Introduction

Diet and physical activity are the major lifestyle determinants of the risk of type 2 diabetes, as demonstrated in diabetes prevention studies, e.g., in the Finnish Diabetes Prevention study (Knowler et al. 2002; Tuomilehto et al. 2001). Recent genome-wide association analyses have identified >70 gene variants that contribute to type 2 diabetes (Huyghe et al. 2012; Morris et al. 2012) and >30 genes that contribute to obesity (Sandholt et al. 2012). However, none of these variants explain >5 % of the risk of these conditions, and altogether, they explain <10 % of the total disease risk (Manolio et al. 2009). Interactions between gene variants and the environment, i.e., our inherited responses to environmental changes, are likely to be crucial in the development of obesity and type 2 diabetes (Cornelis and Hu 2012). Despite a relatively small sample size, interactions between several genes, including PPARG, and lifestyle intervention have been demonstrated in randomized lifestyle intervention trials (Lindi et al. 2002; Uusitupa 2005).

We were the first to identify the Pro12Ala polymorphism of PPARG as a risk factor for type 2 diabetes. This polymorphism is expressed only in the adipose tissue-specific isoform PPARG2 of PPARG, and the Ala12 allele associates with lower transcriptional activity (Deeb et al. 1998). Low prevalence of type 2 diabetes (Altshuler et al. 2000) and dyslipidemias (Huang et al. 2011) has been associated with the Ala12 allele of PPARG. However, other studies suggest that association between the genotype and BMI may depend on obesity (Ek et al. 1999), age (Pihlajamaki et al. 2004), dietary uptake of monounsaturated fatty acids (Garaulet et al. 2011), and dietary ratio of polyunsaturated to saturated fatty acid (P/S ratio) (Luan et al. 2001). Interestingly, dietary P/S ratio has also been shown to have an interaction with PPARG genotype on the effect on serum lipids in cross-sectional studies (Alsaleh et al. 2011; Bouchard-Mercier et al. 2011), and also in an intervention trial in which saturated fatty acid was replaced by monounsaturated fatty acid (AlSaleh et al. 2012). Accordingly, animal studies suggest a complex interaction between PPARG2 and dietary fat. Mice lacking the PPARG2 isoform are more insulin-resistant compared to wild-type mice on normal chow diet but not on high-fat diet (Medina-Gomez et al. 2005). Furthermore, in Pro12Ala knock-in mice, the beneficial effects of the Ala12 allele on adiposity, plasma lipids, and insulin sensitivity are lost with a high-fat diet (Heikkinen et al. 2009).

In this study, we hypothesized that subjects with Pro12Pro and Ala12Ala genotypes respond differently to a diet supplemented with high saturated (SAFA) or polyunsaturated fatty acid (PUFA). This hypothesis was based on the knowledge that polyunsaturated fatty acids are ligands for PPARG (Kliewer et al. 1997; Xu et al. 1999). As opposed to previous studies investigating the putative modifying effect of the PPARG genotype in randomized dietary interventions independent of the genotype, we randomized 17 men with the Pro12Pro genotype and 14 matched controls with the Ala12Ala genotype, from a population-based METSIM study (Stancakova et al. 2012), to receive PUFA and SAFA diets for 8 weeks in a crossover setting. The specific objective of the study was to investigate whether SAFA and PUFA supplementation interferes with genetic regulation of lipid metabolism by Pro12Ala genotype.

Materials and methods

Subjects

We recruited lean and overweight subjects (BMI > 20 kg/m2 < 29 kg/m2) with the Pro12Pro and Ala12Ala genotypes of PPARG from the METSIM study (METabolic Syndrome in Men), including 10,197 men originally identified from the population register of the Kuopio town (Stancakova et al. 2012). Obese individuals were excluded to avoid the confounding effect of PPARG polymorphism on obesity and thus the secondary effects of obesity on serum lipids. One hundred and forty-seven non-diabetic subjects with the Ala12Ala genotype and an equal number of age- and BMI-matched subjects with the Pro12Pro genotype were selected from the METSIM study. A total of 31 subjects (17 with the Pro12Pro genotype and 14 with the Ala12Ala genotype) were willing to participate and were recruited. These participants did not differ from non-participants with respect to anthropometric and laboratory measurements (age 59.2 ± 6.0 vs. 59.0 ± 6.0 years, BMI 26.5 ± 2.9 vs. 27.2 ± 3.1 kg/m2, fasting glucose 6.0 ± 0.6 vs. 5.8 ± 0.7 mmol/l, and fasting insulin 49.1 ± 22.9 vs. 53.2 ± 25.6 pmol/l in participants and non-participants, respectively). There were no dropouts during the study.

