Introduction
Metabolic syndrome (MetS) indicates a cluster of cardiovascular disease (CVDs) risk factors associated with high mortality rate, including abdominal obesity, dyslipidemia, hypertension, and insulin resistance [
1]. People with MetS have a five-times higher risk of type 2 diabetes (T2D), and are two to three times more at risk of CVDs than those without MetS [
2]. In today’s advanced, industrial societies, MetS is an important problem in the development of chronic diseases, thus understanding the factors affecting MetS is a key research challenge [
3].
Although food consumption has been correlated with MetS components [
4,
5], the role of diet in the development of MetS is not well understood. The application of dietary patterns has recently attracted interest in epidemiological nutrition [
6,
7]. In addition, dietary patterns have demonstrated significant relationships with certain plasma risk factors for MetS and metabolic dysfunction [
8,
9], such as glycemic indices, plasma lipids and liver enzymes. There are many indications which show that exposure to environmental factors such as dietary patterns, physical activity, and certain genetic traits have effects in the development of this syndrome [
10,
11]. Findings from previous studies have shown that exposure to different environmental factors may alter the effects of genetic factors. The risk of chronic diseases resulting from high-risk alleles is also affected by various dietary patterns. Among the genes that are affected by environmental factors are genes associated with vitamin D binding protein (DBP) [
10].
DBP, originally known as Gc-globulin (Group-specific component) is the transporter of circulating vitamin D and its metabolites, as well as being active in fatty acid binding and actin scavenging [
12]. Proteomic analysis indicates that Gc is a hepatic acute phase reactant and is down-regulated in patients with type I diabetes [
13], hepatocellular carcinoma [
12,
14], primary non-metastatic breast cancer [
15], and sepsis [
16]. It is up-regulated in patients with diabetes type II [
17], early-stage breast cancer [
18], Alzheimer and Parkinson disease [
19].
A remarkable
DBP polymorphism in humans and primates has been demonstrated using various electrophoresis methods [
20]. In addition to the three known alleles (
GC1F,
GC1S and
GC 2), over 120 rare variants have been identified, making the Gc-locus amongst the most polymorphic recognized. The initial structure of three major allele-level polymorphisms of
DBP are according to the combination of two single nucleotide polymorphisms (SNP) located on chromosome 4q12-q13, rs7041 (p. Glutamate 416 Aspartate) and rs4588 (p. Threonine 420 Lysine). These SNPs correspond to allele-level polymorphisms as follows: rs7041 (p. 416 = Asp) + rs4588 (p. 420 = Thr) to
GC-
1F, rs7041 (p. 416 = Glu) + rs4588 (p. 420 = Thr) to
GC-
1S and rs7041 (p. 416 = Asp) + rs4588 (p. 420 = Lys) to GC-2. The combination of rs7041 (p.416 = Glu) + rs4588 (p.420 = Lys) does not correspond to any of the three major alleles [
10]. Of the
GC1 genotypes,
GC1S is most abundant in European populations, whereas
GC1F is most abundant among those of African ancestry, with Asians exhibiting intermediate frequencies of both
GC1 forms. The
GC2 form is exceedingly rare in black ethnic groups and is found at similar frequencies in people of Asian and European ancestry [
21].
Concentrations of 25-hydroxyvitamin D (25(OH)D) are related to DBP concentration, whereas any type of DBP phenotype has its own specific amount of 25(OH)-vitamin D, with the hierarchy of affinity being
GC1F>
GC1S>
GC2. In addition, this hierarchy (
GC1F>
GC1S>
GC2) also applies to the relative abundance of the respective forms in human serum [
22].
There is much evidence about the associations between vitamin D metabolism, lipolysis and MetS, and some studies have shown the association of MetS with changes in vitamin D concentrations [
23‐
25]. However, it is not known whether the DBP polymorphism affects the metabolic syndrome and its components through changes in vitamin D concentrations.
Although several environmental and genetic factors have been established in relation to metabolic dysfunction [
26,
27], few studies have investigated the interactions of dietary patterns and genetic variants with MetS; most of them have focused primarily on nutrients [
28]. In this study, the interactions between dietary patterns and
DBP polymorphism rs7041/rs4588 separately and combined (
GC1F,
GC1S and
GC 2 alleles) were investigated in terms of the odds of MetS and its components.
