Targeting β-Cell Mass in Type 2 Diabetes: Promise and Limitations of New Drugs Based on Incretins

Abstract

Progressive insulin secretory defects, due to either functional abnormalities of the pancreatic β-cells or a reduction in β-cell mass, are the cornerstone of type 2 diabetes. Incretin-based drugs hold the potential to improve glucose tolerance by immediate favorable effect on β-cell physiology as well as by expanding or at least maintaining β-cell mass, which may delay the progression of the disease. Long-term studies in humans are needed to elaborate on these effects.

• I. Introduction

• II. The Incretin Effect and Its Mediators

  • A. Definitions and overview
  • B. Glucose-dependent insulinotropic polypeptide
  • C. Glucagon-like peptide 1
  • D. Similarities and differences between GIP and GLP-1
  • E. The incretins in diabetes
• III. Effects of the Incretins on β-Cell Mass

  • A. Effects of GLP-1 receptor signaling
  • B. Effects of GIP receptor signaling
• IV. Drugs Based on the Incretins and Their Effects in Diabetic Patients

  • A. GLP-1 receptor agonists
  • B. Dipeptidyl peptidase inhibitors
  • C. Potential of incretin-based drugs to affect β-cell mass in humans

• V. Summary

I. Introduction

THE LAST TWO DECADES have seen a major shift in the consensus on the pathogenesis of type 2 diabetes mellitus (T2DM). Although it is clear that resistance of peripheral tissues to insulin is an abnormality common to almost all patients with T2DM, the centrality of β-cell dysfunction as an essential component necessary for development of diabetic glucose tolerance is now widely accepted (1). One cornerstone of this revised model of diabetes is the recognition that the normal β-cell response to insulin resistance is a compensatory increase in insulin secretion sufficient to maintain glucose homeostasis (2,3). Failure of the β-cells to adapt to systemic insulin needs leads to a decrement in insulin- mediated glucose metabolism that causes hyperglycemia (4,5). A second observation underpinning the core role of the β-cell in the genesis of diabetes is the parallel course of insulin secretion and glucose homeostasis. There is evidence that a relative defect in β-cell function, either constitutive or acquired, is present before clinically apparent diabetes (6), suggesting that abnormal insulin secretion is a proximal defect in the process of glucose intolerance. Moreover, progressive β-cell failure is the unfortunate eventuality for most people with established T2DM, and the decline of insulin secretion is associated with worsening metabolic control and the need for more intensive treatment (7). These important clinical observations—that β-cell function can adapt to increased demand, and that β-cell dysfunction is proportional to the degree of glucose intolerance—have led to a currently accepted view that in most people islet function is fluid and amenable to adaptation, but that genetic or environmental factors exist in others that compromise matching of insulin secretion to insulin needs contributing to abnormal glucose metabolism.

Comparisons of insulin secretion in diabetic and nondiabetic humans have identified a number of characteristic abnormalities of β-cell function. That patients with T2DM have impaired insulin responsiveness to stimulation by glucose has long been recognized (8,9,10). This characteristic of diabetic islet function is manifest as the classic absence of first-phase insulin release, an apparent loss of β-cell sensitivity to glucose (11), and delayed insulin secretion in response to oral glucose (8,10). This latter response may be due in part to a decrease in the actions of gastrointestinal (GI) hormones and neural signals engaged after meals to potentiate the insulin response (12). Diabetic patients also have disruption of normal pulsatile insulin secretion (9), abnormal potentiation of nonglucose secretagogues, and a decreased maximal secretory capacity (13). Although the specific molecular mechanisms underlying these abnormalities of β-cell function in T2DM are still under study, taken together they result in delayed insulin responses that are inappropriately low for the degree of plasma glycemia.

Beyond functional defects of insulin secretion, it is also clear that patients with T2DM have anatomical abnormalities of the pancreatic islets. Pathological studies of diabetic and nondiabetic patients have demonstrated decreased islet size and decreased numbers of insulin-producing cells (14,15,16). This has been attributed to the formation of islet amyloid in diabetic islets (16,17,18), with replacement of β-cells, and to increased β-cell apoptosis (16). The reduction in β-cell number seems to precede diabetes, because subjects with impaired glucose tolerance have approximately 50% of the islet mass of weight-matched controls with normal glucose tolerance (16). Islet mass has been shown to correlate with fasting plasma glucose levels (19), suggesting that the number of β-cells, as well as their function, may play a role in physiological regulation. This observation is supported by data from subjects with autologous islet transplants after pancreatectomy, demonstrating that insulin secretory capacity is proportional to the size of the islet graft (20). The accumulated information supporting reduced islet mass in diabetic patients and evidence that processes inherent to the diabetic state such as islet amyloidosis (17,18), glucotoxicity (21), and inflammation (22) cause ongoing β-cell death have led to the widespread belief that the progressive nature of T2DM is due to a steady reduction of insulin secretory capacity.

There has been a general tendency to dichotomize abnormalities in insulin secretion and reduced β-cell mass when considering the role of islet dysfunction in T2DM. This is due in part to studies supporting distinct signaling pathways for the regulation of acute processes like exocytosis of insulin-containing granules and more chronic processes such as cellular hypertrophy, division, and apoptosis (23). In addition, studies of the effect of reduced β-cell mass have often made use of techniques such as partial pancreatectomy of healthy animals with functionally normal islets (24). However, it is plausible and, in fact, seems likely that functional defects in β-cell secretion and reduced numbers of β-cells are linked (25,26). Reduced numbers of β-cells have increased relative demands for insulin secretion and may develop functional defects comparable to diabetic islet function (24,27). Therefore, it can be argued that diminished islet mass has an impact on insulin output beyond simply lowering the overall capacity for insulin production.

