GLP-1R and amylin agonism in metabolic disease: complementary mechanisms and future opportunities

Abstract

The discoveries of the incretin hormone glucagon-like peptide-1 (GLP-1) and the β-cell hormone amylin have translated into hormone-based therapies for diabetes. Both classes of molecules also exhibit weight-lowering effects and have been investigated for their anti-obesity potential. In the present review, we explore the mechanisms underlying the physiological and pharmacological actions of GLP-1 and amylin agonism. Despite their similarities (e.g. both molecular classes slow gastric emptying, decrease glucagon and inhibit food intake), there are important distinctions between the central and/or peripheral pathways that mediate their effects on glycaemia and energy balance. We suggest that understanding the similarities and differences between these molecules holds important implications for the development of novel, combination-based therapies, which are increasingly the norm for diabetes/metabolic disease. Finally, the future of GLP-1- and amylin agonist-based therapeutics is discussed.

LINKED ARTICLES

This article is part of a themed section on Secretin Family (Class B) G Protein-Coupled Receptors. To view the other articles in this section visit http://dx.doi.org/10.1111/bph.2012.166.issue-1

Introduction

Worldwide estimates suggest that over 1 billion individuals are overweight and 300 million are obese. Obesity and excess weight are established risk factors for diseases like type 2 diabetes mellitus. For obesity, available pharmacological therapies are lacking, with more obesity agents being withdrawn from the market over the past decade relative to those actually approved. The last two decades have seen a dramatic increase in our understanding of the neurohormonal controls of energy balance and a variety of neurohormones have been explored for their anti-obesity potential as mono- and combination-therapies (Morton et al., 2006; Field et al., 2009; Cegla et al., 2010). Like obesity, diabetes mellitus is a multi-hormonal disease that afflicted 171 million people worldwide in 2010 and prevalence is projected to rise to 366 million by the year 2030 (World Health Organization, 2011). To maintain normoglycaemia, a complex, well-orchestrated interplay among several biological processes is required. Insulin secretion and action, β-cell health, glucagon secretion, energy intake and gastric emptying all play a significant role in determining glucose concentrations in the circulation. Relative to obesity, the available arsenal for the treatment of diabetes is larger with traditional approaches addressing insulin insufficiency through dietary/lifestyle modification and/or overcoming insulin resistance with oral agents that improve glycaemia through mechanisms that include reduction in hepatic glucose production or increases in the sensitivity and/or release of insulin (e.g. metformin, thiazolidinediones, sulfonylureas). However, some of these therapies are associated with weight gain and cardiovascular risk, and ultimately do not address the progressive decline of the β-cell in diabetes. When these modalities fail individuals ultimately require exogenous insulin therapy.

These traditional views regarding the physiological and pharmacological regulation of glycaemia significantly evolved following the discoveries of the incretin hormone glucagon-like peptide-1 (GLP-1) and the β-cell hormone amylin in the mid-1980s. Both have translated into hormone-based therapies for diabetes in the forms of GLP-1 receptor (GLP-1R) agonists (e.g. exenatide) and the amylin agonist, pramlintide. Both classes of therapeutics also exhibit weight-lowering effects and have been investigated for their anti-obesity potential. On the surface, their glucose- and weight-lowering effects are orchestrated through seemingly overlapping physiologies. For example, although insulinotropism is unique to GLP-1 agonism, both agents decrease food intake, slow gastric emptying and decrease postprandial glucagon excursions. However, a careful examination of the receptor distribution, pharmacology, peripheral physiology and neurobiology of GLP-1- versus amylin-ergic systems reveals important differences between how these agents coordinate their physiological effects. These differences have significant implications for the development of novel, combination-based therapies, which is increasingly the norm for diabetes/metabolic disease. Hence, the focus of this review is to detail the mechanisms underlying the physiological and pharmacological actions of GLP-1 and amylin agonism and speculate on what future therapeutic modalities invoking these systems may hold.

Basic physiology

GLP-1

Glucagon-like peptide-1 is a product of the preproglucagon gene (Lund et al., 1982), produced primarily by the L-cells of the distal intestine, although immunoreactive GLP-1 cells have also been localized in the mid-small intestine (Mortensen et al., 2003). GLP-1 secretion is stimulated by nutrient ingestion and the peptide is present in multiple forms (1-36, 1-37, 7-36, 7-37). The bioactivity of the forms and degradation products of GLP-1 is an ever-changing landscape, with recent reports documenting important effects of a new cleavage product GLP(28–36) targeting mitochondria and modulating oxidative phosphorylation (Tomas et al., 2011). Nevertheless, the most well-researched and important biologically active cleavage product appears to be GLP-1(7–36) amide. GLP-1's physiological effects include increasing glucose-stimulated insulin secretion, suppressing glucagon, controlling the rate of gastric emptying and reducing food intake (Holst, 2007). Following secretion it is rapidly inactivated (half-life ∼2 min in rats) by dipeptidyl peptidase-4 (DPP-4) (Mentlein et al., 1993). To overcome the pharmacological limitations of GLP-1's rapid degradation, drugs that are resistant to DPP-4 degradation (incretin mimetics or GLP-1R agonists) or which increase endogenous GLP-1 concentrations towards the normal range (DPP-4 inhibitors) have been developed.

Amylin

Amylin is a 37 amino acid peptide secreted from the pancreatic β-cells (Ogawa et al., 1990). It is co-secreted with insulin in response to nutrient intake and the secretion of amylin is anatomically linked to insulin secretion with the diurnal patterns of amylin and insulin plasma concentrations being almost super-imposable (Koda et al., 1992; Weyer et al., 2001). Approximately 60% of immunoreactively detected amylin is glycosylated, although these forms were inactive in in vitro assays and their biological relevance is not well understood (Young, 2005b). Nephrectomy studies suggest that amylin is primarily cleared renally but also catabolized by renal peptidases associated with the vascular supply (Vine et al., 1998). Pharmacokinetic studies that the terminal half-life of amylin in rats is ∼13 min and that the half-life for the amylin agonist pramlintide in humans is ∼20–45 min (Colburn et al., 1996; Young, 2005a). Amylin's gluco-regulatory actions complement those of insulin by regulating the rate of appearance of glucose in the circulation and are achieved through three primary mechanisms: slowing the rate of gastric emptying, suppression of post-meal glucagon secretion and suppression of food intake. The clinical utility of human amylin is limited by instability and a propensity for self-aggregation. These challenges were overcome by the substitution of prolines at positions 25, 28 and 29 of human amylin (Colburn et al., 1996). This resulted in a synthetic amylinomimetic peptide, pramlintide, with improved stability and decreased potential for aggregation.

