DOI: 10.2337/db06-S010 © 2006 by the American Diabetes Association
Signaling Mechanisms Underlying the Release of Glucagon-Like Peptide 1From the Department of Clinical Biochemistry, Cambridge Institute for Medical Research, Addenbrookes Hospital, Cambridge, U.K Address correspondence and reprint requests to Dr. Fiona Gribble, Cambridge Institute for Medical Research and Department of Clinical Biochemistry, Wellcome Trust/MRC Building, Addenbrookes Hospital, Hills Rd., Cambridge CB2 2XY, U.K. E-mail: fmg23{at}cam.ac.uk
Abbreviations:
CGRP, calcitonin gene–related protein; DPPIV, dipeptidyl peptidase IV; GABA,
Glucagon-like peptide-1 (GLP-1) is released from intestinal L-cells in response to a range of nutrients, hormones, and neurotransmitters. Its potency as an insulin secretagogue has led to pharmaceutical interest in developing strategies to enhance GLP-1 receptor activation in type 2 diabetes. A complementary approach, to stimulate endogenous release of GLP-1, would be facilitated by a better understanding of L-cell physiology. Using GLP-1–secreting cell lines such as GLUTag and STC-1, mechanisms underlying GLP-1 release have been identified at a single-cell level. A number of stimuli, including glucose and certain amino acids, result in membrane depolarization and Ca2+ entry through voltage-gated Ca2+ channels. Glucose triggers membrane depolarization both by closing ATP-sensitive potassium channels and because of its uptake by Na+-coupled glucose transporters. Whereas glutamine also triggers depolarization by Na+-coupled uptake, glycine opens Cl– channels on the surface membrane. A number of agents, including fatty acids and hormones, enhance GLP-1 secretion by acting at stages downstream of depolarization. Some of these target G protein–coupled receptors, triggering elevation of cAMP or release of Ca2+ from intracellular stores. Understanding these different pathways and how they could be targeted to maximize GLP-1 secretion may be a step toward developing therapeutic GLP-1 secretagogues. A hundred years ago, in 1906, Moore et al. (1) first reported the successful use of acid extracts of porcine duodenal mucosa to treat three diabetic patients. Their work formed the basis for what is now known as the incretin effect: a concept describing the ability of gastrointestinal hormones, released in response to food intake, to stimulate insulin release from the endocrine pancreas (2). Two hormones, glucagon-like peptide-1 (GLP-1) and glucose-dependent insulinotropic peptide (GIP), are thought to underlie the incretin effect, which in healthy individuals, accounts for up to 50% of the normal insulin release after a meal (3,4). In type 2 diabetes, the incretin response is impaired, as a result of reduced postprandial GLP-1 concentrations (5) and a decreased responsiveness of the pancreatic ß-cells to GIP (6). Pharmacological doses of GLP-1, however, have been found to restore insulin release to near-normal levels (7), leading to the development of GLP-1–based therapies for the treatment of type 2 diabetes.
GLP-1 is secreted from specialized L-cells in the epithelial layer of the intestine. It is produced by posttranslational cleavage of proglucagon, which is processed by the prohormone convertase PC2 in pancreatic L-cells are scattered throughout the epithelial layer of the gut from the duodenum to the rectum, with the highest numbers occurring in the ileum, colon, and rectum (11,12). They are characterized by an open-cell morphology, with apical microvilli facing into the gut lumen and secretory vesicles located adjacent to the basolateral membrane. As well as GLP-1 and GLP-2, L-cells also secrete the anorexigenic hormone, peptide YY, and glutamate (11,13,14). The cells are just one member of a much larger family of enteroendocrine cells that secrete a range of hormones, including ghrelin, GIP, cholecystokinin, somatostatin, and secretin, which are involved in the local coordination of gut physiology, as well as in playing wider roles in the control of insulin release and appetite. GLP-1 is secreted in response to the ingestion of carbohydrates, fat, and protein (15,16). Levels in the circulation increase within about 15 min of food ingestion and remain elevated for up to 3 h, depending on the composition of the meal. The timing of the early secretory phase has led to the speculation that it occurs too rapidly to be attributable to the arrival of nutrients at proximal L-cells and might indicate the involvement of a local neural or hormonal pathway. Studies in humans, however, support the idea of direct glucose sensing, since L-cells have now been identified in sites more proximal than hitherto believed (17), and glucose flux into the small intestine after ingestion or infusion has been shown very quickly to saturate the uptake capacity of the duodenum, allowing sugars to overflow into the jejunum where the L-cell density is greater (18). Nevertheless, the relative roles of direct versus indirect signals to the L-cells, and the involvement of different L-cell pools in shaping the pattern of GLP-1 release, remain largely unestablished.
