Stimulation of incretin secreting cells

Abstract

The incretin hormones glucose-dependent insulinotropic polypeptide (GIP) and glucagon like peptide-1 (GLP-1) are secreted from enteroendocrine cells in the gut and regulate physiological and homeostatic functions related to glucose control, metabolism and food intake. This review provides a systematic summary of the molecular mechanisms underlying secretion from incretin cells, and an understanding of how they sense and interact with lumen and vascular factors and the enteric nervous system through transporters and G-protein coupled receptors (GPCRs) present on their surface to ultimately culminate in hormone release. Some of the molecules described below such as sodium coupled glucose transporter 1 (SGLT1), G-protein coupled receptor (GPR) 119 and GPR40 are targets of novel therapeutics designed to enhance endogenous gut hormone release. Synthetic ligands at these receptors aimed at treating obesity and type 2 diabetes are currently under investigation.

The incretin effect

After the proposal of ‘secretin’ as a blood-borne factor coordinating the arrival of food in the upper small intestine with exocrine pancreatic secretion [Bayliss and Starling, 1902], a similar ‘hormonal’ axis promoting insulin release from the endocrine pancreas was proposed by La Barre in 1932, coining the term ‘incretin’. The incretin effect describes the fact that, even when the insulin-secreting and inherently glucose-sensitive pancreatic β cell is exposed to matching blood glucose levels, only approximately half as much insulin is secreted when glucose is provided by an intravenous infusion compared with oral ingestion/intestinal absorption of glucose [Perley and Kipnis, 1967]. We now know of a number of gut-derived hormones that boost glucose-dependent insulin release, including for example, sulfated cholecystokinin (CCK8S) [Zawalich et al. 1986], but the ‘true incretins’ known to date are glucose-dependent insulinotropic polypeptide (GIP) and glucagon like peptide-1 (GLP-1) [Dupre et al. 1973; Kreymann et al. 1987]. GIP and GLP-1 are produced and secreted from specialized endocrine cells of the gut, enteroendocrine cells (EECs), namely K and L cells, respectively (Figure 1).

Figure 1.

Incretin hormone secretion from K and L cells.

Glucose dependent insulinotropic polypeptide (GIP) is secreted from K cells, which are predominantly found in the duodenum, whereas glucagon-like peptide-1 (GLP-1) is secreted from L cells, which increase in numbers in the distal intestine. Both cell types are so-called open enteroendocrine cells with direct contact to the intestinal lumen, allowing sampling of the chyme. Nutrient transporters expressed in K and L cells play an important part in this chemosensation (see text for details). The difference in location, however, is likely to impact on the observed hormone responses to nutrient ingestion. It is now thought that sufficient L cells are present in the proximal intestine to account for early rises of GLP-1 under normal conditions (left hand side). However, when absorption is blocked (right hand side), the increased delivery to the distal intestine brings more L cells in contact with nutrients and/or their fermentation/putrification products, resulting in elevated GLP-1 levels. This in turn slows gastric emptying, potentially thereby prolonging exposure of proximal cells to chyme. While glucose stimulated GIP secretion is thus abolished in Sglt-1 knockout mice, emphasizing the importance of this transporter in K cells, GIP (as well as GLP-1) levels in B⁰at1 knockout animals are elevated in comparison with wildtype litter mates 60 min after food consumption.

Incretin secreting cells

K and L cells are subsets of enteroendocrine cells found scattered in the intestinal epithelium that, in their totality, contribute ~1% of intestinal epithelial cells. In recent years, transgenic technology has allowed the development of mice which express fluorescent protein reporters under the control of enteroendocrine hormone promoters [Reimann et al. 2008; Parker et al. 2009; Chandra et al. 2010; Wang et al. 2011; Suzuki et al. 2013]. This technique enabled the isolation, purification and systematic characterization of these otherwise elusive cells and led to rapid strides in our understanding of EEC biology. For one, the concept of these cells being uni/bi hormonal has been challenged and recent evidence points towards them being more plurihormonal than previously thought [Egerod et al. 2012; Habib et al. 2012; Sykaras et al. 2014]. Flow cytometric (FACS) analysis and immunostaining have revealed that, while in the colon, most L cells contained GLP-1 and PYY, the picture is different in the upper small intestine. Most small intestinal L cells contained CCK, ~10% were GIP positive and ~20% were PYY positive [Habib et al. 2012]. Recently, we identified insulin-like peptide-5 (INSL5) to be a product of colonic but not small intestinal L cells and showed that its levels were increased in calorie-restricted mice and reduced after feeding. INSL5 administration increased food intake in wildtype mice but not in mice lacking its receptor RXFP4, contrasting with the anorexic properties of other L-cell hormones [Grosse et al. 2014].

Like other EECs, K and L cells are polarized and exhibit an ‘open-type’ morphology with an apical pole consisting of microvilli in direct contact with the lumen and a broad basolateral side from which dense core secretory vesicles exocytose [Kieffer and Habener, 1999]. A unique feature of these EECs was revealed using laser scanning confocal microscopy of CCK-GFP and PYY-GFP cells. Pseudopod-like processes were observed at the base of the cells, extending towards adjacent cells and forming synapse-like structures, thereby presumably exerting a paracrine effect on neighbouring enterocytes [Chandra et al. 2010; Bohórquez et al. 2011]. However, the specific function of these pseudopod structures is yet to be established.

Another important aspect of the polarization of EECs is that their apical and basolateral surfaces differ in their accessibility to luminal and vascular factors; whether sensory receptors are located on the apical or basolateral membrane can be functionally critical, as for example the exclusive expression of the sodium coupled glucose transporter 1 (SGLT-1) on the apical membrane (see below), easily explains why incretin secreting cells should be ‘blind’ to elevation of vascular glucose concentrations. However, basolateral localization of receptors might be important to shield them from saturating ligand concentrations in the intestinal lumen, a scenario likely for the bile acid sensing receptor, G-protein coupled bile acid receptor 1 (GPBAR1) and the short chain fatty acid sensing receptors FFAR2/3 (see below).

