The oral glucose tolerance test (oGTT) is a common tool to provoke a metabolic challenge for scientific purposes, as well as for diagnostic reasons, to monitor the kinetics of glucose and insulin. Here, we aimed to follow the variety of physiological changes of the whole metabolic pattern in plasma during an oGTT in healthy subjects in a nontargeted reversed-phase ultra performance liquid chromatography coupled to electrospray ionization quadrupole time of flight mass spectrometric metabolomics approach. We detected 11,500 metabolite ion masses/individual. Applying multivariate data analysis, four major groups of metabolites have been detected as the most discriminating oGTT biomarkers: free fatty acids (FFA), acylcarnitines, bile acids, and lysophosphatidylcholines. We found in detail 1) a strong decrease of all saturated and monounsaturated FFA studied during the oGTT; 2) a significant faster decline of palmitoleate (C16:1) and oleate (C18:1) FFA levels than their saturated counterparts; 3) a strong relative increase of polyunsaturated fatty acids in the fatty acid pattern at 120 min; and 4) a clear decrease in plasma C10:0, C12:0, and C14:1 acylcarnitine levels. These data reflect the switch from β-oxidation to glycolysis and fat storage during the oGTT. Moreover, the bile acids glycocholic acid, glycochenodeoxycholic acid, and glycodeoxycholic acid were highly discriminative, showing a biphasic kinetic with a maximum of a 4.5- to 6-fold increase at 30 min after glucose ingestion, a significant decrease over the next 60 min followed by an increase until the end of the oGTT. Lysophosphatidylcholines were also increased significantly. The findings of our metabolomics study reveal detailed insights in the complex physiological regulation of the metabolism during an oGTT offering novel perspectives of this widely used procedure.
- oral glucose tolerance test
- bile acids
- fatty acids
food intake is a central point of life and one of the major regular challenges for metabolism. Time-dependent variations in the hormonal and metabolic responses to food are tightly balanced and of great importance to human health. These time-dependent variations can be investigated using a standard meal or if the focus is primarily on the metabolic effects of a glucose load by an oral glucose tolerance test (oGTT). The oGTT can be seen as a standardized meal of pure carbohydrates that evokes a variety of physiological responses to carbohydrates. Numerous scientific animal and human studies implemented this standardized setting to investigate metabolic responses to this carbohydrate challenge of the organism (1, 11, 13, 46, 51), and moreover it is also used as a clinical test for the diagnosis of diabetes mellitus (58). Currently, the oGTT is generally applied to follow the time-dependent changes of glucose and insulin, giving the opportunity to draw conclusions on β-cell function, insulin sensitivity, and glucose disposal (1, 11, 37, 58).
To gain new insights in complex metabolic processes, nonselective but specific information-rich analytical approaches are required. Metabolomics is the nontargeted analysis of metabolites typically carried out to generate a specific fingerprint of a current metabolic state at a given time point or to study several time points thereby following the dynamic changes of the metabolic pattern of an organism (32, 33, 35). It is a rapidly advancing field that complements genomics and proteomics, promised to add significant information to the understanding of physiological and pathophysiological processes (20, 33). The very complex data are evaluated by pattern recognition techniques, i.e., multivariate statistic methods (16, 27). Furthermore, metabolomic investigations have the potential to identify molecular species' differentiating physiological states (10). Thus mass or nuclear magnetic resonance spectra of biofluids in metabolomic studies serve in two distinct but closely related modes: as a quantitative metabolic fingerprinting tool and as means of metabolite elucidation.
The aim of our study was to investigate for the first time metabolic pathways of the physiological response during the oGTT by a nontargeted metabolomics approach. Employing state-of-the-art metabolomics techniques for the analysis of plasma samples of healthy individuals collected in the fasting state and at four additional time points during an oGTT, we aimed to discover metabolite biomarkers and pathways that are influenced by this metabolic carbohydrate challenge, thereby opening new perspectives in the study of the physiological reaction of the human body on glucose ingestion.
MATERIALS AND METHODS
HPLC-grade acetonitrile was purchased from Merck (Merck, Darmstadt, Germany). Formic acid was purchased from Tedia (Fairfield, OH). Leucine-enkephalin was obtained from Sigma-Aldrich (St. Louis, MO).
