Role of ornithine decarboxylase antizyme inhibitor in vivo

Authors

  • Hua Tang,

    Corresponding author
    1. Key Laboratory of Molecular Infectious Diseases, Ministry of Education, Chongqing Medical University, Chongqing 400016, China
    2. Department of Developmental Genetics, Institute of Molecular Embryology and Genetics, Kumamoto University, Kumamoto 860-0811, Japan
      * Correspondence: tanghua86162003@yahoo.com.cn or yamamura@gpo.kumamoto-u.ac.jp
    Search for more papers by this author
  • Kimi Ariki,

    1. Department of Developmental Genetics, Institute of Molecular Embryology and Genetics, Kumamoto University, Kumamoto 860-0811, Japan
    Search for more papers by this author
  • Makiko Ohkido,

    1. Department of Molecular Biology, The Jikei University School of Medicine, Minato-ku, Tokyo 105-8461, Japan
    Search for more papers by this author
  • Yasuko Murakami,

    1. Department of Molecular Biology, The Jikei University School of Medicine, Minato-ku, Tokyo 105-8461, Japan
    2. Faculty of Pharmacy, Research Institute of Pharmaceutical Sciences, Musashino University, Tokyo 202-8585, Japan
    Search for more papers by this author
  • Senya Matsufuji,

    1. Department of Molecular Biology, The Jikei University School of Medicine, Minato-ku, Tokyo 105-8461, Japan
    Search for more papers by this author
  • Zhenghua Li,

    1. Department of Developmental Genetics, Institute of Molecular Embryology and Genetics, Kumamoto University, Kumamoto 860-0811, Japan
    Search for more papers by this author
  • Ken-ichi Yamamura

    Corresponding author
    1. Department of Developmental Genetics, Institute of Molecular Embryology and Genetics, Kumamoto University, Kumamoto 860-0811, Japan
      * Correspondence: tanghua86162003@yahoo.com.cn or yamamura@gpo.kumamoto-u.ac.jp
    Search for more papers by this author

  • Communicated by: Masayuki Yamamoto (Tohoku University)

* Correspondence: tanghua86162003@yahoo.com.cn or yamamura@gpo.kumamoto-u.ac.jp

Abstract

Ornithine decarboxylase (ODC) antizyme inhibitor (AZI) has been shown to regulate ODC activity in cell cultures. However, its biological functions in an organism remain unknown. An embryonic stem (ES) cell clone was established, in which the Azin1 gene was disrupted by the gene trap technique. To identify the function of Azin1 gene in vivo, a mutant mouse line was generated using these trapped ES cells. Homozygous mutant mice died at P0 with abnormal liver morphology. Further analysis indicated that the deletion of Azin1 in homozygous mice resulted in the degradation of ODC, and reduced the biosynthesis of putrescine and spermidine. Our results thus show that AZI plays an important role in regulating the levels of ODC, putrescine and spermidine in mice, and is essential for the survival of mice.

Introduction

The polyamines are abundant multivalent organic cations, largely bound to RNA and DNA in cells. They are essential for growth and differentiation of mammalian cells, and their synthesis, catabolism and transport are highly regulated (Heby & Persson 1990; Wallace et al. 2003; Gerner & Meyskens 2004). Ornithine decarboxylase (ODC) is a key enzyme in the biosynthesis of the polyamines including putrescine, spermidine and spermine. ODC gene expression is stimulated by various mitogens (Sreevalsan et al. 1980; Janne et al. 1984; Greenberg et al. 1985; Heby & Persson 1990; Butler et al. 1991). Elevated ODC activity was observed in tumor cells, and over-expression of ODC could lead to neoplastic transformation, how this enzyme is regulated has attracted much attention (Auvinen et al. 1992; Moshier et al. 1993; Shantz & Pegg 1994; Auvinen et al. 1997; O’Brien et al. 1997; Feith et al. 2001; Tang et al. 2004; Pegg 2006). The polyamines, mainly at the translational and post-translational levels, involved feedback-regulate ODC: high polyamine concentrations decrease, and low polyamine concentrations increase, ODC activity (Stjernborg et al. 1991; Wallace et al. 2003).