Methods and materials
Study population
The study population was recruited using advertising from July 2015 to June 2016 across the West and Central regions of Tehran according to cluster sampling techniques, and using community-based sampling. 363 participants (18–50 years) were enrolled, 98 participants were excluded and 265 participants who were apparently healthy and not known to have MetS, took part in this cross-sectional study and were included with the following criteria: aged 18–50 years, no smoking and alcohol use, absence of any acute or chronic inflammatory disease, and not being pregnant during the study. Participants with a history of CVDs, T2D, cancer or stroke and dietary supplement use (including vitamin D), were excluded because of possible disease-related changes in diet, as well as those taking any therapeutic medications. Also, all participants who did not respond to more than 70 food items from the FFQ, or whose energy consumption lay outside the range 800 kcal/day (3347 kJ/day) to 4200 kcal/day (17,573 kJ/day), [
29] were not included in the study.
Measurement of biochemical parameters
After completing the consent letter, the blood samples were collected into tubes containing 0.1% EDTA after 10–12 h fasting, then immediately centrifuged, aliquoted and stored at a temperature of − 80 °C. All samples were evaluated by the same method and under similar conditions. Serum 25(OH)D3 was measured using a Biosource kit (Bio-source Europe S.A, Belgium) by the radioimmunoassay (RIA) method. Serum glucose level was measured by the glucose oxidase phenol 4-aminoantipyrine peroxidase (GOD/PAP) method, triglycerides (TG) with the glycerol-3-phosphate oxidase phenol 4-aminoantipyrine peroxidase (GPO-PAP) method, total cholesterol by endpoint enzymatic method, and high-density lipoprotein cholesterol (HDL-C) with an enzymatic clearance assay. All of the above measurements were made using Randox laboratory kits (Hitachi 902 Analyzer; Hitachi LTD, Tokyo, Japan). Quality control was carried out for all biochemical assessments using the control serum, as in the manufacturer’s instructions.
MetS was defined according to the adult treatment panel III (ATPIII) criteria [
30]. A subject has MetS when at least 3 of the fol-lowing 5 risk factors are present: (1) hyperglycemia as FBS ≥ 100 mg/dl (5.6 mmol/l); (2) hypertriglyceridemia as serum TG ≥ 150 mg/dl (1.69 mmol/l); (3) low HDL-C serum < 40 mg/dl (1.04 mmol/l) for men, and < 50 mg/dl (1.29 mmol/l) for women; (4) hypertension as BP ≥ 130/85 mmHg; and (5) abdominal adiposity (defined as waist circumference 88 cm [women] or 102 cm [men]).
Assessment of anthropometric measures
Height was measured in a standing position without shoes, using a tape meter, while the shoulders were in a normal state (with a precision of 0.5 cm). Weight was measured while the subjects were minimally clothed without shoes, using digital scales (with an accuracy of 0.1 kg). WC was measured at the narrowest point, and hip circumference at the widest point, over light clothing, using a non-elastic tape meter, without any pressure to the body surface [
31].
Assessment of blood pressure
After the participants sat in a relaxed position for 15 min, blood pressure was measured using a standardized sphygmomanometer [
32].
Assessment of other variables
Additional covariate information regarding physical activity, smoking habits, medical history, and current use of medication was obtained using questionnaires completed during the screening. The validated International Physical Activity Questionnaire (IPAQ) was used to obtain data on physical activity [
33]. This questionnaire has also been validated in Iran [
34]. It assesses physical activity across a complete set of domains, including domestic, leisure time and gardening activities, as well as work-related and transport-related activities. Data were obtained on the frequency and time spent on light, moderate and hard intensity activities, according to the list of common daily activities over the past year [
35].
Dietary intake assessment and extraction of dietary patterns
Information about food intake was collected with a semi-quantitative food frequency questionnaire (FFQ). The FFQ, originally developed for this study, was a Tehran Lipid and Glucose Study (TLGS) format questionnaire, and contained questions about the average consumption and consumption frequency of 147 food items during the past year [
36]. The food items were chosen according to the most frequently consumed items in the national food consumption survey in Iran [
37]. Subjects illustrated their food consumption frequencies on a daily basis (bread), weekly basis (rice and meat), monthly basis (fish), yearly basis (organ meats), or a never/seldom basis according to portion sizes provided in the FFQ. The FFQ was based on food items rather than dishes, since different recipes are used in food preparation (for example, beans, different meats and oils, and rice). In order to complete the FFQ, the interview session took about 45 min. The interviewer read out the food items on the FFQ, and recorded their serving size and frequency. The weight of seasonal fruits was estimated according to the number of seasons within which each food was available. Daily intakes of each food item were determined based on the portion size or household measures, and multiplied by the consumption frequency of each food item [
38].
