CCR5 Plays a Critical Role in Obesity-Induced Adipose Tissue Inflammation and Insulin Resistance by Regulating Both Macrophage Recruitment and M1/M2 Status

C57BL/6J mice and ob/ob mice were purchased from Charles River Laboratories (Yokohama, Japan). Ccr5^−/− mice were provided by K. Matsushima (Tokyo University, Tokyo, Japan) (21). C57BL/6J mice and Ccr5^−/− mice were fed a normal chow (NC) consisting of 10% of calories from fat (CRF-1; Charles River Laboratories) or an HF diet consisting of 60% fat (Research Diets, New Brunswick, NJ). All animal procedures were performed in accordance with the standards set forth in the Guidelines for the Care and Use of Laboratory Animals at Kanazawa University.

Quantitative real-time PCR was performed on a CFX384 (Bio-Rad) using the SYBR Green Master Mix (Takara, Japan) as described previously (22). Primers used in the real-time PCR are shown in Supplementary Table 1.

Epididymal fat pads from male C57BL/6J mice fed an NC or HF diet were minced and digested for 30 min at 37°C with type II collagenase (Sigma-Aldrich) in PBS containing 2% BSA (pH 7.4). The cell suspension was filtered and then spun at 300g for 5 min to separate the floating adipocyte fraction from the stromal vascular cell (SVC) pellet. SVCs were resuspended in PBS supplemented with 2% FBS and incubated with Fc-Block (BD Bioscience), followed by an incubation with fluorochrome-conjugated antibodies (Supplementary Table 2). Cells were analyzed using FACSAriaII (BD Bioscience). Data analysis and compensation were performed using FlowJo (Tree Star).

Total tissues were homogenized and sonicated in radioimmunoprecipitation assay lysis buffer. The primary antibodies used are shown in Supplementary Table 3.

The paraffin-embedded adipose tissue sections were stained with the combinations of anti-F4/80 (AbD Serotec) and anti-CCR5 (a gift from N. Mukaida) antibodies. Alexa Fluor 488 donkey anti-rat IgG (Invitrogen) was used as a secondary antibody to detect F4/80⁺ cells. Simultaneously, Alexa Fluor 594 goat anti-rabbit IgG (Invitrogen) was used to detect CCR5⁺ cells. The adipocyte area was measured in 1,000 or more cells from 20 randomly selected fields at 200-fold magnification for each mouse by analyzing six mice per group.

Plasma triglyceride (TG), free fatty acid (FFA), cholesterol, glucose, and insulin levels and hepatic TG concentration were measured as described previously (22,23). Glucose tolerance test (GTT) was conducted after an overnight fast. After baseline blood collection, mice were injected intraperitoneally with glucose (2 g/kg). Insulin tolerance test (ITT) was performed after a 4-h fast. Mice were injected intraperitoneally with human insulin (0.75 units/kg).

Bone marrow (BM) cells were harvested from femurs and tibias of 7- to 8-week-old Ccr5^−/− or wild-type (WT) mice and transplanted via tail vein into lethally irradiated (10 Gy) recipient WT mice with a minimum cell dose of 10⁷ cells. Transplanted mice were allowed 4 to 5 weeks to reconstitute their hematopoietic systems with Ccr5^−/− or WT BM.

All data are presented as means ± SEM. Differences in the mean values between two groups were assessed by two-tailed Student t test. Differences in mean values among more than two groups were determined by ANOVA. P < 0.05 was considered statistically significant.

