Potential conflict of interest: Nothing to report.
Diabetes is characterized by high blood glucose levels and dyslipidemia. Bile salt sequestration has been found to improve both plasma glycemic control and cholesterol profiles in diabetic patients. Yet bile salt sequestration is also known to affect triglyceride (TG) metabolism, possibly through signaling pathways involving farnesoid X receptor (FXR) and liver X receptor α (LXRα). We quantitatively assessed kinetic parameters of bile salt metabolism in lean C57Bl/6J and in obese, diabetic db/db mice upon bile salt sequestration using colesevelam HCl (2% wt/wt in diet) and related these to quantitative changes in hepatic lipid metabolism. As expected, bile salt sequestration reduced intestinal bile salt reabsorption. Importantly, bile salt pool size and biliary bile salt secretion remained unchanged upon sequestrant treatment due to compensation by de novo bile salt synthesis in both models. Nevertheless, lean and db/db mice showed increased, mainly periportally confined, hepatic TG contents, increased expression of lipogenic genes, and increased fractional contributions of newly synthesized fatty acids. Lipogenic gene expression was not induced in sequestrant-treated Fxr−/− and Lxrα−/− mice compared with wild-type littermates, in line with reports indicating a regulatory role of FXR and LXRα in bile salt–mediated regulation of hepatic lipid metabolism. Conclusion: Bile salt sequestration by colesevelam induces the lipogenic pathway in an FXR- and LXRα-dependent manner without affecting the total pool size of bile salts in mice. We speculate that a shift from intestinal reabsorption to de novo synthesis as source of bile salts upon bile salt sequestration affects zonation of metabolic processes within the liver acinus. (HEPATOLOGY 2010.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Diabetes is a multifactorial disease characterized by increased fasting blood glucose levels and dyslipidemia—that is, high plasma triglyceride (TG) and low-density lipoprotein cholesterol levels. Controlling blood glucose and cholesterol levels in diabetic patients is critical for delaying the progression of clinical complications such as neuropathy and cardiovascular disease. An efficient way to reduce plasma cholesterol levels is to induce cholesterol secretion in bile, either as bile salt or as free cholesterol. Bile is secreted into the ileum to facilitate absorption of lipids and lipid-soluble vitamins. About 95% of secreted bile salts are reabsorbed in the terminal ileum and transported back to the liver through the portal vein (enterohepatic circulation). In addition to their function in the absorption of dietary fats, bile salts are signaling molecules that play an important role in the regulation of lipid metabolism.1 Interestingly, bile salt metabolism is affected in diabetes, which might contribute to the altered lipid profile observed in diabetic patients.2 Knowledge of potential disturbances in bile salt metabolism in type 2 diabetic humans and animal models is still very limited, however.3
Increasing fecal bile salt loss by preventing their intestinal reabsorption (sequestration) increases bile salt synthesis and, hence, hepatic cholesterol turnover. As a consequence, low-density lipoprotein cholesterol levels are reduced in hyperlipidemic subjects.4, 5 Interestingly, bile salt sequestration also improves glucose levels in type 2 diabetic patients.6–8 Yet use of bile salt sequestrants has been associated with elevated plasma TG levels.9, 10 Bile salt feeding, on the other hand, has been shown to improve plasma lipid profiles in these patients.11, 12 The regulation of the interrelationship between bile salt and lipid metabolism is still only partly understood. At a molecular level, a key regulatory role is assigned to the bile salt–activated nuclear receptor FXR (NR1H4).13 Pharmacological activation of FXR has been shown to improve hypertriglyceridemia in mouse models of insulin resistance,14, 15 whereas Fxr−/− mice have increased serum TG levels.16 Moreover, administration of the natural FXR-ligand cholate improved plasma TG levels of high-fat diet–fed mice through SHP-dependent modulation of the lipogenic gene Srebp1c.17 In the same study, it was shown that the nuclear oxysterol receptor LXRα (NR1H3) is involved in the regulation of lipogenic gene expression upon bile salt feeding.
At a physiological level, bile salt–activated signaling pathways are regulated by bile salt concentrations in the liver. We hypothesized that an altered flux of bile salts returning to the liver underlies, at least in part, the consequences on hepatic metabolism observed upon bile salt sequestration. We quantitatively assessed the kinetics of bile salt and hepatic fatty acid metabolism in lean C57Bl/6J mice and in obese and diabetic db/db mice treated with the bile salt sequestrant colesevelam HCl.18 Additionally, we studied the contribution of FXR and LXRα to sequestrant-induced changes in lipogenic gene expression.
