Growth factor-mediated hepatocyte proliferation is crucial in liver regeneration and the recovery of liver function after injury. The nuclear receptor, pregnane X receptor (PXR), is a key transcription factor for the xenobiotic-induced expression of genes associated with various liver functions. Recently, we reported that PXR activation stimulates xenobiotic-induced hepatocyte proliferation. In the present study, we investigated whether PXR activation also stimulates growth factor-mediated hepatocyte proliferation. In G0 phase-synchronized, immortalized mouse hepatocytes, serum or epidermal growth factor treatment increased cell growth and this growth was augmented by the expression of mouse PXR and co-treatment with pregnenolone 16α-carbonitrile (PCN), a PXR ligand. In a liver regeneration model using carbon tetrachloride, PCN treatment enhanced the injury-induced increase in the number of Ki-67-positive nuclei as well as Ccna2 and Ccnb1 mRNA levels in wild-type (WT) but not Pxr-null mice. Chronological analysis of this model demonstrated that PCN treatment shifted the maximum cell proliferation to an earlier time point and increased the number of M-phase cells at those time points. In WT but not Pxr-null mice, PCN treatment reduced hepatic mRNA levels of genes involved in the suppression of G0/G1- and G1/S-phase transition, e.g. Rbl2, Cdkn1a and Cdkn1b. Analysis of the Rbl2 promoter revealed that PXR activation inhibited its Forkhead box O3 (FOXO3)-mediated transcription. Finally, the PXR-mediated enhancement of hepatocyte proliferation was inhibited by the expression of dominant active FOXO3 in vitro. The results of the present study suggest that PXR activation stimulates growth factor-mediated hepatocyte proliferation in mice, at least in part, through inhibiting FOXO3 from accelerating cell-cycle progression.
Hepatocyte proliferation is an important adaptive response to liver diseases and plays a critical role in liver regeneration and repair . When hepatocytes are injured, they are replaced with proliferated mature cells to recover and maintain adequate liver function . Several types of cell-proliferating signals, such as cytokines and growth factors, have been identified as promoting this process [1,2].
The nuclear receptor, pregnane X receptor (PXR), is a key regulator of xenobiotic metabolism and export in the liver. In response to xenobiotics, PXR induces the transcription of genes encoding drug-metabolizing enzymes and transporters [3,4]. Recent studies indicate that physiological functions of PXR are not limited to the regulation of xenobiotic disposition but extend to the regulation of the inflammatory pathway  and cholesterol/lipid homoeostasis .
Recently, we reported a unique role of PXR in hepatocyte proliferation. Although PXR activation alone did not promote hepatocyte proliferation, it enhanced hepatocyte proliferation induced by other nuclear receptors–constitutive androstane receptor (CAR) and peroxisome proliferator-activated receptor α (PPARα) in mice . In addition, PXR ligand treatment enhanced the regeneration of rat livers from partial hepatectomy , and liver regeneration after partial hepatectomy was delayed in PXR-deficient mice . These results have raised the possibility that PXR might accelerate the hepatocyte proliferation induced by growth factors and cytokines observed during liver regeneration.
One of our recent studies has also demonstrated that PXR activation makes G0-phase hepatocytes enter the G1 phase . In that study, we found that PXR activation reduced the expression of G0/G1 phase-transition inhibitors in mouse livers, possibly through the inhibition of Forkhead box O (FOXO) transcription factors. FOXO transcription factors play important roles in diverse physiological processes, such as the cell cycle, apoptosis and metabolism [10,11]. In cell-cycle regulation, FOXOs tightly control G0/G1 and G1/S phase progression by up-regulating the expression of genes associated with cell-cycle inhibition, including Rbl2 (p130), Cdkn1b (p27), Cdkn1c (p57) and Cdkn1a (p21) [11,12].
In the present study, we investigated whether PXR activation enhanced the growth factor-mediated hepatocyte proliferation using in vivo and in vitro models.
