SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References
  9. Supporting Information

Objective

The aim of this study was to search for novel markers of visceral adiposity.

Methods

Visceral (omental) and subcutaneous adipose tissues were obtained from 43 Japanese men. Microarray analysis using total RNA from visceral and subcutaneous adipose tissues obtained from five men with abdominal obesity and five nonobese men was first conducted. Then the expression pattern of candidate genes identified in the human study in mouse models of adiposity was examined.

Results

Among 30,500 genes evaluated, the mRNA expression of CCDC3 (encoding coiled-coil domain-containing protein 3) was upregulated in omental adipose tissues from abdominally obese subjects (3.07-fold) but not in subcutaneous adipose tissues (0.89-fold). Similar expression patterns were found in two distinct mouse models of obesity. In the analysis of all 43 men, CCDC3 mRNA levels in omental, but not in subcutaneous adipose tissue, were positively correlated with waist circumference and body mass index. CCDC3 was predicted to be a secretory protein, which was confirmed by western blotting, as overexpressed CCDC3 was secreted into the culture media.

Conclusions

The expression of CCDC3 is specifically increased in visceral adipose tissues in abdominally obese subjects. These results suggest that CCDC3 is a potential biomarker for estimating visceral adiposity.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References
  9. Supporting Information

Obesity is a major health problem associated with various comorbidities, including type 2 diabetes, dyslipidemia, hypertension, and cardiovascular diseases. There is increasing evidence to show that altered body fat distribution, particularly an increase in visceral adiposity, is critical for the development of obesity-related morbidities rather than obesity itself. Several studies have demonstrated that increased intra-abdominal fat is a predisposing risk factor for the development of metabolic syndrome, a cluster of disorders comprising glucose intolerance, dyslipidemia, and hypertension, which contribute to an increased risk for cardiovascular disease [1, 2]. Adipose tissues are a source of signaling molecules, known as adipokines, that link obesity with its comorbidities [3, 4]. Several proinflammatory adipokines, including tumor necrosis factor-α [3, 4], interleukin-6 (IL-6) [5], monocyte chemoattractant protein-1 [6], plasminogen activator inhibitor-1 [7], and resistin [8], are directly linked to tissue and systemic insulin resistance, and are predominantly secreted from visceral adipose tissues. Another adipokine, adiponectin shows insulin-sensitizing and anti-atherosclerotic properties, and its plasma levels are decreased in obese individuals [1].

Therefore, it is clinically important to evaluate visceral adiposity. Currently, visceral adiposity is estimated by measuring waist circumference [2]. However, waist circumference does not always provide an accurate estimate of visceral adiposity. For example, there are clear differences among different ethnicities in the relationship between overall adiposity, abdominal obesity, and visceral fat accumulation [2]. Therefore, it would be valuable to identify more accurate markers for visceral adiposity than waist circumference; however, no such markers have yet been identified.

In this study, we performed comprehensive mRNA expression analyses using visceral (i.e., omental) and subcutaneous adipose tissues obtained from Japanese men to search for a novel marker for visceral adiposity.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References
  9. Supporting Information

The Ethics Review Committee of Shiga University of Medical Science, RIKEN Yokohama Institute, and Shionogi & Co. approved all of the protocols described in this study. All of the participants gave written informed consent.

Participants

We recruited 43 males who were scheduled for elective abdominal surgery for early-stage gastric cancer at Shiga University Hospital. Subjects with high-sensitive C-reactive protein levels >5 mg/dl or a >10% reduction in body mass index (BMI) within 6 months before admission were excluded. Abdominal obesity was defined according to the criteria proposed by the Japan Society for the Study of Obesity [9]. The characteristics of the participants are described in our previous study [10].

RNA Preparation from Adipose Tissue Biopsies

Omental and abdominal subcutaneous adipose tissue biopsies were obtained from each subject during elective abdominal surgery, as previously described [10]. Each sample (2-5 g) was immediately frozen in liquid nitrogen. Total RNA was extracted using QIAzol Lysis buffer (Qiagen, Valencia, CA) and RNeasy Lipid tissue Mini Kit (Qiagen).

