Timing and source of incretin secretion

The enteroendocrine L-cells that synthesise and release GLP-1 are found scattered throughout the small and large intestine, with the highest density in the distal ileum and colon (Ref. Reference Eissele9). GIP, by contrast, is produced by an anatomically distinct cell type known as K-cells, which are predominantly located in the duodenum (Refs Reference Polak10, Reference Buchan11). Perhaps in part reflecting the different locations of K- compared with L-cells, the circulating concentrations of GIP and GLP-1 exhibit distinct profiles following food ingestion. Depending on the meal composition, GLP-1 secretion can exhibit a biphasic profile, with an early phase occurring soon after food intake and lasting for 15–30 min, and a late phase persisting for 1–2 h or longer (Ref. Reference Herrmann12). GIP secretion occurs with a similar short time delay, and may also remain elevated for several hours (Refs Reference Cataland13, Reference Elliott14). In humans, glucose can be detected in the proximal duodenum within 5 min of the ingestion of a liquid glucose load, correlating with the time taken to see the first rapid elevation of GLP-1 and GIP concentrations in the bloodstream (Refs Reference Balks15, Reference Schirra16).

While the early release of GIP is readily explained by the arrival of nutrients in the duodenum with its high local density of K-cells, it has long posed a dilemma that the rapid secretion of GLP-1 occurs before nutrients reach the majority of the L-cells in the ileum and colon. This gave rise to the idea that the rapid phase of GLP-1 secretion must be triggered indirectly by neuronal signals or hormonal factors arising in the proximal intestine and stimulating the distal L-cell population (Refs Reference Hansen and Holst17, Reference Rocca and Brubaker18). However, L-cells have been identified at low density in the duodenum (Ref. Reference Mortensen19) and in higher numbers in the jejunum (Ref. Reference Eissele9). Glucose infusion studies in human subjects showed that GLP-1 secretion was stimulated only when the rate of glucose entry into the duodenum exceeded the duodenal absorptive capacity – that is, when nonabsorbed glucose would leak through to the jejunum, and potentially stimulate L-cells in that region (Ref. Reference Schirra16). In the same studies, GIP secretion was stimulated by even low duodenal glucose infusion rates, consistent with the more proximal location of the majority of K-cells. Interestingly, a proportion of cells in the proximal small intestine reportedly express both GLP-1 and GIP (Refs Reference Mortensen19, Reference Theodorakis20). Taken together with the finding that the rapid phase of GLP-1 secretion is preserved in patients after ileal resections or proctocolectomy, the data suggest that the early release of GLP-1, which accounts for the majority of the incretin effect in normal individuals, is triggered by local luminal nutrient-sensing pathways (Ref. Reference Nauck21) (Fig. 1). Nevertheless, the relative contribution of the ileal and colonic L-cells to the incretin effect remains surprisingly unclear.

Figure 1.

Sites of secretion of GLP-1 and GIP. Plasma levels of glucagon-like peptide 1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP) increase rapidly after the ingestion of a meal as a result of the direct stimulation of both L- and K-cells located in the duodenum. The later phase of GLP-1 secretion coincides with the time taken for nutrients to reach the L-cells in the lower ileum. As well as entering the circulation and having direct effects on their target tissues, such as the pancreatic β-cell (incretin effect), GLP-1 might also mediate its effects via receptors in close proximity to the enteroendocrine cells, such as those expressed by vagal afferent neurons in the portal vein. Enteroendocrine cells secreting hormones like GLP-1 and GIP are found scattered through the epithelial layer, as shown in the inset (bottom left). L-cells from the distal ileum and colon secrete peptide YY (PYY), GLP-1 and GLP-2. The extent to which GLP-1 from distal L-cells contributes to the incretin effect is unclear, but it also suppresses appetite and slows peristalsis (the ileal brake). PYY acts centrally as a satiety signal, and GLP-2 stimulates regeneration and repair of the intestinal epithelium.

