In the cross-sectional study we analysed data from a total of 502 people of European extraction who had NGT or IGT. The individuals were recruited from the southern part of Germany and participated in the ongoing Tübingen Family Study for type 2 diabetes. The participants were not taking any medication known to affect glucose tolerance or insulin sensitivity. They were considered healthy according to a physical examination and routine laboratory tests. All subjects first underwent an OGTT. When they were found not to have diabetes they underwent further tests. Using self-reported data, only 1.2% of the subjects were related and 72% had a family history of type 2 diabetes. A subgroup of 295 individuals had measurements of insulin sensitivity obtained during a euglycaemic–hyperinsulinaemic clamp.
In the longitudinal study, a group of 45 subjects underwent dietary counselling and increased their physical activity to decrease adiposity and to prevent type 2 diabetes. Data from these individuals were also included in the cross-sectional study. The individuals had a family history of type 2 diabetes and/or a BMI greater than 27 kg/m2. These subjects underwent baseline measurements and had regular visits to our clinic thereafter. The dietary counselling aimed to reduce the total intake of calories, particularly the intake of calories from fat. Subjects were instructed to increase physical activity and to perform at least 3 h of sports per week. The participants did not take any medication known to affect glucose tolerance or insulin sensitivity. Tests were performed at 07.00 h after an overnight fast of 12 h. Informed written consent was obtained from all participants and the local medical ethics committee had approved the studies.
Body composition and body fat distribution
Body composition was measured by bioelectrical impedance as the percentage of body fat. BMI was calculated as weight divided by the square of height (kg/m2). Waist and hip circumferences were measured in the supine position, and WHR was calculated as an index of body fat distribution.
All subjects underwent a 75 g OGTT and venous blood samples were obtained at 0, 30, 60, 90 and 120 min for determination of plasma glucose, insulin and C-peptide. Glucose tolerance was determined according to the 1997 World Health Organization diagnostic criteria . Insulin sensitivity was calculated from glucose and insulin values during the OGTT, as proposed by Matsuda and DeFronzo . The 30-min C-peptide plasma concentrations during the OGTT and the first phase of insulin secretion determined from the OGTT (1283+1.829 Ins30−138.7 Glu30+3.772 Ins0), as described earlier , were used as an estimate of beta-cell function.
After a 12-h overnight fast, an antecubital vein was cannulated for infusion of insulin and glucose. A dorsal hand vein of the contralateral arm was cannulated and placed under a heating device to permit sampling of arterialised blood. Subjects received a primed insulin infusion at the rate of 1.0 mU kg−1 min−1 for 2 h. Blood was drawn every 5 min for determination of blood glucose, and a glucose infusion was adjusted appropriately to maintain the fasting glucose level. An insulin sensitivity index (ISI; in μmol kg−1 min−1 [pmol/l]−1) for systemic glucose uptake was calculated as the mean infusion rate of glucose (in μmol kg−1 min−1) necessary to maintain euglycaemia during the last 60 min of the euglycaemic–hyperinsulinaemic clamp divided by the steady-state plasma insulin concentration.
Intrahepatic and intramyocellular lipid content
Liver fat was determined by localised proton magnetic resonance spectroscopy using a 1.5 T whole-body imager (Magnetom Sonata, Siemens Medical Solutions, Erlangen, Germany). For volume selection, a single-voxel stimulated echo acquisition mode (STEAM) technique was applied (repetition time [TR]=4 s, echo time [TE]=10 ms, 32 acquisitions) and a voxel of 3×3×2 cm3 was placed in the posterior part of the seventh segment of the liver. Subjects were asked to breathe within the TR interval and to be in expiration during data acquisition. The liver fat was assessed quantitatively by analysing the signal integrals of methylene and methyl resonances (between 0.7 and 1.5 ppm), using the liver water signal integral at 4.8 ppm as internal reference.
Intramyocellular lipid content (IMCL) and lipid content interlaced between the muscle fibres (extramyocellular lipid content, EMCL) was differentiated because of their geometrical arrangement using proton magnetic resonance spectroscopy as previously described . In brief, localised image-guided proton spectra of the tibialis anterior muscle representing a muscle of mixed type I and II fibres and of the soleus muscle representing a muscle of predominantly type I fibres with high oxidative capacity were acquired on a 1.5 T whole-body imager (Magnetom Vision; Siemens Medical Solutions, Erlangen, Germany). For volume selection, a single-voxel STEAM technique was applied. Measurement parameters were TE=10 ms, TR=2 s, volume of interest 11×11×20 mm3, 40 acquisitions. IMCL and EMCL were quantified on the separation of two resonances in the lipid–CH2-region.