Several model systems have been developed for the study of GLP-1 secretion in vitro, including partially purified canine or rat L-cells prepared by elutriation (19,20), cultured fetal rat intestinal cells (21), and GLP-1–secreting cell lines such as GLUTag (22), STC-1 (23), and NCI-H716 (24). GLUTag cells were derived from a colonic tumor taken from a transgenic mouse expressing SV40 large T antigen under the control of the proglucagon promoter and secrete GLP-1 in response to a range of physiological stimuli (25). The STC-1 line was also derived from mouse, but is a pluripotent model of enteroendocrine cell function that has been used to investigate the secretion of a wide range of enteric hormones including secretin, cholecystokinin, GIP, and GLP-1. NCI-H716 is a human cell line derived from a caecal adenocarcinoma, which has been reported to differentiate into endocrine cells containing GLP-1 under certain tissue culture conditions. Although this would potentially be a valuable tool, as the only human in vitro model of GLP-1 release, further studies have questioned whether its behavior mirrors what is known about the physiology of L-cells (26).
Preparations of adult L-cells enriched by elutriation or derived by fetal rat intestinal culture might theoretically be superior to cell lines as a model of GLP-1 release because they are not immortalized. However, because they contain, at best, only
GLUTag and STC-1 cells are electrically excitable cells that depolarize and fire action potentials in response to a range of nutrient secretagogues, including monosaccharides and amino acids (27–32). In common with other endocrine cells, such as pancreatic ß-cells, membrane depolarization is coupled with the opening of voltage-gated calcium channels and an elevation of intracellular Ca2+, which is necessary for the stimulation of GLP-1 release (30). Although GLUTag cells, like ß-cells, can act as glucose sensors, the underlying mechanisms are only partially overlapping. To understand how nutrients trigger electrical activity in GLUTag cells, it is helpful to consider a simple model showing the basis of the membrane potential (Fig. 1). In the absence of glucose, GLUTag cells are relatively hyperpolarized at approximately –50 mV, resulting from the small excess of outward (Iout) over inward (Iin) currents (Fig. 1A). Depolarization from this resting state occurs when the balance is tipped in favor of the inward currents, either because outward currents have decreased in magnitude (Fig. 1B) or because the inward currents have increased in magnitude (Fig. 1C). As in pancreatic ß-cells, the outward current is carried largely by K+ ions flowing, at least in part, through ATP-sensitive K+ (KATP) channels. The identity of the inward current has not yet been established but appears to be sodium dependent, since replacement of Na+ with impermeant N-methyl-D-glucamine in perforated patch experiments on GLUTag cells (methods as in Gribble et al. [28]) resulted in membrane hyperpolarization from –50.8 ± 2.8 to –79.1 ± 4 mV (n = 7; including a 5 mV correction for the change in liquid junction potential), close to the predicted K+ reversal potential of –84.8 mV.
