Funding agencies: Supported by grants from the Ministerio de Educacion y Ciencia (SAF 2010/19527), Generalitat de Catalunya (SGR2009/1426), CIBERDEM (Centro de Investigación en Redes de Diabetes y Enfermedades Metabolicas), an initiative of the Instituto de Salud Carlos III, and the Ist Award Boston Scientific of the Sociedad Espanola de Diabetes 2009. FAH was a recipient of a Rio-Hortega contract from the Instituto de Salud Carlos III (ISCIII, Spain) and KP a recipient of an IDIBAPS fellowship.
Obesity rather than regional fat depots marks the metabolomic pattern of adipose tissue: An untargeted metabolomic approach
Article first published online: 18 SEP 2013
Copyright © 2013 The Obesity Society
Volume 22, Issue 3, pages 698–704, March 2014
How to Cite
Hanzu, F.A., Vinaixa, M., Papageorgiou, A., Párrizas, M., Correig, X., Delgado, S., Carmona, F., Samino, S., Vidal, J. and Gomis, R. (2014), Obesity rather than regional fat depots marks the metabolomic pattern of adipose tissue: An untargeted metabolomic approach. Obesity, 22: 698–704. doi: 10.1002/oby.20541
Disclosure: The authors declare no conflict of interest.
Author contributions: MV, SS, and AP designed and carried out experiments. MV, MP, XC designed experiments and carried out the analysis. JV, RG, SD, and FC designed procedures for patient recruitment, clinical phenotyping and analysis. FAH was involved in all steps of the study. All authors were involved in writing the article and had final approval of the submitted and published versions.
- Issue published online: 5 MAR 2014
- Article first published online: 18 SEP 2013
- Accepted manuscript online: 26 JUN 2013 01:07PM EST
- Manuscript Accepted: 26 MAY 2013
- Manuscript Revised: 25 MAY 2013
- Manuscript Received: 17 DEC 2012
This study compares the patterns of visceral (VIS) and subcutaneous (SC) adipose tissue (AT)-derived metabolites from non-obese (BMI 24-26 kg/m2) and obese subjects (BMI > 40 kg/m2) with no major metabolic risk factors other than BMI.
SC- and VIS- AT obtained from obese (Ob) and non-obese (NOb) subjects during surgery were incubated to obtain their metabolites. Differences related to obesity or anatomical provenances of AT were assessed using an untargeted metabolomics approach based on gas chromatography-mass spectrometry.
The overall effect of obesity on the metabolite profile resulted more remarkable than the effect of regional AT. Only the depletion of 2-ketoisocaproic (2-KIC) acid reached statistical significance for the SC-AT alone, although it was observed in both depots. Obesity induced more significant changes in several amino acids levels of the VIS-AT metabolites. On the one hand, higher released levels of glutamine and alanine were detected in the VIS- obese AT, whereas on the other, the VIS- obese AT presented a diminished uptake of essential amino acids (methionine, threonine, lysine), BCAAs, leucine, and serine.
This study shows that obesity markedly affects the amino acid metabolic signature of the AT before the clinical onset of other significant metabolic alterations aside from BMI.
Obesity, an epidemically increasing phenotype, and, in particular, the visceral accumulation of adipose tissue, is closely related to the alterations of glucose and lipid metabolism which lead to dyslipidemia, insulin resistance and diabetes mellitus . Changes in adipose tissue plasticity, with the activation of constitutive adipose tissue cell fractions, locally excessive accumulation of lipids and disturbed cellular metabolism, are thought to be early determinants of these processes, as they are present long before the clinical expression of the disease .
Nowadays, metabolites produced by obese adipose tissue are emerging as principal mediators of systemic metabolic disease. Adipose tissue-secreted metabolites are not only indicative of the obesity-induced psychopathological state but may also act as initiators of pathological cascades and play endocrine roles as regulatory signals . In this regard, untargeted metabolomic studies empower a hypothesis-free overview of the highly abundant metabolites in biological matrices that are substrates and products of the metabolic pathways most affected by the causative disease. Although alterations of carbohydrate and lipid homeostasis in obesity have been widely recognized , the role of specific metabolites [5, 6] remains under study, with controversy surrounding the question of whether protein metabolism is affected in human obesity. Recent targeted metabolomic sera studies have reported notable differences in the metabolic sera signatures of obese non-diabetic and lean individuals [7-9]. Moreover, specific plasmatic metabolites are emerging as strong predictors of insulin resistance and the onset of diabetes in obesity [10-12]. However, the contribution of the metabolites produced by the different obese adipose tissue depots to the systemic altered metabolite pool has not yet been defined.