Microarray Analyses of Subjects with or without Abdominal Obesity

We chose nine men with abdominal obesity (waist circumference ≥85 cm; mean ± SD, 90.1 ± 3.0 cm) and nine control nonobese men (waist circumference <85 cm; mean ± SD, 78.5 ± 3.4 cm) out of 43 subjects who underwent biopsy [10]. To exclude the effects of glucose intolerance on gene expression, we selected 5 obese and 5 nonobese men who had normal plasma glucose levels (HbA1c <6.0%) out of the 18 men and compared their gene expression profiles. Table 1 summarizes the characteristics of the ten men whose adipose tissue samples were used in microarray analysis.

Table 1. Characteristics of subjects included in microarray analysis
VariablesNon-obeseAbdominal obesityP–value
  1. We selected five obese and five nonobese men who had normal plasma glucose levels (HbA1c < 6.0%) for microarray analysis (5:5 analysis).

  2. Values are mean ± SD.

Age (years)69.4 ± 7.166.0 ± 7.70.49
BMI (kg/m2)21.5 ± 0.725.3 ± 0.9>0.001
Waist circumference (cm)78.5 ± 5.390.7 ± 3.2>0.001
HbA1c (%)5.34 ± 0.255.52 ± 0.250.30
Leptin (ng/ml)2.88 ± 1.254.80 ± 2.010.11
Adiponectin (g/ml)7.12 ± 1.393.46 ± 1.84>0.01
hsCRP (mg/l)0.82 ± 0.670.39 ± 0.220.23
IL-6 (pg/ml)2.16 ± 0.881.46 ± 0.470.17

Total RNA (500 ng) from each tissue sample was reverse transcribed, and Cy3-labeled cRNA was synthesized and hybridized onto a whole human genome Oligo Microarray (Agilent Technologies, Santa Clara, CA) for human samples and a whole mouse genome Oligo Microarray (Agilent Technologies) for murine samples. Microarrays were processed according to the manufacturer's instructions.

Data analysis was performed using GeneSpring version 7.0 (Agilent Technologies) for global analysis of each chip. Each chip was scaled globally to the target intensity value of the mean value of all probes to enable interarray comparisons. We used Welch's t-test to determine differences in signal levels between obese and nonobese subjects. The genes that were significantly upregulated (≥2.0-fold) or downregulated (≤0.67-fold) at P < 0.05 and whose signal intensity was >100 were selected and used for hierarchical clustering analysis or gene ontology analysis.

Quantitative Real-Time PCR (qPCR)

Total RNA samples were prepared and reverse-transcription (RT)-PCR was performed as previously described [10]. The thermal profile consisted of 2 min at 50°C, 10 min at 95°C, followed by 40 cycles of 15 s at 95°C, 60 s at 60°C, and 30 s at 72°C. The following primers were used: human CCDC3 (encoding coiled-coil domain-containing protein 3) 5′-TGACTGGGAAATCCAGGAAGA-3′ and 5′-CGTGGTCCTCCTCCTCAAAC-3′; mouse CCDC3, 5′-TGGTTCATCGCCTCTAATGGTG-3′, and 5′-TGGAATGGGTTCTAGGTGGTCAA-3′. The primers for adiponectin, leptin, and IL-6 are described in our previous study [10]. β-actin mRNA was also amplified to normalize gene expression levels. The relative quantitative data were expressed as the ratio of CCDC3 mRNA to β-actin mRNA in arbitrary units. All PCRs were performed in duplicate and repeated twice.

Animals

All animal experiments were approved by the local ethics committee and were conducted in accordance with the NIH Principles of Laboratory Animal Care and the Japanese law. All animals were purchased from Clea (Tokyo, Japan). Male c57BL/6J mice were fed a normal chow (NC) (13.5% fat) (n = 5) or high-fat diet (HFD) (60% fat) (n = 5) for 13 weeks from 7 weeks of age. Male db/db (n = 5) and male db/m+ mice (n = 5) were fed NC until 15 weeks of age. We included two independent groups of animals, with five mice in each group. Mice were anesthetized, and the subcutaneous and omental adipose tissues were excised, washed with phosphate-buffered saline, and frozen in liquid nitrogen. We extracted total RNA using the pooled adipose tissue samples from five mice in each group and performed mRNA expression analyses using both independent sets of adipose tissue samples.

Cell Culture

3T3-L1 adipocytes were cultured and differentiated as previously described [11].

