Amylin: Pharmacology, Physiology, and Clinical Potential

A. Amylin and Related Peptides

Amylin, also known as islet/insulinoma amyloid polypeptide or diabetes-associated peptide, is a 37-amino-acid secreted hormone (Fig. 3), derived from an 89-amino-acid precursor (Westermark et al., 2011). The processing of the prohormone to yield amylin occurs via carboxypeptidase E and prohormone convertases 1/3 and 2 (Westermark et al., 2011). The gene encoding this peptide is found on chromosome 12 in humans (12p12.1). Two major post-translational modifications are required for full bioactivity; these are a C-terminal amide (Tyr37) and a disulfide bond formed between the two cysteine residues found at positions 2 and 7 (Fig. 3) (Young et al., 1996b). In terms of amino acid sequence, amylin is most closely related to calcitonin gene–related peptide (CGRP). CGRP also shares a similarly positioned disulfide bond and an amidated C terminus. This is also the case for calcitonin, adrenomedullin, and adrenomedullin 2. Together, these peptides form a small family, united by these characteristic features. Consequently, there is a degree of overlap in binding the cognate receptors for each peptide and pharmacological activity.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Amino acid sequences of amylin and related peptides. The disulfide bond between N-terminal cysteine residues is indicated as a dashed line. All peptides have a C-terminal amide (NH₂) and some are acetylated at the N terminus (Ac). Human amylin (hAmy) is colored blue and conserved residues between hAmy and other peptides are in bold. Rat amylin (rAmy) is yellow, salmon calcitonin (sCT) is green, and human calcitonin (hCT) is orange. Many peptides are chimeras, as indicated by the coloring.

Commonalities and differences in amino acid sequence and pharmacology between these peptides and across species have been used to develop agonists or antagonists that mimic or block amylin activity. The most notable of these is pramlintide, which is a relatively stable, bioactive synthetic analog of human amylin. Pramlintide is at least as potent as human amylin and is an amylinomimetic agent. It is also a 37-amino-acid polypeptide and differs in amino acid sequence from human amylin by replacement of amino acids with proline at positions 25 (alanine), 28 (serine), and 29 (serine) (Fig. 3). As a result of these substitutions, pramlintide is soluble, nonadhesive, and nonaggregating, thereby overcoming a number of the physicochemical liabilities of native human amylin (Janes et al., 1996; Young et al., 1996b). Pramlintide exhibits similar biologic activity to that of amylin. Additional peptides are described in section III.B as well as in Fig. 3 and Table 1.

B. Amylin Expression, Secretion, and Metabolism

The β-cells found within the islets of Langerhans of the pancreas are the most well known site for amylin production. The discovery of amylin was made here, in the form of amyloid fibrils observed in patients with type 2 diabetes and in diabetic cats (Westermark et al., 1986, 1987a,b; Cooper et al., 1987). The pathophysiological role of amylin-containing fibrils has received much attention and was recently reviewed (Westermark et al., 2011; Abedini and Schmidt, 2013). Therefore, these fibrils will not be considered further in this review. It is the activity of amylin as a physiologically important circulating hormone that warrants this review.

The isolated-perfused pancreas, primary islets, or β-cell models have shown that amylin secretion can be stimulated by many factors, including glucose, arginine, and fatty acids, and that amylin secretion tracks insulin secretion (Kanatsuka et al., 1989; Ogawa et al., 1990; Qi et al., 2010). Pancreatic β cells are the major source of circulating plasma amylin, which ranges from 3–5 pM in the fasting state to postprandial concentrations of 15–25 pM (Lutz, 2006; Boyle et al., 2011). Consequently, amylin is cosecreted with insulin in a consistent molar ratio of approximately 15:1 (insulin:amylin). In a study of healthy women, an oral glucose load produced a significant rise in plasma amylin within 15 minutes. Fasting decreased plasma amylin concentrations (Hwang et al., 2008). Consistent with the concept of an elegant and multifactorial hormonal control system that controls metabolism, the incretin glucagon-like peptide 1 (GLP-1) can increase plasma amylin concentrations in healthy subjects (Asmar et al., 2010). Meal-associated fluctuations of circulating amylin levels are thought to directly reflect changes in β-cell secretion; the contribution of other amylin secreting cells to circulating amylin levels is considered minor. Furthermore, fluctuations in β-cell secretion, like the increases observed during the postprandial period, are believed to elicit the physiologic effects of amylin on eating and energy homeostasis (Lutz, 2010). Changes in circulating concentrations of amylin in obesity are covered later in this review. It is unclear whether amylin synthesis and secretion (either at baseline or postprandially) are altered by age because this aspect has not been studied systematically. The same is true for amylin’s effects across different age ranges, but amylin’s effect on eating has been shown to be present in young and old rats (Lutz et al., 1994).

Additional sites of amylin gene expression are neuroendocrine cells of the mammalian stomach (including humans), the dorsal root ganglion (Gebre-Medhin et al., 1998b; Tingstedt et al., 1999; Zaki et al., 2002), and the developing kidney (Wookey et al., 1998). Detailed information about the function of amylin derived from these sources is not available. However, given the important effects of amylin on food intake and gastric emptying, the intriguing observation that amylin binding sites (see sections III and V) are also present in the stomach warrants further investigation.

Amylin and pramlintide both have a circulating plasma half-life of approximately 13 minutes in rats after an intravenous bolus injection (Young, 2005d). Pramlintide pharmacokinetics is discussed in more detail in section X.A. As with many secreted peptides, amylin (and pramlintide) is metabolized primarily via the kidneys, most likely by proteolytic degradation, and nephrectomized rats exhibit a marked decrease in plasma clearance of amylin (Leckström et al., 1997; Vine et al., 1998a). In vivo, both amylin and pramlintide are rapidly cleaved to their respective des-Lys metabolites. These N-terminally truncated metabolites are both fully active as amylin agonists, and they exhibit similar in vivo potency and efficacy to the full-length peptides (Center for Drug Evaluation and Research Approval Package for Application Number 21-332. Clinical Pharmacology and Biopharmaceutics Review. http://www.accessdata.fda.gov/drugsatfda_docs/nda/2005/21-332_Symlin%20Injection_biopharmr.pdf). Other amylin cleavage products are found in plasma, including amylin_17–37, which is likely to be inactive (Nakazato et al., 1990). It is not clear whether a strategy to inhibit mechanisms of amylin metabolism would enhance plasma amylin levels sufficiently as an alternative clinical intervention to exogenous amylin agonist therapy. Approaches to enhancing amylin agonist circulating half-life are covered in section VIII.D.

A. Molecular Composition

Amylin receptors belong to the large superfamily of cell surface G protein–coupled receptors (GPCRs) (Poyner et al., 2002). This family is further subdivided; amylin receptors reside in class B. A prominent feature of amylin receptors is their close relationship to calcitonin receptors. This is an important consideration when interpreting the pharmacology, expression, and physiologic relevance of amylin receptors.

Some of the earliest attempts to characterize amylin receptors focused on the observations that amylin could displace CGRP binding and activate CGRP receptors in liver membranes or other preparations that endogenously expressed CGRP receptors. Cross-reactivity of amylin with CGRP receptors was plausible because these two peptides share approximately 40% amino acid sequence identity. Much of this early work has now been disregarded because the concentrations of amylin needed to displace CGRP binding were often too high for this receptor to be physiologically relevant to amylin action. However, it may be important to reconsider some of this work in light of the now-known molecular identity of amylin receptors (see section III.B). For instance, a pivotal 1988 study describing salmon calcitonin–sensitive CGRP receptors, which we now know were probably amylin receptors, is an important example of the value of these earlier studies in understanding amylin receptor biology (Sexton et al., 1988).

To identify additional candidates for amylin receptors, radioligand binding assays have been developed to determine the quantity of specific amylin binding in different tissues (Bhogal et al., 1992; Beaumont et al., 1993). High levels of amylin binding are present in the lung, stomach fundus, spleen, and brain. To date, a similar study has not been conducted in human tissues.

Further investigation of amylin binding sites in the brain began to show overlap with calcitonin binding sites, and dense binding sites are detected in the nucleus accumbens, the area postrema (AP) of the brainstem, and the hypothalamus. Subsequent work showed that amylin could stimulate cAMP production in cells expressing calcitonin receptors and both amylin and calcitonin could be cross-linked to calcitonin receptors (Perry et al., 1997). A crucial observation was made that amylin did not interact with all cells that expressed calcitonin receptors and other factors were probably required for this to occur.

In 1999, two groups reported that the single transmembrane spanning receptor activity–modifying proteins (RAMPs) were the answer to this conundrum (Christopoulos et al., 1999; Muff et al., 1999). Although the calcitonin receptor does have some affinity for and can be activated by amylin, the presence of RAMPs can enhance the binding and potency of amylin to such a degree that postprandial concentrations of amylin that are in the picomolar range can produce functional responses (Tilakaratne et al., 2000). There are three RAMPs and several splice variants of the calcitonin receptor; consequently, there are many possible amylin receptor subtypes with potentially unique pharmacology, signaling, and regulation profiles. Another important consideration is that there are species differences in terms of pharmacology and receptor splice variants (Poyner et al., 2002). There has been no investigation as to the potential role of amylin receptors in the clearance and degradation of amylin, and as described in section III.D, it is not even clear whether amylin receptors are internalized. Furthermore, whether amylin degradation at the cellular level is also important as a mechanism to terminate receptor signaling is not known.

B. Pharmacology

Amylin receptors are heterodimers of the calcitonin receptor and a RAMP (Fig. 4; Table 2). The calcitonin receptor represents the core receptor protein that obtains high-affinity amylin binding properties by coexpression of one of the three RAMPs in the same cell (McLatchie et al., 1998; Christopoulos et al., 1999; Muff et al., 1999). RAMPs alter calcitonin receptor pharmacology from calcitonin-preferring to amylin-preferring receptors. The pharmacology of these receptors is summarized in Table 2. RAMPs can regulate the transport to the cell surface of other receptors with which they interact, but this is not reported to occur with the calcitonin receptor (Christopoulos et al., 2003; Hay et al., 2005). The precise site(s) of interaction between the calcitonin receptor and RAMPs is not yet known.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

A schematic diagram of the amylin receptors, which are formed by the interaction of the calcitonin receptor with RAMP1, RAMP2, or RAMP3 to generate the AMY₁, AMY₂, or AMY₃ receptors.

Numerous splice variants of the calcitonin receptor have been reported, which differ between species (Sexton et al., 1999). Only a few of these have been characterized with respect to amylin receptor pharmacology; however, on the basis of these observations, it is likely that their coexpression with RAMPs could also yield amylin receptors. The most common splice variant, and the one upon which most research has focused, is now known as CT_(a) (formerly CTR2, CTR_I1−), as ratified by the relevant Nomenclature Committee of the International Union of Pharmacology (Poyner et al., 2002). This is the insert negative form of the receptor that has equivalent receptors in humans and rodents. The pharmacology of calcitonin and amylin receptors is regularly updated as new literature becomes available via the joint initiative of the Nomenclature Committee of the International Union of Pharmacology and the British Journal of Pharmacology Guide to Receptors and Channels (http://www.guidetopharmacology.org).

A complication with the interpretation of many pharmacological studies of these receptors is that most of this work has likely used a mixed population of receptors, encompassing RAMP-coupled calcitonin receptors as well as calcitonin receptors alone. This means that although in binding assays human calcitonin has low affinity for [¹²⁵I]amylin binding sites, cells transfected with calcitonin receptors and RAMPs display potent human calcitonin functional responses. In some instances, the expression of RAMP3 with the calcitonin receptor results in a reduction in human calcitonin potency (Armour et al., 1999; Hay et al., 2005). The explanation for this is not clear but it could involve a modification to the conformation of the calcitonin receptor or a reduction in the number of “free” calcitonin receptors (Armour et al., 1999; Hay et al., 2005). Salmon calcitonin, on the other hand, has high affinity for [¹²⁵I]amylin binding sites.

