Ghrelin Is a Novel Regulator of GLP-1 Secretion

GLP-1 is a gastrointestinal hormone secreted from the enteroendocrine L-cell in response to nutrient ingestion. Once released into the circulation, GLP-1 elicits a potentiation of glucose-stimulated insulin secretion from the β-cells within the pancreatic islets, known as the incretin effect (1,2). The actions of incretin hormones, including GLP-1 as well as glucose-dependent insulinotropic peptide (GIP), on insulin secretion result in improved glucose clearance, and as such, incretin-based approaches are an important therapeutic tool in the treatment of patients with type 2 diabetes mellitus (T2DM). Current incretin therapies include long-acting GLP-1 receptor (GLP-1R) agonists and also inhibitors of incretin hormone degradation; however, GLP-1 secretagogues represent a potential third approach to enhancing incretin action in T2DM (1–3).

GLP-1 secretion is regulated by a combination of nutrient-, neural-, and hormonal-activated pathways. Although nutrients have been shown to directly enhance GLP-1 release from the intestinal L-cell (4,5), the enteric and parasympathetic nervous systems (6,7) and other endocrine hormones are likely more critical mediators of the very rapid effect of meal ingestion on circulating levels of GLP-1. Several examples of the hormonal regulation of GLP-1 have been demonstrated. GIP enhances GLP-1 secretion from the rodent L-cell in vivo (7) and in vitro (8,9), whereas cholecystokinin appears to be more important in humans (10). The satiety factor, leptin, also stimulates GLP-1 release by rodent and human L-cells (11), as does the metabolic hormone, insulin (12). Interestingly, several studies have indicated that L-cell responses can also be “primed,” such that preexposure to one secretagogue modulates the subsequent response to a heterologous effector. Thus, pretreatment of the murine intestinal GLUTag L-cell line with the nutrient oleic acid enhances the subsequent GLP-1 secretory response to GIP (13). Conversely, prolonged exposure to leptin and insulin suppresses the GLP-1 secretory response to oral glucose administration in vivo in rodents (11,12).

One hormone that has not been investigated with respect to effects on the L-cell is the stomach-derived, acylated peptide, ghrelin. Plasma ghrelin levels, like those of GLP-1, fluctuate in response to nutrient ingestion. However, in contrast to GLP-1, ghrelin levels are highest immediately before habitual mealtimes (14). Although ghrelin is best known for its appetite-inducing effects (15), several groups have also demonstrated a role for ghrelin in enhancing disposition of ingested calories (16,17). Hence, ghrelin promotes the synthesis of fatty acids and triglycerides in the liver (18) as well as their storage within human adipocytes (19). These actions suggest a role for ghrelin in preparing the body for an incoming meal. Because GLP-1 enhances the insulin response necessary for the disposition of these ingested nutrients, we hypothesized that the L-cell response to meal ingestion may be “primed” by the high levels of ghrelin that occur immediately before a meal. Consistent with this notion, premeal ghrelin levels (20,21) and the GLP-1 response to a meal (22) are concurrently reduced in obese individuals. Thus, ghrelin treatment may play a role in enhancing GLP-1 secretion in this condition.

The current study examined the effects of acylated ghrelin pretreatment on GLP-1 secretion and associated glucose tolerance in response to oral glucose administration in normal mice, mice with blockade of the ghrelin receptor and genetic disruption of GLP-1R signaling, and mice with diet-induced obesity. To determine the mechanism of action of ghrelin on the L-cell, murine and human L-cell models were used to also examine its effects in vitro on GLP-1 release and downstream signaling. Collectively, the results indicate that preexposure to ghrelin enhances GLP-1 release in response to a subsequent oral glucose load and that this effect is mediated through a ghrelin receptor– and extracellular signal–related kinase (ERK)1/2–dependent pathway.

Male C57BL/6 mice (4 or 7 weeks old) were purchased from Charles River Laboratories (St. Constant, Quebec, Canada). Male GLP-1R wild-type (WT) and knockout (Glp1r^−/−) mice on a C57BL/6 background (10–12 weeks old) have been described previously (23). The University of Toronto and Mount Sinai Hospital Animal Care Committees approved all animal work.

