The enzyme DGAT (acyl-CoA:diacylglycerol acyltransferase) catalyses the final step of triacylglycerol (triglyceride) synthesis. Mice overexpressing hepatic DGAT2 fed a high-fat diet develop fatty liver, but not insulin resistance, suggesting that DGAT2 induces a dissociation between fatty liver and insulin resistance. In the present study, we investigated whether such a phenotype also exists in humans. For this purpose, we determined the relationships between genetic variability in the DGAT2 gene with changes in liver fat and insulin sensitivity in 187 extensively phenotyped subjects during a lifestyle intervention programme with diet modification and an increase in physical activity. Changes in body fat composition [MR (magnetic resonance) tomography], liver fat and intramyocellular fat (1H-MR spectroscopy) and insulin sensitivity [OGTT (oral glucose tolerance test) and euglycaemic–hyperinsulinaemic clamp] were determined after 9 months of intervention. A change in insulin sensitivity correlated inversely with changes in total body fat, visceral fat, intramyocellular fat and liver fat (OGTT, all P<0.05; clamp, all P≤0.03). Changes in total body fat, visceral fat and intramyocellular fat were not different between the genotypes of the SNPs (single nucleotide polymorphisms) rs10899116 C>T and rs1944438 C>T (all P≥0.39) of the DGAT2 gene. However, individuals carrying two or one copies of the minor T allele of SNP rs1944438 had a smaller decrease in liver fat (−17±10 and −24±5%; values are means±S.E.M.) compared with subjects homozygous for the C allele (−39±7%; P=0.008). In contrast, changes in insulin sensitivity were not different among the genotypes (OGTT, P=0.76; clamp, P=0.53). In conclusion, our findings suggest that DGAT2 mediates the dissociation between fatty liver and insulin resistance in humans. This finding may be important in the prevention and treatment of insulin resistance and Type 2 diabetes in subjects with fatty liver.
- acyl-CoA:diacylglycerol acyltransferase (DGAT)
- fatty liver
- insulin resistance
- lifestyle intervention
NAFLD (non-alcoholic fatty liver disease) represents a strong predictor of insulin resistance, the metabolic syndrome, cardiovascular disease and Type 2 diabetes [1–5]. However, despite the strong association of fatty liver with insulin resistance [6–8], subjects can be identified who, for the same amount of liver fat, have very high and low insulin resistance. When accumulation of liver fat is not accompanied by a respective increase in insulin resistance, the term ‘dissociation’ is used [9,10]. Recent animal findings suggest that a possible explanation for this dissociation between fat accumulation in the liver and insulin resistance may be variation in the expression and/or activity of the enzyme DGAT (acyl-CoA:diacylglycerol acyltransferase) , the enzyme catalysing the final step of triacylglycerol (triglyceride) synthesis [11,12]. There are two isoforms of the enzyme namely DGAT1 and DGAT2. DGAT2 is specific for triacylglycerol synthesis and is predominantly expressed in the liver . Mice overexpressing DGAT2 develop fatty liver but, unexpectedly, not insulin resistance when fed a high-fat diet . In agreement, treatment with antisense oligonucleotide causing a reduction in the expression of Dgat2 resulted in an improvement in hepatic steatosis without a change in insulin sensitivity . In another study of similar design in rats, insulin sensitivity improved as well, but this may have been due to the observed reduction in body weight .
In the present study, we set out to delineate whether DGAT2 plays a role in the dissociation of fatty liver and insulin resistance in humans. We studied the relationships between common SNPs (single nucleotide polymorphisms) in DGAT2 with liver fat and insulin sensitivity during a lifestyle intervention programme. Precise phenotyping methods, such as MRT [MR (magnetic resonance) tomography] to measure total body fat and fat content in the visceral as well as in the subcutaneous depots, and 1H-MR spectroscopy to determine fat deposition in ectopic tissues, including liver and skeletal muscle, were used.