The prevailing consensus is that the AMY₁ receptor is formed by calcitonin receptors expressed with RAMP1, generating a receptor with high affinity for amylin, salmon calcitonin, and CGRP (Poyner et al., 2002). In the case of the delta 47-splice variant of the calcitonin receptor, its expression with RAMPs results in a receptor that is activated with greater potency by CGRP than amylin (Qi et al., 2013). The formation of an AMY₂ (calcitonin receptor/RAMP2) receptor is highly dependent on cell background and calcitonin receptor splice variant (see section III.C on signaling). The AMY₃ receptor (calcitonin receptor/RAMP3) has high affinity for amylin and salmon calcitonin and lower affinity for CGRP. Although most studies have been conducted with rat amylin, a recent report shows that human amylin and pramlintide have very similar pharmacology to rat amylin at each receptor complex (Gingell et al., 2014).

The activity of CGRP at receptors that are classified as amylin receptors is intriguing and may underlie some of the difficulties in interpreting early studies of both CGRP and amylin pharmacology. For example, it has been suggested that the AMY₁ receptor could be a molecular explanation for reports of a “CGRP₂” receptor (Hay et al., 2008). A key question is what the physiologic relevance of each amylin receptor subtype is, and in particular what role the AMY₁ receptor might play in amylin and/or CGRP biology. Furthermore, salmon calcitonin is often used as an amylin agonist, but it is also a calcitonin receptor agonist; therefore, in in vivo studies, its effects may be mediated via both amylin and calcitonin receptors. Therefore, reference to salmon calcitonin derivatives as “amylinomimetics” should be treated with caution because it is not known that all activities are occurring directly through amylin receptors.

Antagonists are important tools for determining the contribution of amylin to a particular physiologic function or for determining the receptor via which amylin acts. Several peptides have been reported as amylin receptor antagonists, including salmon calcitonin 8-32 and AC187 (Fig. 3), and they are widely used in the literature (Table 1). Characterization of these antagonists at molecularly defined calcitonin and amylin receptor subtypes shows that they are able to antagonize both receptor classes and, therefore, should be used carefully. However, when used at appropriate concentrations to block the endogenous actions of amylin, their effects are likely to be specific to amylin receptors. However, these antagonists do not have sufficient subtype selectivity to be able to determine the specific amylin receptor that may be mediating a given effect (Hay et al., 2005; Bailey et al., 2012).

The mechanism of amylin binding to its receptors is likely to follow the general scheme of class B GPCRs in which the peptide C terminus engages with the receptor extracellular domain, whereas the disulfide ring at the N terminus triggers receptor activation through engagement with the transmembrane portion of the receptor (Archbold et al., 2011). However, little is known about the details of this for amylin specifically and what the relative role of calcitonin receptor or RAMP is in peptide engagement. Based on recent crystal structures of the CGRP and adrenomedullin receptors, it is anticipated that amylin will bind in the most part to the calcitonin receptor, with a small contribution from RAMP (Booe et al., 2015). Studies using RAMP1/RAMP2 chimeras and site-directed mutagenesis have clearly identified that the N terminus of RAMP is important for peptide interactions (Udawela et al., 2006a,b; Qi et al., 2008). A specific RAMP1 residue (tryptophan 84) may directly contribute to amylin and CGRP interactions with the AMY₁ receptor (Gingell et al., 2010; Booe et al., 2015). Some data for CGRP at the AMY₁ receptor show that a conserved threonine at position 6 in the peptide is essential for receptor activation (Hay et al., 2014). An alanine analog of amylin at this position had abolished efficacy in vivo, implying a similar effect at amylin receptors (Roth et al., 2008b).

C. Signaling

Most studies of amylin receptor signaling derive from transfected model cellular studies. These studies show that RAMPs can influence the ability of the calcitonin receptor to interact with different G proteins (Morfis et al., 2008). In terms of molecular mechanisms of amylin receptor signaling, the RAMP C terminus is important for defining G protein coupling (Udawela et al., 2006a,b, 2008). The calcitonin receptor splice variants also have different capacities to interact with intracellular signaling molecules (Poyner et al., 2002). The resulting complexity in amylin receptor signaling is substantial.

Consistent with amylin receptors containing the core GPCR, the calcitonin receptor, amylin results in acute activation of Gs-coupled pathways, resulting in an acute increase in intracellular cAMP. Gq coupling has also been reported and acute phosphorylation of extracellular signal–regulated kinase 1 and 2 (ERK1/2) is also evident, which may relate to protein kinase C rather than protein kinase A activation (Morfis et al., 2008).

However, many of these experiments are difficult to interpret because of the intrinsic ability of amylin to activate the calcitonin receptor. Therefore, mixed populations of free and RAMP-complexed calcitonin receptors, which are likely to exist in many experimental systems, may be activated simultaneously by amylin (depending on the concentration applied). Thus, the precise receptor complex through which amylin results in activation of a particular signaling pathway is often unknown. For phosphorylation of ERK1/2 and increases in intracellular Ca²⁺, amylin can certainly trigger these events but interestingly, the potency of amylin only increases slightly in the presence of RAMP when measuring these pathways compared with cAMP (Morfis et al., 2008; Qi et al., 2013). Nevertheless, the time course of rapid ERK1/2 phosphorylation by amylin in cells transfected with calcitonin and amylin receptors is consistent with its time course for activation in AP neurons (see section VI.B) and osteoclasts (Dacquin et al., 2004; Morfis et al., 2008; Potes et al., 2012; Qi et al., 2013). Activation of ERK1/2 is clearly a pathway that amylin can use, but more work is needed to determine the precise circumstances under which this occurs and the mechanism involved. There has been no specific investigation of whether amylin-induced ERK activation is G protein dependent or arrestin mediated; however, the calcitonin receptor can recruit arrestin in response to human and salmon calcitonin (Andreassen et al., 2014). Amylin was recently reported to either increase or decrease ERK and Akt phosphorylation in dispersed islets, depending on the amylin and glucose concentration (Visa et al., 2015).

cGMP has also been linked to amylin’s effects in the AP but has not been studied at defined amylin receptor subtypes in cell lines (Riediger et al., 2001). Conversely, cAMP has not been measured in response to amylin in AP neurons but is the prototypical pathway measured in transfected cell studies. Therefore, there is still a great deal of research needed to clarify which intracellular signals are triggered by amylin within its target cells, and which of these signals are necessary or sufficient to mediate amylin’s effects.

D. Regulation

GPCR signaling is carefully regulated to maintain cellular responsiveness to natural ligands. Acutely, this can be in the form of receptor phosphorylation, uncoupling from signaling proteins, and removal of the receptor from the cell surface (internalization), followed by recycling of the receptor back to the cell surface or degradation of the receptor. In the longer term, such as in the case of prolonged exposure of the receptor to an agonist drug, the synthesis of the receptor protein can be downregulated (Luttrell, 2006). The short- or long-term regulation of defined amylin receptor subtypes has not been studied. However, there is evidence that the calcitonin receptor can be regulated by these mechanisms in a cell- and ligand-dependent manner. The calcitonin receptor is also subject to downregulation, influencing the utility of calcitonin agonists as drugs for bone diseases (Tashjian et al., 1978; Bouizar et al., 1987). Although it is tempting to infer that amylin receptors will behave similarly, these processes are highly dependent on cellular context; thus, how the calcitonin receptor behaves in osteoclasts is unlikely to translate directly to other cell types, such as AP neurons. Importantly, although this has not been studied with the calcitonin receptor, RAMPs are known to alter receptor fate in terms of regulation; therefore, the amylin receptors could be regulated quite differently from the calcitonin receptor alone (Bomberger et al., 2005a,b). Continuous infusion studies of 3–7 days with osmotic minipumps in rats and mice show that although there is a gradual decline in effect on food intake over time, amylin responsiveness is still evident, suggesting that receptor responsiveness is maintained under these conditions and that fluctuations of circulating amylin levels will still trigger relevant effects (Isaksson et al., 2005; Roth et al., 2006; Mack et al., 2007; Boyle et al., 2011; Olsson et al., 2012). This is also apparent during conditions of prolonged hyperamylinemia (associated with hyperinsulinemia), such as type 2 diabetes and/or obesity, in which administration of exogenous amylin analogs remains effective to produce both glucoregulatory and weight-lowering actions (Boyle et al., 2011). There is little evidence that responsiveness to pramlintide wanes over time in clinical studies. In fact, pramlintide (two to three daily injections) consistently produces significant and durable actions to decrease body weight in obese patients for up to 12 months (Smith et al., 2008). Although these changes do plateau in the mid-single-digit percentage range, whether this reflects direct receptor desensitization and/or the emergence of counter-regulatory mechanisms to a weight/calorie-reduced state is difficult to dissect.

E. Expression and Physiologic Relevance of Amylin Receptor Subtypes

Although there are many potential amylin receptor subtypes, discerning which subtype is most relevant to each action of amylin is a considerable challenge. A range of factors, including a lack of receptor-selective pharmacological tools, the heteromeric nature of the receptors, their discrete neuronal expression, and a lack of reliable or widely available RAMP antibodies all contribute to this absence of information. It is also important to point out that RAMPs have several other GPCR partners, so their expression is only indicative of a potential interaction with the calcitonin receptor (McLatchie et al., 1998; Christopoulos et al., 2003; Harikumar et al., 2009; Lenhart et al., 2013).

Amylin’s actions are mediated by a membrane-bound receptor that is present in the AP, the presumed primary site for most amylin actions (see section VI). Amylin binds strongly to the AP, and calcitonin receptor mRNA and protein are reported to be expressed here (Beaumont et al., 1993; Sexton et al., 1994b; Nakamoto et al., 2000; Barth et al., 2004; Becskei et al., 2004). RAMP1, RAMP2, and RAMP3 are all stated to be present in the AP, but thus far only data from in situ hybridization studies are available (Ueda et al., 2001; Stachniak and Krukoff, 2003). Amylin-induced neuronal activation, as assessed by c-Fos mRNA, colocalizes with CT_(a) and RAMP3 mRNA in the rat AP (Sexton et al., 1994b; Christopoulos et al., 1995; Barth et al., 2004; Becskei et al., 2004). Laser capture microscopy will be useful for demonstrating that all necessary amylin receptor components are in fact expressed in single AP neurons or other types of neurons (Le Foll et al., 2015). This will also be a useful technique for determining which may be the most relevant amylin receptor subtype in different brain areas.

However, despite the evidence that all structural components of functional amylin receptors are present in the AP and that the AP is necessary for amylin action (see below), it still remains to be shown that the calcitonin receptor protein colocalizes with any specific RAMP in amylin-sensitive AP neurons. Furthermore, whether site-specific ablation of one of the critical components of the functional amylin receptor in AP neurons is sufficient to abolish amylin action has not been tested. Neural RAMP1 overexpression can augment amylin activity (see section V.C), suggesting that the AMY₁ receptor subtype could be particularly important. Calcitonin-sensitive CGRP binding sites in the AP and nucleus accumbens support this hypothesis (Sexton et al., 1988). However, the data are insufficient to make judgement calls about which receptor might be most effectively targeted with amylin agonists.