Hence, in order to study the biological mechanisms underlying obesity and the particularity of regional adipose tissue depots, we analyzed the adipose tissue-derived metabolites in obese and non-obese individuals using a gas chromatography-mass spectrometry (GC-MS) untargeted metabolomics approach. Our ultimate goal was to decipher how obesity affects the profile of the adipose tissue metabolites. Thus, we compared the metabolic patterns of visceral and subcutaneous adipose tissue in lean and obese subjects with no metabolic risk factors other than their BMI.
A total of eight morbid obese subjects (BMI > 40 kg/m2) from the Hospital Clínic about to undergo gastric bypass surgery were included. The study was approved by the hospital's Ethics Committee. Written informed consent was obtained from all participants. Clinical history, physical data, blood pressure, and ECG were registered. Obese subjects that fulfilled the IDF diagnosis criteria of metabolic syndrome (www.IDF.org) were excluded from the study. Six non-obese age- and sex-matched subjects (BMI: 24-26 kg/m2) undergoing elective procedures of abdominal surgery to correct benign conditions served as controls. Participants of both study groups fulfilled the inclusion criteria previously mentioned .
Main biochemical parameters were measured in serum after overnight fasting. Serum insulin and adiponectin levels were measured by immune-radiometric assays. Insulin resistance was estimated by calculating HOMA-IR index, with a threshold of 2.9, the reference value for the control population from our hospital.
The baseline clinical and metabolic characteristics of study participants are shown in Table 1. Obese patients presented moderate insulin resistance as estimated by HOMA-IR, no alteration of glucose homeostasis, and, as predicted from BMI, systemic signs of inflammation, increased high sensitive C reactive protein (hs-CRP) and decreased adiponectin levels.
|Obese subjects; N = 8||Non-obese subjects; N = 6|
|Age (years)||41.4 ± 10||44.4 ± 14|
|BMI (kg/m2)||44.1 ± 2.2||24.6 ± 2.3|
|Waist circumference (cm)||126.3 ± 4.02||82.9 ± 2.1|
|SBP/DBP (mmHg)||126.1/75.4 ± 2.7/1.8||116.1/70.6 ± 1.5/1.3|
|Fasting glucose (mmol/L)||5.5 ± 0.2||4.8 ± 0.5|
|2h-OGTT plasma glucose (mmol/L)||7.5 ± 1.5||-|
|Cholesterol total (mmol/L)||5.0 ± 1.1||4.3 ± 1.3|
|HDL/LDL (mmol/L)||1.3/3.3 ± 0.1/0.7||1.5/3 ± 0.1/0.5|
|Triglycerides (mmol/L)||1.4 ± 0.2||1 ± 0.3|
|Leukocytes (x 109/L)||8.2 ± 1.7||6.3 ± 0.8|
|High Sensitive-CRP (mg/L)||13.2 ± 4.5||(<5)a|
|HbA1c (NGPS/DCC)||5.2 ± 0.3||(4.0-5.5)a|
|Insulin (pmol/L)||201.9 ± 77.2||(15.3-98.6)a|
|HOMA-IR||5.7 ± 1.9||(<2.93)a|
Samples of adipose tissue biopsies. Preparation of the conditioned media containing the obese adipose tissue-derived metabolites
The conditioned media containing the adipose tissue metabolites, were obtained as follows: visceral (VIS, omental) and subcutaneous (SC, periumbilical) adipose tissue biopsies (AT) from the morbid obese (Ob, n = 8) and non-obese (NOb, n = 6) individuals were isolated after 12-h fasting as paired samples during surgery and immediately transported to the laboratory for analysis. Each experimental replicate involved tissue from a separate individual. All subsequent procedures were carried out under laminar airflow and sterile conditions. Studies were performed following the protocol of Fain and Rodbell [14, 15] with minor modifications. Subcutaneous (SC) and visceral (VIS) adipose tissue samples were carefully cleaned, cut into small pieces (10 mg), incubated in Medium 199 (M199, Sigma–Aldrich) with pH 7.4 for 2-5 min, and then centrifuged (400g for 30 sec) to reduce contamination with blood cells and soluble factors and pieces of tissue containing insufficient adipocytes to float. The cleaned and cut tissues were kept as intact fat pads and placed in M199, pH 7.4 (1g AT/5 mL). After 2 h, the medium was replaced in every dish, and incubation continued for 24 h at 37°C in the same serum-free medium to obtain the corresponding fat pad-conditioned media (NOb-VIS, NOb-SC, Ob-VIS, Ob-SC). After 24 h of incubation, the conditioned media were separated from the fat pads by centrifugation (400g for 30 sec) and the debris removed. The conditioned media were passed through a 200-µm filter, divided in 150-µL aliquots, directly frozen and stored at −80°C until use. Both conditioned media and fat pads were, immediately after processing, separately stored at −80°C until use. The total time spent on sample preparation following incubation and until freezing did not surpass the 15 min. The conditioned media containing the adipose tissue-derived metabolites were employed, respectively, in the subsequent experiments.