Dietary interventions

Both the SAFA diet and the PUFA diet were isocaloric and contained 30 % of energy as fat, 18 % as protein, and 52 % as carbohydrates. The SAFA diet had a polyunsaturated/saturated (P/S) ratio of 0.3 and the PUFA diet 1.0. This was pursued in the SAFA diet by the consumption of medium-fat liquid dairy products, i.e., 1.5 % milk and sour milk, 2–2.5 % yoghurt, and fatty cheese (fat 24–30 %) and by using a butter and vegetable oil mixture as a spread and in cooking. Vegetable oil-based salad dressings were not allowed. Meals made of low-fat fish, e.g., perch and pike, were allowed once per week. In the PUFA diet, the subjects were instructed to use fat-free liquid dairy products (fat less than 1 %) and low-fat cheese (fat max. 17 %). Soft vegetable oil-based margarine was used as a spread and vegetable oil (low erucic acid rapeseed oil) and liquid margarine in cooking. Vegetable oil-based salad dressings were used in salads. Butter and butter-based products were not allowed during this period. Three fish meals per week (150 g of fish) with fatty fish (e.g., rainbow trout, salmon) were to be consumed. Low-fat fish was not recommended. In order to ensure compliance, the participants were given spreads, vegetable oils, liquid margarine, and cheeses free of charge during both periods. The extra costs for fish consumption during the PUFA period were reimbursed.

The two 8-week diet periods were started after a habitual diet and separated by a 2-week washout period with a habitual diet. To monitor the dietary intake, the subjects kept 4-day food records during four predefined consecutive days (Sunday–Wednesday or Wednesday–Saturday) in week 0 and twice during the diet periods (third and seventh week). The food records were checked at return by a clinical nutritionist. The physical activity was instructed to remain unchanged during the study.

The diets were planned, and the nutrient intake during the diet periods was calculated by Diet32 dietary analysis software (AivoFinland Ltd, Turku, Finland).

Clinical measurements and laboratory measurements

Body mass index (BMI) was calculated as weight (kg) divided by height (m) squared. Body composition was determined by bioelectrical impedance (RJL Systems) in subjects in the supine position after a 12-h fast. A 2-h OGTT (75 g of glucose) was performed, and samples for plasma glucose and insulin were drawn at 0, 30, and 120 min. Plasma glucose was measured by enzymatic hexokinase photometric assay (Konelab Systems Reagents, Thermo Fischer Scientific, Vantaa, Finland). Serum insulin was determined by immunoassay (ADVIA Centaur Insulin IRI, no 02230141, Siemens Medical Solutions Diagnostics, Tarrytown, NY). The InsAUC30/GluAUC30 ratio and Matsuda Insulin Sensitivity Index (ISI) (Matsuda and DeFronzo 1999) were used as surrogate markers of insulin secretion and insulin sensitivity, as previously reported (Stancakova et al. 2009).

Serum fatty acid composition

Serum lipids were extracted with chloroform–methanol (2:1), and the lipid classes were separated by solid-phase extraction with an aminopropyl column as a marker of dietary compliance (Agren et al. 1992; Vessby et al. 1980; Vidgren et al. 1997). Fatty acids were analyzed with Agilent GC 7890A gas chromatograph (Agilent Technologies Inc., Santa Clara, CA, Wilmington, DE) equipped with a 25-m FFAP column (I = 25 m, ID = 0.2 mm, film thickness 0.33 µm; Agilent Technologies Inc.).

Adipose tissue biopsies

After an overnight fast and 30-min resting, adipose tissue samples were taken as needle biopsies from subcutaneous abdominal adipose tissue at baseline, after the first dietary period (8 weeks), and at the end of the study (18 weeks) under local anesthesia (lidocaine 10 mg/ml without adrenaline). Samples were immediately frozen in liquid nitrogen.