Antizyme (AZ), a breviate word for “anti-enzyme for ornithine decarboxylase”, was first reported as an inhibitor of a key enzyme in polyamine biosynthesis (Coffino 2001). It is a regulatory protein of ODC. AZ binds directly to ODC (Murakami et al. 1985) and the AZ-ODC complex is degraded by the 26S proteasome. In vitro, ODC antizyme inhibitor (AZI) is involved in the regulation of ODC. It is a homolog of ODC. AZI has a higher affinity for antizyme than for ODC (Fujita et al. 1982; Kitani & Fujisawa 1989; Murakami et al. 1989a), and as such can displace ODC from the ODC-antizyme complex and prevents ODC from being degraded by the 26S proteasome. Therefore, AZI can prevent ODC from degradation initiated by antizyme (Nilsson et al. 2000). AZI is induced in cells and tissues following growth stimulation (Kim et al. 2006; Mangold 2006), and is elevated in gastric tumors (Jung et al. 2000). The ability to regulate AZ (Mangold & Leberer 2005) indicates that AZI may play an important role in regulating cellular polyamine levels and cell growth. Thus far, however, no reports on its function in vivo have been published.

The gene trap strategy, using embryonic stem cells (ES cells), is a powerful method for both identification of genes and subsequent establishment of mutant mice line (Majlinda & Nicholas 2000; Wiles et al. 2000). The Azin1 gene is highly conserved in invertebrate and vertebrate animals. For example, the rat gene is 97% homologous to the mouse gene and the human gene is 96% homologous. This conservation across the species indicates that this gene plays important functions. Some studies showed that its protein (AZI) was important for cell growth in vitro (Nilsson et al. 2000). However, the expression pattern of this gene during embryogenesis is unclear and no knock out mice have been created to study the biological functions of this gene. Thus, the function of this gene in vivo remained unclear. To address this issue, we used the gene trap technology and the trap vector pU-17 to generate a mouse line with a disrupted Azin1 gene (Taniwaki et al. 2005), in this mouse line, Ayu17-689, the trap vector was inserted into the Azin1 gene. The homozygous mutant mice died at P0. Further analysis revealed the reduction of ODC, putrescine and spermidine in these Azin1 homozygous mutant mice. Our results thus indicated an important role of AZI in mouse embryogenesis.

Results

Integration of the trap vector into the Azin1 gene

We had obtained a trapping ES colony named Ayu17-689 that strongly expressed β-galactosidase and neomycin phosphotransferase (data not shown). Southern blot analysis using a partial pSP73 vector fragment as the probe showed that this ES cell line possessed a single trapping vector integration in the genome (Fig. 1A). When the integration of this trapping vector and the mouse genomic DNA were analyzed by plasmid rescue, the trapping vector was found to have a 522-bp deletion at its 5′ terminus (data no shown). The trap vector was inserted in the intron between exon 2 and exon 3 before the translation start codon of Azin1 gene. There were no other mutations detected in the mouse genome. The relationship between integrated vector DNA and mouse genomic DNA is illustrated in Fig. 1B.

Figure 1.

Single integration and the location of trapping vector in Azin1 gene. (A) Southern blotting analysis for single copy of trapping vector. Genomic DNA from Ayu17-689 ES cells were digested with several of restriction enzymes and hybridized with a part fragment of trapping vector. (B) Location of trapping vector in Azin1 gene.

Expression pattern of AZI

To characterize the role of AZI during mouse development, Azin1 gene trapped ES cells were injected into blastocysts to generate germ-line chimeras. Heterozygous mice were fertile, showed no morphological abnormalities, and had a normal life span. As β-galactosidase was transcribed under the control of the Azin1 promoter, the expression pattern of LacZ was the same as Azin1 in these mice. LacZ expression analysis was carried out in five heterozygous mice at 3 weeks of the developmental stages. The expression of the reporter LacZ gene was ubiquitously detectable in all tissues checked (Fig. 2A). Northern-blot analysis also showed that Azin1 ubiquitously expressed in all tissues of adult mice (Fig. 2B), indicating that Azin1 might have a housekeeping function in mouse development. To better understand in vivo expression pattern of AZI in adult mice, β-galactosidase activity was additionally examined in several organ sections dissected from five heterozygous Azin1 mice. AZI expression was found in several organs tested, including valves of heart, mesenchymal cells in kidney and liver cells (Fig. 2C).

Figure 2.

