the noise exposure.

Open in a separate window

Figure 6

Noise-induced lipid peroxidation in the cochlea is suppressed by CDDO-Im administration in an NRF2-dependent manner.

CDDO-Im or vehicle was administered to the mice 6hr before the noise exposure, and the mice were sacrificed at 4hr post-exposure. (a) Immunofluorescence detection of 4HNE in the organ of Corti of wild-type and Nrf2^–/– mice. All of the experiments were performed at least three times. Myosin 7a (Myo7a) is a marker of hair cells (green). The nuclei were counter-stained with DAPI (blue). 4HNE signals are shown in red. Inner hair cells and outer hair cells are indicated with arrows and arrowheads, respectively. The scale bar corresponds to 40μm. (b) Immunoblot analysis of 4HNE accumulation in the whole cochlea. Tubulin was detected as a loading control. Full-length blots are presented in Supplementary Figure S2.

Inner and outer hair cells, which were marked by positive staining with an anti-myosin 7a antibody, gave no detectable 4HNE signals in the absence of the noise exposure, irrespective of genotype. Upon noise exposure, the wild-type hair cells accumulated 4HNE adducts. This accumulation was decreased by the administration of CDDO-Im. In contrast, the Nrf2^–/– hair cells accumulated dramatically high levels of 4HNE adducts, against which CDDO-Im was ineffective. Immunoblot detection of 4HNE adducts in the whole cochlea produced results consistent with the immunofluorescence data (Fig. 6b). The Nrf2^–/– cochleas exhibited abundant accumulation of 4HNE adducts after noise exposure, and this accumulation was not altered by CDDO-Im treatment. These results indicate that NRF2 deficiency exacerbates noise-induced oxidative damage of the cochlea and that CDDO-Im reduces cochlear oxidative stress in an NRF2-dependent manner. Thus, NRF2 protects the cochlea against acoustic overexposure by reducing oxidative stress. Of note, anti-4HNE antibody yielded three or four discrete bands ranging from 50 to 70kDa instead of smears ranging over wide molecular weights, which was also observed in previous reports^23,24,25. This result suggest that specific proteins, which are yet unidentified, have a greater chance of being modified by 4HNE than others.

NRF2-deficient cochlea exhibits significant decrease of reduced glutathione (GSH)

To characterize the effect of the noise-induced oxidative stress on cochlea, we quantified reduced (GSH) and oxidized (GSSG) forms of glutathione (Fig. 7). We divided wild-type mice and Nrf2^–/– mice into three groups; without treatment, with noise exposure, and with CDDO-Im pre-treatment and noise exposure. Following the same protocol employed for the histological analysis, we collected inner ears and processed them for the glutathione quantification.

Open in a separate window

Figure 7

Quantification of glutathione in cochleas with and without noise exposure.

CDDO-Im or vehicle was administered to the mice 6hr before the noise exposure, and the mice were sacrificed at 4hr post-exposure. Reduced and oxidized glutathione and their ratios are shown. *P<0.05, **P<0.01. Unpaired two-tailed Student’s t-test was applied. The data represent the mean±s.d. (n=3).

Basal levels of GSH and GSSG were significantly lower in the Nrf2^–/– inner ears than those of wild-type mice, indicating that NRF2 contributes to the maintenance of the total glutathione level and its redox balance even under conditions without any treatments. After the noise exposure, GSH levels were decreased in both wild-type and Nrf2^–/– inner ears, resulting in the increase of GSSG/GSH ratios. These results are consistent with the increased levels of 4HNE adducts, which is an oxidative stress indicator (see Fig. 6), verifying the accumulation of oxidative stress after the noise exposure, particularly in the Nrf2^–/– inner ears. Importantly, administration of CDDO-Im prior to the noise exposure prevented the GSH decrease in wild-type inner ears, which was not observed in Nrf2^–/– inner ears. These results are explained by the major contribution of NRF2 to glutathione reduction and glutathione synthesis. Therefore, pre-activation of NRF2 increases the GSH level, a parameter of cellular anti-oxidant capacity, achieving the effective protection of cochlea from noise-induced oxidative damage.