Based on the similarity of nutrients, food items were grouped into 24 pre-defined food groups. This classification was based on the similarity of their nutrients, according to previous studies [
39]. If the nutrients composition of a food item was considerably different from other items (like eggs, margarine, tea and coffee), or if its consumption was a special eating habit (like butter), it was considered as an independent group. Then, the adjusted means for energy were calculated for each food group using the residual method [
40]. In this way, the adjusted energy values were obtained.
In the next step, to determine the suitability of the model, the Kaiser–Meyer–Olkin (KMO) and the Bartlett test were used. Dietary patterns were determined by factor analysis (FA). The principal component analysis method (PCA) with varimax rotation was applied to these energy-adjusted food groups. The factors obtained were checked on the basis of the Eigen-values of energy-adjusted food groups; each factor having an Eigen-value of greater than 1.5 was considered as a major dietary pattern. The designation of patterns was based on the interpretation of food items in each factor, which together accounted for 27.9% of the total variation, on the basis of the scree plot and varimax rotation in the 24 food groups [
41]. It should be noted that other food patterns were identified, but were not considered due to their low variance. Then, the subjects were categorized into quartiles, according to dietary patterns. The major dietary patterns were identified as follows, on the basis of previous knowledge:
A.
Healthy dietary pattern: rich in green vegetables, liquid oils and olive oil, fish and poultry, fruits, legumes and eggs, but low in high-energy drinks, sweets and snacks.
B.
Traditional dietary pattern: high-fat dairy products, dry fruits, starchy vegetables, seeds, legumes, and seasonings were the main components. The main feature of this dietary pattern was fiber consumption.
C.
Western dietary patterns: high consumption of refined grains, high-energy drinks and beverages, solid fats, mayonnaise, low-fat dairy products; limited consumption of unrefined grains, vegetables, liquid oils and low-fat dairy products.
Genomic DNA was extracted from whole blood using the Mini Columns Type kit. The Gene ALL DNA kit (Gene ALL, Korean) was used according to the manufacturer’s protocol. The concentration and purity of extracted DNA was measured using a Nano Drop ND-1000 spectrophotometer. DNA was stored at − 20 °C until ready for use. The
DBP polymorphisms in exon 11 [Asp416Glu and Thr420Lys] were determined by polymerase chain reactions–restriction fragment length polymorphism (PCR–RFLP). The PCR procedures were carried out in a total volume of 30 µl with 50 ng of genomic DNA, 2.5 U Blue Master Mix-Tembase Buffer, 1 mmol Dimethyl sulfoxide, 0.2 mM of each dNTP, and 10 pmol of each primer (forward: 5′-CAAGTCTTATCACCATCCTG-3′ and reverse: 5′-GC CAAGTTACAATAACAC-3′) [
42]. PCR amplifications were performed according to the following cycling conditions: initial denaturation at 94 °C for 5 min, followed by 35 cycles at 94 °C for 45 s, 60 °C for 1 min, and 72 °C for 1 min, with a final extension at 72 °C for 10 min.
Then, while being kept at 37 °C overnight, 10 µl of the amplified products were digested in a 20 µl reaction containing either 2.5 U
HaeIII or 2.5 U
StyI restriction enzymes (Fermentas, Germany) for both Asp, 416, Glu and Thr, 420, Lys polymorphisms, respectively. The digestion products were visualized, stained with ethidium bromide on 2% agarose gel. Individuals homozygous for the Glu allele showed two bands (577 and 232 bp) in the DBP Asp, 416, Glu polymorphism while individuals homozygous for the Asp allele had a nondigested band at 809 bp. The homozygous Thr allele of the Thr, 420, Lys polymorphism appeared as a single band (809 bp), while the Lys allele showed two bands (584 and 225 bp).
DBP haplotypes were determined by observing the digestion products of both restriction enzymes. The
GC1S had the
HaeIII but not the
StyI site. The
GC1F had neither the
HaeIII nor the
StyI site. The
GC2 had the
StyI but not the
HaeIII site. The existence of both restriction sites on a single haplotype has not yet been described [
43]. We checked DNA purity by nanodrop and to confirm the PCR-RFLP results, 10% of the PCR samples were directly sequenced.
The minor allele (risk allele) frequency for rs7041T was 0.488, while that of rs4588A was 0.235. They were divided into six main genotypes, which included three homozygotes,
GC1F-
1F, GC1S-
1S and
GC2-
2, and heterozygotes,
GC1F-
1S,
GC2-
1F and
GC2-
1S. The genotypes were coded as follows: code 0 for (
GC1F/F), code 1 for (
GC1F/S), code 2 for (
GC1F/2), code 3 for (
GC1S/S), code 4 for (
GC1S/2) and code 5 for (
GC2/2). The Haplotypes were coded as follows: code 0 for (
GC1F), code 1 for (
GC1S) and code 2 for (
GC2), based on the risk allele carrier (Table
1).