Mice were fasted overnight (16 h) before receiving an oral gavage of 5 g/kg d-glucose in water (preliminary studies indicated that a 16-h fast enabled detection of a greater GLP-1 response than found after a 6-h fast; data not shown). Treatments or vehicle (sterile 0.9% NaCl) were administered intraperitoneally 15 min before the glucose gavage. All peptides were obtained from Bachem Americas Inc. (Torrance, CA). Blood samples (1 drop) for glucose measurements were obtained at 0, 10, 60, 90, and 120 min after oral glucose loading and were analyzed using a OneTouch Ultra 2 glucose meter (LifeScan Canada, Ltd., Burnaby, British Columbia, Canada). Additional blood samples (100 μL) were collected in EDTA-coated capillary tubes (Sarstedt, Montreal, Quebec, Canada) at 0, 10, and 60 min after oral glucose administration and were immediately spiked with 10 μL of a concentrated protease inhibitor solution containing aprotinin (5,000 KIU), diprotin A (0.1 mmol/L), and EDTA (1.2 mg/mL) and kept on ice until centrifugation at 4,000g for 5 min and collection of plasma. Total GLP-1 and insulin concentrations were determined using kits from Meso Scale Discovery (Rockville, MD). Plasma samples for fasting ghrelin were collected without the addition of the protease inhibitor solution and were analyzed using an unacylated ghrelin ELISA kit (Cayman Chemical, Ann Arbor, MI).

To induce weight gain and glucose intolerance, 4-week-old mice were fed a high-fat diet (HFD; 60% calories from fat; Research Diets, New Brunswick, NJ) for 10 weeks. Because preliminary tests showed that these mice were unable to rapidly clear a glucose load of 5 g/kg (i.e., glucose levels were higher than the limit of glucometer detection of 33 mmol/L during an oral glucose tolerance test [OGTT]; data not shown), a lower dose of 2 g/kg was used for these mice.

Mouse GLUTag cells were originally derived by single-cell cloning of a proglucagon-SV40 large T antigen–induced tumor (24). Cells were maintained in high-glucose (25 mmol/L) DMEM supplemented with 10% FBS and were passaged after trypsinization. Human NCI-H716 cells, derived from a cecal tumor, were obtained from American Type Culture Collection (Manassas, VA) and maintained in suspension culture in RPMI medium with 11 mmol/L glucose and supplemented with 10% FBS and penicillin/streptomycin. Both cells lines have been extensively validated as models for GLP-1 secretion by the intestinal L-cell (4,5,11,12,25).

GLUTag cells were split into 24-well plates at a density of 200,000 cells per well in 1 mL culture medium. NCI-H716 cells plated at the same cell density required that cell culture plates be precoated with 0.5 mg/mL Matrigel basement membrane solution in Hank’s Balanced Salt Solution (HBSS) for 1 h to allow attachment. Cells were rinsed with HBSS 48 h after plating and treated in the corresponding culture medium with reduced serum (0.5% FBS). Some GLUTag cells were treated in media containing 0–5 mmol/L glucose. After a 2-h incubation, media were collected and acidified (final concentration 0.1% trifluoroacetic acid), and cells were collected in acid lysis buffer (1% trifluoroacetic acid, 1 N HCl, 5% formic acid, and 0.1 mmol/L NaCl). Peptides were collected by reversed-phase adsorption to C18 silica (Sep-Pak, Waters, Ontario, Canada), and samples were dried in vacuo before radioimmunoassay using a total GLP-1 kit (Millipore Canada Ltd.).

GLUTag cells were split into 12-well plates at 500,000 cells per well. After 2 days in culture, cells were rinsed in HBSS and incubated in HBSS containing 5.5 mmol/L glucose for 1 h under standard culture conditions. Experiments where inhibitors were used received an additional 30-min preincubation of vehicle, U-0126 (10 µmol/L), or PD98059 (10 µmol/L). Cells were then incubated for 10 min in HBSS containing various treatments. After this, cells were lysed and collected in radioimmunoprecipitation assay cell lysis buffer supplemented with PhosphoStop and Complete Mini (EDTA-free) enzyme inhibitors (Roche Diagnostics, Indianapolis, IN). Proteins were quantified using the Bradford method, and protein (50 µg) was loaded onto 8% acrylamide SDS-PAGE Tris-glycine gels. Proteins were transferred to polyvinylidene fluoride membranes, blocked with 5% skim milk in Tris-buffered saline, and then incubated with primary antibodies (phospho and total ERK1/2; Cell Signaling Technologies, Danvers, MA) overnight in 5% BSA. Proteins were visualized using appropriate horseradish peroxidase–conjugated anti-rabbit secondary antibodies (Cell Signaling Technologies) and were imaged and quantitated using Luminata Forte chemiluminescence substrate (Millipore) with a Kodak digital imaging system.