Expression pattern of Azin1 and AZI. (A) RT-PCR analysis for expression of Azin1 and LacZ. Glyceraldehyde-3-phosphate dehydrogenase (G3PDH) was the loading control. (B) Northern blotting analysis for Azin1 expression. RNA from tissues were isolated and hybridized with a partial Azin1 cRNA. G3PDH was the loading control. (C) Analysis for AZI expression. Tissue sections from Azin1 heterozygous mice were immunostained with anti-X-gal monoclonal antibody (brown color). Similar results were obtained from a duplicate experiment.

Generation and genotyping of the mutant mouse line

We established a mouse line by using this trapped ES colony. Mouse genotyping was confirmed by PCR using the primers derived from the genomic sequence flanking the trapped vector and primers of the trapping vector (Fig. 3A). Northern blot analysis showed that the expression levels of Azin1 in mutant mice were strongly suppressed, especially in homozygous mutant mice, comparing with the wild-type mice (Fig. 3B). RT-PCR also revealed that Azin1 expression was inhibited in all tissues of heterozygous mice (Fig. 3C).

Figure 3.

Generating Azin1 mutant mice. (A) Genotyping with PCR. NeoF, NeoR primers on the trapping vector. GF4, GR3 primers on the Azin1 gene, flanking the trapping vector. Genotyping for each mouse was repeated twice. (B) Northern-blot analysis for Azin1 mRNA in mutant mice.+/+, +/– and –/– indicate wild-type, heterozygous and homozygous mice, respectively. (C) RT-PCR analysis for Azin1 expressions in several tissues from wild-type and heterozygous mice. Similar results were obtained from a duplicate experiment.

Phenotype of homozygous mutant mice

Twenty F3 heterozygous mice were mated to analyze the phenotype of homozygous mutant mice. No obvious phenotypic abnormalities were observed in any of heterozygous mutant mice. However, almost all of the homozygous mice died at P0 (8 mice) (Fig. 4A) with slightly reduced body weight (Fig. 4B). In order to understand the cause of death, the histological examinations of many tissues of six homozygous mutant mice at E19.5 were carried out. Only the liver in homozygous mice was found abnormal compared with normal littermates. The liver of homozygous mutant mouse was filled with round cells (as the black arrows indicated) (Fig. 4C). These cells were not stained by PECAM-1, a specific marker of endothelial cells (Fig. 4D). No significant change of proliferation was observed in homozygous mutant mice as revealed by immunostaining with murine Ki67 monoclonal antibody, a marker of cell proliferation (Fig. 4E).

Figure 4.

Phenotype of Azin1 mutant mice. (A) Azin1 homozygous mutant mice died at newborn stage. Genotyping results for mice at different developmental stages. (B) Body weight comparison of wild-type and mutant mice (+/+ wild-type 3 mice, +/– heterozygous 4 mice, –/– homozygous 4 mice). Unit: gram. Error bars represent the standard error of the mean. *P < 0.05 on comparing wild-type with homozygous mice; ‡P < 0.05 on comparing heterozygous with homozygous mice. (C) Hematoxylin and Eosin (HE) staining of the livers from wild-type (+/+) and homozygous mutant (–/–) mice. (D) PECAM-1 immunostaining of the livers from wild-type (+/+) and homozygous mutant (–/–) mice. (E) Ki67 immunostaining of the livers from wild-type (+/+) and homozygous mutant (–/–) mice. Black arrows in C and D indicating the round cells.

Reduced ODC in mutant mice

Because the expression of Azin1 in homozygous mice was strongly suppressed, we presumed that the expression of ODC might have changed. For this reason, we conducted Northern blot and Western blot analyses to examine whether the expression of ODC was affected in our three mutant mice at E19.5. The result showed that the ODC expression was increased at the RNA level faintly (Fig. 5A) but significantly decreased at the protein level in mutant mice when they are compared with normal mice (Fig. 5B). These data indicated that AZI could regulate ODC at both RNA and protein levels.

Figure 5.

ODC expression was altered in mutant mice. (A) ODC transcriptions were up-regulated in mutant mice. Northern blotting analysis for ODC RNA in wild-type (+/+) and mutant (–/–) mice (brain, left three lanes; liver, right three lanes). G3PDH was the loading control. (B) The amount of ODC protein was reduced in mutant mice. Western blotting analysis for ODC protein in liver from wild-type and mutant mice. Actin was the loading control. Similar results were obtained from a duplicate experiment.