Bile salt sequestration reduced intestinal reabsorption of bile salts by 30%. Nevertheless, the bile salt pool size remained unchanged in both models due to a compensatory increase in de novo synthesis of bile salts. Remarkably, sequestrant treatment significantly increased hepatic TG contents, which primarily accumulated in periportal areas. Expression levels of lipogenic genes as well as the fractional contribution of de novo synthesized fatty acids were increased. This lipogenic response appeared to be FXR- and LXRα-dependent. We speculate that a shift from reabsorption to de novo synthesis as the source of biliary bile salts underlies the lipogenic phenotype observed upon bile salt sequestration.
CA, cholate; CDCA, chenodeoxycholate; FXR, farnesoid X receptor; GC/MS, gas chromatography/mass spectrometry; LXRα, liver X receptor α; TG, triglyceride.
Materials and Methods
Male lean C57Bl6/J and obese, diabetic db/db mice on a C57Bl6/J background (B6.Cg-m +/+ Leprdb/J) were purchased from Charles River Laboratories (L'Arbresle, France, and Brussels, Belgium, respectively). Fxr−/−19 and Lxrα−/−20 mice were generated as described. All animals were housed individually in a temperature- and light-controlled facility. Mice were fed commercially available laboratory chow (RMH-B; Arie Blok, Woerden, The Netherlands)—supplemented with 2% (wt/wt) colesevelam HCl (Daiichi Sankyo, Inc., Parsippany, NJ) when indicated—for 2 weeks. Mice were used for experimental procedures at 12 weeks of age. All experiments were approved by the Ethical Committee for Animal Experiments of the University of Groningen.
Postprandial blood glucose levels were measured at the start of the experiment and subsequently after 1 week and after 2 weeks of treatment. Additionally, body weight and food intake were determined at these time points. [1-13C]-acetate (2% wt/vol in drinking water) was provided for 24 hours (7 AM to 7 AM), starting at day 13 of the experiment. Blood spots were collected from the tail on filter paper (Schleicher and Schuell No2992, ‘s Hertogenbosch, The Netherlands) before and after administration of the label. Blood spots were air-dried and stored at room temperature until analysis. After 2 weeks on the diets, mice were sacrificed by way of heart puncture under isoflurane anesthesia. Plasma was stored at −20°C until analyzed. The liver was removed, weighed, and snap-frozen in liquid nitrogen. The intestine was excised, flushed with cold (4°C) phosphate-buffered saline, and snap-frozen in liquid nitrogen. Both the liver and the intestine were stored at −80°C until biochemical analysis and RNA isolation.
In a separate experiment, 400 μg [2H4]-cholate (in 0.5% NaHCO in phosphate-buffered saline [pH = 7.4]) was intravenously administered on day 10 of the 2-week period. Subsequently, retro-orbital blood samples (75 μL) were obtained at 12, 24, 36, 48, and 60 hours after injection of [2H4]-cholate in chow-fed lean and db/db mice. A pilot study in colesevelam-treated animals indicated that, as expected, turnover of [2H4]-cholate was markedly increased. Therefore, blood samples were obtained at 12, 18, 24, 30, and 36 hours after administration of [2H4]-cholate in colesevelam-treated lean and db/db mice. Plasma was stored at −20°C until analyzed. Feces were collected over the 60-hour experimental period and, after air-drying, kept at room temperature until analysis. After 60 hours, mice were anesthesized through intraperitoneal injection of Hypnorm (1 mL/kg) and Diazepam (10 mg/kg) and subjected to gallbladder cannulation for 30 minutes. During bile collection, body temperature was stabilized using a humidified incubator. Bile was stored at −20°C until analyzed. Animals were sacrificed by cervical dislocation.