Other potential sites of relevance to amylin biology include the subfornical organ (SFO), nucleus of the solitary tract (NTS), ventral tegmental area (VTA), nucleus accumbens, and hypothalamus. Amylin binding has been reported in each of these brain areas, accompanied by calcitonin receptor protein and/or mRNA (Olgiati et al., 1983; Sexton et al., 1994a,b; Christopoulos et al., 1995; Hilton et al., 1995; Skofitsch et al., 1995; Nakamoto et al., 2000; Ueda et al., 2001; Stachniak and Krukoff, 2003; Barth et al., 2004; Becskei et al., 2004; Eftekhari and Edvinsson, 2011). The presence of RAMPs and thus functional amylin receptor subtypes is subject to the same caveats described above. Nevertheless, the SFO reportedly expresses mRNA for all three RAMPs, whereas the NTS expresses only low levels of RAMP2. Hypothalamus expression of RAMP1 and RAMP2 is evident in several nuclei. RAMP2 mRNA has been reported in the preoptic hypothalamic area, suprachiasmatic nucleus, paraventricular nucleus, periventricular nucleus, supraoptic nucleus, arcuate nucleus, ventromedial nucleus, and dorsomedial nucleus. RAMP1 has been reported in fewer hypothalamic nuclei, namely the arcuate nucleus, ventromedial nucleus, and dorsomedial nucleus (Oliver et al., 2001; Stachniak and Krukoff, 2003). Recently, all three RAMPs were detected in the ventromedial hypothalamus (VMH) (Le Foll et al., 2015). In the nucleus accumbens, only RAMP1 has been reported. The VTA was recently reported to express all necessary receptor components and is suggested to be another area of physiologic amylin relevance (Mietlicki-Baase et al., 2013). All of these are rodent studies. There is no information on the expression of RAMP2 or RAMP3 protein in the human brain and there are very limited data for RAMP1, as a consequence of it being a subunit of the CGRP receptor, which is a major target for treating migraine (Edvinsson and Warfvinge, 2013). There are some primate data for calcitonin receptor expression, and calcitonin receptor expression was recently shown in the human brainstem (Walker et al., 2015). This appeared to overlap with RAMP1 in several places; however, there was limited colocalization performed. Nevertheless, the coexpression of both proteins was found in some blood vessels and in the spinal trigeminal tract (Walker et al., 2015). This is the first report of AMY₁ receptor expression at the protein level in human tissues. Overall, the current data are unsatisfactory due to the challenges outlined in the opening paragraph of this section, and specific amylin receptor subtypes cannot be attributed to amylin activities at the present time.

A. Genetic Models of Amylin Deficiency or Overexpression

Amylin knockout animals have been developed to study the role of amylin in nutrient (in particular glucose) metabolism and bone metabolism (Gebre-Medhin et al., 1998a). These mice were established on a mixed 129Ola/B6 background in which the amylin coding sequence in exon 3 of the amylin gene was replaced by a neomycin resistance cassette. The specific phenotypes of these amylin-deficient mice that are not related to amylin’s role in the control of eating include a lower bone mass due to a higher number of osteoclasts, and reduced late phase nociception in the paw formalin test; the latter implies that endogenous amylin increases the perception of chemical pain (Table 3), despite findings from pharmacology studies reporting that amylin agonists can have analgesic properties (Huang et al., 2010). Furthermore, these mice had increased insulin secretion and a more rapid glucose elimination, which may indicate that endogenous amylin exerts an inhibitory tone on insulin secretion (Gebre-Medhin et al., 1998a).

In the mixed genetic background, male amylin knockout mice show a slightly higher rate of body weight gain compared with corresponding wild-type controls (Gebre-Medhin et al., 1998a). This knockout model has also been backcrossed onto the C57/Bl6 background strain more commonly used in metabolic disease studies. As reported by Olsson et al. (2012), body weight did not differ between knockout and wild-type animals over a 27-week study. This study also tracked food intake, meal number, and meal size over this period and did not report any differences between groups. In another study, body weight, body composition, and plasma leptin levels were compared in male and female wild-type and knockout mice maintained on either a low-fat or high-fat diet (Turek et al., 2010). With the exception of a modest increase in adiposity in female knockout mice maintained on the low-fat diet, there were no consequences of amylin gene deletion; these results imply that at least in the endogenous system, mice can compensate for an amylin deficit.

Interestingly, amylin knockout mice on the 129Ola/B6 background differed from control animals in their interaction with other factors controlling eating, indicating that endogenous amylin may have a facilitating role for other controls of eating, such as cholecystokinin (CCK) (Mollet et al., 2003a). Likewise, low-fat–fed amylin knockout mice on the C57/Bl6 background were found to be less responsive to exogenous administered leptin. Leptin-induced p-signal activator of transcription 3 (STAT3) signaling and leptin receptor mRNA expression were reduced in the ventromedial and arcuate hypothalamus of amylin knockout mice, and the weight-lowering effects of minipump-administered leptin were attenuated (Turek et al., 2010). Together, these studies suggest a modest role for endogenous amylin in the overall control of energy balance, most likely via an effect on eating in interaction with other satiating and adiposity signals (Lutz, 2010).

Numerous rodent models with overexpression of human amylin have been generated to investigate the role of islet amyloid deposition in the destruction of pancreatic β-cells and the pathophysiology of diabetes (Westermark et al., 2011). To our knowledge, however, no detailed studies are available on the effect of amylin overexpression on eating or body weight. Rats that were transgenic for the overexpression of human amylin gained weight comparably until 5 months of age but then had a slightly lower body weight than wild-type controls (Butler et al., 2004). The authors argued that lower body weight may have been a consequence of energy loss due to the development of glycosuria and diabetes in the transgenic rats, but no detailed data on food intake were provided. Hence, reduced eating, which in principle would be consistent with the effects of exogenous amylin on eating and body weight, cannot be excluded. The rather small difference between transgenic and control animals may have been due to receptor desensitization or a redundant system controlling body weight. Because the major difference in body weight between amylin knockout and wild-type mice occurred mainly within the first 3 to 4 months of life, the same may also hold true for animals overexpressing amylin. Another recent article with mice that were transgenic for the overexpression of human amylin did not report data on eating and body weight (Wong et al., 2008).

B. Genetic Models of Amylin Receptors

Given the heteromeric nature of endogenous amylin receptors, the phenotypes of both calcitonin receptor– and RAMP-deficient/overexpressing animals need to be considered. In neither case can the phenotype be considered specific to amylin; models of altered calcitonin receptor expression will likely exhibit phenotypes relevant to both calcitonin and amylin, whereas with RAMPs the phenotype could be relevant to many hormonal systems.

Several models of altered calcitonin receptor expression have been generated. Deletion of exons 6 and 7 of the calcitonin receptor gene calcr results in no viable offspring (Dacquin et al., 2004). However calcr^+/− mice were viable with expression of the receptor reduced by 2-fold by quantitative polymerase chain reaction. The genetic background of these mice was not clear. The phenotype reported was that of high bone mass due to increased bone formation, with normal bone resorption (Dacquin et al., 2004). These mice were cross-bred with amylin^+/− mice, but neither data on eating nor data on body weight were reported. Because of the nonidentical phenotypes of calcr^+/− and amylin-deficient mice, it was concluded that amylin may act through a receptor other than the calcitonin receptor. However, studies of this type are often very difficult to interpret due to the complicated receptor system and because sites of calcitonin receptor expression during development are not identical to those in adulthood, as evident from human calcitonin receptor/Lacz transgenic mice (Jagger et al., 2000). Subsequent studies have developed mice with 94%–100% deletion of the calcitonin receptor, using the Cre-LoxP system, targeting exons 13 and 14 or exons 6 and 7 (Davey et al., 2008; Keller et al., 2014). There were no differences in body weight in male or female mice compared with wild-type mice, although the ages at which these measurements were taken were not specified. The first attempt at osteoclast-specific deletion of the calcitonin receptor resulted in no viable offspring, using floxed calcitonin receptor mice crossed with TRACP-Cre mice (Davey et al., 2008). Later, however, a cross with cathepsin K-Cre transgenic mice resulted in viable mice. These mice showed increased bone resorption (Turner et al., 2011). Keller et al. (2014) also developed mice with osteoclast-specific disruption of calcr, which showed increased bone formation. Although these models give intriguing insight into the physiologic role of the calcitonin receptor in bone biology, none of these models have been carefully evaluated with respect to food intake, body weight, or sensitivity to diet-induced obesity.

Knockout models lacking each RAMP have been generated and their phenotypes were recently summarized (Kadmiel et al., 2012). Most work has focused on understanding their role in adrenomedullin or CGRP physiology. RAMP1 and RAMP3 global deletion leads to viable offspring, whereas loss of RAMP2 is lethal and therefore heterozygous mice are used to study its function. In all cases, there are no reports in relation to glucose homeostasis. Haploinsufficiency of RAMP2 results in defects of bone homeostasis (Kadmiel et al., 2011). Only for RAMP3 is a body weight phenotype reported, whereby there is an age-dependent lean phenotype at approximately 6 months of age, compared with wild-type mice. Food and water intake was apparently unaltered in these animals and the reason for the difference in body weight is not yet known (Dackor et al., 2007). Because of the lack of investigation into parameters relevant to amylin, no conclusions can be drawn as to the relative importance of each RAMP at this time.

A few RAMP overexpression models have also been generated. There is a global human RAMP1-overexpressing model and a neural-specific human RAMP1-overexpressing model (Zhang et al., 2007; Sabharwal et al., 2010). Characterization of the second of these does potentially provide insight into amylin physiology and we describe the relevant aspects in section V.C. There is also a model of RAMP2 overexpression in smooth muscle, but this is unlikely to have particular relevance to amylin (Tam et al., 2006). Overall, the extant literature on RAMP transgenics has included only a limited investigation of each model with respect to amylin physiology.

A. Amylin as a Satiation Signal

One of the most extensively investigated functions of amylin focuses on its role as a signal of satiation (Lutz, 2010). It is believed that amylin is a physiologic regulator of meal size (Geary, 2005; Lutz and Geary, 2008). Eating leads to a rapid increase in circulating amylin levels and exogenous amylin reduces eating within minutes after application. The effect of amylin in reducing meal size is dose dependent and the effect is effectively and specifically blocked by amylin antagonism; on the other hand, amylin antagonists alone increase eating by a meal size effect (Mollet et al., 2004). Importantly, amylin and its analogs reduce meal size without producing signs of conditioned taste aversion or visceral illness (Lutz et al., 1995b; Morley et al., 1997; Mack et al., 2007, 2010). Overall, the data indicate that amylin acts as a physiologic short-term satiating hormone in rats.

Reduced drinking that has been observed in some studies after the administration of exogenous amylin seems to be secondary to the reduction in eating. No evidence is available that would support the opposite conclusion (i.e., that amylin reduces eating secondary to an effect on water intake) (Lutz et al., 1994; Arnelo et al., 1996; Morley et al., 1997; Asarian et al., 1998).

B. Amylin as an Adiposity Signal

The concept of “homeostatic eating controls” classically distinguishes between adiposity or “tonic” signals that enhance the effect of satiation and other meal-associated or “episodic” signals; adiposity signals are secreted in proportion to body adiposity by adipose tissue (leptin) or the pancreas (insulin, amylin), and are thought to affect eating by enhancing the effects of satiation signals. For example, amylin, similar to leptin and insulin, enhances CCK’s satiating effect (Riedy et al., 1995; Barrachina et al., 1997; Bhavsar et al., 1998; Mollet et al., 2003a). Even though the concept of distinct satiation versus adiposity signals is an oversimplification of the physiologic situation, it helps to structure the signals involved in the control of energy balance. The situation is, however, complicated by the fact that some signals actually may be part of both categories; insulin is considered a classic adiposity signal (Woods, 2005), but it is also released during meals, and blockade of endogenous insulin by insulin antibodies increased meal size (Surina-Baumgartner et al., 1995).