Blood samples were collected after an overnight fasting and during the following oGTT from 16 healthy individuals (age 48.3 ± 2.5 yr) under standardized conditions at the metabolic ward at the University Clinic in Tuebingen (Germany). Furthermore, to study the fatty acid composition of the very low density lipoprotein (VLDL)-triglycerides (TGs), eight individuals underwent a glucose infusion experiment (blood glucose was maintained at 180 mg/dl for 4 h). All individuals had no medications known to influence glucose metabolism. The plasma samples were immediately stored in 500-μl aliquots at −80°C. The subjects were Caucasians from the south western part of Germany. The protocol of the study was approved by the Ethics Committee of the University Tuebingen and conformed to the 1975 Declaration of Helsinki before commencement. All subjects gave written informed consent. The investigation was conducted in accordance with the ethical principles of Good Clinical Practice.
oGTT to calculate the insulin sensitivity index.
All 16 individuals underwent a 75-g oGTT according to the recommendations of the World Health Organization (WHO)/International Diabetes Federation IDF (58). Venous blood samples were obtained directly before (= 0 min) and at 30, 60, 90, and 120 min during the oGTT. Insulin sensitivity was calculated from glucose and insulin values during the oGTT as proposed by Matsuda and DeFronzo (37).
Analysis of routine parameters in blood.
Blood was collected in sodium fluoride (for blood glucose)- and lithium-heparin vacuum containers. Plasma was separated by low-speed centrifugation at 2,000 g for 7 min. Glucose, TGs, cholesterol, and total nonesterified fatty acids were measured by the ADVIA 1650 clinical chemical analyzer; insulin was analyzed with the ADVIA Centaur immunoassay system (both Siemens Healthcare Diagnostics, Fernwald, Germany).
VLDL-TGs fatty acid analysis.
Determination of the VLDL-TG fatty acid composition was performed as recently described (52). In brief, the plasma VLDL fraction was isolated by ultracentrifugation (817,480 g) for 18 h. Next, the VLDL fraction was divided into five subfractions [phospholipids, diacylglycerol, free fatty acid (FFA), TG, and cholesteryl esters] using thin-layer chromatography (TLC). The TG fraction was scraped off the TLC plate, dissolved in methanol-toluol mixture (4:1 vol/vol) and subjected to direct transesterification. The fatty acid methyl esters were quantified with cis-13,16,19-docosatrienoic acid as the internal standard using gas chromatography with a flame ionization detector.
Sample preparation for the metabolomic analysis.
The 500-μl plasma aliquots were thawed, deproteinized with 2 volume parts of acetonitrile (final concentration 66%), centrifuged (13,000 revolutions/min for 20 min), run to dryness in a vacuum centrifuge, and stored at −20°C. For the analysis, the samples were reconstituted in 200 μl acetonitrile and water (2+8) and analyzed by reversed-phase ultra performance liquid chromatography (UPLC) coupled to electrospray ionization quadrupole time of flight (qTOF) mass spectrometry (MS).
Reversed-phase UPLC analysis.
We applied our recently reported UPLC-qTOF-MS approach, established and validated for the metabolomics analysis of serum (60). In brief, the chromatographic separation was performed on a 100 × 2.1 mm ACQUITY 1.7 μm/C18 column using an ACQUITY-UPLC system (Waters). The column was maintained at 35°C, and the gradient program at a flow rate of 0.35 ml/min was 95% of 0.1% formic acid in water for 0.5 min, changed to 100% acetonitrile linearly within 24 min, then held for 4 min, finally back to 95% of 0.1% formic acid in water.
Mass spectrometric procedures.
The UPLC system was coupled to a Micromass qTOF-MS (Manchester, UK) equipped with an electrospray source operating in either positive or negative ion mode. The capillary voltage was set at 3,100 volts in the positive and negative ion mode and the cone voltage to 35 volts. All analyses were acquired using the lock spray to ensure accuracy and reproducibility; leucine-enkephalin was used as the lock mass. The mass spectrometric data were collected from 0 to 30 min in the full scan mode [mass-to-charge ratio (m/z) 100–750]. The metabolites were identified following our recently published analytical strategy for the identification of biomarkers in metabolomics studies, described in detail previously (10).