Blood glucose concentrations were measured using EuroFlash test strips (LifeScan Benelux, Beerse, Belgium). Hepatic lipids were extracted according to Bligh and Dyer.21 Plasma and liver TG and cholesterol contents were determined using commercially available kits (Roche Diagnostics, Mannheim, Germany; DiaSys Diagnostic Systems, Holzheim, Germany). Plasma-free fatty acids were determined using a NEFA-C kit (Waco Chemicals, Neuss, Germany). Pooled plasma samples from each group were used for lipoprotein separation by fast protein liquid chromatography on a Superose 6 column using an Akta Purifier (GE Healthcare, Diegem, Belgium). Triglycerides in each fraction were determined. Total bile salts in bile and feces were determined by an enzymatic fluorimetric assay.22 Liver morphology was assessed by Masson’s trichrome staining of parafin-embedded material.
Gas Chromatography/Mass Spectrometry Analysis and Mass Isotopomer Distribution Analysis.
Biliary and fecal bile salts were determined by way of gas chromatography as described.23 The isotope dilution technique as well as the preparation of plasma samples for analysis of bile salts by gas chromatography/mass spectrometry (GC/MS) were described in detail by Hulzebos et al.23 Fecal neutral sterols were analyzed as described.24 Labeling of acetyl-coenzyme A pools with orally provided [1-13C]-acetate was described by Jung et al.25 Cholesterol was extracted from blood spots and prepared for GC/MS analysis as described.26 Lipids in liver homogenates were hydrolyzed in HCl/acetonitril. Fatty acids were extracted in hexane and converted to their pentafluorobenzyl derivatives. The fatty acid–pentafluorobenzyl isotopomer patterns (mass fragments C16:0 m/z 255–259, C18:0 m/z 283–287, C18:1 m/z 281–285) were analyzed using a Agilent 5975 series GC/MS (Agilent Technologies, Santa Clara, CA). GC/MS measurements of fatty acids and mass isotopomer distribution analyses were performed essentially as described.27, 28 See also Supporting Materials and Methods.
Total RNA was isolated from liver and intestine using TRI-reagent (Sigma, St. Louis, MO) according to the manufacturer's protocol. Complementary DNA was produced as described.29 Real-time polymerase chain reaction was performed with a 7900HT FAST system using FAST PCR master mix and MicroAmp FAST optical 96-well reaction plates (Applied Biosystems Europe, Nieuwekerk ad IJssel, The Netherlands). Primer and probe sequences have been published before (www.labpediatricsrug.nl). Polymerase chain reaction results were normalized to 18S (liver) and β-actin (intestine).
All values are presented as the mean ± standard deviation. Statistical analysis was assessed using the Mann-Whitney U test (SPSS 12.0.1 for Windows). P values were corrected for multiple comparison errors. Level of significance was set at P < 0.05.
Effects of Colesevelam Treatment on Food Intake, Body Weight, and Plasma Metabolites in Lean and db/db Mice.
Lean and db/db mice were treated with the bile salt sequestrant colesevelam for 2 weeks. Food intake was increased in colesevelam-treated lean and db/db mice during treatment compared with untreated controls (Table 1). Body weight gain was unaffected in colesevelam-treated lean mice but decreased in colesevelam-treated db/db mice. Blood glucose levels increased in control db/db mice during the 2-week intervention period but remained stable in colesevelam-treated db/db mice. Blood glucose levels remained unchanged upon treatment in lean mice. Fasting insulin levels (not shown) were unchanged and decreased, respectively, in colesevelam-treated lean and db/db mice. Nonesterified fatty acid and very low-density lipoprotein TG levels (Supporting Fig. 1) were significantly reduced in colesevelam-treated db/db mice compared with untreated controls but remained unchanged in lean mice.
Table 1. Basal, Plasma, and Liver Parameters of Lean and db/db Mice (Control and Colesevelam-Treated)
Values are presented as the mean ± standard deviation (n = 6 animals per group).
Biliary Bile Salt Flux and Cholate Pool Size Remain Unchanged in Colesevelam-Treated Lean and db/db Mice.
Control db/db mice showed increased feces production and a higher fecal bile salt output, representing hepatic bile salt synthesis, compared with lean controls (Fig. 1A,B). As expected, colesevelam treatment led to massive increases in fecal bile salt output (Fig. 1B). Untreated lean and db/db mice had similar bile flow rates and biliary bile salt output rates (Fig. 1C,D) that remained unchanged in both models upon sequestrant treatment.
