Figures
Abstract
Circadian clock genes are critical regulators of energy homeostasis and metabolism. However, whether variation in the circadian genes is associated with metabolic phenotypes in humans remains to be explored. In this study, we systemically genotyped 20 tag single nucleotide polymorphisms (SNPs) in 8 candidate genes involved in circadian clock, including CLOCK, BMAL1(ARNTL), PER1, PER2, CRY1, CRY2, CSNK1E,, and NOC(CCRN4L) in 1,510 non-diabetic Chinese subjects in Taipei and Yunlin populations in Taiwan. Their associations with metabolic phenotypes were analyzed. We found that genetic variation in the NOC gene, rs9684900 was associated with body mass index (BMI) (P = 0.0016, Bonferroni corrected P = 0.032). Another variant, rs135764 in the CSNK1E gene was associated with fasting glucose (P = 0.0023, Bonferroni corrected P = 0.046). These associations were consistent in both Taipei and Yunlin populations. Significant epistatic and joint effects between SNPs on BMI and related phenotypes were observed. Furthermore, NOC mRNA levels in human abdominal adipose tissue were significantly increased in obese subjects compared to non-obese controls.
Citation: Chang Y-C, Chiu Y-F, Liu P-H, Hee SW, Chang T-J, Jiang Y-D, et al. (2013) Genetic Variation in the NOC Gene Is Associated with Body Mass Index in Chinese Subjects. PLoS ONE 8(7): e69622. https://doi.org/10.1371/journal.pone.0069622
Editor: Nicolas Cermakian, McGill University, Canada
Received: March 17, 2013; Accepted: June 11, 2013; Published: July 26, 2013
Copyright: © 2013 Chang et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported in part by a grant (NSC 97–2314-B-002-047-MY3) from the National Science Council of Taiwan and a grant (99C101–601) from the National Taiwan University College of Medicine, Taipei, Taiwan. This study is also partially supported by the grant PH-102-PP-04 from the National Health Research Institutes in Taiwan. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
The rotation of the earth around its axis generates the dark and light cycles, and organism living on earth developed circadian rhythms to adapt their activities to the dark and light cycles. In mammals, the central clock is located in the suprachiasmatic nucleus (SCN). The SCN clock is composed of single-cell neuronal oscillators that generate a coordinated output, which controls the systemic circadian rhythm. SCN neurons respond to external stimuli such as light or nutrients by adjusting their oscillatory activity [1]–[3]. In addition to external stimuli, the central circadian rhythm is also controlled by a intrinsic transcriptional auto-regulatory feedback loop. This feedback loop is composed of the transcriptional activators CLOCK and BMAL1 and their target genes PERs and CRYs. The CLOCK gene encodes a transcription factor, CLOCK, which dimerizes with BMAL1 to activate downstream genes, including PERs and CRYs [1]–[3]. PERs and CRYs oligomerize and translocate to the nucleus to inhibit CLOCK- BMAL1-mediated transactivation, thereby forming a negatively feedback loop. This autoregulatory loop is post-translationally regulated by the casein kinases, which target the PER proteins for degradation [1]–[3]. Another important circadian gene is the NOC gene, which encodes a deadenylase, nocturnin, that removes the 3 ′ adenosine residue from the mRNA transcripts of target genes at a post-transcriptional level to regulate gene function [4].
Several clinical studies have shown association between disturbed circadian rhythm and adverse metabolic consequence [5], [6]. Furthermore, mouse models with a targeted ablation of circadian genes also displayed abnormal energy metabolism. For example, Clock mutant mice developed hyperphagia, obesity, hyperlipidemia, and hyperglycemia [7]. Per2 knockout mice showed an alteration in lipid metabolism, with a decrease in levels of triglyceride and free fatty acid [8]. Nocturnin knockout mice exhibited a resistance to diet-induced obesity, decreased triglycerides levels and improved insulin sensitivity [9]. These data indicate a critical role of circadian genes in energy homeostasis, glucose, and lipid metabolism.
In this present study, we systemically analyzed the relationship between variations in circadian genes and metabolic phenotypes in general populations in Taiwan. We found that genetic variation in the NOC gene was associated with body mass index (BMI). NOC expression in adipose tissue was increased in obese subjects. These data suggested a potential role for NOC in obesity.
Results
The clinical and biochemical characteristics of study participants are summarized in Table 1. The nucleotide composition, chromosomal position, and minor allele frequencies for all genotyped SNP are summarized in Table 2. On average, the genotype call rate was 97.6%. All SNP were in Hardy-Weinberg equilibrium. The LD patterns between SNPs are shown in Figure 1.