Construction of FLAG-Tagged Human CCDC3 Expression Vector and Transfection

cDNA for human CCDC3 with a FLAG-tag at the C-terminal end was inserted into pReceiver vector (GeneCopoeia, Rockville, MD). The resultant plasmid was transfected into HEK293T cells, and the cells were cultured for 48 h. The culture media and cell lysates were subjected to western blotting with an anti-FLAG M2-peroxidase antibody (Sigma, St. Louis, MO).

Statistical Analysis

Data are expressed as means ± standard deviation. Two-tailed unpaired Student's t-test or Wilcoxon's rank test was used to compare mean values. Pearson's correlation coefficient was used to determine the relationship between gene expression levels and other quantitative variables. Values of P < 0.05 were considered statistically significant. Analyses were conducted using JMP IN 5.1 software (SAS Institute, Cary, NC).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References
  9. Supporting Information

Microarray Profiles of Omental and Subcutaneous Adipose Tissues from 10 Japanese Men

Out of 41,000 probe pairs for 30,500 genes, we detected significant signals for around 50% of transcripts in both omental and subcutaneous adipose tissues (signal intensity >100, data not shown).

The expression levels of 55 genes were upregulated in omental or subcutaneous adipose tissues in obese subjects compared with nonobese subjects. Of these 55 genes, 38 were increased in omental adipose tissues, and 32 were specifically increased in omental adipose tissues (Figure 1 and Supplementary Table S1).

image

Figure 1. Strategy used to identify the target genes as possible markers of visceral obesity. We first searched for genes whose expression levels were specifically increased in omental adipose tissues of obese men. Microarray analysis revealed that the expression of 23 genes was upregulated in analyses of five nonobese and five obese men (5:5 analysis) and in five nonobese and two obese men (5:2 analysis). Of the 23 genes identified, 8 were predicated to encode secreted proteins. Only one of these genes, CCDC3 was specifically increased in omental adipose tissue of HFD-fed mice.

Download figure to PowerPoint

We then selected two of the five subjects with abdominal obesity who met the criteria of International Diabetes Federation for abdominal obesity (waist circumference >90 cm), the criteria for overweight in Asian populations (BMI > 24 kg/m2) [12], and the criteria of American Diabetes Association for normal glucose tolerance (HbA1c < 5.7%) (Supplementary Table S2), and evaluated the expression of the 32 genes in these two subjects (Figure 1). Among the 32 genes, 23 genes were specifically upregulated in the omental adipose tissues in the two subjects (Figure 1 and Supplementary Table S3). To evaluate whether any of these 23 genes were candidate markers for visceral obesity, we performed gene ontology analyses to identify those that potentially encoded secretory products. In this analysis, eight genes (CCDC3, CES1, CLSTN2, COCH, EPDR1, LAMC3, LEP, and NPR3) were predicted to encode secreted proteins. Therefore, we focused on these eight genes in subsequent analyses (Table 2).

Table 2. Expression profiles of eight candidate genes in two abdominally obese subjects and in a mouse model of obesity
 HumanHFD-fed mice
 Fold increaseSignal intensityFold increaseSignal intensity
 (obese/nonobese)(obese/nonobese)
Gene symbolOmentalSubcutaneousOmentalOmentalSubcutaneousOmental
  1. Expression levels are shown as the fold increase relative to those in lean individuals or control mice-fed normal chow.

CES13.761.1714,3107.079.4235
LAMC33.501.762801.000.6835
CCDC33.070.89114,9152.151.281,530
LEP2.730.8819,29711.375.8141,613
EPDR12.521.489,0801.541.372,234
COCH2.481.503650.090.764
NPR32.361.132894.952.6837,637
CLSTN22.301.315051.050.83709

Expression Levels of the Eight Candidate Genes as Markers of Visceral Adiposity in Mouse Models of Adiposity

In HFD-fed mice, the mRNA expression levels of CCDC3, CES1, LEP, and NPR3 were increased in omental adipose tissues and those of CES1, LEP, and NPR3 were also increased in subcutaneous adipose tissues as compared with their expression levels in control mice (Table 2).