Direct end products of de novo bile salt synthesis are the primary bile salts cholate (CA) and chenodeoxycholate (CDCA). Modifications of these bile salts in the liver and intestine give rise to differentially structured primary and secondary bile salts, respectively. Supporting Table 1 provides details on biliary and fecal bile salt compositions. In short, sequestrant treatment resulted in a strongly increased relative content of fecal deoxycholate in both groups. Cholate remained the major biliary bile salt species in both models upon sequestrant treatment. Next, we determined relevant kinetic parameters of CA,23 the major primary bile salt species in mice. Untreated db/db mice displayed a larger pool size and a higher synthesis rate of CA compared with untreated lean mice (Fig. 2). Importantly, CA pool size remained unchanged upon colesevelam treatment in both models. Synthesis rates of CA were massively increased upon sequestrant treatment (+375% and +172%, lean and db/db mice, respectively) and completely compensated for the increased fecal bile salt loss induced by colesevelam. The calculated amount of CA reabsorbed from intestines of colesevelam-treated lean and db/db mice was reduced by about 30% compared with untreated controls (Fig. 2D). Decreased plasma bile salt levels further reflect a reduced flux of bile salts returning to the liver (Fig. 2E).
To gain insight into colesevelam-induced changes in total bile salt pool composition and synthesis of bile salts derived from the primary bile salt species CA and CDCA, we calculated the amount of CA- and CDCA-derived bile salts in the pool as well as their synthesis rates (for details on calculation, see Supporting Materials and Methods). Upon sequestrant treatment, the total pools of bile salts remained unchanged in both models (Fig. 3A). Nevertheless, the pool size of CDCA-derived bile salts was decreased. The synthesis of CA-derived bile salts was massively increased, whereas synthesis of CDCA-derived bile salts remained unchanged in sequestrant-treated mice compared with untreated controls (Fig. 3B). Corresponding with increased CA synthesis, hepatic expression of Cyp7a1, encoding the first and rate-limiting enzyme in bile salt synthesis, and Cyp8b1, encoding the specific enzyme required for CA synthesis, were massively increased. Cyp27a1 directs bile salt synthesis toward CDCA. Although expression of Cyp27a1 was increased in colesevelam-treated mice, this was not reflected in increased CDCA synthesis (Fig. 3C). Despite the fact that bile salt reabsorption was not completely abolished, expression levels of the FXR target gene Fgf15 were undetectable in distal ilea of colesevelam-treated lean and db/db mice (Fig. 3D).
Cholesterol synthesis is massively increased in colesevelam-treated lean and db/db mice, and colesevelam treatment increased fecal cholesterol excretion (Fig. 4A). Together with a strongly increased synthesis of bile salts, this finding translates into an increased turnover of cholesterol. However, this did not result in reduced plasma concentrations or hepatic contents of cholesterol (Fig. 4B,C). Increased hepatic expression of HmgCoAr, encoding the rate-controlling enzyme in cholesterol synthesis, and of Ldlr (Fig. 4D) indicated the anticipated hepatic compensatory response in cholesterol metabolism after colesevelam treatment. To quantify this, the fraction of newly synthesized cholesterol was determined by analysis of the incorporation of [1-13C]-acetate into plasma cholesterol. Fractional cholesterol synthesis was indeed robustly increased in colesevelam-treated mice (Fig. 4E).
Hepatic TG Content Is Increased Due to Enhanced Contribution of Lipogenesis in Colesevelam-Treated Mice.
Both colesevelam-treated lean and db/db mice had modestly increased (lean +50%, db/db +23%) hepatic TG contents compared with untreated controls (Table 1). Remarkably, fat accumulated primarily in periportal areas upon bile salt sequestration (Fig. 5A,B). Increased hepatic expression of key lipogenic genes (Srebp1c, Acc1, Fas, and Scd1) (Fig. 5C) was highly suggestive of enhanced synthesis of fatty acids. Indeed, the total fractions of newly synthesized C16:0, C18:0, and C18:1, as determined by incorporation of [1-13C]-acetate followed by mass isotopomer distribution analysis, confirmed that synthesis of these major hepatic fatty acid species was increased. Additionally, we calculated the contribution of de novo synthesis and chain elongation to the total fractional C18:0 and C18:1 synthesis.28 The increased total fraction of newly synthesized fatty acids was mainly attributable to increased chain elongation in colesevelam-treated lean and db/db mice (Fig. 5D).