GLUTag and NCI-H716 cells were seeded into 10-cm plates at 2,000,000 cells per plate. After 2 days in culture, RNA was harvested using the Qiagen RNeasy total RNA extraction kit with QIAshredder (Qiagen, Toronto, Ontario, Canada). Total RNA (2 µg) was converted to cDNA using 5× All-in-One RT MasterMix (Applied Biological Materials, Richmond, British Columbia, Canada). Primers for the ghrelin receptor were ordered from IDT (Coralville, IA): human forward: GTGAAAATGCTGGCTGTAGTGG, and reverse: TGATGGCAGCACTGAGGTAG; and mouse forward: GTCTCATCGGGAGGAAGCTG and reverse: AGCGCTGAGGTAGAAGAGGA. Mouse brain cDNA was used as a positive control. PCR products were run and visualized on a 1% agarose gel.

All data are expressed as the mean ± SE. Studies comparing two groups were analyzed by Student t test. All animal experiments where comparisons were made between two independent variables (i.e., time and drug treatment) were analyzed by two-way ANOVA, followed by Bonferroni post hoc tests at individual time points where applicable. Vertical parentheses with # symbols to the right of the graphs indicate significant interaction, and * over individual time points indicate significance at that time point for the post hoc test. Area under the curve (AUC) analysis was calculated using the trapezoidal method and was assessed by Student t test or one-way ANOVA, followed by Bonferroni post hoc tests where applicable. The relative insulin response was calculated as ∆Insulin^{AUC(0–60min)}/∆glucose^{AUC(0–60min)}. Cell culture data were expressed as percent GLP-1 secretion [media content/(media + cell content)] normalized to vehicle control. Cell culture experiments in which multiple doses of ghrelin were tested were analyzed by one-way ANOVA, followed by a Bonferroni post hoc test. Group and post hoc effects with a P value of <0.05 were considered significant.

To determine the effect of ghrelin pretreatment on GLP-1 secretion, mice were injected with acylated ghrelin (200 nmol/kg) or vehicle at t = −15 min, and blood was sampled before (t = 0 min) and during an OGTT. Ghrelin administration had no effect on baseline GLP-1 levels (Fig. 1A); however, 10 min after oral glucose loading, ghrelin-pretreated mice had significantly higher levels of circulating GLP-1 than mice given vehicle alone (P < 0.001, Fig. 1A). Although insulin levels were not significantly increased by ghrelin pretreatment (Fig. 1B), the relative insulin response was elevated in response to ghrelin (P < 0.05) in association with significantly improved glucose tolerance (P < 0.05 at t = 60 min and P < 0.01 at t = 90 min, Fig. 1C and D).

To determine the importance of endogenous ghrelin in GLP-1 secretion and glucose homeostasis, mice were pretreated with the ghrelin receptor antagonist, D-Lys GHRP-6 (10 µmol/kg), or vehicle alone, 15 min before an OGTT. Blocking the ghrelin receptor caused a significant reduction in glucose-stimulated levels of circulating GLP-1 (Fig. 2A). These mice also had a significant decrease in the insulin response to oral glucose compared with saline controls 10 min after oral glucose administration (P < 0.01 at t = 10 min and P < 0.05 for the effects of ghrelin compared with saline over the entire time course; Fig. 2B), as well as impaired clearance of the oral glucose load (P < 0.001 at t = 60 min, P < 0.01 at t = 90 min, and P < 0.01 overall by two-way ANOVA; Fig. 2C).

To establish that the effects of ghrelin on oral glucose handling were mediated through the GLP-1R pathway, mice were pretreated with ghrelin (200 nmol/kg) and the GLP-1R antagonist exendin-4^9–39 (5 nmol/kg) or with vehicle and exendin-4^9–39 alone. The stimulatory effect of ghrelin pretreatment on GLP-1 secretion during an OGTT was maintained in the presence of GLP-1R antagonism (P < 0.01 vs. vehicle alone; Fig. 3A). However, the ability of the ghrelin injection to subsequently enhance insulin secretion and glucose tolerance was completely lost in mice cotreated with exendin-4^9–39 (Fig. 3B and C).