Receptor pharmacology

GLP-1 receptors

The GLP-1R was originally cloned from a rat pancreatic islet cDNA library in 1992 (Thorens, 1992). The human receptor was isolated by several groups in 1993 and found to be 95% homologous to the rat protein (Dillon et al., 1993; Graziano et al., 1993; Thorens et al., 1993). The GLP-1R is a member of the class B GPCRs (secretin family) and is 463 amino acids in length. Briefly, Class B receptors are classic seven transmembrane proteins whose C-terminus interacts with a signalling G protein and whose large N-terminal extracellular domain plays an important role in ligand binding (Parthier et al., 2009). GLP-1R signalling mainly occurs through a Gαs coupled subunit that leads to increased cAMP (cyclic adenosine monophosphate) and Ca²⁺ concentrations. However, these second messengers can also be increased via signalling through other G protein subunits such as G αq, G αo and Gαi, activating other downstream signalling pathways [e.g. protein kinase A, protein kinase C, Phosphatidylinositol 3-kinase, exchange proteins activated by cAMP and MAPK pathways (Doyle and Egan, 2007)]. Subsequent expression studies have detected receptor transcripts in multiple tissues and in various species. In mice, mRNA was found in pancreatic islets, brain, heart, kidney and the GI tract (Campos et al., 1994). In rats, a similar but expanded profile of mRNA expression identified transcript in pancreatic islets, lung, hypothalamus, stomach, heart and kidney (Bullock et al., 1996). In situ hybridization studies have localized GLP-1R's in regions of the brain classically implicated in regulating energy homeostasis (e.g. brainstem and hypothalamus) as well as other regions (Merchenthaler et al., 1999). More recently, a role for lingual GLP-1 receptors has been proposed (Shin et al., 2008). Notable species differences in receptor expression include the observation that rats have higher expression in lung tissue and thyroid C cells than humans (Korner et al., 2007; Bjerre Knudsen et al., 2010). Although dogs have detectable GLP-1R expression in muscle and adipose tissue (Sandhu et al., 1999), there is controversy about whether or what form of GLP-1 receptors are present in human and rodent muscle, adipose tissue and liver in view of GLP-1 activity on those tissues (Campos et al., 1994; Egan et al., 1994). The biological relevance of all of these inter-species comparisons will also depend upon a host of other factors including the prevailing endogenous hormone concentrations and receptor affinities within each species.

Amylin receptors

As with GLP-1, amylin stimulates receptors containing a class B GPCR. However, with the amylin receptor there is an added layer of complexity due to the fact that the base GPCR heterodimerizes with a receptor activity modifying protein (RAMP). RAMPs are a family of three type I transmembrane proteins composed of a large N-terminal extracellular domain, a single alpha helical transmembrane domain and a short intracellular section. These proteins regulate receptor pharmacology, receptor signalling and receptor trafficking [originally described in McLatchie et al. (1998)] and for a recent review see Sexton et al. 2009). RAMPs were initially shown to regulate and modify the calcitonin receptor-like receptor to form distinct receptors for calcitonin gene-related peptide and adrenomedullin. Subsequent work found that RAMPs also modify the pharmacology of the calcitonin receptor such that specific amylin receptors emerge from the dimerization of various splice variants of the calcitonin receptor (CT(a)/CT(b)) with either RAMP1, RAMP2 and RAMP3. These receptors are commonly referred to as AMY₁, AMY₂ and AMY₃ with either an a or b in the subscript to define which splice variant of the calcitonin receptor is in the complex (Poyner et al., 2002). Overall, the distribution of amylin receptors appears to be more localized relative to GLP-1R. Amylin receptors have been localized via binding studies to the nucleus accumbens, the dorsal raphe and the area postrema (AP) in the hind brain (Beaumont et al., 1993). In situ studies in rats revealed that AMY_2(a) and AMY_3(a) are the amylin receptor subtypes localized to the AP suggesting these subtypes are responsible for the satiogenic effect of amylin (Barth et al., 2004) (discussed in further detail below). Within the AP, the key second messenger system appears to be cGMP (Riediger et al., 2001). Messenger RNA for calcitonin receptor-like receptor + RAMP1 or RAMP2 are found in the subfornical organ (SFO) and are likely responsible for amylin's induction on drinking (Barth et al., 2004). RT-PCR studies demonstrate that RAMP1 and 2 but not RAMP3 are present in the rat nucleus accumbens suggesting that the amylin receptor in the nucleus accumbens is either AMY₁ or AMY₂ (Li et al., 2004). The role of these amylin receptors, as well as their cognate ligand, has not been well established but has been proposed to potentially link amylin's regulation of motor activity and ingestive behaviour (Baldo and Kelley, 2001). Infusion studies suggest that amylin crosses the blood–brain barrier potentially accessing a number of regions (e.g. cerebellum, midbrain, frontal cortex, parietal cortex and occipital cortex) (Banks et al., 1995; Banks and Kastin, 1998), although the physiological relevance of these findings merits further investigation.

Overlap of CNS GLP-1 and amylin receptors and signalling

Amylin-activated cells (by c-Fos (Potes et al., 2010) as well as GLP-1R-positive cells (Yamamoto et al., 2003) have been localized to tyrosine hydroxylase containing cells within the hindbrain AP, although double labelling of amylin and GLP-1R's has never been performed. Both amylin and GLP-1 appear to activate somewhat overlapping neural circuits. GLP-1 agonism induces c-Fos in the AP, caudal nucleus of the solitary tract (cNTS), the lateral parabrachial nucleus, hypothalamic paraventricular nucleus (PVN), bed nucleus of the stria terminalis and central nucleus of the amygdala (Rowland et al., 1997; Baggio et al., 2004). Likewise, amylin induces c-Fos expression in the AP, NTS, lateral parabrachial nucleus and the central nucleus of the amygdala. The AP is far more important for transducing the effects of amylin relative to GLP-1. AP lesions abolish the anorexigenic effects of amylin (Lutz et al., 1998b; Rowland and Richmond, 1999; Riediger et al., 2004; Becskei et al., 2007; Mack et al., 2010), whereas those of GLP-1 are maintained (Baraboi et al., 2010). Studies using [Ser8]GLP-1 (a more stable analogue of GLP-1) suggest that peripheral GLP-1 can cross the blood–brain barrier by simple diffusion (Kastin et al., 2002). However, because of its rapid degradation by DPP-IV, to what extent under physiological conditions peripheral GLP-1 actually reaches the CNS has been debated. Gut-derived GLP-1's central effects are most likely propagated via interactions with afferent vagal nerve fibres arising from the nodose ganglion, sending impulses to the NTS and onwards to the hypothalamus (Vahl et al., 2007). The NTS does contain a population of GLP-1 synthesizing neurons that project to the PVN (Larsen et al., 1997), representing another potential point of overlap between amylin and GLP-1 signalling. Irrespective of their primary mechanisms for activating CNS neural circuits, both peptides have been shown to regulate the activity of a number of key neurohormonal signals (e.g. neuropeptide Y, pro-opiomelanocortin, leptin, cholecystokinin) in a variety of nuclei (e.g. AP, NTS, ARC, PVN, ventromedial hypothalamus) implicated in energy homeostasis (Roth et al., 2006; Seo et al., 2008; Becskei et al., 2009; Skibicka et al., 2009; Williams et al., 2009; Turek et al., 2010).