Studies in animals suggest that islets hypertrophy and involute as necessary for varying states of insulin sensitivity (28). Examples include pregnancy (29), insulin resistance (30), and the response to pancreatectomy (31). These studies form the basis for considering β-cell mass as a regulable parameter. In this review, we will consider the potential of pharmacological interventions to promote β-cell mass as a means of treating T2DM. The focus of our discussion will be recently developed drugs that act through the incretin signaling pathways. The incretins are insulinotropic hormones derived from the GI tract that link nutrient absorption with β-cell secretion; the major incretins are glucose-dependent insulinotropic polypeptide (GIP) and glucagon-like peptide 1 (GLP-1). The incretins activate specific receptors on the β-cell that stimulate insulin release but also activate β-cell growth and inhibit apoptosis in vitro, and chronic stimulation of these pathways in rodents enhances β-cell mass. We will provide an overview of the incretin system and discuss the effects of GLP-1 and GIP to regulate β-cell proliferation, growth, and death. We will also discuss the effects of drugs that work through the incretin pathways in T2DM patients and whether any of these therapeutic effects can be explained by changes in β-cell mass.

II. The Incretin Effect and Its Mediators

A. Definitions and overview

It has long been known that the assimilation of glucose is more efficient when it is given orally than when it is given iv (32). After the development of the RIA and the ability to measure insulin in circulating plasma, investigators noted significantly greater insulin concentrations when glucose was ingested compared with it was given iv (33). The primary explanation for this observation is that factors released from the GI tract regulate insulin secretion and glucose homeostasis. These hormones have been termed incretins, and the relatively greater insulin secretion observed with oral compared with iv glucose is termed the incretin effect. There are two known peptides that act as incretins, GIP and GLP-1. The incretins are produced by specialized cells in the intestinal mucosa and secreted primarily in response to carbohydrate- and lipid-containing meals. There are specific receptors for GIP and for GLP-1 that are expressed on islet cells, as well as in other tissues, and deletion of these receptors in mouse models leads to glucose intolerance (34). These data, and corroborating studies using GIP and GLP-1 receptor antagonists, have demonstrated that so-called “enteroinsular” signaling is a physiological process that is necessary for normal glucose homeostasis.

B. Glucose-dependent insulinotropic polypeptide

GIP (originally referred to as gastric inhibitory polypeptide) is a highly conserved 42-amino acid peptide produced by enteroendocrine K cells in the duodenum and jejunum (35). GIP secretion is stimulated by enteral glucose and lipids in a dose dependent manner (36), and increased plasma levels augment glucose-stimulated insulin secretion (32). Consistent with the incretin effect, GIP release is directly regulated by the products of meal digestion and acts as a feed-forward mechanism to signal the endocrine pancreas of impending substrate fluxes from the gut (37).

After release into the circulation, GIP is rapidly metabolized by the enzyme dipeptidyl peptidase IV (DPP-IV), a ubiquitous protease located on capillary endothelial cells as well as in the circulation. DPP-IV cleaves specifically between residues 2 and 3 leaving GIP_(3-42) (35,38). This conversion occurs rapidly, so that the circulating half-life of full-length GIP is only 5–7 min in mammals. Metabolism by DPP-IV inactivates GIP because GIP_(3-42) does not stimulate insulin secretion.

A single GIP receptor (GIPr) has been identified and is currently believed to mediate all of the physiological effects of the peptide (39). The GIPr has substantial homology with the secretin-vasoactive intestinal peptide receptor family of seven-transmembrane G protein-coupled receptors. The GIPr is expressed in β-cells of the pancreatic islet, the upper GI tract, adipocytes, adrenal cortex, bone, pituitary, and a variety of brain regions (40,41). Binding of GIP to its receptor on the β-cell is insulinotropic only when ambient glucose concentrations are elevated, typically to greater than 5–6 mm (42). This action has been demonstrated uniformly in a wide range of experimental settings including cultured β-cells, animal models, and humans. In humans, both porcine and synthetic human GIP stimulate insulin secretion when infused to concentrations mimicking those occurring after meals as long as some degree of hyperglycemia is present.

The incretin role of GIP has been demonstrated by several experimental techniques such as immunoneutralization of circulating GIP and by administration of competitive antagonists of the GIP receptor, both of which cause glucose intolerance (43,44). Similarly, targeted gene deletion of the GIP receptor in a line of mice resulted in animals with normal glucose levels in the fasting state and in response to ip glucose loads (45). However, after oral glucose loading GIP receptor knockout mice had significant glucose intolerance and impaired insulin secretory responses. Taken together these studies in rodents indicate that the incretin action of GIP is necessary for normal glucose tolerance.

