Upper- and lower-body fat depots exhibit opposing associations with obesity-related metabolic disease. We defined the relationship between DEXA-quantified fat depots and diabetes/cardiovascular risk factors in a healthy population-based cohort (n = 3,399). Gynoid fat mass correlated negatively with insulin resistance after total fat mass adjustment, whereas the opposite was seen for abdominal fat. Paired transcriptomic analysis of gluteal subcutaneous adipose tissue (GSAT) and abdominal subcutaneous adipose tissue (ASAT) was performed across the BMI spectrum (n = 49; 21.4–45.5 kg/m2). In both depots, energy-generating metabolic genes were negatively associated and inflammatory genes were positively associated with obesity. However, associations were significantly weaker in GSAT. At the systemic level, arteriovenous release of the proinflammatory cytokine interleukin-6 (n = 34) was lower from GSAT than ASAT. Isolated preadipocytes retained a depot-specific transcriptional “memory” of embryonic developmental genes and exhibited differential promoter DNA methylation of selected genes (HOTAIR, TBX5) between GSAT and ASAT. Short hairpin RNA–mediated silencing identified TBX5 as a regulator of preadipocyte proliferation and adipogenic differentiation in ASAT. In conclusion, intrinsic differences in the expression of developmental genes in regional adipocytes provide a mechanistic basis for diversity in adipose tissue (AT) function. The less inflammatory nature of lower-body AT offers insight into the opposing metabolic disease risk associations between upper- and lower-body obesity.

Lower-body fat accumulation, as opposed to central obesity, is inversely associated with metabolic risk factors, including hyperinsulinemia, dyslipidemia, and hypertension (1), and is also associated with a reduced incidence of type 2 diabetes mellitus (T2DM) and cardiovascular disease (CVD) (2,3). Conversely, loss of lower-body fat during weight reduction relates to adverse changes in blood lipid and glucose profiles as well as blood pressure (4). Most of these associations have been ascertained using conventional anthropometrics with limitations in providing accurate measurements of regional fat mass. However, a need to use more refined imaging techniques to accurately quantify fat distribution on a large scale has been identified (5).

The epidemiological and clinical associations between lower- and upper-body fat distribution raise the question of whether gluteofemoral adipose tissue (AT) is functionally different from abdominal AT. The uptake of meal-derived fatty acids occurs less efficiently in lower-body AT compared with upper-body AT (6). This is balanced by less fatty acid release from lower-body AT in the postabsorptive state (7,8) or during adrenergic lipolytic stimuli (9). However, fatty acids presented to femoral AT by VLDL seem to be taken up more avidly (10). We have also demonstrated that release of the insulin-sensitizing lipokine palmitoleate is greater from lower- than upper-body AT (11). These physiological observations suggest there are fundamental differences in lipid handling that allow lower-body AT to act as a metabolic sink and protect insulin-sensitive tissues from excessive fatty acid exposure.

Regional variation in AT function may reflect the presence of distinct preadipocyte populations. Supporting this view is the finding that isolated preadipocytes exhibit intrinsic differences in many functional characteristics, including proliferation, apoptosis, and adipogenesis (12). Transcriptional profiling has also identified unique regional mRNA profiles characterized by differential expression of developmental genes (1315), some of which correlate with obesity (HOXA5, TBX15) (16) and appear to be important regulators of adipocyte development and function (SHOX2, TBX15) (17,18). The “memory” of these transcriptional signatures can persist in vitro across many cell generations (13), implicating the involvement of epigenetic regulation in regional AT development (19).

Perturbed lipid storage (20), enhanced inflammation (21,22), and altered adipokine signaling (23) have all been proposed as mechanisms for obesity-related metabolic complications. The cellular or mechanistic basis for these pathways may involve a reduced capacity to recruit and renew adipocytes, thus limiting AT expandability (24), induction of innate immune mechanisms, or the activation of cellular stress pathways (25). Chronic AT inflammation (21,22) coincides with the silencing of lipid metabolism genes (14,20). However, the majority of these findings come from studies performed on upper-body AT (visceral, subcutaneous), and it is unclear whether these mechanisms also apply to gluteofemoral AT.

We have previously applied transcriptional profiling and expression quantitative trait loci analyses of gluteal subcutaneous AT (GSAT) and abdominal subcutaneous AT (ASAT) to identify genetic polymorphisms that correlate with gene expression networks and metabolic syndrome–related phenotypes (26). We also identified genetic loci influencing fat distribution and explored regional expression of associated genes (RSPO3, TBX15, ITPR2) in ASAT and GSAT (27). However, a direct gene-by-gene comparison of transcriptional profiles between GSAT and ASAT was not performed. Given the apparent beneficial nature of gluteofemoral AT, we hypothesized that this tissue may, to some extent, resist the proinflammatory and hypometabolic changes associated with obesity otherwise found in abdominal AT. To investigate this, we first defined the relationship between DEXA-quantified regional fat masses and metabolic risk factors in a large healthy population-based cohort. We then selected a subset of AT samples from the cohort previously used (26), comparing transcriptional profiles of individual pairs of GSAT and ASAT. Supporting functional information was added in terms of human preadipocyte studies and in vivo arteriovenous adipokine measurements.

Recruitment and Sample Collection

All studies received ethical approval from Oxfordshire Clinical Research Ethics Committee (08/H0606/107), and all subjects gave written informed consent.

Oxford Biobank

At the time of this study the Oxford Biobank (OBB; http://www.oxfordbiobank.org.uk/) contained basic anthropometric and biochemical data for 5,828 subjects and DEXA data for 3,399 subjects. Subjects were Caucasian, aged 29–54 years at recruitment, and considered healthy (defined by the absence of CVD and T2DM) (28).

MolOBB Cohort (Microarray Analysis)

International Diabetes Federation metabolic syndrome criteria were applied to recruit subjects from the OBB into four groups (male/female, metabolic syndrome+/−), thus enriching for obesity-associated cardiovascular and diabetes risk factors. Paired needle biopsies were obtained from ASAT (from the abdominal wall at the level of the umbilicus) and GSAT (from the upper outer quadrant of the buttock) from 70 subjects (40 males) from the OBB. BMI ranged from 21.4 to 45.5 kg/m2.

Preadipocyte Studies

Paired ASAT and GSAT biopsies were obtained from five healthy subjects (four males) recruited from the OBB, aged 36–49 years with BMI ranging from 23.2 to 37.9 kg/m2.

Arteriovenous Study

Depot-specific adipokine release was measured in 42 age- and BMI-matched subjects (23 males) recruited from the OBB. Anthropometric and biochemical characteristics for the MolOBB and arteriovenous cohorts are presented in Table 1.