CDDO-Im treatment

Mice were intraperitoneally administered vehicle (DMSO) or CDDO-Im (30μmol/kg body weight, obtained from Mochida Pharmaceuticals Co., Ltd., Tokyo, Japan). In the ABR study, CDDO-Im was given to the mice according to the regimen shown in Fig. 3a. For the quantitative RT-PCR, immunohistochemistry, immunoblot analysis, and glutathione measurement, the mice received CDDO-Im at 6hr pre-exposure or at 6hr pre-sacrifice.

Acoustic overexposure

Mice were kept awake and unrestrained in an anechoic chamber and were exposed to an 8–16kHz octave-band noise at 96dB sound pressure level (SPL) for 2hr. The sound was generated by a waveform generator (SF-06, Random Noise Generator; RION), amplified by an audio amplifier (D-75A; Crown), and filtered by an audio filter (Multifunction Filter; NF Corporation). The sound was presented in an open field by a dome tweeter (2446H; JBL) positioned at the center of the cage. Sound level was measured using a sound level meter (2250L; Brüel & Kjær.)

Hearing function test

Mice were anesthetized with ketamine (100mg/kg body weight) and xylazine (20mg/kg body weight) by intraperitoneal administration. ABR recordings were performed using a TDT System 3 auditory-evoked potential workstation and BioSigRP software (Tucker-Davis Technologies). ABR responses were evoked using bursts of pure tones at frequencies of 4, 8, 12, 16, and 32kHz. Evoked responses were averaged across 1,000 sweeps. Responses were collected for stimulus levels in 5-dB steps from 100dB SPL to 10dB SPL. The ABR threshold was defined as the lowest sound intensity sufficient to elicit at least one peak in the averaged ABR.

Real time quantitative RT-PCR

Mice were deeply anesthetized and transcardially perfused with ice-cold 10mM phosphate-buffered saline (PBS) (pH 7.4). The cochleas were quickly removed from the skull, collected on ice, and stored at –80°C in RNAlater (Life Technologies). The cochleas were homogenized in ISOGEN (Nippon gene), and for each sample, total RNA was purified from two whole cochleas. cDNA was synthesized using reverse transcriptase (ReverTra Ace, Toyobo). Quantitative PCR was performed on an Applied Biosystems 7300 sequence detector system with Power SYBR Green PCR Master Mix (Applied Biosystems), THUNDERBIRD SYBR qPCR Mix (Toyobo) for the SYBR Green system and THUNDERBIRD Probe qPCR Mix (Toyobo) for the Taqman probe system. The amplification reaction mixture (25μl) contained 400nM of each primer in the SYBR Green system, or 300nM of each primer and 200nM of a probe in the Taqman probe system. A thermal cycling condition used in this study is as follows; 2min at 50°C and 10min at 95°C followed by 50 cycles of 95°C for 15s and 60°C for 60s. The primers utilized in this study are shown in Supplementary Table S1.

Morphological analysis of hair cells

Mice were deeply anesthetized and transcardially perfused with 10mM PBS and 4% paraformaldehyde (PFA) in 10mM PBS (pH 7.4). The inner ears were quickly removed from the skull and immediately placed in 4% PFA. Small openings were made at the round window, oval window, and apex of the cochlea. The inner ears were fixed with 4% PFA at 4°C overnight and then decalcified in 10% EDTA for 2days at 4°C. For hair cell counting, surface preparations of the basilar membrane and the organ of Corti were processed. The hair cells were stained for F-actin with rhodamine-conjugated phalloidin (1:100, Invitrogen) for 1hr at room temperature under light-protected conditions. High-power fluorescence images were obtained using a microscope (BZ-9000, Keyence) and BZ-H1 software (Keyence), and the full-length cochlea was assembled and analyzed in Photoshop CS4 (Adobe). ImageJ software (NIH) was used to assemble and analyze the computed frequency map. Damaged and undamaged outer hair cells were counted in each 400-μm length of the four frequency-specific regions (8.0, 11.3, 16.0, and 32.0kHz) of the cochlea.