Table 1
Allele and genotype frequencies of rs7041 and rs4588 polymorphisms among subjects
G | 51.12 | C | 76.42 | GC | 44.2 |
GA | 6.8 |
T | 48.88 | A | 23.57 | TC | 32.1 |
TA | 16.7 |
GG | 26.4 | CC | 52.8 | TT CC | 4.9 |
TG CC | 21.1 |
GT | 48.7 | CA | 46 | TT CA | 19.9 |
GG CC | 26.4 |
TT | 24.2 | AA | 0.4 | TG CA | 27.5 |
TT AA | 0.4 |
Haplotype (GC1F/F) | TT CC (13) | | | | |
Haplotype (GC1F/S) | TG CC (57) | | | | |
Haplotype (GC1F/2) | TT CA (50) | | | | |
Haplotype (GC1S/S) | GG CC (70) | | | | |
Haplotype (GC1S/2) | TG CA (72) | | | | |
Haplotype (GC2/2) | TT AA (1) | | | | |
Total | 263 | | | | |
Haplotype 1 (GC1F) | T-C (133) | | | | |
Haplotype 2 (GC1S) | G-C (269) | | | | |
Haplotype 3 (GC2) | T-A (124) | | | | |
Total | 526 | | | | |
Statistical analysis
The normality of distribution was tested by applying Kolmogorov–Smirnov’s test. Data on quantitative characteristics were reported as the mean ± SD, and data on qualitative characteristics were expressed as a percentage. Qualitative variables were compared using an analysis of variance (ANOVA), and an independent t-test was used to compare the quantitative variables.
To investigate the interactions between dietary patterns and haplotypes of rs7041 and rs4588, binary logistic regression models were used on the qualitative variables of MetS and its components. Moreover, age, sex, physical activity (PA) and vitamin D 25(OH)D3 concentrations were considered as confounders, and adjusted analysis was performed in these models. The level of significance was set at a probability of ≤ 0.05 for all tests, and P < 0.1 was considered as marginal significance. Statistical analysis was performed using SPSS version 22.0 (SPSS, Chicago, IL, USA).
Discussion
In the present cross-sectional study of Tehrani adults, the interactions between dietary patterns and genetic variants of DBP SNPs (rs7041, rs4588) were assessed in relation to MetS risk levels and MetS components. The two SNPs (rs7041 and rs4588) were found to be in linkage disequilibrium (D′ = 1, r2 = 0.23), which meant that there was a nonrandom association between alleles of the two SNPs.
The interactions between all three dietary patterns and rs7041 showed noteworthy effects on the odds of MetS. In two of the regression models, even after adjusting for confounders, the significant interaction between the healthy pattern, rs7041, and MetS did not change. Beyond this, two strong interactions were observed between rs4588 and both the healthy pattern and western pattern. An interaction between the quartiles of the healthy pattern and rs4588 alleles had strongly significant effects on reducing MetS odds, exactly like the interaction effect between the healthy pattern and rs7041. On the contrary, a higher intake of the western pattern among the risk allele carriers of rs4588 was associated with higher odds of MetS. However, significant interactions between traditional and western patterns with rs7041 were seen only after adjusting for the effect of vitamin D. This can be seen as a demonstration of the effects of
DBP gene variants on the bioavailability of vitamin D (according to past studies [
44,
45]), and an indication of vitamin D’s subsequent effects on MetS components. The mechanisms of this relationship are not clear. But, 1,25(OH)
2 vitamin D3 is essential for normal insulin secretion and there is a possibility that DBP is a carrier protein for vitamin D and the affinity of DBP for 25-OH-vitamin D3 and 1,25 (OH)
2 vitamin D3 varies depending on the genotype of DBP [
46]. DBP genotype have been previously shown to be strongly associated with circulating 25-OH-vitamin D3 in diverse populations [
47,
48] and in genome-wide association studies [
49,
50].
In addition, subsequent to Baier et al.’s [
51] initial report describing the linkage of this locus to prediabetic phenotypes in Pima Indians, the Gc locus has been reported to be associated with noninsulin-dependent diabetes mellitus in Japan [
52].
In the ARIC study [
53], lower compared with higher serum concentrations of 25(OH)D were associated with major incidents of strokes, over nearly 20 years of follow-up. This association was similar by race, but there was suggestive evidence (though not statistically significant), that associations between low 25(OH)D and stroke incidence were stronger among participants genetically disposed to higher serum DBP (i.e. those with presence of rs7041 G allele and rs4588 A allele). Individuals with a rs7041G allele (versus a T allele) and individuals with a rs4588 A allele (versus a C allele) have been shown to have higher DBP levels, and therefore lower levels of bioavailable 25(OH)D for a given 25(OH)D level. In this study, people who had a higher adherence to the healthy pattern and had rs7041 G allele and rs4588 A allele were less likely to be at risk of MetS, and vice versa.