Enteroendocrine cell sensing

For decades it has been known that the presence in the lumen of food or its macronutrient components (carbohydrates, fats and proteins) regulates incretin hormone secretion. Elevation of circulating GLP-1 levels can be detected within 10–15 minutes of eating and persists for several hours, depending on the nutritional composition of the meal. The molecular mechanisms behind this nutrient-dependent secretion have become clear from studies conducted by our group and others, aided by the engineering of transgenic mice wherein the EECs can be identified, isolated and studied. Several studies have shown that EECs directly sense nutrients in the lumen, and this has been attributed to the activity of nutrient transporters, receptors and metabolism. In addition, hormone secretion has been linked to non-nutrient stimulation by factors such as cytokines, bile acids and even gut microbial metabolites.

This review focuses on the nutrient and non-nutrient sensing mechanisms employed by GLP-1 and GIP-secreting L and K cells, respectively. While most of the mechanisms described here hold true for both cell types, some differences exist [Parker et al. 2009].

Nutrient regulation

Carbohydrate sensing

Glucose is a well-recognized stimulus of GLP-1 secretion and has consistently been demonstrated to cause GLP-1 release in in vitro enteroendocrine cell lines, primary murine intestinal culture systems, and in vivo mouse models and humans. The concentrations of glucose in the lumen required to trigger GLP-1 release in vivo are not clear and concentrations ranging from 5 to 1000 mM have been reported by various groups to trigger release from perfused ileum [Shima et al. 1990; Sugiyama et al. 1994; Ritzel et al. 1997]. The pathway underlying glucose-dependent secretion has been well characterized and requires the cotransport of sodium ions by the electrogenic sodium/glucose cotransporter 1 (SGLT1). The influx of Na⁺ ions depolarizes the plasma membrane, opening voltage sensitive calcium channels and exocytosis of GLP-1-containing secretory vesicles [Gribble et al. 2003].

Depletion of sodium chloride in the lumen impaired glucose-mediated GLP-1 secretion in isolated perfused rat small intestine [Kuhre et al. 2015], suggesting that secretion requires Na-coupled uptake by the L cells. Furthermore, support for the involvement of SGLT1-mediated glucose influx come from studies where co-administration of the SGLT inhibitor, phloridzin, with glucose in the upper small intestine blocked glucose-induced GLP-1 and GIP secretion in mice [Moriya et al. 2009] and rats [Kuhre et al. 2015]. Phloridzin also blocked hormone secretion triggered by nonmetabolizable substrates of SGLT: α-methyl-d-glucopyranoside (α-MDG) and 3-O-methyl-d-glucose (3-OMG) [Moriya et al. 2009].

Sglt1 deficient mice displayed impaired GIP and GLP-1 release early after oral glucose administration [Gorboulev et al. 2012], but surprisingly had higher GLP-1 levels at later time points following glucose gavage [Powell et al. 2013a]. One hypothesis to explain these divergent effects on early and late GLP-1 secretion would be that impaired glucose absorption in the small intestine causes increased glucose delivery and microbial fermentation in the distal intestine to produce short chain fatty acids which in turn stimulate secretion from L cells via alternative pathways [Powell et al. 2013a, 2013b]. This exemplifies an important difference between K and L cells – as K cells are mostly found in the duodenum, no late glucose stimulated GIP secretion was seen in Sglt1 knockout mice [Powell et al. 2013b] whereas, consistent with the increased L-cell number in the distal intestine, interventions that shift nutrients down the gut tend to increase GLP-1 secretion.

Whether cellular metabolism contributes to glucose-induced incretin secretion is debatable. In GLUTag cells, a murine enteroendocrine cell line that expresses the proglucagon gene and secretes GLP-1, glucose metabolism raises intracellular adenosine triphosphate (ATP) levels, causing closure of ATP-sensitive potassium (K_ATP) channels, membrane depolarization, intracellular Ca²⁺ elevation and GLP-1 secretion [Parker et al. 2012]. L and K cells express glucokinase and K_ATP channel subunits [Reimann et al. 2008; Parker et al. 2009] and, while this machinery was shown to be required for glucose-stimulated GLP-1 secretion in GLUTag cells, in humans treated with the K_ATP channel inhibitors glibenclamide and repaglinide, GLP-1 secretion after an oral glucose tolerance test was not affected [Stephens et al. 2011]. A recent study investigated the role of K_ATP channels in GIP secretion and concluded that, in healthy mice, the channels are mostly closed and are therefore unlikely to play a major role in glucose-stimulated hormone secretion [Ogata et al. 2014]. Other sugars like fructose, which is not a substrate for SGLT1, trigger GLP-1 release in GLUTag cells by closure of K_ATP channels in response to increased metabolism [Gribble et al. 2003]. Fructose also induced GLP-1, CCK, PYY and neutrotensin (NTS) secretion but not GIP secretion in healthy young humans and GLP-1 but not GIP secretion in mice and rats [Kuhre et al. 2014]. The authors speculated that K cells might be less excitable than L cells and so while, the weak fructose stimulus might activate L cells, it is not sufficient to activate K cells. However, reduced excitability of K cells has not been experimentally demonstrated and a recent report demonstrated fructose-stimulated GIP secretion in mice with streptozotocin-induced diabetes which seemed, however, also not dependent on K_ATP-channel closure [Seino et al. 2015].