MCP-1 and its receptor CCR2 are important for ATM recruitment and the development of insulin resistance (15–17). However, other chemokine systems also may play a role in these processes. Therefore, we analyzed gene expression levels of several chemokines and their receptors in WAT of ob/ob and DIO mice using quantitative real-time PCR (Fig. 1). We studied mRNA expression of CCR2 and its ligands, CCL2, also known as MCP-1 (CCL2/MCP-1), CCL7/MCP-3, and CCL8/MCP-2,and that of CCR5 and its ligands, CCL3/MIP-1α, CCL4/MIP-1β, CCL5/RANTES, and CCL8/MCP-2 in epididymal WAT from lean C57BL/6J mice maintained on an NC (WT), from C57BL/6J mice made obese on an HF diet starting at age 5 weeks for 10 weeks (DIO), and from ob/ob mice at age 15 weeks. Compared with lean mice, mRNA expression for CCR5 and its ligands and protein expression for CCR5 were markedly increased in WAT in both DIO mice and ob/ob mice at age 15 weeks (Fig. 1A and B). Consistent with previous reports (6,7,16), expression of mRNA for CCR2 and its ligands was also increased in WAT in both DIO mice and ob/ob mice at age 15 weeks (Fig. 1A). DIO mice had increased expression of the macrophage marker CD68 in WAT as compared with WT mice as reported previously (6,7) (Fig. 1C). Increased expression of CCR5 and its ligands in WAT of DIO mice persisted until at least age 20 weeks (Fig. 1C). In addition to the marked upregulation of CCR5 mRNA in WAT, there was a significant increase in expression in liver, although absolute expression was much lower than in WAT (Fig. 1D). No other tissue showed an increase in expression of CCR5 mRNA.

We next examined the cellular sources responsible for CCR5 expression in obese and lean WAT. Expression of mRNA for CCR5 was higher in the stromal vascular (SV) fraction than adipocyte fraction both from WT and DIO mice (Fig. 2A). Moreover, CCR5 expression in the SV fraction in DIO was markedly increased compared with WT mice. In addition, CCR5 mRNA was highly expressed in 3T3L1 preadipocytes, whereas its expression progressively decreased during adipocyte differentiation (Supplementary Fig. 1). These data suggest that obesity-induced CCR5 expression in adipose tissue is primarily derived from SVCs. Furthermore, immunofluorescence analysis of WAT in DIO mice revealed that CCR5 was expressed by F4/80⁺ macrophages in crown-like structures (Fig. 2B).

To quantify CCR5⁺ ATMs in lean and obese mice, we performed fluorescence-activated cell sorter (FACS) analysis on SVCs isolated from epididymal WAT of WT or DIO mice. In FACS analysis, ATMs were identified as propidium iodine (PI)⁻CD45⁺NK1.1⁻CD3⁻CD19⁻TER119⁻CD11b⁺ F4/80⁺ cells (Supplementary Fig. 2). Consistent with previous studies (6,10), ATMs were markedly increased in HF diet–fed mice compared with NC-fed mice (Fig. 2C). The total number of ATMs isolated from epididymal fat pads increased in mice fed an HF diet by 12.2-fold compared with NC-fed mice. When gated for CCR5, ATMs exhibited a high level of CCR5 expression in response to HF diet, whereas ATMs in NC-fed mice showed a low level of expression (Fig. 2D). Quantification of the mean fluorescence intensity indicated that expression of CCR5 in ATMs from DIO mice was significantly higher than that in ATMs from WT mice (Fig. 2E). Only a small percentage of ATMs coexpressed CCR5 in WT mice. However, DIO mice had a significant increase in the percentage of CCR5⁺ cells within ATMs (Fig. 2F). DIO mice had an 11.9-fold increase in the total number of CCR5⁺ cells within ATMs (Fig. 2G). Normalizing these data to fat pad weight indicated that there was a 2.9-fold increase in the content of CCR5⁺ ATMs in DIO mice compared with WT mice (Fig. 2H). Collectively, these results indicate that CCR5⁺ macrophages accumulate in WAT of obese mice.

To determine whether CCR5 is required for obesity-induced ATM recruitment and insulin resistance in vivo, we examined metabolic phenotype of Ccr5^−/− mice. We placed Ccr5^−/− and control mice on an NC or HF diet for 16 weeks, beginning at age 5 weeks. WT and Ccr5^−/− mice had similar body weights (Fig. 3A) and consumed similar quantities of food (Fig. 3B) on each diet. At 20 weeks of age, body weight, adipose tissue weight, and fasting plasma glucose levels did not differ significantly between Ccr5^−/− and WT mice maintained on either an NC or HF diet (Table 1). However, on an HF diet, Ccr5^−/− mice had plasma insulin levels in a fed state that were decreased by 50% compared with WT mice (Fig. 3C). Plasma FFA and TG levels were also decreased in Ccr5^−/− mice on both NC and HF diets compared with WT controls (Table 1). Hepatic TG content and the expression of lipogenic genes were reduced in Ccr5^−/− mice on an HF diet (Fig. 3D and E). GTTs indicated that Ccr5^−/− mice fed NC had slightly better glucose tolerance (Fig. 3F). Moreover, HF diet–induced glucose intolerance and hyperinsulinemia were significantly improved in Ccr5^−/− mice (Fig. 3G). ITTs also showed that Ccr5^−/− mice on an HF diet had increased insulin sensitivity (Fig. 3H). Therefore, CCR5 deletion is associated with protection from obesity-induced glucose intolerance, insulin resistance, hypertriglyceridemia, and hepatic steatosis.

