Creating the amylin story

This paper is based on a presentation given at the Annual Meeting of the Society for the Study of Ingestive Behavior in July 2021 and provides a personal view on some of the milestones in the discovery of amylin as a constituent of pancreatic islet amyloid deposits, as a pancreatic beta-cell hormone, and on its role in physiology and pathophysiology. Only selected effects of amylin are discussed here because we recently published extensive reviews on the physiology and pathophysiology of amylin. Amylin was discovered as the main constituent of islet amyloid that is predominantly found in pancreatic islets in type 2 diabetics. These deposits, and in particular small oligomer aggregates of amylin seem to contribute to the progressive beta-cell damage seen in type 2 diabetics. Amylin is also a physiologically relevant circulating hormone with diverse metabolic functions, e.g. inhibition of eating, of pancreatic glucagon secretion and of gastric emptying. Knowledge of these types of functions and amylin's mechanisms of action lead to the development of amylin analogues that are now among the most promising anti-obesity targets in clinical testing. With this review, I want to give a short overview of 35 exciting years of amylin research.


The dichotomy of amylin in physiology and pathophysiology
The mature 37 amino acid peptide amylin is derived from the IAPP (islet amyloid polypeptide) gene and is produced in pancreatic beta-cells andin much lower amountsin other tissues, like the stomach, spinal ganglia and in the brain (Cooper, 1994;Hay, Chen, Lutz, Parkes, & Roth, 2015). Amylin is characterized by an interesting dichotomy because this peptide, on one hand, contributes significantly to the development of type 2 diabetes (T2D) by its propensity to aggregate into fibrils (Westermark, Andersson, & Westermark, 2011), but on the other hand also to the physiological control of metabolism . These dichotomous roles seem to be "functionally independent" in that some amylin forms aggregate into oligomers and eventually mature amyloid fibrils; this process likely involves multiple mechanisms and is independent of the cell membrane bound amylin receptor because aggregation is initiated intracellularly (Lorenzo, Razzaboni, Weir, & Yankner, 1994;Westermark et al., 2011). On the other hand, the soluble monomeric form of mature amylin activates the amylin receptor (AMY) in the brain to produce hormonal effects on glucose metabolism (inhibition of glucagon secretion, control of gastric emptying) and nutrient intake (induction of satiation), hence beneficial weight-lowering and anti-diabetic effects.