Pregnenolone 16α-carbonitrile (PCN), propidium iodide (PI), Pyronin Y, 7-aminoactinomycin D (7-AAD) and collagenase (type IV) were obtained from Sigma-Aldrich. Saline for injection, RNase A and epidermal growth factor (EGF) were purchased from Otsuka Pharmaceuticals, Nacalai Tesque and Calbiochem, respectively. Recombinant fibroblast growth factor 19 (FGF19) had been prepared previously . Anti-V5-tag, anti-FOXO3 and anti-PCNA antibodies were purchased from MBL, Cell Signaling Technology and Thermo Fisher Scientific, respectively. Anti-DDDDK-tag and anti-β-actin antibodies were obtained from Sigma-Aldrich. Oligonucleotides were commercially synthesized by Fasmac. All other chemicals were of the highest grade available from Wako Pure Chemical Industries, Nacalai Tesque and Sigma-Aldrich.
Animals and primary hepatocytes
All experiments were performed in accordance with the guidelines for animal experiments of Tohoku University. Male wild-type (C57BL/6, Charles River Japan) and Pxr-null mice  were maintained in a temperature- and light-controlled environment (24°C, 12-h light–dark cycle). Mice (around 8 weeks old) were intraperitoneally treated with PCN (100 mg/kg) or vehicle (corn oil). For the FGF19 co-treatment experiment, mice were subcutaneously treated with FGF19 (40 nmol/kg) once a day for 3 days. On day 2, PCN (100 mg/kg) was co-administered intraperitoneally; 24 h after the last FGF19 treatment, mice were sacrificed by cervical dislocation and the livers collected. For liver injury experiments, mice were intraperitoneally treated with carbon tetrachloride (CCl4, 0.5 ml/kg, 10%, v/v, in corn oil); 24 h after CCl4 administration, mice were treated with PCN (100 mg/kg) and sacrificed at 24, 36, 48, 72, 96 or 144 h after CCl4 treatment by cervical dislocation; blood and livers were collected. Plasma alanine aminotransferase (ALT) activity was determined using Transaminase CII-B-test Wako (Wako Pure Chemical Industries). Mouse primary hepatocytes were prepared by a two-step collagenase perfusion method and cultured for further experiments in William's E medium supplemented with 10% FBS for 12 h.
Plasmids and adenoviruses
A mouse Rbl2 luciferase reporter construct was prepared by inserting Rbl2 promoter DNA, which was amplified by PCR using KOD-FX (Toyobo) and mouse genomic DNA as a template, into HindIII and XhoI sites of pGL4.10 (Promega). The cDNAs of mouse PXR and FOXO3 were amplified by PCR and inserted into pTargeT (Promega). Mutated Rbl2 luciferase reporter constructs, and expression plasmids for the dominant active form of FOXO3 (FOXO3-3A), in which all three protein kinase B (Akt) phosphorylation sites (Thr32, Ser253, Ser315) were mutated to alanine, and FLAG-tagged FOXO3-3A was prepared using a KOD-Plus-Mutagenesis Kit (Toyobo).
The mPXR-V5-expressing adenovirus (Ad-mPXRV5) was generated using mPXR cDNA, which was amplified by PCR using KOD-FX with specific primers and cloned into Shuttle–CMV vector (pShuttle–CMV–mPXR). The PmeI-linearized plasmid was integrated by homologous recombination into the adenoviral genomic plasmid (pAdEasy-1) in Escherichia coli bacterial strain BJ5183 to generate a recombinant, replication-deficient, adenoviral genome plasmid. Insertion of V5 at the C-terminal of pShuttle–CMV–mPXR was performed using a KOD-Plus-Mutagenesis Kit with specific primers. The LacZ-expressing adenovirus was reported previously . The resultant expression viruses were amplified in HEK-293 cells (RIKEN BRC), and separated into titres using Adeno-X Rapid Titer Kit (Clontech).
The primers used are shown in Supplementary Table S1.
Cell-cycle synchronization and adenovirus infection
AML12 cells (American Type Culture Collection) were seeded on to plates at a density of 10000 cells/cm2, cultured for 24 h and washed with PBS before additional culture in serum-free Dulbecco's modified Eagle's medium (DMEM)/F-12 containing Antibiotic-Antimycotic (Invitrogen), ITS-Premix (BD Biosciences) and 100 nM dexamethasone for 48 h. To allow cells to enter the cell cycle, cells were stimulated with serum by replacing the medium with DMEM/F-12 containing ITS, 100 nM dexamethasone and 10% FBS. In some experiments, cells were infected with 50 multiplicity of infection (MOI) of Ad-mPXRV5 or Ad-LacZ in FBS-free medium for 48 h.