MATERIALS AND METHODS
Data from 187 Caucasians (72 men and 115 women) without Type 2 diabetes from the southern part of Germany were analysed. These individuals participated in an ongoing study to reduce adiposity and prevent Type 2 diabetes [15,16]. Individuals were included in the study when they fulfilled at least one of the following criteria: a family history of Type 2 diabetes, a BMI (body mass index) >27 kg/m2, a previous diagnosis of IGT (impaired glucose tolerance) and/or of gestational diabetes. Subjects were considered healthy according to a physical examination and routine laboratory tests, had no history of liver disease and did not consume more than two alcoholic drinks per day. Measurements were taken at baseline and after a mean of 9 months of the lifestyle intervention. After the baseline measurements have been taken, individuals underwent dietary counselling and had up to ten sessions with a dietician. During each visit, participants presented a 3-day food diary and discussed the results with the dieticians. Diet composition was determined with a validated computer program using two representative days of the 3-day diary (DGE-PC 3.0; Deutsche Gesellschaft für Ernährung). Counselling was aimed to reduce body weight by ≥5%, to reduce the intake of calories from fat to <30% and, particularly, the intake of saturated fat to ≤10% of energy consumed, and to increase the intake of fibre to at least 15 g/4.184 kJ. Individuals were asked to perform at least 3 h of moderate exercise per week. Aerobic endurance exercise (e.g. walking and swimming) with an only moderate increase in heart rate was encouraged. Participants were seen by the staff on a regular basis to ensure that these recommendations were accomplished.
The research has been carried out in accordance with the Declaration of Helsinki (2000) of the World Medical Association. Informed written consent was obtained from all of the participants, and the local medical ethics committee approved the protocols.
Body fat distribution, liver fat and intramyocellular fat
Waist circumference was measured at the midpoint between the lateral iliac crest and lowest rib. Total body and visceral fat were measured by MRT, and liver fat and intramyocellular fat of the tibialis anterior muscle by 1H-MR spectroscopy, as described previously .
OGTT (oral glucose tolerance test)
All individuals underwent a 75 g OGTT. Venous plasma samples were obtained at 0, 30, 60, 90 and 120 min for determination of plasma glucose and insulin. Glucose tolerance was determined according to the 1997 World Health Organization diagnostic criteria .
Insulin sensitivity from the OGTT was calculated as proposed by Matsuda and DeFronzo  (10000/√(mean insulin×mean glucose)×(fasting insulin×fasting glucose).
A subgroup (n=142 at baseline and n=46 at follow-up) of subjects additionally underwent a euglycaemic–hyperinsulinaemic clamp (subsequently referred to as clamp). In the smaller group of subjects in our analyses who underwent the clamp (n=46), we had a power of 81% in the additive model to detect an effect of the DGAT genotype on the change in liver fat. The lowest number of subjects to detect an effect at an α level of 0.05 was 30. In the dominant model, the power was 89% and the lowest number of subjects to detect an effect at an α level of 0.05 was 20. In this group, we had a power of 98% to detect an effect size of 20% in the change in insulin sensitivity at the α level of 0.05 in the dominant model and of 97% in the additive model. Insulin sensitivity was determined with a primed insulin infusion at a rate of 40 milli-units·m−2·min−1 for 2 h as described previously . In these subgroups of individuals, insulin sensitivity calculated from the OGTT was strongly correlated with insulin sensitivity measured during the clamp (r=0.72, P<0.0001 at baseline; and r=0.78, P<0.0001 at follow-up).
SNPs that are representative in a region of the genome with high linkage disequilibrium (the non-random association of alleles at two or more loci-tagging SNPs) allow the identification of genetic variation without genotyping every SNP in a chromosomal region. Tagging SNPs for DGAT2 were selected from HapMap (HapMan Public Release number 22). Applying a pairwise tagging SNP r2 threshold of 0.8, two SNPs (rs10899116 and rs1944438) had a MAF (minor allele frequency) ≥0.05 and covered 100% of the alleles with a D′ of 1 (Haploview Program). Thus these two SNPs are representative of all SNPs in DGAT2 that have a relevant frequency in the general population. These SNPs were genotyped in our population by direct sequencing and using the TaqMan® assay (Applied Biosystems).