For an efficient evaluation of the metabolic variability within 120 min after a single oral glucose challenge, mass spectra were digitally analyzed using the Micromass MarkerLynx Applications Manager version 4.0 (Waters, Manchester, UK). The numbers of variables exported from MarkerLynx were ∼11,500 in total (7,800 in the positive and 3,700 in the negative ionization mode). The statistical calculation was performed using the intensity of the metabolite ions. The data were combined into a single matrix by aligning peaks with the same mass and retention time together from each data file in the dataset. The intensity for each peak was normalized to the sum of the peak intensities for each dataset to enable the comparison of the relative mass intensities of metabolites between the different data files.
The preprocessed UPLC-qTOF-MS data were exported into Soft Independent Modeling of Class Analogy-P (version 11.0; Umetrics, Umea, Sweden) for analysis and visualization by multivariate statistical methods. Three representative time points before and during the oGTT were selected for the primary data analysis and identification of biomarkers of this metabolic challenge. We chose the fasted state (0 min; representing the starting point), the time point 30 min after the glucose ingestion where the increase in glucose and insulin is most pronounced, and 120 min after the glucose load since this is the most important time point in the diagnostic oGTT reflecting the “recovery” of the metabolism after a meal. After Pareto scaling and orthogonal signal correction (OSC) filtering according to Wold et al. (59), partial least-squares discriminant analysis (PLS-DA) was applied. The OSC-filtered PLS-DA loading plot was used to identify metabolites with major influence on the group membership. The predictive ability of the model was assessed by internal validation using sevenfold cross-validation and response permutation testing. Clinical chemical and anthropometric data of the individuals were computed using the statistical software packet JMP (SAS Institute, Cary, NC). P < 0.05 was considered significant.
oGTT glucose and insulin levels.
All 16 individuals of the study group underwent an oGTT. Based on the results of the oGTT and according to the criteria of the WHO/IDF [fasting glucose <7.0 mmol/l, 120 min glucose <7.8 mmol/l; (58)], all participants were classified as normal glucose tolerant. Furthermore, the insulin sensitivity calculated according to Matsuda and DeFronzo (37) was determined to be normal (insulin sensitivity index Matsuda = 11.9 ± 1.8). The kinetics of the glucose and insulin levels during the oGTT are displayed in Fig. 1.
Analysis of the plasma metabolic pattern by reversed-phase UPLC-qTOF-MS.
The plasma metabolic pattern of the study group was analyzed by UPLC-qTOF-MS at five time points during an oGTT. To increase the number of detected metabolite ions, we analyzed the 80 samples two times, both in the positive and negative electrospray ionization (ESI) mode. Figure 2A shows a typical base peak intensity chromatogram (BPC) recorded in the positive ionization mode, and the BPC of the negative ionization mode is shown in Fig. 2B. In total, ∼11,500 metabolite ion masses had been detected (∼7,800 in the ESI+ mode and 3,700 ions in the ESI− mode). For the statistical calculation, the intensity of each peak was normalized to the sum of the peak intensity for each dataset thus enabling the comparison of the relative mass intensities of metabolites between the different data files. The preprocessed UPLC-qTOF-MS data were further investigated using multivariate statistical analyses.
Identification of metabolites reflecting changes in the metabolite pattern during the oGTT.
To detect plasma metabolites reflecting metabolic changes at different time points during the oGTT on the basis of the UPLC-qTOF-MS spectra, we constructed a model using PLS-DA with the OSC data filter using randomized sample order, according to Wold et al. (59). Three representative time points of the metabolomics analysis of the oGTT were selected to generate the model, namely the fasted state (0 min), which is directly followed by the time point of the most pronounced increase in glucose and insulin levels (30 min; Fig. 1), and the most important time point in the diagnostic oGTT reflecting the recovery of the metabolism after a meal (120 min after the glucose load). As indicated by the PLS-DA scores plots, the time points 0, 30, and 120 min could be separated into clearly distinct clusters in both the positive (Fig. 3A) and the negative ionization mode (Fig. 3B). The clustering indicates considerable changes of the metabolome in the plasma at each time point represented by 11,500 metabolite ions. To ensure that the calculated model is reliable and the observed clustering is not due to chance, we performed an internal validation using sevenfold cross-validation (17). The estimated goodness of prediction was 0.569 for ESI+ and 0.582 for ESI−, which underlines the robustness of the model.