Gas chromatography-mass spectrometry analysis
The adipose tissue-conditioned media samples containing the adipose tissue metabolites and the non-supplemented culture medium were thawed at 4°C. The whole gas chromatography-mass spectrometry (GC-MS) analysis was performed according to Agilent's specifications . Briefly, 100-μL aliquots of each biological specimen were spiked with 20-μL internal standard solution (1 μg μ/L succinic-d4 acid; Sigma–Aldrich), subsequently used for normalization. Metabolites were extracted by adding 900 μL cold methanol/water (8:1 v/v) followed by 4 min of ultrasonication and 10 sec of vortexing. After centrifugation (10 min at 19,000g at 4°C), three technical replicates of each specimen were prepared transferring 200 μL supernatant to a GC autosampler vial. Samples were subsequently spiked with 20 µL myristic acid-d27 (Sigma–Aldrich), used as the internal standard for retention time lock (RTL system provided in Agilent's ChemStation Software), and lyophilized overnight (Lyotrap freeze dryer). Metabolites were derivatized in order to make them volatile and less polar and, hence, amenable to GC-MS analysis. Derivatization consisted of two different steps. The first one, namely methoximation, was performed to prevent ring formation and to stabilize carbonyl moieties. Thus, lyophilized sample residues were incubated with 50 μL methoxyamine in pyridine (0.3 μg μ/L) for 16 h at room temperature. The next step was performed to increase the volatility of the compounds. Samples were silylated using 30 μL N-methyl-N-trimethylsilyltrifluoroacetamide with 1%trimethylchlorosilane (MSTFA + 1% TMCS, Sigma) for 1 h at room temperature.
Derivatized samples were automatically injected into a GC–MS system (HP 6890 Series gas chromatograph coupled to a mass selective detector model 5973) equipped with a J&W Scientific DB 5-MS+DG stationary phase column (30 m × 0.25 mm i.d., 0.1 μm film) (Agilent Technologies). The injector temperature was set at 250°C, and the helium carrier flow rate was kept constant at 1.1 mL/min. The column temperature was held at 60°C for 1 min, then increased to 325°C at a rate of 10°C/min and held at 325°C for 10 min. The detector operated in the electron impact ionization mode (70 eV), and mass spectra were recorded after a solvent delay of 4 min with 2.46 scans per second (mass scanning range of mass-charge ratio, m/z, 50-600; threshold abundance value of 50 counts). The source temperature and quadrupole temperatures were 230 and 150°C, respectively. To reduce systematic error associated with instrumental drift, samples were entirely randomized.