Glucose transporter 2 (GLUT2), a facilitative glucose transporter expressed in K and L cells, has been proposed to play a role in incretin release upstream of metabolism. GLUT2 is thought to translocate to the apical membrane in response to elevated luminal glucose [Kellett and Helliwell, 2000], although the importance of this phenomenon is controversial [Gorboulev et al. 2012]. Phloretin, an inhibitor of facilitative glucose but not sodium coupled transport, abolished glucose uptake into L cells in primary murine intestinal cultures, although in contrast to phloridizin it had little effect on glucose stimulated secretion [Parker et al. 2012]. Nonetheless, phloretin impaired GLP-1 secretion in the GLUTag cell line model [Parker et al. 2012] and perfused rat small intestine [Kuhre et al. 2015], and reduced GIP and GLP-1 secretion in isolated rat small intestine loops [Mace et al. 2012]. GLUT2 knockout mice have been reported to have impaired GLP-1, but not GIP responses, possibly due to reduced GLP-1 content in the intestine [Cani et al. 2007a]. However, as a more recent study found no significant differences in GIP and GLP-1 secretion in GLUT2 knockout mice after an oral glucose gavage [Röder et al. 2014], it seems that GLUT2 and facilitative glucose transport in general is of less importance in K and L cells compared with its well established role in the stimulus secretion coupling of pancreatic β cells.

Sweet taste receptors have also been proposed as glucose sensors in incretin secreting cells. On the one hand, expression of members of the T1R sweet taste receptors and the α-subunit of the G-protein gustducin have been reported in the small intestine of mice and the enteroendocrine cell line, STC-1 [Dyer et al. 2005] and colocalization of α-gustducin with PYY and GLP-1 was demonstrated in L cells of the human colon [Rozengurt et al. 2006] and a subset of duodenal L and K cells [Young et al. 2013]. α-Gustducin null mice receiving glucose gavage showed deficient release of GLP-1 [Jang et al. 2007] in comparison with wildtype mice. Similar disruption of glucose-stimulated GLP-1 secretion was observed in T1r3 null mice [Kokrashvili et al. 2009] and a sweet taste receptor antagonist interfered with glucose-stimulated GLP-1 and PYY elevation in human volunteers [Steinert et al. 2011]. On the other hand, ingestion of artificial sweeteners did not trigger GLP-1 in mice [Fujita et al. 2009] or humans [Ma et al. 2009], and their intraluminal administration in the upper small intestine of mice did not trigger hormone release [Moriya et al. 2009].

The finding that SGLT-1 inhibition or knockout completely abolished GIP secretion in vivo [Powell et al. 2013b] argues against an apical location of a glucose sensing (‘taste’) receptor in K cells, as the resulting elevated (unabsorbed) luminal sugar should then result in increased secretion. It could be argued that SGLT-1 might transport glucose across the epithelium where it is then sensed by basolaterally located sweet taste receptor, a mechanism described later for bile acid detection by TGR5 (see below). However, this seems unlikely as incretin secretion from mixed primary epithelial cultures, which give unrestricted access to both sides of an enteroendocrine cell, was also abolished by SGLT-1 knockout/inhibition and was not responsive to saturating concentrations of artificial sweeteners [Parker et al. 2012; Reimann et al. 2008]. It remains, however, possible that sweet taste receptor expression is especially sensitive to culture conditions, although a recent study using a perfused intestinal preparation failed to observe effects of artificial sweeteners [Kuhre et al. 2014]. An alternative explanation for the impaired incretin secretion observed in mice with altered taste receptor pathway activity could be that the well documented taste receptor dependent increase in SGLT-1 expression and glucose absorption [Zietek and Daniel,2015] should be impaired in these mice, possibly reducing SGLT-1 expression in enteroendocrine cells. More work is needed to substantiate a role of sweet taste receptors in incretin hormone secretion.

Protein sensing

Dietary proteins are known to promote satiety and weight loss, and the satiating effects of proteins were found to be associated with the release of the anorectic hormone, PYY, and suppression of the orexigenic hormone, ghrelin. In age-matched normal weight and obese volunteers, a high protein diet caused a greater reduction in hunger and a larger increment in PYY compared with isocaloric high-carbohydrate or high-fat meals [Batterham et al. 2006]. Similar effects were seen in rodents, in which increasing the dietary protein content reduced ad libitum calorie intake and increased circulating PYY levels, and long-term high-protein feeding resulted in significantly less weight gain, less white adipose tissue, and lower plasma leptin levels compared with the corresponding isocaloric normal-protein diets. Additionally, Pyy null mice were found to be resistant to the satiating and weight-reducing effects of high protein diets [Batterham et al. 2006]. Oral whey protein in combination with glucose augmented GLP-1 responses in mice but had no effect on GIP secretion [Gunnarsson et al. 2006]. Interestingly, the same study showed that whereas total GIP secretion was unaffected, the active form of GIP, measured with N-terminally directed antibodies, was increased by whey protein. The authors suggested that ingested protein resulted in slower GIP inactivation, leading to an increase in the active form, but no change in circulating total GIP levels. Several other studies reported differential effects of oral protein ingestion on the two incretins, with greater effects observed on GLP-1 than GIP secretion [Elliott et al. 1993].

In the perfused rat intestine, GLP-1 release was triggered by luminal perfusion of a meat protein hydrolysate known as peptone, and the response magnitude was similar in the upper and lower half of the small intestine [Svendsen et al. 2015]. Meat peptones strongly stimulated GLP-1 secretion from primary murine colonic cultures [Diakogiannaki et al. 2013] and we recently demonstrated GLP-1 release from primary murine duodenal cultures after treatment with meat, vegetable and casein-derived peptones [Pais et al. 2015]. Mechanistic studies performed with meat peptones revealed the involvement of the calcium sensing receptor (CASR) and voltage gated calcium channels (VGCC) in mediating peptone-stimulated GLP-1 release [Pais et al. 2015]. CASR’s primary role as a calcium sensor regulating parathyroid hormone release [Brown and Herbert, 1995] is well recognized and established, but increasing evidence supports the idea that CASR also acts as an amino acid sensor in the gut, mediating macronutrient-dependent secretion of gut hormones such as gastrin [Feng et al. 2010], CCK [Liou et al. 2011; Wang et al. 2011], GIP and GLP-1 [Mace et al. 2012; Diakogiannaki et al. 2013; Pais et al. 2015]. Meat hydrolysate has been linked to the activation of mitogen-activated protein kinases (MAPK) and stimulation of GLP-1 secretion from murine-derived STC-1 and human-derived NCI-H716 cell lines [Reimer, 2006]. In GLUTag cells, albumin egg hydrolysate also increased transcription of the proglucagon gene [Cordier-Bussat et al. 1998]. The extent to which these effects are downstream of CASR activation remains to be established.

