Molecular Mechanisms of Glucose-Stimulated GLP-1 Secretion From Perfused Rat Small Intestine

Glucagon-like peptide 1 (GLP-1) is an incretin hormone with many effects that can be exploited for the treatment of type 2 diabetes and obesity (1). As an incretin, GLP-1 potentiates glucose-induced insulin secretion, and it has long been known that intake of macronutrients (triglycerides, proteins, and carbohydrates) stimulates GLP-1 secretion. Among these, the effects of glucose on the incretin system have been studied in the greatest detail, and glucose appears to be one of the most effective incretin stimuli in humans and rodents (2–4). On the basis of studies of the GLP-1–expressing cell lines, GLUTag, NCI-H716, and STC-1 (5–7), primary L cells and cultures of dispersed intestinal cells (8), several molecular mechanisms have been suggested to link glucose exposure to GLP-1 secretion. These include 1) sweet taste receptors, which are expressed in GLUTag cells and in primary murine L cells (8,9); 2) electrogenic uptake through the sodium-glucose transporter 1 (SGLT1), which couples uptake of one glucose molecule and two Na ions and causes cell depolarization (10,11); and 3) by electroneutral GLUT2-mediated glucose uptake with K_ATP-channel closure secondary to intracellular glucose metabolism causing cell depolarization by preventing leakage of K ions. Because intestinal epithelial cells in culture have lost their polarity and all relevant paracrine and neural influences, whether these findings translate to intact physiological systems is uncertain; however, the mechanisms of secretion appear to be activated in the gut before glucose has reached the systemic circulation because glucose stimulates GLP-1 secretion from isolated canine ileal loops (12). For instance, it is uncertain whether glucose stimulates GLP-1 secretion by apically or basolaterally activated mechanisms and it is equally uncertain to what extent the enteric nervous system, which is known to have a synergistic role on GLP-1 secretion, may affect the suggested mechanisms of secretion outlined above.

GLUT2 is predominantly located on the basolateral side of the enterocyte but may be recruited to the apical surface upon the presence of glucose in the lumen (13,14). Theoretically, glucose could therefore stimulate GLP-1 secretion by basolateral GLUT2-facilitated glucose uptake rather than or in addition to apical SGLT1- and/or GLUT2-mediated uptake. In contrast, SGLT1 has not been localized to the basolateral side of enterocytes or L cells, making this scenario unlikely for SGLT1–mediated glucose-stimulated GLP-1 release. The aim of the current study was, therefore, to investigate the mechanisms of glucose-stimulated GLP-1 secretion in a more physiological setting, which allows discrimination between apical and basolateral activated mechanisms. To achieve this, we studied the effects of glucose on the isolated perfused rat small intestine. The advantage of this preparation is that cell polarity is retained, nerve and blood vessel connections are intact, and the local cell environment (including contact with neighboring cells) is undisturbed, meaning that data from this model are more likely to be transferrable to intact physiological systems. We examined the effects of vascular (intra-arterial) glucose and luminal glucose stimulation and the importance of intracellular glucose metabolism, K_ATP-channel closure, voltage-sensitive Na (Na_V) and voltage-gated (V-gated) Ca channels, sweet taste receptors, facilitative (GLUT2-mediated) glucose uptake, and electrogenic (SGLT1)-mediated glucose uptake for GLP-1 secretion.

Studies were conducted with permission from the Danish Animal Experiments Inspectorate (2013-15-2934-00833) and the local ethics committee (EMED, P-12-099 and P-13-240) in accordance with the guidelines of Danish legislation governing animal experimentation (1987) and the National Institutes of Health (publication number 85-23). Male Wistar rats (∼250 g) were obtained from Taconic (Ejby, Denmark) and housed two per cage, with ad libitum access to standard chow and water, and kept on a 12:12-h light-dark cycle.