Raw GC/MS files were exported into the platform-independent netCDF (*.cdf) and loaded into XCMS software (version 1.6.1) based on R-program version 2.4.0 (R-Foundation for statistical computing, www.Rproject.org), where peak finding, integration and alignment in the time domain were performed. XCMS data was exported to Matlab (version 6.5.1, Release 13, The Mathworks, 2003), where normalization to internal standard succinic acid-d4 was performed and the average integrated intensities for the three analytical replicates of each biological specimen were computed. The samples were compared based on either multivariate or systematic multiple univariate statistical test performed on XCMS-derived dataset. The false discovery rate (FDR) procedure described by Storey et al.  was used to account for multiple testing. The AMDIS program (Automated Mass Spectral Deconvolution and Identification System, National Institute of Standards and Technology, Gathersburg, MD) was run for peak annotation, and both the Fiehn GC/MS Metabolomics RTL Library and NIST mass spectral databases were used for identification. Data processing, analysis, and statistical calculations were performed using Matlab.
Obesity rather than fat regional provenance affects the metabolic signature of adipose tissue-derived metabolites
After raw data filtering, retention time correction, peak alignment in the time domain and peak integration, XCMS generates a table containing the resulting integrated intensities for each m/z-retention time pair spectral feature detected. A feature is defined as a molecular entity with a unique m/z and a specific retention time. This data table was scaled to unit variance and used to fit four different PCA (principal component analysis) multivariate models aimed at investigating the effect of obesity and fat pad regional differences on the metabolic profiles. PCA is a method used for the unsupervised exploratory analysis of multivariate datasets derived from high throughput analysis technologies. It involves a mathematical transformation of a number of possibly correlated input variables into a smaller number of uncorrelated variables called principal components. The projection of sample data into this new lower-dimensional variables subspace is called the PCA scores plot.
This provides a simplified qualitative overview on how the adipose tissue-conditioned media samples relate to each other in terms of their metabolic profiles i.e., similarity or dissimilarity among the metabolic profiles of the studied samples. Figure 1 shows the PC1/PC2 score plots derived from the PCA comparison of the obese (Ob) and non-obese (NOb) adipose tissue metabolites from either visceral (VIS-AT, left panel) or subcutaneous (SC-AT, right panel) fat pad depots. In the case of visceral adipose tissue (left panel), samples belonging to obese subjects (gray dots) presented a prevailing trend to cluster altogether and apart from the non-obese subjects (white dots). This trend resulted less pronounced in the case of subcutaneous adipose tissue (right panel). Nevertheless, PCA analysis of the visceral versus the subcutaneous adipose tissue metabolites did not reveal any clustering trends, neither for non-obese nor for obese subjects (data not shown).
We also analyzed the influence of obesity and of the anatomical provenance of fat pads on the metabolic profile of the adipose tissue-derived metabolites from an univariate data analysis perspective. Thus, we applied multiple paralleled two-way ANOVA on the entire list of metabolic features detected using XCMS. Data were rank-transformed before being submitted to two-way ANOVA. After FDR correction for multiple testing, 31% and just 19% out of the initially detected features resulted significantly different due to obesity and regional fat provenance, respectively.
Overall, our exploratory analysis indicates that obesity induces a differential metabolic secretory pattern of adipose tissue, an effect that is more pronounced in visceral adipose tissue.
Visceral obese fat pads present the greatest differences in the pattern of adipose tissue-derived metabolites
We retained those metabolic features significantly varied according to obesity for further metabolite annotation and identification. Obese and non-obese conditioned media samples containing the metabolites derived from either visceral or subcutaneous adipose tissue were compared using the Mann-Witney test and FDR correction. Metabolite identification parameters together with statistical analysis details are summarized in Table 2. Strikingly, most of the metabolites found to be significantly varied according to obesity were identified as amino acids: the essential ones, namely alanine, lysine, methionine, threonine; the branched-chain amino acid (BCAA) leucine, as well as glutamine and serine (Table 2). Subsequently, we comparatively determined the levels of these amino acids in the initial nonconditioned culture medium. Glutamine, alanine and 2-KIC acid (2 ketoisocaproic) were only detectable in the obese adipose tissue-conditioned samples and not in the non-conditioned media, therefore, they were considered a product of fat pads.
|Retention time (min)||Quantitative ion (m/z)||P values||q values||Fold change Ob vs NOb|
On the other hand leucine, lysine, threonine, methionine, and serine were detected in both the adipose tissue conditioned media and in the non-conditioned culture medium, and they were therefore considered as uptake metabolites.