A number of studies have shown that luminal, but not systemic, glucose is a potent trigger of GLP-1 release in humans and animals (16,33–37). The classic and most extensively studied glucose-sensing cell is the pancreatic ß-cell, which responds to the rising plasma glucose concentration by increasing its metabolic rate (38). Consequent changes in the concentrations of adenine nucleotides, ATP and ADP, result in the closure of KATP channels on the surface membrane, thereby tipping the overall balance of currents in favor of the inward currents and causing membrane depolarization (Fig. 1B). Early studies on the glucose sensitivity of GLP-1 release, however, suggested that the L-cell glucose sensor was not identical to that of the pancreatic ß-cell (34,36). In perfused rodent and canine intestine preparations, the glucose response was sodium dependent and mimicked by a number of monosaccharides, both metabolizable and nonmetabolizable, including galactose, 3-O-methylglucose and -methylglucopyranoside, but not 2-deoxyglucose. This profile is characteristic of the sensitivity of the sodium glucose cotransporter (SGLT), leading to the proposal that recognition of the sugar by an SGLT is, in some way, linked to the release of GLP-1 (36). However, a number of studies also showed that fructose, which enters cells passively via GLUT5 transporters, can trigger GLP-1 release in humans as well as in rodent intestine, but does not require the presence of luminal Na+ (36,37,39). Clearly, neither the ability of a sugar to be metabolized nor its affinity for a particular transporter can on its own determine its ability to trigger GLP-1 release. In the GLUTag cell line, we found that application of glucose resulted in membrane depolarization, action potential firing, a rise in intracellular calcium (Fig. 2), and the release of GLP-1 (27,28), suggesting that L-cells might themselves contain the necessary glucose-sensing machinery to trigger GLP-1 release in vivo. Although the previous studies on whole intestine had shown that the mechanism was unlikely to be identical to that found in pancreatic ß-cells, we found that GLP-1 secretion from GLUTag cells could be triggered by the sulfonylureas tolbutamide and glibenclamide, which stimulate insulin secretion from ß-cells by closing KATP channels. The presence of a KATP channel–dependent pathway in GLUTag cells is supported by a number of observations, including the ability of tolbutamide to trigger action potentials and a rise in intracellular calcium (Fig. 2), the presence of a tolbutamide-sensitive wash-out current in whole-cell GLUTag recordings, and the detection of ß-cell type KATP channel subunits (Kir6.2 and sulfonylurea receptor 1 [SUR1]) by RT-PCR (27). Furthermore, dynamic ATP measurements in GLUTag cells expressing firefly luciferase showed a prompt increase in the cytoplasmic ATP concentration after the application of glucose (29).
The ability of nonmetabolizable monosaccharides to trigger GLP-1 release from the perfused intestine, however, indicates that KATP channel closure is not the complete story of L-cell glucose sensing. Indeed, further studies in our laboratory showed that nonmetabolizable glucose analogs, such as -methyl-glucopyranoside, could also trigger electrical activity and GLP-1 secretion in GLUTag cells when applied at high enough concentrations (up to 100 mmol/l, compared with a half-maximal effective concentration for glucose of 0.5 mmol/l) (28). The explanation lies in the intrinsic properties of the sodium glucose cotransporter. SGLTs are used by the intestinal brush border to take up sugars from the gut lumen, since they couple the influx of glucose to the uptake of Na+ ions, and thereby use the Na+ gradient to drive the uphill movement of the sugar. The inward Na+ flux associated with substrate uptake at higher sugar concentrations generates an inward current of sufficient magnitude to tip the balance of the membrane potential in favor of depolarization, even without any KATP channel closure, as illustrated in Fig. 1C. Interestingly, the human isoform of SGLT3 was subsequently cloned and characterized and found to behave more like a glucose-regulated ion channel than a transporter, making it a good candidate glucose sensor (40). Because SGLTs form only part of a much larger solute transporter family (41), the possibility exists that other family members might similarly act as sensors of different nutrient substrates.