Table 1

Anthropometric and clinical characteristics of study participants

MaleFemale
MolOBB Affymetrix study participants   
 n 31 18 
 Age, years 48 (40–56) 50 (39–56) 
 BMI, kg/m2 28.2 (21.4–40.5) 30.2 (22.6–45.5) 
 Waist, cm 98.2 (79.0–134.0) 95.4 (75.0–131.0) 
 Hip, cm 106.0 (94.0–126.0) 110.0 (93.0–141.0) 
 WHR 0.92 (0.82–1.08) 0.86 (0.77–1.05) 
 Fasting glucose, mmol/L 5.4 (4.7–7.1) 5.4 (4.6–8.6) 
 Fasting insulin, mU/L 13.8 (5.8–53.3) 12.3 (4.4–31.8) 
Arteriovenous study participants   
 n 23 19 
 Age, years 41 (22–59) 41 (23–57) 
 BMI, kg/m2 25.4 (20.8–30.6) 25.4 (21.0–39.3) 
 WHR 0.92 (0.79–1.05) 0.84 (0.72–0.97) 
 Abdominal fat mass, kg 10.6 (4.5–17.0) 12.6 (4.0–25.5) 
 Leg fat mass, kg 5.8 (3.5–8.7) 8.6 (4.7–15.3) 
 Fasting glucose, mmol/L 5.4 (4.8–6.6) 5.0 (4.4–5.6) 
 Fasting insulin, mU/L 11.6 (7.2–20.8) 11.7 (3.8–27.3) 
 Fasting nonesterified fatty acids, µmol/L 513 (300–1,084) 659 (297–1,012) 
 Serum IL-6, pg/mL 1.6 (0.26–9.14) 0.96 (0.31–2.69) 
 Serum MCP-1, pg/mL 7.67 (4.79–14.76) 6.86 (2.81–10.93) 
 Serum leptin, pg/mL 991 (69–5,375) 3,481 (333–18,000) 
 Abdominal ATBF 3.4 (0.3) 3.5 (0.4) 
 Gluteofemoral ATBF 2.5 (0.3) 3.4 (0.4) 
MaleFemale
MolOBB Affymetrix study participants   
 n 31 18 
 Age, years 48 (40–56) 50 (39–56) 
 BMI, kg/m2 28.2 (21.4–40.5) 30.2 (22.6–45.5) 
 Waist, cm 98.2 (79.0–134.0) 95.4 (75.0–131.0) 
 Hip, cm 106.0 (94.0–126.0) 110.0 (93.0–141.0) 
 WHR 0.92 (0.82–1.08) 0.86 (0.77–1.05) 
 Fasting glucose, mmol/L 5.4 (4.7–7.1) 5.4 (4.6–8.6) 
 Fasting insulin, mU/L 13.8 (5.8–53.3) 12.3 (4.4–31.8) 
Arteriovenous study participants   
 n 23 19 
 Age, years 41 (22–59) 41 (23–57) 
 BMI, kg/m2 25.4 (20.8–30.6) 25.4 (21.0–39.3) 
 WHR 0.92 (0.79–1.05) 0.84 (0.72–0.97) 
 Abdominal fat mass, kg 10.6 (4.5–17.0) 12.6 (4.0–25.5) 
 Leg fat mass, kg 5.8 (3.5–8.7) 8.6 (4.7–15.3) 
 Fasting glucose, mmol/L 5.4 (4.8–6.6) 5.0 (4.4–5.6) 
 Fasting insulin, mU/L 11.6 (7.2–20.8) 11.7 (3.8–27.3) 
 Fasting nonesterified fatty acids, µmol/L 513 (300–1,084) 659 (297–1,012) 
 Serum IL-6, pg/mL 1.6 (0.26–9.14) 0.96 (0.31–2.69) 
 Serum MCP-1, pg/mL 7.67 (4.79–14.76) 6.86 (2.81–10.93) 
 Serum leptin, pg/mL 991 (69–5,375) 3,481 (333–18,000) 
 Abdominal ATBF 3.4 (0.3) 3.5 (0.4) 
 Gluteofemoral ATBF 2.5 (0.3) 3.4 (0.4) 

All data other than ATBF are presented as mean (minimum–maximum). ATBF data are presented as mean (± SEM) in mL · min–1 · 0.100 g tissue–1.

Analytical Methods

Plasma glucose, triacylglycerol (TAG), HDL, and wide-range C-reactive protein (CRP) concentrations were measured enzymatically using commercially available kits on an ILab 650 clinical analyzer (Instrumentation Laboratory, U.K.). Plasma insulin concentrations were measured by radioimmunoassay (Millipore UK Ltd.). Insulin resistance was estimated using the HOMA of insulin resistance (HOMA-IR) (29).

Regional Body Fat Measurements by DEXA

Regional and total fat mass (TFM) were measured using a Lunar iDXA (GE Healthcare, Bucks, U.K.). A daily quality assurance procedure was performed using a phantom block. Android and gynoid regions of interest were automatically defined by Encore software (version 14; GE Healthcare, Bucks, U.K.) as previously described (30).

In Vivo Measurements of Adipokine Release From GSAT and ASAT

Release of interleukin-6 (IL-6), monocyte chemoattractant protein-1 (MCP-1), and leptin from GSAT and ASAT was measured using the arteriovenous difference technique (31,32). Blood samples were obtained following an overnight fast from the superficial epigastric vein (draining ASAT), the saphenous vein (draining GSAT), and the femoral artery. Regional AT blood flow (ATBF) was measured by 133Xe washout (32). MCP-1 and leptin concentrations were determined using xMAP Technology (Luminex) with a Milliplex kit (Millipore). IL-6 concentrations were measured by ELISA (R&D Systems). Regional adipokine release was calculated by multiplying the arteriovenous difference by the ATBF.

Measurement of Cell Size

AT biopsies were fixed in paraformaldehyde, embedded in paraffin wax, cut into 5 μm sections, and stained with hematoxylin and eosin. Sections were viewed at ×10 magnification, and adipocyte cross-sectional area was calculated using Adobe Photoshop 5.0.1 (Adobe Systems, San Jose, CA) and Image Processing Tool Kit (Reindeer Games, Gainesville, FL).

Microarray Experiment: Sample Preparation and Analysis

Total RNA was extracted from AT biopsies, and Affymetrix HGU133 Plus 2 arrays were prepared as previously described (26). Gene expression data can be accessed at ArrayExpress (E-MTAB-54). After quality control, there remained 65 GSAT samples (five in replicate) and 54 ASAT samples (four in replicate) (26). Of these, 49 GSAT-ASAT pairs were selected for analysis. Microarray intensity data were preprocessed using gcrma software (33), including quantile normalization. Custom chip-definition files (34) were used to group probes into Entrez Gene–specific probe sets. Only the 18,122 probe sets annotated to autosomal genes were retained. In replicate samples, each gene’s pair of log2 (expression) was averaged. Statistical models were fitted to each gene’s expression data to determine whether expression significantly varied between AT depots (models MDepot and MDepot×Sex in the Supplementary Data) and/or significantly correlated with clinical traits within each AT depot (models MObesity and MObesity×Sex in the Supplementary Data). The number of significant genes cannot be compared directly across sex-specific analyses because of differences in sample size (Table 1); however, sex-specific effect estimates may be usefully compared with reference to SEs and/or CIs. Genes showing statistically significant differences in expression were analyzed for enrichment in gene ontology (GO) terms.

Preadipocyte Experiment: Culture of GSAT and ASAT Preadipocytes

Primary human preadipocytes were isolated as previously described (35). GSAT and ASAT preadipocyte cell lines were generated from a male subject according to Darimont et al. (36) by coexpressing human telomerase reverse transcriptase (hTERT) and human papillomavirus type-16 E7 oncoprotein (HPV16-E7) using the pLenti6.3/V5-DEST lentiviral expression system (Invitrogen). Preadipocytes were cultured in DMEM/F12 Ham nutrient mixture (DMEM/F12), 10% FCS, 2 mmol/L glutamine, 0.25 ng/mL fibroblast growth factor, 100 units/mL penicillin, and 0.1 mg/mL streptomycin. Confluent preadipocytes were stimulated for 14 days with an adipogenic cocktail comprising DMEM/F12, 2 mmol/L glutamine, 17 μmol/L pantothenate, 100 nmol/L human insulin, 10 nmol/L triiodo-L-thyronine, 33 μmol/L biotin, 10 μg/mL transferrin, 1 μmol/L dexamethasone, 100 units/mL penicillin, and 0.1 mg/mL streptomycin. For the first 4 days, 0.25 mmol/L 3-isobutyl-1-methylxanthine and 4 μmol/L troglitazone were added.