Nonfasted rats were anesthetized by subcutaneous injection with Hypnorm/midazolam (0.079 mg fentanyl citrate + 2.5 mg fluanisone + 1.25 mg midazolam) and placed on a 37°C heated table. The abdominal cavity was opened, and the entire large intestine and approximately two-thirds of the small intestine were carefully removed, leaving the most proximal part of the small intestine (∼32 cm) in situ. The intestinal length retained varied little (coefficient of variation = 9.70%) between experiments. A plastic tube was placed in the lumen, the intestinal contents were carefully removed by flushing with isotonic (0.9%; 0.31 Osmol/L) NaCl (room temperature), and the intestine was luminally perfused with a steady flow of NaCl (0.250 mL/min). A catheter was inserted into the cranial/superior mesenteric artery, and the intestine was vascularly perfused (7.5 mL/min) with perfusion buffer gassed with 95% O₂ and 5% CO₂ (Krebs-Ringer bicarbonate buffer femented with 0.1% BSA, 5% dextran T-70 [Pharmacosmos, Holbaek, Denmark], 3.5 mmol/L glucose, 10 μmol/L 3-isobutyl-1-methylxanthine, and 5 mmol/L pyruvate, fumarate, and glutamate; pH 7.4). The perfusion medium was collected through a metal catheter placed in the hepatic portal vein, and the rat was killed by perforation of the diaphragm as soon as proper flow was apparent. The perfusion buffer was warmed (37°C) by passing it through a Uniper UP-100 perfusion system (Hugo Sachs; Harvard Apparatus, March-Hugstetten, Germany), and perfusion pressure was measured continuously.

The luminal stimuli were glucose (5% or 20% [w/v]); 0.36 and 1.42 Osmol/L, respectively), mannitol (20% [w/v]; 1.41 Osmol/L), or the nonmetabolizable sugar, methyl-α-d-glucopyranoside (α-MGP; 20% [w/v]; 1.34 Osmol/L), perfused in presence or absence of the SGLT1 inhibitor phloridzin (10 mmol/L) or the GLUT2 inhibitor phloretin (1 mmol/L) applied to the lumen. Moreover, the lumen was stimulated with the artificial sweeteners sucralose and acesulfame K, which are 800 and 260 times sweeter than glucose, respectively, meaning that the applied concentrations of 0.25% (w/v) sucralose and 0.78% (w/v) acesulfame K were each equivalent to a sweetness 10 times greater than 20% (w/v) glucose. For all luminal stimulation experiments except one (luminal glucose with and without NaCl), stimulants were diluted in isotonic saline (0.9% w/v) applied at an initial rate of 2.5 mL/min for the first 2 min to replace the saline solution in the lumen and then at 0.250 mL/min throughout the rest of the stimulation period. Immediately after each stimulation, the lumen was flushed with a similar bolus of 0.9% NaCl (2.5 mL/min) for 2 min, followed by infusion at a flow rate of 0.250 mL/min between stimulations.

In the luminal glucose with and without the NaCl experiment, the gut was perfused as outlined above, but immediately after the first glucose stimulation and until the end of the second glucose stimulation, isotonic saline was replaced with NaCl-deprived Milli-Q water balanced with iso-osmolar KCl (0.11% [w/v]). In other experiments, the gut was vascularly (arterially) perfused with KCl (50 mmol/L), the K_ATP-closing sulfonylureas tolbutamide (500 μmol/L) or gliclazide (500 μmol/L), the K_ATP-channel opener diazoxide (500 μmol/L), the Na_V channel blocker lidocaine (100 μmol/L), the Na_V activator veratridine (10 μmol/L), or the V-gated Ca channel blocker nifedipine (10 μmol/L). Additional experiments assessed GLP-1 responses when glucose was administered intra-arterially at physiological (5 mmol/L and 10 mmol/L) and pathophysiological (15 mmol/L and 25 mmol/L) concentrations. Bombesin (10 nmol/L; BBS), a known GLP-1 secretagogue, was included in all experiments as a positive control. All drugs were obtained from Sigma-Aldrich (Brøndby, Denmark). Drugs applied to the luminal side of the gut were diluted in 0.9% NaCl, and stimulants for the vascular side were diluted in perfusion buffer. To aid solubility, diazoxide, gliclazide, lidocaine, nifedipine, phloretin, and tolbutamide were first dissolved in DMSO and then diluted further in perfusion medium, with DMSO concentrations never exceeding a final concentration of 0.1%, which did not cause GLP-1 secretion in control experiments (n = 6; data not shown).