Only CCDC3 was specifically increased in omental adipose tissues (2.15-fold in omental adipose tissues and 1.28-fold in subcutaneous adipose tissues), similar to the expression pattern in humans (3.07-fold in omental adipose tissues and 0.89-fold in subcutaneous adipose tissues) (Table 2 and Figure 2a and b). Ccdc3 expression was also specifically increased in omental adipose tissues of db/db mice compared with db/m+ mice (1.9-fold in omental adipose tissues and 0.7-fold in subcutaneous adipose tissues) (Figure 2c). The absolute expression level of CCDC3 tended to be greater in omental adipose tissue than in subcutaneous adipose tissue in humans (Figure 2d), whereas the absolute expression of Ccdc3 tended to be greater in subcutaneous adipose tissue than in omental adipose tissue in mice (Figure 2e and f). The omental adipose tissue-specific increase in CCDC3 expression in subjects with abdominal obesity was also observed in the first ten men analyzed (five men with abdominal obesity and five nonobese men; 5:5 analysis, Figure 1 and Table 1). The increase in CCDC3 expression was statistically significant in omental adipose tissue (P < 0.05) but not in subcutaneous adipose tissue (P > 0.05) (Figure 3a and b).

image

Figure 2. CCDC3 mRNA expression levels in omental and subcutaneous adipose tissues obtained from abdominally obese (n = 2) and lean men (n = 5) (a and d), normal chow (NC) and high-fat diet (HFD)-fed mice (b and e), and db/db and db/m+ mice (c and f). Expression levels are shown as the fold increase relative to those in lean individuals or control mice (a-c) or as the signal intensity (d-f).

Download figure to PowerPoint

image

Figure 3. CCDC3 mRNA expression levels in omental and subcutaneous adipose tissues in ten individual men (5 abdominally obese men and 5 lean men; 5:5 analysis). Expression levels are shown as the signal intensity (a) or the fold-increase relative to those in lean individuals (b). Data are means ± SE; P-values are for obese vs. lean subjects.

Download figure to PowerPoint

We also determined Ccdc3 mRNA expression in epididymal adipose tissue, which was commonly used as a model of “visceral fat” in mouse studies. We found that Ccdc3 mRNA expression in epididymal adipose tissue was increased in HFD mice but was unchanged in db/db mice (1.65-fold in HFD mice and 0.88-fold in db/db mice) (Supplementary Figure S4). Considering that a few reports have demonstrated differences between epididymal fat and mesenteric fat in animals [13], omental adipose tissue might be more suitable than epididymal adipose tissue in mouse studies aimed at inferring the pathological changes in “human visceral adipose tissues.”

With regard to the mRNA expression levels of adipokines, we found that the mRNA expression of adiponectin was unchanged in omental adipose tissue, but was decreased in subcutaneous adipose tissue, in both the mouse models of obesity (Supplementary Figures S1a and b). By contrast, the mRNA expression of leptin was increased in omental and subcutaneous adipose tissues in both the mouse models (Supplementary Figure S1c and d).

The Protein Product of CCDC3 Is a Secreted Protein

Searches of the Swiss-Prot database suggested that CCDC3 encodes a secreted protein because it contains a signal peptide at the N-terminal region, and is cut between amino acid residues 21 and 22. The predicted molecular weight was 38 kDa. To confirm this, we constructed a FLAG-tagged human CCDC3 expression vector and transiently transfected the constructs into HEK293T cells. The culture media and cell lysates were subjected to western blotting with an anti-FLAG antibody. As shown in Figure 4, a signal at ~38 kDa was detected when we used culture media or cell lysates from cells expressing CCDC3, but not from cells transfected with the empty vector. These results imply that CCDC3 is a secreted protein.

image

Figure 4. Western blotting of CCDC3 protein. The FLAG-tagged expression vector encoding human CCDC3 (CCDC3+) or empty vector (CCDC3–) were transfected into HEK293T cells. The culture media and cell lysates were subjected to western blotting using an anti-FLAG antibody.