Among amino acids, l- glutamine (l-Gln) was consistently shown to trigger GLP-1 secretory responses in human volunteers and patients with type 2 diabetes [Greenfield et al. 2009; Samocha-Bonet et al. 2011] and also in in vitro studies performed in primary mouse cultures and GLUTag cells. The effect of l-Gln has been attributed to its ability both to trigger membrane depolarization via electrogenic Na⁺ dependent amino acid uptake leading to opening of L- and Q-type voltage gated calcium channels, and to elevate cytoplasmic cyclic adenosine monophosphate (cAMP) concentrations [Reimann et al. 2004; Tolhurst et al. 2011]. In primary small intestinal cultures, glutamine-triggered GLP-1 secretion was inhibited by antagonists of CASR [Pais et al. 2015]. Other amino acids, such as glycine and alanine, are strong stimulants of GLP-1 secretion from GLUTag cells through activation of the ionotropic glycine receptor [Gameiro et al. 2005], but this mechanism was not observed in colonic L cells in primary culture. l-Phenylalanine (l-Phe), however, was a highly effective stimulus of GLP-1 release in mouse small intestinal cultures [Pais et al. 2015], but was not an exceptional stimulus in colonic cultures, and was ineffective in GLUTag cells [Reimann et al. 2004; Tolhurst et al. 2011]. The identity of the L-cell sensor underlying l-Phe triggered secretion remains to be established, and although some studies have proposed the involvement of CASR [Mace et al. 2012], we were unable to block l-Phe triggered GLP-1 secretion from primary intestinal cultures using CASR antagonists [Pais et al. 2015]. Jiang and colleagues reported higher meal-dependent GLP-1 levels in mice lacking the neutral amino acid transporter B⁰AT1 than in controls and suggested that reduced neutral amino acid absorption in the null mice causes an overload of amino acids in the lower gut and stimulation of the large reservoir of distally located GLP-1 secreting cells [Jiang et al. 2015]. These results mirror the elevation of glucose-triggered GLP-1 release in mice lacking SGLT-1, which are similarly postulated to reflect increased glucose delivery to the distal gut, as discussed above.

Several other candidate sensory pathways for protein digestion products have been described. Activation of the G-protein coupled receptor GPRC6A in GLUTag cells is triggered by ornithine [Oya et al. 2013]. The proton-coupled electrogenic peptide transporter PEPT1 [Daniel, 2004] was shown to be critical for dipeptide triggered GLP-1 secretion in studies using agonists and antagonists of the transporter and pept1 deficient mice [Diakogiannaki et al. 2013]. Another receptor activated by polypeptides, lysophosphatidic acid receptor 5 (LPAR5, GPR92/93) was shown to mediate CCK [Choi et al. 2007] but not GLP-1 secretion [Diakogiannaki et al. 2013]. Further, the umami taste receptor dimer Tas1R1/Tas1R3 was shown to be involved in amino acid triggered CCK secretion [Daly et al. 2013].

Fat sensing

It was observed in the 1980s that infusion of lipid emulsion into the ileum of humans reduced food intake and promoted satiety, and it was later found that these effects were accompanied by an increase in plasma CCK levels [Welch et al. 1985; Drewe et al. 1992]. Indeed, fats also potently stimulate GLP-1 and PYY release [Feltrin et al. 2004; Little et al. 2005]. The main products of lipid digestion are monoglycerides (MGs) and fatty acids, and inhibition of fat digestion by lipase inhibitors attenuated the rise in plasma levels of CCK, PYY and GLP-1 after a fat-meal [Feinle et al. 2003; Feinle-Bisset et al. 2005; Ellrichmann et al. 2008; Beglinger et al. 2010]. CCK, GLP-1 and PYY release were found to be dependent on fatty acid chain length with only fatty acids greater than C10 being effective in stimulating hormone secretion [Feltrin et al. 2004; Feinle-Bisset et al. 2005; Little et al. 2005].

Several GPCRs have been implicated in the sensing of fatty acids and the concomitant release of gut hormones by EECs. Those identified until now include GPR40 (FFAR1), GPR120 (FFAR4) and GPR119. GPR40 is activated by medium to long chain free fatty acids (FFAs) [Briscoe et al. 2003] and found to be present in GIP and GLP-1 secreting cells in mice. Following an oral high-fat diet (HFD), Gpr40 deficient mice exhibited diminished secretion of GLP-1 and GIP with concomitant reduction in insulin levels [Edfalk et al. 2008]. GPR120 is also activated by unsaturated long chain FFAs and was found co-localized with GLP-1 in human colon. In HEK cells stably expressing GPR120, long chain FFAs evoked an increase in intracellular calcium [Ca²⁺]_i in a dose-dependent fashion and, in STC-1 cells, α-linoleic acid strongly stimulated GLP-1 release [Hirasawa et al. 2005]. The increase in [Ca²⁺]_i in STC-1 cells induced by α-linolenic acid was eliminated by siRNA specific for mouse Gpr120 confirming the role of this receptor in long chain FFA-mediated GLP-1 release [Hirasawa et al. 2005].