Central and peripheral signalling pathways

In this section we compare and contrast (when known), the primary pathways mediating the effects of GLP-1 and amylin agonism to suppress glucagon, slow gastric emptying and decrease food intake. These effects are summarized in Table 1.

Table 1.

Summary of Features and Mechanisms Reviewed

	GLP-1	Amylin
Source	• Intestinal L-cells	• Pancreatic β-cells
	• Neurons in NTS
Receptor and distribution	• Class B GPCR	• Class B GPCR + RAMP
	• Widespread receptor distribution (e.g. islets, brain, heart, kidney, gastrointestinal tract)	• Localized CNS distribution – key receptors in hindbrain AP. Also expressed in nucleus accumbens, dorsal raphe and SFO
β-cell	• Insulinotropic/Amylinotropic	• Co-secreted and co-released with insulin
	• Increased β-cell mass via direct and indirect mechanisms
Glucagon secretion	• Glucagonostatic (with ‘hypoglycaemic’ override)	• Glucagonostatic (with ‘hypoglycaemic’ override)
	• Direct and indirect/paracrine mechanisms reported	• Activation of AP and then vagally transmitted (proposed)
Gastric emptying	• Decreases rate	• Decreases rate
	• Current evidence favours neural reflex triggered by peripheral GLP-1 activating vagal afferents innervating the gut/portal area	• Activation of AP, then transmitted by vagal efferents
Food intake	• Decreases intake	• Decreases intake
	• Central and peripheral mechanisms	• Mediated via direct effects on AP
		• Intact vagus not required
Available knockout models	• Proglucagon peptides (Gcggfp/gfp)	• Amylin deficient mice
	• GLP-1R−/−	• No specific receptor knockouts
	• Double incretin receptor knockout (GLP-1R−/− and GIP-R−/−)	• RAMPS (only) have been deleted or over-expressed
		• Calcitonin receptor knockouts
Future therapeutic approaches in metabolic disease	• Analogueing and improved delivery	• Analogueing and improved delivery
	• Hormonal combinations	• Hormonal combinations (e.g. amylin/leptin synergy evident in preclinical and clinical studies)
	• Allosteric modulators
	• ‘Phybrids’/Chimeras	• ‘Phybrids’/Chimeras
Other disease areas of note	• Neurodegenerative diseases (Parkinson's, Alzheimer's)	• Neuropsychiatric diseases (anxiety, depression)

Insulinotropic and direct β-cell effects

Early in vitro studies demonstrate that GLP-1 can act directly upon pancreatic β-cells to stimulate insulin secretion. These mechanisms are likely invoked under pharmacological conditions with stable GLP-1R agonists; however, under normal physiological conditions whether endogenous GLP-1 from L-cells can access β-cells before degradation is often debated (Williams, 2009). Other intermediary mechanisms have been proposed. For example, the observation that hepatic portal GLP-1 increased activity of pancreatic vagal fibres (Nakabayashi et al., 1996) led to a series of studies that suggest that β-cell stimulation by portal GLP-1 is evoked by a neural reflex (shown to be non-muscarinic) that may be triggered in the liver (Balkan and Li, 2000). Central GLP-1 systems also activate pancreas-projecting preganglionic vagal motoneurons (Wan et al., 2007). Within the pancreas, the stimulatory effects involve multiple signal transduction events, including ion channel activity, intracellular Ca²⁺ handling, membrane docking and exocytosis of insulin-containing granules. In preclinical models, a number of reports also support the concept that GLP-1 agonism can increase β-cell mass via stimulation of β-cell neogenesis, stimulation of β-cell proliferation and suppression of β-cell apoptosis. These effects have all been extensively reviewed elsewhere (Gromada et al., 2004; Nielsen et al., 2004; Doyle and Egan, 2007). The effects of amylin have been explored in isolated β-cells, isolated pancreatic islets and perfused pancreas, and amylin has not been shown to possess insulinotropic properties. Amylin exerts insulinostatic effects in vitro, potentially mediated via an amylin-like receptor linked to Gi-mediated inhibition of cAMP in islets [for a historical/methodological perspective on the in vivo relevance of these studies see Young (2005d)].

Glucagon suppression

The glucagonostatic effects of both peptides are important from a therapeutic perspective as patients with diabetes have inappropriate elevations in fasting and postprandial glucagon. Although GLP-1 is clearly a potent inhibitor of glucagon, the direct mechanisms by which GLP-1 reduces glucagon secretion have not been well established. Whether GLP-1Rs are even expressed on α-cells is controversial. Receptors have been detected by some techniques, but not others, and surprisingly, GLP-1 application tends to increase glucagon secretion in ex vivo preparations [isolated rat and mouse α-cells; reviewed in Gromada and Rorsman (2004)]. The action of GLP-1 on glucagon secretion has been proposed as a paracrine mechanism, mediated by stimulated release of somatostatin that in turn suppresses exocytosis in the α-cell (Ding et al., 1997; de Heer et al., 2008). The potent insulinotropic actions of GLP-1 have been suggested as contributing to direct insulin-mediated suppression of glucagon (Unger and Orci, 2010). Like GLP-1, amylin's glucagonostatic properties also do not appear to be achieved by a direct effect of amylin upon α-cells as evidenced by a lack of effect in ex vivo (e.g. isolated perfused pancreas) or in vitro[e.g. isolated pancreatic islets Silvestre et al. (2001)] preparations. Rather, amylin's effects are most likely initially mediated via activation of hindbrain receptors, which in turn activate the vagus nerve, although this remains to be demonstrated experimentally. It is important to note that for both hormones, under conditions when blood glucose is low (e.g. fasting), their actions to inhibit glucagon are not present representing an important protective effect against developing hypoglycaemia (Gedulin et al., 1997; Nauck et al., 2002). This phenomenon is referred to as a ‘hypoglycaemic override mechanism’ and is a desirable feature from a therapeutic perspective.

