Fructose stimulates GLP-1 but not GIP secretion in mice, rats, and humans

Fructose stimulates GLP-1 but not GIP secretion in mice, rats, and humans. Am J often stimulate gut hormone secretion, but the effects of fructose are incompletely under-stood. We studied the effects of fructose on a number of gut hormones with particular focus on glucagon-like peptide 1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP). In healthy humans, fructose intake caused a rise in blood glucose and plasma insulin and GLP-1, albeit to a lower degree than isocaloric glucose. Cholecystokinin secretion was stimulated similarly by both carbohydrates, but neither peptide YY 3–36 nor glucagon secretion was affected by either treatment. Remarkably, while glucose potently stimulated GIP release, fructose was without effect. Similar patterns were found in the mouse and rat, with both fructose and glucose stimulating GLP-1 secretion, whereas only glucose caused GIP secretion. In GLUTag cells, a murine cell line used as model for L cells, fructose was metabolized and stimulated GLP-1 secretion dose-dependently (EC 50 (cid:2) 0.155 mM) by ATP-sensitive potassium channel closure and cell depolarization. Because fructose elicits GLP-1 secretion without simultaneous release of glucagonotropic GIP, the pathways underlying fructose-stimulated GLP-1 release might be useful targets for type 2 diabetes mellitus and obesity drug development.

enteroendocrine axis; gastric inhibitory peptide; glucagon-like peptide-1 THE GUT IS THE LARGEST hormone-producing organ of the body. While it has long been known that the presence of nutrients such as glucose, fat, and protein in the intestinal lumen stimulates secretion of many gut hormones, the effects of fructose are poorly understood. While some studies have shown that fructose does not stimulate insulin secretion in: 1) humans when perfused intraduodenally, 2) in rat pancreatic preparations, or 3) in isolated human and rat pancreatic islets (3,6,27), other studies showed that fructose causes insulin secretion in humans after oral ingestion (9,24). Nowadays, fructose is a major sweetener in Western diets (2,10). However, increased dietary intake of fructose has been suspected to be partly responsible for the growing rates of obesity and the metabolic syndrome (hypertension, hypertriglyceridemia, hyperlipidemia, insulin resistance, and type 2 diabetes mellitus) (17,20), possibly due to fructose-induced perturbation of cell signaling and inflammatory reactions in insulin-sensitive tissues (21). We conducted the present study to investigate the effect of fructose on appetite-and metabolism-regulating hormones from the gut, with a particular focus on the incretin hormones glucose-dependent insulinotropic polypeptide (GIP) and glucagon-like peptide 1 (GLP-1), since recent expression profiles of some gut endocrine cells suggested expression of the fructose transporter, GLUT5 (14,19). Because it turned out that fructose clearly stimulated GLP-1 secretion in humans, we investigated intracellular mechanisms behind fructose-stimulated secretion in the murine L cell model, the GLUTag cell line (4), after confirming that GLP-1 secretion was also stimulated in rats and mice.

Effects of Fructose on Gut Hormone Secretion in Humans
Subjects and study protocol. Nine healthy volunteers (4 men and 5 women; mean age 27.7 Ϯ 1.2 yr, range: 23.4 -36.8, mean body mass index 21.7 Ϯ 0.4 kg/m 2 , range: 18. 2-26.3) participated in the study, which was approved by the local ethical committee and conducted according to the principles of The Helsinki Declaration. All subjects had normal fasting blood glucose levels (5.29 Ϯ 0.39 mM, range: 4.5-6.1 mM), and none had parents or siblings diagnosed with any type of diabetes. No subjects received medication known to interfere with glucose homeostasis. Each subject was studied on two occasions within 3 wk after the first day of study. Subjects were instructed to refrain from vigorous exercise and alcohol for at least 24 h before each study. Study days began at 0830 preceded by a 10-h overnight fast. Venous blood samples were collected at time Ϫ10, 0, 15,30,45,60,90, 120 min in prechilled EDTA (10.8 mg) tubes (catalog no. 367864; BD Biosciences, Albertslund, Denmark) through a polyethylene cannula placed in a cubital vein. At time 0 min, the seated, single-blinded, subjects drank a sugar solution containing 75 g fructose or glucose dissolved in 300 ml water (RT) within 2 min. For palatability, solutions were refreshed with lemon juice. Upon collection, blood samples were instantly chilled on ice and centrifuged (2,400 g, 15 min, 5°C) within 30 min. Plasma were stored at Ϫ20°C until analysis.