A recent report emphasized the greater importance of GPR120 for GIP compared with GLP-1 secretion [Iwasaki et al. 2015] and the same group reported specific enrichment of the fatty acid binding protein FABP5 in K cells, suggesting a role in fatty acid detection [Shibue et al. 2015]. Ligands of GPR119 include the fatty acid derivatives oleoylethanolamide (OEA) [Overton et al. 2006] and lysophophatidylcholine (LPC) [Soga et al. 2005], as well as monoacylglycerides [Hansen et al. 2011; Mandøe et al. 2015; Lauffer et al. 2009]. OEA administration in rats reduced food intake and weight gain [Rodríguez De Fonseca et al. 2001] and stimulated GLP-1 secretion from perfused rat ileum [Lauffer et al. 2009], GLUTag cells [Lauffer et al. 2009] and primary cultured murine L cells [Moss et al. 2015]. GPR119 is Gs-coupled and increases intracellular cAMP levels upon activation. GLUTag cells transfected with specific Gpr119 siRNA failed to increase cAMP levels in response to OEA [Lauffer et al. 2009]. Gpr119 knockout mice showed diminished nutrient-stimulated GLP-1 secretion [Lan et al. 2009]. Also, oral administration of the GPR119 agonist AR231453 increased plasma concentrations of GLP-1 and GIP in wildtype but not Gpr119 knockout animals [Chu et al. 2008]. 2-Oleoyl glycerol, a product of triglyceride digestion, also acts on GPR119 and its administration to humans significantly increased plasma GLP-1 and GIP levels [Hansen et al. 2011; Mandøe et al. 2015]. Pharmacological GPR119 agonists developed for human studies and tested in patients with type 2 diabetes have not, however, shown significant metabolic benefits [Katz et al. 2012]. The reason behind this is uncertain and the therapeutic potential of GPR119 is still under investigation.

A range of non-GPCR machinery has also been implicated in free-fatty-acid-triggered incretin secretion. Protein kinase Cζ was implicated in oleic acid triggered GLP-1 secretion in GLUTag cells [Iakoubov et al. 2007] and rodents [Iakoubov et al. 2011], and fatty acid transport protein was found to contribute to oleic acid induced GLP-1 secretion in vitro and in vivo [Poreba et al. 2012]. Triglycerides (TGs) are resynthesized in the enterocytes of small intestine from absorbed fatty acids and MGs by the sequential action of monoacylglycerol acyltransferase (MGAT) and diacylglycerol acyltransferase (DGAT). The newly formed TGs are then incorporated into chylomicrons (CM) by microsomal triglyceride transfer protein (MTP). MGAT2 and DGAT1 deficient mice displayed impaired GIP secretion after oral triglyceride gavage, but increased GLP-1 and PYY levels which remained elevated even 2 hours after gavage in comparison with wildtype controls [Okawa et al. 2009]. In addition, administration of a pharmacological inhibitor of DGAT1 prior to oil gavage reduced postprandial TG levels and led to prolonged elevated levels of GLP-1 in mice [Ables et al. 2012] Furthermore, an inhibitor of intestinal specific MTP was shown to elevate GLP-1 and PYY in rats fed a high fat diet [Hata et al. 2011]. CM formation and secretion were shown to be critical for GLP-1 [Lu et al. 2012] and GIP [Shimotoyodome et al. 2009] release, as revealed in rodent studies using surfactants that inhibited CM formation. Recently, Wang and colleagues showed that Apolipoprotein A-IV (ApoA-IV) knockout mice have increased lymphatic and plasma GLP-1 levels 30 minutes after intraduodenal administration of Ensure, a commercially available liquid meal, in comparison with wildtype controls. Interestingly, in another experiment conducted by the same group, acute administration of ApoA-IV prior to intraduodenal Ensure infusion did not alter GLP-1 levels in wildtype mice or the ApoA-IV-/- mice. They attributed the increased GLP-1 levels in the ApoA-IV knockout mice to increased expression of Gpr119 in the ileum [Wang et al. 2015].

Whereas fats acutely stimulate GLP-1 secretion as evidenced from above, chronic high-fat feeding in mice reduced numbers of GLP-1 and GIP positive cells and significantly decreased the expression of enteroendocrine hormones, nutrient sensing machinery and enteroendocrine-specific transcription factors [Richards et al. 2015]. Hayashi and colleagues demonstrated that endoplasmic reticulum stress, induced with excessive doses of palmitate, increased the expression of endoplasmic reticulum stress markers, Chop and BiP, reduced protein levels and function of prohormone convertase 1/3, and decreased GLP-1 secretion from GLUTag cells [Hayashi et al. 2014]. Aranias and colleagues, however, observed an increase in L-cell density in the jejunum of morbidly obese subjects and elevated GLP-1 secretion in HFD fed mice after an oral glucose challenge. They hypothesized that increased GLP-1 secretion would favour insulin secretion and counter balance diet-induced insulin resistance [Aranias et al. 2015]. The influence of diet on L cells and gut hormone secretion is clearly not completely understood, and more work needs to be done in this field.