C. Glucagon-like peptide 1

GLP-1 is cleaved from proglucagon in specific intestinal mucosal cells termed L cells and secreted primarily as an amidated 30-amino acid peptide GLP-1(7-36)NH₂ (46). The distribution of the L cells that produce GLP-1 is greatest in the distal small intestine and colon. Similar to GIP, GLP-1 is also released in response to nutrient carbohydrate and fat. Plasma levels peak in the first 60 min after eating (36) and are proportional to meal size (47). Upon reaching the bloodstream, GLP-1, like GIP, is metabolized by DPP-IV, which cleaves the N-terminal dipeptide His-Ala leaving the circulating congener GLP-1(9-36)NH₂ (38,48). This rapid metabolism of bioactive GLP-1 results in a plasma half-life of 1–2 min and leaves inactive GLP-1(9-36)NH₂, which does not affect insulin secretion or glucose tolerance in humans at physiological concentrations (49).

GLP-1 binds to a specific GLP-1 receptor (GLP-1r) that is expressed on β-cells, as well as the GI tract, heart, lung, kidney, and neurons in the hypothalamus, hindbrain and several other brain regions, and the vagus nerve (50,51,52); controversy exists over whether the GLP-1r is also expressed by pancreatic α-cells. In β-cells the classic action of GLP-1 is to augment glucose-stimulated insulin secretion (23,53). Like GIP, GLP-1 stimulates insulin release in cultured β-cells, animals, and humans in a glucose-dependent manner. However, beyond acting as an acute insulin secretagogue, GLP-1 increases the biosynthesis of important β-cell products including insulin, glucokinase, and the GLUT 2 glucose transporter (54).

The important effects of GLP-1 on β-cell function have been elegantly demonstrated in studies of mice with a targeted deletion of the GLP-1r gene. These mice have glucose intolerance and delayed and diminished insulin secretion (55). Islets from GLP-1r −/− mice have a slight reduction in size, with a relative increase in the percentage of α-cells (56). In addition, the islets of these mice are more susceptible to the toxic effects of streptozotocin than control mice (57). Finally, absence of the GLP-1r blunts the compensatory growth after partial pancreatectomy (58). In sum, studies with GLP-1r −/− mice demonstrate broad effects of GLP-1 signaling on β-cell function, including effects on insulin secretion and growth/death.

D. Similarities and differences between GIP and GLP-1

The incretins share a number of similarities. GIP and GLP-1 have modest degrees of sequence homology as do their receptors. Both peptides are released in response to meal consumption with a dose-dependent relationship to meal size. Both GIP and GLP-1 are metabolized, and inactivated as incretins, by DPP-IV. Not surprising given the general similarities of the GIPr and GLP-1r, early intracellular signaling events after ligand binding are comparable. In the β-cell this causes enhanced insulin secretion during periods when blood glucose is elevated. The general commonalities shared by GIP and GLP-1 suggest the evolution of redundant systems to mediate the incretin effect. In fact, there is evidence that one system compensates in the absence of normal functioning of the other incretin (59,60).

Despite the many similarities between the incretins, there are several key points of distinction, including site of synthesis and the manner in which nutrient stimuli is coupled to secretion, mechanism of action, and extra-incretin effects. GIP is produced primarily in the upper intestine, and there is convincing data connecting K cell secretion of GIP with the digestion and absorption of lipid and carbohydrate (36,61,62). GLP-1 is made throughout the small intestine, but the highest concentrations of L cells are in the distal gut (63). In rats, the case has been made that stimulation of GLP-1 secretion involves signaling through vagal pathways (64,65), although it is not clear that there is a neural component to GLP-1 release in humans. GIP has characteristics more typical of a hormone, with a 5- to 10-fold increase in circulating concentrations after meals and a longer half-life comparable with other peptides that have endocrine actions. In contrast, GLP-1 circulates in lower molar amounts than GIP and increases only 1.5- to 2-fold after meal stimuli (66). The enhanced susceptibility of GLP-1 to DPP-IV and its abbreviated half-life raises the question of whether significant amounts of intact, bioactive, peptide actually reach the arterial circulation and target tissues (67). Therefore some controversy remains over how GLP-1 mediates its actions. Recent evidence supports a neural mechanism initiated by sensors in the hepatic portal vein that would have access to relatively high and dynamic concentrations of GLP-1 (52,56). Finally, beyond acting as an incretin, the only other prominent actions of GIP seem to be stimulus of glucose uptake and triglyceride storage in adipocytes (68,69), and possibly alteration of bone mass and bone turnover (70,71). In contrast, GLP-1 has a variety of effects that all seem to contribute to minimizing elevations in blood glucose. For example, GLP-1 reduces glucagon secretion (72), slows gastric emptying (73,74), induces satiety (75), suppresses hepatic glucose production (76), and promotes insulin-independent glucose disappearance (77,78). Research on the incretins over the last decade has demonstrated the range of actions that these peptides have on postprandial metabolism and supports the notion that GIP and GLP-1 have physiological roles that are distinct and complementary.

E. The incretins in diabetes

The incretin effect is impaired in persons with T2DM (8,12). This has raised the question of the sufficiency of circulating incretins in diabetic individuals. In fact, there is evidence from several groups that persons with type 2 diabetes have increased plasma concentrations of GIP after glucose or meal ingestion, compared with nondiabetic individuals (37,79). Although this finding has not been universal, there is no evidence for reduced GIP secretion as an explanation for a reduced incretin effect in diabetes. The case for GLP-1 is more complex. Different studies have reported that plasma GLP-1 levels are increased (80,81), decreased (47,66,82,83), and similar (84,85,86) in type 2 diabetic and nondiabetic subjects. These studies have used different assays of plasma GLP-1 so that the lack of consensus may be due to technical discrepancies. However, it seems more likely that these differences are the result of the relatively small sample sizes studied with distinct characteristics that could independently affect the results. Regardless, in studies where GLP-1 secretion has been reported as reduced in diabetic subjects, the relative decrease has been modest, suggesting that GLP-1 deficiency probably does not contribute significantly to β-cell dysfunction in T2DM.