Given that obesity-associated increases of CCR5 and its ligands were largely restricted to adipose tissue (Fig. 1), we next investigated the impact of CCR5 deficiency on inflammation and insulin action of adipose tissue. Infiltration of macrophage into adipose tissue and crown-like structure formation induced by HF diet were markedly reduced in Ccr5^−/− mice (Fig. 4A and B). It is important that there was no significant difference in adipocyte size between Ccr5^−/− and WT mice (Fig. 4C). Examination of gene expression in epididymal WAT showed that obese Ccr5^−/− mice had significantly decreased expression of macrophage markers F4/80 and Cd68. It is noteworthy that markers for M1-type macrophages, Cd11c, and inflammatory cytokines derived from M1-type macrophages Tnf, iNOS, and PAI1 were decreased in WAT of Ccr5^−/− mice compared with WT mice (Fig. 4D). However, no significant differences were seen in gene expression of adipokines, such as adiponectin and leptin, between genotypes. Plasma levels of TNF-α were significantly lower in Ccr5^−/− mice than in WT mice fed an HF diet (Fig. 4E). However, plasma levels of other cytokines or adipokines, including IL-6, IL-1β, adiponectin, and leptin, were not significantly different between Ccr5^−/− and WT mice (Supplementary Fig. 3). Furthermore, inflammatory signals were attenuated in WAT of Ccr5^−/− mice (Fig. 4F). Phosphorylation of Jun NH₂-terminal kinase (JNK) and p38 mitogen-activated protein kinase (MAPK) was decreased in Ccr5^−/− mice on both NC and HF diet. Nuclear factor-κB (NF-κB) p65 phosphorylation was also attenuated in Ccr5^−/− mice on an HF diet. Decreased inflammatory signals resulted in improved insulin signaling characterized by decreased Ser phosphorylation of insulin receptor substrate (IRS)-1 in WAT of Ccr5^−/− mice compared with WT mice (Fig. 3F). In addition, insulin-stimulated Tyr phosphorylation of insulin receptor (IR)-β subunit (p-IRβ), IRS-1 (p-IRS1), and Akt (p-Akt) was enhanced in WAT of Ccr5^−/− mice compared with WT mice (Fig. 4G). However, the differences of insulin signaling were lost in primary adipocytes (Supplementary Fig. 4). In addition, the increase in insulin signaling in either the liver or muscle of Ccr5^−/− mice was smaller than that observed in adipose tissue (Supplementary Fig. 5). These results suggest that CCR5 is required for obesity-induced ATM recruitment and subsequent adipose tissue inflammation and insulin resistance.

A reduction in adipose tissue inflammation and improved insulin action as a result of CCR5 deficiency prompted us to investigate the ATM subsets (10,11,24). To quantify M1- and M2-type ATMs in lean and obese Ccr5^−/− mice, we performed FACS analysis on SVCs isolated from epididymal WAT of Ccr5^−/− and WT mice on different diets. The gating strategies to determine populations of M1- and M2-type ATMs are depicted in Supplementary Fig. 6. The total numbers of ATMs were lower in Ccr5^−/− mice compared with WT mice fed either an NC (Fig. 5A and E) or HF diet (Fig. 5B and F). However, Ccr5^−/− mice fed an HF diet had markedly decreased ATM content when normalized to fat pad weight (Fig. 5F). Next, we determined the phenotypes of ATMs with antibodies to CD11c and MGL1 (CD301) (Fig. 5C and D) (10,24,25). On an NC diet, no differences were observed in either CD11c⁺ MGL1⁻ (M1-type) or CD11c⁻ MGL1⁺ (M2-type) expression within ATMs from WT and Ccr5^−/− mice (Fig. 5C and E). However, on an HF diet, in addition to reduction of total ATM content, Ccr5^−/− mice had 39% fewer M1-type ATMs and 33% more M2-type ATMs than WT mice, resulting in a predominance of M2 over M1 ATM population (Fig. 5D and F). This suggests that loss of CCR5 causes a shift to an M2-dominant ATM phenotype, which contributes to attenuation of obesity-induced insulin resistance.