Non nutrient regulation

Gut microbiota

Gut bacteria have a profound impact on the host and influence, in many ways, host physiology and metabolism [Tremaroli and Bäckhed, 2012]. EECs lie in close proximity to microbes and/or their metabolic products in the gut and the resulting crosstalk regulates hormone secretion. Bacterial fermentation of undigested prebiotics and complex carbohydrates produces short chain fatty acids (SCFAs) such as acetate, butyrate and propionate. SCFAs induced GLP-1 secretion from primary colonic cultures and this was shown to be mediated by GPR43 (FFAR2) and GPR41 (FFAR3). GPR43 is Gi- and Gq-coupled and triggered an increase in [Ca²⁺]_i in colonic L cells following activation by SCFAs. Mice deficient in Ffar2 and Ffar3 displayed reduced SCFA-triggered GLP-1 secretion [Tolhurst et al. 2012]. Additionally, butyrate, acetate and propionate administration protected mice from developing diet-induced obesity and insulin resistance and stimulated gut hormone release [Lin et al. 2012]. A recent report, however, observed improved glucose tolerance in Ffar2/3 double knockout mice, which was explained by the direct inhibitory effect of these receptors in pancreatic β cells, and found no significant effect of Villin-Cre mediated conditional knockout of these receptors in the intestine [Tang et al. 2015]. In contrast, impaired glucose tolerance of Ffar3-/- mice was recapitulated, but a selective FFAR3 agonist also impaired glucose tolerance, presumably by direct action in the islet [Forbes et al. 2015], and it was concluded that the observed anorectic effect of this compound was mostly mediated through elevation of PYY.

It is now well established that obesity and type 2 diabetes affect gut microbiome diversity and composition [Ley et al. 2005; Qin et al. 2012]. Prebiotic feeding in ob/ob mice increased L-cell number, proglucagon mRNA levels and portal plasma GLP-1 levels, and this was mediated by the increased production of acetate and propionate [Everard et al. 2011]. Characterization of the microbiome revealed that the abundance of Akkermansia muciniphila was reduced in obese and diabetic mice [Everard et al. 2013] and feeding of this bacteria was positively associated with increased L-cell number and secretion [Everard et al. 2011]. Several other studies showed beneficial effects of prebiotics treatment on GLP-1, GLP-2 and PYY secretion in rodents and humans [Cani et al. 2007b, 2009a, 2009b].

Although a large body of evidence suggests that SCFAs produced by gut bacterial fermentation stimulate GLP-1 secretion, Wichmann and colleagues demonstrated that in germ-free (GF) mice, having severely reduced SCFAs, basal plasma GLP-1 levels were 3-fold higher than conventional raised mice (CONV-R). They attributed this observation to increased proglucagon (Gcg) expression in the caecum and colon. In addition, GF mice had more GLP-1 positive cells in the caecum and colon than the CONV-R mice. Colonization of GF mice with SCFA-producing bacteria increased acetate and propionate and correspondingly decreased Gcg expression and GLP-1 positive cells. Conversely, depletion of microbiota in CONV-R with antibiotics treatment over 3 days decreased SCFA levels and increased Gcg expression [Wichmann et al. 2013]. Other bacterial metabolites known to affect GLP-1 secretion include indole. Indole is produced from tryptophan by gut bacteria and, while acutely stimulated GLP-1 secretion by inhibiting voltage gated K⁺ channels, after prolonged exposure it slowed ATP production by blocking NADH dehydrogenase leading to a delayed reduction of GLP-1 secretion [Chimerel et al. 2014].

Bile acids

Bile acids play an important role in the digestion and absorption of dietary fats by facilitating the formation of micelles. Primary as well as secondary bile acids (formed as a result of bacterial modification) act as signalling molecules and strongly stimulate GLP-1 and PYY secretion [Ullmer et al. 2013]. They are known to act at two specific receptors: the nuclear farnesoid X receptor (FXR); and the transmembrane G-protein coupled receptor TGR5 (GPBAR1). Bile acid triggered GLP-1 secretion was found to be TGR5 dependent in STC-1 cells [Katsuma et al. 2005], GLUTag cells and primary murine intestinal cultures [Parker et al. 2012] and in vivo [Thomas et al. 2009]. Downstream signalling involved elevation of cAMP and [Ca²⁺]_i [Parker et al. 2012]. Deoxycholic acid infusion in the colon increased enteroglucagon (GLP-1) and PYY secretion in humans undergoing colonoscopy [Adrian et al. 1993] and rectal taurocholate infusion reduced food intake, lowered glucose and increased GLP-1, PYY and insulin in obese type-2 diabetic volunteers [Adrian et al. 2012].

However, there has been debate about whether bile acids target their receptors on L cells from the apical or basolateral compartment. Brighton and colleagues measured GLP-1 release from murine ileum mounted in small volume Ussing chambers; they showed that basolateral taurodeoxycholate (TDCA) or a TGR5 agonist stimulated GLP-1 secretion but that the effects of apically-applied TDCA were abolished by inhibition of bile acid uptake across the epithelium. Preferential stimulation of GLP-1 release by basolateral versus apical bile acids was confirmed in the perfused rat gut, leading to the conclusion that bile acid triggered GLP-1 secretion is predominantly mediated by basolaterally located TGR5 on the L cell [Brighton et al. 2015]. These findings dampen the hope of developing TGR5 agonists as GLP-1 secretagogues, as TGR5 ligands entering the systemic circulation have been found to trigger gall bladder overfilling [Li et al. 2011]. It is interesting to note that a recent report suggests that FFAR1 (GPR40) is also located on the basolateral rather than the apical membrane of enteroendocrine cells [Christensen et al. 2015] and one might speculate that this will also extend to the related short-chain fatty acid receptors FFAR2 (GPR43) and FFAR3 (GPR41), as physiological observed luminal concentrations of short chain fatty acids are well above their reported potencies for these receptors.

A number of studies have shown that some of the beneficial effects of gastric bypass surgery on glucose metabolism in diabetic subjects can be attributed to increased bile acid delivery to the distal gut and enhanced release of gut hormones such as GLP-1 and PYY [Patti et al. 2009]. Bile acid sequestrants (BAS) are anion exchange resins that trap bile acids in the lumen and have cholesterol lowering and antidiabetic properties [Prawitt et al. 2014]. Additionally, they were shown to enhance GLP-1 secretion in diet-induced obese mice and increased proglucagon gene expression downstream of TGR5 [Harach et al. 2012]. Recently, Fxr was shown to be expressed in mouse ileal and colonic L cells and colocalized with GLP-1 in the human jejunum. Surprisingly, treatment with the FXR agonist GW406 decreased gcg expression and FXR activation was demonstrated to inhibit glycolysis, leading to reduced intracellular ATP levels and impaired glucose-dependent GLP-1 secretion [Trabelsi et al. 2015].

