Introduction

Metabolic cycles are influenced by the circadian clock to such an extent that the demarcation between metabolic and circadian oscillations can be regarded as somewhat arbitrary.1 Although the links from genetic variants through transcription to physiologic functions are most likely less linear than predicted,2 a landmark study identified obesity and features of metabolic syndrome as characteristics of clock circadian regulator (CLOCK)-deficient mice.3 Since then, human studies have identified genetic variants and expression patterns of circadian clock genes, such as aryl hydrocarbon receptor nuclear translocator-like (ARNTL or BMAL1), CLOCK, neuronal PAS domain protein 2 (NPAS2) or period circadian clock 2 (PER2), that are associated with metabolic syndrome, hypertension or type 2 diabetes.4, 5, 6, 7, 8, 9, 10, 11

However, as cryptochrome circadian clock 1 (CRY1) and cryptochrome circadian clock 2 (CRY2) have a key role in the reciprocal regulation of the metabolic and circadian networks,12, 13 we wanted to analyze whether CRY1 or CRY2 genetic variants are associated with metabolic syndrome or its components. In addition, because protein kinase C, delta binding protein (PRKCDBP or CAVIN-3) regulates not only the circadian period but also interactions and abundance of the period-cryptochrome protein dimers,14 we studied the associations of PRKCDBP genetic variants with metabolic syndrome and its components. Furthermore, because CRY2 has been indicated in the control of fasting glucose levels,15 to confirm and extend these findings, we analyzed whether CRY2 genetic variants are associated with a range of glucose metabolism indicators in our sample.

Methods

Subjects

Our sample included 5910 individuals (56% women) with available blood samples, the Munich-Composite International Diagnostic Interview (M-CIDI)16 and the self-report on seasonal changes in mood and behavior. The sample is part of a national health survey, Health 2000, of the population aged ⩾30 years (n=8028; http://www.terveys2000.fi/indexe.html) that was approved by the ethics committees of the National Public Health Institute and the Helsinki and Uusimaa Hospital District. All participants provided written informed consent.

Phenotypes

Routine fasting laboratory tests included the concentrations of blood glucose, serum insulin, serum total cholesterol, triglycerides and high-density lipoprotein cholesterol. Blood glucose was measured one hour after the oral glucose intake. The Homeostasis Model Assessment insulin resistance and beta-cell function indexes were then computed. Blood pressure and waist circumference were also measured.

Metabolic disorder was assessed using two sets of criteria (Table 1): those of the US Adult Treatment Panel III of the National Cholesterol Education Program (NCEP-ATPIII)17 and those of the International Diabetes Federation (IDF).18 As a post-hoc test, elevated blood pressure was also analyzed, to dissect the parameter from the metabolic syndrome components, using the current WHO (World Health Organization)-based definition of 140/90 mm Hg or over.19

Table 1 Criteria for metabolic disorder and its risk factors used in this study and number of subjects in each of them
Full size table

Gene and SNP selection

CRY1, CRY2 and PRKCDBP single-nucleotide polymorphism (SNP) selection was based on HapMap phase 3 data (http://www.hapmap.org/) and performed using the Tagger program in the Haploview 4.1 software.20 The linkage disequilibrium within the genes and within 10 kb of their 5′ and 3′ flanking regions, that is, 122 kb for CRY1 (chr12:105 899–106 021 kb, NCBI36/hg18 assembly), 56 kb for CRY2 (chr11:45 815–45 871 kb) and 22 kb for PRKCDBP (chr11: 6286–6308 kb), was used to select SNPs capturing most of the genetic variation.

The aim was to capture all the SNPs with a minor allele frequency >5% in the European population (CEU and TSI) in the HapMap database by setting the pair-wise r2 to ⩾0.9. Ten out of 21 CRY1 and 10 out of 34 CRY2 SNPs fulfilled the criterion and were all successfully included in the genotyping multiplexes. Of the 12 out of 19 PRKCDBP SNPs fulfilling the criterion, 8 were successfully included. In addition to the aforementioned tag-SNPs, 20 potentially functional CRY1 (12) and CRY2 (8) variants were selected using Pupasuite,21 Variowatch,22 database of SNPs affecting miR Regulation (dbSMR)23 and microRNA SNP24 databases and were included in the study. Table 2 presents the 48 successfully genotyped SNPs.