After an equilibration period of 30 min, effluent samples were collected each minute, immediately chilled on ice, and transferred to −20°C within 30 min. In randomly chosen experiments, Po₂, Pco₂, lactate concentrations, and pH were measured with a Radiometer ABL 700 blood gas analyzer (Radiometer Copenhagen, Copenhagen, Denmark) in perfusate samples collected from the arterial and venous sides of the preparation at the start and end of the experiment, respectively. The intestine was excised and stored in formaldehyde (4°C) for later histological examination (embedded in paraffin and stained with hematoxylin and eosin, as described previously [15]). The perfused preparations showed regular peristaltic activity (recorded as changes in perfusion pressure), which increased with respect to frequency and amplitude in response to stimulation with BBS, suggesting preserved enteric nervous function.

Venous effluent glucose concentration was measured by the glucose-oxidase method using the hand-held Accu-Chek Compact Plus meter (Roche, Mannheim, Germany). Effluent GLP-1 concentrations were measured using a total GLP-1 radioimmunoassay (measuring intact peptide + the primary metabolite, GLP-1 [9–36_amide]) using antiserum code no. 89390, recognizing the amidated C-terminus of the molecule, as described previously (16). Because the gut was perfused at a constant rate, concentrations parallel output (= secretion). The choice of targeting the amidated (x-36_amide) rather than the glycine extended (x-37) isoform was based on a recent study showing that GLP-1 is predominantly amidated in rats (17). The experimental detection limit was 1 pmol/L for the 89390 assay, with an intra-assay coefficient of variation of 6%.

Data are expressed as means ± 1 SEM. Responses were assessed by comparing averaged concentrations during the stimulation period with mean basal levels from a period of similar duration: immediately before the stimulation (5–8 consecutive 1-min observations) and at the end of the following equilibrium period (5–8 observations). The response period was defined as the period from the start of stimulant application until GLP-1 concentrations were back to baseline levels. Statistical significance was assessed by paired t test; P < 0.05 was considered significant.

Luminal stimulation of the intestine with 5% (w/v) glucose caused a small, but significant, elevation in glucose and GLP-1 (total) concentrations in the venous effluent compared with basal levels (glucose: from 3.53 ± 0.1 to 4.42 ± 0.3 mmol/L, GLP-1: from 14.9 ± 0.5 to 20.2 ± 0.6 pmol/L, P < 0.0001, n = 6; Fig. 1A–C), with greater responses being elicited by 20% (w/v) glucose (glucose: from 4.46 ± 0.1 to 7.87 ± 0.4 mmol/L, GLP-1: from 19.6 ± 0.9 to 39.5 ± 2.6 pmol/L, P < 0.0001, n = 6; Fig. 1E–G). A positive linear relationship was found between venous glucose and GLP-1 concentrations for both treatments (R² = 0.68 for 5% [w/v] glucose, R² = 0.87 for 20% [w/v] glucose, P < 0.001; n = 6; Fig. 1D and H). Control experiments demonstrated similarly enhanced GLP-1 responses to two consecutive stimulation periods with luminal glucose (20% [w/v]) compared with basal levels (stimulation 1: from 22.7 ± 1.2 to 42.3 ± 2.3 pmol/L, P < 0.0001; stimulation 2: from 27.1 ± 0.8 to 43.2 ± 2.5 pmol/L, P < 0.0001; P > 0.05 for stimulation 1 vs. stimulation 2, n = 6; Fig. 2A and B). The responses were not related to osmolality, because luminally administered mannitol (20% [w/v]) did not stimulate secretion (from 14.6 ± 0.5 to 14.8 ± 0.4, P = 0.73, n = 6; Fig. 2E and F). All experiments included administration of BBS as a positive control, in each case resulting in a robust GLP-1 response (Figs. 1–7).