Hormones/cytokines

An emerging body of evidence shows that there is a crosstalk between the enteroendocrine system and other organs in the body and that gut hormone secretion is influenced by factors such as hormones and cytokines which act on the EECs to influence secretion either positively or negatively. Somatostatin (SST) is produced from D cells in the oxyntic and pyloric regions of the stomach and its secretion is stimulated by GIP, GLP-1 and CCK but not gastrin [Adriaenssens et al. 2015]. SST in turn exerts a suppressive tone over neighbouring EECs in the stomach as well as in the gut. Acting at least in part via the Gαi-coupled SSTR5 receptor, SST inhibited GIP and GLP-1 secretion in primary murine intestinal cultures and acts by suppressing cAMP levels in EECs [Moss et al. 2012].

Intestinal endocannabinoids exert orexigenic effects by signalling through vagal afferent neurons to satiety centres in the central nervous system [Izzo and Sharkey, 2010]. Levels of the endocannabinoid anandamide have been reported to increase in the small intestine of rats upon food deprivation and levels returned to baseline upon refeeding [Gómez et al. 2002]. The cannabinoid receptor CB1 was found to be highly expressed in GIP secreting K cells and, consistent with this finding, GIP but not GLP-1 secretion was reduced by the CB1 synthetic agonist methanandamide (mAEA) [Moss et al. 2012].

A number of studies have identified a stimulatory effect of GIP on GLP-1 secretion in rodents. Intravenous infusion of GIP, for example, elevated plasma levels of proglucagon-derived peptide (GLP-1) in rats [Roberge et al. 1993] and GIP treatment triggered GLP-1 release from primary canine L cells [Damholt et al. 1998]. It is not, however, currently believed that this plays an important role in humans, as even pharmacologically raised GIP levels had little effect on plasma GLP-1 [Nauck et al. 1993]. Ghrelin, an orexigenic hormone released from the stomach, enhanced glucose-induced GLP-1 release in vivo in mice and in vitro in rodent and human L cells through an extracellular-signal-regulated kinase ERK 1/2-dependent pathway [Gagnon et al. 2015a].

Obesity is a pro-inflammatory condition associated with elevated levels of circulating pro-inflammatory cytokines which are released from the adipose tissue and infiltrating immune cells. The endocrine and secretory behaviour of adipose tissue was revealed with the discovery that adipocytes synthesize and secrete the cytokine tumour necrosis factor α (TNFα) [Hotamisligil et al. 1993] and the hormone leptin [Zhang et al. 1994]. Since then, several adipose-derived proteins have been discovered that have profound physiological effects; they have collectively been given the term adipokines. Leptin has well established effects on food intake and elevated circulating levels in obesity are proportional to fat mass [Considine et al. 1996]. The long form of the leptin receptor, Ob-Rb, was found on human and rodent L cells and leptin was observed to stimulate GLP-1 secretion in vitro and in vivo in rats and mice [Anini and Brubaker, 2003]. Among other adipokines, interleukin (IL) 6 is regarded as a key player contributing to obesity linked insulin resistance [Senn et al. 2002; Eder et al. 2009]. IL-6 is also secreted from exercising skeletal muscle and was shown to stimulate GLP-1 secretion not only from intestinal L cells but also pancreatic α cells [Ellingsgaard et al. 2011]. Exercise has been shown to be associated with elevated levels of GLP-1, PYY and pancreatic polypeptide [Martins et al. 2007] and IL-6 is a candidate mediator of this effect.

Recently, TNFα was demonstrated to affect GLP-1 secretion. Whereas acute application of TNFα increased GLP-1 secretion from a human GLP-1 secreting cell line, TNFα pretreatment of both human and murine cell lines reduced gcg expression and impaired GIP-triggered GLP-1 secretion. Co-administration of a nuclear factor-κB (NFκβ) inhibitor abolished the inhibitory effect of TNFα pretreatment on GLP-1 responses and treatment with the TNFα inhibitor etanercept improved GLP-1 secretion in HFD-induced obese mice [Gagnon et al. 2015b]. Another adipokines, RANTES (CCL5) which is upregulated in obesity [Wu et al. 2007; Keophiphath et al. 2010] was shown to reduce glucose-stimulated GLP-1 release in vitro in NCI-H716 cells and plasma levels of GLP-1 and GLP-2 after an oral glucose load in mice [Pais et al. 2014].

Progesterone increased plasma levels of GIP and GLP-1 in mice, and was shown to be mediated by the G-protein coupled receptors PAQR5 and PAQR7 by receptor knockdown studies conducted in GLUTag cells [Flock et al. 2013]. Insulin receptor expression was also found on murine and human L cells, and insulin was found to stimulate GLP-1 secretion in vitro from murine L cells by activating the phosphatidylinositol 3 kinase-Akt and MAPK kinase (MEK)-ERK1/2 pathways. Pretreatment with high concentrations of insulin for 24 hours caused insulin resistance and reduced basal and insulin induced GLP-1 secretion [Lim et al. 2009]. Nesfatin-1, an anorexigenic peptide found in the hypothalamus [Oh-I et al. 2006], colocalized with GIP and GLP-1 positive cells in mouse small intestine and stimulated GLP-1 secretion from STC-1 cells [Ramesh et al. 2015].