Gastric emptying

Several groups have explored the mechanisms mediating the gastric emptying effects of GLP-1. The role/s of peripheral versus central GLP-1 and vagal transmission of the gastric emptying effects has received much attention. In rats, vagotomy decreased intestinal secretion of GLP-1 (Rocca and Brubaker, 1999) as well as its gastro-inhibitory effects (Holmes et al., 2009). A series of experiments characterized the actions of GLP-1 on gastric projecting dorsal motor nucleus of the vagus (Holmes et al., 2009). Using electrophysiology, subpopulations of gastric-projecting dorsal motor nucleus of the vagus neurons were found to be depolarized by exogenously applied GLP-1 and exenatide. Microinjection of GLP-1 or exenatide to the brainstem decreased gastric tone, and selective vagotomy studies implicated ipsilaterally projecting motoneurons in mediating these gastric effects. These effects appear to be transmitted via activation of non-cholinergic non-adrenergic inhibitory pathways rather than by inhibiting tonic excitatory cholinergic pathways (Holmes et al., 2009). As noted above, although intestinally secreted GLP-1 can cross the blood–brain barrier (Kastin et al., 2002), given its rapid degradation, this scenario is unlikely. Under normal physiological conditions, the proposed source of GLP-1 is from proximal GLP-1 synthesizing neurons within the NTS acting in a neurotransmitter-like manner upon the dorsal vagal complex. At higher pharmacological levels, GLP-1 and/or its more stable analogues could engage both central and peripheral mechanisms. In pigs, the effect of GLP-1 on gastric motility does not appear to require an intact vagus nerve suggesting there may be species differences (Nagell et al., 2006). In humans, clinical observations suggested the importance of an intact vagal pathway, as the actions of GLP-1 on other gastric functions such as decreasing gastric acid secretion were no longer evident in individuals that had undergone vagotomy due to duodenal ulcers (Wettergren et al., 1997). Clinical studies have gauged the contribution of endogenous GLP-1 on gastric emptying using the GLP-1R antagonist exendin(9–39). Unfortunately, owing to the use of different populations and tests meals, results to date have been contradictory, with reports of gastric emptying either being accelerated, slowed or unchanged (Salehi et al., 2008; Deane et al., 2010; Nicolaus et al., 2011; Witte et al., 2011). In contrast to GLP-1, amylin inhibition of gastric empting is thought to be primarily centrally mediated. First, aspiration of the AP abolishes amylin's effects on gastric emptying (Young, 2005c). More recently, amylin inhibition of gastric emptying was demonstrated in rats with a sub-diaphragmatic vagal deafferentation (e.g. in which vagal afferent fibres are destroyed but efferent fibres remain intact) supporting the hypothesis that amylin acts via hormonal mechanisms to stimulate the AP and transmits signals to the gut via efferent vagal fibres (Wickbom et al., 2008).

Food intake

The inhibitory effects of peripherally applied GLP-1 on food intake are attenuated in surgically vagotomized rats (Abbott et al., 2005) and chemically (capsaicin)-treated mice (Talsania et al., 2005). Careful pharmacological studies have compared peripheral routes of administration of GLP-1 on meal size in sham and vagotomized rats. Anorexigenic effects of intraperitoneal but not hepatic portal vein GLP-1 required intact vagal afferent signalling, but intravenous GLP-1 most likely acts centrally and not via activation of hepatic GLP-1 receptor systems (Ruttimann et al., 2009). The effects of centrally administered GLP-1 were abolished by perinatal ARC lesions [using monosodium glutamate (Tang-Christensen et al., 1998)]. Overall, CNS lesions have not yielded specific nuclei that are necessary for the expression of peripheral GLP-1's anorexigenic effects. For example, a recent series of studies investigated the role of two circumventricular organs, the AP and the SFO, both of which are known to possess GLP-1 binding sites (Orskov et al., 1996; Yamamoto et al., 2003), in mediating the metabolic and CNS-stimulating effects of GLP-1. Neither electrolytic lesion of the AP, the SFO or combined AP + SFO meaningfully altered the anorexigenic effects of peripherally applied exenatide, a GLP-1R agonist (Baraboi et al., 2010). To what extent GLP-1 or its receptor agonists need to cross the blood–brain barrier to exert any of their physiological effects is debated. For example, intraperitoneal injections of albumin-bound GLP-1 molecules, which are of sufficient size to be excluded from the blood–brain barrier, still decrease food intake and delay gastric empting and activate a similar pattern of neuronal activity as exenatide (Baggio et al., 2004). These findings point to a model whereby peripheral activation of GLP-1R-dependent vagal afferents is capable of activating CNS centres and transducing the effects of GLP-1 in the brain. Nevertheless, there may be important differences when considering pharmacological effects of agonism using a GLP-1 ‘backbone’ versus exenatide. For example, in contrast to GLP-1, the central but not peripheral anorexigenic properties of exenatide were shown to be insensitive to GLP-1R antagonism suggesting that there may be important differences between these ligands (Barrera et al., 2009). CNS GLP-1 systems have been implicated in mediating non-homeostatic influences on ingestive behaviour, most notably malaise (Thiele et al., 1997). GLP-1R agonism increases taste aversion learning and pica behaviour, and GLP-1R antagonism inhibits the formation of taste aversion in response to known emetic agents such as lithium chloride (Seeley et al., 2000). Separate populations of GLP-1 receptors may mediate the anorexigenic versus aversive effects of GLP-1 agonism (Kinzig et al., 2002) as intra-amygdala injections of GLP-1 produce taste aversion learning but not anorexigenic effects whereas brainstem injections of GLP-1 reduce food intake but do not condition aversions. To what extent the satiogenic and malaise effects can be dissociated is frequently debated (Tang-Christensen and Cowley, 2007). Clinical experience with GLP-1 agonism suggests that there is nausea that occurs during the early days of treatment and, in most individuals, decreases with time and/or gradual dose escalation.

The neural circuitry required for the expression of amylin's anorexigenic effects is well understood. As noted above, an intact AP is necessary for the expression of amylin's anorexigenic effects. Direct AP application of low doses of amylin inhibits food intake, whereas antagonism of AP receptors (with AC187) increases food intake and abrogates the anorexigenic effects of exogenous peripheral amylin (Mollet et al., 2004). Rats with a specific hepatic branch vagotomy (Lutz et al., 1995), total sub-diaphragmatic vagotomy (Lutz et al., 1994) or capsaicin-induced lesions of peripheral neural afferents that project to the brain (Lutz et al., 1998a) still respond to amylin implying that an intact vagus is not required for amylin's anorexigenic effects. Under normal physiological conditions vagal transmission and gastric emptying likely serve to enhance amylin's central effects. For example, amylin suppression of ‘sham’ feeding (e.g. a technique where liquid food drains from the stomach so that gastric food stimuli related to the accumulation of food in the stomach, intestinal food stimuli and post-absorptive food stimuli are absent or greatly reduced) requires higher doses compared with those required to decrease ‘real’ feeding (Asarian et al., 1998). An interesting preclinical difference between amylin versus GLP-1 agonism is noted in preclinical assays of malaise. Amylin is not associated with malaise (taste aversion or pica) or competing locomotor behaviours (e.g. hyperactivity or immobility would prevent food consumption) in rats or mice (Chance et al., 1992; Morley et al., 1997; Rushing et al., 2002; Mack et al., 2007). Nausea is the most common tolerability-related adverse event with pramlintide treatment; however, it is generally mild and transient and is dissociable from weight loss (Aronne et al., 2007). In a clinical study approximately 76% of patients who received pramlintide achieved a significant reduction in body weight but did not report any malaise (Hollander et al., 2004). This is another example, wherein amylin and GLP-1 seem to strongly activate similar brain regions and circuits (e.g. the AP a region known to be important for emesis), with different whole organism effects. Another interesting feature with amylin agonism relates to dietary selection. When amylin-treated diet induced obesity (DIO) rats were allowed to self-selecting calories from either a low- or high-fat diet, they consistently chose a larger percentage of calories from the low-fat diet relative to controls (Mack et al., 2007). In clinical studies, pramlintide has also been shown to reduce acute and sustained reductions in the intake of highly palatable high-fat, high-sugar foods at a ‘fast-food challenge’ (Smith et al., 2007) suggesting that some of these nuances of feeding behaviour observed in preclinical models may be recapitulated in the clinic. To what extent these characteristics are unique to amylin agonism warrants further investigation.