Time axis in amylin discovery
The first description of amylin-derived deposits in pancreatic islets dates back to 1900 when "hyaline appearance" was reported within islets from diabetic patients (Opie, 1901). In 1943 (Ahronheim, 1943) (and then confirmed in 1961 (Ehrlich & Ratner, 1961)), this extracellular material was histologically characterized as amyloid; similar deposits were described in diabetic cats in 1981 (Yano, Hayden, & Johnson, 1981a;1981b). A milestone was the characterization of the partial sequence of the main constituent of pancreatic islet amyloid in 1986; the underlying peptide was originally called insulinoma amyloid peptide (IAP), and was renamed later as islet amyloid polypeptide (IAPP) (Westermark, Wernstedt, Wilander, & Sletten, 1986). About at the same time, Cooper and co-workers Cooper et al., 1987) identified the sequence of the full 37 amino acid peptide from the pancreas of type 2 diabetics which was originally called diabetes-associated peptide (DAP) but soon renamed as amylin because it was realized that the presence of amylin is not restricted to T2D. I will use the term "amylin" for the rest of the article, realizing that some authors prefer the term "IAPP", in particular when a specific chemical structure is meant (Young, 2005d. Several groups identified the amylin gene (Mosselman, Höppener, Lips, & Jansz, 1989;Roberts et al., 1989;Sanke, Bell, Sample, Rubenstein, & Steiner, 1988) and it soon became clear that amylin is a beta-cell product, just as insulin, and is in fact co-packed and co-released with insulin Johnson et al., 1988;Nakazato, Asai, Kangawa, Matsukura, & Matsuo, 1989;Nakazato et al., 1990;O'Brien, Westermark, & Johnson, 1991). The discovery of amylin as a "normal" beta-cell hormone was soon followed by the first reports on amylin's eating inhibitory effect (Balasubramaniam, Renugopalakrishnan, Stein, Fischer, & Chance, 1991;Chance, Balasubramaniam, Chen, & Fischer, 1992;Chance, Balasubramaniam, Stallion, & Fischer, 1993;Chance, Balasubramaniam, Zhang, Wimalawansa, & Fischer, 1991;Morley, Flood, Horowitz, Morley, & Walter, 1994). Based on amylin's metabolic actions, the first full-length amylin analogue, pramlintide, was developed in 1995/1996 (Kong et al., 1997;Thompson, Pearson, & Kolterman, 1997) and still today is the only approved amylin-based drug (Symlin R ) for the treatment of T1D and T2D (Weyer, Maggs, Young, & Kolterman, 2001). Another key event in amylin discovery was the characterization of the structure of the receptors of the calcitonin family of peptides which, next to calcitonin, includes amylin, calcitonin gene-related peptide (CGRP), adrenomedullin and adrenomedullin 2, also called intermedin (Christopoulos et al., 1999;McLatchie et al., 1998;Muff, Buhlmann, Fischer, & Born, 1999). Since then, we know that the AMY consists of the calcitonin receptor core (CTR) which heterodimerizes with receptor-activity modifying proteins (RAMP) 1, 2 or 3 to form the AMY 1-3 receptors (Hay, Christopoulos, Christopoulos, Poyner, & Sexton, 2005;Hay, Garelja, Poyner, & Walker, 2018). About at the same time, it became clear that amylin's metabolic actions are mediated by direct activation of the central nervous system, in particular the area postrema (AP) which is characterized by an open blood brain barrier Lutz, Senn, Althaus, Del Prete, Ehrensperger, & Scharrer, 1998).
Our own laboratory actively contributed to the increase in knowledge about amylin's physiological and pathophysiological role in metabolism. To give some examples, we were the first to quantify amylin-derived amyloid deposition in diabetic cats (Lutz & Rand, 1993, showed for the first time that amylin is a satiating hormone and reduces eating by a meal size effect (Lutz, Geary, Szabady, Del Prete, & Scharrer, 1995), that this effect is of physiological relevance (Mollet, Gilg, Riediger, & Lutz, 2004), and that amylin's effect does not seem to depend on afferent nerve signalling to the brain but rather a direct brain effect (Lutz, Althaus, Rossi, & Scharrer, 1998;Lutz, Del Prete, & Scharrer, 1994. Our studies which were then confirmed by other laboratories (Braegger, Asarian, Dahl, Lutz, & Boyle, 2014;Edwards et al., 1998;Mack et al., 2010;Smith et al., 2016) indicated that the AP plays a predominant role in mediating amylin's effect in the homeostatic control of eating Lutz, Senn, et al., 1998). Based on our findings that amylin also has characteristics of an adiposity signal Rushing, Hagan, Seeley, Lutz, & Woods, 2000), we subsequently showed that amylin synergizes with insulin and leptin (Osto, Wielinga, Alder, Walser, & Lutz, 2007); in particular the latter effect led to a large number of studies indicating that amylin seems to act as a sensitizer to leptin action, most likely via interaction in the hypothalamus . We were also able to define neurotransmitters in the caudal hindbrain which are critical to mediate amylin's effect Braegger et al., 2014;Lutz, Del Prete, Walzer, & Scharrer, 1996;Lutz, Tschudy, Mollet, Geary, & Scharrer, 2001;Mollet et al., 2001;Potes, Turek, et al., 2010), and early on identified the lateral parabrachial nucleus (LPBN) as an important projection site of amylin activated AP neurons (Becskei, Grabler, Edwards, Riediger, & Lutz, 2007;Riediger, Zuend, Becskei, & Lutz, 2004). Interestingly, the primary AP dependent pathway activated by amylin seems to differ from pathways activated by other eating inhibitory stimuli (Zuger, Forster, Lutz, & Riediger, 2013), hence there seems to be a differentiation of processing of various incoming signals already at the level of the AP. Even though the structure of the AMY had been known for many years, it was not until 2016 that all necessary AMY components were co-localized in single neurons in native brain tissue . Finally, we gave the first indications of AMY subtype specific actions of amylin, with the AMY3 receptor most likely mediating amylin's effect on eating and glucose metabolism, and the AMY1 being involved in the control of adiposity (Coester, Pence, et al., 2020).