Another major difference between GLP-1 and GIP is their activity in persons with diabetes. GIP is characteristic of a number of other GI peptides as being an insulin secretagogue in healthy subjects but relatively inactive in patients with type 2 diabetes. Thus, a number of studies have shown that diabetic subjects given iv GIP during a hyperglycemic stimulus have only minimal augmentation of insulin release (87,88,89). Several recent studies have suggested that the reduction in the GIP response in T2DM patients is not a specific defect but is proportional to other abnormalities in β-cell function (90,91). Nonetheless, it appears that in diabetic subjects with moderate glycemic control, acute infusions of GIP are not likely to be effective for reducing blood glucose.

The case with GLP-1 is completely different. Administration of iv GLP-1 to hyperglycemic T2DM subjects decreased blood glucose levels to nondiabetic values over 3–4 h (88). This effect was at least partly the result of increased insulin secretion and reductions in circulating glucagon. Similarly, continuous infusions of GLP-1 normalized fasting blood glucose overnight and caused postprandial levels to be near-normal as well (92). Administration of GLP-1 caused partial restitution of first-phase insulin release in subjects with established T2DM (76,93), demonstrating that this peptide reverses one of the hallmark lesions of diabetes. And although there is evidence that the β-cell sensitivity to GLP-1 is reduced in T2DM compared with nondiabetic subjects (94), pharmacological amounts of GLP-1 cause a robust improvement in stimulated insulin secretion (87,88,95). The potent effects of acute administration of GLP-1 to correct hyperglycemia and β-cell function in T2DM are persistent because 6 wk of continuous sc infusion caused significant improvements in these parameters in diabetic subjects (96). These findings and numerous other studies have shown that GLP-1 is unique among the insulinotropic gut hormones in maintaining potent effects on the diabetic islet, and for being able to correct hyperglycemia.

III. Effects of the Incretins on β-Cell Mass

A. Effects of GLP-1 receptor signaling

There is thought to be some remodeling of size and function of islet β-cells occurring throughout life that plays a role in the maintenance of normal glucose tolerance (28). Maintenance of islet cell mass, primarily through changes in numbers and size of β-cells, is generally thought to rely on the interplay of division and growth of established β-cells and on rates of apoptosis and β-cell loss (39). The role of neogenesis of endocrine precursors from pancreatic duct cells has been called into question as a significant mechanism for changes in β-cell mass in vivo by several recent studies (97,98,99). Although the role of these processes in normal physiology is still debated, especially with regard to human islets, there is now strong evidence that GLP-1 affects all three, at least in cultured cells and in rodent models. It has been established for some time that GLP-1 signaling in β-cells increases the transcription of several key genes related to glucose sensing and insulin secretion (54,100,101). However, more recent work has demonstrated that activation of the GLP-1r enhances β-cell replication and diminishes β-cell apoptosis, leading to an overall increase in β-cell mass. There is also evidence that GLP-1 can increase neogenesis of β-cells from precursor cells in the pancreas, although the significance of this as a developmental or adaptive mechanism is less clear. Although basic research in this area has proceeded with great speed over the past several years, translation of this work to humans has not yet been accomplished. At this point in time the work on GLP-1 and islet mass is strictly preclinical.

Activation of the GLP-1r in β-cells initiates signaling through the generation of cAMP (100) and activation of phosphatidylinositol-3 kinase (PI-3K) (Fig. 1). There are some data to suggest involvement of the epidermal growth factor receptor as a mediator of GLP-1-induced PI-3K activity (57). Acute effects of GLP-1 on insulin secretion, including enhanced intracellular calcium, membrane depolarization, and insulin exocytosis, are largely mediated through cAMP and subsequent activation of either protein kinase A (PKA) (102) or cAMP-regulated guanine nucleotide exchange factor II (Epac 2) (103). This pathway also affects gene expression through the actions of cAMP response element-binding protein. However, chronic effects of GLP-1, such as gene transcription, activation of cellular growth, and inhibition of apoptosis, can also be attributed to PI-3K downstream pathways, such as protein kinase B and protein kinase C (104,105). Besides cAMP response element-binding protein, the key intermediaries of GLP-1 signaling that mediate more chronic β-cell effects include insulin receptor substrate 2 and the transcription factor pancreatic duodenal homeobox factor-1 (PDX-1). In particular, GLP-1-mediated stimulation of β-cell gene transcripts, such as proinsulin, requires PDX-1 (106,107), although recently support for involvement of the transcription factor NFAT (nuclear factor of activated T cells) has been presented (108,109).

Figure 1.

Signaling pathways triggered in response to GLP-1 receptor activation. The actions of GLP-1 signaling have been subdivided into those mediating primarily acute responses, such as potentiation of insulin secretion, and those mediating chronic responses, such as gene transcription and cell replication. Although a general distinction between acute and chronic effects can be made, there is some degree of overlap and interaction between these signaling pathways.