Table 2 Successfully genotyped SNPs, their selection criteria, allele and genotype frequencies and Hardy–Weinberg equilibrium P-values
Full size table

Genotyping

Genomic DNA was isolated from whole blood according to standard procedures. The SNPs were genotyped at the Institute for Molecular Medicine Finland, Technology Centre, University of Helsinki, using the MassARRAY iPLEX method (Sequenom, San Diego, CA, USA),25 with excellent success (>95%) and accuracy (100%) rates.26 For quality control purposes, positive (CEPH) and negative water controls were included in each 384-plate. Genotyping was performed blind to phenotypic information.

For CRY1 and CRY2 and for PRKCDBP and CRY1 rs11113153 and rs10861695, 173 and 238 individuals were removed, respectively, due to a high missing genotype rate (that is, >0.1). The total genotyping rate in the remaining individuals was 0.999. Three SNPs were removed because the minor allele frequency was <0.01: CRY2 rs3747548, CRY2 rs35488012, CRY1 rs7294758. Finally, there were 5737 (CRY1 and CRY2) or 5672 (PRKCDBP and CRY1 rs11113153 and rs10861695) individuals and 45 SNPs used in the statistical analysis.

Statistical analysis

Statistical analysis was performed using logistic or linear regression and additive genetic models controlling for age and sex with PLINK software v1.07.27 Haplotype blocks were defined using Haploview software20 and the confidence interval algorithm. For continuous phenotypes, 10 000 permutations were used to produce empirical P-values to relax the assumption of normality. The results of each set of analyses were corrected for multiple testing by calculating false discovery rate q-values28 using R software (http://www.r-project.org/). The q-values of <0.05 were considered to be significant, and the P-values of <0.05 were considered to be nominally significant.

Results

Genotypes, allele frequencies and Hardy–Weinberg equilibrium estimates are shown in Table 2. The associations presented in the text refer to the NCEP-ATPIII criteria of metabolic syndrome, and the results for IDF criteria are presented in the Supplementary Material. Both criteria produced similar results. Five intronic CRY1 SNPs (rs4964513, rs11613557, rs59790130, rs4964518 and rs12821586) and two CRY2 SNPs (rs7121611 and rs7945565) had nominally significant associations with hypertension (Table 3 and Supplementary File 1). Three CRY1 SNPs (rs2888896, rs10746077 and rs2078074) and one CRY2 SNP (rs75065406) had nominally significant associations with elevated triglycerides, and the CRY2 SNP (rs75065406) had a nominally significant association with metabolic syndrome as well. However, after correcting for multiple testing, none of the associations remained significant.

Table 3 Nominally significant single SNP associations using the NCEP-ATPIII criteria for metabolic syndrome
Full size table

In the haplotype analysis, three haplotype blocks were formed for CRY1 (Figure 1), and one each for CRY2 (Figure 2) and PRKCDBP (Supplementary File 2). Table 4 (see also Supplementary File 3) presents the nominally significant haplotype associations (P<0.05). The haplotype analysis supported the association of CRY1 5′-CGG-3′ and 5′-TGA-3′ (Block 1) and 5′- CTTCGTCCTTAG -3′ (Block 3) haplotypes with hypertension. CRY1 5′- CCCCACTCTCAG -3′ and 5′- GTCTGCCCCCAT -3′ haplotypes associated with elevated triglycerides. CRY2 5′- ATTTGCGGTGGCACG -3′ haplotype associated with elevated triglycerides and metabolic syndrome.

Figure 1
figure 1

The analyzed circadian clock 1 single-nucleotide polymorphisms (CRY1 SNPs) in this study, their location and the haplotype block structure of the area formed, based on our sample showing r2-values.

Full size image
Figure 2
figure 2

The analyzed circadian clock 2 single-nucleotide polymorphisms (CRY2 SNPs), their location and the haplotype block structure constructed using the Haploview program showing r2-values.

Full size image
Table 4 Nominally significant haplotype associations using the NCEP-ATPIII criteria for metabolic syndrome
Full size table

A priori, we planned to extend the metabolic findings of CRY2 and analyze CRY2 SNPs in relation to indicators of glucose metabolism. CRY2 SNPs had nominally significant associations with the indicators, but these associations did not survive after correction for multiple testing (Supplementary Files 4 and 5).