Because ODC was the key enzyme that regulates the synthesis of putrescine and spermidine, its reduced expression should also affect the synthesis of putrescine and spermidine. We analyzed (3 wild-type mice, 4 heterozygous mice, 4 homozygous mice) the levels of putrescine, spermidine and spermine in the liver, brain and kidney of wild-type, heterozygous and homozygous mice at E19.5. The level of putrescine was greatly reduced in all three tissues of mutant mice (Fig. 6A). Spermidine was also reduced in all of these tissues of mutant mice (Fig. 6B). However, spermine was increased (Fig. 6C). These data further showed that in our mutant mice, the amount of ODC was reduced and ODC played a critical role in the synthesis of putrescine and spermidine.

Figure 6.

Polyamine in mutant mice. 3 wild-type mice, 4 heterozygous mice and 4 homozygous mice at E19.5 were analyzed. (A) Putrescine in mutant mice. (B) Spermidine in mutant mice. (C) Spermine in mutant mice. The tissues of mice were combined for each genotype. These data were on pooled samples, they represent the average of tissue levels of each genotype.

Discussion

We have generated a mouse line which analyzes loss of function of AZI using the gene trap method. The expression of AZI was decreased in this mutant line. This mouse line allowed us to analyze the function of AZI gene in vivo.

In our studies, the expression pattern of AZI was analyzed both at the RNA level and at the protein level. The Azin1 mRNA expression patterns were demonstrated by RT-PCR and Northern blot (Fig. 2A,B). The principle of gene trapping is the random insertion of a reporter gene, which can only be expressed when this reporter gene has been integrated into an active mouse gene locus. Thus, the expression of the reporter gene will mimic the expression pattern of the endogenous gene that is inserted by the reporter (Evans et al. 1997). As the trap vector was inserted into Azin1 gene, the expression of report fragment β-galactosidase was controlled by the promoter of Azin1 gene. Thus, LacZ expression pattern resembled that of the Azin1 gene (Fig. 2A). Furthermore, the AZI protein expression pattern could be analyzed using the anti β-galactosidase monoclonal antibody instead of the AZI antibody. By using this approach, we demonstrated that the AZI protein was strongly expressed on heart valve cells, and kidney mesenchymal cells (Fig. 2C).

In Northern-blot hybridization we show that mice have two isoforms of Azin1 mRNA. In the rat, cDNAs encoding two isoforms have been cloned (accession numbers D50734 and D89983) by Murakami et al. (1996) and Koguchi et al. (1997). Despite their major differences in size (2.3 and 4.2 kb), the two rat Azin1 cDNAs have identical amino acid coding regions. Our trap vector can knock out these two isoforms of Azin1 mRNA (Fig. 3B). Both Azin1 mRNAs exhibit similar patterns of expression, suggesting that they are transcribed from one gene, rather than two. Therefore, this mouse line could be used as an Azin1 gene knock out model.

ODC is a highly regulated enzyme in eukaryotic organisms. The accumulation of ODC protein is controlled by gene transcription, mRNA translation and enzyme degradation. In our experiments, ODC was found be regulated by AZI at both the protein level and the RNA level. In our mutant mouse, the expression of AZI was reduced and the ODC RNA level was slightly higher than normal mice (Fig. 5A). The transcription of ODC RNA could be feedback regulated by polyamines such as putrescine, spermidine and spermine (Davis et al. 1992; Shantz & Pegg 1999). In the mutant mice, the ODC transcription could be stimulated by the low levels of putrescine and spermidine. This may be the reason why the ODC mRNA level was slightly higher in mutant mice than in the normal mice. However, at the protein level, the amount of ODC was reduced in homozygous mice (Fig. 5B). This may be because when the expression of AZI is eliminated there will be fewer competitive molecules to bind to antizyme, which plays an important role to facilitate the degradation of ODC by the 26S proteasome. This speculation is supported by the observation that AZI can suppress ODC degradation initiated by antizyme in vitro (Nilsson et al. 2000). Our data also showed that AZI could inhibit the degradation of ODC by antizyme.