Lipogenic Gene Expression Upon Sequestrant Treatment Is FXR- and LXRα -Dependent.
Bile salt–mediated changes in expression of one of the major regulators of lipogenesis, Srebp1c, have been reported to be regulated by both FXR- and LXRα-regulated pathways.17 Surprisingly, expression levels of well-defined FXR and LXRα target genes were differentially or not at all affected in colesevelam-treated lean and db/db mice (Supporting Figs. 3 and 4). To further address the role of hepatic FXR and LXRα in the lipogenic response to bile salt sequestration, Fxr−/− and Lxrα−/− mice and wild-type littermates were treated with colesevelam for 2 weeks. The key lipogenic genes Srebp1c, Acc1, Fas, and Scd1 were significantly increased in livers of sequestrant-treated wild-type mice compared with untreated controls (Fig. 6). Lipogenic genes, however, were barely affected in sequestrant-treated Fxr−/− and Lxrα−/− mice. These results support earlier observations of the regulatory roles of these nuclear receptors in the response to bile salt–mediated changes in lipid metabolism.17
This paper reports novel insights in the interrelationship between bile salt and lipid metabolism in lean and diabetic db/db mice treated with the bile salt sequestrant colesevelam. To the best of our knowledge, this is the first report to quantitatively show that, despite massively induced fecal bile salt loss upon sequestrant treatment, bile salt pool sizes and biliary bile salt secretion rates remain unaffected. Additionally, we show that bile salt sequestration induces hepatic fatty acid synthesis and elongation. An altered hepatic bile salt gradient due to decreased reabsorption but increased de novo synthesis of bile salts likely affects specific aspects of hepatic bile salt signaling. The lipogenic response appears to be dependent on FXR and LXRα signaling, as was evident from studies in the respective knockout mice.
Knowledge of possible disturbances in bile salt metabolism in type 2 diabetic humans and animal models is very limited.3 To our knowledge, this study reports the first data on kinetic alterations of bile salt metabolism in diabetic db/db mice and shows that db/db mice have an increased pool size and synthesis rate of bile salts compared with lean controls. As suggested for db/db mice15 and liver-specific insulin receptor knockout mice,30 disturbed hepatic insulin signaling may directly contribute to changes in bile salt synthesis. Indeed, insulin was shown to reduce plasma bile salts in type 1 diabetic rats,31 possibly through FOXO1-mediated regulation of Cyp7a1.32 Further studies beyond the scope of this study are needed to further unravel underlying mechanisms of disturbed bile salt metabolism in type 2 diabetes.
We observed that db/db mice responded favorably to sequestrant treatment: blood glucose levels stabilized, whereas nonesterified fatty acid and very low-density lipoprotein–TG levels decreased. These parameters were unchanged in lean mice. Importantly, the pool size of the primary bile salt species CA as well as the total pool size of bile salts remained unchanged in sequestrant-treated lean and db/db mice. Remarkably, only the synthesis of CA was massively increased: synthesis of CDCA-derived bile salts was not affected at all. In humans, an increased CA-to-CDCA ratio would result in a more hydrophilic bile salt pool that has been associated with decreased susceptibility for gallstone disease.33 Colesevelam treatment might therefore be beneficial for prevention of gallstone formation in type 2 diabetic humans who have an increased prevalence of gallstones.34 Bile salt reabsorption was reduced by 30% in both models. Although bile salt reabsorption was not fully impaired, Fgf15 expression levels were not detectable in the ilea of sequestrant-treated wild-type mice. It is possible that bile salt sequestration decreases the cellular content of bile salts below a certain threshold value necessary to activate FXR in enterocytes, as observed in an in vivo study in rabbits.35
Interestingly, hepatic TG contents of colesevelam-treated lean and db/db mice were enhanced, which appeared to be mediated by an increased de novo synthesis of hepatic fatty acids and chain elongation. In contrast to our data, other studies addressing the effects of bile salt sequestration on lipid metabolism showed that bile salt sequestration prevented TG accumulation in the liver.36, 37 It should be realized that those studies were performed in high-fat diet–fed mice in which the beneficial effects of bile salt sequestration are likely partly attributable to sequestrant-induced malabsorption of lipids. In addition, strain-specific responses to sequestrant treatment cannot be ruled out.