Insights from knockout models

GLP-1 receptor knockout

Proglucagon encodes multiple bioactive products (e.g. glucagon, GLP-1, GLP-2) precluding the generation of a GLP-1 specific knockout. Knock-in mice deficient in all peptides were generated by introduction of a green fluorescent protein cDNA (Gcg^gfp/gfp). Overall their phenotype was fairly mild with α-cell hyperplasia noted, somewhat lowered blood glucose concentrations and slight increases in body weight (Hayashi et al., 2009). Mice with a targeted disruption of the GLP-1R gene have also been created (GLP-1R^−/− (Scrocchi et al., 1996) and reviewed in (Hansotia and Drucker, 2005). Islets isolated from these mice exhibit abnormal basal and glucose-stimulated cytosolic Ca⁺ accumulation (Flamez et al., 1999). Histological irregularities were noted in islet cell size and number as well as in the distribution of α-cells (Ling et al., 2001). The phenotype of GLP-1R^−/− mice includes modest gluco-regulatory perturbations such as mild fasting hyperglycaemia, glucose intolerance in response to oral and intraperitoneal glucose tolerance tests and diminished glucose-stimulated insulin secretion. Given the glucagonostatic effects of GLP-1 it was surprising that fasting and postprandial glucagon concentrations were normal in the knockouts. With respect to energy balance, GLP-1R^−/− mice maintain normal body weight and food intake, even in aged mice maintained on a high-fat diet (Scrocchi and Drucker, 1998). Intriguingly, GLP-1R^−/− mice are protected from high-fat diet induced muscle and hepatic insulin resistance (Ayala et al., 2010). Central or peripheral leptin responsiveness of GLP-1R^−/− mice was also explored and shown to be similar to that in WT controls (Scrocchi et al., 1997). Because the GLP-1R knockout phenotype was so modest, it was speculated that endogenous gastric inhibitory polypeptide (GIP; the other incretin) was compensating for these effects and double GLP-1/GIP receptor knockouts were created. Although oral glucose tolerance was somewhat worsened in GLP-1/GIP double knockout mice further perturbations in food intake, body weight, insulin-induced hypoglycaemia and glucagon effects were not noted (Hansotia et al., 2004).

Amylin knockout (AmyKO)

With respect to amylin, the complexity of multiple RAMPs and calcitonin receptors makes the exploration of specific receptor knockouts difficult. Global deletion of the calcitonin receptor has revealed physiological roles for the receptor in bone biology (e.g. to protect against hypercalcaemia), but no effects on body weight or metabolic disease-related endpoints were noted (Davey et al., 2008; Turner et al., 2011). Individual RAMPs have been disrupted revealing a range of phenotypes. Mice with a disrupted RAMP1 gene were hypertensive and exhibited a dysregulated immune response, removal of RAMP2 was lethal and RAMP3 knockout mice appear normal until old age when they were not as heavy as their wild-type littermates (Sexton et al., 2009). An interesting recent report demonstrated that mice with over-expression of RAMP1 had increased energy expenditure as well as enhanced responsiveness to amylin (Zhang et al., 2011). The role of endogenous amylin, however, was originally investigated in peptide deficient (AmyKO) mice originally created on a 129Ola/B6 background (Gebre-Medhin et al., 1998). Basal insulin and glucose concentrations were unchanged in AmyKO mice relative to wild types throughout development. If anything, the knockouts demonstrated slight gender-specific effects to increase insulin secretion and glucose tolerance. The lack of endogenous amylin was noted to worsen alloxan-induced diabetes pointing to a potential benefit in situations of β-cell damage (Mulder et al., 2000). In terms of weight regulation, modest, transient elevations in weight gain were noted especially in AmyKO males maintained on standard low fat diet (Gebre-Medhin et al., 1998). To further gauge the impact of amylin insufficiency on energy balance the mixed background AmyKO mice were backcrossed onto the more commonly used C57/B6 background, which readily develops high-fat diet-induced obesity (Turek et al., 2010). With the exception of a temporary increase in adiposity in AmyKO females, no differences in body weight, body composition or plasma leptin concentrations were observed when mice were maintained on a low-fat diet (6 weeks). AmyKO mice did not have an increased propensity to develop obesity after being switched to a high-fat diet (27 weeks). However, by contrast to GLP-1R^−/− mice, marked differences were noted in response to exogenous leptin challenges (Turek et al., 2010). Lean AmyKO exhibited diminished central leptin signalling in response to peripheral leptin injections. Leptin-induced phosphorylated signal transducer and activator of transcription levels in the hypothalamus and hindbrain only achieved 60% the levels of stimulation observed in wild-type controls. While initially effective, the weight-reducing effects of exogenous leptin were ultimately less durable (in female AmyKO) or completely absent (in male AmyKO) over the course of a 4 week peripheral infusion. AmyKO mice also had a 50% reduction in mRNA for the long form of the leptin receptor (Turek et al., 2010), and amylin has been recently proposed to play a role as a trophic factor (Potes and Lutz, 2010).

The relatively modest phenotypes observed in the GLP-1R^−/− and AmyKO models may also be attributable to the reliance on germ line knockouts in which multiple compensatory mechanisms may be engaged during development. For a variety of reasons (e.g. difficulty in measuring food intake, frequency of meal-taking and responses to food deprivation) mice may also not represent the optimal species for discerning the role/s of these signals in energy homeostasis. The importance of amylin and GLP-1 in normal physiology and their relevance as therapeutic classes warrants the development of inducible (and potentially tissue-specific) knockouts and/or developing knockouts in other species (e.g. rats).

Human imaging studies

Complementary to the immunohistochemical neurobiological assessments of CNS pathways in rodents clarifying differences between GLP-1R and amylin agonism is the emerging application of neuroimaging data in humans. The effects of i.v. GLP-1(7–36) infusion on cerebral glucose metabolism were evaluated using positron emission tomography. Peripheral administration reduced carbohydrate metabolism within discrete regions of the brain, most notably in the hypothalamus and brainstem (Alvarez et al., 2005). To assess whether postprandial changes in GLP-1 modulate neuronal activity, another study performed a post hoc correlation analysis of changes in H₂¹⁵O-positron emission tomography measurements of regional cerebral blood flow relative to changes in plasma concentrations of GLP-1 after a meal (Pannacciulli et al., 2007). The peak postprandial increases in plasma GLP-1 concentrations were associated with increases in rCBF in the left dorsolateral prefrontal cortex and the hypothalamus, both before and after adjustment for sex, age, body fat, and changes in plasma glucose, insulin, and serum free fatty acid concentrations. Although there are clear methodological and interpretational caveats associated with both techniques and study design (spatial resolution, contrast resolution of individual subtraction images, covariate analyses, etc.), the observation that the former study was associated with changes in hindbrain but not cortical regions whereas the latter implicated hypothalamic and cortical regions but not hindbrain points to regions that may be activated by pharmacological versus endogenous (meal-stimulated) GLP-1. In response to food cues (image viewing), individuals with type 2 diabetes mellitus have also been shown to exhibit enhanced activation of dopaminergic reward and motivation centres without activation of cholinergic self-awareness centres. A recent report suggests that exenatide can shift these responses to be more akin to that observed in non-diabetic individuals (Nathan et al., 2010).