Genotyping

PPARG2 polymorphism (rs1801282) was determined with the TaqMan Allelic Discrimination Assay (Applied Biosystems, Foster City, CA, USA). The TaqMan genotyping call rate was 100 %, with an error rate of 0 % in 4.5 % of DNA samples genotyped in duplicate.

mRNA expression

Total RNA was extracted using Tri-Reagent [Applied Biosystems (ABI) Foster City, CA, USA] and reverse transcribed using High Capacity cDNA Reverse Transcription Kit (ABI) according to the manufacturer’s protocol. Quantitative real-time PCR was carried out in the Applied Biosystems 7500 Real-Time PCR System. PPARG2, ADIPOR1, and ADIPOR2 expressions were determined by the TaqMan Gene Expression Assays (Applied Biosystems) (Hs01115510_m1, Hs00360422_m1, and Hs00226105_m1, respectively) according to the protocol. PPARG2, ADIPOR1, and ADIPOR2 expressions were normalized to RPLP0 expression (Hs99999902_m1). Expression of LPIN1, FASN, SREBP-1c was determined using SYBR Green (KAPA SYBR FAST qPCR Kit, Kapa Biosystems, Woburn, MA), and gene expression was normalized to RPLP0. Primer information is presented in Supplementary Table 1.

Statistical analysis

Variables with non-normal distribution were logarithmically transformed before statistical analysis. Differences between the genotype groups at baseline and after the dietary periods were examined with general linear model using the SPSS version 19 programs (SPSS, Chicago, IL, USA).

To adjust for possible confounding factors, mixed-model analyses were performed using linear mixed effect models (nlme package version 3.1-105) (Pinheiro et al. 2015). Models were fit using a restricted maximum likelihood method ignoring missing observations. For investigating the association of a particular genotype with selected phenotypes, we included type of diet, phase of the intervention period (start or end of the period), and the week of the study (0, 8, 10, or 18) as fixed effects. For investigating the association of diet with selected phenotypes, we examined diet–phase interaction, which includes type of diet, phase of the intervention period, and the week of the study as fixed effects. In all models, subject identifier was used as a random effect. We used Benjamini–Hochberg false discovery rate (FDR) to adjust for multiple testing (Benjamini and Hochberg 1995). All analyses were performed using R version 2.15.2.

Results

Thirty-one non-diabetic men (17 with the Pro12Pro genotype and 14 with the Ala12Ala genotype) were randomized to receive either the PUFA or SAFA diets for 8 weeks, separated by a 2-week washout period. No significant difference in clinical parameters indicating a carryover effect after the first diet was observed. Therefore, we combined the data for SAFA and PUFA diets from the first and second periods according to the original study design. The calculated intake of energy nutrients, dietary fiber, cholesterol, and sucrose from food records during the SAFA and PUFA diets is presented in Supplementary Table 2. Estimated P/S ratio was 0.25 during the SAFA diet and 0.63 during the PUFA diet. The intake of dietary cholesterol was higher in the SAFA diet due to the use of higher fat dairy products as compared with the PUFA diet, as instructed. The intake of dietary fiber and sucrose did not differ between the diets.

The results of serum fatty acid composition analysis indicated a good compliance of the study subjects with the diets (Supplementary Table 3). Levels of docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA) increased, as expected in agreement with previous studies (Karvonen et al. 2002; Vessby et al. 1980; Vidgren et al. 1997), both in cholesterol esters (by 32 and 108 %, respectively, P < 1 × 10−4) and in phospholipids (by 32 and 101 %, P < 0.001) during the PUFA diet. In contrast, levels of DHA and EPA decreased in cholesterol esters (by 14 and 16 %, P < 0.001) and in phospholipids (by 12 and 15 %, P < 0.001) during the SAFA diet. In addition, levels of α-linolenic acid (ALA) increased by 11 % in cholesterol esters during the PUFA diet (P = 2 × 10−5). Overall, FDRs for a dietary effect on DHA and EPA were <10−10, verifying that participants followed the instructed diets.