The next step in the multivariate data analysis, the PLS-DA loading plot to detect potential biomarkers (Fig. 4), shows the m/z values of ions responsible for the separation of the three different time points in the positive (Fig. 4A) and the negative (Fig. 4B) ionization mode. The metabolite ions with the greatest influence on the PLS components and therefore on the clustering in the PLS-DA scores plot (given in Fig. 3) were those metabolite ions with the largest distance from the origin. The most discriminative ions are presented in Table 1 for the positive and Table 2 for the negative ionization mode. Applying our recently developed strategy for the unequivocal identification of biomarkers in metabolomics studies (10), we identified four major groups of metabolites: FFAs, acylcarnitines, bile acids, and lysophosphatidylcholines (lyso-PCs), as well as two amino acids (Table 1 and 2). Because we applied reversed-phase chromatography, highly polar biomarkers, like glucose, could not be detected in our approach. In the next step, the kinetic of the intensity changes of metabolites belonging to the four identified groups were analyzed at all five time points of the oGTT.
FFAs, in particular oleate (C18:1; m/z 281.2 in Fig. 4B), are the most discriminative ions in the negative ESI mode. Besides oleate, we identified also fatty acids C16:0, C16:1, C18:0, C18:2, C20:4, and C22:6 as metabolite biomarkers relevant for the differentiation of the three oGTT time points. Furthermore, we selected additional FFAs (C14:0, C18:3, C20:2, C20:3, C20:5, C22:4, C22:5) based on previous findings (48) and the corresponding monounsaturated fatty acids (MUFAs) C14:1 and C20:1 by extracting their peak intensities from the ESI− BPC. The plasma kinetic, i.e., the strong decrease of the sum of these 16 different fatty acids, was comparable to that of the total FFAs measured by a clinical chemical analyzer (Fig. 5A). Of note, besides the measured reduction of 90% of total FFA (Fig. 5A), we detected distinct differences in the decrease of FFA species during the oGTT depending on their degree of desaturation, as shown in Fig. 5, B and C. MUFA levels C16:1 and C18:1 decreased significantly faster in the plasma than their saturated counterparts C16:0 and C18:0 (Fig. 5, B and C). In general, the pattern of plasma FFA species showed a rapid shift from MUFA toward saturated FFA (SFA; Table 3). Accordingly, the ratio of MUFA to SFA (C14, C16, and C18) decreased by 79% during the oGTT. To the best of our knowledge, a rapid change in this SFA/MUFA-ratio in plasma during an oGTT, demonstrated in Table 3 and Fig. 5, B and C, has not been described so far. A second striking alteration during the oGTT in the plasma pattern of the 16 FFAs is also shown in Table 3. Although the plasma levels of all 16 FFAs decreased from the fasted state (= 0 min) to 120 min after the glucose ingestion, the decrease in polyunsaturated fatty acids (PUFA) levels is less pronounced compared with the strong reduction in SFA and MUFA. The major PUFA C18:2, C20:4, and C22:6 showed only a reduction of 30–40% (data not shown). This resulted in a strong relative increase in long-chain PUFA compared with the SFA and MUFA from 0 to 120 min in the fatty acid plasma pattern (Table 3).
Whether this dramatic change in plasma FFA composition also affects the fatty acid composition of newly synthesized TGs was studied next. Plasma FFAs are rapidly taken up by the liver and incorporated into TG, which can be detected in plasma VLDL within 20–30 min (43). VLDL have a typical plasma half-life of 2–3 h (26). Because this rapid turnover may also result in rapid changes in the VLDL-TG composition during an oGTT, we decided to use an experimental setup of a continuous glucose infusion for these investigations. Blood glucose levels were maintained at 180 mg/dl for 4 h, and the VLDL-TG fatty acid composition was determined in eight healthy subjects in the fasted state immediately before and after the glucose infusion by gas chromatographic analysis. The proportion of MUFA (C16:1 and C18:1) that is strongly reduced in the plasma FFA fraction during the oGTT (Table 3) was also reduced to 89 ± 2.4% (P = 0.004) in VLDL TG after the glucose infusion. Furthermore, the amount of SFA (C14:0, C16:0, and C18:0), which is unchanged in the plasma FFA fraction during the oGTT (Table 3), was also unchanged in the VLDL-TG after the glucose infusion (P = 0.6). Finally, we examined the long-chain PUFAs, which showed the strongest relative increase in the plasma FFA fraction during the oGTT (Table 3). The proportion of PUFAs in VLDL-TG showed an relative increase of 62 ± 13% (P = 0.004) after the glucose infusion. Of note, similar to the FFA composition in plasma, these long-chain PUFAs (C20:3, C20:4, C20:5, C22:4, C22:5, and C22:6) only make up for 2.3% of fasting VLDL-TG fatty acids and 1% of hepatic TG fatty acids (Peter, unpublished observations). The data from the FFA composition in VLDL-TG suggest that the rapid alterations of plasma FFA composition that occur during the oGTT are in fact translated into the TG synthesis leading to an enrichment of long-chain PUFAs in newly synthesized VLDL-TG.