Preadipocyte Experiment: mRNA Preparation and Real-time PCR

Total RNA was extracted from preadipocytes (35). Real-time PCR was performed on an Applied Biosystems 7900HT, using TaqMan Assays-on-Demand (Applied Biosystems) and Klear Kall Master Mix (KBiosciences). mRNA expression values for target genes were calculated using the ΔCt transformation method (37). The ΔCt was calculated as follows: ΔCt = efficiency(minimumCt–sampleCt). Values were normalized to endogenous control genes (PPIA and PGK1) (38). Target mRNA log2 (expression) values are reported as the mean of the PPIA- and PGK1-normalized ratios.

Preadipocyte Experiment: Bisulfite Pyrosequencing

Genomic DNA was bisulfite converted using the EZ DNA Methylation-Gold Kit (Zymo Research). PCR of bisulfite-converted DNA was performed using 5′-biotinylated reverse primers, Titanium Taq reagents (Clontech) and 1 mmol/L betaine. PCR and sequencing primers (Supplementary Table 1) were designed using Methyl Primer (Applied Biosystems) and MethMarker (Max Planck Institute), respectively. Bisulfite pyrosequencing was performed on a PSQ 96ID (Qiagen), using PyroGold reagents (Qiagen). Results were subject to quality control checks, including positive controls to assess bisulfite conversion. Data were analyzed using Pyro Q-CpG software (version 1.0.9).

Short Hairpin RNA–Mediated Silencing of TBX5

TBX5 was silenced in primary and hTERT/HPV16-E7 preadipocytes derived from ASAT, using TBX5-short hairpin RNA (shRNA) lentiviral transduction particles (SHCLNV-NM_000192; Sigma-Aldrich). Control cells were transduced with nontarget shRNA lentiviral particles (SHC002V; Sigma-Aldrich). Quantification of intracellular lipid content was measured using the AdipoRed Assay (Lonza). For histological examination, differentiated preadipocytes were fixed with 4% paraformaldehyde and stained with AdipoRed (Lonza) and Oregon Green 488 phalloidin (Invitrogen). Western blot analysis was performed using anti-TBX5 (sc-17866; Santa Cruz) and anti-α-tubulin (ab15246; Abcam) followed by horseradish peroxidase–conjugated secondary antibodies and enhanced chemiluminescence detection (GE Healthcare, Little Chalfont, U.K.).

OBB Observations: Lower-Body AT Is Associated With Metabolic Protection

DEXA measurements of regional fat mass (Fig. 1A) were used to examine the hypothesis that preferential lower-body fat storage is associated with decreased metabolic risk. Proxy markers of insulin resistance (fasting insulin, HOMA-IR), dyslipidemia (TAG, HDL), and inflammation (CRP) were regressed on gynoid or android fat mass (Fig. 1B), with and without adjustment for TFM. Unadjusted regression analyses identified significant correlations between both fat depots and all risk factors (fasting insulin, HOMA-IR, TAG, HDL, CRP). However, when TFM was adjusted for, the direction of association between gynoid fat mass and each risk factor was reversed, with the reversed associations being statistically significant for insulin, HOMA-IR, TAG, and HDL in both sexes and for CRP in males. By comparison, associations between android fat mass and risk factors were not reversed under the TFM-adjusted model. These observations are consistent with the hypothesis that lower-body fat confers metabolic protection (insulin sensitizing, TAG lowering) when corrected for overall adiposity. Next we examined whether gynoid fat displays a distinct transcriptional profile that could support the hypothesis that preferential lower-body fat is metabolically protective.

Figure 1

Associations between regional fat distribution and risk factors for metabolic disease. A: Representative DEXA scan indicating the demarcation of android and gynoid fat as defined by Encore software. The android region lies between the ribs and iliac crest. The gynoid region includes the upper thighs and the hips. B: Regression of risk factors (fasting insulin, HOMA-IR, TAG, HDL, and CRP) on depot-specific fat mass (gynoid or android) unadjusted and adjusted for TFM. Estimates of regression coefficients are shown as points with 95% CIs as whiskers. The regression was performed separately for each combination of risk factor, fat depot, and sex. All risk factor and fat mass data were log transformed; prior to log transformation, CRP >10 μg/mL were omitted and a single zero data point was transformed to 0.01 µg/mL—the value of the next smallest data point. wrCRP, wide-range CRP.

Figure 1

Associations between regional fat distribution and risk factors for metabolic disease. A: Representative DEXA scan indicating the demarcation of android and gynoid fat as defined by Encore software. The android region lies between the ribs and iliac crest. The gynoid region includes the upper thighs and the hips. B: Regression of risk factors (fasting insulin, HOMA-IR, TAG, HDL, and CRP) on depot-specific fat mass (gynoid or android) unadjusted and adjusted for TFM. Estimates of regression coefficients are shown as points with 95% CIs as whiskers. The regression was performed separately for each combination of risk factor, fat depot, and sex. All risk factor and fat mass data were log transformed; prior to log transformation, CRP >10 μg/mL were omitted and a single zero data point was transformed to 0.01 µg/mL—the value of the next smallest data point. wrCRP, wide-range CRP.

Close modal

Intrinsic Depot-Specific Expression Profiles of GSAT and ASAT

In both sexes, 500 genes were differentially expressed between GSAT and ASAT (model MDepot), and 887 genes were differentially expressed in at least one sex (model MDepot×Sex). Across the two analyses, 942 unique genes were identified (in each case the nominal significance level was 5 × 10−4; see Supplementary Tables 2 and 3 for gene lists). Using the MDepot model, the transcripts displaying greatest differential expression were the long noncoding RNA, HOX transcript antisense RNA (HOTAIR; P = 3.8 × 10−18; GSAT-higher), and Homeobox gene (HOXA5; P = 2.9 × 10−19; ASAT-higher). The MDepot gene list was subdivided into GSAT-higher (253 genes) and GSAT-lower (247 genes) and tested for GO term enrichment. Four GO terms were significantly over-represented in GSAT-lower genes (Table 2), the most significant of which was “sequence-specific DNA binding” (P = 3.7 × 10−6; see Fig. 2A for annotated genes).

Table 2

GO terms enriched with genes differentially expressed in GSAT relative to ASAT

ContextHigher/lower expression in GSAT (+/−)OntologyGO IDGO termSignificant genes at termTotal genes at termSignificant genes in ontologyTotal genes in ontologyP value
Combined — MF GO:0043565 sequence-specific DNA bindinga 26 564 174 10,111 3.7E-06 
Combined — BP GO:0009952 anterior/posterior pattern specification 81 171 9,791 1.1E-05 
Combined — BP GO:0048704 embryonic skeletal system morphogenesis 34 171 9,791 2.3E-05 
Combined — MF GO:0003700 sequence-specific DNA-binding transcription factor activity 33 939 174 10,111 5.3E-05 
ContextHigher/lower expression in GSAT (+/−)OntologyGO IDGO termSignificant genes at termTotal genes at termSignificant genes in ontologyTotal genes in ontologyP value
Combined — MF GO:0043565 sequence-specific DNA bindinga 26 564 174 10,111 3.7E-06 
Combined — BP GO:0009952 anterior/posterior pattern specification 81 171 9,791 1.1E-05 
Combined — BP GO:0048704 embryonic skeletal system morphogenesis 34 171 9,791 2.3E-05 
Combined — MF GO:0003700 sequence-specific DNA-binding transcription factor activity 33 939 174 10,111 5.3E-05 

BP, biological process; MF, molecular function.

aSignificant genes annotated to “sequence-specific DNA binding” were selected for confirmation in ASAT and GSAT primary preadipocytes and hTERT/HPV16-E7 preadipocyte cell lines.