Neural regulation

The enteric nervous system is sometimes referred to as a second brain because of its degree of autonomous function. It exerts several actions in the gastrointestinal tract such as control of gut motility, fluid exchange between the gut and the lumen, regulation of gastric and pancreatic secretion and gut defence, and it interacts with EECs in the gut epithelium to regulate hormone release. Previous immunohistochemistry studies performed on rat and human tissue detected GLP-1-positive cells predominantly in the lower small intestine and colon [Eissele et al. 1992] and led to the proposal that the initial rapid rise in GLP-1 following a meal was indirectly mediated by neural/endocrine pathways arising from nutrient detection in the proximal gut, rather than by direct nutrient stimulation of L cells.

More recent studies, however, have concluded that enough GLP-1 cells are present in the proximal small intestine of a number of species to account for the rapid postprandial rise of plasma GLP-1 concentrations [Theodorakis et al. 2006; Habib et al. 2012] and that they directly respond to nutrient stimulation [Pais et al. 2015]. L cells lie in close proximity to neurons and microvasculature in the gut and studies performed on isolated vascularly perfused rat colon and ileum demonstrated stimulatory roles for neurotransmitters, calcitonin-gene related peptide (CGRP) and gastrin releasing peptide (GRP) on GLP-1 secretion [Plaisancie et al. 1994; Dumoulin et al. 1995]. GRP is released from GRPergic neurons and its role in GLP-1 secretion was confirmed in studies using GRP agonists [Roberge and Brubaker, 1996] and GRP deficient mice [Persson et al. 2000]. Cholinergic regulation of GLP-1 secretion was shown in the murine STC-1 cell line, mediated by the muscarinic receptor M3 [Abello et al. 1994]. Other studies have shown roles for M1 and M2 in stimulating GLP-1 release from L cells in vitro, and M1 in mediating GLP-1 release from rat in vivo [Anini et al. 2002]. Pituitary adenylate cyclase-activating polypeptide (PACAP) increased GLP-1 secretion from GLUTag cells by increasing intracellular cAMP levels [Simpson et al. 2007]. γ-Aminobutyric acid (GABA) receptors were detected on the STC-1 and GLUTag cell lines: GABA triggered CCK and GLP-1 secretion mediated by the GABA_A receptor subtype in STC-1 cells [Glassmeier et al. 1998] and by both GABA_A and GABA_C receptors in GLUTag cells [Gameiro et al. 2005].

Cell number and sensitivity

An alternative or supplementary approach to the acute stimulation of incretin secretion would be the promotion of increased incretin expression in the intestine. FXR-mediated effects of bile acids on pro-glucagon expression have already been mentioned. Alternatively, SCFA has been shown to increase the number of gcg-expressing cells in organotypic intestinal organoid cultures [Petersen et al. 2014], which might underlie the reported increase in L-cell density in response to prebiotic fibre supplementation in ob/ob mice [Everard et al. 2011] and the reduced gcg mRNA content of Ffar2-/- derived intestines [Tolhurst et al. 2012]. More recently Petersen and colleagues demonstrated that inhibition of NOTCH signalling pathways with the γ-secretase inhibitor dibenzazepine (DBZ), increased L-cell numbers in in vitro models of intestinal organoids cultures from mouse and human and augmented glucose-stimulated GLP-1 secretion. DBZ also increased L-cell numbers in the intestines of HFD fed mice and type 2 diabetes mouse models and improved glucose-stimulated insulin release and glucose tolerance [Petersen et al. 2015].

Another potentially interesting aspect of incretin secretion investigated only recently is its link to circadian rhythm and the possibility that it plays a role in entraining these rhythms to food intake. Gil-Lozano and colleagues demonstrated circadian rhythmicity in the secretion of GLP-1 in response to different secretagogues in vitro and well as in vivo in rats fasted for 4 hours before being administered a glucose load. They also showed rhythmic expression of clock genes in GLUTag cells and demonstrated that knockdown of two rhythmically expressed target genes (thyrotroph embryonic factor and protein tyrosine phosphatase 4a1) resulted in altered GLP-1 release [Gil-Lozano et al. 2014].

Physiological relevance

Therapies targeting the incretin axis have proved highly effective in treating type 2 diabetes and provide the additional benefits of weight loss and cardioprotection [Drucker and Nauck, 2006; Gejl et al. 2012]. Roux-en-Y gastric bypass (RYGB) is currently the most widely used bariatric procedure for treating severe obesity. Anastomosis of the jejunum to a small gastric pouch causes rapid postprandial delivery of nutrients to the distal gut with its large population of L cells. The consequent exaggerated release of GLP-1 is believed to underlie some of the remarkable improvements in glucose tolerance found in people undergoing surgery, in many cases resulting in remission of type 2 diabetes [Korner et al. 2007; Dirksen et al. 2010; Falkén et al. 2011; Jiménez et al. 2012]. High accompanying PYY levels after surgery likely contribute to the observed reductions in appetite [Korner et al. 2005; Morínigo et al. 2008]. Interestingly, nutrient malabsorption in mice deficient in Sglt1 [Powell et al. 2013a], B⁰at1 [Jiang et al. 2015] and Dgat1 [Okawa et al. 2009] seems to present a scenario that mimics some aspects of bariatric surgery. In each case elevated postprandial GLP-1 levels have been reported, probably reflecting the increased delivery of nonabsorbed nutrients to the hindgut (Figure 1), although not reproducing the rapid peaks found after surgery.

Conclusion

The enteroendocrine system secretes a myriad of hormones that are pivotal to glucose homeostasis and gut function. The incretins, in particular, have garnered a great deal of interest owing to their antidiabetic and satiating effects. GLP-1 based drugs (GLP-1 mimetics and inhibitors of GLP-1 degradation) and bariatric surgery have shown immense beneficial effects on glycaemia and weight loss in humans. Fuelled by our understanding of the intricate pathways by which EEC cells sense nutrients and other factors in their milieu, strategies to harness the tremendous potential of the enteroendocrine system by targeting endogenous gut hormone production offer great hope for the future treatment of diabetes and obesity.