In the initial experiments, we assessed whether cell depolarization causes GLP-1 secretion from the perfused rat intestine. Vascular administration of KCl increased GLP-1 secretion (from 8.60 ± 0.4 to 47.8 ± 13 pmol/L, P < 0.05, n = 6), but in the presence of the V-gated (V-type) Ca channel inhibitor nifedipine, the response was markedly reduced (from 7.83 ± 0.5 to 15.3 ± 1.4 pmol/L, P < 0.01; Fig. 3C and D) and significantly smaller than without nifedipine (P < 0.05). Control studies showed that the secretory responses to two consecutive KCl stimulations resulted in responses that both were significantly different from their respective baselines (stimulation 1: from 12.7 ± 1.7 to 42.2 ± 12.3 pmol/L, stimulation 2: from 6.29 ± 3.7 to 22.2 ± 4.1 pmol/L, both P < 0.05) but were not significantly different from each other (P = 0.08; Fig. 3A and B), although the second response tended to be smaller. Moreover, BBS-induced secretion was unaffected by nifedipine (BBS: from 13.9 ± 1.0 to 65.6 ± 10 pmol/L, BBS + nifedipine: from 12.7 ± 0.3 to 49.9 ± 8.9 pmol/L, P < 0.05, n = 6; Fig. 3E and F).

The effect of luminal glucose (20% [w/v]) was also lost in the presence of nifedipine (glucose: from 9.19 ± 0.6 to 25.1 ± 0.6 pmol/L, P < 0.01; glucose + nifedipine: from 7.90 ± 0.6 to 10.6 ± 1.9 pmol/L, P > 0.05, n = 6; Fig. 2C and D). The addition of the hyperpolarizing agent diazoxide also abolished the GLP-1 response to luminal glucose (20% [w/v] glucose: from 12.0 ± 0.6 to 31.8 ± 1.9 pmol/L, P < 0.0001; glucose + diazoxide: from 16.9 ± 1.5 to 14.6 ± 0.5, P > 0.05, n = 6; Fig. 3G and H).

We next investigated the importance of Na_V channel activation for GLP-1 secretion by stimulating the gut with luminal glucose (20% [w/v]) in the presence or absence of lidocaine, an analgesic that blocks Na_V channels (18). Again, luminal glucose (20% [w/v]) stimulated GLP-1 secretion (from 14.8 ± 1.2 to 26.7 ± 2.7 pmol/L, P < 0.0001), but the response was fully maintained in the presence of lidocaine (from 15.0 ± 1.5 to 27.1 ± 2.0 pmol/L, P < 0.0001, n = 6; Fig. 4A and B). Control experiments confirmed that this was not due to inadequate Na_V-channel blockade, because channel activation by veratridine (19) caused GLP-1 secretion (veratridine: from 16.8 ± 0.5 to 33.5 ± 7.3, P < 0.01, n = 6), but not in the presence of lidocaine (veratridine + lidocaine: from 18.3 ± 1.1 to 22.9 ± 0.6, P > 0.05, n = 6; Fig. 4C and D). In other experiments, veratridine-stimulated GLP-1 secretion (from 10.4 ± 0.4 to 22.3 ± 4.2 pmol/L, P < 0.05) was abolished by costimulation with nifedipine (from 11.6 ± 0.6 to 13.9 ± 1.1 pmol/L, P > 0.05, n = 6; Fig. 4E and F), indicating that the response to Na⁺ channel activation also involved the V-gated Ca channels.