Like insulin, a similar “dual role” may also apply to amylin; it is released during meals and a number of experiments suggest a role for amylin as an adiposity signal. For example, basal plasma levels of amylin are higher in obese rats than in lean rats (Pieber et al., 1994), and high-fat–fed obese rats have higher baseline amylin levels than age-matched lean controls (Boyle et al., 2010, 2011). However, not all studies showed such an effect (Gloy et al., 2010) and it is possible that extended periods of being overweight may be required (Boyle et al., 2011) for baseline amylin levels to increase after the onset of obesity. In general, amylin secretion from the β-cell correlates very tightly with insulin secretion.

More direct evidence for amylin being an adiposity signal is provided by experiments showing that chronic peripheral (Roth et al., 2006; Mack et al., 2007) or central (Rushing et al., 2001) amylin decreases body weight and fat gain, whereas amylin antagonists increase body adiposity (Rushing et al., 2001). Furthermore, centrally administered amylin levels seem to directly influence the target body weight of rats (Wielinga et al., 2010). Together, these data suggest that central amylin, like leptin or insulin, may encode the regulated level of body weight and hence may contribute to the relative constancy of body weight throughout adult life.

Even though the underlying mechanisms are not yet clear, amylin action shows interesting differences to the action of CCK, which may be a “pure” satiation signal. First, rats receiving a continuous infusion of CCK do not show sustained reductions in eating or body weight (Crawley and Beinfeld, 1983). Second, CCK given before each spontaneous meal effectively reduced meal sizes, but the effect seems to be totally compensated by an increase in meal frequency (West et al., 1984). Similar compensations do not occur during chronic amylin infusions (Lutz et al., 1995b; Arnelo et al., 1996), perhaps because of the tonic eating-inhibitory effect of amylin (Grabler and Lutz, 2004). It would be interesting to know whether tonic and episodic effects of amylin rely on similar or distinct mechanisms. Both, however, seem to require an intact AP (Lutz et al., 1998c, 2001).

C. Amylin and Energy Expenditure

In some of the studies mentioned above, chronic amylin reduced body fat more than in pair-fed controls (Roth et al., 2006), as also shown in other experiments (Isaksson et al., 2005). This indicated that amylin infusions most likely reduced eating and increased energy expenditure (Mack et al., 2007; Wielinga et al., 2007, 2010). Hence, amylin shares another similarity with leptin because it influences energy balance by an eating and an energy expenditure effect (Isaksson et al., 2005; Roth et al., 2006; Mack et al., 2007; Osaka et al., 2008). In our own studies, amylin increased energy expenditure acutely when given at low doses into the third cerebral ventricle or directly into the AP (Wielinga et al., 2008, 2010). Furthermore, the decrease in energy expenditure that is the expected metabolic reaction in weight-reduced rats is prevented by chronic central amylin infusion (Wielinga et al., 2007, 2010).

Some questions in respect to amylin’s effect on energy expenditure remain. First, it is not yet clear whether the effect is of physiologic relevance because the role of endogenous amylin in the control of energy expenditure has not been tested. Second, the site and neurotransmitter mechanisms of amylin’s actions on energy expenditure also remain to be determined. Finally, further studies are necessary to define the underlying intracellular processes and whether these are similar to the ones activated in respect to amylin’s effect on eating (e.g., pERK, cGMP).

Two recent studies provided interesting mechanistic input on amylin’s effect on energy metabolism (Zhang et al., 2011; Fernandes-Santos et al., 2013). In rodents, as well as in humans, total energy expenditure depends on the thermogenic activity of the brown adipose tissue (BAT); BAT activity is controlled by the sympathetic nervous system and heat production relies on the organ specific expression of uncoupling protein (UCP)-1. Although earlier studies failed to demonstrate a clear effect of amylin on the expression of UCP-1 (Roth et al., 2006, 2008c), recent studies indicate that BAT activity is increased by amylin, that this effect is mediated by the sympathetic nervous system, and that the effects are enhanced in transgenic mice that overexpress RAMP1 of the AMY₁ receptor complex (Zhang et al., 2011; Fernandes-Santos et al., 2013). In brief, the neuronal overexpression of human RAMP1 led to sensitization of mice to the metabolic effects of amylin; human RAMP1 transgenic mice were lighter, had less fat, had increased energy expenditure and body temperature, and showed at least temporary hypophagia compared with control animals (Zhang et al., 2011). The higher energy expenditure was mediated by an increased sympathetic tone in efferents subserving BAT, and this was associated with increased expression of the peroxisome proliferator–activated receptor γ coactivator 1a as well as UCP-1 and UCP-3 (Fernandes-Santos et al., 2013). Although these studies are presented in the context of amylin, the potential role of CGRP should also be considered in any study investigating RAMP1. This is because RAMP1 can form two high-affinity receptors for CGRP and CGRP knockout mice are reportedly protected against diet-induced obesity and have elevated core body temperature (Walker et al., 2010). It should be noted that CGRP can also decrease food intake (Lutz et al., 1998c; Dhillo et al., 2003).

A. Amylin Action in the Brainstem

Several brain areas have high-affinity binding sites for amylin, as shown by amylin receptor autoradiography studies (Sexton et al., 1994b). Amylin binding is particularly strong in the circumventricular organs such as the SFO and the AP. Amylin action in the SFO may be linked to an effect of amylin to increase drinking (Riediger et al., 1999; Fry et al., 2007) but whether it is necessary for the expression of amylin’s effect on eating has not been demonstrated.

In the context of amylin’s effect on energy metabolism, the AP has received the most attention, although the contribution of other brain areas such as the VTA should also be considered (see below; Mietlicki-Baase et al., 2013). A transporter-mediated process may facilitate the transfer of amylin through the blood–brain barrier into the brain, also outside of the circumventricular organs (Banks et al., 1995; Banks and Kastin, 1998). Of note, the primary receptor sites for the eating-inhibitory effect of centrally administered amylin have thus far not been studied in great detail.

A large number of independent studies indicate that the AP plays an important role in mediating effects of peripheral amylin. Amylin appears to activate AP neurons by direct humoral action, whereas vagal or nonvagal afferents do not seem to be involved (Lutz et al., 1994, 1995a, 1998a,c, 2001; Morley et al., 1994; Mack et al., 2010). Furthermore, local amylin administration into the AP recapitulates peripheral amylin’s effect, whereas the amylin antagonist AC187 has the opposite effect; AC187 also abolished the eating-inhibitory effect of peripheral amylin (Mollet et al., 2004). The AP also seems to be necessary for amylin's action to inhibit gastric emptying (Young et al., 1995; Gedulin et al., 1997; Wickbom et al., 2008; Mack et al., 2010; Young, 2005b). Electrophysiological and immunohistochemical studies supported a direct influence of amylin on the AP (Riediger et al., 2001, 2002, 2004; Potes et al., 2010b, 2012). As discussed in section III of this article, the receptor components (calcitonin receptor, RAMPs) necessary for amylin function appear to be expressed in the AP. The available evidence is not sufficient to specify which AMY receptor subtype is particularly important in the AP, or indeed in any part of the brain or other tissue.

B. Brainstem Mechanisms Mediating Amylin Signaling

A frequent marker to define stimulus-activated brain areas is based on the expression of the immediate early gene product c-Fos (Rowland et al., 1997; Becskei et al., 2004; Riediger et al., 2004; Mack et al., 2010; Potes et al., 2010a). It is important to note that c-Fos expression does not necessarily have a functional correlate for a behavioral effect of the same stimulus (e.g., see Züger et al., 2013). Nonetheless, c-Fos expression helps define brain areas that may be involved in responses to the respective stimulus. According to these studies, endogenous and exogenous amylin activates a neuroaxis (see section VI.D) comprising the AP, the NTS, and the LPB; activation of the latter areas seems to depend on an intact AP because the effect is absent in AP lesioned rats, and hence in rats that also do not eat less after amylin injection.

Markers that may also be functionally linked to amylin are the formation of the second messenger cGMP (Riediger et al., 2001) or the phosphorylation of ERK1/2 (Potes et al., 2012). Local AP injection of a membrane permeable analog of cGMP decreased eating similar to amylin by a meal size effect (Mollet et al., 2004). Furthermore, phosphorylation of ERK occurred rapidly in neurons carrying the calcitonin receptor after peripheral amylin (Lutz et al., 1995b; Potes et al., 2010b), and the prior administration of U0126 (1,4-diamino-2,3-dicyano-1,4-bis[2-aminophenylthio]butadiene), an inhibitor of ERK phosphorylation, into the fourth brain ventricle reduced the acute eating inhibitor effect of amylin under certain experimental conditions (Potes et al., 2010b, 2012). Hence, cGMP and the ERK cascade seem to be directly involved in amylin's anorectic effect (see also section III.C).

A downstream consequence of amylin-induced activation of AP neurons seems to be an increased expression of dopamine-β-hydroxylase (DBH), the key enzyme in noradrenaline synthesis, and the subsequent release of noradrenaline possibly in the NTS or LPB (Potes et al., 2010c). DBH-positive neurons in the AP seem to be necessary for peripheral amylin to reduce eating because a partial chemical lesion of these neurons was sufficient to abolish the eating-inhibitory effect of peripheral amylin (Potes et al., 2010c). Interestingly, the number of amylin-activated neurons correlated with the number of remaining noradrenergic neurons in the AP, indicating that rats with larger noradrenergic lesions and hence fewer remaining DBH-positive neurons also displayed lower numbers of amylin-activated AP neurons. A recent electrophysiological study suggested that glutamatergic neurotransmission in the AP seems to play a role in mediating amylin effects. According to this study, amylin receptors appeared to be located mainly on presynaptic glutamatergic terminals connecting to the AP neurons (Fukuda et al., 2013); the association of this effect with the effects in AP neurons described above has not yet been clarified.

C. Amylin-Sensitive Area Postrema Neurons Are Sensitive to Nutrients

The activity of amylin-sensitive AP neurons seems to be modulated by nutrients. The first evidence was provided by electrophysiological in vitro studies that showed that amylin and glucose coactivate AP neurons (Riediger et al., 2002); neurons that were activated by amylin reduced their spontaneous activity in a concentration-dependent manner when exposed to low glucose concentrations. This cosensitivity finds a correlate in an amylin-mediated effect that is reduced at low glucose levels; amylin’s effect to reduce the rate of gastric emptying was absent under hypoglycemic conditions (Gedulin and Young, 1998). It will be interesting then to compare amylin’s effects on eating under euglycemic, hyperglycemic, and hypoglycemic conditions. Teleologically, the glucose dependence of amylin’s effects seems plausible because it guarantees that amylin should not limit the supply of nutrients at a time when they are urgently needed.

Amino acids have also been shown to modulate amylin’s effect on eating and, at the cellular level on the activity of AP neurons. Interestingly, and opposite of the effect of glucose, amino acids seem to inhibit rather than activate these amylin responses. We recently showed that rats exposed to low-protein (1%–8%) diets reduce eating more when injected with amylin than rats exposed to a diet with normal protein content (18%). Similarly, dietary protein seemed to inhibit rather than to enhance the amylin-induced c-Fos activation in the AP, and the same was seen in rats that received parenteral injections of amino acids (Züger et al., 2013). The mechanisms underlying these phenomena and the functional implications are not clear; whether this mechanism is meant to prevent the intake of amino acid–deficient diets and thus amino acid imbalance is not clear and has not been tested because all rats included in these experiments were protein replete. In the same context, it is also not clear whether the effects are due to specific amino acids directly reducing the effect of amylin to activate AP neurons using AMY receptors themselves or via other cellular mechanisms. Finally, these studies also indicated that the effect of amylin to reduce eating and to induce c-Fos expression may be dissociated under certain conditions.