For statistical analyses, non-normally distributed parameters were logarithmically transformed. Relationships between parameters at baseline and follow-up were tested using a paired Student's t test. To test the effect of genotype on the metabolically relevant parameters, multivariate linear regression models were used with the parameter to be considered set as the dependent variable. In cross-sectional analyses, differences in parameters at baseline between genotypes were tested. Body weight, waist circumference and total body fat were adjusted for age and gender. The other parameters were additionally adjusted for total body fat. In longitudinal analyses, changes in parameters during the intervention between genotypes were tested. Changes in body weight, waist circumference and total body fat were adjusted for the respective baseline values, gender and age at follow-up. Changes in the other parameters were additionally adjusted for body fat at baseline and follow-up. The genotype was included in both analyses as an independent nominal variable. For each dependent variable, two models were applied. In the additive model, the effects of all possible genotypes on the dependent variable were compared; in the dominant model, homozygotes for the major allele were compared with heterozygotes and homozygotes of the minor allele. A P value ≤0.05 was considered statistically significant. The statistical software package JMP 4.0 (SAS Institute) was used.
The anthropometrics and metabolic characteristics of the subjects at baseline are shown in Table 1. All of the parameters at baseline covered a wide range. A total of 71 subjects (38%) were obese (BMI >30 kg/m2), and 49 (26.2%) had IGT at baseline. According to the cut-off of liver fat at 5.56% , 54 individuals were found to have fatty liver at baseline.
Liver fat was higher in males than females [4.28 (2.48–11.19) compared with 2.26 (1.17–4.46)%, P<0.0001; values are medians (interquartile range)] and correlated significantly with age (r=0.15, P=0.04), total body fat (r=0.28, P=0.0001), visceral fat (r=0.62, P<0.0001), intramyocellular fat (r=0.18, P=0.02), and inversely correlated with insulin sensitivity (OGTT, r=−0.54, P<0.0001; clamp, r=−0.58, P<0.0001).
Insulin sensitivity was higher in females than males [OGTT, 12.97 (8.47–18.04) compared with 9.37 (6.81–14.12) arbitrary units, P=0.02; clamp, 0.066 (0.047–0.090) compared with 0.054 (0.037–0.071) μmol·kg−1 of body weight·min−1·(pmol/l)−1, P=0.009] and inversely correlated with total body fat (OGTT, r=−0.39, P<0.0001; clamp, r=−0.39, P<0.0001), visceral fat (OGTT, r=−0.47, P<0.0001; clamp, r=−0.51, P<0.0001) and intramyocellular fat (OGTT, r=−0.28, P=0.0004; clamp, r=−0.27, P=0.002).
Both SNPs rs10899116 and rs1944438 were in Hardy–Weinberg equilibrium (χ2 test, P≥0.75). The MAF of SNPs rs10899116 C>T and rs1944438 C>T were 0.09 and 0.39 respectively. SNP rs10899116 was not associated with anthropometric or metabolic characteristics. In a dominant model, SNP rs1944438 was not associated with age or gender. No relationship of this SNP with insulin sensitivity, visceral fat, liver fat or intramyocellular fat, all adjusted for age, gender and total body fat, were observed (all P≥0.13; Table 2). Carriers of the minor T allele of SNP rs1944438 had higher total body fat, adjusted for age and gender, compared with homozygous carriers of the C allele (P=0.04).
During the 9 months of the intervention, there were significant mean decreases in body weight (−3.1%), total body fat (−12.1%), visceral fat (−18.7%) and liver fat (−29.1%). Over baseline, significantly fewer subjects were still obese (59 subjects; 31.6%), had IGT (40 subjects, 21.4%) and had fatty liver (39 subjects, 20.9%) at follow-up (all P<0.0001). Insulin sensitivity estimated from the OGTT increased by 15.5%, and insulin sensitivity measured by the clamp increased by 11% (Table 1). The change in insulin sensitivity correlated inversely with changes in total body fat, visceral fat and intramyocellular fat (OGTT, all P≤0.046; clamp, all P≤0.02). Furthermore, changes in insulin sensitivity correlated inversely with the change in liver fat (OGTT, r=−0.32, P<0.0001; clamp, r=−0.32, P=0.03).