To our knowledge, although comparable data do not yet exist for amylin agonism as a monotherapy, a functional magnetic resonance imaging study recently compared activation following the administration of metreleptin with the combination of metreleptin + pramlintide. In this study, pramlintide significantly potentiated leptin activation in mesolimbic brain regions such as the hippocampus and amygdala (Klopfenstein et al., 2010). To what extent these findings explain the established weight loss synergy between amylin and leptin agonism [discussed below (Roth et al., 2008b; Trevaskis et al., 2008)] remains to be determined. Functional neuroimaging in humans is clearly an area that warrants further research. Not only do these kinds of studies provide rich translational insights into the neurobiological substrates activated by these neuropeptides in humans but they may also serve as an important tool in identifying key pathways to target with combination-based therapeutics.

GLP-1/amylin agonism and additivity/synergy

The therapeutic potential of neurohormonal combinations has garnered significant recent attention. In this section we briefly review additive and synergistic interactions that have been described between other neurohormonal signals with GLP-1 or amylin, as well as interactions between GLP-1 and amylin.

Neurohormonal interactions with GLP-1 agonism

The marked anti-diabetic and weight-lowering effects of bariatric Roux-en-Y surgery have been associated with elevations in endogenous peptide YY(3–36) [PYY(3–36)] and GLP-1. As such, the combined effects of these hormones or their agonists have been studied by several groups. In fasted lean mice, single-dose combination studies revealed that co-injection of PYY(3–36) with exenatide exerted greater and more durable anorexigenic and gastric emptying effects relative to either agent alone over an 8 h period (Talsania et al., 2005). Short-term additive anorexigenic effects have been reported with this combination in lean and genetically obese mice, as well as in a clinical study in lean, fasted volunteers (Neary et al., 2005). However, the anti-obesity potential of this combination remains to be demonstrated as repeated injection of the combination failed to suppress weight gain in high-fat fed mice (Talsania et al., 2005). Chronic intermittent administration of exenatide and PYY(3–36) at doses likely to reproduce plasma concentrations of GLP-1 and PYY(3–36) observed in Roux-En-Y gastric bypass patients also failed to produce a more prolonged reduction in daily food intake and body weight in DIO rats (Reidelberger et al., 2011). Several groups have explored interactions between GLP-1 and leptin with mixed results. In lean rats, leptin pretreatment (at a low dose that had no effects when given alone) substantially increased the anorexigenic effects of GLP-1 or exenatide (Williams et al., 2006). Counter to these behavioural effects was the surprising observation that leptin pretreatment blocked exenatide-induced c-Fos activation in several hindbrain regions pointing to the need for further mechanistic investigation/s. Another group tested multiple dose combinations of both agents in lean rats and observed transient additive anorexigenic effects noted only when exenatide was combined with a low dose of leptin to suppress 24 h food intake. Upon repeated administration (5 days) no significant effects beyond monotherapy were noted in lean rats (Bojanowska and Nowak, 2007). In our lab, in a single dose combination study in DIO leptin resistant rats, the exenatide analogue AC3174 exerted less than additive anorexigenic and weight lowering effects when combined with leptin (Roth et al., 2008b). As discussed below, the same dose of leptin was shown to markedly and repeatedly synergize with amylin. Two short-term clinical studies have compared the individual and combined effects of GLP-1 and cholecystokinin infusion in lean healthy individuals and noted less than additive interactions to suppress acute food intake (Gutzwiller et al., 2004; Brennan et al., 2005).

Neurohormonal interactions with amylin agonism

Multiple combination studies have been conducted with amylin and other hormones including CCK, PYY(3–36) and leptin. Several of these studies used rigorous tests for synergy such as isobolographic analysis or response surface methodology (RSM) (Roth et al., 2010). Using isobolographic analysis amylin was shown to synergistically suppress acute food intake in lean mice when combined with cholecystokinin (Bhavsar et al., 1998). With sustained administration, durable anorexigenic and weight-lowering effects were demonstrated in leptin resistant DIO rats (Trevaskis et al., 2010b). Neurobiology studies implicated cooperative activation of the AP particularly in noradrenergic neurons (Trevaskis et al., 2010b). Amylin and PYY(3–36) exerted complementary effects to suppress food intake for up to 24 h in high-fat fed rats whereas the effects of the individual agents were no longer evident after 5 h (Roth et al., 2007). Likewise, in DIO mice a single dose combination suggested mathematical additivity. An RSM study in DIO rats revealed that this combination inhibited food intake synergistically and reduced body weight additively (Roth et al., 2007). As with amylin/cholecystokinin, co-activation of AP neurons may be an important feature of amylin/PYY(3–36) additivity (Potes and Lutz, 2010). Finally, a clinically relevant and reproducible synergy has been noted with combined amylin and leptin agonism. Central administration of leptin was shown to increase the acute eating-inhibitory effect of peripheral amylin in lean rats (Osto et al., 2007). In co-administration studies we found that doses of peripherally administered leptin that had no appreciable effect on body weight in DIO leptin resistant rats enhanced amylin-mediated weight loss. An RSM study confirmed that the combination synergistically decreased both food intake and body weight (Trevaskis et al., 2008). Although the combination of leptin with other anorexigenic peptides such as PYY(3–36) and a GLP-1 analogue also resulted in significant weight loss, neither combination induced the same extent of fat-specific weight loss as the combination of amylin and leptin. Several mechanisms of amylin and leptin synergy have been elucidated (Roth et al., 2008b; Trevaskis et al., 2008; Turek et al., 2010) and recently reviewed (Trevaskis et al., 2010a). In brief, amylin appears to restore and/or augments central leptin responsiveness in selected hypothalamic regions (most notably the ventromedial hypothalamus) by increasing leptin receptor number and signalling capacity. Another lab recently examined the cardiovascular effects of combined amylin and leptin treatment in lean and obese rats. In addition to decreasing body weight and adiposity the combination lowered blood pressure, and prevented bradycardia and metabolic suppression typically observed with negative energy balance (Seth et al., 2010). The clinical significance and potential of amylin/leptin synergy was established in a translational clinical research study in which obese/overweight individuals treated with pramlintide/metreleptin induced significantly greater weight loss (12.7%) than treatment with pramlintide or metreleptin alone (Roth et al., 2008b; Ravussin et al., 2009).

Amylin-dependent actions of GLP-1?