In the positive ESI mode, decanoyl-, dodecanoyl-, and tetradecenoyl-carnitine were identified as discriminative biomarkers between the different time points during oGTT (Fig. 6 and Table 1). We found that the levels of C10:0, C12:0, and C14:1 acylcarnitine drop by 60–70% during the oGTT, reaching statistical significance after 60–90 min (Fig. 6).
Bile acids were highly discriminative in the positive ESI and in the negative ESI mode and appeared as potential biomarkers representing metabolic changes during oGTT. We identified glycocholic acid (m/z 412.28 in ESI+), glycochenodeoxycholic acid (m/z 414.28 in ESI+, m/z 448.29 in ESI−), and glycodeoxycholic acid (m/z 448.29), which account for ∼50% of the total bile acids in normal human plasma (7, 18). The plasma kinetic of all three bile acids showed a uniform, characteristic biphasic pattern during the oGTT, reaching a maximum of a 4.5- to 6-fold increase 30 min after the glucose application and significantly decreased over the next 60 min (Fig. 7). Finally, there is a second increase after 120 min.
Another group of metabolites detected as discriminative markers between the different oGTT time points in the PLS-DA loading plots were the lyso-PCs, showing increasing plasma levels during the oGTT (Fig. 8). Two isomers of C16:0 lyso-PC, C16:1 lyso-PC, and C18:2 lyso-PC increase by 80–100% during the oGTT (0 vs. 120 min). The most prevalent lyso-PCs, C18:0 lyso-PC and C18:1 lyso-PC, increase less pronounced but significantly from 0 to 120 min (∼20%) during the oGTT, whereas C18:0 lyso-PC peaked at 90 min and decreased thereafter (Fig. 8).
The physiological changes of the whole metabolic pattern in plasma during an oGTT in healthy subjects, analyzed by a nontargeted metabolomics approach, revealed four major groups of metabolites as the most discriminating oGTT biomarkers: FFA, acylcarnitines, bile acids, and lyso-PCs.
The plasma pattern of FFA showed a strong decrease of all SFA and MUFA studied during the oGTT and a significant faster decline of C16:1 and C18:1 FFA levels than their saturated counterparts. SFA have been shown to exert lipotoxic effects like the induction of apoptosis, inflammation, and endoplasmatic reticulum stress that are associated with metabolic diseases (6, 15, 44, 47, 52, 57). Because MUFAs are reported to have the ability to prevent the lipotoxic effects of SFA (6, 34, 49), our finding may be of particular interest in this context. The identified shift from MUFA toward SFA during the oGTT might be a potential mechanism for the lipotoxic effects of SFA independent from the absolute concentration of SFA. A possible explanation for these changes in plasma FFA composition could be a more efficient suppression of lipolysis of TGs containing C16:1 and C18:1 than their saturated counterpart. In fact, insulin inhibits hormone-sensitive lipase activity against TGs and cholesterol esters, but not against diglycerides (30). Thus, because TGs contain substantially more MUFAs and less SFAs than diglycerides (30), this may contribute, at least in part, to the selective decrease in plasma MUFAs during the oGTT. Additionally, a preference of some enzymes in TG synthetic pathways for unsaturated fatty acid substrates (34) may lead to a more rapid clearance of MUFAs from the circulating FFA pool that results in increased MUFA content in TGs compared with plasma FFA (22).