Effects of Fructose and Glucose Gavage in Rats
Animals and study protocol. Studies were approved by the Danish Animal Experiments Inspectorate (2013-15-2934-00833). Male Wistar rats were obtained from Taconic (Ejby, Denmark) and allowed at least 1 wk of acclimatization before the day of experiment. Rats followed a 12:12-h light-dark cycle with ad libitum access to standard chow and drinking water. Every day for a week up to the experiment, rats were handled to minimize stress-related responses on the day of study. Handling included restraint and gavage feeding with drinking water. Experiments were carried out on two occasions on nonfasted rats (294.9 Ϯ 3.8 g) just before their nocturnal feeding period (1700). Weight did not differ between groups or study occasions (P Ͼ 0.05) (data not shown). Rats were divided into weight-matched groups (n ϭ 4/occasion) and given a bolus of either glucose or fructose (2 g/1,000 g body wt) diluted in milli-Q water to a final concentration of 50% (wt/vol) or a matched volume vehicle (milli-Q water). Rats from the same cage received different treatments. Blood (200 l) was collected into prechilled EDTA-coated capillary tubes (catalog no. 200 K3E, Microvette; Sarstedt, Nümbrecht, Germany) by sublingual vein puncture and instantly transferred onto ice. The zero sample was collected 10 min before bolus administration. At time 0 min rats were stimulated with either vehicle, glucose, or fructose, and blood was collected at time 15, 30, and 60 min. Blood glucose was measured instantly after collection, and samples were centrifuged (1,650 g, 4°C, 10 min) to obtain plasma, which was transferred to fresh Eppendorf tubes and immediately frozen on dry ice. Samples were stored at Ϫ20°C until analysis. At the end of the experiment rats were anesthetized by subcutaneous injection of hypnorm/midazolam (0.2 ml/ 100 g, concentration 5 mg/ml) and killed by diaphragm perforation.

Effects of Fructose and Glucose Gavage in Mice
Animals and study protocol. Studies were carried out with permission from the Danish Animal Experiments Inspectorate (2012-15-2934-00207). Female C57BL/6 mice were obtained from Charles River (Sulzfeld, Germany) and allowed 1 wk of acclimatization before the day of experiment. Mice (16 mice/group, weight: 20.30 Ϯ 0.31 g, 8 -10 wk) were fasted overnight but allowed free access to water. Mice were prestimulated with a gavage of 100 l drinking water at time Ϫ30 min and again at time 0 min with fructose or glucose (3 g/kg) diluted in drinking water to a final concentration of 20% (wt/vol). At time 0 and 6 min, blood (75 l) for hormone measurement was collected into prechilled EDTA-coated capillary tubes (catalog no. 164213; Vitrex, Herlev, Denmark) supplemented with dipeptidyl peptidase 4 inhibitor (2 l, 1 mM diprotinin A) (catalog no. I9759; Sigma Aldrich). After collection, blood samples were immediately transferred to Eppendorf tubes supplemented with 7.5 l diprotinin A (1 mM) and centrifuged (3,200 g, 20 min, 4°C) within 30 min. Plasma was transferred to fresh Eppendorf tubes, snap-frozen on dry ice, and stored at Ϫ20°C until analysis. Blood for glucose measurements (3 l) was obtained by tail puncture immediately after blood sampling and directly processed for blood glucose measurements as described below. After the first day of study (13 wk), the procedure was repeated. At both occasions, mice were randomly divided into glucose-and fructose-receiving groups. Weights did not differ between fructose and glucose groups at the respective days of study (weight: week 0: F ϭ 20.30 Ϯ 0.313 g vs. glucose ϭ 20.59 Ϯ 0.267 g; week 13: F ϭ 27.26 Ϯ 1.013 g vs. glucose ϭ 28.49 Ϯ 0.892 g, P Ն 0.37). At the end of the experiment, mice were killed by cervical dislocation.