Determination of mRNA and protein levels
Total RNA isolation and cDNA synthesis were carried out as described previously . Quantitative RT-PCR was performed using GoTaq qPCR Master Mix (Promega) and primer pairs for genes of interest (see Supplementary Table S2). Target mRNA levels were normalized to Actb mRNA or 18S rRNA levels.
Cultured cells were lysed in 20 mM Tris/HCl (pH 7.6), 150 mM NaCl and 1% NP-40. Protein concentrations were determined using a protein assay dye reagent (Bio-Rad Laboratories). Western blot analysis was carried out as described previously .
AML12 cells were seeded on a 48-well plate (BD Biosciences) at 30000 cells/well 24 h before transfection, transfected with reporter construct, expression plasmid and phRL-TK Control Vector (Promega) with Jet-PEI (Polyplus transfection), and treated with 10 μM PCN. The cell lysate was subjected to a Dual Luciferase Assay System (Promega) 12 h after transfection. Firefly luciferase activity was normalized to Renilla luciferase activity.
Co-immunoprecipitation was performed as described previously  with minor modifications. AML12 cells were cultured with serum-free DMEM/F-12 containing 50 MOI of Ad-mPXRV5 or Ad-LacZ for 36 h. Whole-cell lysates or nuclear extracts were prepared and incubated with protein G-coupled Dynabeads (Invitrogen) at 4°C for 3 h, and then the beads were removed. The supernatant was then incubated with normal rabbit immunoglobulin G (Millipore), anti-DDDDK-tag or anti-V5-tag antibody at 4°C overnight. Protein G-coupled Dynabeads were added to the reaction and the mixture was further incubated at 4°C for 3 h. The immune complexes were eluted with SDS/PAGE sample buffer and subjected to immunoblotting with Clean-Blot IP Detection Reagent (Thermo Fisher Scientific).
Flow cytometry for cell-cycle analysis
Cell-cycle analysis was performed by staining DNA with PI or double staining DNA and RNA with 7-AAD and Pyronin Y, respectively, as described previously .
Statistical analysis was performed using GraphPad Prism (GraphPad Software). All data are provided as the means±S.D.s. The significance of difference between control and treated groups was assessed using Student's t-test and ANOVA, followed by Dunnett's test for data from two and multiple groups, respectively. The Tukey–Kramer test or Bonferroni's correction was used to compare multiple groups.
Influence of PXR activation on the serum-induced cell proliferation of AML12 cells
We first investigated the influence of PXR activation on the growth factor-mediated hepatocyte proliferation using an in vitro model. AML12 cells, immortalized mouse normal hepatocytes, were cultured in serum-free medium for 48 h to be synchronized at G0 phase and then cultured in fresh medium containing 10% FBS to force the cells to enter the cell cycle. Flow cytometry analysis demonstrated that the culturing in serum-free medium increased the number of cells in the G0/G1 peak and decreased their RNA contents (see Supplementary Figures S1A and S1B), suggesting synchronization at the G0 phase. Either 12 or 24 h after serum treatment, the ratio of the G0/G1 peak was decreased and those of the S (12 h and 24 h) and G2/M (24 h) phases were increased (see Supplementary Figures S1C and S1D), suggesting re-entry into the cell cycle.
AML12 cells were next infected with adenovirus expressing V5-tagged mPXR (Ad-mPXRV5). The mPXRV5 and cytochrome P450 (CYP)3A (PXR target) protein levels were robustly increased in the cells infected with Ad-mPXRV5 (Figure 1A). PCN treatment after Ad-mPXRV5 infection augmented the serum-mediated decrease in the ratio of the G0/G1 peak, and their increase in the S and G2/M phases (Figures 1B and 1C). In addition, PXR activation enhanced the serum-induced increase in PCNA protein levels (Figure 1D), and Ccnd1 and Mcm2 mRNA levels (Figure 1E). It also significantly accelerated serum-induced cell growth (Figure 1F). The treatment of AML12 cells with EGF was reported to promote cell growth . This EGF-mediated cell growth was also accelerated by PXR activation, as demonstrated by the increase in Ccnd1 and Mcm2 mRNA levels and the cell number (see Supplementary Figure S2).