Pairwise LD coefficients D'×100 are shown in each cell (D' values of 1.0 were not shown). The standard color scheme of Haploview was applied for LD color display (LOD score ≧2 and D' = 1 in bright red; LOD score ≧2 and D'<1 in blue; LOD score <2 and D' = 1 in shade of pink; LOD score <2 and D'<1 in white).
SNP association with metabolic phenotypes
Single-locus SNP associations with metabolic phenotypes are summarized in Table 3. Among the 20 genotyped SNPs, the A allele at rs9684900 in the NOC gene was associated with higher BMI (P = 0.0016, Bonferroni adjusted P = 0.032) (Table 4). The estimated effect size associated with each A allele was 0.46 kg/m2. This direction associations are consistent in both Taipei and Yunlin populations (P = 0.0045 and P = 0.091, respectively)(Table S1). We also observed nominal associations of rs17050679 in the NOC gene with BMI (P = 0.0057, Bonferroni corrected P = 0.11) and fasting plasma triglycerides (P = 0.0037, Bonferroni corrected P = 0.074). The directions of association with BMI (P = 0.021 in Taipei and P = 0.074 in Yunlin population) and fasting plasma triglycerides (P = 0.026 in Taipei and P = 0.046 in Yunlin population) are also consistent in both study populations (Table S1).
Another variant, the A allele at rs135764 in the CSNK1E gene was significantly associated with fasting plasma glucose (P = 0.0023, Bonferroni corrected P = 0.046) (Table 5) with concordant directions of association in Taipei and Yunlin populations (P = 0.058 and P = 0.021, respectively) (Table S1).
The effect of multi-locus interaction between SNPs on metabolic phenotypes
We next evaluated the effect of interaction between SNPs on metabolic phenotypes. Results of the multilocus interaction analyses using GMDR are summarized in Table 6. The most significant interactions with a good cross-validation consistency (CVC) were found between NOC rs17050679 and NOC rs9684900 (P<0.001, CVC: 8/10) for BMI and between NOC rs9684900 and BMAL1 rs2290035 (P = 0.001, CVC: 7/10) for fasting glucose level.
Joint effects of risk alleles on metabolic phenotypes
We further evaluated the joint effects of risk alleles with nominal association (P<0.1) with metabolic phenotypes. The distribution of BMI, blood pressures, fasting triglycerides, and fasting glucose according to different number of risk alleles is shown in Figure 2. Increasing risk alleles were significantly associated with higher BMI (P<0.0001), systolic blood pressure (SBP) (P = 0.0041), DBP (P = 0.031), fasting triglycerides (P = 0.0033) and fasting glucose levels (P = 0.00027).
NOC expression in adipose tissue between obese and non-obese subjects
Nocturnin was previously reported to be a regulator of adipogenesis highly expressed in adipocytes [10]. We next explored whether NOC expression in adipose tissue was associated with obesity. We found that NOC expression was higher in obese subjects in either subcutaneous or visceral adipose tissue (P<0.001) (Figure 3).
Discussion
In this present study, we systemically analyzed the association of tag SNPs in 8 circadian genes with metabolic phenotypes in 1,510 otherwise healthy Taiwanese individuals. We found that variation in the NOC gene was associated with BMI. Interestingly, NOC expression in abdominal adipose tissue was positively associated with obesity. Consistent with our findings, nocturnin has been shown to regulate adipogenesis by stimulating nuclear translocation of PPARγ [11] or by modulating mitotic clonal expansion during early adipogenesis [10]. Importantly, Nocturnin knockout mice were protected from diet-induced obesity [9]. These findings indicate nocturnin is an important regulator of adiposity. The molecular mechanism by which an intronic variant in the NOC gene influences phenotype is currently not known. It is most likely that the functional causative variant is in LD with this associated variant.
We also found a significant association between CSNK1E genetic variant and fasting glucose. The CSNK1E encodes for the casein kinase 1 epsilon, a serine/threonine kinase that phosphorylates PER proteins in the cytoplasm and triggers their proteasome-dependent degradation. In yeasts, glucose-dependent activation of casein kinase I catalyzes phosphorylation of Mth1, a glucose transporter gene (HXT) repressor, triggering degradation of Mth1 by the proteasome and leads to de-repression of HXT gene expression [12], [13]. These data suggest a potential role of casein kinase in glucose homeostasis.
Interestingly, we also found that an interaction between NOC and BMAL1 genetic variants may affect fasting glucose levels. Loss of Bmal1 in mice has been shown to cause hyperglycemia due to impairment in β-cells function while deletion of Noc in mice causes higher fasting glucose. Although no main effect of BMAL1 and NOC genetic variants on fasting glucose was observed, fasting glucose were significantly elevated in the presence of both genetic variants. These data indicates that a clinically evident phenotype may require a close interaction between 2 or more genetic factors.