Incretin hormone processing

GLP-1 is post-translationally cleaved from the product of the proglucagon gene (GCG), which is expressed not only in the L-cell but also in pancreatic α-cells and subsets of neurons in the brainstem. Alternative processing of the prohormone is largely attributable to the differential expression of prohormone convertase (PC) enzymes. Whereas pancreatic α-cells express PC2 (PCSK2) and generate glucagon, L-cells in the gut express PC1/3 (PCSK1) and produce the bioactive peptides GLP-1, GLP-2 and oxyntomodulin (Ref. Reference Holst7) (Fig. 2). Proglucagon processing in the brainstem is believed to follow a similar pathway to that in the intestine. In fact, GLP-1 exists in two active forms, GLP-1(7–36)amide and GLP-1(7–37), which are produced in varying proportions across species but have equipotent insulinotropic activity (Ref. Reference Orskov, Wettergren and Holst22). The C-terminal moiety of proglucagon is processed almost completely to GLP-1 and GLP-2 in L-cells, but only 20–40% of the N-terminal region is released as the active peptide oxyntomodulin, with the remainder taking the form of glicentin (60–80%) (Ref. Reference Meier and Nauck23). Certain doubts therefore remain about the physiological importance of oxyntomodulin, and its broadly similar effects to those of GLP-1 probably reflect the fact that its action in vivo is largely mediated through the GLP-1 receptor (Refs Reference Baggio24, Reference Druce25). By contrast to proglucagon, the proGIP precursor in K-cells is not believed to contain any bioactive peptide sequences in addition to that of GIP. ProGIP processing in K-cells, like that of proglucagon in L-cells, is attributed to the action of PC1/3 (Ref. Reference Ugleholdt26).

Figure 2.

Proglucagon processing in different cell types. Proglucagon processing in pancreatic α-cells generates the major products glucagon, GRPP (glicentin-related pancreatic polypeptide) and the major proglucagon fragment. In the brainstem and intestinal L-cells, by contrast, processing generates glucagon-like peptide 1 (GLP-1), GLP-2, glicentin, oxyntomodulin and GRPP. Removal of the first six amino acids of GLP-1 generates the active forms GLP-1(7–36)amide and GLP-1(7–37). The major processing enzymes are believed to be prohormone convertase 2 (PC2) in the α-cells, and PC1/3 in the L-cells. Solid arrows depict points where cleavage is largely complete; dashed arrows depict sites of incomplete processing (in L-cells ~20–40% of the N-terminal region is released as the active peptide oxyntomodulin). The uncoloured segments of proglucagon between glucagon and GLP-1 and between GLP-1 and GLP-2 correspond to intervening peptides 1 and 2, respectively. For further details see Ref. Reference Rouillé, Martin and Steiner135.

Incretin hormone degradation

Following their secretion, both GLP-1 and GIP are rapidly degraded by the enzyme dipeptidyl peptidase 4 (DPP4), which is ubiquitously expressed on the surface of endothelial and epithelial cells and exists as a soluble form in plasma (Refs Reference Demuth, McIntosh and Pederson27, Reference Hansen28). DPP4 has a preference for substrate peptides containing a proline or alanine residue at position 2 (Ref. Reference Mentlein29) and readily cleaves N-terminal dipeptides from GLP-1 and GIP, forming GLP-1(9–37) or GLP-1(9–36) amide and GIP(3–42), which lack activity at their respective receptors (Ref. Reference Holst7). As a result, it is estimated that only 10–15% of the secreted active form of GLP-1 passes into the systemic circulation, where it has a short half-life of <2 min (Refs Reference Baggio and Drucker6, Reference Hansen28), and only after larger meals does the active GLP-1 level rise sufficiently to be detected in the peripheral circulation. GIP inactivation by DPP4 is slightly slower (t_½ ~6 min in humans and 2 min in rodents) (Refs Reference Deacon30, Reference Kieffer, McIntosh and Pederson31), and the peak postprandial GIP levels are correspondingly higher – factors that probably contribute to the effectiveness of GIP as a circulating physiological incretin hormone.

Site of action of incretins

While the best-characterised effect of GLP-1 is its stimulatory action on the pancreatic β-cell, GLP-1 receptors are found in a variety of extrapancreatic sites, including the lungs, kidneys, stomach, heart, central nervous system, major blood vessels (including the portal vein) and vagal neurons (Refs Reference Baggio and Drucker6, Reference Vahl32, Reference Bucinskaite33). The rapid clearance of GLP-1 from the circulation suggests that there may also be primary sites of GLP-1 action located closer to its site of secretion. The nodose ganglion, which contains the cell bodies of afferent vagal neurons innervating the viscera, expresses receptors for GLP-1, GLP-2 and GIP, although only GLP-1 has been shown to reliably elicit changes in the electrical activity of the neurons themselves (Refs Reference Nakagawa34, Reference Kakei35). The findings suggest that GLP-1, like cholecystokinin (CCK), might exert a proportion of its effect by acting on nerve endings located in close proximity to the enteroendocrine cells themselves (Refs Reference Simasko and Ritter36, Reference Simasko37). This idea is supported by a recent report that the appetite-suppressive effect of systemic GLP-1 was inhibited when the GLP-1 receptor antagonist exendin(9–39) was administered peripherally but not centrally (Ref. Reference Williams, Baskin and Schwartz38).