As amylin and insulin are co-localized and co-secreted, GLP-1 can also be regarded as having ‘amylinotropic’ properties. To understand the contribution of endogenous amylin to the effects of GLP-1 recent elegant clinical investigations compared the effects of IV infusion of a low and high dose of GLP-1 (that achieved plasma concentration of ∼100 pM or ∼170 pM, respectively) in healthy individuals and individuals with type 1 diabetes mellitus. Because endogenous amylin is present in healthy individuals and absent in individuals with type 1 diabetes mellitus differences in GLP-1 modulation of gastric emptying, food intake and glucagon was attributed by the authors to the presence or absence of endogenous amylin (Asmar et al., 2010). In these studies, GLP-1 exerted similar effects on gastric emptying, food intake and appetite in both subject populations implying that GLP-1 can exert these effects in an amylin-independent manner. From a therapeutic perspective this supports the notion that these systems are not completely overlapping and points to the potential for combined GLP-1 and amylin agonism (discussed below).

GLP-1 and amylin combinations

Although both GLP-1 and amylin are released in response to a meal, and exert gluco- and food-intake regulatory effects, as discussed above, they do so through different mechanisms and sites of action. If GLP-1 and amylin pathways were truly overlapping (and redundant) then simultaneous agonism of both pathways should not achieve greater effects than maximal agonism of either pathway alone. On the other hand, if their pathways were complementary, it might be possible to achieve greater effects than that obtained by either pathway alone. Bello et al. (Bello et al., 2010) tested this hypothesis in a nonhuman primate model by co-administering the amylin agonist salmon calcitonin alone and in combination with exenatide. Monkeys were maintained on a schedule of 6 h daily access to food and multiple dose combinations were used, which enabled the authors to use RSM to formally test for statistical additivity or synergy for suppression of food intake. The response surface indicated synergistic anorexigenic effects on hours 1–4 and additive effects during hours 5–6. Importantly, the monkeys did not display outward signs of malaise or nausea. Although the only dependent measure in this study was acute food intake, these non-human primate findings raise the possibility for additional gluco- and weight-regulation with combined GLP-1 and amylin agonism in the clinic.

Future therapeutic opportunities and considerations

Improved delivery

Glucagon-like peptide 1 receptor and amylin agonists are each currently dosed as QD/TID injection regimens. From a patient convenience perspective, this injection burden, while potentially adding efficacy, is clearly not attractive. The GLP-1 class is at a more advanced stage relative to amylin agonism and GLP-1 agents with weekly to monthly injection intervals at various stages of development (Baggio et al., 2008; Malone et al., 2009; Christensen and Knop, 2010; Madsbad et al., 2011). These technologies rely either on formulation- (e.g. microspheres) or scaffold- (e.g. albumin binding or conjugation, Fc fusions or conjugates) based approaches to prolong the pharmacokinetic half-life of the peptide. Oral (Marino et al., 2010; Steinert et al., 2010) or inhaled (e.g. fumaryl diketopiperazine microspheres capable of carrying small peptides deep into the lungs (Marino et al., 2010) preparations of GLP-1R agonists are being vigorously pursued but are still at very early stages.

Increase efficacy/potency

A classic drug development approach to improve upon these molecules would be to increase either the efficacy or potency of the molecule at its cognate receptor so that either fewer overall injections and/or less drug product would be required. As with many endogenously circulating peptides, GLP-1 already binds to the GLP-1R receptor with high affinity. However, there have been several studies aimed at elucidating the peptide–receptor interaction via both mutational and structural analyses in an effort to provide information to support rationale design of improved molecules. These have revealed that the N-terminal region is important for receptor activation while the C-terminal region has a greater contribution to binding. ((Siegel et al., 1999; Xiao et al., 2001; Green et al., 2004). Stabilization of the α-helical regions in the N- and C-terminus using lactam bridges to create bi-cyclic GLP-1 analogues have been shown to enhance potency as well as render the molecule more resistant to degradation by the neutral endopeptidase NEP 24.11 (Murage et al., 2008; 2010). Efforts have also been made to shorten the peptide from the native 30 amino acids of GLP-1. A novel class of 11 mer peptides consisting of a structurally modified 9 mer derived from the N-terminal region of GLP-1 linked to a substituted C-terminal biphenylalanine dipeptide exhibited enhanced pharmacokinetic profiles compared with GLP-1 (Mapelli et al., 2009). However, these molecules will likely require further optimization in terms of their in vivo potency to be considered clinical candidates. Finally, efforts have attempted to extend half-life by synthesizing ester-based prodrug molecules that convert at extended rates to the parent GLP-1 analogue under physiological conditions due to their inherent chemical instability (De and DiMarchi, 2010). These efforts were successful at extending in vitro half-life (as measured by chemical stability) but it remains to be demonstrated whether these advantages translate in vivo.

With respect to amylin agonism increasing the efficacy or potency of the molecule at its cognate receptor is challenging for two reasons. First, as with GLP-1, amylin already binds its receptors in the pM range. Moreover, as discussed above, the amylin receptor complex is multidimensional (CTR + RAMP), and is not adequately captured in in vitro cell systems nor is the relative importance in vivo of various CTR + RAMP complexes involved in amylin's energy balance effects understood. Recently, we reviewed strategies towards developing improved amylin analogues (Roth et al., 2008a). The helical region and the N-terminus of amylin were felt to represent the most promising regions for modifications. Our modified analogues were tested for binding in rat nucleus accumbens and decisions around selection of optimized analogues relied on in vivo efficacy and duration. This approach ultimately led to the development of a novel amylinomimetic, davalintide (AC2307). Davalintide was shown to bind with high affinity to amylin, calcitonin and calcitonin gene-related peptide receptors, yet suppress food intake for up to 23 h (compared with 6 h for native amylin administered at an equimolar dose (Mack et al., 2010). Davalintide stimulated c-Fos expression studies as well as the dependence of an intact AP to observe anorexigenic effects suggested that the metabolic effects of davalintide and amylin are transduced via similar neuronal pathways. These findings suggest that with respect to amylin agonism, although difficult to extensively examine in vitro, it may be possible to design amylinomimetics with a longer duration of action that would require fewer daily injections.

Small molecule agonists/modulators

Although this review has focused on peptide-based therapeutics, small molecules with good bioavailability after oral administration remain attractive, patient-friendly options. Recent structural studies using photo-labelled probes confirmed that the amino terminal region of GLP-1 interacts with the amino terminal region of the receptor adjacent to the transmembrane domain as well as the 1st extracellular loop defining a pocket that could be targeted by small molecule agonists in the future (Chen et al., 2010). Another approach that has been considered is to design allosteric modulators (e.g. molecules that bind at a site distinct from the endogenous ligand and that can impact ligand binding and/or modulate signalling pathways) of the GLP-1R (Christopoulos and Kenakin, 2002). Several of these molecules have been developed (Teng et al., 2007; Irwin et al., 2010; Sloop et al., 2010). In theory these molecules can either be used alone or in combination with peptide ligands. An exciting aspect of some of these compounds is that they can selectively modulate various signalling pathways in a peptide-agonist-dependent manner highlighting the potential to develop molecules for distinct clinical efficacy (Koole et al., 2010).