Influence of PXR activation on serum-induced hepatocyte proliferation
Influence of PXR activation on hepatocyte proliferation during acute liver injury
We next investigated whether PXR activation could promote hepatocyte proliferation during the recovery from liver injury. FGFs, known to induce hepatocyte proliferation in rodents [19,20], are one of the pivotal factors for liver regeneration. We observed that treatment of mice with recombinant FGF19 tended to increase the number of Ki-67-positive nuclei as well as hepatic mRNA levels of Mcm2, Ccna2 and Ccnb1 (see Supplementary Figure S3). PCN co-treatment synergistically increased these levels (see Supplementary Figure S3).
Using a CCl4-induced liver injury model, we investigated the influence of PXR activation on hepatocyte proliferation during liver regeneration. Wild-type (WT) and Pxr-null mice were treated with PCN 24 h after CCl4 administration and hepatocyte proliferation was investigated 48 h after CCl4 administration. Immunohistochemical analyses of the livers using antibody against Ki-67, a marker for cell proliferation, demonstrated that CCl4 treatment alone increased the number of Ki-67-positive nuclei (Figures 2A and 2B), as well as hepatic mRNA levels of Ccna2 and Ccnb1 (Figure 2C), as expected. Intriguingly, PCN co-treatment further increased these levels in WT but not Pxr-null mice, suggesting that PXR activation enhances hepatocyte proliferation during liver regeneration after injury. In this model, plasma ALT activities were increased and PCN co-treatment had no influence on them in either strain (see Supplementary Figure S4A), suggesting that the extent of liver damage was similar regardless of PCN treatment. We obtained similar results for the mRNA levels of inflammatory cytokines such as Tnfa and Il6 (see Supplementary Figure S4B).
Influence of PCN treatment on hepatocyte proliferation after CCl4-induced liver injury
Next we investigated the influence of PCN treatment on the chronological change of CCl4-induced hepatocyte proliferation (Figure 3A). The number of Ki-67-positive nuclei (Figure 3B) and mRNA levels of Ccna2 and Ccnb1 (Figure 3C) were remarkably increased with a peak at 72 h after CCl4 administration. In mice co-treated with PCN, these levels were increased more rapidly than in mice treated with CCl4 alone (48 h; Figures 3B and 3C). In fact, at 48 h after CCl4 administration, the number of M-phase cells was significantly increased by PCN co-treatment in WT but not Pxr-null mice (Figures 3D and 3E). These results suggest that PXR activation accelerates cell-cycle progression during liver regeneration.
Influence of PXR activation on cell-cycle progression in CCl4-induced liver injury
Influence of PXR activation on cell-cycle-related gene expression in hepatocytes
In mature livers hepatocytes normally remain at a quiescent stage (G0 phase) and achieve competence for proliferation on proliferating stimuli. A couple of cell-cycle suppressors tightly regulate the cell-cycle transition from G0 to G1 and from G1 to S phases, and these expression levels are directly linked to the sensitivity to proliferative stimuli [21,22]. We thus hypothesized that PXR activation might regulate the expression of those suppressors and investigated the influence of PXR activation on mRNA levels of these genes. PCN treatment decreased mRNA levels of all the genes investigated including Cdkn1b, Rbl2, Cdkn1a, Cdkn1c, Cdkn2a and Cdkn2b in WT but not Pxr-null mice (Figure 4A). Consistently, PXR activation significantly decreased these mRNA levels in AML12 cells as well (Figure 4B).
Influence of PXR activation on the mRNA levels of cell-cycle suppressor genes
Cross-talk between PXR and FOXOs in hepatocytes
As the cell-cycle suppressor genes are regulated by FOXO family transcription factors , we determined mRNA levels of these FOXO family members and their target genes, and found that PCN treatment decreased the mRNA levels of Foxo3 and target genes Bim and Mxi1 in WT but not Pxr-null mice (Figure 5A).