However, our results were not consistent with a previous genome-wide association analysis reporting an association between CRY2 variant and fasting glucose [14]. In our study, all tag SNPs in the CRY2 gene were not associated with fasting plasma glucose. Our results are also not consistent with previous candidate-gene association studies reporting significant associations between CLOCK genetic variation and obesity [15]–[17]/metabolic syndrome [18], PER2 genetic variation and obesity [19]/fasting plasma glucose [20], and BMAL1 genetic variation and type 2 diabetes/hypertension [21]. Nevertheless, since we did not genotype the same SNPs (or tagSNPs) with previous studies, a direct comparison between studies is not possible. Heterogeneity between different ethnic groups may also confound the association.
The major limitation of this study is the lack of replication. However, the directions of association between NOC variant and BMI and between CSNK1E variant and fasting glucose are the same in both Taipei and Yunlin populations, indicating consistency in different populations. Further study is still warranted to confirm the association.
In conclusion, we systemically analyzed the association of tag SNPs in circadian genes with metabolic phenotypes. We identified associations of NOC genetic variation with BMI. A strong correlation of NOC expression in human adipose tissue with obesity was also observed. These data, together with previous animal models, suggest a substantial role of nocturnin in obesity.
Materials and Methods
Subjects
The studied population comprised 760 subjects recruited from the Health Management Center of National Taiwan University Hospital (NTUH) in Taipei and 750 subjects recruited from a community-based screening program in the Yunlin County in southern Taiwan. The detailed description of study subjects have been described previously [22], [23]. Patients with a history of diabetes or other major systemic diseases were excluded [22], [23]. Blood pressure was measured by mercury sphygmomanometer to the nearest 2 mmHg. Trained nurses took three separate readings at 1-minute intervals. Fasting glucose and triglycerides levels were measured by an automatic analyzer (Toshiba TBA 120FR, Toshiba Medical System Co., Japan). The study was approved by the National Taiwan University Hospital Research Ethics Committee. Written informed consent was obtained from each participant.
Subjects for measurement of gene expression in adipose tissue
We recruited 60 non-diabetic subjects undergoing bariatric surgery or elective abdominal surgery such as cholecystectomy or partial hepatectomy at the Ming-Sheng General Hospital and the Yunlin branch of NTUH in Taiwan. Abdominal subcutaneous adipose tissues were sampled from patients in a fasting state prior to surgery and were immediately placed in liquid nitrogen until processing. The study was approved by the institutional review board of Ming-Sheng General Hospital and NTUH. Written informed consent was obtained from each patient.
SNP genotyping
The Tagger program implemented in Haploview version 4.0 software (http://www.broad.mit.edu/mpg/haploview/) [24] was used to select tag SNPs (with a minor allele frequency threshold of 0.1) from the HapMap Chinese Beijing (CHB) databank (Rel 24/phase II). In total, 20 tag SNPs were selected to capture the genetic variation of 8 genes involved in circadian clock, including CLOCK (clock circadian regulator, Gene ID: 9575), BMAL1 (ARNTL, aryl hydrocarbon receptor nuclear translocator-like, Gene ID: 406) PER1 (period circadian clock 1, Gene ID: 5187), PER2 (period circadian clock 2, Gene ID: 8864), CRY1 (cryptochrome 1, Gene ID: 1407), CRY2 (cryptochrome 2, Gene ID: 1408), CSNK1E (casein kinase 1, epsilon, Gene ID: 1454), and NOC (CCRN4L, CCR4 carbon catabolite repression 4-like, Gene ID: 25819). These 20 tag SNPs capture average 71.4% of all SNPs in these 8 genes with a minor allele frequency greater than 0.1 at r2 of 0.7. Genotyping was performed using the GenomeLab SNPstream genotyping platform (Beckman Coulter, Brea CA, USA) and its accompanying SNPstream software suite. The concordance rate of genotyping based on this platform was 99.62%. The raw genotype and phenotype data were deposited as Data S1.