In vitro models

Three cell line models are widely used to investigate the secretion of GLP-1, GIP and other gut hormones. The GLUTag cell line was originally derived from endocrine colonic tumours taken from a transgenic mouse expressing SV40 large T antigen under the control of the proglucagon promoter (Ref. Reference Drucker39), and has proved to be a reliable model of GLP-1 secretion, responding to a range of physiologically relevant stimuli (Ref. Reference Brubaker, Schloos and Drucker40). The secretin tumour cell line (STC-1), derived from an endocrine tumour of the mouse small intestine, secretes a range of gut hormones including GLP-1, GIP, CCK and secretin (Ref. Reference Rindi41). It has been used as a model to investigate the release of each of these peptides, although it is unclear whether its ability to secrete such a range of hormones is due to a loss of cell differentiation that may also impact on its expression of stimulus–secretion coupling components, or whether this actually reflects the behaviour of distinct populations of enteroendocrine cells that similarly release a range of hormonal products. As there are very few model systems for studying GIP secretion in vitro, subclones of STC-1 cells have been generated with increased GIP expression, such as GTC-1, STC6-14 and GIP-Ins (Refs Reference Ramshur, Rull and Wice42, Reference Cheung43, Reference Kieffer44). NCI-H716 is a GLP-1-secreting human cell line derived from a caecal adenocarcinoma (Ref. Reference Reimer45). It responds to a limited range of secretagogues, but was found to exhibit impaired physiological regulation of the proglucagon promoter (Ref. Reference Cao46). The validity of this cell line as a model of the human L-cell remains to be established.

Primary cultured enteroendocrine cells provide an alternative system for the in vitro study of gut hormone secretion. Fetal rat intestinal cell (FRIC) cultures secrete GLP-1 in response to a wide array of signalling pathways and extracellular mediators, although are not glucose responsive (Ref. Reference Brubaker and Vranic47). While this lack of glucose responsiveness could be used to argue that L-cells are not themselves glucose sensitive, it is possible that fetal L-cells, like pancreatic β-cells at the same age, do not express a full adult repertoire of glucose-sensing machinery. Primary adult canine K- and L-cells, partially purified by counterstream centrifugal elutriation, have been the topic of a few in vitro studies, although they are not a readily available tool for the majority of laboratories (Refs Reference Kieffer48, Reference Damholt, Buchan and Kofod49). Mixed primary intestinal cultures from the adult mouse have recently been developed as an alternative tool to assay secretion of a range of intestinal hormones in vitro. Together with the generation of transgenic mice with fluorescently labelled K- or L-cells, this technology provides new opportunities to purify, identify and characterise primary adult enteroendocrine cells (Refs Reference Reimann50, Reference Parker51).

Enteroendocrine detection of sweet stimuli

The incretin hormones augment insulin secretion after oral glucose intake, yet how sugars in the gut lumen elicit secretion of GLP-1 and GIP from L- and K-cells remains controversial. Early studies using the GLUTag cell line showed that glucose elicits electrophysiological events that trigger action potential firing, Ca²⁺ entry via L-type Ca²⁺ channels, and GLP-1 secretion (Ref. Reference Reimann and Gribble52).

Metabolism and K_ATP channel closure

As in the pancreatic β-cell, a component of the glucose responsiveness can be attributed to the closure of ATP-sensitive K⁺ channels (K_ATP channels), brought about by changes in the relative proportions of ATP and Mg²⁺–ADP consequent upon glucose metabolism (Fig. 3). Recent studies have confirmed that (1) the K_ATP channel subunits Kir6.2 (KCNJ11) and SUR1 (sulfonylurea receptor; ABCC8) are expressed at high levels in purified mouse K- and L-cells, (2) functional K_ATP currents are found in the primary colonic L-cells, (3) sulfonylureas stimulate GLP-1 and GIP secretion in primary cultures, and (4) Kir6.2 is detectable by immunostaining in human K- and L-cells (Refs Reference Reimann50, Reference Parker51, Reference Nielsen53).