ODC has a short half-life in vivo and is induced during the initiation of growth processes (Tabor & Tabor 1984; Luk et al. 1986; Luk & Casero 1987; Pegg 1988). In non-cycling or non-growing cells, ODC activity is low or undetectable, whereas it is increased in growth. Increases in ODC activity are important for hepatic regeneration, and polyamine metabolism plays an important role in the increased DNA and protein synthesis in hepatic proliferation (Russell & Snyder 1968; Luk 1986). In rat liver regeneration after partial hepatectomy, ODC activity increases tenfold within 4 h (Russell & Snyder 1968). ODC is important for liver development. This may be the reason why AZI deficient mouse displays only liver abnormality.

ODC is a key enzyme in the biosynthesis of the polyamines including putrescine, spermidine and spermine, which are essential for growth and differentiation of mammalian cells. The results of our work showed that the expression levels of putrescine were lower in the homozygous mutant mice, particularly in the brain and kidney. In these two organs, the heterozygous knockout of AZI might have affected the putrescine level, too. In addition, the homozygous mutants showed lower spermidine and higher spermine levels. Because the syntheses of spermidine and spermine compete with each other for the supply of aminopropyl group from decarboxylated S-adenosylmethionine, these results are consistent with the observed decrease of putrescine level in tissues. The AZI knock down resulted in the reduction of ODC level, which then caused the decrease of polyamines. It is well-known that the polyamines are essential for many molecular processes such as DNA replication, RNA synthesis and protein translation (Snyder 1989; Pendeville et al. 2001). Lack of polyamines (induced by knocking down the AZI) might be one reason of the death of homozygous mutant mice. More work will be needed to further understand the mechanism.

Experimental procedures

Animals

Mice were kept the SPF facility (Center for Animal Resources and Development, Kumamoto University) at 24 °C and 55% relative humidity on a12/12-h light (0800–1900 h)/dark (1900–0800 h) cycle and fed with breeding quality ad libitum. Chimeric mice were produced by the aggregation method using gene trap ES clones and subsequently mated with C57BL/6 (CLEA) females to obtain F1 heterozygotes with gene trap insertions. Five heterozygous males were mated with fifteen heterozygous females to produce wild-type mice, heterozygous mice, and homozygous mice. Offspring mice were weaned at 3 weeks of age. At that time, tail biopsies were taken for genotyping.

Trapping vector

The trapping vector, pU17 (Taniwaki et al., 2005), consists of a mouse En2 splicing acceptor, a lox71 site, β-galactosidase, neomycin phosphotransferase, loxP, polyA, and a lox2272 site in the pSP73 plasmid. When the pU17 is inserted into the mouse endogenous gene locus, the β-geo gene will be expressed by the promoter of trapped gene, thus makes it possible to select the trap clones in the presence of G418. Because this trapping vector has lox71 at the 5′ flanking region and loxP at the 3′ flanking region of the β-geo, the β-geo can be replaced with a plasmid DNA with lox66 and loxP at 5′ and 3′ ends, respectively, by expression of the Cre recombinase, Thus, the plasmid rescue can be easily carried out with the aim of cloning the mouse genomic fragments even when pSP73 is deleted upon integration of the pU17 (Araki et al. 2002).

Cell culture and electroporation

TT2 is a hybrid ES cell line derived from a C57BL/6 female mouse and a CBA male mouse. The TT2 cell line was cultured as described previously (Yagi et al., 1993). For co-electroporation experiments, a targeting plasmid and pCAGGS-Cre were used together in their circular forms. The ES cells [1 × 107 cells/0.8 mL in phosphate-buffer saline (PBS)] were electroporated at 200 V and 960 µF, and after 24 or 48 h they were selected with G418 at 200 µg/mL or chloramphenicol or ampicillin for 7 days. Colonies were picked and expanded for staining with X-gal and DNA analysis.

Northern blot analysis

RNA was isolated from various tissues at E19.5 with Trizol reagent (Invitrogen, Carlsbad, CA). Briefly, mouse tissues were homogenized in Trizol reagent and RNA extracted with chloroform and precipitated with isopropanol. The resulting pellet was finally resuspended in TE buffer (10 mm Tris–HCl, pH 7.5, 1 mm EDTA). Total RNA was separated by gel electrophoresis and quantified by spectrophotometry. For Northern blot analysis, 20 µg of total RNA was run on a 1% formaldehyde-agarose gel and transferred to a nylon membrane (N-Hybond). Hybridizations were carried out under stringent conditions with a DIG-labeled Azin1 partial cRNA, which was prepared from a PCR product that had been cloned using the pGEM-T easy vector kit (Promega, Madison, WI). Primers for the PCR amplification of Azin1 were 5′-ttgacgatgcgaactactcc-3′ and 5′-tgaatgccagatccttcagg-3′; and primers of ODC were 5′-taaggacgagtttgactgcc-3′ and 5′-tcttcagatccaggaaagcc-3′. The hybridization signals were detected using FuJi film.