Amylin, amyloidogenesis and type 2 diabetes mellitus
In certain species and under certain pathological conditions such as T2D or in insulinoma, amylin can self-aggregate and eventually form insoluble amylin amyloid plaques. This property is mainly seen in amylin from humans, other primates and cats, but not in rodent forms of amylin (Westermark et al., 2011).

Mechanisms of amyloidogenesis
The amino acid sequence of amylin is highly conserved among mammalian species. Formation of amylin derived amyloid is favored by differences in the amino acid sequence in an amyloidogenic region of the molecule that is essential for amyloid formation and toxicity (Betsholtz, Johnson, & Westermark, 1989;Hoppener et al., 1993;Matveyenko & Butler, 2006a, 2006bMoriarty & Raleigh, 1999); this heterogeneity mainly depends on the amino acid residues 20-29 of the amylin sequence and influences the formation of ordered secondary structures such as the β-sheet conformation as an essential prerequisite for amyloidogenesis.
Clear data about the first intracellular localizations of the aggregates are still missing but our own unpublished data indicate that amylin oligomer aggregates can be found just outside of secretory granules in the cytoplasma of beta-cells in diabetic cats; such aggregates were not seen in healthy control cats. Similarly, amyloid deposits in baboons were observed in the cytoplasm of β-cells as well as bound to the outer β-cell membrane (Guardado-Mendoza et al., 2009). However, amylin-derived amyloid is also observed in the endoplasmic reticulum (ER), Golgi apparatus and secretory vesicles (Westermark et al., 2011).

Pathological effects of amylin deposits on pancreatic β-cells
There is consensus in that abnormalities in the processing of amylin and its deposition as amyloid in the islets contribute to the progressive loss of pancreatic β-cells in T2D and feline diabetes Henson & O'Brien, 2006;Jurgens et al., 2011;Westermark et al., 2011). The toxic effects induced by amylin amyloid deposits depend on a combination of mechanisms that act both intra-and extracellulary, and are in fact mechanistically similar to the intracellular processes that lead to neuronal damage by protein aggregates in Alzheimer disease (Lorenzo et al., 1994;Lutz & Meyer, 2015;Westermark et al., 2011).
Amylin-derived beta-cell toxicity is associated with ER stress and the unfolded protein response (UPR) (Preston, Gurisik, Bartley, Laybutt, & Biden, 2009). When degradation of misfolded proteins fails, protein aggregation and amyloidogenesis may eventually lead to the induction of cell apoptosis (Westermark et al., 2011). These processes may be operational especially during the early phase of T2D, during which amylin and insulin syntheses are markedly increased as a consequence of insulin resistance.
Once fibrillar material is formed, this may lead to a disruption of the vesicles and result in the release of fibrillar material into the cytosol. Such aggregates can be dissolved by autophagy but in the early phase of T2D, the formation of autophagolysosomes seems to be attenuated, which in turn causes the accumulation of material in autophagic vacuoles, inducing a cascade of intracellular toxic effects (Westermark et al., 2011). Finally, toxic oligomers consisting of pro-amylin or amylin during the early phases of T2D may disrupt mitochondrial function, increased cellular oxidative stress and the formation of reactive oxygen species (ROS) and ultimately damage to cell structures and cause cell loss (Chen, Taylor, & Verchere, 2018;Marzban et al., 2004;Westermark et al., 2011). All these processes can also activate the inflammasome eventually contributing to beta-cell death (Donath, 2013;Donath, Dalmas, Sauter, & Boni-Schnetzler, 2013).