Figure 3.

Proposed glucose-sensing mechanisms in primary L-cells. (a) L-cells show ‘open-type’ morphology with a long apical process extending towards the gut lumen, as shown by the pattern of Venus fluorescence of a Venus-labelled L-cell from a colonic crypt. (b) This is a typical electrophysiological trace from a primary L-cell treated with the K_ATP-channel inhibitor tolbutamide, showing membrane depolarisation and action potential firing triggered by the drug. (c) Electrogenic transport of glucose via the apically expressed sodium-coupled glucose transporter SGLT1 generates a sufficient inward current to depolarise the cell membrane (ΔΨ) and trigger action potentials, leading to the opening of voltage-gated Ca²⁺ channels. The consequent rise in intracellular Ca²⁺ is the primary trigger of exocytosis and glucagon-like peptide 1 (GLP-1) secretion. In addition, glucose metabolism results in the closure of ATP-sensitive K_ATP channels, contributing to the excitability of the cell and GLP-1 secretion.

In a variety of glucose-sensitive tissues, it is believed that glucokinase (GCK), the hexokinase with the lowest affinity for glucose, acts as a master controller of the metabolic rate and thereby governs the activity of downstream signalling pathways such as K_ATP channel activity. The low affinity of GCK for glucose (S_0.5 of ~8 mm) (Ref. Reference Matschinsky54) and its lack of product inhibition enable it to link metabolic flux to the extracellular glucose concentration over a physiologically relevant concentration range. Activating and inactivating mutations of the GCK gene cause syndromes of hypo- and hyperglycaemia in humans (Refs Reference Glaser55, Reference Stoffel56), reflecting the critical role of this enzyme in the control of insulin secretion. Immunostaining and expression analysis have demonstrated that GCK is also expressed at high levels in enteroendocrine K- and L-cells (Refs Reference Theodorakis20, Reference Cheung43, Reference Reimann50, Reference Parker51, Reference Jetton57).

Despite the clear presence of GCK and K_ATP channels in primary K- and L-cells, the signalling role of metabolism in enteroendocrine cells remains uncertain as there is little evidence for a stimulatory effect of sulfonylureas or elevated blood glucose on incretin secretion in an intact animal. Indeed, (1) sulfonylureas had no effect on the peak GLP-1 and GIP responses to an oral glucose tolerance test in humans (Ref. Reference El-Ouaghlidi58), (2) patients with inactivating mutations in GCK did not exhibit impaired GLP-1 and GIP secretion following glucose ingestion (Ref. Reference Murphy59), (3) mice lacking the Kir6.2 subunit of the K_ATP channel exhibited increased, rather than decreased, plasma GIP levels after an oral glucose challenge (Ref. Reference Miki60), and (4) an acute episode of hyperglycaemia reduced rather than enhanced GLP-1 and GIP responses to glucose ingestion (Ref. Reference Vollmer61). While the latter finding was in part influenced by the slowing of gastric emptying by hyperglycaemia itself, and it could be argued that the altered blood glucose levels in patients with GCK mutations might offset the shifted glucose sensitivity of GCK in enteroendocrine cells, the balance of evidence does not favour a significant role for L- and K-cell GCK and K_ATP channels in determining the peak GLP-1 and GIP responses to glucose ingestion. However, in a perfused pig intestinal preparation, elevating the vascular perfusate glucose concentration from 5 to 11 mm in the presence of a constant luminal glucose load (3.5 mm) did stimulate GLP-1 secretion (Ref. Reference Hansen62), raising the possibility that glucose arriving by the vascular route may be able to modulate secretion at lower luminal nutrient levels.

Bitter stimuli

Incretin-secreting cells have been found to express bitter-taste receptors of the type 2 taste receptor (TAS2R) family, and bitter agents were shown to stimulate STC-1 cells, leading to the suggestion that bitter tastants might modulate incretin release in vivo (Refs Reference Rozengurt73, Reference Chen78). In this respect it is interesting that a genetic association between a single-nucleotide polymorphism (SNP) in TAS2R9 and impaired glucose homeostasis was recently reported in the Amish population, and that dose-dependent GLP-1 secretion could be triggered from NCI-H716 cells by the TAS2R9 agonist ofloxacin (Ref. Reference Dotson79). However, it remains possible that some of the bitter agents employed in these studies might have off-target effects, as for instance fluoroquinolones like ofloxacin have been reported to depolarise cells by blocking K⁺ channels (Refs Reference Kang80, Reference Saraya81). In human subjects, the bitter tastants quinine and naningin did not affect gastric emptying or satiety when given by intragastric infusion (Ref. Reference Little75), and there are currently no reports of bitter compounds affecting incretin secretion in vivo.