Although both KATP channel closure and SGLT action are involved in glucose sensing in GLUTag cells, which of these pathways, if either, is primarily responsible for glucose-sensitive GLP-1 release in vivo remains to be established. The potential role of KATP channels in L-cells has raised a lot of interest, since their closure by sulfonylureas, at concentrations similar to those required to stimulate insulin secretion from pancreatic ß-cells, could contribute to the therapeutic efficacy of these compounds as antidiabetic agents. However, in GLUTag cells, KATP channels only opened at glucose concentrations below
Protein-rich meals and protein hydrolysate are well-established triggers of GLP-1 release in whole animals as well as perfused intestinal preparations (15,16,46–48). However, which products of protein digestion are primarily responsible for triggering secretion remains unclear. Studies on humans and isolated rodent intestine have variously found that amino acid mixtures either stimulate GLP-1 release or are ineffective (16,48). GLUTag cells, however, secrete GLP-1 in response to a range of individual amino acids, including alanine, serine, glutamine, asparagine, and glycine (Fig. 3) (29,31).
Glutamine was a more effective stimulus of GLP-1 release from GLUTag cells than the other amino acids tested. At a single-cell level, this amino acid acted as a reliable trigger of action potentials and intracellular calcium elevation (29). Membrane depolarization could be attributed to the small inward currents generated by glutamine uptake on Na+-coupled transporters such as ATA-2, which, although only a few pico-amperes in magnitude, are sufficient to tip the balance in favor of the inward currents, as illustrated in Fig. 1C.
Interestingly, the marked secretory response to glutamine could not be entirely attributed to its effect on the membrane potential and cytoplasmic calcium concentration. Asparagine shares amino acid uptake pathways with glutamine and triggered similar-sized inward currents and calcium changes in GLUTag cells, but was a much less potent stimulus of GLP-1 release (29). The results suggested that glutamine has additional effects on secretion, downstream of the elevation of intracellular calcium. In support of this idea, glutamine still enhanced GLP-1 release under conditions where transporter currents could not alter the membrane potential and the intracellular Ca2+ concentration was elevated (e.g., in cells depolarized with KCl and diazoxide) or permeabilized with
Although glutamine triggered a greater increment in GLP-1 release than the other amino acids tested, alanine and glycine were even more effective at generating membrane depolarization and intracellular Ca2+ elevation in GLUTag cells (31). The mechanism used by these amino acids was clearly distinct from that used by either glucose or glutamine, since it was independent of extracellular Na+ ions. It became clear that GLUTag cells express ionotropic glycine and Protein hydrolysate and peptones are well documented as stimuli of GLP-1 release from perfused intestine as well as STC-1 and NCI-H716 cells (24,47,48,51). The finding that small transporter currents are sufficient to depolarize GLUTag cells raises the question of whether PEPT1 and PEPT2 transporters, which are responsible for intestinal proton-coupled dipeptide uptake, could generate currents of sufficient magnitude to depolarize GLP-1–secreting cells. STC-1 cells transfected with PEPT1 were found to depolarize and secrete GLP-1 on addition of dipeptide substrates (32), suggesting that either dipeptide uptake or the concomitant H+ influx can trigger membrane depolarization. However, as STC-1 cells did not themselves express PEPT1, it remains to be established whether this mechanism operates in GLP-1–releasing cells. Peptones have also been reported to stimulate proglucagon gene transcription via elevation of cAMP in STC-1 cells (48,52,53).
In common with other endocrine cells, GLUTag and STC-1 cells express voltage-gated Ca2+ channels, which open as the membrane depolarizes and allow Ca2+ ions to enter the cell down their electrochemical gradient (32,54). More detailed characterization of the Ca2+ current in GLUTag cells showed that it is activated when the membrane potential is depolarized to approximately –40 mV and above and is comprised of both L-type and N-type components. Although inhibition of both L- and N-type Ca2+ channels was necessary to abolish most of the intracellular Ca2+ response to glucose, GLP-1 secretion was largely eliminated by nifedipine, which blocks only the L-type component. The results suggested that the L-type Ca2+ channels might be more closely linked to the secretory machinery, as has been reported previously for pancreatic ß-cells (55). Although the nutrients discussed above have been shown to act predominantly at the level of the membrane potential, triggering membrane depolarization is not a universal feature of agents that stimulate GLP-1 release. Indeed, in secretion assays performed in the presence of diazoxide and 30 mmol/l KCl, or in permeabilized cells, certain nutrients such as glucose and glutamine were found to enhance GLP-1 release, even when they were not able to stimulate further elevation of the intracellular calcium concentration (29). The L-cell might therefore be compared with the pancreatic ß-cell, in which the effect of depolarizing agents is enhanced by factors that target downstream pathways such as the exocytotic machinery. Although amplification of the secretory response in pancreatic ß-cells is incompletely understood, a range of pathways have been implicated, including elevation of intracellular cAMP and activation of protein kinases A and C, as well as less well-characterized effects of metabolites such as glucose (56). In the L-cell, agents that might predominantly trigger secretion by targeting pathways downstream of membrane depolarization include hormones, neurotransmitters, and fatty acids.