Several lines of evidence suggest that GLP-1 has a role in β-cell and islet development. For example, activation of GLP-1r signaling promotes conversion of human or rat pancreatic ductal cell lines to islet-like cells that produce insulin in a glucose-dependent fashion (69,110,111). Moreover, GLP-1 stimulates the differentiation and maturation of fetal islet cells to insulin-producing islet cells (112). The effect of GLP-1 on β-cell differentiation involves the transcription factors PDX-1 and its upstream regulator Foxa2 (formerly HNF3β) (69,110,111) and appears to be mediated by MAPKs (69,104), including ERK (113), and an isoform of protein kinase C (104).

GLP-1 also seems to stimulate the proliferation of developed β-cells (104,114,115). Moreover, GLP-1 and GLP-1r agonists reduce apoptosis in primary rodent islets or islet cell lines exposed to cytotoxic agents, such as cytokines, free fatty acids, hydrogen peroxide, or streptozotocin (57,107,116,117). Activation of PDX-1 appears to be a shared component in all molecular pathways involved in cytoprotective effects of GLP-1. Enhanced PI-3K activity as a result of insulin receptor substrate 2 phosphorylation or epidermal growth factor receptor transactivation leads to the up-regulation of PDX-1 via nuclear exclusion of Foxo1 (118,119).

The overall effect of GLP-1r signaling on β-cell proliferation and apoptosis results in increased islet mass (105) and has demonstrable benefits in rodent models of β-cell insufficiency. Treatment with GLP-1r agonists has caused parallel increases in β-cell growth and glucose control in several rodent models of diabetes (115,120,121,122). A similar response has been attributed, at least partially, to endogenous GLP-1 protected from metabolism by DPP-IV inhibitors (117,119). Administration of the GLP-1r agonist exendin-4 during the neonatal period preserves β-cell mass and prevents the development of diabetes in rats with experimentally induced intrauterine growth retardation (58) or exposure to streptozotocin (123). Finally, exendin-4 treatment improves outcomes in diabetic mouse recipients of transplanted rat islets (124,125), as does treatment of mouse islets with adenoviral vectors to promote GLP-1 production by α-cells (126). These studies raise the interesting possibility that activation of the GLP-1r could increase β-cell mass, a novel and potentially effective approach to the treatment of diabetes. Table 1 summarizes the studies in rodents treated with compounds that increase GLP-1r activity that have assessed parameters of β-cell proliferation, apoptosis, or islet mass.

Table 1.

Studies in rodents demonstrating effects of GLP-1r agonists and DPP-IV inhibitors on parameters of islet mass

Incretin-based drug	Method of administration	β-Cell function	β-Cell histology/mass	Apoptosis
GLP-1	Continuous infusion for 5 d in old and young Wistar rats (121)	Improved glucose tolerance	Increased PDX-1 expression [reversed with infusing Ex-(9-39)], proliferation, and neogenesis.	NA
	Continuous infusion for 2 d in diabetic Zucker rats (120)	Improved glucose tolerance	Increased proliferation	↓
	Continuous infusion for 4–8 wk in 8-wk-old female NOD mice (158)	Improved glucose tolerance (lasted for 2–5 wk from the last dose of GLP-1 depending on the duration of treatment)	Increased proliferation and neogenesis	↓
		Diabetes did not occur in those treated for 8 wk by 21 wk of age (termination of study) and diabetes onset was delayed in rats treated for 4 wk.
	Increased GLP-1 expression in β−cells of Goto-Kakizaki type 2 diabetic rats secondary to ileal transposition (159)	Improved glucose tolerance	Improved morphology, absence of islet cell degeneration and interstitial fibrosis, and increased neogenesis.	NA
	Enhanced GLP-1 expression by im plasmid-based gene transfer in STZ-treated mice (122)	Improved glucose tolerance	Increased β-cell mass with improved morphology	NA
Exendin-4	Once daily for 10 d in diabetic rats with partial pancreatectomy (115)	Improved glucose tolerance in diabetic rats	Increased replication, neogenesis, and β-cell mass.	NA
	Once daily for 7–12 d in streptozocin-induced diabetic rats (160)	Improved glucose tolerance lasted 2 wk from discontinuing Ex-4	Increased expression of IDX-1 in the pancreatic ducts [reversed with infusing Ex-(9-39)], differentiation, and neogenesis	↓
	Daily for 2 wk in db/db mice (114)	Improved glucose tolerance		NA
	Once daily for 1 wk in diabetic mouse recipient of transplanted rat islets, which were cultured with exendin-4 (124)	Diabetes was cured in 80% of treated animals vs. 10% of controls at 4 wk.	Increased insulin-stained cells in the grafts of treated vs. untreated animals.	NA
	Once daily for 11 d (started on d 2) in diabetic mouse recipients of fetal islet-like cell cluster transplant (125)	Improved glucose tolerance in treated animals at 3-month follow-up	Enhanced maturation and differentiation of transplanted fetal islet cells into insulin-producing β-cells	NA
	Twice daily for 6 wk in fa/fa Zucker rats (161)	Improved insulin secretion and insulin sensitivity, assessed by euglycemic clamp	Increased β-cell mass*insulin sensitivity index, suggestive of trophic effect of GLP-1	NA
DPP-IV inhibitors	Des-fluoro-sitagliptin once daily for 11 wk in high-fat diet, streptozocin-induced diabetic mice (162)	Improved glucose tolerance.	Increased proliferation and neogenesis to restore β-cell mass to normal size. Enhanced pancreatic insulin content	NA
		Enhanced pancreatic insulin secretion
	P32/98 twice daily for 7 wk in STZ-treated Wistar rats (117)	Improved glucose tolerance and insulin secretion	Enhanced proliferation and β-cell mass by 120% (intervention group gained weight by 230%)	NA
	NVP DPP728 for 8 wk in C57BL/6J mice fed normal or high-fat food in comparison with controls in each group (163)	Improved glucose tolerance and insulin secretion in both high-fat and normal food groups receiving DPP-IV inhibitors compared with that in their counterpart controls.	No changes in islet morphology in either group, although islet cell size was significantly reduced in high-fat mice compared with that in their controls	NA