D. Neuroaxis Activation by Amylin

The evidence for an important (although perhaps not the sole [see below]) role of the AP for the mediation of the eating-inhibitory effect of peripheral amylin is strong (e.g., Sexton et al., 1994b; Lutz et al., 1998c; Becskei et al., 2004; Mollet et al., 2004; Mack et al., 2010; Potes et al., 2010c). The activation of AP neurons may then trigger the activation of a neuroaxis that projects rostrally to the forebrain and that includes the NTS, the LPB, and the central amygdala (CeA) (Lutz et al., 1998c; Becskei et al., 2007) (Fig. 5). Lesions of the respective brain areas (AP, NTS, and LPB) abolished amylin’s effect, and activation (as indicated by c-Fos) was absent in brain areas rostral to the lesion (e.g., in the NTS, LPB, and CeA in AP lesioned rats, or in the CeA in LPB lesioned rats). The direct link between the respective brain areas was confirmed by the use of anterograde and retrograde neuronal tracers; these studies also identified the LPB as the primary relay between the hindbrain and the hypothalamus, including the lateral hypothalamic area, where amylin reduces fasting-induced c-Fos expression (Riediger et al., 2004; Potes et al., 2010a) and the VMH (Mollet et al., 2003b; Roth et al., 2008a; Turek et al., 2010). The latter area received more attention in recent years due to the possible role of the VMH in mediating the interaction between amylin and leptin (see section IX.A).

E. Amylin Action in Other Brain Areas

Despite the clear evidence indicating an important role of the AP in mediating peripheral amylin’s actions, it also needs to be realized that amylin binding sites have widespread distribution throughout the brain (Sexton et al., 1994b), and the same is true for the expression of the amylin receptor components, the calcitonin receptor (Becskei et al., 2004) or the RAMPs (Ueda et al., 2001; Barth et al., 2004). Central amylin produces very potent effects on eating (e.g., Rushing et al., 2000; Osaka et al., 2008; Wielinga et al., 2010) that at least under certain conditions are independent of an intact AP (Lutz et al., 1998b), and most reports indicate that lower doses of amylin are required to reduce eating (and body weight) after infusion into the lateral or third than in the fourth cerebral ventricle (Lutz et al., 1997, 1998b; Rushing et al., 2000). Nonetheless, the physiologic relevance of the eating-inhibitory effect of central amylin is less clear, because for most of the potential sites of action, activation by peripheral amylin has not been demonstrated and it is also unclear which areas are primarily targeted by centrally infused amylin. Furthermore, evidence for an action of centrally produced amylin is scarce and may only occur under very specific conditions in females, which may not be related to effects of amylin on eating (Dobolyi, 2009).

One interesting study recently provided evidence that amylin receptors in the VTA may be direct targets for peripheral amylin. The VTA is well known for its role in the control of eating and in particular for the intake of palatable food (Mietlicki-Baase et al., 2013). That study showed that all amylin receptor components are present in the VTA and that direct administration of the amylin/calcitonin receptor agonist salmon calcitonin into the VTA reduced eating of normal chow as well as palatable sucrose. Similar to peripheral amylin (Lutz et al., 1995b) or amylin given into the AP (Mollet et al., 2004), reduced eating appeared to result at least in part from a meal size effect. The authors suggested that amylin may reduce meal size by modulating the reward value of food throughout the meal. The observation that administration of the antagonist AC187 into the VTA increased eating and that intra-accumbens AC187 reduced the effect of peripheral salmon calcitonin on eating indicates that salmon calcitonin may in fact exert a direct action on VTA neurons and that the effects seem to be of physiologic relevance (Mietlicki-Baase et al., 2013). The relations between the VTA-mediated effect and the previously described AP-mediated effects are currently unclear. As noted in section III.C, salmon calcitonin and AC187 are not selective tools for only amylin receptors, having activity also at calcitonin receptors. Therefore, authors must be careful in their data interpretation using these molecules.

The nucleus accumbens, which also plays a role in reward/hedonics, has also been tested for its potential role in the control of food intake by amylin (Kelley, 1999; Baldo and Kelley, 2001). The experiments were based on the knowledge that the nucleus accumbens contains a high density of high-affinity amylin binding sites (Beaumont et al., 1993; Sexton et al., 1994b). In fact, nucleus accumbens membranes had originally been used to characterize the properties of amylin binding sites. It was shown that administration of amylin into the shell of the accumbens reduced eating and drinking; however, because amylin also markedly reduced locomotor activity and because the latter effects were observed at lower doses than the eating effect, these motor effects may perhaps be part of an amylin effect on the “satiety sequence,” which includes a number of typical postprandial behaviors like rearing and cleaning. A direct contribution of the nucleus accumbens to the amylin effect on meal termination seemed to be less likely (Halford et al., 1998).

F. Summarizing Remarks

Overall, the neuroaxis that is activated by peripheral amylin or its analogs has been reasonably well defined (Potes and Lutz, 2010). Evidence for a critical role of specific hindbrain structures, in particular the AP, NTS, and LPB, is compelling; activation of these structures may then affect specific hypothalamic nuclei (Fig. 5), but their individual roles in the effects of amylin on eating are still not clear (Potes et al., 2010a; Trevaskis et al., 2010a; Turek et al., 2010). Furthermore, more studies are required to define the potential contribution of other circumventricular organs such as the SFO (Fry et al., 2007; Hoyda et al., 2009; Mimee et al., 2013). Recent studies indicate that in addition to the AP, peripheral amylin may activate a more complex and distributed network that includes the VTA (Mietlicki-Baase et al., 2013, 2015). However, how the VTA and the other brain areas may interact is also not clear. Furthermore, although central amylin produces potent eating-inhibitory effects, the brain structures mediating these effects and their relevance under physiologic conditions are not clear.

A. Amylin in the Cardiovascular System

In general, the majority of amylin’s actions on the cardiovascular system are believed to be mediated by activation of CGRP receptors, widely expressed within the vasculature (for details, see Young, 2005a). These actions are believed to be largely pharmacological due to the high plasma concentrations required to observe a response. Consequently, in rodent models, bolus intravenous injection of amylin produces a potent vasodilatation to decrease arterial blood pressure, accompanied with short-term baroreflex increases in heart rate (Young et al., 1991). Little work has been reported investigating how well these actions translate to human pharmacology. However, it is likely that clinically relevant doses of pramlintide for treatment of diabetes or obesity are insufficient to increase plasma exposure to levels required to activate CGRP receptors in humans, because it is estimated that the EC₅₀ for blood pressure lowering exceeds therapeutic pramlintide concentrations by three orders of magnitude. Confirming this hypothesis, 12-month dosing with pramlintide in humans did not significantly change blood pressure (Young et al., 1999).

Direct cardiac inotropic effects, potentially CGRP receptor mediated, have been reported for amylin in isolated rodent cardiomyocytes (Bell and McDermott, 1995), isolated rat hearts (Kaygisiz et al., 2010), and isolated cardiac tissue from pigs (Saetrum Opgaard et al., 1999). However, because of the high concentrations required to observe efficacy, it is again unlikely that these actions are clinically relevant at therapeutic doses.

B. Amylin and Lipolysis

Given the observation that amylin administration decreases adiposity, the extent to which amylin exerts direct effects on BAT and white adipose tissue has been explored. In isolated adipocyte preparations, amylin stimulates neither lipolysis nor basal or insulin-stimulated rate of glucose incorporation into either CO₂ or triacylglycerol (Cooper et al., 1988; Lupien and Young, 1993; Roth et al., 2006). When epididymal fat pads were excised from amylin-treated animals, the basal rate of lipolysis did not differ relative to a vehicle-treated control group. Sustained in vivo administration of amylin also failed to augment the ex vivo lipolytic sensitivity of adipose tissue to adrenergic stimulation. Thus, at least in epididymal fat, amylin did not regulate basal or stimulated lipolysis. However, more recent in vitro studies reported that amylin activated multiple intracellular pathways in human adipose tissue such as the STAT3, AMP-activated protein kinase, Akt, and ERK signaling pathways (Moon et al., 2011). Of note, these effects were obtained at very high pharmacological levels of amylin in the culture media and the physiologic significance of these findings remains to be determined.

C. Amylin and Maternal Behaviors

Microarray studies detected a marked increase (approximately 25-fold) in amylin expression within the preoptic area of the hypothalamus of rat dams (Dobolyi, 2009). These mRNA findings were confirmed by quantitative polymerase chain reaction and in situ hybridization methods. Next, a series of experiments detailed the time course (during late pregnancy and throughout lactation) and potential mechanisms for amylin expression (Szabó et al., 2012). First, mRNA levels remained elevated as long as the pups were not removed from the dams. Amylin expression was also induced in maternally behaving (sensitized) nonlactating females but not in nonsensitized nulliparous females or in females that did not become maternal despite the sensitization procedure. This phenomenon was sex steroid hormone independent because ovariectomy had no effects. In other studies, mothers were also separated from their pups for 22 hours. On return of the pups, neuronal activation was found in the mother's preoptic area, with a distribution pattern similar to amylin-expressing neurons. The authors hypothesize that amylin may play a role in the physiologic regulation of maternal adaptations potentially endocrine and emotive, although these remain to be formally demonstrated.

D. Amylin and Stress-Induced Eating

Preclinical findings suggest that amylin agonism may be beneficial in the treatment of neuropsychiatric disease (see Table 4). 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 laboratory 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 after restraint stress, while maintaining their intake of standard laboratory chow (Roth et al., 2009). Beneficial effects have also been noted in other preclinical stress models. For example, during recovery from social stress in a visible burrow system (VBS), during which a dominance hierarchy is formed among the males, rats display hyperphagia and gain weight preferentially as visceral adipose tissue. By proportionally increasing visceral adiposity, social stress may contribute to the establishment of metabolic disorder. Amylin was administered to rats fed ad libitum during recovery from VBS stress in an attempt to prevent hyperphagia and the resultant gain in body weight and fat mass. Amylin treatment reduced food intake, weight gain, and accumulation of fat mass in male burrow rats but not in male controls that spent time housed with a single female rather than in the VBS (Smeltzer et al., 2012).

E. Antipsychotic-Like Actions of Amylin

Recently, antipsychotic-like actions of amylin were demonstrated after central injections (Baisley et al., 2014). Amylin infusion into the accumbens shell, but not the dorsal striatum, reversed amphetamine-induced prepulse inhibition without affecting baseline startle. Prepulse inhibition is a commonly used cognitive test paradigm of the reaction of an organism, such as a mouse, to “startling” stimuli of different magnitude, such as a sound (Castagné et al., 2009). Coinfusion of AC187 blocked the ability of amylin to normalize amphetamine-induced prepulse inhibition disruption. These findings suggest that amylin receptors may be a potential target for the development of putative antipsychotics and future work should explore the potential utility of systemically administered amylin in this regard.

F. Amylin and Alzheimer’s Disease

Two recent reports suggest that peripheral treatment with either amylin or pramlintide can have therapeutic effects in Alzheimer’s disease (AD) (Adler et al., 2014; Zhu et al., 2015). First, lower amylin levels were noted among subjects with AD and mild cognitive impairment compared with the cognitively intact subjects (Adler et al., 2014). This led the authors to test whether chronic infusions of the amylin analog pramlintide would be beneficial in rodent models of AD. Chronic infusions of pramlintide in the senescence-accelerated prone mouse (a model of sporadic AD) improved performance in the novel object recognition task (assay for memory and cognition). These behavioral effects were accompanied by increased markers of synaptic formation and decreased markers of inflammation and oxidative stress within the hippocampus (Adler et al., 2014). Likewise, both amylin and pramlintide administration in a transgenic model of AD improved learning and memory in the Y-maze and Morris swim tests. These effects were coupled with reduced amyloid burden and reduced Aβ in brain tissue. The authors suggest that amylin and pramlintide enhance the removal of Aβ from the brain and its transfer into the blood, probably through their effects on cerebral vasculature, and amylin and pramlintide may represent a novel therapeutic avenue for the treatment of AD. However, more work is needed to determine whether amylin is a beneficial or contributory factor to AD (Yang and Song, 2013).