Download figure to PowerPoint

Expression Profiles of the CCDC3

We examined the expression profiles of CCDC3 by microarray analysis using total RNA from various human tissues (Human Total RAN Master Panel II, Clontech). This analysis revealed that CCDC3 expression was highest in the aorta, followed by the adipose tissues (Figure 5a). Furthermore, we found that the expression of Ccdc3 increased in mouse 3T3-L1 cells with ongoing differentiation (Figure 5b).

image

Figure 5. (a) Expression profiles of CCDC3 in various human tissues. (b) Ccdc3 mRNA expression during adipocyte differentiation. Cultured 3T3-L1 fibroblasts were induced to differentiate, and cells were harvested at the indicated times. Data are expressed as the fold increase relative to that on day 0 (n = 3; data are means ± SD; *P < 0.001 vs. day 0).

Download figure to PowerPoint

Correlations of CCDC3 mRNA Expression Levels in Omental or Subcutaneous Adipose Tissues with Waist Circumference or BMI

We examined the correlations between the expression of CCDC3 mRNA and obesity-related characteristics using all 43 men. As shown in Figure 6a-d, the CCDC3 mRNA level in omental adipose tissue was significantly correlated with waist circumference (r = 0.483, P < 0.001) and BMI (r = 0.364, P < 0.05) (Figure 6a and b). By contrast, the CCDC3 mRNA level in subcutaneous adipose tissue was not correlated with either waist circumference or BMI (Figure 6c and d).

image

Figure 6. Correlations of CCDC3 mRNA expression levels in omental or subcutaneous adipose tissues with waist circumference (a and c) or BMI (b and d).

Download figure to PowerPoint

CCDC3 mRNA expression levels in omental and subcutaneous adipose tissues were not significantly correlated (r = 0.169, P = 0.256) (Supplementary Figure S2).

Correlations between CCDC3 mRNA Expression and mRNA Expression of Adipokines in Omental or Subcutaneous Adipose Tissues

The mRNA expression of CCDC3 was significantly correlated with that of leptin in subcutaneous (r = 0.703, P < 0.001) and omental (r = 0.860, P < 0.001) adipose tissues (Supplementary Figure S3). The mRNA expression of CCDC3 was also modestly correlated with that of adiponectin in subcutaneous (r = 0.423, P < 0.001) and omental (r = 0.251, P < 0.05) adipose tissues. By contrast, the mRNA expression of CCDC3 was negatively correlated with that of IL-6 in subcutaneous (r = –0.440, P < 0.001) and omental (r = –0.662, P < 0.001) adipose tissues.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References
  9. Supporting Information

In this study, we examined the gene expression profiles of omental or subcutaneous adipose tissues obtained from Japanese individuals, and found that the expression of CCDC3 was upregulated in omental adipose tissues, but not in subcutaneous adipose tissues, in subjects with abdominal obesity compared with those in nonobese subjects.

Although the degree of obesity was reported to be associated with obesity-related complications, several reports have demonstrated that altered fat distribution, including the accumulation of visceral fat, is more closely related to the development of systemic metabolic disorders than obesity itself [14-16]. The accumulation of visceral fat can be directly determined using several radiographic methods, but these procedures cannot be applied for all individuals. Therefore, it is clinically important to identify simple markers that accurately reflect the accumulation of visceral fat. Although waist circumference is widely used as an indicator of abdominal obesity, it does not accurately estimate the amount of visceral fat. Therefore, useful and accurate markers for visceral adiposity need to be identified.

In this study, we found that the expression levels of eight genes, which were predicted to encode secreted proteins, were increased in omental adipose tissues, but not in subcutaneous adipose tissues, in abdominally obese people (Table 2). Two of these genes, LEP and CES1, were previously reported to be associated with visceral adiposity in humans, suggesting that this study was appropriately designed to identify specific markers for visceral obesity in humans. LEP encodes leptin, an adipokine whose expression is increased in obesity [17]. CES1 encodes carboxylase-1, and its expression was reported to be increased in omental and subcutaneous adipose tissues of obese individuals [18]. In this study, subsequent analyses using animal models of obesity revealed that the expression levels of three genes (CES1, LEP, and NPR3) were increased in omental and subcutaneous adipose tissues in obese mice, whereas the expression levels of four genes (CLSTN2, COCH, EPDR1, and LAMC3) were not increased in omental or subcutaneous adipose tissues in obese mice. From these observations, the effects of obesity on the expression patterns of these genes in subcutaneous and visceral adipose tissues vary among individual genes, ethnicities, or species, and that the increased expression levels of seven genes (CES1, LEP, NPR3, CLSTN2, COCH, EPDR1, and LAMC3) in adipose tissues are not necessarily specific to visceral adipose tissues. In contrast, the expression of CCDC3 was consistently increased in omental adipose tissues, but not in subcutaneous adipose tissues in abdominally obese people and in two animal models of obesity. In addition, it was reported that CCDC3 mRNA expression in epididymal white adipose tissues was increased in obese mice compared with that in control mice. Therefore, it is notable that CCDC3 expression was specifically upregulated in visceral adipose tissues in abdominal obesity. Moreover, CCDC3 expression levels in subcutaneous and omental adipose tissues were not significantly correlated (Supplementary Figure S2), and in an expanded analysis using 43 Japanese men, CCDC3 mRNA expression in omental adipose tissue, but not in subcutaneous adipose tissue, was positively correlated with waist circumference and BMI (Figure 5). These results suggest that different mechanisms regulate CCDC3 expression in omental and subcutaneous adipose tissues.