As summarized in Table 2, in the case of obese subcutaneous adipose tissue metabolites, the only difference detected was the decreased levels of secreted 2-KIC acid for the obese subjects. However, this was not the case for the visceral adipose tissue metabolites from obese subjects, where we consistently detected higher released levels of glutamine and alanine and a lower net uptake of leucine, lysine, threonine, methionine, and serine (Figure 2).
To the best of our knowledge, this is the first comprehensive differential report on the metabolomics of different regional adipose depots isolated from obese subjects. Study participants presented no metabolic disturbances other than their BMI, thus, none of them could be diagnosed as suffering from metabolic syndrome following the IDF consensus criteria (www.idf.org). As expected, due to their BMI, they presented signs of systemic inflammation, such as increased high sensitive C reactive protein, insulin resistance, and decreased adiponectin.
Although untargeted GC-MS-based metabolomics allows for the non-biased screening of a wide range of chemical classes (sugars, short and medium chain fatty acids, steroids, amino acids or organic acids, among others), our results demonstrated that amino acids accounted for the main differences observed in the secretory patterns of visceral and subcutaneous fat depots of non-obese and obese individuals. Our work reports significant differences in essential (alanine, lysine, methionine, threonine), BCAA (leucine), the non-essential amino acids (glutamine and serine) and 2-ketoisocaproic acid, pointing out that obesity overall blunts amino acid metabolism in adipose tissue and profoundly affects BCAA catabolism. Moreover, our results demonstrate that the effect of obesity upon the metabolic patterns of adipose tissue is more profound than the effect of the regional provenance of fat tissue, being the more affected the metabolism of obese visceral adipose tissue, in line with our previously reported results regarding holistic changes in obese adipose tissue .
Amino acids are not stored in the body. There is a highly active interorgan transport, and dietary amino acids in excess of those required for protein synthesis or gluconeogenesis are rapidly catabolized . Dysfunction of tisular amino acid metabolism has been previously reported to be correlated with insulin resistance in obesity. In this sense, the liver's incapacity to take the essential amino acids for gluthatione synthesis (GSH) has been considered to have repercussions on postprandial peripheral insulin sensitivity [19, 20]. In the adipose tissue, BCAA metabolism has recently gained renewed interest, as it has been demonstrated that the coordinated regulation of adipose tissue BCAA enzymes in fasting and in feeding may modulate circulating BCAA levels. In this sense, a transcriptional study on subcutaneous adipose tissue biopsies from a cohort of monozygotic twins discordant for obesity reported decreased mitochondrial BCAA catabolism for the obese twins in parallel with the onset of systemic and local adipose tissue inflammation. The down-regulation of BCAA oxidation enzymes was paralleled by decreased levels in plasma of 2-KIC acid and increased levels of leucine. These changes correlated with elevated fasting insulin levels and insulin resistance [21, 22]. Another recent study on type 2 diabetic and obese youth contradicts these findings, indicating the presence of an adaptive metabolic plasticity in youth that equilibrates fatty acid and amino acid metabolism .
We found both decreased secreted levels of 2-KIC and diminished leucine uptake in the obese subcutaneous or visceral fat depots, respectively. 2-KIC is the transamination product of leucine, i.e., a leucine breakdown product. This confirms data from indirect functional in vivo and ex vivo proteomic and transcriptional studies on adipose tissue in the fasting state [24-26]. The dysregulation of leucine metabolism seems to be higher in the visceral obese adipose tissue, eluding the formation of 2-KIC.
Furthermore, we have determined increased released levels of both glutamine and alanine from the obese visceral adipose tissue. Adipose tissue alanine and glutamine released in quantities sufficient to make a significant contribution to the whole body economy of these amino acids has been previously reported [27, 28]. Catabolic pathways of BCAA leucine through the KIC acid route involve the formation of important amounts of alanine, glutamine and glutamate. These pathways are the route for disposal of amino groups released in the transamination of BCAA [8, 27]. Therefore, as alanine and glutamine are highly gluconeogenic amino acids, it could be hypothesized that the increased amount of alanine released by the visceral adipose tissue to the systemic circulation could contribute to hyperinsulinemia and to the development of insulin resistance.