We next investigated the role of Na-coupled glucose uptake through SGLT1 for GLP-1 secretion. Luminal glucose 20% (w/v) stimulated a GLP-1 response (glucose: from 12.5 ± 0.3 to 21.1 ± 2.0 pmol/L, P < 0.01) but not in the absence of luminal NaCl (glucose − NaCl: 12.5 ± 0.3 to 12.7 ± 0.4, P > 0.05, n = 6; Fig. 5A and B). In other experiments, 20% (w/v) luminal glucose resulted in a robust GLP-1 response (glucose: from 17.3 ± 0.4 to 37.8 ± 2.1 pmol/L, P < 0.0001), but this was completely lost in the presence of the SGLT1 inhibitor, phloridzin (glucose + phloridzin: from 20.1 ± 1.0 to 20.9 ± 0.9 pmol/L, P > 0.05, n = 6; Fig. 5C and D). Perfusion of the gut with the nonmetabolizable SGLT1 substrate, α-MGP, also significantly increased GLP-1 secretion (from 15.7 ± 0.5 to 29.4 ± 0.7 pmol/L, P < 0.0001, n = 6), which could be completely blocked by the administration of phloridzin (from 17.5 ± 0.7 to 17.7 ± 0.4 pmol/L, P > 0.05, n = 6; Fig. 5E and F). When applied in the same experiment, luminal glucose and luminal α-MGP (both 20% [w/v]) both stimulated GLP-1 secretion (glucose: from 14.6 ± 1.1 to 30.6 ± 2.9, α-MGP: 16.1 ± 0.5 to 22.0 ± 0.8, both P < 0.001, n = 6), but the response to glucose was significantly higher than the response to α-MGP (P < 0.01; Fig. 6G and H).

We next investigated the effect of GLUT2 inhibition on glucose-stimulated GLP-1 secretion. Phloretin, a widely used GLUT2-specific inhibitor (20,21), significantly reduced, but did not eliminate, glucose-stimulated GLP-1 secretion (glucose: from 13.9 ± 0.6 to 30.9 ± 2.8 pmol/L, glucose + phloretin: from 12.4 ± 1.3 to 17.2 ± 3.4 pmol/L, P < 0.001 for both stimulations compared with respective baselines, P < 0.0001 between responses, n = 6; Fig. 6A and B), whereas the secretory response to α-MGP was not sensitive to phloretin (α-MGP: from 14.1 ± 0.3 to 29.4 ± 2.4 pmol/L, α-MGP + phloretin: from 14.7 ± 0.4 to 25.1 ± 1.0 pmol/L, P < 0.0001 between responses and respective baselines, P > 0.05 between responses; Fig. 5G and H).

To investigate the consequences of K_ATP-channel closure, we stimulated the gut with the two sulfonylurea drugs: gliclazide and tolbutamide. Both drugs significantly elevated GLP-1 levels compared with basal levels (tolbutamide: from 14.7 ± 0.2 to 28.6 ± 5.2 pmol/L, gliclazide: from 15.2 ± 0.3 to 24.6 ± 2.4 pmol/L, P < 0.05, n = 8; Fig. 6C and D). Furthermore, glucose-stimulated GLP-1 secretion was lost by blockage of ATP synthesis with 2,4-dinitrophenol (glucose: from 7.49 ± 0.7 to 23.5 ± 3.0 pmol/L, P < 0.0001, glucose + 2,4-dinitrophenol: from 9.24 ± 1.3 to 6.61 ± 0.4 pmol/L, P > 0.05, n = 6; Fig. 6E and F).

Intravascularly administered glucose within the physiological range (5 and 10 mmol/L) had no effect on GLP-1 secretion (P > 0.05), but a minor increase was observed at the 15 mmol/L concentration (from 21.7 ± 0.5 to 25.2 ± 0.6 pmol/L, P < 0.001, n = 6; Fig. 7A and B). However, 25 mmol/L vascular glucose concentrations were ineffective in a separate experiment (P > 0.05, n = 6; data not shown). Some reports have indicated that sweet taste receptor activation causes GLP-1 secretion. In our hands, however, neither sucralose (0.25% w/v) nor acesulfame K (0.78% w/v) increased GLP-1 secretion (sucralose: from 16.9 ± 0.4 to 16.5 ± 0.4 pmol/L, acesulfame K: from 17.9 ± 0.5 to 16.7 ± 0.3 pmol/L, P > 0.05, n = 6) when administered luminally (Fig. 7C and D).