Post-hoc, we wanted to dissect the parameter of elevated blood pressure from the metabolic syndrome components. Four CRY1 SNPs (rs17289712, rs4964513, rs59790130 and rs11613557), one CRY2 SNP (rs75065406) and two PRKCDBP SNPs (rs2947030 and rs4758095) had nominally significant associations with elevated blood pressure (Supplementary File 6). After correcting for multiple testing, none of the associations remained significant. The haplotype analysis, however, supported the association of CRY1 and CRY2 with elevated blood pressure, as the 5′-CGG-3′ (CRY1 Block 1), 5′- CCCCACTCTCGG -3′ and 5′- CTTCGTCCTTAG -3′ (CRY1 Block 3) and 5′- ATTTGCGGTGGCACG -3′ (CRY2) haplotypes associated with elevated blood pressure (Supplementary File 7).

Discussion

Our results suggest that CRY1 genetic variants may have a role in elevated blood pressure and hypertension. Previously, other core circadian clock genes were implicated in the regulation of blood pressure whose systolic component follows a circadian rhythm.29 Circadian clock disruption has been implicated in the pathogenesis of cardiovascular disease, for which hypertension is a major factor. We found no evidence in our study sample to support the findings that relate CRY2 to fasting glucose levels or indices of glucose metabolism.

Our study has some limitations. Our systematic screening for the metabolic syndrome and its components in relation to the SNPs covering three genes yielded results that did not reach study-wide significance. Our results indicated that metabolic syndrome, as such, is not associated with the genetic variants of CRY1, CRY2 or PRKCDBP. However, the false discovery rate procedure used does not take into account the linkage disequilibrium between the SNPs, and our haplotype analysis gave further support to the one-phenotype association of hypertension and elevated blood pressure with the CRY1 SNPs.

Cryptochromes act as key repressors in the core of the circadian clocks.30, 31, 32, 33 Both CRY1 and CRY2 act as repressors, but the actions of CRY1 are opposed by CRY2.33 Furthermore, PER1 antagonizes CRY2, through which PER1 target genes are activated.34 Actions of PER1 have a potential contribution to visceral fat accumulation,35 functions of beta-cells in the pancreas36 and synchronization of the peripheral liver clock by insulin.37 In addition to actions in the nucleus, the cryptochrome proteins act as inhibitors of adenylyl cyclase, thereby limiting cyclic adenosine monophosphate production,38 and they act as inhibitors of G protein coupled receptor activity through a direct interaction with the G(s)alpha subunit.39

Genetic loss of cryptochromes does change physiology. Cryptochromes appear relevant to the pathogenesis of metabolic syndrome, as cryptochromes participate in glucocorticoid regulation of gluconeogenesis and steroidogenesis.40 Cryptochrome-deficient mice have elevated sympathetic nerve activity and impaired glucose tolerance,41 and increased susceptibility to glucocorticoid-induced hyperglycemia with glucose intolerance and constitutively high levels of circulating corticosterone.40 When cryptochrome-deficient mice are challenged with a high-salt diet, they have hypertension due to abnormally high synthesis of the mineralocorticoid aldosterone by the adrenal gland,42 and when challenged with a high-fat diet, obesity develops due to the increased insulin secretion and lipid storage in white adipose tissue.43 Genetic loss of cryptochromes constitutively activates proinflammatory cytokine expression, and then the innate immune system becomes hypersensitive, the NF-κB signaling pathway is constantly activated and the PKA signaling activity is constitutive.38

In addition to these findings, in CRY2 knock-down experiments, genes contributing to inflammation are upregulated, the proinflammatory cytokine activity through the actions of interleukin-6 and interleukin-18 is increased, and genes contributing to immune responses are upregulated.44 All of these features are also part of metabolic syndrome, and here, we hypothesize that dysfunction of cryptochromes might be an overarching factor with a shared effect that contributes to the changes in physiology. Disruption of circadian clocks seems to affect not only the metabolic and cell-division cycles,45, 46 but also mood and behavior.47, 48 It is therefore likely that in humans, cryptochromes have a role in the pathogenesis of these medical conditions and mental disorders.49, 50 Based on our current results, the role of CRY1 in the pathogenesis of cardiovascular diseases, and its contribution to elevated blood pressure deserve further study.