Figure 2

Depot-specific characteristics of developmental genes in human preadipocytes. A: Affymetrix mRNA expression of the most differentially expressed genes in GSAT and ASAT (HOTAIR and HOXA5, respectively) and all significant genes annotated to the enriched GO term “sequence-specific DNA binding.” B: Confirmation of selected depot-specific genes (HOXA5, IRX2, TBX5, HOXA6, HOXC11, SHOX2, HOTAIR) in primary human preadipocytes (n = 5) at day 0 and day 14 of adipogenic differentiation and in undifferentiated hTERT/HPV16-E7 human preadipocyte cell lines (n = 5) derived from GSAT and ASAT. Data in panels A and B are expressed as the GSAT/ASAT ratio of log2 expression ± 95% CIs. C: DNA methylation of individual CpG sites in the promoter regions of HOTAIR and TBX5 in hTERT/HPV16-E7 human preadipocyte cell lines derived from GSAT and ASAT. Bars represent percentage of DNA methylation ± 95% CIs (n = 5). D: Depot-specific mRNA expression of TBX5 during adipocyte differentiation in primary preadipocytes (n = 5; mean ± SEM) and hTERT/HPV16-E7 preadipocyte cell lines (n = 3; mean ± SEM). E: Protein expression of TBX5 and α-tubulin determined by Western blot in TBX5-shRNA preadipocyte cell lines derived from ASAT. F: Proliferation of TBX5-shRNA preadipocyte cell lines derived from ASAT assessed over 5 days (n = 3; mean ± SEM). G: Cellular lipid content of TBX5-shRNA preadipocyte cell lines derived from ASAT assessed at day 10; data are presented relative to nontarget shRNA controls (n = 3). H: TBX5-shRNA and nontarget shRNA primary preadipocytes derived from ASAT at day 10 labeled with AdipoRed (lipid marker) and Oregon Green 488 phalloidin (actin marker; magnification ×200). I: mRNA expression of adipogenic genes (CEBPA, PPARG2, FASN, FABP4, PLIN1, ADIPOQ) in TBX5-shRNA and nontarget shRNA ASAT preadipocytes over a 14-day adipogenic time course (n = 3; mean ± SEM). Affymetrix data were analyzed using model MDepot. All cellular data were analyzed by paired t test. The P values in panels B and C are Benjamini-Hochberg FDR-adjusted; other P values are nominal (*P < 0.05; **P < 0.005; ***P < 0.0005). TSS, transcription start site.

Figure 2

Depot-specific characteristics of developmental genes in human preadipocytes. A: Affymetrix mRNA expression of the most differentially expressed genes in GSAT and ASAT (HOTAIR and HOXA5, respectively) and all significant genes annotated to the enriched GO term “sequence-specific DNA binding.” B: Confirmation of selected depot-specific genes (HOXA5, IRX2, TBX5, HOXA6, HOXC11, SHOX2, HOTAIR) in primary human preadipocytes (n = 5) at day 0 and day 14 of adipogenic differentiation and in undifferentiated hTERT/HPV16-E7 human preadipocyte cell lines (n = 5) derived from GSAT and ASAT. Data in panels A and B are expressed as the GSAT/ASAT ratio of log2 expression ± 95% CIs. C: DNA methylation of individual CpG sites in the promoter regions of HOTAIR and TBX5 in hTERT/HPV16-E7 human preadipocyte cell lines derived from GSAT and ASAT. Bars represent percentage of DNA methylation ± 95% CIs (n = 5). D: Depot-specific mRNA expression of TBX5 during adipocyte differentiation in primary preadipocytes (n = 5; mean ± SEM) and hTERT/HPV16-E7 preadipocyte cell lines (n = 3; mean ± SEM). E: Protein expression of TBX5 and α-tubulin determined by Western blot in TBX5-shRNA preadipocyte cell lines derived from ASAT. F: Proliferation of TBX5-shRNA preadipocyte cell lines derived from ASAT assessed over 5 days (n = 3; mean ± SEM). G: Cellular lipid content of TBX5-shRNA preadipocyte cell lines derived from ASAT assessed at day 10; data are presented relative to nontarget shRNA controls (n = 3). H: TBX5-shRNA and nontarget shRNA primary preadipocytes derived from ASAT at day 10 labeled with AdipoRed (lipid marker) and Oregon Green 488 phalloidin (actin marker; magnification ×200). I: mRNA expression of adipogenic genes (CEBPA, PPARG2, FASN, FABP4, PLIN1, ADIPOQ) in TBX5-shRNA and nontarget shRNA ASAT preadipocytes over a 14-day adipogenic time course (n = 3; mean ± SEM). Affymetrix data were analyzed using model MDepot. All cellular data were analyzed by paired t test. The P values in panels B and C are Benjamini-Hochberg FDR-adjusted; other P values are nominal (*P < 0.05; **P < 0.005; ***P < 0.0005). TSS, transcription start site.

Close modal

To determine whether differential expression was detectable in preadipocytes and if a depot-specific transcriptional memory was retained ex vivo, mRNA expression of HOTAIR and HOXA5, as well as selected genes from the GO term “sequence-specific DNA binding” (IRX2, TBX5, HOXA6, HOXC11, SHOX2) was quantified in primary human preadipocytes (having undergone up to five passages) and hTERT/HPV16-E7 human preadipocyte cell lines (having undergone up to 20 passages), both originating from GSAT and ASAT. All selected genes showed GSAT-ASAT expression consistent with the whole tissue Affymetrix analysis (Fig. 2B). Furthermore, depot-specific expression was retained over a 14-day adipogenic differentiation time course in primary preadipocytes. The depot-specific transcriptional memory observed in isolated preadipocytes and cell lines suggested the involvement of epigenetic mechanisms. CpG islands were identified in the promoter regions of HOXA5, IRX2, HOXA6, TBX5, HOXC11, and HOTAIR, and differential DNA methylation of the HOTAIR and TBX5 promoters was investigated in hTERT/HPV16-E7 preadipocyte cell lines (Fig. 2C). CpG methylation was significantly higher in GSAT preadipocytes across 10 of the 14 CpG sites examined in the TBX5 promoter (−26 to −244 bp). By comparison, DNA methylation of CpG sites in the HOTAIR promoter (−55 to −132 bp) was significantly higher in ASAT preadipocytes. These data indicate that isolated preadipocytes exhibit intrinsic depot-specific DNA methylation across promoter regions of differentially expressed genes.