It is worth mentioning that, in the visceral adipose tissue of obese subjects, the uptake of leucine and other reported glucogenic (methionine, serine and threonine) and exclusively ketogenic (leucine) amino acids was found to be diminished. This is indicative of profound deviations in local amino acid metabolism that may determine an increased release in the circulatory flux of amino acid metabolites with gluconeogenic potential, as well as unmetabolyzed BCAA with demonstrated direct insulin-resistant cellular consequences.
Contrary to the spare data available from tissue metabolism studies, there are very interesting reports from recent metabolomic sera studies. Obesity has been associated with hyperaminoacidemia , and, recently the plasma amino acid profile has been found to be altered according to visceral fat accumulation . Generally it is accepted that hyperinsulinemia, a common feature of obesity, has as principal determinant the reduced peripheral metabolism and clearance of insulin . Moreover, there is evidence that amino acids like alanine and glutamine may contribute to hyperinsulinemia through direct pancreatic beta-cell stimulation and gluconeogenesis, affecting also insulin action and glucose cellular uptake [32-35]. The role of exogeneous amino acid overload, particularly of highly represented BCAA, in the etiopathogenesis of metabolic alterations associated with obesity has been under debate for decades [8, 22, 36-38]. Recently, notable differences have been reported in the amino acid metabolic signatures between obese, mild insulin-resistant but nondiabetic subjects and lean individuals , while a decrease in a cluster group of metabolites comprising BCAAs and related analytes predicts improvement in insulin resistance (HOMA-IR), independent of the amount of weight lost . Very recent clinical evidence from the Framingham offspring study indicated increased levels of a group of five BCAA and aromatic amino acids (namely isoleucine, leucine, valine, phenylalanine and tyrosine) as candidate metabolic risk markers for diabetes in obesity, predating the clinical onset of diabetes by years . On the contrary, a high glutamine to glutamate ratio would exert a preventive role in the cardiometabolic status .
Altogether, our findings point to a blunted amino acid metabolism and overload of BCAA catabolism in the inflammated and lipid-overcharged obese adipose tissue. Our results demonstrate that the early stages of obesity, including the presence of systemic inflammation and insulin resistance, before the clinical onset of significant metabolic alterations other than the BMI, are characterized by a markedly affected adipose tissue- amino acid metabolism and secretion. The contribution of the obese adipose tissue-emerged metabolites to the altered systemic amino acid pool and to the onset of metabolic carbohydrate disturbances awaits further functional studies.
We thank Y. Esteban for technical help.
- 1Relation of body fat distribution to metabolic complications of obesity. J Clin Endocrinol Metab 1982;54/2:254-260., , , et al.
- 4Role of lipid oxidation in pathogenesis of insulin resistance of obesity and type II diabetes. Diabetes 1987;36/11:1341-1350., , , et al.
- 16Metabolite identification in blood plasma using CG/MS and the agilent fiehn GC/MS metabolomics RTL library. Agilent Application Note 2009., .
- 18Interorgan amino acid transport and its regulation. J Nutr 2003;133/6(Suppl 1):2068S-2072S..
- 20Hepatic glutathione and nitric oxide are critical for hepatic insulin-sensitizing substance action. Am J Physiol Gastrointest Liver Physiol 2003;284/4:G588-G594., , , et al.
- 26Regional muscle and adipose tissue amino acid metabolism in lean and obese women. Am J Physiol Endocrinol Metab 2002;282/4:E931-E936., , , et al.
- 27Leucine degradation and release of glutamine and alanine by adipose tissue. J Biol Chem 1980;255/17:8074-8081., .
- 28Amino acid metabolism in human subcutaneous adipose tissue in vivo. Clin Sci (Lond) 1991;80/5:471-474., , , et al.
- 30Plasma amino acid profile is altered by visceral fat accumulation and is a predictor of visceral obesity in humans. From Nature Proceedings Available at: http://hdl.handle.net/10101/npre.2011.5560.1 (2011)., , , et al.
- 33Metabolic effects of amino acid mixtures and whey protein in healthy subjects: studies using glucose-equivalent drinks. Am J Clin Nutr 2007;85/4:996-1004., , .
- 37Insulin and hyperaminoacidemia regulate by a different mechanism leucine turnover and oxidation in obesity. Am J Physiol 1996;270/2(Part 1):E273-E281., , .