Phybrids/chimeras

The development path for a combination product faces several hurdles, both scientific and practical. Clinical development costs and complexity are increased by the need to compare the efficacy and safety of fixed dose combinations to the individual comparator arms. Further compounding these expenses is the likelihood that a combination product will require two separate manufacturing processes. In the case of injectable peptides, depending upon their compatibility for co-formulation, combinations will also likely require multiple daily injections. One way to mitigate some of these hurdles is to use chemical ligation strategies to link two molecules together into a single, novel chemical entity. Such peptide hybrids (‘phybrids’) would act as dual agonists by activating distinct receptors to induce separate but complementary pharmacological effects. Based on our observations that amylin and PYY(3–36) agonism exerted anorexigenic synergy and weight loss additivity (Roth et al., 2007), we investigated whether this could be replicated in a proof-of-concept amylinomimetic : PYY(3–36) phybrid molecule. We demonstrated that single administration of the two parent molecules each elicited ∼5% body weight loss in DIO rats over 1 week, their co-infusion induced ∼12% weight loss, and the phybrid albeit at a somewhat higher dose, was able to approximate the same degree of weight loss (Roth et al., 2010). The phybrid approach has also been applied to combined GLP-1 and amylin agonism (Mack et al., 2008). When a single peptide entity composed of a GLP-1 agonist covalently attached to an amylinomimetic was assessed in vitro, the GLP-1R : amylinomimetic phybrid stimulated cAMP through both the GLP-1 and the amylin/calcitonin receptors confirming each domain was functionally active, albeit with decreased potency for the CT receptor compared with the amylinomimetic parent. Twenty-eight days of SC infusion of the phybrid to Lep^ob/Lep^ob mice reduced haemoglobin A1c similar to the GLP-1R agonist parent component of the phybrid and in DIO rats reduced body weight comparably to the sum of the combination of the GLP-1R and amylinomimetic parent portions of the molecule, providing proof-of-concept for dual pharmacological action and anti-diabetic and anti-obesity properties in rodents. Another platform for combinations as single chemical entities involves co-agonist ‘chimeric’ molecules in which a single molecule can serve as an agonist at two (or more) distinct receptors. The chimeric platform is most applicable to partner molecules with sequence similarities. Recent examples include classes of compounds that agonize both GLP-1 and glucagon receptors. These molecules retained their in vitro potencies and improved glucose tolerance and reversed obesity in preclinical models (Day et al., 2009; Pocai et al., 2009). Future clinical studies on these novel classes of molecules are warranted.

GLP-1R and amylin agonism in other disease areas

In addition to metabolic disease, it is important to briefly note that both GLP-1R and amylin agonism may hold utility in other disease areas. The observations that GLP-1 agonism may play a role in pancreatic protection and regeneration coupled with the presence of GLP-1Rs in CNS regions implicated in neurodegenerative diseases (substantia nigra, sub-ventricular zone, etc.) has led to the exploration and potential capitalization of GLP-1R stimulation in neuroprotection. The most mature of these areas of research involves salutary effects of GLP-1R agonism in multiple preclinical models of Parkinson's disease (e.g. 6-hydroxydopamine, 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine, lipopolysaccharide-induced) (Bertilsson et al., 2008; Harkavyi et al., 2008; Li et al., 2009). Improvements are noted on a behavioural, biochemical and histological level and even link GLP-1R agonism to potential neurogenic effects [for recent reviews and discussion see (Harkavyi and Whitton, 2010; Perry and Greig, 2005)]. A similar body of research is emerging on beneficial effects of GLP-1 agonism in preclinical models of Alzheimer's disease (Li et al., 2010; Hamilton et al., 2011). The possibility that the pathology underlying neurodegenerative diseases is linked to common pathways and/or mechanisms has been raised given the increased link between the incidence of some of these neurodegenerative diseases in individuals with diabetes (Holst et al., 2011).

Preclinical findings suggest that amylin agonism may hold utility in the treatment of neuropsychiatric disease. Amylin exerted anxiolytic and antidepressive effects in several preclinical assays (reviewed in Roth et al. (2009)). In a chronic stress model, the effects of amylin administration were assessed in rats that were given access to standard lab chow and sucrose solutions and were also exposed to daily restraint stress. Vehicle-treated rats increased their consumption of sucrose and increased visceral adiposity. By contrast, amylin-treated rats decreased their stress-induced sucrose consumption following restraint stress, while maintaining their intake of standard lab chow (Roth et al., 2009). We also recently noted antidepressive and neurogenic potential of amylin agonism in estrogen-deficient animals. These studies were originally designed to explore the weight-lowering and metabolic effects of amylin in ovariectomized (OVX) and intact (SHAM) animals (Trevaskis et al., 2010c). To our surprise, amylin was approximately twofold more efficacious in reducing body weight in OVX rats compared with SHAM as well as OVX control rats that received estradiol replacement therapy. In exploring the underlying mechanisms we found that amylin restored neurogenesis in the hippocampus of OVX rats and increased (approximately twofold) neurogenesis within the AP. These alterations were not evident in SHAM rats treated with amylin. Amylin-treated OVX rats also displayed decreased immobility (a depressed phenotype) as compared with SHAM-operated controls in the forced swim test. These findings warrant further exploration, but are noteworthy, as estradiol concentrations decrease in postmenopausal women who make up a high percentage of the obese population and incidences of depression are reported to increasing during menopause. To what extent amylin agonism may hold utility in neurodegenerative diseases remains to be elucidated.

Conclusions

The research reviewed herein on GLP-1 and amylin agonism represents a poignant example of how our understanding of the physiological actions and interactions and therapeutic potential of islet- and gut-hormones has grown dramatically in the past decades. In step with our deeper mechanistic understanding of these neurohormonal signals is progress in protein and peptide chemistry, formulation and drug delivery. The continued convergence of basic science and drug delivery efforts to develop mono- and combination-based peptide therapeutics that leverage this complex yet elegant physiology will hopefully yield important additions to the armamentarium for the treatment of metabolic disease/s.

Glossary

AmyKO

amylin knockout

AP

area postrema

ARC

arcuate nucleus of the hypothalamus

DIO

diet induced obesity

DPP-4

dipeptidyl peptidase-4

GLP-1

glucagon-like peptide-1

GLP-1R

glucagon-like peptide 1 receptor

GLP-2

glucagon-like peptide-2

NTS

nucleus of the solitary tract

OVX

ovariectomized

PVN

paraventricular nucleus of the hypothalamus

PYY(3–36)

peptide YY(3–36)

RAMP

receptor activity modifying protein

RSM

response surface methodology

SFO

subfornical organ

Conflict of interest

Jonathan D. Roth, Mary R. Erickson, Steve Chen and David G. Parkes are all employed by Amylin Pharmaceuticals, Inc. and own stock in Amylin Pharmaceuticals, Inc.