G. Amylin in Renal Physiology

The effects of amylin in the kidney have been overlooked in recent years but amylin has interesting effects on renal physiology. In the spontaneously hypertensive rat model, the density of amylin binding sites was greater before the development of hypertension, compared with control Wistar-Kyoto rats. As blood pressure increased with age in the spontaneously hypertensive rat model group, so too did the density of amylin binding sites (Wookey et al., 1997). Thus, amylin and its receptor(s) may contribute to the development or maintenance of hypertension (Wookey and Cooper, 1998). Amylin may also act as a growth factor in the kidney and may control water and sodium reabsorption (Harris et al., 1997; Wookey et al., 1998). In anesthetized rats, amylin infusion can also promote diuresis and natriuresis (Vine et al., 1998b).

A. Diabetes

B. Obesity

The studies summarized in section V indicate that amylin physiologically contributes to the control of eating, energy expenditure, and body weight. On the basis of these actions, it could be hypothesized that reduced amylin secretion, action, or sensitivity may contribute to less satiation, increased meal sizes and overall eating, and lower energy expenditure, together eventually contributing to the development of obesity.

C. Identifying Amylin Agonist-Responsive Populations in Preclinical Models

With the arsenal of available therapies for metabolic diseases continuing to evolve, it is likely that subgroups of target populations will exhibit differential responsiveness to any given pharmacological agent. Just as large differences in the regulation of energy homeostasis exist between males and females, it is not unreasonable to assume that therapeutic modalities will be more, or less, efficacious as concentrations of gonadal hormones fluctuate during development and aging. Sexually dimorphic effects are especially relevant with respect to neurohormone-based therapies for metabolic diseases, because pharmacological and neurobiological studies have convincingly demonstrated a role for estradiol in the central control of energy balance. These studies generally predict that the presence of estradiol enhances the anorexigenic properties of short-term signals of satiation (e.g., CCK) and long-term signals of adiposity (e.g., leptin), while concomitantly decreasing the potency of orexigenic signals (e.g., melanin concentrating hormone, neuropeptide Y, ghrelin; reviewed in Asarian and Geary, 2006; Brown and Clegg, 2010; Butera, 2010). We recently explored whether estradiol signaling affected the weight regulatory and metabolic effects of amylin agonism (see also Fig. 5) (Trevaskis et al., 2010c). We observed a surprising approximately 2-fold increase in amylin’s weight-lowering efficacy in a state of estradiol deficiency [DIO ovariectomized (OVX) rat model] compared with sham-operated controls. Detailed metabolic studies revealed that amylin reversed the decline in energy expenditure and fat oxidation linked to an estradiol-deficient state via food intake–dependent (reduction in respiratory exchange ratio) and food intake–independent (maintenance of metabolic rate) mechanisms. Because postmenopausal women comprise a large majority of patients for whom weight loss agents are prescribed, the extent to which these preclinical observations translate into the clinical use of the amylin agonist pramlintide warrants further investigation. These studies should also comprise detailed mechanistic experiments that clarify the apparent paradox that while under chronic conditions, the lack of estradiol enhances amylin action, but exogenous and endogenous amylin were more effective when given acutely to OVX rats that received physiologic estrogen replacement therapy (Asarian et al., 2011).

Interestingly, an antidepressive and neurogenic profile of amylin agonism was also noted in these studies. First, amylin restored neurogenesis in the hippocampus of OVX rats and increased (approximately 2-fold) neurogenesis within the AP. These alterations were not evident in sham rats treated with amylin. Amylin-treated OVX rats also displayed decreased immobility (i.e., improvements in this depressed phenotype) compared with sham-operated controls in the forced swim test. These findings warrant further exploration but they are noteworthy because estradiol concentrations decrease in postmenopausal women, who make up a high percentage of the obese population, and incidences of depression are reported to be increasing during menopause (Asarian and Geary, 2006). To what extent amylin agonism may hold utility in neurodegenerative diseases remains to be elucidated. Collectively, these preclinical findings raise the intriguing possibility that the integrated mechanisms of amylin may improve metabolic and behavioral processes and represent an important area for future clinical research.

D. Next-Generation Drugs for the Amylin System

Pramlintide is a useful mimetic of human amylin but other amylin agonists with improved potency and pharmacokinetics may provide optimized therapeutics for the treatment of obesity. Several additional analogs of human amylin have been synthesized and investigated. The feasibility of orally administered amylin has also recently been explored. There are also alternative approaches that could be used, such as inhibition of amylin degradation.

A. Amylin and Leptin

The most extensively investigated amylin-based combination is with the adipokine leptin (Fig. 6). Leptin is regarded as the prototypical long-term signal of energy balance; however, obese rodents and humans are largely nonresponsive to exogenous leptin administration. Early studies demonstrated that acute third ventricular leptin increased the eating-inhibitory effect of peripheral amylin in lean rats (Osto et al., 2007). Next, exogenous pharmacological amylin was shown to synergize with peripherally administered leptin to induce marked and sustained fat-specific body weight loss in leptin-resistant DIO rats (Roth et al., 2008a; Trevaskis et al., 2008). These findings suggested that amylin may help restore leptin responsiveness in obesity. The effects appear to be pharmacological in nature; in other words, both agonists need to be present for leptin’s pharmacological effects to be evident in obesity. DIO rats pretreated with amylin or with the combination of amylin and leptin did not maintain their weight loss when switched to a regimen of leptin monotherapy (Roth et al., 2008a; Turek et al., 2010). Studies exploring caloric restriction, surgery, and coadministration of other anorexigens suggest that amylin appears to be one of the best modalities for restoring leptin responsiveness. Caloric restriction alone does not appear to be sufficient to restore leptin responsiveness in preclinical models or obese human subjects. For example, rats pair fed and weight matched to an amylin-treated group did not lose additional weight when leptin was coadministered (Roth et al., 2008a).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

(A) Effect of amylin on body weight in DIO rats. Change in body weight (percent vehicle corrected) of rats receiving amylin (3–300 μg/kg per day) for 4 weeks via subcutaneously implanted osmotic minipumps. Mean ± S.E.M. *P < 0.05 versus vehicle. (B) Effect of pramlintide in obese subjects. Changes in body weight from baseline for placebo (n = 17), 120 μg pramlintide (n = 24), 240 μg pramlintide (n = 17), or 360 μg pramlintide (n = 21) twice daily (evaluable population). Mean ± S.E.M. *P < 0.05; **P < 0.01 versus placebo. (C) Effect of amylin, leptin, or amylin plus leptin on body weight in DIO rats. Change in body weight (percent vehicle corrected) of rats pretreated for 14 days with 100 μg/kg per day amylin and then maintained on 100 μg/kg per day amylin switched to either 500 μg/kg per day leptin monotherapy or 100 μg/kg per day amylin plus 500 μg/kg per day leptin combination therapy (n = 6–8/group). Compounds were delivered for 6 weeks via subcutaneously implanted osmotic minipumps. Mean ± S.E.M. *P < 0.05 versus vehicle; ^#P < 0.05, compared with monotherapies. (D) Effect of metreleptin, pramlintide, or metreleptin plus pramlintide combination on body weight in obese subjects. Changes in body weight from enrollment (week 4) for subjects pretreated with pramlintide (titrated from 180 μg to 360 μg) twice and then treated with 360 μg pramlintide twice daily, 5 mg metreleptin twice daily, or pramlintide plus metreleptin combination treatment (evaluable population, n = 93). Mean ± S.E.M. ^#P < 0.05; ^##P < 0.01; ^###P < 0.001, compared with monotherapies. This figure has been redrawn from the individual publications in which these data were presented. (A) contains data from Mack et al. (2010); (B) contains data from Smith et al. (2008); and (C and D) are from Roth et al. (2008a).

In terms of other pharmacologies, doses of PYY(3-36) or an exenatide analog were selected that gave a similar magnitude of weight loss (6%) that was sufficient for amylin to restore leptin responsiveness (Roth et al., 2008a). In these studies, only additive to less than additive effects were observed. In line with these findings, leptin replacement alone or with exendin-4 failed to promote greater weight loss in weight-reduced DIO rats (Reidelberger et al., 2012). Although a leptin/exendin synergy has been reported in DIO mice, it was only evident after an initial body weight loss of 30% (Müller et al., 2012), far greater than the 6% required for amylin/leptin synergy. Mechanistic studies suggest that the presence of amylin “primes” the hypothalamus to respond to leptin by enhancing leptin receptor number and/or signaling capacity (for a recent review of the physiologic, neurobiological, and molecular mechanisms mediating these interactions, see Trevaskis et al., 2010b). This may be due to amylin increasing interleukin-6 expression (Le Foll et al., 2015). It was recently reported that leptin and amylin may also interact functionally in the VTA (Mietlicki-Baase et al., 2015).

B. Amylin and PYY(3-36)

PYY is a 36-amino-acid peptide hormone that is secreted from intestinal L cells after a meal and is cleaved by dipeptidyl peptidase-IV to generate PYY(3-36), the major circulating form (Grandt et al., 1994). Experiments in rodents have shown that PYY(3-36) inhibits food intake, reduces body weight gain, and increases utilization of fat stores for energy. A series of rodent studies examined the effects of combined administration of amylin and PYY(3-36). Overall, these studies revealed that amylin and PYY(3-36) together were able to elicit a greater reduction in 24-hour food intake after a single acute injection than either monotherapy without significantly changing locomotor activity. With sustained administration, the combination synergistically reduced food intake and additively reduced body weight in DIO-prone rats (Roth et al., 2007b).

C. Amylin and Cholecystokinin

CCK is a short-term satiation signal released from intestinal L cells after nutrient consumption. After its release, CCK is thought to exert a paracrine effect by its actions on vagal afferents that project to the NTS. CCK acutely synergizes with amylin to suppress food intake in lean mice (Bhavsar et al., 1998; Mollet et al., 2003a). To expand on these findings, sustained administration studies (7 days) assessed whether the combination of amylin and CCK would synergize for weight loss in DIO rats. As expected, amylin treatment induced significant body weight loss. CCK, although ineffective alone, significantly enhanced body weight loss when coadministered with higher doses of amylin. c-Fos activation was also assessed in various brain nuclei after a single intraperitoneal injection of amylin and/or CCK. Amylin and CCK additively increased c-Fos within the AP, predominantly in noradrenergic (e.g., DBH-containing) cells.

D. Amylin and Melanocortins

An important downstream regulator of leptin is the hypothalamic melanocortin (MC) system. Convergent lines of evidence underscore the importance of the melanocortin subtype-4 receptor (MC-4R) as a key regulator of feeding behavior. The impact of impaired MC signaling on amylin’s efficacy has been evaluated in two species. Chow-fed overnight fasted agouti and wild-type mice respond similarly to an acute dose of amylin (Roth et al., 2006). However, MC-4R–knockout mice displayed quantitatively reduced responsiveness to salmon calcitonin at a dose range that was effective in chow-fed C57Bl/6J mice (overnight fasted) (Eiden et al., 2002), and Mc4r^K314X/K314X rats were nonresponsive to an acute dose of amylin that was effective in wild-type rats. Overall, it is unclear whether amylin's acute food intake–lowering effects are MC-4R dependent (Roth et al., 2012). One explanation for these discrepant findings may be that these are two different models of MC dysregulation; agouti mice are a model of ectopic endogenous MC-3R/MC-4R antagonism (by agouti overexpression), whereas the Mc4r^K314X/K314X rat is a receptor loss-of-function model. The methodological and procedural differences (e.g., species, diet composition, nutritive status) across these different studies may also play a role. Nevertheless, coadministration studies clearly demonstrate the antiobesity potential for combined agonism of these MC and amylinergic systems. Amylin coadministration enhanced the acute food intake–lowering effects of Ac-R[CEH-dF-RWC]-amide in mice and with repeated coadministration had durable, additive anorexigenic, weight-lowering, and fat-lowering effects in DIO rats (Roth et al., 2012).