Reverse transcription and quantitative real-time PCR
Total RNA was isolated using REzol™C&T Reagent (Protech, Taipei, Taiwan) and reverse transcribed with Superscript III kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. PCR amplification was performed using LightCycler FastStart DNA master Plus SYBR (Roche, Mannheim, Germany). Each sample was analyzed in duplicate and the gene expression was normalized to that of the PPIA (cyclophilin A), a housekeeping gene. The primers used for NOC were PPH23949A (SA BioScience, Frederick, MD, USA). The primers used for PPIA (cyclophilin A) were forward: 5′-GCATACGGGTCCTGGCATCTTGTCC-3′ and reverse: 5′-ATGGTGATCTTCTTGCTGGTCTTGC-3′, respectively. The correlation (R2) between Ct value and log cDNA input was 0.99 for NOC primers and 0.99 for PPIA primers (Figure S1A&S1B). The efficiency of primers for NOC and PPIA are 0.99 and 0.93, respectively. The slope between delta Ct (Ct of PPIA minus Ct of NOC) and log cDNA input was 17.1% (Figure S1C). There is no difference of Ct of PPIA between obese and non-obese subjects in either subcutaneous or visceral fat (Figure S1D).
Statistical analyses
Data that were not normally distributed were logarithmically transformed to approximate normal distribution. A Hardy-Weinberg equilibrium (HWE) test was performed for each sequence variant before marker-trait association analysis was estimated by using the Haploview software. Linear regression with an additive genetic model was used to analyze SNP association with quantitative traits. Inter-marker linkage disequilibrium (LD) was measured by pairwise D' and r2. Multi-locus interaction between SNPs on metabolic phenotypes was analyzed using the Generalized Multifactor Dimensionality Reduction (GMDR) computing package (http://www.ssg.uab.edu/gmdr/) with adjustment for age and sex. GMDR is a model-free combinatorial approach for detecting multi-locus interactions [25]. The combinations of all factors were classified into high- or low-risk groups based on the score-based statistic. The best n-factor model that yielded a minimum misclassification error was chosen (n = 1, 2, 3 in this study). The significance of the best n-factor model was assessed based on 1,000 permutations. To test for cumulative effects of genetic variants on the individual phenotypes, a subset of SNPs that had nominal associations (P<0.1) with metabolic phenotypes were collected as a risk-SNP set. Consequently, an individual can carry the number of risk alleles ranging from 0 to 2 times of the total number of risk SNPs. For instance, one individual can carry 0 up to 10 risk alleles for BMI, 0 up to 6 risk alleles for SBP or DBP, 0 up to 8 risk alleles for fasting serum triglycerides, and 0 up to 8 risk alleles for fasting plasma glucose. Subjects were then divided into 4, 5 or 6 groups based on the number of risk alleles for each phenotype. An ordinal variable coded as 1–4 for the 4 groups (or 1–5 for the 5 groups; 1–6 for 6 groups) was used as a covariate for testing the trend effect from the cumulative risk alleles on the individual phenotype using a linear regression with adjustment for age and sex.
Supporting Information
Figure S1.
Correlation between Ct value and log (cDNA input) for (A) NOC primers and (B) PPIA primers. (C) slope (dotted line) between delta Ct (Ct of PPIA minus Ct of NOC) and log (cDNA input) (D) Ct of PPIA (internal control) in subcutaneous and visceral fat in obese and non-obese subjects.
https://doi.org/10.1371/journal.pone.0069622.s001
(TIF)
Table S1.
SNP association with metabolic phenotypes according to study populations.
https://doi.org/10.1371/journal.pone.0069622.s002
(DOC)
Acknowledgments
We thank all the patients for their cooperation in this study. All authors declare no conflict of interest.
Author Contributions
Conceived and designed the experiments: YCC YFC LMC. Performed the experiments: YCC HYK SWH. Analyzed the data: PHL. Contributed reagents/materials/analysis tools: TJC YDJ WJL PCL JJH. Wrote the paper: YCC LMC.
References
- 1. Huang W, Ramsey KM, Marcheva B, Bass J (2011) Circadian rhythms, sleep, and metabolism. J Clin Invest 121: 2133–2141.
- 2. Rutter J, Reick M, McKnight SL (2002) Metabolism and the control of circadian rhythms. Annu Rev Biochem 71: 307–331.
- 3. Bass J, Takahashi JS (2010) Circadian integration of metabolism and energetics. Science 330: 1349–1354.
- 4. Stubblefield JJ, Terrien J, Green CB (2012) Nocturnin: at the crossroads of clocks and metabolism. Trends Endocrinol Metab 23: 326–333.
- 5. Buxton OM, Cain SW, O'Connor SP, Porter JH, Duffy JF, et al. (2012) Adverse metabolic consequences in humans of prolonged sleep restriction combined with circadian disruption. Sci Transl Med 4: 129ra143.
- 6. Gangwisch JE (2009) Epidemiological evidence for the links between sleep, circadian rhythms and metabolism. Obes Rev 10 Suppl 237–45.