Another interesting finding of our study was the less pronounced decrease in PUFA plasma levels compared with the strong reduction in SFA and MUFA, yielding in a reduction of 30–40% of the major PUFA (C18:2, C20:4, and C22:6) compared with the 90% reduction of total FFA. PUFAs are the precursors of so-called eicosanoids, proinflammatory signaling molecules like prostaglandins, thromboxanes, and leukotrienes. They are typically released from the A2 position of phospholipids by phospholipase A2. Unlike hormone-sensitive lipase, which releases the majority of FFA from adipose tissue TGs during fasting, phospholipase A2 is not inhibited by insulin. In fact, the strong relative increase (up to 4.85-fold for C20:4; 4.5-fold for C20:5) in PUFAs within the fatty acid plasma pattern at 120 min does not reflect an absolute increase but lack of suppression during the oGTT. The continued release of PUFAs due to continued phospholipase A2 activity may also result in an increased rate of PUFA incorporation in newly synthesized TGs. Because FFAs are rapidly taken up by the liver and incorporated into VLDL-TG within less than one hour (42), accordingly we could detect a significant, >60% increase of PUFAs in VLDL-TG after the glucose infusion in our experimental setting.
Another class of metabolite biomarkers detected in our nontargeted metabolomics study are the acylcarnitines. The detected C8:0-, C10:0-, C12:0-, and C14:1-carnitines dropped by 60–70% from the fasted state to the time point 120 min of the oGTT. Acylcarnitine esters are generated from their respective acyl-CoA intermediates by carnitine acyltransferases, which are primarily localized in mitochondria catalyzing the exchange of CoA for carnitine. Long-chain fatty acids are activated to CoA esters in the cytosol and shuttled into the mitochondria by a carnitine-dependent transport system where the β-oxidation proceeds. Acylcarnitines represent byproducts of mitochondrial β-oxidation, but, in contrast to acyl-CoAs, which cannot cross the mitochondrial membrane, the acylcarnitines can pass efficiently into the cytosol and subsequently into the bloodstream. Therefore, the plasma acylcarnitines reflect the substrate flux through β-oxidation and are commonly used to detect inborn metabolic disorders (54). Because the oGTT starts in a fasting situation with fatty acid β-oxidation as the major energy source, where the intramitochondrial flux of acyl-CoA esters is increased and elevated plasma acylcarnitine levels can be observed (5, 12, 19, 21), we detected a high amount of acylcarnitines in plasma. As glucose and insulin levels rise, fatty acid β-oxidation and lipolysis are essentially blocked, and carbohydrates become the major energy source. Thus decreasing plasma acylcarnitine levels reflect the switch from fatty acid β-oxidation to glycolysis.
The bile acids glycocholic acid, glycochenodeoxycholic acid, and glycodeoxycholic acid, which represent the next discriminative metabolite class in our approach, showed a uniform, characteristic biphasic kinetic during the oGTT, reaching a maximum increase 30 min after the glucose application, significantly decreasing over the next 60 min, and finally increasing again within the last 30 min of the oGTT. The levels of plasma bile acids are determined by bile release from the gallbladder, intestinal absorption, and hepatic extraction. During a meal, the gallbladder releases the bile into the small intestine. After reabsorption, they recirculate to the liver. In healthy individuals, 70–90% of reabsorbed bile acids are cleared from the portal vein blood by an efficient first pass extraction in the liver (7, 56). The bile release is stimulated by sensory gustatory sensation of food, by gastric filling, and most potently by the intestinal hormone cholecystokinin (CCK). In the regulation of bile acid synthesis, insulin may play a key role, since changes in bile acid pool size were directly dependent on the insulin deficiency state and were reversed with insulin treatment in different diabetic animal models (38, 55). Additionally, insulin is involved in the modulation of the hepatobiliary bile acid secretion and the composition of the bile acid profile, very recently reported by Biddinger et al. (3). They found an impaired synthesis of bile acids and a high expression of the biliary cholesterol efflux pump ABCG5/ABCG8 in liver-specific insulin receptor knockout mice resulting in a lithogenic bile salt profile and the formation of gallstones (3). Of note, since the composition of the bile acid pool between mice and humans is different [being more hydrophobic in humans (23)] and patients with different forms of diabetes appear to have differences in the bile acid profile (45), further human studies will be required to determine how insulin influences the human bile acid metabolism and profile under physiological and pathophysiological conditions.