↓, Decreased; NA, not addressed. 

It is important to note that despite the accumulation of an impressive body of evidence that the GLP-1r signaling promotes the expansion or protection of β-cell mass, this information is limited to studies of cultured cells or rodent models. There are important differences in rates and capacity for islet cell turnover and growth between rodents and humans (127), such that extension of findings from animal studies cannot be assumed. Because it is not currently possible to evaluate β-cell mass noninvasively, the question of GLP-1 effects on β-cell proliferation and apoptosis in human subjects cannot be reliably addressed in longitudinal or intervention studies. Rather, the best extensions of the interesting findings in rats and mice to humans are inferential and based on functional studies. For example, administration of exenatide for 3 months in a group of C-peptide-positive islet cell transplant recipients with uncontrolled glycemia resulted in increased insulin secretion and improved meal tolerance, which was sustained for 1 month after the last dose of exenatide (128). Although this finding is compatible with effects of GLP-1r signaling on growth and viability of the islet grafts, it could also be explained by indirect effect on glycemic control or other aspects of metabolism. Thus, whereas preclinical research has provided hope and excitement for GLP-1 based drugs, the critical evidence from clinical studies is not yet available.

B. Effects of GIP receptor signaling

GIP signaling in the β-cell shares many similarities to GLP-1. GIP binds to a specific G protein-coupled receptor that is related to the GLP-1r (129) and enhances glucose-stimulated insulin secretion. The general molecular mechanisms of GIP activation of the β-cell are similar to those of GLP-1, including production of cAMP and activation of PKA, increased intracellular Ca²⁺, and closure of K_ATP channels, causing insulin exocytosis (53). However, like GLP-1, a significant component of GIP signaling is independent of PKA and acts through the Epac 2 pathway (130). Administration of GIPr antagonists (44) or targeted deletion of the GIPr gene (45,59) causes a deterioration of glucose tolerance and reduced glucose-stimulated insulin secretion in rodents. Importantly, islet cells from GIPR −/− mice exhibit reduced proinsulin gene expression and insulin content compared with wild-type mice (59), indicating that GIP does more than simply stimulate insulin release during hyperglycemia.

GIP also has proliferative and antiapoptotic effects on β-cells. GIP enhances β-cell proliferation synergistically with glucose (131) through pleiotropic mechanisms that include PKA activation, PI-3K/protein kinase B activity, and p38 MAPK. GIPr signaling protects β-cell lines from the cytotoxicity of streptozotocin (132), glucolipotoxicity (133), and starvation from glucose or serum (134). The cytoprotective effects of GIP are at least partly mediated by a suppression of caspase-3, a mediator of apoptosis (132).

Several studies in animal models have demonstrated the potential importance of GIPr signaling in maintaining islet cell mass. Administration of GIP-antagonist for 11 d to ob/ob mice partially reversed islet hypertrophy with a concurrent improvement of hyperinsulinemia and insulin resistance (68). Compatible with this finding, continuous administration of GIP to Zucker diabetic fatty rats for 2 wk diminished β-cell apoptosis (133). There are as yet no studies of chronic effects of GIPr agonists in humans, so the intriguing results of these animal studies have not been extended to possible clinical applications. However, it is important to note that new drugs that act by blocking the action of DPP-IV increase levels of GIP as well as GLP-1. Therefore, manipulations that increase islet mass in rodents by reducing DPP-IV action are potentially due to actions of both incretins (117,135).

IV. Drugs Based on the Incretins and Their Effects in Diabetic Patients

The last several years have seen the introduction of two classes of antidiabetic drugs based on incretin signaling for clinical use. These classes include GLP-1r agonists and DPP-IV inhibitors, which increase endogenous levels of GLP-1 and GIP (136). With the progress of these compounds to market and their use by large numbers of patients with diabetes, it has been possible to gain insights into the effects of chronic stimulation of the incretin pathways on β-cell function (137). The results of clinical studies of incretin-based drugs has engendered considerable discussion over their potential effects to promote β-cell mass. However, given that these studies have been either randomized trials with clinical outcomes or short-term physiological experiments with functional measures, the matter remains mostly conjectural at this point.