It has long been known that oral glucose intake stimulates GLP-1 secretion in humans, but detailed analysis of the mechanisms involved, including discrimination between apically and basolaterally activated mechanisms is, of course, not possible in vivo. Significant insight into the molecular mechanisms of GLP-1 secretion has been gained from studies of GLP-1–secreting cell lines and, more recently, through studies of primary L cells and primary murine gut epithelial cell cultures. According to the classical model of intestinal glucose absorption, glucose crosses the apical membrane of the enterocyte via SGLT1 and exits across the basolateral membrane through GLUT2. More recent evidence indicates, however, that the presence of glucose in the lumen recruits GLUT2 to the apical side of the enterocyte within minutes, and during assimilation of a meal, this component of absorption may be several times greater than the active SGLT1–mediated absorption, as reviewed (13,14). Because GLUTag and primary murine L cells express both SGLT1 and GLUT2 transporters (8), glucose may, therefore, stimulate GLP-1 secretion by a basolateral uptake secondary to intestinal glucose absorption (by neighboring enterocytes) rather than by apical mechanisms of the L cells themselves, activated by luminal glucose. We, therefore, investigated if vascular glucose stimulated GLP-1 secretion and the role of apical SGLT1 and GLUT2 activity for luminal glucose-stimulated GLP-1 secretion. Vascular glucose was without effect in physiological concentrations and did not consistently stimulate secretion at supraphysiological doses, whereas luminal glucose robustly and dose-dependently stimulated secretion.

Glucose absorption has been shown to be more important for GLP-1 secretion than the mere presence of glucose in the lumen (22). We therefore investigated the effect of luminal NaCl depletion (substituting Na with K). In absence of NaCl, the stimulatory effect of luminal glucose was completely lost, suggesting that the mechanisms of secretion depend on Na-coupled uptake. A similar observation was made by another group studying perfused rat ileum (23). Indeed, glucose uptake by SGLT1 was both necessary and sufficient to cause glucose-stimulated GLP-1 secretion because α-MGP and glucose both caused GLP-1 secretion, whereas blockade of the transporter (with phloridzin) completely abolished the response to both secretagogues. Consistent with this, studies from other groups on GLUTag cells and primary mouse intestine cultures have shown that α-MGP and glucose both stimulate phloridzin-sensitive GLP-1 secretion and that α-MGP also stimulates GLP-1 secretion after oral administration in mice, whereas glucose did not stimulate GLP-1 secretion in SGLT1^−/− mice (8,11,24–27).

However, in our model (Fig. 8), glucose-stimulated GLP-1 secretion was also sensitive to apical GLUT2 blockage, but in contrast to SGLT1 inhibition, phloretin did not eliminate the response to luminal glucose, although the applied dose was sufficient to inhibit GLUT2 activity in other models (21,27). In support of this, glucose-stimulated GLP-1 secretion to an oral gavage has been found reduced, but not abolished, in GLUT2^−/− knockout mice compared with wild-type littermates, and the same pattern has been observed in studies on GLUTag cells (27,28). Our results therefore are in good agreement with work performed on isolated/single cells, but only our approach allows analysis of the differential effects caused by apical or basolateral GLUT2 inhibition. Although SGLT1-mediated transport involves cotransport of two Na ions for each glucose molecule, glucose uptake via GLUT2 is electroneutral. In this case, secretion might depend on glucose metabolism and formation of ATP, with actions on the K_ATP channels, similar to glucose-activation of pancreatic β-cells. Indeed, expression analysis in primary murine L cells (and GLP-1–secreting cell lines) indicated that L cells do express the K_ATP-channel subunits SUR1 and Kir6.2 and that tolbutamide also stimulates GLP-1 secretion from primary dispersed murine cultures and GLUTag cells (8,29). In contrast, in humans, sulfonylureas do not seem to influence GLP-1 secretion (27), although SUR1 was found colocalized with GLP-1 in human L cells (30). Like the β-cells, human L cells also express the rate-limiting glucose sensor, glucokinase, but glucose-stimulated GLP-1 secretion is not reduced in individuals with glucokinase mutations (MODY 2) (31–33), although these individuals require abnormally high glucose concentrations to generate adequate metabolism-dependent insulin secretion.