While fat ingestion has long been known to be a potent trigger of GLP-1 release, more recent studies in humans as well as isolated cells have attributed the response primarily to the subgroup of longer-chain unsaturated fatty acids (57,58). In fetal rat intestinal cells, GLP-1 release triggered by long-chain monounsaturated fatty acids was not dependent on fatty acid oxidation (57). The recent identification of G protein–coupled receptors with specificity for fatty acids of particular chain length and degree of saturation has raised the possibility that these surface receptors might be involved in fatty acid recognition in GLP-1–secreting cells. Two receptors that have received recent attention are GPR40 and GPR120, both of which preferentially respond to longer-chain unsaturated fatty acids. GPR40 is highly expressed in insulinoma cells and is postulated to play a role in fatty acid modulation of insulin release (59). GPR120, by contrast, is strongly expressed in the gut and STC-1 cells, but not in pancreas. In STC-1 cells, knock down of GPR120 but not GPR40 by siRNA led to a reduction in both -linolenic acid–induced calcium increments and GLP-1 secretion (60). Both GPR40 and GPR120 are also expressed in GLUTag cells, as determined by RT-PCR (F.R., unpublished data). The downstream pathways triggered by GPR120 activation, and the possible involvement of other fatty acid–sensing mechanisms in GLP-1–secreting cells, remain to be fully established. The cases for short- and medium-chain fatty acids as physiological stimuli of GLP-1 release are less well established. Medium-chain fatty acids have been reported to trigger release of calcium from intracellular stores and hormone secretion from the STC-1 cell line (61), whereas short-chain fatty acids were found to stimulate GLP-1 and peptide YY release from perfused rodent intestine (62). Whether the latter finding is related to the recent identification of a receptor for short-chain fatty acids, GPR43, in peptide YY–positive cells (63) remains to be investigated. Bile acids, like fatty acids, also act as GLP-1 secretagogues in perfused intestinal preparations (62). A recent study suggests that the newly identified G protein–coupled receptor, TGR-5 (GPR131), might be involved in bile acid–triggered GLP-1 release (64). Knock down of TGR-5 in STC-1 cells by small interference RNA led to a significant reduction in lithocholic acid–triggered GLP-1 release. Further studies suggested that the downstream pathway linking TGR-5 activation to GLP-1 release might be mediated by elevated cAMP levels.
Rodent L-cells secrete GLP-1 in response to several gut peptides such as GIP and calcitonin gene–related protein (CGRP), the receptors for which are coupled to the elevation of cAMP (65,66). CGRP is a peptide found throughout the enteric nervous system, which, if released in the vicinity of the L-cells, might also play a role in neurally triggered GLP-1 secretion (67). Although effective in the perfused rodent intestine (68,69), administration of supraphysiological concentrations of GIP in humans does not elevate GLP-1 concentrations (4), and the hormonal responsiveness of human L-cells remains unclear. Elevation of cAMP has been shown to stimulate GLP-1 release in a variety of preparations, including GLUTag and fetal rat intestinal cells. In the GLUTag cell, cAMP elevation after application of forskolin and isobutylmethylxanthine is a potent trigger of both proglucagon gene transcription and GLP-1 release (27,70). Thus, forskolin and isobutylmethylxanthine trigger approximately a threefold elevation of GLP-1 release under a range of conditions, not only in the presence of nutrients such as glucose and glutamine, or the sulfonylurea tolbutamide, but also in the absence of added nutrient (27). It also enhanced GLP-1 release in the presence of 30 mmol/l KCl plus diazoxide (Fig. 4E), indicating that it amplifies secretion downstream of the Ca2+ signal.