To further study the effects of CCR5 deficiency in monocyte-derived macrophages, we transplanted lethally irradiated WT mice with BM from age-matched WT or Ccr5^−/− mice to generate myeloid cell–specific chimeric mice. To detect Ccr5^−/− cells in the blood of chimeric mice lacking CCR5 only in myeloid cells (Ccr5^−/− to WT), we performed PCR on genomic DNA from whole-blood samples with primers unique to the deletion mutant neo gene in the Ccr5 targeting vector (Fig. 6A) (21). The neo amplification product was not detected in samples from BM transplant (BMT) of WT into WT (WT to WT) mice but was present in each of the samples from Ccr5^−/− to WT mice (Fig. 6A). The loss of CCR5 did not alter hematopoietic cell lineage distribution (Fig. 6B).

To determine the effect of CCR5 deficiency in monocyte-derived macrophages on glucose metabolism, WT to WT and Ccr5^−/− to WT mice were fed an HF diet for 12 weeks. Both groups of transplanted mice gained weight and developed a similar level of moderate DIO (Fig. 6C). Ccr5^−/− to WT NC-fed mice showed slightly better glucose tolerance (Fig. 6D). On an HF diet, however, Ccr5^−/− to WT mice exhibited further enhanced glucose tolerance (Fig. 6D) and were protected from hyperinsulinemia (Ccr5^−/− to WT, 2.6 ± 0.3 vs. WT to WT, 5.1 ± 1.1 ng/mL; P < 0.05; fed state). It is important that Ccr5^−/− to WT mice fed an HF diet had markedly decreased macrophage infiltration and crown-like structure formation in their adipose tissue (Fig. 6E). Likewise, mRNA expression for M1 macrophage markers and proinflammatory cytokines were reduced in WAT of HF diet–fed Ccr5^−/− to WT mice (Fig. 6F). BM transplanted from Ccr5^−/− to WT mice conferred resistance to HF diet–induced hepatic steatosis (Fig. 6G).

The role of chemokine systems in obesity-associated ATM recruitment and insulin resistance has become increasingly evident during the past several years, with most studies focusing on the CC chemokine MCP-1 and its receptor CCR2. Previous work by several groups demonstrates that inhibition of MCP-1–CCR2 signaling by either genetic or pharmacological approaches can ameliorate obesity-induced insulin resistance in mice (15–17,26), whereas mice that overexpress MCP-1 have increased ATM accumulation and impaired glucose metabolism (15,17). More recent studies, however, show conflicting results and indicate greater complexity than suggested by those earlier reports. Thus, it has been demonstrated that the absence of MCP-1 neither attenuates obesity-associated macrophage recruitment to WAT nor improves metabolic function, suggesting that MCP-1 is not critical for obesity-induced ATM recruitment and systemic insulin resistance (27,28). Furthermore, although Ccr2^−/− mice fed an HF diet have fewer macrophages in WAT compared with WT mice (16), CCR2 deficiency does not normalize ATM content and insulin resistance to the levels observed in lean animals, indicating that ATM recruitment and subsequent insulin resistance are regulated by MCP-1–CCR2 independent signals as well.

Most chemokines bind to several G protein–coupled receptors, and chemokine receptors have overlapping ligand specificities (9). The complexity and redundancy of chemokine signaling may account for those conflicting results. However, other chemokine systems have also been implicated in ATM infiltration in models of obesity (29–31). These latter studies suggest that additional, unidentified chemokine/chemokine receptor pathways that could play significant roles in ATM recruitment and insulin sensitivity remain to be fully identified. Our goal in the current study was to identify and characterize such additional chemokine systems.

Our study revealed that CCR5 plays a crucial role in the regulation of adipose tissue inflammatory response to obesity and the development of insulin resistance. First, expression of CCR5 and its ligands is markedly increased in WAT of both genetically obese (ob/ob) and DIO mice. Expression of CCR5 and its ligands is equal to the levels of CCR2 and its ligands, and FACS analysis clearly demonstrated that CCR5⁺ macrophages accumulate in WAT of obese mice. Second, and most important, the loss of CCR5 ameliorates obesity-induced insulin resistance in mice. Two distinct models, Ccr5^−/− mice fed an HF diet and mice deficient in Ccr5 BM-derived cells, show improved insulin sensitivity and protection from obesity-induced insulin resistance through, at least in part, reduction of ATM accumulation.