We therefore investigated the effects of high doses of sulfonylureas on the perfused gut. Tolbutamide and gliclazide both stimulated GLP-1 secretion, supporting that K_ATP-channel closure does stimulate GLP-1 secretion from the murine L cell in a physiological setting. Although not investigated in present study, the discrepancy between our findings and the lack of GLP-1 response to sulfonylureas in humans could be related to the level of K_ATP-channel expression in L cells. Thus, the K_ATP-channel subunits Kir6.2 and SUR1 are expressed to a greater degree in isolated murine β-cells compared with isolated murine L cells and GLUTag and STC-1 cells (8). It is, therefore, possible that the therapeutic doses of sulfonylureas used to control blood sugar levels in type 2 diabetic patients may not be high enough to have demonstrable effects on the L cell in vivo. Furthermore, the lack of effect of glucokinase mutations may simply reflect a predominant role of SGLT1 in GLP-1 secretion after oral glucose intake. However, at high concentrations, K_ATP-channel closure, subsequent to L-cell glucose uptake, may function to further enhance SGLT1-driven secretion. Indeed, we found that when the rat gut was luminally stimulated with similarly high concentrations (20% [w/v]) of α-MGP and glucose, respectively, in the same experiments, glucose stimulated GLP-1 secretion to a larger extent than α-MGP.

Studies of GLUTag and primary L cells have shown that L cells are electrically excitable, firing action potentials in response to glucose (8,11). Propagation of the action potentials across the membranes of the L cells would then result in depolarization and influx of Ca through V-gated Ca channels. Action potentials cannot easily be recorded from L cells in our preparation, but we used lidocaine to examine the consequences of blocking the Na_V channels responsible for the action potentials. This, however, did not affect luminal glucose-induced GLP-1 secretion, although the applied dose has been shown to powerfully inhibit channel activity in other studies (18,34). This is consistent with reported findings on GLUTag cells (35). Furthermore, we used veratridine to investigate the secretory responses to Na_V-channel activation. Veratridine powerfully stimulated GLP-1 secretion, and this response could be blocked by the same doses of lidocaine, suggesting that they were indeed sufficient to inhibit channel activity. A probable consequence of Na entry after Na_V-channel activation and entry via SGLT1 would be depolarization, activation of V-gated Ca channels, and secretion. This concept was supported by the observation that nifedipine abolished veratridine-induced GLP-1 secretion.

To directly investigate the influence of depolarization for GLP-1 secretion in our preparation, we added 50 mmol/L KCl to the perfusion medium. This caused clear GLP-1 secretion, which could be abolished by nifedipine, indicating that cell depolarization causes GLP-1 secretion by V-gated Ca-channel activation and uptake of Ca from the extracellular space. In further support of this concept, nifedipine and diazoxide (a K_ATP-channel opener, causing efflux of K and hyperpolarization of the cell) both abolished glucose-stimulated GLP-1 secretion. In contrast, nifedipine did not affect the secretory responses to the phospholipase C activator, BBS, which stimulates GLP-1 secretion by mobilization of Ca from intracellular stores rather than by depolarization. Thus, in the isolated perfused rat intestine, glucose-stimulated GLP-1 release does not depend on firing of action potentials, generated by activation of Na_V channels, but does require cell membrane depolarization and activation of L-type Ca channels with subsequent entry of extracellular Ca²⁺. Diazoxide also has inhibitory effects on insulin secretion from the pancreatic β-cell, and because of this effect was previously used to treat hypoglycemia. From our data it may, however, be speculated diazoxide may also have inhibited GLP-1 secretion in the human subjects receiving this treatment, but to our knowledge, no data are available on this.