To investigate the contribution of GSAT-ASAT differentially expressed genes to adipocyte development and function, TBX5 was selected for shRNA-mediated silencing in primary and hTERT/HPV16-E7 preadipocytes. Expression of TBX5 remained higher in ASAT preadipocytes throughout a 14-day adipogenic time course (Fig. 2D); therefore, all shRNA studies were performed in ASAT preadipocytes. TBX5 mRNA expression was constitutively silenced by 60–80%, corresponding to a 45% reduction in TBX5 protein levels (Fig. 2E). TBX5-shRNA preadipocytes displayed a 1.5-fold reduction in cellular proliferation compared with nontarget shRNA control preadipocytes (Fig. 2F). A 50% reduction in lipid content was identified in TBX5-shRNA preadipocytes following adipogenic differentiation (Fig. 2G), and histological examination of TBX5-shRNA preadipocytes revealed a marked reduction in lipid-containing cells compared with controls (Fig. 2H). Consistent with impaired adipogenic differentiation, the expression of proadipogenic transcription factors (CEBPA, PPARG) and adipocyte functional genes (FASN, FABP4, PLIN1, ADIPOQ) was markedly ablated in TBX5-shRNA preadipocytes (Fig. 2I).

Relationships Between Obesity-Associated Traits and AT Transcriptional Profiles Are Sex- and Depot-Specific

Next, the transcriptional profile of each AT depot was characterized by identifying genes positively or negatively associated with obesity-descriptive traits (BMI, waist, hip, waist-to-hip ratio [WHR], HDL, insulin). As well as a simple model (MObesity) in which the trait association gradient was constant for both sexes, the data were also analyzed using a sex-specific model (MObesity×Sex; Supplementary Data). Fewer genes were significantly associated with obesity-associated traits in GSAT than in ASAT; in GSAT, there were 713 BMI-associated genes, compared with 1,803 in ASAT (nominal significance level of 5 × 10−4; see Supplementary Tables 4 and 5 for gene details). The majority of genes positively associated with obesity typically exhibited their strongest relationship in ASAT compared with GSAT, with the effect more pronounced in males (Fig. 3); for genes positively associated with BMI, the proportion exhibiting a stronger effect in ASAT was higher in males (0.82; 95% CI 0.78–0.83) than females (0.68; 0.60–0.74). GO analysis of genes positively associated with obesity identified the “extracellular region” as the site of strongest enrichment in GSAT, whereas the “lysosome” displayed strongest enrichment in ASAT (Fig. 4 and Supplementary Table 6). In both depots, there was significant enrichment of immune pathways, including “inflammatory response,” but specifically, enrichment of “platelet degranulation” was observed in ASAT only, whereas enrichment of “complement activation, classical pathway” was observed in GSAT only. Furthermore, there was no enrichment in GSAT of terms, including “positive regulation of neutrophil chemotaxis” and “platelet activation.”

Figure 3

Obesity trait and gene associations in GSAT and ASAT. For each gene significantly associated with an obesity-descriptive trait (BMI, waist, hip, WHR, HDL, or insulin) in either AT depot, the gradient of association was compared between depots (separately for males and females). The estimated proportion of genes for which the strongest obesity trait association occurred is displayed (central line in box) with bootstrapped 50% (represented by boxes) and 95% CIs (represented by error bars). A value of 1 indicates that all gene associations were strongest in ASAT; a value of 0 indicates that all gene associations were strongest in GSAT.

Figure 3

Obesity trait and gene associations in GSAT and ASAT. For each gene significantly associated with an obesity-descriptive trait (BMI, waist, hip, WHR, HDL, or insulin) in either AT depot, the gradient of association was compared between depots (separately for males and females). The estimated proportion of genes for which the strongest obesity trait association occurred is displayed (central line in box) with bootstrapped 50% (represented by boxes) and 95% CIs (represented by error bars). A value of 1 indicates that all gene associations were strongest in ASAT; a value of 0 indicates that all gene associations were strongest in GSAT.

Close modal
Figure 4

GO analysis of genes positively associated with obesity traits. Genes positively associated with obesity in GSAT and ASAT were tested for overrepresentation at GO terms at the 10−4 significance level (significant term-trait pairs are boxed, and enrichment P values are presented using the accompanying color scale). Examples of genes are presented in cutouts listing the most significantly obesity-associated genes, up to a maximum of 30. Supplementary Table 5 provides full gene names, and Supplementary Table 6 lists all significant GO terms and enrichment P values. BP, biological process; CC, cellular component; MF, molecular function.

Figure 4

GO analysis of genes positively associated with obesity traits. Genes positively associated with obesity in GSAT and ASAT were tested for overrepresentation at GO terms at the 10−4 significance level (significant term-trait pairs are boxed, and enrichment P values are presented using the accompanying color scale). Examples of genes are presented in cutouts listing the most significantly obesity-associated genes, up to a maximum of 30. Supplementary Table 5 provides full gene names, and Supplementary Table 6 lists all significant GO terms and enrichment P values. BP, biological process; CC, cellular component; MF, molecular function.

Close modal

Similarly, the majority of genes that were negatively associated with obesity exhibited their strongest response in ASAT compared with GSAT. This finding was consistent for all obesity-associated traits examined and was again stronger in males than in females (Fig. 3). In both depots, for genes negatively associated with BMI, waist, WHR, and HDL, GO analysis identified significant overrepresentation of genes involved in multiple energy-generating metabolic processes (including “respiratory electron transport chain,” “branched chain amino acid catabolism,” and “fatty acid β oxidation”), with the mitochondrion identified most clearly as the site of strongest enrichment (Fig. 5 and Supplementary Table 6). In summary, both GSAT and ASAT exhibited an obesity-associated transcriptional profile characterized by altered expression of immune and metabolic genes, and these associations were consistently weaker in GSAT.

Figure 5

GO analysis of genes negatively associated with obesity traits. Genes negatively associated with obesity in GSAT and ASAT were tested for overrepresentation at GO terms at the 10−4 significance level (significant term-trait pairs are boxed, and enrichment P values are presented using the accompanying color scale). Examples of genes are presented in cutouts listing the most significantly obesity-associated genes, up to a maximum of 30. Supplementary Table 5 provides full gene names, and Supplementary Table 6 lists all significant GO terms and enrichment P values. BP, biological process; CC, cellular component; MF, molecular function; TCA, tricarboxylic acid.

Figure 5

GO analysis of genes negatively associated with obesity traits. Genes negatively associated with obesity in GSAT and ASAT were tested for overrepresentation at GO terms at the 10−4 significance level (significant term-trait pairs are boxed, and enrichment P values are presented using the accompanying color scale). Examples of genes are presented in cutouts listing the most significantly obesity-associated genes, up to a maximum of 30. Supplementary Table 5 provides full gene names, and Supplementary Table 6 lists all significant GO terms and enrichment P values. BP, biological process; CC, cellular component; MF, molecular function; TCA, tricarboxylic acid.

Close modal

Depot-Specific Release of Adipokines from GSAT and ASAT

Next we investigated whether depot differences at the transcriptional level translated into functional differences measurable in vivo. Using arteriovenous sampling, we have previously reported GSAT-ASAT differences relating to lipid metabolism (10,11). To explore whether inflammation signaling also displays regional variation, we directly measured the release of two previously described proinflammatory cytokines (IL-6 and MCP-1 [39]) from GSAT-ASAT pairs. One individual showed extremely high release of IL-6 from GSAT (>60 pg · 100 g tissue−1 · min−1) and was removed from the analysis. In the remaining subjects, the release of IL-6 was substantially lower from GSAT than ASAT (Fig. 6A) (n = 34; Wilcoxon signed-rank, P = 5.1 × 10−6). Similar differences between tissues were seen in both sexes. The difference in release of MCP-1 between depots was less marked and not statistically significant (Fig. 6A) (n = 16; P = 0.13). Release of leptin was similar across both depots (ASAT, 1,601 ± 334 pg · 100 g tissue−1 · min−1; GSAT, 1,673 ± 366 pg · 100 g tissue−1 · min−1; n = 41; P = 0.86), indicating that lower IL-6 release from GSAT was not attributable to a more widespread reduction in peptide secretion. No difference in adipocyte cell size was observed between the two depots upon histological examination (Fig. 6B).