Our results demonstrating elevated expression of CCR5 and its ligands in WAT of obese mice are consistent with CCR5 expression in adipose tissue of human obesity. Huber et al. (19) showed similarly increased visceral fat expression of CCRs (CCR1, CCR2, CCR3, and CCR5) and their ligands in obese subjects with insulin resistance (homeostasis model assessment of insulin resistance) compared with lean subjects. Obese subjects with metabolic syndrome also have been shown to have higher mRNA levels of CCR5 and CCL5/RANTES in visceral adipose tissue (18). Consistent with previous reports (6,7), our data indicate that transcripts for CD68 and F4/80 are markedly increased, and infiltration of ATMs is confirmed histologically, after 15 weeks on an HF diet. Of import, CCR5 and its ligands are significantly expressed in WAT of DIO mice at as early as 10 weeks of HF-diet feeding (aged 15 weeks), suggesting that upregulation of CCR5 and its ligands precedes ATM recruitment in DIO mice. Moreover, the time course of upregulation of CCR5 and its ligands is similar to that for MCP-1–CCR2 during the course of DIO, supporting a causative role of CCR5 in the observed ATM infiltration.

It has been suggested that elevated CCR5 and its ligands in WAT of obese mice may derive from several types of SVCs, including preadipocytes, infiltrated macrophages, and/or other hematopoietic cells. Our results, however, indicate that CCR5 is expressed mainly in the macrophage fraction (Fig. 2). Therefore, we next determined the significance of CCR5 in macrophages in propagation of the inflammatory response to obesity by generating chimeric mice lacking CCR5 only in myeloid cells derived from BMT. The lack of CCR5 expression in myeloid cells alone was sufficient to protect mice from HF diet–induced insulin resistance and fatty liver in association with a marked reduction of ATM infiltration, thus recapitulating the enhanced glucose metabolism, reduced hepatic steatosis, and anti-inflammatory state present in Ccr5^−/− mice. Therefore, CCR5⁺ ATMs are important in the development and maintenance of obesity-induced adipose tissue inflammation, fatty liver, and insulin resistance. Reduced de novo lipogenesis indicated by downregulation of sterol regulatory element–binding protein 1c and its target genes in liver and decreased plasma FFA levels lead to the amelioration of hepatic steatosis in Ccr5^−/− mice. It is interesting that hepatic CCR5 expression was also increased in response to an HF diet, particularly in the macrophage fraction (Supplementary Fig. 7), whereas CD11c mRNA expression was markedly decreased in liver of Ccr5^−/− mice fed an HF diet (1.0 ± 0.25 vs. 0.47 ± 0.2, P < 0.05 vs. WT; n = 5). Thus, recruited CCR5⁺ myeloid cells or Kupffer cells in liver may play a role in obesity-induced hepatic steatosis similar to CCR2⁺ myeloid cells recruited to the liver (32).

It is important that decreased ATM recruitment does not appear to be secondary to changes in adiposity because adipocyte size was similar between HF-fed Ccr5^−/− mice and age-matched controls. Moreover, expression of adipocyte-derived factors, such as leptin and adiponectin, in WAT and their plasma levels were similar between genotypes. Of note, HF diet–induced increases in fat mass and adipocyte size were minimally affected by Ccr2 deficiency, and obese Ccr2^−/− mice matched for adiposity with controls had reduced ATM recruitment and improved systemic insulin sensitivity (16).