Detection of other nutrients

Because of the insulinotropic action of GLP-1 and GIP and the associated importance of the incretins in glucose homeostasis, much research has focused on the mechanisms of carbohydrate sensing by L- and K-cells. However, it is widely accepted that other macronutrients can also stimulate hormone release from these cells (Refs Reference Herrmann12, Reference Elliott14).

G_s-coupled receptors

The activation of G_s-coupled receptors and consequent elevation of intracellular cyclic AMP (cAMP) has been shown to be a strong stimulus of GLP-1 release in many in vivo and in vitro studies (Refs Reference Reimer45, Reference Kieffer48, Reference Damholt, Buchan and Kofod49, Reference Simpson94, Reference Brubaker95, Reference Buchan96). cAMP acts through several downstream effectors, including protein kinase A (PKA), exchange proteins directly activated by cAMP (EPACs; RAPGEF3 and RAPGEF4) and ion channels such as cyclic-nucleotide-gated channels. In GLUTag cells, cAMP elevates proglucagon gene transcription via PKA and EPAC-dependent pathways, but also directly stimulates GLP-1 secretion (Refs Reference Drucker39, Reference Brubaker, Schloos and Drucker97) (Fig. 4). The enhancement of vesicle release not only includes a stimulatory effect on the secretory machinery, but also involves a depolarising action of cAMP on the plasma membrane, where it acts synergistically with glucose (Ref. Reference Simpson94). Although the molecular targets of cAMP have not yet been identified in primary K- and L-cells, it is clear that the combination of the adenylate cyclase activator forskolin and the phosphodiesterase inhibitor IBMX (3-isobutyl-1-methylxanthine) both elevates Ca²⁺ in mouse colonic L-cells and stimulates GLP-1 and GIP secretion synergistically with glucose in primary intestinal cultures (Refs Reference Reimann50, Reference Parker51).

Figure 4.

G-protein-coupled receptor signalling pathways in the GLUTag cell line. Nutrient signalling via G-protein-coupled receptor pathways mediated by G_s and G_q stimulates glucagon-like peptide 1 (GLP-1) secretion. G_q stimulation results in activation of protein kinase C (PKC) and inositol (1,4,5)-trisphosphate-dependent release of Ca²⁺ from intracellular stores. G_s activation elevates intracellular cAMP levels, with consequent effects on the membrane potential (ΔΨ) and Ca²⁺ entry, as well as targets affecting secretion downstream of the Ca²⁺ signal. Inhibitory, G_i-coupled receptors have also been implicated in downregulation of incretin secretion (not shown).

Hormones and neurotransmitters known to modulate GLP-1 and/or GIP secretion via G_s-coupled receptors include pituitary adenylate cyclase activating polypeptide (PACAP/ ADCYAP1), calcitonin-gene-related peptide (CGRP; encoded by CALCA) and GIP itself (Refs Reference Hansen and Holst17, Reference Simpson94, Reference Brubaker, Schloos and Drucker97, Reference Roberge and Brubaker98). It should be noted, however, that GIP administration, although repeatedly found to stimulate GLP-1 release in rodents, does not raise GLP-1 levels in humans, indicating that species differences need to be considered. More recently, GPR119 and G-protein-coupled bile acid receptor 1 (GPBAR1/TGR5), also believed to be G_s-coupled, have come under the spotlight as potential novel drug targets to stimulate hormone secretion from L-cells. GPR119 is reportedly a receptor for the natural ligands oleoylethanolamide (OEA) and lysophosphatidylcholine (Ref. Reference Overton99), and its knockout in mice results in reduced postprandial GLP-1 levels (Ref. Reference Lan100). A small-molecule agonist of GPR119 has been shown to stimulate both GLP-1 and GIP secretion in vivo (Ref. Reference Chu101), consistent with the finding that this receptor is expressed at relatively high levels in purified mouse K- and L-cells (Refs Reference Reimann50, Reference Parker51). GPBAR1 has been identified as a receptor for bile acids (Ref. Reference Kawamata102) and is postulated to contribute to the enhanced secretion of GLP-1 observed after gastric bypass surgery (Ref. Reference le Roux103), which increases the delivery of bile components to L-cell-rich regions of the ileum and colon (Ref. Reference Patti104). In isolated perfused rat colon, the luminal infusion of bile acids increased GLP-1 levels in the portal venous effluent (Ref. Reference Plaisancie105), and in STC-1 cells GPBAR1 activation resulted in both elevated cAMP levels and GLP-1 secretion (Ref. Reference Katsuma, Hirasawa and Tsujimoto106). Consistent with this idea, GPBAR1 was also found to be highly expressed in purified mouse colonic L-cells (Ref. Reference Reimann50).