Figure 6

Measurement of depot-specific adipokine release from GSAT and ASAT. A: The release of IL-6 and MCP-1 from GSAT-ASAT pairs was determined by arteriovenous measurements. Data are shown as mean values within each depot ± 95% CIs. Statistical significance was assessed using a Wilcoxon signed-rank test. B: Cell size (area µm2) of GSAT and ASAT adipocytes was determined from hematoxylin and eosin–stained histological sections in males (n = 5) and females (n = 5). Mean ± SEM.

Figure 6

Measurement of depot-specific adipokine release from GSAT and ASAT. A: The release of IL-6 and MCP-1 from GSAT-ASAT pairs was determined by arteriovenous measurements. Data are shown as mean values within each depot ± 95% CIs. Statistical significance was assessed using a Wilcoxon signed-rank test. B: Cell size (area µm2) of GSAT and ASAT adipocytes was determined from hematoxylin and eosin–stained histological sections in males (n = 5) and females (n = 5). Mean ± SEM.

Close modal

Lower-body obesity is independently associated with reduced CVD and T2DM risk compared with central obesity (2,3). Using DEXA-quantified fat mass, we provide further support for the protective nature of lower-body fat in a large healthy population-based cohort—the OBB. Earlier DEXA studies have generally been limited by small sample size (40) or have not been performed in relevant populations (children, elderly) (41,42). Our findings highlight the importance of correcting for overall adiposity when examining the relationship between lower-body fat and metabolic health; due to colinearity, gynoid fat mass per se was positively related to metabolic risk factors, but this was reversed by adjusting for TFM and may thus provide an explanation for conflicting results in the literature (42). The transcriptional profile of GSAT and ASAT differs markedly in obesity, with GSAT demonstrating a level of “resistance” to the proinflammatory and hypometabolic expression profile that accompanies obesity in ASAT. This contrasts with an earlier report that visceral and abdominal subcutaneous AT exhibit similar transcriptional profiles with increasing adiposity (14) and demonstrates the value of studying two tissues that display opposing associations with metabolic disease risk. Translating our findings to a functional level, we demonstrate that proinflammatory cytokine release is lower from GSAT than ASAT, and propose that this regional variation arises from intrinsic developmental differences in preadipocyte populations.

Regional Inflammatory and Metabolic Profiles of AT

Obesity is associated with a chronic low-grade inflammatory state characterized by increased production of proinflammatory cytokines in AT (21,39) and is closely related to insulin resistance (21,43). Given the metabolic protection conferred by lower-body AT, we hypothesized that GSAT may exhibit a less inflammatory profile than ASAT in obesity. Both fat depots exhibited increased expression of genes relating to inflammation. However, when the strength of these associations was examined, the signal from GSAT was consistently weaker than ASAT. The finding that GSAT displays a less inflammatory profile in obesity contrasts with previous reports that inflammation is either the same as (44) or higher than (45) ASAT. However, unlike the global approach used here, those observations were based on gene-specific approaches (44,45). Furthermore, in addition to the transcriptional analyses, we performed in vivo measurements of regional cytokine release. For the first time, we report that IL-6 release from GSAT is consistently lower than from ASAT. Previous estimates suggested that subcutaneous AT contributes as much as 30% of systemic IL-6 concentrations, assuming equal release from all depots (46). Our findings indicate that GSAT makes a markedly reduced contribution compared with ASAT, and thus the overall impact of subcutaneous AT on systemic IL-6 levels may have previously been overestimated.

Macrophage infiltration is a widely reported feature of obese AT (22). GO analysis provided evidence to support an increased macrophage presence in ASAT (e.g., enrichment of “positive regulation of neutrophil chemotaxis”), whereas no corresponding enrichment was observed in GSAT. However, a lack of significance could also result from insufficient statistical power to detect enrichment. Owing to the heterogeneous cellular composition of AT, we cannot identify macrophage-specific effects; however, this would clearly be of interest in future studies. The factors that promote activation of AT inflammation remain unclear, but there appears to be a strong link between inflammation and metabolic dysfunction (47). In both GSAT and ASAT, enrichment of mitochondrial and energy-generating pathways was identified among genes negatively associated with obesity. These findings support the view that in obesity, the adipocyte becomes “hypometabolic” (14). GSAT exhibited a markedly weaker hypometabolic profile compared with ASAT, pointing to depot-specific differences in energy metabolism in obesity.

Developmental Differences Between GSAT and ASAT Preadipocytes

It is well documented in humans and rodents that genes involved in developmental processes, in particular, members of the HOX family, display striking differences in expression between regional (subcutaneous, omental, mesenteric) fat depots (13,14,16). We also observed significant enrichment of embryonic development genes and saw good agreement with previous reports of differential HOX gene expression in GSAT and ASAT (GSAT-lower genes in both studies included HOXA5, IRX2, HOXA3, HOXB8) (15). Furthermore, preadipocyte cell lines displayed a strong depot-specific transcriptional memory of GSAT-ASAT differentially expressed genes and showed distinct differences in the degree of promoter DNA methylation consistent with the differential expression (HOTAIR, TBX5). DNA methylation has previously been implicated in the depot-specific transcription of HOX genes in whole ASAT and GSAT (48). Together, these findings suggest epigenetic mechanisms may play an important role in regulating AT depot-specific characteristics.

The view that preadipocytes from different regions are developmentally distinct may be an important factor that contributes to regional differences in AT function (13,16). Indeed, two developmental genes (SHOX2, TBX15) have recently been identified as important regulators of adipocyte development and function (17,18). We have previously reported differential expression of TBX15 in GSAT and ASAT (27), which has since been shown to regulate differentiation of brown and “brite” adipocytes (18). In this study, we chose to investigate the functional role of TBX5, which, like TBX15, contains a common (T-box) DNA-binding domain and belongs to the Brachyury (T) family. A role for TBX5 in AT development has not previously been suggested; however, we hypothesized that TBX5 may regulate development of upper-body AT depots since mutations arising in TBX5 are responsible for the developmental disorder Holt-Oram syndrome, which is characterized by upper-body organ defects affecting the heart and upper limbs (49). The expression profile of TBX5 in GSAT and ASAT preadipocytes was consistent with this view; TBX5 expression was markedly higher in ASAT preadipocytes, with expression in GSAT preadipocytes barely detectable. Silencing TBX5 in ASAT preadipocytes resulted in a reduced proliferative capacity and a failure to differentiate along the adipogenic lineage. A role for TBX5 as a positive regulator of cell proliferation has previously been suggested in the developing heart (50), but this is the first evidence to support a role for TBX5 in adipogenic differentiation. Our findings suggest that TBX5 may act as an ASAT-specific regulator of preadipocyte proliferation and differentiation. Further studies are now required to examine expression of TBX5 in other AT depots, including visceral.

In conclusion, our findings support the view that, in obesity, AT displays a hypometabolic-proinflammatory profile. However, intrinsic developmental differences in the regional adipocyte populations may provide resistance of lower-body AT to immunological and metabolic derangements, thus contributing to the opposing associations between upper- and lower-body obesity and metabolic disease risk.