To further validate the importance of the GLP-1 pathway in mediating the effects of ghrelin on glucose tolerance, GLP-1 levels were assessed in Glp1r^−/− mice. Similar to the responses observed in C57BL/6 mice, WT mice demonstrated a significant increase in the GLP-1 response (P < 0.01 at t = 10 min and P < 0.05 overall by two-way ANOVA; Fig. 4A) and an improvement in glucose tolerance (P < 0.05; Fig. 4B) when injected 15 min before an OGTT with ghrelin (200 nmol/kg) compared with vehicle alone. Interestingly, fasting levels of GLP-1 were extremely elevated in Glp1r^−/− mice (160 pg/mL, data not shown), consistent with previous reports that demonstrated elevated levels of plasma exendin-4 after exogenous administration in studies of Glp1r^−/− mice (26). However, the ghrelin-induced improvement in glucose tolerance was completely abrogated in Glp1r^−/− mice (Fig. 4C).

Because obesity is often associated with decreased fasting ghrelin levels and impaired glucose tolerance, we next determined whether ghrelin pretreatment of obese mice could improve oral glucose tolerance in association with enhanced GLP-1 secretion. After 10 weeks of being fed the HFD, mice had a significant increase in body weight compared with paired chow-fed controls (42.6 ± 1.9 vs. 26.8 ± 1.2 g, P < 0.5). Obese mice had significantly lower pre-OGTT levels of ghrelin (Fig. 5A). During the OGTT, obese mice demonstrated significantly impaired glucose tolerance compared with control mice (P < 0.01 at t = 60 min, P < 0.05 at t = 90 min, and P < 0.05 overall; Fig. 5B). However, when obese mice were pretreated with ghrelin (100 nmol/kg), glucose tolerance was restored to a level similar to that of chow-fed controls (Fig. 5B). This improvement in glucose tolerance was associated with a significant increase in GLP-1 levels compared with obese mice treated with vehicle alone (P < 0.01 at t = 10 min and P < 0.05 overall; Fig. 5C).

Finally, to determine whether the effects of ghrelin on the L-cell are exerted directly, GLP-1 secretion was examined in two different L-cell models: the mouse GLUTag and human NCI-H716 cell lines. Ghrelin receptor mRNA expression was detected in both cell lines by RT-PCR of total RNA (Fig. 6A). Both cell lines also demonstrated significant increases in GLP-1 secretion in response to treatment with varying doses of acylated ghrelin (1–100 nmol/L) in regular culture media (P < 0.05 at 1 and 10 nmol/L and P < 0.01 at 100 nmol/L for GLUTag cell, and P < 0.05 at 1 nmol/L and P < 0.01 at 10 nmol/L for NCI-H716 cells; Fig. 6B and C). To investigate the role of glucose in ghrelin-stimulated GLP-1 secretion, GLUTag cells were also examined at 0, 1, and 5 mmol/L glucose. Cells incubated in 0 and 1 mmol/L glucose did not respond to ghrelin, whereas ghrelin significantly stimulated GLP-1 release in the presence of 5 mmol/L glucose (1.4 ± 0.2-fold, P < 0.05).

To investigate the signaling pathway involved in ghrelin-stimulated GLP-1 release, GLUTag cells were treated with ghrelin for 10 min, followed by immunoblot for phospho- and total ERK1/2. Ghrelin induced a significant increase in the levels of phospho-ERK1/2 (P < 0.05), an effect that was completely blocked by pretreatment with the mitogen-activated protein kinase (MAPK)/extracellular signal–regulated kinase (MEK)/ERK inhibitors U-0126 (Fig. 6D) and PD98059 (data not shown). Finally, to confirm a role for the MAPK pathway in ghrelin-mediated GLP-1 secretion, GLUTag cells were preincubated with U-0126 before being treated with ghrelin. ERK1/2 inhibition completely blocked the ability of ghrelin to stimulate GLP-1 secretion (Fig. 6E).