At a molecular level, the interrelationship between bile salt and lipid metabolism is generally accepted to be mediated by FXR. Nevertheless, data to explain the exact mechanisms of this relationship are still very inconsistent. Pharmacological activation of FXR has been shown to reduce free fatty acid levels in insulin-resistant rodents.15, 38 Absence of FXR signaling in Fxr−/− mice leads to increased very low-density lipoprotein–TG levels in plasma of these mice,39 suggestive of a role for FXR in control of very low-density lipoprotein assembly.
Colesevelam treatment induced hepatic expression levels of the lipogenic gene Srebp1c in lean and db/db mice. Hepatic expression levels of the lipogenic gene Srebp1c were reduced in Fxr−/− mice compared with controls.39, 40 Conversely, FXR activation was also shown to repress the expression of Srebp1c in a pathway involving SHP.17, 40 Expression levels of the FXR target gene Shp were unaffected and decreased in colesevelam-treated lean and db/db mice, respectively. These results are suggestive of SHP-independent regulation of Srebp1c upon sequestrant treatment. Supportive of SHP-independent regulation of lipogenic gene expression by FXR was the observation that FXR regulates the transcription of the lipogenic gene Fas through direct binding to the Fas promoter.41 Because expression levels of well-known FXR target genes were differentially affected in lean and db/db mice, we studied the role of FXR in the lipogenic response of sequestrant treatment in Fxr−/− mice and found that in contrast to wild-type mice littermates, lipogenic gene expression levels were barely affected. Srebp1c, is strongly regulated by the oxysterol receptor LXRα.42 Increased synthesis of cholesterol, as occurs in sequestrant-treated mice, could possibly lead to increased hepatic levels of oxysterols and, hence, activation of LXRα. Expression levels of established LXRα target genes, however, were unaffected in sequestrant-treated lean and db/db mice, suggestive of unchanged LXRα signaling. Yet, investigation in sequestrant-treated Lxrα−/− mice revealed that lipogenic gene expression was not increased in these mice compared with untreated wild-type littermates. Our results from colesevelam-treated Fxr−/− and Lxrα−/− mice confirm earlier findings that FXR and LXRα are both involved in regulation of bile salt–mediated changes in lipogenic pathways.17 The exact molecular mechanisms through which these nuclear receptors signal regulate the lipogenic response to bile salt sequestration exceed the scope of this report.
At a physiological level, bile salt–mediated signaling pathways are dependent on the concentration of bile salts in the liver acinus. We speculate that the concept of metabolic zonation43 might add to the understanding of the observed hepatic effects upon bile salt sequestration. Hepatocytes localized around the portal vein display different metabolic activities than those lining the central vein; for example, bile salt and fat synthesis are pericentrally localized processes, whereas cholesterol synthesis is performed mainly by portal hepatocytes.43 As we show in the current report, the amount of bile salt molecules reabsorbed in the ilea of colesevelam-treated mice was decreased by ≈30% with a subsequent reduction in plasma bile salt levels and, hence, reduced bile salt signaling in periportal hepatocytes. Newly synthesized bile salts, which accommodate a much larger fraction of the bile salt pool of colesevelam-treated mice compared with controls, are primarily secreted by pericentrally localized cells and possibly exert differential signaling functions. Selective periportal fat accumulation and differentially affected expression levels of hepatic FXR target genes support this hypothesis. Additionally, it was shown that Cyp7a1, which is exclusively expressed in pericentral hepatocytes,44 translocates to a larger area of the liver lobulus with more involvement of periportally localized cells in sequestrant-treated rats.45 Our working model is summarized in Supporting Fig. 5. It should be stressed that this hypothesis requires dedicated investigation.
In conclusion, we show that colesevelam treatment increases lipogenesis and chain elongation in mice that, at least at the level of gene expression, is dependent on FXR and LXRα. A shift from reabsorption to de novo synthesis as the source of biliary bile salts affects the sinusoidal gradient of bile salts.46 This shift modifies the regulation of genes and proteins involved in bile salt synthesis and bile salt–mediated regulation of metabolism and possibly underlies the phenotypical response to colesevelam treatment in mice.
We are indebted to Rick Havinga for excellent contributions to the mouse studies performed. Additionally, we are grateful to Theo Boer and Elles Jonkers for their excellent technical assistance with GC/MS analyses.