Immunohistochemistry

Cochlear samples were prepared using a procedure similar to that used for the histological hair cell analysis. The decalcified samples of cochlea were rinsed in 10mM PBS containing 10% and 30% sucrose, embedded in Tissue-Tek O.T.C. Compound (Sakura Finetechnical), and frozen in liquid nitrogen. Thin sections (8μm) were obtained with a cryostat (CM3000, Leica Instruments). Immunohistochemistry was performed according to a previously described procedure with modification⁴². Briefly, tissue sections were rinsed in 0.1% Triton X-100/Tris buffered saline (TBS), blocked with 3% bovine serum albumin/0.3% Triton X-100/TBS (TBST) for 30min at room temperature, and incubated with an unconjugated AffiniPure Fab fragment anti-mouse IgG (1:10, Jackson ImmunoResearch) for 2hr at room temperature. The sections were incubated overnight with primary antibodies (anti-4HNE; 1:400, JaICA. anti-Myosin 7a; 1:500, Abcam) diluted in 0.3% TBST at 4°C. All of the sections were washed three times in 0.1% TBST and incubated with secondary antibodies (Cy3 anti-mouse; 1:500, Jackson ImmunoResearch, Alexa Fluor 488 anti-rabbit; 1:500, Jackson ImmunoResearch) for 1hr at room temperature under light-protected conditions. The nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (1:2,000).

Equipment and settings

Fluorescence images were obtained using a laser-scanning confocal microscope (LSM 780, Zeiss) equipped with a Plan-Neofluar 40x/1.3 oil immersion objective lens and Blue-diode 405nm, Multi-Argon 458/488/514nm and DPSS 561nm lasers. The images were analyzed using ZEN software (Zeiss).

Immunoblot analysis

Mice were deeply anesthetized and transcardially perfused with ice-cold 10mM PBS (pH 7.4). The cochleas were quickly removed from the skull, collected on ice and stored at –80°C in 10mM PBS (pH 7.4). As described previously²⁴, both cochleas of each mouse were homogenized in 400μl of RIPA buffer [10mM Tris-HCl (pH 7.4), 150mM NaCl, 0.1% sodium deoxycholate, 0.1% SDS, 1% Nonidet P-40, 1mM dithiothreitol, and 1x complete protease inhibitor cocktail tablets (Roche)] for 15s. After centrifugation for 15min at 12,000rpm at 4°C, 250μl of the supernatant was collected and 6x SDS sample buffer was added. The solution was then boiled at 95°C for 5min. Each sample was separated by SDS-PAGE on a 10% gel. Proteins were then transferred onto PVDF membranes (Millipore). After blocking with 5% skim milk in 0.05% Tween 20/TBS, the membranes were incubated overnight at 4°C with an anti-4HNE antibody (1:100, JaICA) or anti-NRF2 antibody (1:100)⁴³. After rinsing in 0.05% Tween 20/TBS, the membranes were incubated for 30min at room temperature with a horseradish peroxidase-conjugated anti-mouse IgG antibody (1:2,000, Life Technology) for 4HNE or anti-rat IgG antibody (1:1,000) for NRF2. The immunoreactive bands were detected with Chemi-Lumi One L (Nacalai Tesque). Equal protein loading among individual lanes was confirmed by reprobing the membrane with an anti-Tubulin mouse monoclonal antibody (1:2,000, Sigma Aldrich) or anti-Lamin B goat polyclonal antibody (1:1,000, Santa Cruz Biotech).

Measurement of glutathione in the mouse cochlea

Both temporal bones (including the cochlea and vestibule, approximately 30mg) of each sample were dissected 4hr after the noise exposure and snap-frozen in liquid nitrogen. The frozen tissue was homogenized in 500μL of cold methanol containing 50μM internal standard solution (Human Metabolome Technologies, Tsuruoka, Japan). The homogenate was mixed with 500μL chloroform and 200μL Milli-Q water and further homogenized and centrifuged at 750g for 10min at 4°C. The aqueous layer was collected by filtration through a Millipore 5-kDa cutoff filter and centrifuged at 9,100g to remove proteins. The filtrate was lyophilized, suspended in 25μL Milli-Q water and analyzed using CE-TOFMS (Human Metabolome Technologies, Tsuruoka, Japan).