GLUTag Cell Studies
Cell culture and secretion studies. GLUTag cells were cultured in T75 cell culture flasks in low-glucose (1.0 g glucose/l) DMEM 5564 medium (Sigma Aldrich, Buchs, Germany) supplemented with 10% (vol/vol) FCS, 1% (vol/vol) penicillin/streptomycin, and 1% (vol/vol) glutamine. Cells were incubated at 37°C, 5% CO 2 until 70 -80% confluent, then trypsinized and plated (ϫ10, 1 ml/well) on a 24-well plate precoated with matrigel (1:100) (catalog no. 354234; BD Biosciences, Bedford, MA) as described previously (15). The following day, cells were thoroughly washed with saline buffer (138 mM NaCl, 4.5 mM KCl, 4.2 mM NaHCO 3, 1.2 mM NaH2PO4, 2.5 mM CaCl2, 1.2 mM MgCl2, and 10 mM HEPES) supplemented with 0.1% (wt/vol) fatty acid-free BSA (A-603-10G; Sigma Aldrich) and incubated for 2 h (37°C, 5% CO2) with 250 l test reagents (fructose, 100 M gliclazide/tolbutamide, 340 M diazoxide) dissolved in saline buffer. All reagents were supplied by Sigma Aldrich. Supernatants were collected and centrifuged (1,500 g, 5 min, RT) to remove floating cells and debris and then either snap-frozen on dry ice and stored at Ϫ80°C or directly processed for GLP-1 active measurement. NAD(P)H imaging. Intracellular NAD(P)H concentration levels were measured using NAD(P)H's autofluorescence properties (22,23). GLUTag cells were plated on glass bottom dishes (catalog no. P35G-0-14-C; MatTek, Ashland, UK), and experiments were performed using an inverted fluorescence microscope (Olympus IX71) with an ϫ40 oil-immersion objective as described previously (13,15). In brief, cells were excited at 360/15 nm using a 75-W xenon arc lamp and monochromator, and emission was recorded at 510/80 nm with an Orca-ER CCD camera (Hamamatsu, UK), controlled by MetaFluor software. Measurements were obtained every 10 s and background corrected using parallel measures from a cell-free location. Cells were perifused with test reagent dissolved in saline buffer (0.1, 1, 3, 10, or 30 mM fructose, 30 mM mannitol, or 10 mM glucose) at a rate of ϳ1 ml/min. Glucose was included in all experiments as positive control because of previous studies demonstrating that glucose is metabolized by the GLUTag cell (19). Negative glucose responders (Ϸ1/3) were excluded from the data analysis. Between test agents, cells were washed with saline buffer as above until readings returned to baseline. NAD(P)H imaging data are expressed as background-subtracted average values taken over a period of 5 min before and after test substance application.

Data Presentation and Assessment of Statistical Significance
Data are expressed as means ϩ 1 SE. Graphs were constructed using Graphpad Prism 5 software, and statistical significance was calculated with the same program. Data were evaluated statistically by one-way ANOVA followed by Bonferroni post hoc test, Student's t-test, paired Student's t-test, or one-sample t-test, testing values against a hypothetical value of zero, as indicated. P Ͻ 0.05 was considered significant. For the human study, baseline values are presented as the mean of the Ϫ15 and 0 min levels. EC 50 value for fructose-stimulated GLP-1 secretion from GLUTag cells (see Fig. 5D) was obtained by a five-parameter logistic equation using GraphPad Prism software.

Fructose Elevates Blood Glucose Levels and Stimulates Insulin Secretion in Healthy Young Humans
Glucose. There was no difference in blood glucose concentrations at baseline between the glucose and fructose treatment days (P Ͼ 0.05, n ϭ 9) (Fig. 1A). Both treatments increased blood glucose (P Ͻ 0.001) with significantly higher area under the curve (AUC) values after oral glucose vs. fructose (P Ͻ 0.05, n ϭ 9) (Fig. 1B). Maximum blood glucose concentrations were also higher in the glucose group (glucose ϭ 7.5 Ϯ 0.4 pM vs. fructose ϭ 6.1 Ϯ 0.2 pM, P Ͻ 0.05, n ϭ 9). For both groups, blood glucose concentrations returned to basal levels by the end of the study period (2 h) (Fig. 1A).
CCK. CCK concentrations did not differ significantly between treatments at baseline (P Ͼ 0.05, n ϭ 9) (Fig. 2G). Glucose and fructose both caused a rise in CCK concentrations (P Ͻ 0.01, n ϭ 9) with peak values occurring 15 min after ingestion for both treatments (glucose ϭ 2.3 Ϯ 0.5 pM vs. fructose ϭ 2.8 Ϯ 0.6 pM, P Ͻ 0.05, n ϭ 9). Thereafter, CCK concentrations decreased for both treatments, albeit with a tendency toward higher values in the glucose treatment group at later time points (Fig. 2G). CCK AUC values did not differ between treatments (P Ͼ 0.05) (Fig. 2H).