Influence of PXR activation on FOXO3-mediated gene transcription
To investigate whether PXR could affect the transcriptional activities of FOXOs, we performed reporter assays using mouse Rbl2 promoter (Rbl2-pGL4.10), containing two FOXO-binding sites . In AML12 cells, reporter activities were increased by ectopic FOXO3 expression in a dose-dependent manner (Figure 5B). As expected, the increase was more obvious with a constitutively active FOXO3 mutant (FOXO3-3A) (Figure 5C). PXR activation significantly repressed the FOXO3- and FOXO3-3A-mediated increase in luciferase activities (Figure 5C). With a FOXO-binding site-mutated reporter, such suppressing effects of PXR were not observed (Figure 5D). In mouse primary hepatocytes, FOXO3 expression also increased reporter activities of Rbl2-pGL4.10 and PCN treatment repressed them (see Supplementary Figure S5A). In addition, PCN treatment decreased Rbl2 mRNA levels, whereas it increased those of Cyp3a11, a PXR target gene (see Supplementary Figure S5B). Finally, FOXO3-3A expression significantly increased mRNA levels of Rbl2 and Mxi1 in mouse primary hepatocytes, and PCN treatment decreased their levels (see Supplementary Figure S5C).
PXR has been reported to interact with FOXO1 to repress its target gene expression . We thus investigated whether PXR could interact with FOXO3 in mouse hepatocytes by co-immunoprecipitation assays. Even though PXR overexpression decreased endogenous FOXO3 protein levels as expected from the results shown in Figure 5A, FOXO3 was detected in the immunoprecipitates with anti-V5 (for PXR) antibody (Figure 5E). To confirm the interaction, both Flag-FOXO3-3A and PXRV5 were co-expressed and co-immunoprecipitation assay was performed with anti-flag (for FOXO3) antibody. As shown in Figure 5F, PXR was detected in the immunocomplex containing Flag-FOXO3-3A.
Influence of FOXO3 overexpression on the PXR-mediated enhancement of hepatocyte proliferation
Finally, we investigated the influence of FOXO3 overexpression on the PXR-mediated augmentation of cell proliferation. As expected, the PXR-mediated enhancement of serum-induced cell growth disappeared via FOXO3-3A overexpression (Figure 6A). Serum treatment increased Ccnd1 and Mcm2 mRNA levels and PXR activation further increased these levels in mock-transfected cells. In the FOXO3-3A-overexpressed cells, however, the PXR-mediated enhancement was not observed (Figure 6B).
Influence of FOXO3 overexpression on the PXR-mediated enhancement of hepatocyte proliferation
We have recently reported in mice that PXR enhances xenobiotic-mediated hepatocyte proliferation, although PXR activation alone does not induce hepatocyte proliferation . In the present study, we found that PXR activation increased the FBS- or EGF-mediated proliferation of AML12 cells. Moreover, hepatocyte proliferation in mice induced by FGF19 treatment was accelerated by PCN treatment. These results suggest that PXR activation can enhance the growth factor-mediated hepatocyte proliferation as well as the xenobiotic-mediated one. Growth factor-mediated hepatocyte proliferation is a key process during liver regeneration after injury. In the present study, hepatocyte proliferation triggered by CCl4-induced liver injury was enhanced by PXR ligand treatment as expected.
We assessed the effect of PCN treatment on the chronological change of CCl4-induced hepatocyte proliferation. On CCl4 treatment, hepatocyte proliferation begins around 24 h after treatment  and, at 48 h after CCl4 treatment, most of the proliferating hepatocytes are in the G1, S or G2 phase, and barely in the M phase . In the present study we found that the number of M-phase hepatocytes was increased by PCN treatment at this time point. It seemed that cell-cycle progression was accelerated by PXR activation. Consistent with this, the number of Ki-67-positive nuclei was increased by PCN co-treatment at 48 h after CCl4 treatment, but reduced at 72 h. It suggested that the hepatocyte proliferation peak was shifted to an earlier time point by PXR activation in liver regeneration. Taken together, the results of the present study suggest that PXR activation not only increases the number of proliferating hepatocytes but also accelerates the cell-cycle progression of murine hepatocytes after growth factor stimuli.
In the present study, we found that PXR activation decreased the mRNA levels of G0/G1 or G1/S cell-cycle inhibitors. It is consistently reported that loss of these genes activates cell proliferation [21,22,28,29]. On the other hand, we had previously reported that PXR activation did not induce the genes associated with cell-cycle progression in mouse livers . Thus, PXR activation-dependent stimulation of hepatocyte proliferation may be caused by a decrease in these inhibitory factors. FOXO maintains self-renewal of cells through control of G0/G1 and G1/S phase progression . The FOXO-mediated cell-cycle regulation depends on the transcriptional up-regulation of the cell-cycle inhibitor genes [10,30]. In addition, the enhancing effect of PXR on serum-mediated cell proliferation was prevented by FOXO3-3A overexpression. From these observations, it is suggested that activated PXR prevents the FOXO3-mediated transcription of cell-cycle suppressor genes by interacting with FOXO3, because G0/G1 and G1/S checkpoints are known as rate-determining stages during cell-cycle progression. This might make hepatocytes pass these checkpoints more easily to accelerate cell-cycle progression.