The views expressed are those of the authors and not necessarily those of the National Health Service, the National Institute for Health Research, or the Department of Health.

See accompanying article, p. 3575.

Acknowledgments. The authors thank Marjorie Gilbert (University of Oxford) for histological assistance.

Funding. This work was supported by funding from the European Commission to the MolPAGE Consortium (LSHGCT-2004-512066), the European Union FP7 LipidomicNet (202272), the Wellcome Trust, and the National Institute for Health Research Oxford Biomedical Research Centre based at Oxford University Hospitals National Health Service Trust and University of Oxford. The authors acknowledge all of the contributing partners within these consortia.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. K.E.P. and G.N. were responsible for planning and conducting experiments, data analysis and interpretation, and writing the manuscript. K.N.M. and S.E.M. conducted the clinical experiments and reviewed the manuscript. P.V. performed adipokine measurements. K.N.F. and K.T.Z. advised on the study design and reviewed the manuscript. N.D. performed bisulfite pyrosequencing experiments. J.L.M. performed data quality control for the microarray experiment and reviewed the manuscript. J.F. performed microarray experiments and reviewed the manuscript. M.I.M., C.C.H., and F.K. advised on the study design and statistical analysis and reviewed the manuscript. F.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Adipokine data were presented by K.N.M. at the Academy of Medical Sciences Meeting for Clinician Scientists in Training, London, U.K., 27 February 2013, and were published as a scientific abstract in The Lancet.