A. GLP-1 receptor agonists

There are a number of compounds developed to activate the GLP-1r that are resistant to DPP-IV and so have extended activity. The most established of these new drugs are synthetic exendin-4 (exenatide; Amylin Pharmaceuticals Inc., La Jolla, CA), and liraglutide (formerly NN2211, Novo Nordisk, Copenhagen, Denmark). Exenatide has been available to treat patients since 2005, and several hundred thousand patients have been treated with this drug to date; liraglutide is in advanced clinical trials.

Exendin-4 is a naturally occurring reptilian peptide of 39 amino acids with considerable homology to GLP-1 (136), which mimics to a great extent the actions of GLP-1 (138). Exenatide has a half-life of 2–3 h in humans after sc injection (139), and so provides a relatively extended activation of the GLP-1r. When administered iv to healthy humans, exenatide affects fasting glucose and insulin levels minimally, demonstrating the glucose dependence that is a noteworthy component of GLP-1 action (140). When administered during the consumption of a standard meal, exenatide infusions reduce the glycemic excursion significantly, at least partially due to enhanced β-cell function. The effects of exendin-4 on insulin secretion persist for up to 3 h after administration (141) and do not affect the normal endocrine responses to hypoglycemia (142).

In diabetic subjects, administration of exenatide stimulates insulin secretion, reduces glucagon levels, reduces body weight and presumably insulin resistance, and delays gastric emptying (139,143), all actions that would tend to mitigate postprandial hyperglycemia. Exenatide partially restores first-phase insulin secretion in T2DM subjects (144), correcting one of the classic abnormalities of diabetes. In type 2 diabetic patients participating in clinical trials, generally subjects under moderate glycemic control with conventional therapies, exenatide improved glycemic control as reflected in reduced hemoglobin A1c levels (133,145,146,147). These findings indicate that the effect of GLP-1r signaling to enhance insulin secretion in acute studies of diabetic subjects persists with chronic intermittent administration of exenatide. However, these results do not necessarily support chronic effects of GLP-1r signaling, such as gene transcription, increased β-cell replication, or reduced apoptosis, which cannot be readily assessed in human studies. In a recent examination of subjects treated with exenatide for 2 yr, the reduction of hemoglobin A1c levels was maintained at a stable level (148). Because of the known progression of type 2 diabetes over time (7), with a general tendency for hemoglobin A1c to worsen with fixed treatment, these results raise the possibility that treatment with exenatide can stabilize the progressive course of T2DM. Such results are consistent with, but do not prove, an effect of exenatide on β-cell function beyond that of an acute secretagogue. However, the 2-yr duration of follow-up is in actuality relatively short to comment on effects to change the natural course of diabetes. In addition, because this was an open-label trial of subjects continuing participation from randomized trials of exenatide, it lacks the proper controls to make a firm conclusion about effects on diabetes progression.

The other GLP-1 mimetic in advanced trials is liraglutide. This compound is a modified GLP-1 molecule that includes a C-16 fatty acyl derivative that promotes binding to albumin. Liraglutide is resistant to metabolism by DPP-IV, is absorbed gradually from the sc space after injection, and reaches peak levels 12 h after administration (149), with an elimination half-life of 12 h (150). Similar to GLP-1, liraglutide improves the sensitivity of the β-cell to glucose in persons with type 2 diabetes (151) and partially restores first-phase insulin release (142). The clinical trials with liraglutide that have been reported are relatively short (138,152), so there is insufficient information to make conclusions about chronic effects of this compound.

A long-acting microsphere-based preparation of exenatide (long-acting release exenatide, LAR-exenatide, Amylin Pharmaceuticals Inc.) has been developed to extend the half life of exendin-4. Administration of 2 mg of LAR-exenatide in 45 patients with poorly controlled type 2 diabetes who were on metformin or lifestyle intervention lowered hemoglobin A1c levels by 1.7%, relative to placebo-treated patients, and body weight by 3.7 kg over a 15-wk period (153). The advent of an extended-release form of exenatide is a view into the future of this class of drugs. There has been a broad effort to develop GLP-1 mimetics that are released into the circulation slowly after injection and have activity that extends over days to weeks. The recent findings with LAR-exenatide provide a proof-of-concept for this approach. Further work will be needed to determine the ultimate efficacy of drugs that provide protracted exposure of GLP-1r to ligand, and whether they provide other advantages to diabetic patients.

B. Dipeptidyl peptidase inhibitors

The action of DPP-IV provides a robust constraint on the incretin effect because it rapidly inactivates GLP-1 and GIP. Several compounds are currently available that provide nearly complete and long-lasting inhibition of DPP-IV, which increases the proportion of active GLP-1 and GIP from 20–30% of the total circulating incretins to 75–90% (154). One DPP-IV inhibitor, sitagliptin (Merck, Rahway, NJ), is now available to treat type 2 diabetic patients, with a number of other compounds in development; of these vildagliptin (Novartis, Basel, Switzerland) has been the subject of most research published on these drugs.

Sitagliptin is an orally available drug that can lower the measurable activity of DPP-IV by more than 95% for 12 h (154). Sitagliptin treatment causes a greater than 2-fold elevation of active GIP and GLP-1, and these are associated with increased insulin secretion, reduced glucagon levels, and improvements in both fasting and postprandial hyperglycemia. In randomized, placebo-controlled studies of sitagliptin in type 2 diabetic patients, glycemic control is improved, as reflected by decreased hemoglobin A1c by 0.5–1%, after several months of monotherapy or addition to other oral agents (154,155,156). The effects of sitagliptin have been ascribed primarily to effects on islet cell function. This drug does not cause weight loss and presumably has minimal effects on insulin sensitivity.