In the present study, we evaluated cell proliferation induced by growth factors in an in vitro system using G0 phase-synchronized AML12 cells, because commonly used hepatoma cell lines, such as HepG2 cells, were not responsive to growth factor signals [31,32]. Moreover, they were not synchronized at the G0 phase (data not shown). Primary mouse hepatocytes are often used as a cell model reflecting in vivo situations, but neither EGF/FBS nor known liver tumour promoter (CAR or PPARα ligands) treatment increased mRNA levels of cell-cycle progression markers in primary hepatocytes (data not shown). Thus, AML12 cells might be one of the useful models for investigating hepatocyte proliferation in vitro.
Our results confirm a role of hepatic PXR as a cell-cycle regulator. We found that PXR regulated the cell cycle by preventing expression of cell-cycle inhibitors and increased sensitivity to proliferative stimuli. Thus, PXR activation enhances two types of proliferation, such as liver regeneration or abnormal proliferation by exposure to liver cancer promoters, e.g. CAR and PPARα activators. Growth factor-mediated cell proliferation has fundamental roles not only in liver regeneration but also in tumour cell growth. PXR enhanced cancer cell growth , and higher expression levels of PXR were detected in several cancerous tissues, such as endometrial cancer , prostate cancer , oesophageal squamous carcinoma  and breast cancer , when compared with normal tissues. Based on these facts and our previous findings  and those of the present study, PXR may up-regulate the proliferation of several types of cells. Moreover, these observations suggest that PXR also contributes to some extent to cancer development and progression.
In summary, we have demonstrated that PXR activation enhances growth factor-mediated hepatocyte proliferation using in vitro and in vivo models. Our results suggest that this enhancement results from the PXR-mediated down-regulation of cell-cycle suppressor genes, such as Cdkn1b and Rbl2, through the inhibition of FOXOs’ functions. CAR and PPARα, xenobiotic-responsive nuclear receptors such as PXR, are known to enhance hepatocyte proliferation in rodents, possibly through regulation of the expression of genes involved in cell-cycle progression [38,39]. However, the activation of these two receptors is not considered to be associated with hepatocyte proliferation or liver tumour formation in humans, although there is little understanding of the reason for this [39–41]. Cell-cycle regulation by PXR is considered to have occurred through different mechanisms from CAR and PPARα; therefore, in future studies, whether PXR activation enhances hepatocyte proliferation in human livers and is associated with an increase in sensitivity to chemical hepatocarcinogenesis in humans needs to be investigated. This is very important from the viewpoint of chemical safety evaluation because human PXR is activated by a wide variety of chemical compounds such as drugs, pesticides and food constituents .
K. Yoshinari, S. Benoki, R. Shizu and T. Abe conceived of and designed the experiments; R. Shizu, T. Abe, S. Benoki, M. Takahashi, S. Kodama, M. Miayata and K. Yoshinari performed the experiments and analysed the data; and R. Shizu, T. Abe, A. Matsuzawa and K. Yoshinari wrote the paper. R. Shizu and T. Abe contributed equally to this work.
The authors are grateful to Professor Jeffrey L. Staudinger, PhD (University of Kansas, Lawrence, KS, U.S.A.) for the gift of the Pxr-null mice.
This study was supported in part by Grant-in-Aid from the Ministry of Education, Culture, Sports, Sciences and Technology of Japan [22390027, 24659061] and a grant from Takeda Science Foundation.
protein kinase B
constitutive androstane receptor
Dulbecco’s modified Eagle’s medium
epidermal growth factor
fibroblast growth factor
Forkhead box O
proliferating cell nuclear antigen
peroxisome proliferator-activated receptor α
pregnane X receptor
These authors contributed equally to this work.
Current address: Department of Food Science and Technology, National Fisheries University, 2-7-1 Nagatahonmachi, Shimonoseki, Yamaguchi 759-6595, Japan.