G_q-coupled receptors

Binding of agonists to G_q-coupled receptors leads to activation of PKC and inositol (1,4,5)-trisphosphate [Ins(1,4,5)P ₃]-dependent release of Ca²⁺ from intracellular stores, and both these pathways have been demonstrated to modulate incretin secretion (Fig. 4). Bombesin, gastrin-releasing peptide and muscarinic agonists, for example, stimulate incretin release in vivo and in vitro through G_q-coupled pathways (Refs Reference Balks15, Reference Dumoulin89, Reference Brubaker95, Reference Abello107, Reference Roberge, Gronau and Brubaker108, Reference Anini, Hansotia and Brubaker109). Although the downstream molecular targets of PKC in K- and L-cells remain to be identified, the finding that bombesin triggers rapid Ca²⁺ transients in cell lines and primary L-cells, which at least in the former case are sensitive to the sarcoplasmic reticulum Ca²⁺-ATPase inhibitor thapsigargin (Ref. Reference Reimann, Ward and Gribble110), speaks for a key involvement of Ins(1,4,5)P ₃-dependent Ca²⁺ stores in G_q-coupled incretin secretion. More-recently identified G_q-coupled receptors that are potentially involved in GLP-1 and GIP secretion are GPR40 (FFAR1) and GPR120. These are responsive to long-chain fatty acids and have been implicated in fatty acid sensing by enteroendocrine cells lines (Refs Reference Tanaka111, Reference Hirasawa112). In support of a role for GPR40 in incretin secretion in vivo, GPR40-null mice were found to exhibit diminished GIP and GLP-1 secretion when fed a high-fat diet (Ref. Reference Edfalk, Steneberg and Edlund113). Plasma GLP-1 levels were also elevated in rats after administration of the GPR120 ligand α-linolenic acid (Ref. Reference Tanaka114), although selective activation of GPR120 by this experimental approach seems unlikely. Despite the current lack of strong evidence that GPR40 and GPR120 underlie the normal incretin response to ingested fat, both L- and K-cells undoubtedly express high levels of both these receptors (Refs Reference Reimann50, Reference Parker51).

Related receptors have also been identified for medium-chain (GPR84) (Ref. Reference Wang115) and short-chain (GPR41/FFAR3 and GPR43/FFAR2) fatty acids, with expression of the latter two demonstrated in colonic L-cells by immunostaining (Refs Reference Tazoe116, Reference Karaki117). In this regard, it is interesting that luminal short-chain fatty acids, produced naturally from the bacterial fermentation of dietary fibre, have been reported to stimulate secretion of GLP-1 in perfused rat intestine (Ref. Reference Plaisancie118), and PYY (a peptide found located with GLP-1 in colonic L-cells) in rabbit colon (Ref. Reference Longo119). GPR41 is believed to be coupled to G_i/o and preferentially responsive to C₃–C₅-chain-length fatty acids, whereas GPR43 may couple via G_i/G_q and prefer C₂/C₃ chain lengths (Ref. Reference Covington120). Although there is clearly scope for the differential stimulation of alternative signalling pathways by different short-chain fatty acids, a functional link between these receptors and incretin secretion has not yet been demonstrated.

G_i-coupled receptors

In addition to the GPRs that stimulate incretin secretion, evidence also favours a role for inhibitory, G_i-coupled receptors, responding for example to somatostatin. Somatostatin is a potent inhibitor of a wide range of endocrine cell types, and has been shown to inhibit incretin secretion in both whole intestinal preparations and cell lines (Refs Reference Kieffer44, Reference Kieffer48, Reference Buchan96, Reference Hansen121). Released locally from intestinal D-cells, it is possible that somatostatin provides an additional level of control on the release of GIP and GLP-1, although the complexities of this interaction in vivo remain to be explored.