GLP-1 secretion has been shown previously to be regulated by orally ingested nutrients and by metabolic hormones, including GIP, leptin, and insulin (4,7–13,24,25). Importantly, the L-cell response to these heterologous secretagogues appears to be integrated, such that pretreatment with one effector can modulate the subsequence response to another, thereby ensuring appropriate levels of circulating GLP-1 (11–13). Because ghrelin plays key roles in communicating whole-body energy status and its levels are highest before a meal, a time point that also immediately precedes the GLP-1 response to the ingested nutrients, we aimed to determine if ghrelin pretreatment could affect GLP-1 release in response to oral glucose. Administration of acylated ghrelin alone, in the absence of an oral glucose load, did not modulate basal levels of GLP-1; the effects of ghrelin on GLP-1 were only observed after oral glucose administration. Interestingly, the effects of insulin on GLP-1 release also require the presence of glucose (12). This requirement for glucose in the effects of ghrelin on the L-cell, an effect that was confirmed in vitro, is of particular significance because ghrelin levels are highest during fasting, a period when the insulinotropic effects of GLP-1 are not required. Our findings therefore indicate that ghrelin alone does not lead to increases in GLP-1 but, rather, induces “priming” of the L-cell in preparation for subsequent oral glucose-stimulated GLP-1 secretion. Interestingly, the concept of priming has been established for other endocrine cells, including the effects of acetylcholine, γ-aminobutyric acid, and fatty acids on subsequent insulin release by the β-cell (27–29), as well as of a novel modulatory factor that modulates gonadotrope secretion of luteinizing hormone (30).

The results of the current study demonstrate that priming of the L-cell by ghrelin results in an improvement in oral glucose tolerance in lean and obese mice and that this is mediated through a direct mechanism of action in murine and human in vitro L-cell models. The physiological relevance of these findings in the context of the biological activities of GLP-1 is twofold. First, a small, albeit nonsignificant, increase in insulin levels was found in normal mice after ghrelin treatment before the OGTT. Because this occurred in the setting of reduced blood glucose levels, insulin levels relative to prevailing glycemia were therefore increased. This finding was complemented by the demonstration that insulin levels were decreased by pretreatment with a ghrelin receptor antagonist. Hence, some of the observed actions of GLP-1 to modulate glucose tolerance may have been modulated through the canonical incretin axis. However, an additional mechanism of action of GLP-1 also includes suppression of glucagon release (1,2), which may therefore have contributed to the glucose-lowering effects in our model.

Although the ghrelin doses used in the in vivo experiments are comparable with other studies examining ghrelin action in vivo (31,32), they do not represent physiological concentrations of preprandial ghrelin. To resolve this issue, we used a ghrelin receptor antagonist–based approach to determine the role of endogenous ghrelin in L-cell secretory responses. Treatment of mice with the established ghrelin receptor antagonist, D-Lys GHRP-6, resulted in a decline in oral glucose–stimulated GLP-1 and insulin secretion as well as impaired glucose tolerance. These findings support our initial findings of a positive role for ghrelin in GLP-1 release and downstream glucose tolerance but differ from those of Dezaki et al. (33), who reported that D-Lys GHRP-6 coinjected with IP glucose caused an enhanced insulin response and an improvement in glucose clearance. In addition, a recent study by Tong et al. (34) demonstrated that an infusion of acylated ghrelin into normal humans not only increased fasting glucose levels but also decreased the disposition index (which measures β-cell function) during an intravenous glucose tolerance test. The discrepancy between the results of these studies and our findings on glucose tolerance likely lies in the different routes of glucose administration. The injection of glucose intraperitoneally or intravenously represents a significant difference from the more physiological oral route used in the current study, in that only oral administration of glucose increases GLP-1 release (35,36), thereby inducing the incretin effect. Collectively, these findings suggest that ghrelin may play two distinct roles in the regulation of insulin and, in turn, glycemia. At the level of the β-cell, ghrelin can inhibit glucose-stimulated insulin secretion, a finding that is supported by in vitro β-cell experiments (37,38). However, as shown in our study, ghrelin also enhances oral glucose-induced GLP-1 release. The latter is likely of greater importance in the physiological regulation of glycemia because nutrients normally arrive through the oral route and the associated incretin response is responsible for most of the downstream effect on insulin by the β-cell (36).

Unexpectedly, somewhat variable total GLP-1 levels were found within and between the different strains of mice used in the current study. Hence, basal levels of GLP-1 in saline-injected controls ranged from 3 to 7 pg/mL in the C57BL/6 mice (Figs. 1–3), and this was increased slightly, to 10 pmol/L, in the GLP-1R WT mice (Fig. 4). The GLP-1 increments in response to identical OGTTs also varied somewhat, ranging from onefold (Fig. 1) to 2.5-fold (Figs. 2–4). However, the Glp1r animals were studied in a separate animal facility from the normal C57BL/6 mice, and the GLP-1 data for each figure were also obtained using separate ELISA plates. Thus, whether these differences in GLP-1 levels are physiologically meaningful or simply represent normal experimental variation remains unclear.