According to the authors’ knowledge, this study is the first study to examine the interactions between dietary patterns and
DBP gene variants on MetS. However, among studies that examine the interactions between dietary patterns and other genes on MetS, the study of Flips et al. [
54] should be mentioned; it showed that an interaction between dietary fat, especially PUFA, and ACSL1 gene polymorphisms, is effective in reducing the risk of MetS. Therefore, it can be concluded that—to a large extent—the success of nutritional interventions for controlling weight, diabetes, and MetS components depends on the genetics of different individuals.
A noteworthy point about the traditional dietary pattern in this study is its high fiber content, which according to past studies [
55], especially Whelton et al.’s meta-analysis [
56], seems to lead to reduced systolic blood pressure. Traditional food patterns identified in the present study were not similar to patterns extracted in other studies [
57], in that high-fat dairy products, starchy vegetables and legumes were similar components in this study and other studies, while red meat and butter were not. Furthermore, the consumption of healthy food items was observed in this pattern, including fruits and unrefined grains. Thus, by taking these items into consideration, this pattern could also be termed a fiber-rich pattern, or a neo-traditional pattern; its interaction with DBP leads to reductions in WC and blood pressure.
The protective effect of healthy patterns in interacting with the
DBP gene on FBS levels is also due to its low consumption of sweets, as well as high-energy drinks and beverages. Also, the effect of the healthy pattern on serum TG levels and MetS, in addition to the frequent consumption of items like vegetables and fish, can be seen in milk consumption. Low-fat dairy products score highly in the healthy pattern, and milk appears to be insulinotropic [
58].
In addition to the effect of the patterns, there are at least two reasons why some associations between the DBP haplotypes and MetS components can be suspected. First, the metabolically active form of vitamin D is involved in a feedback system of insulin regulation, because DBP binds vitamin D [
59]. Second, DBP plasma levels are under genetic control: almost 84% of the variation in DBP protein is explained by genes, and the remainder by environmental factors, some of which may be sex-specific [
60].
GC variants have formerly been associated with type 2 diabetes in seven Polynesian populations, and insulin regulation and glucose in a Hispanic-American/Anglo population of Southern Colorado and Dogrib Indians of Canada [
59]. In this study, blood pressure was measured only once, which could suggest low precision in the hypertension estimate, causing attenuated relationships. However, with these conditions, the significant interaction of the traditional dietary pattern and the
DBP gene was only observed with systolic blood pressure in the regression model before adjustment for vitamin D concentrations.
To the authors’ knowledge, one strength of the current study is that it is the first study to evaluate the interactions between
GC gene variants and major dietary patterns on the odds of MetS and its components. Additionally, because it included only Caucasians with Iranian ancestry, its homogeneous population was a strength (and therefore results were less likely to be affected by population stratification). However, in the interpretation of this study, certain limitations should be taken into consideration. First, the main limitation of the present study was its low sample size, which probably led to insufficient sample for the statistical analyses for participants presented with the metabolic syndrome. Second, dietary patterns can vary by socioeconomic status, ethnic group, and culture. Thus, it is necessary to replicate the results of the current study in other populations. Also, the genetic variants identified only account for a very small section of the variability in vitamin D status, and this study is consistent with other studies based on protein isoforms of this binding protein [
61,
62]. Also, there is consistent support for one SNP in the enzyme that converts the storage form of the vitamin (25(OH)D) to the active ligand 1-25(OH)D (CYP27B1; rs10877012) [
63], and one SNP in the vitamin D receptor gene (
VDR; rs10735810) [
64]. As mentioned above, genetic variants, through changes in vitamin D status, cause changes in health outcomes, including in the components of metabolic syndrome.
While the body of evidence is still incomplete, a more comprehensive assessment of gene-diet interactions with large sample size to have enough power, will need to include more genetic risk factors for MetS when they are identified, and to perform analyses on the interactions between specific variants and dietary intakes.
Authors’ contributions
MHR: wrote the manuscript, MM: conceived of the study, and participated in its design and coordination and compiled the data, SY: participated in the design of the study and performed the statistical analysis, AM: helped to draft the manuscript, ZhM: carried out the molecular genetic studies and participated in the sequence alignment, KhM: analysed, reviewed and evaluated all data. All authors read and approved the final manuscript.