Our current understanding of how ATMs promote obesity-associated inflammation and insulin resistance is based largely on the “phenotype switch” model proposed by Lumeng et al. (10). Evidence has accumulated indicating that there is an increase in M1 ATMs and a decrease in M2 ATMs in adipose tissue of both obese mice and obese humans (10,11,24,33–35). Our results suggest that deficiency of CCR5 causes an M2-dominant phenotypic shift in ATMs, which contributes to the attenuation of obesity-induced insulin resistance. Obesity is associated with increased accumulation of not only macrophages but also T cells in adipose tissue. CCR5 is preferentially expressed on Th1 cells (36). Recent studies show that infiltration of CD8⁺ T cells into adipose tissue, which precedes macrophage recruitment, contributes to obesity-induced insulin resistance (37). Wu et al. (18) showed that RANTES/CCL5 mRNA levels are highly correlated with the T-cell marker CD3 in human visceral adipose tissue. Using FACS to quantify subsets of T cells, such as CD3⁺, CD4⁺, and CD8⁺ cells, in WAT of HF diet–fed mice, no significant changes were seen between Ccr5^−/− mice and WT mice (Supplementary Fig. 8). This result suggests that CCR5 deficiency affects the M1-type ATM recruitment more prominently. In mice, Ly6C^high monocytes accumulate in atherosclerotic plaques and show a proinflammatory response (38), whereas Ly6C⁻ monocytes participate in the resolution of inflammation (39,40). A predominance of the Ly6C⁻ over Ly6C^high monocyte population was observed in the peripheral blood of HF diet–fed Ccr5^−/− mice (data not shown). Therefore, the alteration of Ly6C^high and Ly6C⁻ monocyte subsets could contribute to the M2-dominant shift of ATM in obese Ccr5^−/− mice. Further studies are required to determine how the loss of CCR5 enhances the observed M2-dominant shift.

Plasma levels of TNF-α, a key proinflammatory cytokine secreted predominantly by monocytes and macrophages in models of obesity, inflammation, and insulin resistance (3), were markedly reduced in Ccr5^−/− mice on an HF diet. Reduced TNF-α expression in WAT was most likely directly related to the reduced ATM infiltration because M1 macrophages are the primary source of TNF-α in obese adipose tissue (6,7). Activation of the TNF receptor results in stimulation of NF-κB signaling via inhibitor of κB kinase, leading to Ser phosphorylation of IRS-1, which causes insulin resistance (41). Knockout of JNK1 (Jnk1^−/−) ameliorates insulin resistance in DIO mice, at least in part, by decreased IRS-1 phospho-Ser307 in insulin target tissues (42). We demonstrated attenuated signaling via NF-κB and JNK together with diminished Ser phosphorylation of IRS-1 and enhanced downstream insulin signaling in WAT of Ccr5^−/− mice.

In summary, we here provide evidence that CCR5 plays a critical role in adipose tissue inflammatory response to obesity by regulating both macrophage recruitment and M1/M2 status. CCR5 ablation in mice prevents insulin resistance, diabetes, and fatty liver induced by HF feeding. The beneficial effects of CCR5 deficiency are the result of both the decrease of ATM recruitment and the M2-dominant shift of ATM polarization. CCR5⁺ macrophages accumulate in WAT of DIO mice and, importantly, knockout of CCR5 signaling in macrophages attenuates insulin resistance in HF diet–fed mice. Altogether, these data indicate that CCR5 is a novel link between obesity, adipose tissue inflammation, and insulin resistance. Thus, CCR5 could be a promising therapeutic target for insulin resistance, metabolic syndrome, and type 2 diabetes.

This work was supported by the following grants to T.O.: Program for Improvement of Research Environment for Young Researchers from the Special Coordination Funds for Promoting Science and Technology; Grant-in-Aid for Young Scientists (B) (22790854) from the Ministry of Education, Culture, Sports, Science and Technology of Japan; Uehara Memorial Foundation; Kanae Foundation for the Promotion of Medical Science; and Banyu Life Science Foundation International. This work was also supported by the Program for Promotion of Basic and Applied Research for Innovations in Bio-Oriented Industry (to H.I.); and National Institutes of Health R01-HL-55638 and R01-HL-73030 (to H.N.G.).

No potential conflicts of interest relevant to this article were reported.

H.K., K.S., and M.N. researched data. H.I. and Y.Y. researched data and contributed to discussion. Y.S., T.T., H.Y., K.M., N.M., and S.K. contributed to discussion. H.N.G. contributed to discussion and reviewed and edited the manuscript. T.O. researched data and wrote the manuscript. T.O. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Parts of this article were presented in abstract form at the 70th and 71st Scientific Sessions of the American Diabetes Association, Orlando, Florida, 25–29 June 2010, and San Diego, California, 24–28 June 2011, respectively.

The authors thank Dr. Y. Okamoto (Kanazawa University) for technical advice on BMT and Y. Hashimoto (Kanazawa University) for technical assistance and care of the animals.