DISCUSSION
The increased intake of dietary fructose in general and high-fructose corn syrup in particular has been suggested to play a role in the current obesity epidemic (8,20). Moreover, fructose has been suggested to have a lower satiating potency than glucose, but the available studies are conflicting and involve fructose doses greater than normally consumed, as reviewed (12). In this study, we examined the effects of fructose intake on the secretion of gut hormones, known to influence appetite and metabolism. The main finding was that fructose intake caused GLP-1 secretion in healthy humans, whereas it had no effect on GIP secretion. Fructose also stimulated the release of CCK and NTS 1-13 to a similar extent as glucose, and also enhanced insulin secretion, albeit to a lower degree than glucose. In contrast, neither fructose nor glucose altered glucagon or PYY 3-36 levels. With a view to perform mechanistic studies regarding the differential responses of the incretin hormones, we also investigated GIP and GLP-1 responses to fructose in mice and rats. In both animals, fructose stimulated GLP-1 secretion to the same extent as isocaloric glucose, but, similar to humans, GIP secretion was not stimulated. In mice and humans, fructose intake also increased blood glucose levels, and a similar tendency was seen in the rat. It is generally accepted that fructose can be converted to glucose in the liver by mechanisms involving rapid phosphorylation of fructose to fructose 1-phosphate by fructose kinase, which by liver-type-B aldolase activity, yields the intermediate metabolites dihydroxyacetone phosphate and glyceraldehyde, with the latter being a substrate for glucose production by further metabolic activity. However, it has also been shown that the small intestine has capacity to transform fructose into glucose by the same pathway (1,5). Although not investigated here, it may be speculated that the rise in blood glucose after fructose intake reflects a combination of intestinal and hepatic conversion. Previous studies have shown that fructose did not stimulate insulin secretion from perfused rat pancreatic preparations or from isolated human and rat pancreatic islets (3, 6, 27), but two other human studies found, in agreement with our study, that oral fructose stimulates insulin release and showed that fructose is a less effective insulin secretagogue than glucose (9,24). This would suggest that the stimulatory effect in humans might result from the effects of the increased glucose concentrations in combination with the secretion of the insulinotropic hormone, GLP-1. GIP clearly did not contribute.
Because fructose stimulated GLP-1 release also in the mouse, this provided a meaningful reason to study the underlying mechanism of GLP-1 release using the murine L cell model, the GLUTag cell line. A direct mechanism might be expected since expression of the fructose transporter GLUT5 was previously found in both L cells and GLUTag cells. We reported previously that glucose metabolism stimulates GLP-1 secretion from GLUTag cells by K ATP channel closure, causing cell membrane depolarization (13), consistent with high expression levels of K ATP channel subunits Kir6.2 and SUR1 in both GLUTag cells and primary murine L cells (19). We show here that fructose was also metabolized by GLUTag cells and that fructose-triggered GLP-1 release was similar in magnitude to that induced by the K ATP channel inhibitors gliclazide and tolbutamide. That fructose did not trigger further secretion when added in the presence of gliclazide suggests that it acts at least in part via K ATP channel closure. This is further supported Statistical significance was tested by Student's t-test (B) or one-way ANOVA analysis followed by Bonferroni post hoc test (C-F). *P Ͻ 0.05, **P Ͻ 0.01, ***P Ͻ 0.001, and ****P Ͻ 0.0001 relative to base level.
by the finding that the channel opener diazoxide lowered GLP-1 secretion below basal levels even in the presence of fructose, although it should be noted that the membrane hyperpolarization induced by diazoxide would also inhibit secretion stimulated by other stimuli dependent on membrane depolarization. We cannot, therefore, exclude the possibility that fructose, like glucose, also enhances GLP-1 release from GLUTag cells via a metabolic action independent of K ATP channel closure. Whether fructose also targets similar pathways in primary L cells remains to be established. Our study provides no information on the cellular and physiological basis for why fructose stimulates GLP-1 and CCK, but not GIP, secretion. We reported previously that GLP-1 secretion from primary intestinal cultures was readily triggered by glucose (19), whereas glucose-triggered GIP responses were small in the absence of additional agents that increased cAMP concentrations (14). Thus, although primary murine K cells do express GLUT5 and the K ATP channel subunits (14), one possible explanation for the discrepancy between the cells reported here, therefore, is that K cells are generally less excitable than L cells and that the weaker fructose stimulus is sufficient to activate L cells, but not K cells. It is also possible, however, that the more distal location of L cells renders them more responsive to fructose. The sugar may, for example, be fermented by the gut microbiome, producing local stimuli such as short-chain fatty acids (25), although this is unlikely to explain the acute effects of fructose seen 15 min after ingestion. The metabolic consequences of the enhancement of GLP-1 but not GIP secretion may be important, since it appears to be the first time that a nutrient has been shown to activate the L cells while not affecting the K cells in vivo. Fructose might thus be used as a tool to discriminate between the combined actions caused by secretion of both GIP and GLP-1 and the effects caused by GLP-1 alone. Furthermore, because fructose elicits GLP-1 secretion without simultaneous release of glucagonotropic GIP, the pathways underlying fructose-stimulated GLP-1 release might be useful targets for drug development aiming at stimulating GLP-1 secretion for the treatment of type 2 diabetes and obesity. For obesity treatment, this strategy may prove to be a more an effective strategy to reduce food intake than monohormone treatment, since the stimulation of GLP-1 secretion should lead to simultaneous release of the costored anorectic hormones CCK, glicentin, oxyntomodulin, and PYY, as reviewed (26).