Effect of the Ala12Ala genotype on serum lipids and PPARG2 mRNA expression

As expected (Huang et al. 2011), individuals with the Ala12Ala genotype had higher levels of HDL cholesterol and lower levels of LDL cholesterol, total triglycerides, and apolipoprotein B at baseline (Table 1; Fig. 1a, P < 0.05, FDR < 0.01). Although we aimed to match individuals with the Pro12Pro and Ala12Ala genotypes for age and BMI, carriers with the Ala12Ala genotype were slightly older than carriers of the Pro12Pro genotype at baseline (P = 0.048, Table 1). No difference in glucose and insulin levels was observed between the genotypes. In Table 1, we also present the effects of the genotypes on serum lipids taking into account all time points during the study. In that analysis, total and LDL cholesterol, total triglycerides, and apolipoprotein were associated with the genotype. More specifically, total and LDL cholesterol, total triglycerides, and apolipoprotein were associated with the genotype at the end of the diets (P < 0.05).

Table 1 Clinical characteristics at baseline according to the PPARG genotype
Full size table
Fig. 1
figure 1

Levels of serum triglycerides (mmol/l) are lower in individuals with the Ala12Ala genotype compared to individuals with the Pro12Pro genotype of PPARG at baseline (a) and lower in response to PUFA diet. The difference between genotypes was observed also after PUFA and SAFA diets (P < 0.01, not shown) (b). Mean ± SD. ***P < 0.001, t test at baseline

Full size image

The Ala12Ala genotype was associated with higher adipose tissue expression of PPARG2 (Fig. 2a). Levels of adipose tissue PPARG2 mRNA expression were consistently higher in carriers of the Ala12Ala genotype than in carriers of the Pro12Pro genotype regardless of the diet (FDR = 0.001 for genotype effect in mixed-model analysis including all time points). In addition, expression of LPIN1 and SREBP-1c was higher in individuals with the Ala12Ala genotype (P = 0.009 and P = 0.032, respectively) (Fig. 2c, e). Expression of FASN did not associate with the genotype (P = 0.058) (Fig. 2g).

Fig. 2
figure 2

Association of adipose tissue mRNA expression of PPARG2 (a, b), LPIN1 (c, d), SREBP-1c (e, f), FASN (g, h) with the PPARG genotype and PUFA diet. Mean ± SD. *P < 0.05, **P < 0.01 t test

Full size image

PPARG2 mRNA expression correlated negatively with serum total triglycerides at all three time points (all P < 0.05, Table 2, all time points combined in Supplementary Figure 1). We could not observe a significant difference in the correlation between PPARG2 mRNA expression and serum triglycerides between the genotypes (z = 0.71, P = 0.477, Fisher r-to-z transformation test, Supplementary Figure 1). There was an inverse correlation of PPARG2 mRNA expression with total and LDL cholesterol at 8 weeks but not at baseline or at the end of the study. There was no correlation between adipose tissue PPARG2 mRNA expression and the levels of glucose and insulin (data not shown).

Table 2 Spearman correlation of the PPARG2 mRNA with serum lipids at different time points (n = 31 at each time point)
Full size table

PUFA diet improves lipid profile without affecting adipose tissue mRNA expression

Table 3 demonstrates a decrease in fasting glucose, total cholesterol, and total triglycerides in response to the PUFA diet (P < 0.05). In contrast, the SAFA diet led to a significant increase in total cholesterol and LDL cholesterol (P < 0.05). Importantly, at the end of the dietary periods, the levels of total cholesterol, LDL cholesterol, total triglycerides, and apolipoprotein B were higher after the SAFA diet compared to the PUFA diet (P < 0.05). To control for multiple testing and for the crossover setting, we also analyzed the data in a mixed model including diet, phase of the study, and the genotype. Diet had an effect on total cholesterol, LDL cholesterol, total triglycerides, and apolipoprotein B (FDR < 0.1). The PUFA diet did not affect mRNA expression of PPARG2 (Fig. 2b), LPIN1 (Fig. 2d), SREBP-1c (Fig. 2f), FASN (Fig. 2h), ADIPOR1, and ADIPOR2 (data not shown). There was no interaction between the genotype and diet.