E. Amylin and Glucagon-Like Peptide 1

Combined amylin and GLP-1 receptor agonism has also been assessed (Bello et al., 2010). This study evaluated the acute anorexigenic effects of coadministering the amylin/calcitonin receptor agonist salmon calcitonin alone and in combination with GLP-1 receptor agonist exenatide in a non–human primate model. Monkeys were maintained on a schedule of 6-hour daily access to food and multiple dose combinations were used. This experimental design enabled the authors to use response surface methodology and test for statistical additivity or synergy for suppression of food intake. Synergistic anorexigenic effects were noted during hours 1–4 and additive effects during hours 5 and 6. Behavioral observations suggested that the monkeys did not display malaise or nausea. After a washout period, monkeys received repeated daily injections of one of the dose combinations (0.56 μg/kg Exendin-4 + 0.32 μg/kg salmon calcitonin) for 5 days. During this phase of the study, sustained reductions in daily food intake (>70% from saline baseline) were noted for 5 days, suggesting that the additive/synergistic effects of the combination may be durable. Future repeated administration studies should include monotherapy arms to give a better sense for the potential additive/synergistic potential of this combination.

F. Amylin and Small Molecule Anorectics

Amylin has also been combined with various classes of small molecule anorectic agents that are either still approved for the short-term treatment of obesity (phentermine), are withdrawn from market (sibutramine), or are in late stages of regulatory review (naltrexone/bupropion). These studies are reviewed below.

Phentermine is a β-phenethylaminic derivative that acts through monoaminergic systems (noradrenaline and dopamine) to decrease food intake and body weight. Sibutramine inhibits the reuptake of central noradrenaline and serotonin and stimulates thermogenesis in rodents. Mathematically additive food intake, weight-lowering, and fat-lowering effects were observed when amylin was combined with either phentermine or sibutramine. Detailed analyses of feeding patterns with amylin and phentermine suggest that these effects may be achieved by an approximately 2-fold increase in the satiety ratio compared with either agent alone (Roth et al., 2008c). In terms of off-target behavioral effects, amylin neither enhanced nor reduced the known actions of phentermine (which is an amphetamine derivate) to stimulate activity (Roth et al., 2008c).

Antagonism of opioid systems (e.g., with naltrexone) is another antiobesity strategy, and it is particularly effective when coadministered with dual inhibitors of dopamine and norepinephrine reuptake (e.g., bupropion). The interaction of amylin with naltrexone/bupropion on energy balance has recently been explored (Clapper et al., 2013). Wild-type and amylin knockout mice were similarly responsive to the food intake–lowering effects of either naltrexone (1 mg/kg s.c.) or bupropion (50 mg/kg s.c.), suggesting that they act independently of amylinergic systems and could interact additively when given in combination with amylin. To test this, DIO rats were treated (for 11 days) with vehicle, rat amylin (50 μg/kg per day s.c.), naltrexone/bupropion (1 and 20 mg/kg, respectively, by twice-daily s.c. injection), or their combination. Amylin plus naltrexone/bupropion combination therapy exerted additive effects to reduce cumulative food intake, body weight, and fat mass. In a separate study, the effects of amylin and naltrexone/bupropion administered at the same doses (for 14 days) were compared with a pair-fed group. Although the combination and pair-fed groups lost a similar amount of body weight, rats treated with the combination lost 68% more fat and better maintained their lean mass. These findings support the strategy of combined amylin agonism with opioid and catecholaminergic signaling systems for the treatment of obesity.

A. Pharmacology of Amylin Agonists

Pramlintide is a synthetic and equipotent analog of human amylin and is used as an adjunctive therapy to patients treated with mealtime insulin. It differs from the natural peptide by three amino acids; prolines were substituted at positions 25, 28, and 29 (Fig. 3). These substitutions were necessary to overcome several physicochemical properties that make human amylin unsuitable for pharmacologic delivery, including poor stability in solution and a propensity to aggregate and adhere to surfaces. Hence, amylin’s biology and therapeutic potential in humans is almost exclusively assessed using pramlintide. By contrast, reports on the actions of amylin in rodents have been assessed using either rat amylin (which also possesses favorable pharmaceutical properties) or pramlintide. In general, the pharmacological and pharmacokinetic properties of pramlintide are identical to those of rat amylin, when the two peptides are compared in preclinical rodent models (Young et al., 1996b).

The pharmacokinetics of pramlintide in rats reflects a half-life of approximately 13 minutes after intravenous injection, a value identical to that of rat amylin (Young et al., 1996b). A similar profile is seen with intravenous injection of human amylin into normal human males (Clodi et al., 1998).With subcutaneous injection in rodents, half-lives for both rat amylin and pramlintide are in the range of 15–35 minutes, whereas bioavailability ranged between 25% and 40%. In normal humans, subcutaneous injections of pramlintide ranging from 300 to 10,000 μg produced half-life values varying from 26 to 42 minutes (Moyses et al., 1993). In subjects with type 1 and type 2 diabetes, subcutaneous doses ranging from 30 to 180 μg gave similar profiles, with maximal plasma concentrations achieved in approximately 20 minutes and a half-life of around 50 minutes (Colburn et al., 1996; Weyer et al., 2001). Two recent articles can be consulted for more information on the pharmacokinetics of pramlintide (Younk et al., 2011; Fang et al., 2013).

B. Clinical Studies Relating to Diabetes

Collectively, multiple clinical studies in insulin-using patients with type 1 and type 2 diabetes suggest that pramlintide addresses important deficits in the treatment of diabetes, and pramlintide in combination with insulin may provide a more physiologically balanced therapeutic approach (Younk et al., 2011). A 2011 review of the literature pertaining to the clinical data for the use of pramlintide in diabetes covers most of these studies in some depth (Younk et al., 2011). In brief, pramlintide treatment improved hemoglobin A1c, reduced postprandial glucose excursions relative to insulin therapy alone, enabled reductions in mealtime insulin dosing, inhibited postprandial glucagon secretion, and improved overall glycemic control. Although excess weight gain is associated with many antihyperglycemic therapies and can potentially induce further insulin resistance, pramlintide treatment was associated with sustained and significant weight loss. This aspect was apparent without prescribed dietary/lifestyle interventions and is explored in more detail below. Nausea (mild to moderate) is the most frequently reported adverse event for pramlintide, yet the weight loss experienced with pramlintide is largely independent of any reported nausea.

Pramlintide was approved in the United States as an adjunct therapy to mealtime insulin. The subcutaneous dose of pramlintide is fixed with 60 μg for patients with type 1 diabetes and 120 μg for patients with type 2 diabetes, and patients are instructed to dose pramlintide 30 minutes before a meal (Janes et al., 1996; Center for Drug Evaluation and Research Approval Package for Application Number 21-332. Clinical Pharmacology and Biopharmaceutics Review. http://www.accessdata.fda.gov/drugsatfda_docs/nda/2005/21-332_Symlin%20Injection_biopharmr.pdf). Given the fact that pramlintide is delivered as a separate injection to mealtime insulin, compliance for patients may be an issue. One study in patients with type 2 diabetes evaluated the addition of pramlintide versus mealtime insulin as an adjunct to basal insulin to reduce the injection burden and the clinical effect of pramlintide without mealtime insulin. The results showed that pramlintide was similar to rapid-acting mealtime insulin with glycemic reduction but with less hypoglycemia and no weight gain (Riddle et al., 2009).

In patients with type 1 diabetes, not only is compliance an issue, but these patients are also more prone to insulin-induced hypoglycemia. Few studies have looked at ways to make the combination easier and potentially safer for this population. One study showed that acute mixing of pramlintide with rapid-acting or basal insulin did not significantly alter the pharmacokinetics and clinical effectiveness of pramlintide; however, there are no long-term mixing studies to assess the stability of both molecules (Weyer et al., 2005). Insulin-induced hypoglycemia is the most serious complication when using pramlintide with mealtime insulin in the type 1 diabetes population. Anecdotally, clinicians have reported that dosing analog insulin before meals in conjunction with pramlintide leads to more hypoglycemia because patients underestimate the satiety effect of pramlintide and end up overdosing the mealtime insulin. Given this, some clinicians have started dosing mealtime insulin after the meal rather than before the meal. Although it has not yet been officially examined, we can see this trend with some of the clinical trial designs for patients with type 1 diabetes using pramlintide as an adjunct to mealtime insulin (Hassan and Heptulla, 2009).

Because more patients with type 1 diabetes are using pump therapy, investigators have looked at different ways of delivering the mealtime insulin dose in conjunction with pramlintide. Two separate studies have been conducted to evaluate how best to dose insulin in a patient using pump therapy, in conjunction with pramlintide. One study tested four different insulin delivery wave forms: standard, square wave, combination (standard shifting to square), and modified combination (standard combined with square 1 hour into the meal). In that study, it was found that the square and modified combination was the safest way to deliver insulin (King, 2010). Another study found that a square wave delivery for patients with type 1 diabetes near normal control was the best bolus approach, whereas with weight loss, there may need to be adjustments to total basal insulin dose to avoid hypoglycemia (King, 2009).

Amylin and insulin are colocated and cosecreted in a fixed molar ratio from the pancreatic β-cells; therefore, amylin is also secreted in a basal-bolus fashion. However, in clinical practice, insulin is often titrated for clinical effect, whereas pramlintide is delivered in a fixed dose and only at mealtime. Few studies have evaluated the effect of pramlintide delivered in a basal-bolus fashion. A 24-hour infusion study evaluating pramlintide with insulin basal-bolus infusion versus insulin infusion alone found that there was a significant decrease in postprandial glycemic control and robust glucagon suppression (Heptulla et al., 2009). Given the exploratory nature of this study, the pramlintide dose was calculated based on previous experience but had to be changed during the study. Another pilot study evaluated the pramlintide basal-bolus concept over a 16-week period. The basal pramlintide rate was 9 μg/h and the bolus dose was administered per label up to 60 μg/bolus at mealtime. The study showed significant glucose, insulin dose, and weight reduction but an increase in mild hypoglycemia (Huffman et al., 2009). A real-world pilot study with a small cohort of patients with type 1 diabetes (n = 5) and patients with type 2 diabetes (n = 5) with variable duration of diabetes and variable pramlintide to insulin ratios over a span of 1–5 years found that delivery of pramlintide in a separate pump in a basal-bolus fashion resulted in improved glycemic control, insulin dose reduction, and weight loss (Schorr and Ofan, 2012). Therefore, there may be some alternative dosing strategies that are possible for pramlintide in patients with diabetes, beyond current recommendations.

Given its physiologic actions and its clinical profile, pramlintide is an attractive candidate as an additional therapy to be included in the artificial pancreas in the treatment of type 1 diabetes. Excessive glycemic variability and hypoglycemia are major obstacles in the development of an artificial pancreas. A study evaluated dosing of 30 μg pramlintide bolus under a closed loop system and showed that pramlintide not only delayed the time to peak of postprandial hyperglycemia, but it also significantly reduced the glycemic excursion (Weinzimer et al., 2012). There are currently no data to indicate the optimal pramlintide to insulin ratio. Most of the studies done thus far tested variable ratios in combination with a fixed dose bolus. At the time of this writing, there is an active pilot study being conducted by the Juvenile Diabetes Research Foundation to evaluate the optimal insulin to pramlintide ratio in the clinical setting (ClinicalTrials.gov identifier NCT01708044: http://www.clinicaltrials.gov/ct2/show/NCT01708044?term=pramlintide&rank=13). Although much work has been done, more studies are still needed to evaluate the role of pramlintide in an artificial pancreas.