Feeding mice the HFD for 8–10 weeks produced a marked elevation in body weight and a concomitant reduction in oral glucose tolerance. Interestingly, these mice also had significantly lower levels of fasting ghrelin, as reported previously (20,21). Although we did not observe an impairment in the GLP-1 response to oral glucose under this diet paradigm (data not shown), a longer state of obesity may lead to a blunted GLP-1 response, as is the case in chronic human obesity (21). Nevertheless, obese mice treated with ghrelin demonstrated a robust increase in their oral glucose–stimulated GLP-1 response and an improvement in glucose tolerance. These data suggest that in the state of obesity, lower circulating ghrelin levels may contribute to the impaired clearance of oral glucose. Whether these findings hold clinical relevance is unclear, because the administration of ghrelin is known to induce weight gain and adiposity in lean rodents (39,40). However, ghrelin administration to obese rodents has been shown to improve growth hormone responses without causing additional weight gain (41). Future long-term studies examining the chronic effects of ghrelin administration to obese mice on GLP-1 levels and glucose tolerance should resolve this issue.

To establish whether the effects of ghrelin on GLP-1 release in vivo could be mediated through direct actions on the intestinal L-cell, we examined well-established in vitro models of the murine and human L-cell. Ghrelin treatment of the murine GLUTag and human NCI-H716 cells stimulated GLP-1 secretion, an effect that was mediated through the MEK/ERK1/2 pathway. It is well established that the ghrelin receptor activates MEK/ERK signaling in a number of different cell types, including intestinal epithelial cells (42–44). Furthermore, the MEK/ERK pathway has been previously shown to be required for stimulation of GLP-1 secretion by meat hydrolysate (25) and insulin (12), with recent studies further implicating ERK1/2 signaling in the circadian regulation of GLP-1 release by multiple secretagogues (45). Although the precise mechanisms linking MEK/ERK activation to GLP-1 secretion requires further investigation, previous studies have demonstrated essential roles for rho guanosine 5′-triphosphatase, cell division cycle 42, p21-activated kinase-1, and remodeling of the cortical F-actin cytoskeleton in this pathway (46). In line with this, actin-interacting SNARE (SNAP [Soluble NSF Attachment Protein] Receptor) proteins are also expressed in L-cells and are required for the exocytosis of GLP-1 (47,48). Further molecular analyses of intestinal L-cell signaling and exocytosis will be required to elucidate the exact pathways underlying the priming of GLP-1 secretory responses.

In summary, we have demonstrated a role for ghrelin in the regulation of oral glucose-stimulated GLP-1 secretion leading to an improvement in glucose clearance. Ghrelin actions are exerted directly on the L-cell and require the MAPK pathway. Finally, we have shown that ghrelin pretreatment of obese mice with reduced fasting ghrelin levels enhances GLP-1 secretion and restores glucose tolerance. Collectively, these findings indicate that ghrelin plays a novel role in the regulation of GLP-1 secretion.

Acknowledgments. The authors thank Dr. M. Gil-Lozano (Department of Psychology, University of Toronto for expertise and help with the cell culture studies.

Funding. J.G. was supported by a postdoctoral fellowship from the Canadian Institutes of Health Research (CIHR), D.J.D. by a Banting & Best Diabetes Centre-Novo Nordisk Chair in Incretin Biology, and D.J.D. and P.L.B. by the Canada Research Chairs Program. This work was funded by the CIHR (MOP-123391 to D.J.D.), by an operating grant from the Canadian Diabetes Association (OG-3-13-4024 to P.L.B.), and by an equipment grant from the Canadian Foundation for Innovation and Ontario Research Fund (#19442).

Duality of Interest. D.J.D. has consulted for Arisaph Pharmaceuticals Inc., Intarcia Therapeutics, Merck Research Laboratories, MedImmune, Novo Nordisk Inc., NPS Pharmaceuticals Inc., Receptos, Sanofi, and Transition Therapeutics Inc. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. J.G. researched data and wrote the manuscript. L.L.B. researched data and reviewed and edited the manuscript. D.J.D. reviewed and edited the manuscript. P.L.B. obtained funding, contributed to discussion, and reviewed and edited the manuscript. P.L.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

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