CE-TOFMS was carried out on an Agilent CE Capillary Electrophoresis System equipped with an Agilent 6210 time-of-flight mass spectrometer, an Agilent 1100 isocratic HPLC pump, an Agilent G1603A CE-MS adapter kit and an Agilent G1607A CE-ESI-MS sprayer kit (Agilent Technologies, Santa Clara, CA). The system was controlled by Agilent G2201AA Chem- Station software version B.03.01 for CE (Agilent Technologies, Santa Clara, CA).

The separation and detection of the cationic metabolites were carried out as described previously⁴⁴. A fused silica capillary (50lm i.d.; 980cm total length) with cation buffer solution (Human Metabolome Technologies, Tsuruoka, Japan) was used. The sample was injected at a pressure of 50mbar for 10s (approximately 10nL). The applied voltage was set at 27kV. Electrospray ionization-mass spectrometry (ESI-MS) was conducted in positive ion mode, and the capillary voltage was set at 4,000V. The spectrometer was scanned from m/z 50 to 1,000.

The raw data that were obtained using CE-TOFMS were processed with Keio MasterHands, which was developed by Keio University⁴⁵. Signal peaks corresponding to GSH and GSSG were extracted. GSH and GSSG migration times (MTs) were normalized using those of the internal standards. The resultant relative area values were further normalized to the sample amount.

Association study of an NRF2 promoter SNP and susceptibility to SNHL

For the association analysis between the SNP in the NRF2 promoter, rs6721961, and SNHL susceptibility, 602 Japanese males were recruited from health examinations conducted at Self-Defense Forces Central Hospital. We only extracted the examination data of the 49- and 50-year-old participants to exclude the effects of age-related hearing loss. This study protocol was approved by the Ethics Committee of the National Defense Medical College and Self-Defense Forces Central Hospital. Written informed consents were obtained from all of the participants, and all of the investigations were conducted according to the principles expressed in the Declaration of Helsinki. Pure-tone audiometry was conducted with an audiometer (AA-79S, RION Inc., Tokyo, Japan). To determine the precise impact of the SNP on SNHL, participants with conductive hearing loss were excluded through careful examination by two independent otolaryngologists before the data analysis (I.M. and K.M.). These participants were divided into controls (hearing level at 4kHz of the better side ear ≤20dB hearing level (HL)) and cases (>20dB HL). Genomic DNA was extracted from whole peripheral blood cells⁴⁶. Genotyping of the NRF2 promoter SNP, rs6721961, was performed using the TaqMan method (Life Technologies Corporation, Carlsbad, CA USA) with a LightCycler 480 (Roche Diagnostics, Mannheim, Germany)⁴³. The association between NRF2 and SNHL was examined using the Cochran-Armitage trend test by the software R (version 3.1.1) (http://www.r-project.org/).

We thank Dr. Jun Yasuda for providing important advice and instruction regarding the whole study; Drs. Kazuhiro Nomura, Ryoukichi Ikeda and Jun Suzuki for support regarding the mouse experiments; Drs. Tomoyoshi Soga and Masahiro Sugimoto for providing their special skills for metabolite analysis; and Ms. Nao Ota and Yuka Matsuyama for their technical support regarding the metabolite analysis. We thank the Biomedical Research Core of the Tohoku University Graduate School of Medicine for their technical support. We also thank Drs. Takashi Hinata, Takashi Suzuki, Toshimitsu Ito, Takeshi Matsunobu, Takaomi Kurioka, Katsuki Niwa, Akiyoshi Nakayama and Toshihide Higashino for their contributions to human sample collection and genetic analysis. This work was supported by JSPS KAKENHI grant numbers 24390075 (H. Motohashi) and 15K18999 (S.M.), MEXT KAKENHI grant numbers 23116002 (H. Motohashi) and 25293145 (H. Matsuo), the Ministry of Defense of Japan (H. Matsuo, K.M., A.S., I.M.), the Takeda Scientific Foundation (H. Motohashi), the Core Research for Evolutional Science and Technology from the JST (H. Motohashi and M.Y.) and the GSK Japan Research Grant (S.M.).