Plasmid rescue

Genomic DNA (20 µg) was digested with the appropriate restriction enzymes and ligated in a reaction volume of 400 µL to obtain circular molecules. After phenol/chloroform extraction and ethanol precipitation, the DNA was suspended in 10 µL nuclease free water. Half of the DNA solution was used to transform E. coli STBL2 cells (Life Technologies, Gaithersburg, MD) by electroporation using a Bio-Rad Gene Pusler according to the manufacturer's recommendations (Bio-Rad, Hercules, CA). The electroporated cells were incubated in 1 mL of Circle Grow medium (BIO 101, Vista, CA) at 30 °C for 1 h with shaking, and then concentrated and plated on selective LB/agar plates using ampicillin for the plasmids selection. The rescued plasmids were analyzed by restriction mapping and sequencing.

PCR genotyping

DNA was isolated from tail biopsies, as described previously. To genotype the offspring heterozygous mice, we PCR-amplified of 506 bps of the neomycin gene using 5′-AGA GGC TAT TCG GCT ATG AC-3′ as a forward primer and 5′-CAC CAT GAT ATT CGG CAA GC-3′ as a reverse primer. To genotype the offspring wild-type mice, primers on the mouse genome located in the flanking regions of the trapping vector were used: 5′-AGTTAGTTCCACCTTCCACC-3′ as a forward primer and 5′-TAGAGTACTGGAGACAGCTC-3′ as a reverse primer. PCR amplification of 410 bps fragment was carried out in a reaction mixture containing 50 mm KCl, 10 mm Tris–HCl, 2 mm MgCl2, 0.1% Triton X-100, and 0.2 mm dNTPs. The PCR cycling consisted of initial denaturation at 94 °C for 5 min followed by 30 cycles of 1 min at 94 °C (denaturation), 1 min at 60 °C (annealing) and 1 min at 72 °C (extension) and was terminated with a final extension at 72 °C for 10 min. The PCR products were separated on a 1% agarose gel and visualized with ethidium bromide staining.

Western-blot

Liver extracts were isolated from normal and mutant mice at E19.5, separated in 12% SDS-polyacrylamide gels, electroblotted onto nitrocellulose sheets, developed by ECL chemiluminescence system (Amersham), Exposed to a film. Monoclonal anti-ornithine decarbixylase antibody was purchased from Sigma (Saint Louis, MO).

Polyamine measurements

To measure polyamines, livers, kidneys and brains were isolated from 3 wild-type, 4 heterozygous and 4 homozygous mice at E19.5, respectively, and frozen in liquid nitrogen immediately. Tissues were washed twice with phosphate-buffered saline (20 mm sodium phosphate, 140 mm NaCl, pH 7.4). Tissues were homogenized with two volumes (for liver and brain) or four volumes (for kidney) of homogenizing buffer (25 mm Tri-HCl, 0.01% Tween 80, 1 mm dithiothreitol, 1 mm EDTA, pH 7.5). Each homogenate was mixed with an equal volume of 8% perchloric acid. The mixture was vortexed, kept on ice for 5 min, and centrifuged at 15 000 rpm at 4 °C 5 min. The supernatant was subjected to polyamine analysis using a high perhormance liquid chromatography (Shim-Pack ISC-05/S0504, Shimadzu) and fluorometry as described previously (Murakami et al., 1989b).

Quantification analysis

Western and Northern blots results were quantified in densitometry with Quanty One software (Bio-Rad).

Acknowledgements

Authors thank Nakada for staining of all tissue sections and Jing-Hsiung James Ou for critical reading of the manuscript. This study was supported in part by KAKENHI (Grant-in-Aid for Scientific Research) on Priority Areas “Integrative Research Toward the Conquest of Cancer” from the Ministry of Education, Culture, Sports, Science and Technology of Japan, a grant from the Osaka Foundation of Promotion of Clinical Immunology; a grant by the Nature Science Foundation of Chongqing (2005BB5035) and a grant from the National Nature Science Foundation of China (30671080).

Ancillary