1.
Seidell
JC
,
Pérusse
L
,
Després
JP
,
Bouchard
C
.
Waist and hip circumferences have independent and opposite effects on cardiovascular disease risk factors: the Quebec Family Study
.
Am J Clin Nutr
2001
;
74
:
315
321
[PubMed]
2.
Snijder
MB
,
Dekker
JM
,
Visser
M
, et al
.
Associations of hip and thigh circumferences independent of waist circumference with the incidence of type 2 diabetes: the Hoorn Study
.
Am J Clin Nutr
2003
;
77
:
1192
1197
[PubMed]
3.
Yusuf
S
,
Hawken
S
,
Ôunpuu
S
, et al
INTERHEART Study Investigators
.
Obesity and the risk of myocardial infarction in 27,000 participants from 52 countries: a case-control study
.
Lancet
2005
;
366
:
1640
1649
[PubMed]
4.
Okura
T
,
Nakata
Y
,
Yamabuki
K
,
Tanaka
K
.
Regional body composition changes exhibit opposing effects on coronary heart disease risk factors
.
Arterioscler Thromb Vasc Biol
2004
;
24
:
923
929
[PubMed]
5.
Petersen SE, Matthews PM, Bamberg F, et al. Imaging in population science: cardiovascular magnetic resonance in 100,000 participants of UK Biobank - rationale, challenges and approaches. J Cardiovasc Magn Reson 2013;15:46
6.
Romanski
SA
,
Nelson
RM
,
Jensen
MD
.
Meal fatty acid uptake in adipose tissue: gender effects in nonobese humans
.
Am J Physiol Endocrinol Metab
2000
;
279
:
E455
E462
[PubMed]
7.
Jensen
MD
,
Johnson
CM
.
Contribution of leg and splanchnic free fatty acid (FFA) kinetics to postabsorptive FFA flux in men and women
.
Metabolism
1996
;
45
:
662
666
[PubMed]
8.
Tan
GD
,
Goossens
GH
,
Humphreys
SM
,
Vidal
H
,
Karpe
F
.
Upper and lower body adipose tissue function: a direct comparison of fat mobilization in humans
.
Obes Res
2004
;
12
:
114
118
[PubMed]
9.
Manolopoulos
KN
,
Karpe
F
,
Frayn
KN
.
Marked resistance of femoral adipose tissue blood flow and lipolysis to adrenaline in vivo
.
Diabetologia
2012
;
55
:
3029
3037
[PubMed]
10.
McQuaid
SE
,
Humphreys
SM
,
Hodson
L
,
Fielding
BA
,
Karpe
F
,
Frayn
KN
.
Femoral adipose tissue may accumulate the fat that has been recycled as VLDL and nonesterified fatty acids
.
Diabetes
2010
;
59
:
2465
2473
[PubMed]
11.
Pinnick
KE
,
Neville
MJ
,
Fielding
BA
,
Frayn
KN
,
Karpe
F
,
Hodson
L
.
Gluteofemoral adipose tissue plays a major role in production of the lipokine palmitoleate in humans
.
Diabetes
2012
;
61
:
1399
1403
[PubMed]
12.
Tchkonia
T
,
Thomou
T
,
Zhu
Y
, et al
.
Mechanisms and metabolic implications of regional differences among fat depots
.
Cell Metab
2013
;
17
:
644
656
[PubMed]
13.
Tchkonia
T
,
Lenburg
M
,
Thomou
T
, et al
.
Identification of depot-specific human fat cell progenitors through distinct expression profiles and developmental gene patterns
.
Am J Physiol Endocrinol Metab
2007
;
292
:
E298
E307
[PubMed]
14.
Klimcáková
E
,
Roussel
B
,
Márquez-Quiñones
A
, et al
.
Worsening of obesity and metabolic status yields similar molecular adaptations in human subcutaneous and visceral adipose tissue: decreased metabolism and increased immune response
.
J Clin Endocrinol Metab
2011
;
96
:
E73
E82
[PubMed]
15.
Karastergiou
K
,
Fried
SK
,
Xie
H
, et al
.
Distinct developmental signatures of human abdominal and gluteal subcutaneous adipose tissue depots
.
J Clin Endocrinol Metab
2013
;
98
:
362
371
[PubMed]
16.
Gesta
S
,
Blüher
M
,
Yamamoto
Y
, et al
.
Evidence for a role of developmental genes in the origin of obesity and body fat distribution
.
Proc Natl Acad Sci U S A
2006
;
103
:
6676
6681
[PubMed]
17.
Lee
KY
,
Yamamoto
Y
,
Boucher
J
, et al
.
Shox2 is a molecular determinant of depot-specific adipocyte function
.
Proc Natl Acad Sci U S A
2013
;
110
:
11409
11414
[PubMed]
18.
Gburcik
V
,
Cawthorn
WP
,
Nedergaard
J
,
Timmons
JA
,
Cannon
B
.
An essential role for Tbx15 in the differentiation of brown and “brite” but not white adipocytes
.
Am J Physiol Endocrinol Metab
2012
;
303
:
E1053
E1060
[PubMed]
19.
Pinnick
KE
,
Karpe
F
.
DNA methylation of genes in adipose tissue
.
Proc Nutr Soc
2011
;
70
:
57
63
[PubMed]
20.
McQuaid
SE
,
Hodson
L
,
Neville
MJ
, et al
.
Downregulation of adipose tissue fatty acid trafficking in obesity: a driver for ectopic fat deposition?
Diabetes
2011
;
60
:
47
55
[PubMed]
21.
Hotamisligil
GS
,
Arner
P
,
Caro
JF
,
Atkinson
RL
,
Spiegelman
BM
.
Increased adipose tissue expression of tumor necrosis factor-alpha in human obesity and insulin resistance
.
J Clin Invest
1995
;
95
:
2409
2415
[PubMed]
22.
Xu
H
,
Barnes
GT
,
Yang
Q
, et al
.
Chronic inflammation in fat plays a crucial role in the development of obesity-related insulin resistance
.
J Clin Invest
2003
;
112
:
1821
1830
[PubMed]
23.
Wallace
AM
,
McMahon
AD
,
Packard
CJ
, et al
.
Plasma leptin and the risk of cardiovascular disease in the west of Scotland coronary prevention study (WOSCOPS)
.
Circulation
2001
;
104
:
3052
3056
[PubMed]
24.
Danforth
E
 Jr
.
Failure of adipocyte differentiation causes type II diabetes mellitus?
Nat Genet
2000
;
26
:
13
[PubMed]
25.
Gregor
MF
,
Hotamisligil
GS
.
Thematic review series: Adipocyte Biology. Adipocyte stress: the endoplasmic reticulum and metabolic disease
.
J Lipid Res
2007
;
48
:
1905
1914
[PubMed]
26.
Min
JL
,
Nicholson
G
,
Halgrimsdottir
I
, et al
GIANT Consortium
MolPAGE Consortium
.
Coexpression network analysis in abdominal and gluteal adipose tissue reveals regulatory genetic loci for metabolic syndrome and related phenotypes
.
PLoS Genet
2012
;
8
:
e1002505
[PubMed]
27.
Heid
IM
,
Jackson
AU
,
Randall
JC
, et al
MAGIC
.
Meta-analysis identifies 13 new loci associated with waist-hip ratio and reveals sexual dimorphism in the genetic basis of fat distribution
.
Nat Genet
2010
;
42
:
949
960
[PubMed]
28.
Tan
GD
,
Neville
MJ
,
Liverani
E
, et al
.
The in vivo effects of the Pro12Ala PPARgamma2 polymorphism on adipose tissue NEFA metabolism: the first use of the Oxford Biobank
.
Diabetologia
2006
;
49
:
158
168
[PubMed]
29.
Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 1985;28:412–419
30.
Stults-Kolehmainen MA, Stanforth PR, Bartholomew JB, Lu T, Abolt CJ, Sinha R. DXA estimates of fat in abdominal, trunk and hip regions varies by ethnicity in men. Nutr Diabetes 2013;3:e64
31.
McQuaid
SE
,
Manolopoulos
KN
,
Dennis
AL
,
Cheeseman
J
,
Karpe
F
,
Frayn
KN
.
Development of an arterio-venous difference method to study the metabolic physiology of the femoral adipose tissue depot
.
Obesity (Silver Spring)
2010
;
18
:
1055
1058
[PubMed]
32.
Frayn
KN
,
Coppack
SW
,
Humphreys
SM
,
Whyte
PL
.
Metabolic characteristics of human adipose tissue in vivo
.
Clin Sci (Lond)
1989
;
76
:
509
516
[PubMed]
33.
Wu
ZJ
,
Irizarry
RA
,
Gentleman
R
,
Martinez-Murillo
F
,
Spencer
F
.
A model-based background adjustment for oligonucleotide expression arrays
.
J Am Stat Assoc
2004
;
99
:
909
917
34.
Dai
M
,
Wang
P
,
Boyd
AD
, et al
.
Evolving gene/transcript definitions significantly alter the interpretation of GeneChip data
.
Nucleic Acids Res
2005
;
33
:
e175
[PubMed]
35.
Collins
JM
,
Neville
MJ
,
Hoppa
MB
,
Frayn
KN
.
De novo lipogenesis and stearoyl-CoA desaturase are coordinately regulated in the human adipocyte and protect against palmitate-induced cell injury
.
J Biol Chem
2010
;
285
:
6044
6052
[PubMed]
36.
Darimont
C
,
Zbinden
I
,
Avanti
O
, et al
.
Reconstitution of telomerase activity combined with HPV-E7 expression allow human preadipocytes to preserve their differentiation capacity after immortalization
.
Cell Death Differ
2003
;
10
:
1025
1031
[PubMed]
37.
Pfaffl
MW
.
A new mathematical model for relative quantification in real-time RT-PCR
.
Nucleic Acids Res
2001
;
29
:
e45
[PubMed]
38.
Neville
MJ
,
Collins
JM
,
Gloyn
AL
,
McCarthy
MI
,
Karpe
F
.
Comprehensive human adipose tissue mRNA and microRNA endogenous control selection for quantitative real-time-PCR normalization
.
Obesity (Silver Spring)
2011
;
19
:
888
892
[PubMed]
39.
Madani
R
,
Karastergiou
K
,
Ogston
NC
, et al
.
RANTES release by human adipose tissue in vivo and evidence for depot-specific differences
.
Am J Physiol Endocrinol Metab
2009
;
296
:
E1262
E1268
[PubMed]
40.
Wiklund
P
,
Toss
F
,
Weinehall
L
, et al
.
Abdominal and gynoid fat mass are associated with cardiovascular risk factors in men and women
.
J Clin Endocrinol Metab
2008
;
93
:
4360
4366
[PubMed]
41.
Aucouturier
J
,
Meyer
M
,
Thivel
D
,
Taillardat
M
,
Duché
P
.
Effect of android to gynoid fat ratio on insulin resistance in obese youth
.
Arch Pediatr Adolesc Med
2009
;
163
:
826
831
[PubMed]
42.
Kang
SM
,
Yoon
JW
,
Ahn
HY
, et al
.
Android fat depot is more closely associated with metabolic syndrome than abdominal visceral fat in elderly people
.
PLoS ONE
2011
;
6
:
e27694
[PubMed]
43.
Kern
PA
,
Ranganathan
S
,
Li
C
,
Wood
L
,
Ranganathan
G
.
Adipose tissue tumor necrosis factor and interleukin-6 expression in human obesity and insulin resistance
.
Am J Physiol Endocrinol Metab
2001
;
280
:
E745
E751
[PubMed]
44.
Mališová
L
,
Rossmeislová
L
,
Kováčová
Z
, et al
.
Expression of inflammation-related genes in gluteal and abdominal subcutaneous adipose tissue during weight-reducing dietary intervention in obese women
.
Physiol Res
2014;63:73–82
[PubMed]
45.
Evans
J
,
Goedecke
JH
,
Söderström
I
, et al
.
Depot- and ethnic-specific differences in the relationship between adipose tissue inflammation and insulin sensitivity
.
Clin Endocrinol (Oxf)
2011
;
74
:
51
59
[PubMed]
46.
Mohamed-Ali
V
,
Goodrick
S
,
Rawesh
A
, et al
.
Subcutaneous adipose tissue releases interleukin-6, but not tumor necrosis factor-alpha, in vivo
.
J Clin Endocrinol Metab
1997
;
82
:
4196
4200
[PubMed]
47.
Hotamisligil
GS
.
Inflammation and metabolic disorders
.
Nature
2006
;
444
:
860
867
[PubMed]
48.
Gehrke
S
,
Brueckner
B
,
Schepky
A
, et al
.
Epigenetic regulation of depot-specific gene expression in adipose tissue
.
PLoS ONE
2013
;
8
:
e82516
[PubMed]
49.
Basson
CT
,
Bachinsky
DR
,
Lin
RC
, et al
.
Mutations in human TBX5 [corrected] cause limb and cardiac malformation in Holt-Oram syndrome
.
Nat Genet
1997
;
15
:
30
35
[PubMed]
50.
Bruneau
BG
,
Nemer
G
,
Schmitt
JP
, et al
.
A murine model of Holt-Oram syndrome defines roles of the T-box transcription factor Tbx5 in cardiogenesis and disease
.
Cell
2001
;
106
:
709
721
[PubMed]

Supplementary data