The postprandial increase of plasma bile acids is well known; however, a rapid and strong increase after glucose ingestion like in the oGTT was unexpected, since the secretion of CCK, the most potent stimulus for bile acid release, is stimulated by the presence of fatty acids and amino acids in the chymus but not by carbohydrates (24). Therefore, the reason for the release and biphasic plasma kinetic of the plasma bile acids during an oGTT is currently unknown. However, recently novel functions of bile acids as metabolic integrators of whole body energy homeostasis that influence glucose and lipid metabolism have been discovered (2, 4, 8, 28, 36, 56). Because of an efficient extraction in the liver, only micromolar concentrations of bile acids are present in the peripheral blood circulation, and a systemic physiological role was unexpected. The actions of bile acids on glucose and lipid metabolism in extrahepatic tissues are mediated via two distinct pathways: the nuclear farnesoid X receptor and the newly discovered G protein-coupled TGR5 receptor (56, 61). Bile acids were shown to lower TG levels, inhibit gluconeogenesis, activate glycogen synthesis, promote mitochondrial activity, energy expenditure, thermogenesis, and finally improve insulin sensitivity (4, 8, 28, 36, 39, 50). Because bile acids have now evolved from detergents that emulsify nutrient lipids and remove cholesterol from the body to important metabolic signaling molecules (3, 25, 39, 50, 61), the speculation that the biphasic plasma kinetic of the bile acids glycocholic acid, glycochenodeoxycholic acid, and glycodeoxycholic acid may reflect the tight regulation and function of these new players in the orchestra of metabolic acting molecules of glucose homeostasis seems plausible, but needs to be proven.
Several other studies have proposed lyso-PCs as potential biomarkers (14, 53, 62), but they may also be considered as “usual suspects,” occurring as marker molecules under many different physiological and pathophysiological conditions like impaired endothelial function (9, 29), colorectal cancer (53, 62), and in septic patients where they are inversely correlated with mortality (14). Plasma lyso-PCs mainly originate from hydrolysis of phosphatidylcholine by phospholipase A2 and transesterification by lecithin-cholesterol acyltransferase. Depending on the chain length and saturation, lyso-PCs have been shown to induce an inflammatory response and superoxide production in leukocytes (41). These proinflammatory effects are stronger for unsaturated than for saturated lyso-PCs. However, where do the lyso-PCs that appear in the plasma during the oGTT originate from? Plasma lyso-PCs freely exchange with erythrocyte membranes, and alterations are rapidly transferred between both compartments (31). In plasma, lyso-PCs occur bound to albumin or α1-acidic glycoprotein as well as in a free unbound form that is biologically active (40). The albumin-binding capacity is limited, and lyso-PCs compete with FFA for albumin binding (40). As plasma FFAs decrease by 90% in the oGTT, a part of the observed increase in lyso-PC levels could be the net result of a redistribution of lyso-PCs from erythrocyte membranes to vacated plasma albumin-binding capacity without involving de novo synthesis. The applied method for sample pretreatment does not discriminate between free and albumin-bound lyso-PCs; therefore, we cannot distinguish if the increase in plasma lyso-PC results in increased biological activity or reflects redistribution from plasma membranes to albumin-bound biologically inactive forms.
In conclusion, analysis of the kinetic changes of the plasma metabolome of normal subjects induced by a single carbohydrate challenge by an oGTT reveals significant metabolic alterations. The detected metabolite biomarkers reflect a switch of the body's energy source from β-oxidation of FFA released from the adipose tissue in the fasted state toward glycolysis and fat storage. We demonstrate a faster elimination of MUFA from the bloodstream compared with SFA, as well as a rapid change in the patterns of plasma FFA and the fatty acid composition of newly synthesized TGs. Elevated plasma bile acids may modulate this adaptation through their metabolic actions and glucose-sensitizing properties that promote energy expenditure, mitochondrial activity, and insulin sensitivity. Our findings offer new insights in the complex physiological regulation of the metabolism during an oGTT.
This work was supported by the Kompetenznetz Diabetes mellitus (Competence Network for Diabetes mellitus) funded by the Federal Ministry of Education and Research (FKZ 01GI0803-04), by a grant from the Ministry of Education, Youth, and Sport of the Czech Republic (MSM0021627502), the foundation for Distinguished Young Scholars (No. 20425516), the foundation from the National Natural Science Foundation of China (No. 20675082), the Sino-German Center for Research Promotion (DFG and NSFC, GZ 364), the National Basic Research Program of China (No. 2006CB503902), and the National Key Project of Scientific and Technical Supporting Programs (No. 2006BAK02A12) from the State Ministry of Science & Technology of China.
We thank Heike Runge and Iris Mertens for excellent technical assistance.
↵* X. Zhao and A. Peter contributed equally to this work.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2009 the American Physiological Society