SNP rs10899116 was not associated with any changes in the parameters tested (all P≥0.38). SNP rs1944438 was also not associated with changes in total body fat (Figure 1A), visceral fat or intramyocellular fat (all P≥0.39; Table 2). However, subjects carrying the minor T allele had a smaller decrease in liver fat, independently of age, gender, liver fat at baseline and total body fat at baseline and at follow-up, compared with homozygous carriers of the C allele (Table 3, and Figure 1B). In contrast, the change in adjusted insulin sensitivity calculated from the OGTT was not associated with the SNP (Table 3 and Figure 1C). Even in the small subgroup of individuals with measurements of insulin sensitivity by the clamp (n=46), similar results were found for the relationships between SNP rs1944438 and changes in liver fat (P=0.0021) and insulin sensitivity (P=0.53).
In the group of subjects who had fatty liver at baseline, the results were analogous to the larger group. Subjects with fatty liver carrying the minor T allele of SNP rs1944438 had a smaller decrease in adjusted liver fat compared with homozygous carriers of the C allele (P=0.02). No relationship was found between the SNP and the change in insulin sensitivity calculated from the OGTT (P=0.52) in these individuals.
In agreement with animal data [9,13], in the present study we found that dissociation between fat accumulation in the liver and insulin resistance mediated by DGAT2 may also exist in humans. During the lifestyle intervention programme, subjects carrying the minor T allele of SNP rs1944438 had a smaller decrease in liver fat compared with homozygous carriers of the C allele. Importantly, changes in total body fat, visceral fat and intramyocellular fat did not differ between the genotypes, which is in agreement with a predominant effect of DGAT2 on hepatic steatosis . A recent study, which applied a mutation screen in DGAT2, also failed to show an important role of the common genetic variation in the gene for the development of obesity . Furthermore, SNP rs1944438 of DGAT2 was not associated with a change in insulin sensitivity, a finding that is in line with the observations in mice [9,13]. Interestingly, the relationships between the SNP and changes in liver fat and insulin sensitivity were also observed in the subgroup of participants in the present study who underwent the clamp (n=46), supporting that a type 1 error can, most probably, be excluded.
The dissociation between the change in liver fat and insulin sensitivity, which was attributable to the SNP, is apparent in subjects with fatty liver who are at very high risk of metabolic diseases [2,8]. This finding is important particularly as, in the treatment of fatty liver, moderate lifestyle intervention, as it was used in the present study, is considered the first line of therapy [22,23]. It, therefore, underlines the relevance of considering DGAT2 activity during a lifestyle intervention programme to reduce hepatic steatosis and increase insulin sensitivity. In addition, it remains to be established whether DGAT2 is also important in determining the metabolic effect of pharmacological intervention, e.g. with thiazolidinediones , in the treatment of fatty liver.
Mechanisms involved in this dissociation of fatty liver and insulin resistance are not fully understood. As suggested by animal studies [9,13,14], altered amounts of fatty acid metabolites regulating insulin sensitivity, such as diacylglycerol [25,26], long-chain fatty acyl-CoAs and ceramides , may be important in this aspect. Furthermore, unsaturated long-chain fatty acyl-CoAs were found to have a protective function against saturated acyl-CoA-induced lipotoxicity and cellular apoptosis by promoting triacylglycerol accumulation [14,27]. DGAT has the same effect, catalysing triacylglycerol synthesis, but whether SNP rs1944438 of DGAT2 is also associated with such a protective effect in liver in humans needs to be determined.
In conclusion, our findings support the notion that dissociation between fatty liver and insulin resistance exists in humans. As the reduction in liver fat during a lifestyle intervention in subjects with fatty liver did not always result in a beneficial effect on insulin sensitivity, this finding may be of clinical relevance in the prevention and treatment of Type 2 diabetes. Genetic screening for this polymorphism prior to the intervention may help to apply early alternative strategies to improve insulin sensitivity.
The work was supported by the Deutsche Forschungsgemeinschaft [grant number KFO 114]; and the European Community's FP6 EUGENE6 [grant number LSHM-CT-2004-512013]. N.S. is currently supported by a Heisenberg grant from the Deutsche Forschungsgemeinschaft [grant number STE-1096/1-1].
Abbreviations: BMI, body mass index; DGAT, acyl-CoA:diacylglycerol acyltransferase; IGT, impaired glucose tolerance; MAF, minor allele frequency; MR, magnetic resonance; MRT, MR tomography; OGTT, oral glucose tolerance test; SNP, single nucleotide polymorphism
- © The Authors Journal compilation © 2009 Biochemical Society