Vildagliptin seems to be comparable to sitagliptin as an inhibitor of DPP-IV activity and has similar effects on incretin levels and glucose control in moderately controlled type 2 diabetic patients (136). Vildagliptin improves insulin secretion and reduces plasma glucagon. Similar to exenatide, the effects of vildagliptin on glycemic control have been followed in studies of up to 52 wk. Compared with patients taking metformin who had a gradual increase in hemoglobin A1c over the course of the trial, those taking the DPP-IV inhibitor had a stable and durable reduction in average plasma glucose (136).

C. Potential of incretin-based drugs to affect β-cell mass in humans

Based on current clinical and pharmacological studies, it appears that the effects of GLP-1r agonists in humans can be extrapolated from what is known about the basic biology of the GLP-1 system. Thus, administration of exenatide, which results in the equivalence of very high levels of GLP-1, recapitulates many of the known pharmacological effects of this hormone- increased insulin secretion, decreased glucagon output, delayed gastric emptying, and reduced food intake. In addition to these positive effects, high levels of GLP-1r action can also cause the nausea that is a prominent side effect of GLP-1r agonists. In contrast, DPP-IV inhibitors only increase plasma GLP-1 and GIP into the high normal or supraphysiological range. Because DPP-IV inhibitors do not share the effects of exenatide on food intake, gastric emptying, or nausea, it has been assumed that these responses are the result of pharmacological amounts of GLP1–1r signaling. Despite significant differences in the relative amount of circulating GLP-1 activity caused by the administration of exenatide or DPP-IV inhibitors, these drugs appear to be equally efficacious for glucose lowering (137). This interesting observation suggests one of two possibilities. Either the physiological mechanisms activated by exenatide and DPP-IV inhibitors that account for improved glycemic control are not a strict function of the amount GLP-1r agonism, or DPP-IV inhibitors act by protecting factors beyond GLP-1, such as GIP.

It is not clear from animal studies whether the effects of GLP-1r signaling to promote expansion or protection of islet cell mass are dose dependent; for the most part these studies used a single dose of exendin-4. However, a single study in a mouse model of diabetes reported a graded response of islet mass and β-cell number to increasing doses of sitagliptin (119). Based on these limited findings, it is not possible to conclude whether the effects of incretin-based drugs to increase islet mass in animal models is dependent on the relative dose of GLP-1r signaling, chronic physiological increases of GLP-1 and GIP, or both.

There are no direct measures currently available to determine whether an antidiabetes drug has the ability to alter the course of T2DM by increasing β-cell mass or attenuating β-cell apoptosis. The progressive nature of diabetes in humans has been inferred from autopsy studies (16), longitudinal monitoring of glycemia (7), and functional measures of insulin secretion (13). Noninvasive imaging techniques that can assess islet mass are currently being explored but have not yet reached the sensitivity that is required for use in humans (26), and pancreatic tissue for histological examination cannot be ethically procured for research purposes only. This leaves no direct means for testing direct effects of incretin-based drugs on β-cell mass in diabetic patients. Given this limitation, the best alternative currently available to gain insights into the effects of GLP-1r agonists and DPP-IV inhibitors on the course of β-cell function in diabetes may be careful natural history studies. An excellent example of this is the recently published A Diabetes Outcome Progression Trial study (157). In this trial, patients with relatively early stage T2DM were randomized to receive one of three conventional oral agents and followed for 4 yr to assess the relative rate of decline in glycemic control. The results demonstrated that patients receiving glyburide had the most rapid decline in glycemic control, those allocated to rosiglitazone the slowest, with the group getting metformin intermediate between these two. Given the current view that progression of diabetes, as reflected in worsened glycemic control on stable therapy, is thought to be at least partially the result of loss of islet mass, these results are compatible with different levels of β-cell protection among different therapeutic agents. Therefore, the first step in ascertaining whether incretin-based drugs can affect islet mass in humans is likely to come from clinical trials such as ADOPT. Comparison of the course of diabetes in patients treated with GLP-1r agonists or DPP-IV inhibitors, using durability of effect on glycemic control as the primary outcome, would provide some insight into whether these agents have chronic effects on β-cell health. Although such studies cannot prove that the effects of incretin-based stimulation of islet mass in animals also pertain in humans, they would go a long way toward establishing the optimal role for these drugs in clinical medicine.

V. Summary

The progressive nature of type 2 diabetes is one of the major challenges in the treatment of affected patients, and agents that could alter the natural history of this condition would add greatly to current treatment approaches. The incretins, GLP-1 and GIP, appear to stimulate islet growth and protect β-cells from a number of stressors. In rodent models of diabetes, GLP-1r agonists and DPP-IV inhibitors increase islet mass and preserve β-cell function. The recent availability of these drugs in the clinic has generated excitement that the beneficial effects seen in preclinical studies will also be conferred to diabetic patients using them. Exenatide and DPP-IV inhibitors are effective for lowering blood glucose in treated patients but the mechanisms by which they work in humans are not completely understood. Although the promise of in vitro and animal studies certainly warrants continued investigation into the effects of incretin-based drugs on islet mass, at present there is not strong evidence that they alter the course of β-cell function in human diabetes.