C. Clinical Studies Relating to Food Intake, Satiation, and Weight Loss

The aforementioned observation that pramlintide reduced body weight in patients with insulin-treated diabetes, a patient population that generally suffers from weight gain due to insulin, pointed to its potential clinical utility as an antiobesity agent (Chapman et al., 2005; Aronne et al., 2007). In clinical studies in obese subjects, the weight-reducing effects of pramlintide emerged as significant, sustained, and dose dependent (Hollander et al., 2004; Smith et al., 2008; Singh-Franco et al., 2011). Interestingly, weight loss occurred despite the fact that subjects were instructed to maintain their usual diet and exercise routines. Progressive weight loss was also seen in those subjects who did not report nausea, providing further evidence that the weight-lowering effect of pramlintide is dissociable from the occurrence of nausea. The use of pramlintide alongside lifestyle intervention resulted in weight loss and enhanced long-term maintenance of weight loss in obese subjects.

Weight loss is likely to be linked with the food intake–lowering effects of pramlintide, which have been demonstrated by several clinical studies in obese subjects. A single injection of pramlintide 1 hour before a buffet meal decreased food intake by 16% compared with placebo (Smith et al., 2007). Pramlintide also elicited an increase in satiety scores, despite the consumption of less food. This indicates that less food was required to promote meal-related satiation. In other studies, pramlintide reduced total daily food intake (by 15%–20%) in obese subjects when measured both early during the study as well as after 6 weeks of administration (Chapman et al., 2005, 2007). Pramlintide was also shown to cause acute and sustained reductions in the intake of high-fat and high-sugar foods at a “fast-food challenge” and to improve perceived control of eating, as demonstrated by a 45% placebo-corrected reduction in binge eating scores (Smith et al., 2007). These findings suggest that the nuances of changes in feeding behavior (e.g., decreased meal size while concomitantly increasing satiation, reduced intake and preference for high-fat and high-sugar foods), many of which are observed in preclinical models with amylin, may also be recapitulated in the clinic with pramlintide.

Pramlintide was evaluated in conjunction with lifestyle intervention in a phase 2b dose-ranging proof-of-concept study (Smith et al., 2008). Subjects who entered the study were obese [body mass index [BMI] ≥ 30 and ≤ 50 kg/m² for at least 1 year] nondiabetic men and women aged 18–70 years with abdominal obesity [waist circumference >102 cm for men and >88 cm for women]. In this 4-month, double-blind, placebo-controlled, dose-ranging study, 411 obese subjects were randomized to receive pramlintide (six arms: 120, 240, and 360 μg twice daily and three times daily) or placebo in conjunction with a structured lifestyle intervention program geared toward weight loss. At the end of 4 months, 77% of the evaluable subjects (n = 270) opted to continue preexisting treatment during an 8-month single-blind extension with lifestyle intervention geared toward weight maintenance. At month 4, mean weight loss from baseline in the pramlintide arms ranged from 3.8 ± 0.7 to 6.1 ± 0.8 kg (2.8 ± 0.8 kg with placebo). By month 12, initial four-month weight loss was regained in the placebo group but was maintained in all but the 120 μg twice-daily group. Placebo-corrected weight loss with 120 μg three times daily and 360 μg twice daily averaged 3.2 ± 1.2 kg (3.1% ± 1.1% body weight) and 3.3 ± 1.1 kg (3.1% ± 1.0% body weight), respectively, at month 4 (both P < 0.01; 4-month evaluable n = 270) and 6.1 ± 2.1 kg (5.6% ± 2.1% body weight) and 7.2 ± 2.3 kg (6.8% ± 2.3% body weight), respectively, at month 12 (both P < 0.01; 12-month evaluable n = 146). At month 12, 40% and 43% of subjects treated with 120 μg three times daily and 360 μg twice daily, respectively, achieved ≥10% weight loss (versus 12% for placebo). Nausea was the most common adverse event with pramlintide in the 4-month study (9%–29% pramlintide versus 2% placebo) and it was generally mild to moderate and occurred in <10% of subjects during the extension. Collectively, these studies suggest that obese subjects treated with pramlintide alone or as an adjunct to lifestyle intervention can achieve weight loss and enhanced long-term maintenance of weight loss. Fig. 6 compares preclinical and clinical data for pramlintide-induced effects on body weight.

Davalintide has also undergone limited clinical investigation. Translational medicine studies revealed suppression of 24-hour food intake in obese humans treated with davalintide (Nicandro et al., 2008).

D. Clinical Combination Studies (Metreleptin and Small Molecules) Relating to Weight Loss

Obesity is a multifaceted disease. There are multiple and redundant pathways to avoid weight loss as a means of survival. Monotherapy for obesity, although initially effective, often only yields modest and unsustained weight loss. As with other chronic diseases such as hypertension, cancer, and diabetes, combinatorial approaches have gained significant traction for the potential treatment of obesity (Kim et al., 2013). Analogous to diabetes, monotherapy-based approaches for obesity target only a single aspect of a multihormonal disease state. Although it was taken off the market for safety reasons, the history of fenfluramine/phentermine demonstrated that by targeting more than one single pathway in the control of weight, a robust and sustainable weight loss can be achieved. Accordingly, a series of translational research studies have demonstrated the potential utility of combined amylin agonism with small molecule anorexigenic agents (e.g., phentermine or sibutramine) or with an analog of the long-term adiposity signal leptin (e.g., metreleptin) that may be useful in overcoming these adaptations to promote meaningful weight loss. These preclinical studies are described in section IX. Some clinical investigation has also been conducted.

As discussed above, the neurohormonal control of body weight involves a complex interplay between long-term adiposity signals (e.g., leptin), and short-term satiation signals (e.g., amylin). We have discussed that in DIO rodents, amylin/leptin combination treatment led to marked, synergistic, fat-specific weight loss (Fig. 6). To evaluate the weight-lowering effect of combined amylin/leptin agonism (with pramlintide/metreleptin) in human obesity, a 24-week, randomized, double-blind, active drug–controlled, proof-of-concept study was conducted in obese or overweight subjects (n = 177; 63% female; aged 39 ± 8 years; BMI 32.0 ± 2.1 kg/m²; weight 93.3 ± 13.2 kg; mean ± S.D.) (Ravussin et al., 2009). The study started with a 4-week lead-in period for the titration of pramlintide and dietary intervention. After the initial lead-in period with pramlintide (180 μg twice daily for 2 weeks, 360 μg twice daily thereafter) and diet (40% calorie deficit), subjects achieving 2%–8% weight loss were randomized 1:2:2 to 20 weeks of treatment with metreleptin (5 mg twice daily), pramlintide (360 μg twice daily), or pramlintide/metreleptin (360 μg/5 mg twice daily). Combination treatment with pramlintide/metreleptin led to significantly greater weight loss from enrollment to week 20 (−12.7% ± 0.9%; least-squares mean ± S.E.) than treatment with pramlintide (−8.4% ± 0.9%; P < 0.001) or metreleptin (−8.2% ± 1.3%; P < 0.01) alone (evaluable, n = 93). The greater reduction in body weight was significant as early as week 4, and weight loss continued throughout the study, without evidence of a plateau even at the 24-week end point. The most common adverse events with pramlintide/metreleptin were consistent with the two molecules studied and were mainly injection site events and nausea, which were mostly mild to moderate and decreased over time. These results further support the concept that combination treatment that targets separate physiologic pathways is a viable approach for the induction of clinically meaningful weight loss.

A functional magnetic resonance imaging study recently compared activation after the administration of metreleptin with the combination of metreleptin with pramlintide. 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 weight loss synergy between amylin and leptin agonism remains to be determined.

Based on the preclinical finding of amylin in combination with small molecule oral medications like phentermine or sibutramine, a proof-of-concept phase 2 human clinical study was conducted (Aronne et al., 2010). This study was a randomized placebo-controlled trial, and it examined the safety and efficacy of the amylin analog pramlintide alone or in combination with either phentermine or sibutramine. All patients enrolled in the study also received lifestyle intervention. After a 1-week placebo lead-in period, 244 obese or overweight subjects without diabetes (88% female; 41 ± 11 years; BMI 37.7 ± 5.4 kg/m²; weight 103 ± 19 kg; mean ± S.D.) received placebo (subcutaneously three times a day), pramlintide (120 μg subcutaneously three times a day), pramlintide (120 μg subcutaneously three times a day) plus oral sibutramine (10 mg every day before noon), or pramlintide (120 μg subcutaneously three times a day) plus oral phentermine (37.5 mg every day before noon) for 24 weeks. Treatment was single blind for subjects receiving subcutaneous medication only and open label for subjects in the combination arms. Weight loss achieved at week 24 with either combination treatment was greater than with pramlintide alone or placebo (P < 0.001; −11.1% ± 1.1% with pramlintide plus sibutramine, −11.3% ± 0.9% with pramlintide plus phentermine, −3.7% ± 0.7% with pramlintide, −2.2% ± 0.7% with placebo; mean ± S.E.). Given that phentermine and sibutramine all have potential to raise blood pressure or heart rate, this cardiovascular safety parameter was of particular importance in a population at higher risk for cardiovascular events. Elevations in heart rate and diastolic blood pressure from baseline were demonstrated with both pramlintide plus sibutramine (3.1 ± 1.2 beats/min, P < 0.05; 2.7 ± 0.9 mm Hg, P < 0.01) and pramlintide plus phentermine (4.5 ± 1.3 beats/min, P < 0.01; 3.5 ± 1.2 mm Hg, P < 0.001) using 24-hour ambulatory monitoring. However, the majority of subjects receiving these treatments remained within normal blood pressure ranges. These results validated the preclinical observation that amylin agonism combined with another pathway that targets weight management may have additive or synergistic effects.

As we have stated before, obesity is a complex disease and due to the evolutionary need for energy preservation, there are multiple pathways to prevent weight loss. Monotherapy in general does not elicit the amount of weight loss patients and health care providers are looking for. The most robust clinical data have come from combination therapies. However, the development of small molecule obesity drugs has had major safety issues in the past. The combination of phentermine and fenfluramine was taken off the market due to cardiac valvulopathy issues from fenfluramine. The drug rimonabant that leverages the cannabinoid pathway was also taken off the market for suicidal risks. The most recent obesity therapy to be taken off the market was sibutramine due to potential increase in cardiovascular events. Despite these setbacks, the U.S. Food and Drug Administration recently approved another selective serotonin agonist like fenfluramine called lorcaserin. As a monotherapy, it elicits a modest weight loss of around 2 kg (Fidler et al., 2011). Another therapy that was recently approved was a combination of phentermine and topiramate. Despite combining different pathways, the weight loss was also modest and the highest dose did elicit the most significant side effects (Gadde et al., 2011). Despite new approvals, the obesity treatment landscape is still very limited due to potential safety concerns (Kim et al., 2013). The recent U.S. Food and Drug Administration approval of the GLP-1 receptor agonist liraglutide for obesity (Saxenda; Novo Nordisk Inc., Plainsboro, NJ) provides a new opportunity for trialling combinations of pramlintide and liraglutide in patients (Ladenheim, 2015). The attraction of leveraging the physiologic pathway of amylin agonist is that it can potentially deliver a more clinically meaningful weight loss while minimizing the off-target adverse events.