As we also showed that CCDC3 was a secreted protein, it can be expected that its plasma levels would be increased in accordance with the increase of visceral fat in abdominally obese subjects, and that CCDC3 is a good marker for visceral adiposity if its plasma concentrations can be determined.

CCDC3 was first cloned by National Institute of Health Mammalian Gene Collection [19]. Human CCDC3 encodes a 270-amino acid protein with a calculated molecular mass of 32 kDa. CCDC3 protein contains a putative coiled-coil domain in its C-terminal region, a well-characterized structural feature found in many proteins, particularly motor proteins and transcription factors [20, 21]. Proteins containing a coiled-coil domain usually interact with other coiled-coil proteins and participate in many protein–protein interactions [21].

Eberlein et al. [22] reported a significant correlation between intramuscular fat content and Ccdc3 mRNA expression in bovine skeletal muscle, suggesting that bovine Ccdc3 protein may be involved in the metabolism of fat deposition. In this study, we also demonstrated that CCDC3 proteins were secreted from cells overexpressing CCDC3, and that Ccdc3 expression increased during differentiation of murine 3T3-L1 cells into mature adipocytes. Therefore, it is likely that CCDC3 secreted from mature adipocytes acts as an adipokine. Consistent with our present results, Kobayashi et al. [23] reported that murine Ccdc3 was a secretory factor expressed in adipocytes and endothelial cells, and that murine Ccdc3 mRNA expression was regulated in adipose tissues by hormonal and nutritional factors. Combining their results with our present results, we hypothesize that CCDC3 secreted from visceral adipose tissues contributes to the development of obesity-related metabolic disorders in abdominally obese subjects. However, the functional roles of CCDC3 are still unknown. Therefore, further studies are needed to elucidate the functional significance of CCDC3 in the pathogenesis of metabolic disorders in humans.

In conclusion, in this comprehensive analysis of the mRNA expression profiles of omental and subcutaneous adipose tissues, we found that CCDC3 is preferentially expressed in adipose tissues in humans. Furthermore, CCDC3 expression is specifically increased in omental adipose tissues, but not in subcutaneous adipose tissues, in subjects with abdominal obesity. A visceral adipose tissue-specific increase in CCDC3 expression was observed in abdominally obese subjects and in two animal models of obesity. We also confirmed that CCDC3 is a secreted protein. These results suggest that CCDC3 could be a novel marker for visceral adiposity. Further studies are needed to elucidate the functional significance of CCDC3 in the pathogenesis of metabolic disorders in humans, and to evaluate the clinical usefulness of measuring plasma CCDC3 levels in abdominally obese subjects.

Acknowledgement

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References
  9. Supporting Information

We thank Ms. Keiko Kondo for her assistance in performing statistical analyses.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article

FilenameFormatSizeDescription
oby20645-sup-0001-suppinfofig01.tif89KSupporting Information
oby20645-sup-0002-suppinfofig02.tif58KSupporting Information
oby20645-sup-0003-suppinfofig03.tif113KSupporting Information
oby20645-sup-0004-suppinfofig04.tif70KSupporting Information
oby20645-sup-0005-suppinfotab01.xls31KSupporting Information
oby20645-sup-0006-suppinfotab02.doc30KSupporting Information
oby20645-sup-0007-suppinfotab03.xls30KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.