Table 3 Clinical characteristics before and after the diet with high PUFA or the diet with high SAFA content
Full size table

Discussion

In this study, we used a unique recruit-by-genotype approach to study the effect the Pro12Pro and Ala12Ala genotypes of PPARG on serum lipids in a dietary intervention trial. Similar approach has earlier been used to demonstrate an effect of the PPARG genotype on adipose metabolism (Tan et al. 2006). We demonstrated that the Ala12Ala genotype associated with lower levels of serum lipids at the baseline and at the end of the PUFA and SAFA diets and higher expression of PPARG2 mRNA in adipose tissue compared to the Pro12Pro genotype, independent of the diet (Figs. 1, 2; Table 1). In contrast, the PUFA diet improved serum lipids without affecting PPARG2 expression in adipose tissue. These results suggest that dietary PUFAs and the PPARG genotype regulate serum triglycerides through divergent pathways. More specifically, our results suggest that individuals with Pro12Pro genotype, with higher triglyceride levels, are more likely to benefit from the PUFA diet (Fig. 1; Table 2).

Our hypothesis was that the effects of the PUFA diet on glucose and lipid metabolism could depend on the PPARG genotype. This hypothesis was based on the knowledge that polyunsaturated fatty acids, including derivatives of DHA and EPA, are ligands for PPARG (Kliewer et al. 1997; Xu et al. 1999). Therefore, we assumed that the modification of the fatty acid composition in the diet changes the ligands, i.e., PUFAs, for PPARG and thus modifies PPARG2 function depending on the genotype. Given the low frequency of the Ala12Ala genotype in the population (<1 %), we recruited the participants with either the Pro12Pro or the Ala12Ala genotypes of PPARG from a large population-based METSIM cohort including 10,197 participants (Stancakova et al. 2009, 2012). However, there was no effect of the PUFA diet on the expression of PPARG2, and we could also not observe any diet–genotype interaction effect on PPARG2 expression. In addition, we did not observe any significant interaction in the effects of the PPARG genotype and the PUFA diet on serum lipids, or on glucose metabolism, which could be due to our small sample size. The correlation analysis between PPARG2 mRNA expression and serum triglycerides suggested a difference between genotypes (Supplementary Figure 1). More importantly, the beneficial effect of PUFA diet on serum triglycerides was observed in individuals with Pro12Pro genotype, with higher triglyceride levels at baseline. Together with the earlier findings that dietary fat intake correlates with the components with the metabolic syndrome in individuals with the Pro12Pro genotype but not in those with Ala12Ala genotype (Robitaille et al. 2003), our results suggest that although dietary effect on serum lipids is not mediated through PPARG2 mRNA expression in adipose tissue, PPARG genotype may still modify the effects of dietary fat on serum lipids.

The Ala12Ala genotype did associate with higher mRNA expression of PPARG2 which correlated with serum triglyceride levels. These results suggest that the effects of the PPARG genotype on serum lipid levels are mediated through altered PPARG2 expression in adipose tissue (Deeb et al. 1998), while the effect of the PUFA diet on serum triglyceride levels is independent of the PPARG genotype and PPARG2 expression in adipose tissue. The mechanisms how the Ala12 allele, which leads to lower transcriptional activity (Deeb et al. 1998) and associates with higher mRNA expression (Kolehmainen et al. 2003), contributes to lower levels of serum triglycerides remain unknown. However, these mechanisms may include secondary responses in adipose tissue, such as higher expression of LPIN1 and SREBP-1c and potential other targets. Secondary effects also outside the subcutaneous adipose tissue, e.g., in the visceral depot and in the liver, are possible, but cannot be investigated in a dietary intervention. We acknowledge that PCR-based gene expression analysis is always affected by the selection of endogenous control genes, and that the use of other control genes than RPLP0 could have slightly modified the results.

In summary, our study suggests that the beneficial effects of dietary PUFA and Ala12Ala genotype of PPARG on serum lipids are mediated through divergent pathways. Ultimately, larger randomized intervention trials based on genetic background will be needed to fully evaluate the significance of interactions between Pro12Ala polymorphism and environmental factors.