Blood glucose was determined using a bedside glucose analyser (glucose-oxidase method; Yellow Springs Instruments, Yellow Springs, CO, USA). Plasma insulin was determined by microparticle enzyme immunoassay (Abbott Laboratories, Tokyo, Japan) and plasma C-peptide by radioimmunoassay (Byk-Sangtec, Dietzenbach, Germany). Serum samples were frozen immediately and stored at −20°C for determination of adiponectin by radioimmunoassay (LINCO Research, St Charles, MO, USA) and free fatty acids with an enzymatic method (Wako Chemicals, Neuss, Germany). Plasma TNF-α and IL-6 concentrations were determined by enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, MN, USA).
Screening of ADIPOR1/ADIPOR2 and genotyping of single-nucleotide polymorphisms
Human ADIPOR1 is composed of eight exons spanning 17.5 kb on chromosome 1p36.13-q41. Human ADIPOR2 maps to chromosome 12p13.31, comprises eight exons and spans 97.6 kb. Sequencing was done according to the sequences NT_004671.15 (ADIPOR1) and NT_009759.15 (ADIPOR2) in the NCBI database (http://www.ncbi.nih.gov/entrez/query.fcgi?db=gene, last accessed in August 2005). We performed 14 PCR amplifications of ADIPOR1 and 13 of ADIPOR2 with a series of specific primer pairs using DNA extracted from blood leucocytes in 50 subjects. For the two genes, primer sequences cover 2 kb of the proximal promoter, all exons, exon–intron boundaries (100–200 bp of the flanking intronic region), including the 5′ untranslated region and part of the 3′ untranslated region. The sequence of ADIPOR1 in the NCBI database was updated towards the end of 2004, resulting in reorganisation of the exon order. Since we started our sequencing before this, we initially considered a large region upstream of exon 2 as a promoter and, therefore, additionally sequenced intron 1.
The polymorphisms were localised by direct sequencing. PCR products were sequenced bidirectionally, using an ABI Prism dye terminator cycle sequencing ready reaction kit (Applied Biosystems, Foster City, CA, USA), and analysed on an automated sequencer (ABI model 310). Polymorphisms in selected DNA samples were genotyped using the TaqMan assay (Applied Biosystems). The TaqMan genotyping reaction was amplified on a GeneAmp PCR system 7000 (50°C for 10 min, 95°C for 10 min, 92°C for 15 s, 60°C for 1 min for 35 cycles) and fluorescence was detected on an ABI Prism 7000 sequence detector (Applied Biosystems). As a quality standard, we randomly included six positive and two negative (all components excluding DNA) sequenced controls in each TaqMan reader assay. All controls were correctly identified.
Unless otherwise stated, data are given as mean±SE. Except for insulin secretion (analysed only in individuals with normal glucose tolerance [NGT]), subjects with NGT and impaired glucose tolerance (IGT) were analysed together. Hardy–Weinberg equilibrium was tested with the χ
2 test. Statistical comparison was performed using logarithmically transformed data (for non-normally distributed parameters). Differences in anthropometrics and metabolic characteristics between genotypes were tested using general linear regression models. In these models the trait was the dependent variable, whereas age, sex and genotype, for example, were the independent variables. An additive model was used in all cross-sectional analyses for all SNPs. In the longitudinal study, differences between measurements at baseline and follow-up were tested with the two-tailed Student’s t-test. Because of the small number of subjects who had follow-up data, subjects who were homozygous for the mutant alleles (e.g. n=3 for the −8503 G/A polymorphism) and heterozygotes were combined for analyses. In general linear regression models, insulin sensitivity and liver fat at follow-up were the dependent variables and insulin sensitivity and liver fat at baseline, sex and genotype were the independent variables. Pairwise linkage disequilibrium (D′, r
2) and haplotypes were determined using the THESIAS program  and the Java linkage disequilibrium plotter (http://www.genepi.com.au/projects/jlin, last accessed in August 2005). THESIAS performs haplotype-based association analyses in unrelated individuals with phase-unknown genotypes. It is based on a maximum likelihood model and is linked to the SEM algorithm. This allows the simultaneous estimation of haplotype frequencies and of their associated effects on the phenotype of interest. Both quantitative and qualitative phenotypes can be studied. Using this approach the differences in the phenotypes of the most common (frequency >0.02) haplotypes in respect to the so-called wild-type haplotype (this haplotype consisted only of the major alleles) was determined.
Because of the relatively small number of subjects in the longitudinal study, haplotype analyses were performed only in the cross-sectional study. A p value below 0.05 was considered statistical significant. The statistical software package JMP (SAS Institute, Cary, NC, USA) or SPSS version 10.0 software (SPSS, Chicago, IL, USA) was used.