- 7. Turek FW, Joshu C, Kohsaka A, Lin E, Ivanova G, et al. (2005) Obesity and metabolic syndrome in circadian Clock mutant mice. Science 308: 1043–1045.
- 8. Grimaldi B, Bellet MM, Katada S, Astarita G, Hirayama J, et al. (2010) PER2 controls lipid metabolism by direct regulation of PPARgamma. Cell Metab 12: 509–520.
- 9. Green CB, Douris N, Kojima S, Strayer CA, Fogerty J, et al. (2007) Loss of Nocturnin, a circadian deadenylase, confers resistance to hepatic steatosis and diet-induced obesity. Proc Natl Acad Sci U S A 104: 9888–9893.
- 10. Hee SW, Tsai SH, Chang YC, Chang CJ, Yu IS, et al. (2012) The role of nocturnin in early adipogenesis and modulation of systemic insulin resistance in human. Obesity (Silver Spring) 20: 1558–1565.
- 11. Kawai M, Green CB, Lecka-Czernik B, Douris N, Gilbert MR, et al. (2010) A circadian-regulated gene, Nocturnin, promotes adipogenesis by stimulating PPAR-gamma nuclear translocation. Proc Natl Acad Sci U S A 107: 10508–10513.
- 12. Pasula S, Chakraborty S, Choi JH, Kim JH (2010) Role of casein kinase 1 in the glucose sensor-mediated signaling pathway in yeast. BMC Cell Biol 11: 17.
- 13. Moriya H, Johnston M (2004) Glucose sensing and signaling in Saccharomyces cerevisiae through the Rgt2 glucose sensor and casein kinase I. Proc Natl Acad Sci U S A. 101: 1572–1577.
- 14. Dupuis J, Langenberg C, Prokopenko I, Saxena R, Soranzo N, et al. (2010) New genetic loci implicated in fasting glucose homeostasis and their impact on type 2 diabetes risk. Nat Genet 42: 105–116.
- 15. Scott EM, Carter AM, Grant PJ (2008) Association between polymorphisms in the Clock gene, obesity and the metabolic syndrome in man. Int J Obes (Lond) 32: 658–662.
- 16. Garaulet M, Corbalan MD, Madrid JA, Morales E, Baraza JC, et al. (2010) CLOCK gene is implicated in weight reduction in obese patients participating in a dietary programme based on the Mediterranean diet. Int J Obes (Lond) 34: 516–523.
- 17. Sookoian S, Gemma C, Gianotti TF, Burgueno A, Castano G, et al. (2008) Genetic variants of Clock transcription factor are associated with individual susceptibility to obesity. Am J Clin Nutr 87: 1606–1615.
- 18. Garaulet M, Lee YC, Shen J, Parnell LD, Arnett DK, et al. (2009) CLOCK genetic variation and metabolic syndrome risk: modulation by monounsaturated fatty acids. Am J Clin Nutr 90: 1466–1475.
- 19. Garaulet M, Corbalan-Tutau MD, Madrid JA, Baraza JC, Parnell LD, et al. (2010) PERIOD2 variants are associated with abdominal obesity, psycho-behavioral factors, and attrition in the dietary treatment of obesity. J Am Diet Assoc 110: 917–921.
- 20. Englund A, Kovanen L, Saarikoski ST, Haukka J, Reunanen A, et al. (2009) NPAS2 and PER2 are linked to risk factors of the metabolic syndrome. J Circadian Rhythms 7: 5.
- 21. Woon PY, Kaisaki PJ, Braganca J, Bihoreau MT, Levy JC, et al. (2007) Aryl hydrocarbon receptor nuclear translocator-like (BMAL1) is associated with susceptibility to hypertension and type 2 diabetes. Proc Natl Acad Sci U S A 104: 14412–14417.
- 22. Lin JW, Chang YC, Li HY, Chien YF, Wu MY, et al. (2009) Cross-sectional validation of diabetes risk scores for predicting diabetes, metabolic syndrome, and chronic kidney disease in Taiwanese. Diabetes Care 32: 2294–2296.
- 23. Chang YC, Chang TJ, Jiang YD, Kuo SS, Lee KC, et al. (2007) Association study of the genetic polymorphisms of the transcription factor 7-like 2 (TCF7L2) gene and type 2 diabetes in the Chinese population. Diabetes 56: 2631–2637.
- 24.
Barrett JC (2009) Haploview: Visualization and analysis of SNP genotype data. Cold Spring Harb Protoc 2009: pdb ip71.
- 25. Lou XY, Chen GB, Yan L, Ma JZ, Mangold JE, et al. (2008) A combinatorial approach to detecting gene-gene and gene-environment interactions in family studies. Am J Hum Genet 83: 457–467.