GLP-1 secretion has also been reported in response to hormones that activate phospholipase C, such as gastrin-releasing peptide (the mammalian homolog of bombesin) and acetyl choline (acting on certain muscarinic receptors) (69,71,72). Phospholipase C activation results in the formation of inositol-1,4,5-triphosphate (IP3) and diacylglycerol, which could potentially stimulate GLP-1 release by triggering Ca2+ release from IP3-sensitive stores and/or by activating protein kinase C, respectively. In GLUTag cells loaded with fura2, 100 nmol/l bombesin triggered an increase in intracellular [Ca2+] that was abolished by 10 µmol/l thapsigargin (Fig. 4A and B), suggesting that the Ca2+ increment results from the release of Ca2+ from intracellular stores, as has been reported previously in STC-1 cells (61,73). In secretion experiments, bombesin stimulated GLP-1 release approximately threefold in both the presence and absence of glucose (Fig. 4C). To investigate the relative importance of stored Ca2+ and of signals acting downstream of Ca2+ elevation, we measured GLP-1 secretion from GLUTag cells in the presence of thapsigargin and in cells depolarized by 30 mmol/l KCl plus diazoxide. Thapsigargin abolished bombesin-triggered GLP-1 release but did not inhibit glucose-triggered secretion (Fig. 4D), suggesting that the release of stored Ca2+ is an essential component of the response to bombesin in the absence of nutrient. Bombesin also failed to enhance GLP-1 release when Ca2+ was elevated by 30 mmol/l KCl plus diazoxide (Fig. 4E), showing that it does not amplify secretion downstream of the Ca2+ signal. By RT-PCR, GLUTag cells were found to express the three major IP3 receptor isoforms: IP3R1, IP3R2 (also known as IP3R5), and IP3R3 (Fig. 4F), consistent with the involvement of these receptors in the calcium response triggered by activation of hormone receptors linked to phospholipase C.
Experiments on GLP-1–secreting cell lines have identified a number of pathways underlying the detection of nutritional and hormonal stimuli (Fig. 5). Changes in the membrane potential arising from, e.g., Na+-coupled substrate uptake or KATP channel closure, are coupled to the opening of voltage-gated Ca2+ channels, allowing Ca2+ entry and a consequent elevation of the intracellular Ca2+ concentration, which stimulates release of GLP-1–containing secretory vesicles. Hormones and neurotransmitters such as bombesin, acetyl choline, GIP, and CGRP, as well as fatty acids and bile acids, may enhance secretion by increasing cAMP concentrations or by triggering the release of Ca2+ from intracellular stores. Having identified these pathways in GLUTag and STC-1 cells, however, it is now critical that these results are verified in L-cells that are not immortalized.
Finding new ways to enhance GLP-1 secretion in vivo will depend on both identifying the pathways involved in GLP-1 release and establishing their interrelationships. Ultimately, the aim must be to improve the treatment of type 2 diabetes or obesity, by increasing the circulating levels of GLP-1 or peptide YY sufficiently to increase insulin release or modulate satiety. Whether this will be possible using L-cell secretagogues alone, and the effectiveness of combining secretagogues with DPPIV inhibitors, are exciting avenues for future exploration.
F.M.G. is a Wellcome Trust Senior Research Fellow in Clinical Science; F.R. is the Meres Research Associate at St. Johns College, Cambridge; and P.S.W. is a Churchill Scholar. We thank these funding bodies for their support.
This article is based on a presentation at a symposium. The symposium and the publication of this article were made possible by an unrestricted educational grant from Servier. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Received for publication March 23, 2006 and accepted in revised form May 13, 2006
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||