GLUTag cells and primary murine L cells have been shown to express the sweet taste receptor subunits T1R2 and T1R3 as well as the signaling molecule gustducin. Combined with reports demonstrating that artificial sweeteners cause GLP-1 secretion in GLUTag cells and mice (9,36), this has led to the suggestion that the L cell may “taste” sugars and secrete GLP-1 in response (37). In our hands, however, the artificial sweeteners sucralose and acesulfame K (at doses corresponding to 10 times the sweetness of a 20% [w/v] glucose solution) did not enhance GLP-1 secretion, and our study, therefore, does not support the view that glucose (or other sugars) should stimulate GLP-1 secretion by activation of sweet taste receptors, in agreement with evidence from other studies by several independent groups on primary murine intestinal cultures in rodents or humans (8,24,38,39). It might be argued that the sweet taste receptors might generate afferent vagal signals that eventually lead to a reflex-generated GLP-1 response involving the central nervous system. In our preparation, this would not be possible. However, there is little evidence that GLP-1 secretion is vagally regulated (40), and, as mentioned, the artificial sweeteners are ineffective in humans.

Taken together, our data show that glucose most likely stimulates GLP-1 secretion from the perfused rat small intestine by SGLT1-mediated uptake from the lumen, causing GLP-1 secretion by activation of V-gated Ca channels through coupled Na uptake and depolarization. However, glucose may also stimulate GLP-1 secretion via electroneutral GLUT2-mediated uptake with subsequent metabolism, ATP formation, and K_ATP-channel closure. This hypothesis is supported by the elimination of the GLP-1 response after administration of the oxidative phosphorylation blocker 2,4-dinitrophenol. However, because GLUT2 inhibition, in contrast to SGLT1 inhibition, did not completely abolish glucose-stimulated GLP-1 secretion, K_ATP-channel closure may function to enhance SGLT1-driven GLP-1 secretion when glucose concentrations are high. Our data therefore indicate that SGLT1-mediated uptake is the major activator of glucose-stimulated GLP-1 secretion and that closure of K_ATP channels is required to stimulate a full-blown response to glucose.

Acknowledgments. The authors thank Carolyn F. Deacon (Novo Nordisk Foundation Center for Basic Metabolic Research, Department of Biomedical Sciences, the Panum Institute, University of Copenhagen) for editing the manuscript, Nicolai Jacob Wever Albrechtsen (also at Novo Nordisk Foundation Center for Basic Metabolic Research) for many fruitful discussions, and Musa Büyükuslu (Parenkymkirurgisk afsnit, Sydvestjysk Sygehus, Denmark) for making Fig. 8.

Funding. The study was supported by grants from the Novo Nordisk Center for Basic Metabolic Research (Novo Nordisk Foundation, Denmark) and the European Union’s Seventh Framework Programme for Research, Technological Development, and Demonstration Activities (Grant No. 266408).

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

Author Contributions. R.E.K. and J.J.H. contributed to the conception and design of the study and interpreted results. R.E.K. performed experiments, analyzed data, prepared figures, and drafted the manuscript. R.E.K., C.R.F., B.S., and J.J.H. approved final version of the manuscript. C.R.F. and B.S. developed the study model. J.J.H. supervised experiments and edited and revised the manuscript. J.J.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data en the accuracy of the data analysis.

Prior Presentation. The main findings of this study were presented at the 74th Scientific Sessions of the American Diabetes Association, San Francisco, CA, 13–17 June 2014.