Amylin aggregates as a target for T2D therapy
Based on the important role of amylin aggregates in damaging pancreatic beta-cells in T2D, several strategies have been developed that aim at reducing the pathological consequences of these aggregates. As one strategy, recent studies indicated that antibodies targeting aggregated but not monomeric amylin may produce beneficial metabolic effects and maintain functional beta-cell mass. One such approach was based on the generation of an active vaccination against amylin aggregates which was induced by coupling such peptides to virus-like particles. This vaccination lead to the generation of specific antibodies which reduced the deposition of islet amyloid in human-amylin transgenic mice and delayed the onset of hyperglycemia, associated with a reduced pro-inflammatory response in pancreatic islets (Roesti et al., 2020). In a different approach, passive vaccination with antibodies targeting pathologic oligomer amylin aggregates (but not monomeric amylin) showed similar effects, in that normal beta-cell function was better maintained and the occurrence of diabetes symptoms was reduced (https://www.ne urimmune.com/research-development/type-2-diabetes).

Amylin and insulin co-secretion from the pancreatic beta-cell, and other sources of peripheral amylin
Early after the discovery of amylin, it became clear that the pancreatic beta-cells are the main source of circulating amylin (Cooper, 1994). Amylin is co-packaged with insulin into the same secretory vesicles, and the same physiological stimuli lead to their co-release. Hence, circulating plasma amylin which ranges from 3 to 5 pM in the fasting state, increases post-prandially to concentrations of approx. 15-25 pM in rats (Boyle, Rossier, & Lutz, 2010. Further, the incretin glucagon-like peptide-1 (GLP-1) not only increases insulin but also amylin levels (Inoue, Hisatomi, Umeda, & Nawata, 1991). Other sources of peripheral amylin are e.g. neuroendocrine cells along the gastrointestinal tract and dorsal root ganglia, but the release of amylin from these cells and their physiological relevance has been poorly studied (Ferrier et al., 1989;Mulder et al., 1995;Nicholl, Bhatavdekar, Mak, Girgis, & Legon, 1992).

Amylin receptor pharmacology
Amylin receptors are heterodimers of the CTR which is a G-protein coupled receptor, and one (or several) RAMPs being co-expressed in the same cell (Bower & Hay, 2016;Hay et al., 2015;Hay et al., 2016;Morfis et al., 2008;Poyner et al., 2002;Tilakaratne, Christopoulos, Zumpe, Foord, & Sexton, 2000). The CTR has various spice variants, the variant CT (a) is the best characterized. The CTR core receptor changes its specificity and affinity for amylin depending on the co-expression of one of the three RAMPs in the same cell (Christopoulos et al., 1999;Fischer, Muff, & Born, 2002;McLatchie et al., 1998;Muff et al., 1999); in other words, RAMPs change CTR pharmacology from calcitonin-preferring to amylin-preferring receptors by affecting receptor specificity and affinity, and RAMPs may regulate CTR transport to the cell surface (Hay et al., 2015).
Three RAMPs have been defined so far, resulting in the known AMY 1-3 receptors. In a study showing for the first time the presence of CTR and RAMPs in single cells of native brain tissue, we also demonstrated that many AP neurons seem to express more than one RAMPs, at least at the level of mRNA . Theoretically, this could result in even more AMY subtypes with different RAMPs associating with CTR in the same cell, but the potential pharmacology of such putative receptors is currently unknown.
The close relationship of AMY to the CTR and the calcitonin-like receptor (CLR ) makes it imperative to interpret data on amylin pharmacology with caution. Currently, no receptor-subtype specific antagonists exist, and no reliable RAMP antibodies are available that could e.g. be used for immunohistochemical double labeling studies in specific brain areas to demonstrate the co-expression of amylin receptor components in single cells at the protein level.
Various amylin agonists that are under development for the treatment of obesity seem to be less specific for AMY than the only approved amylin-based anti-diabetes drug pramlintide which has a similar AMY specificity and pharmacodynamic as native amylin (Weyer et al., 2001). Most newer agonists seem to qualify as dual receptor (CTR and AMY) agonists (Andreassen et al., 2014;Andreassen et al., 2021;Fletcher et al., 2021;Gydesen et al., 2017;Thomas Kruse et al., 2020). I believe that it is not yet clear whether these characteristics are necessary for improved body weight lowering activities; interestingly, human and rat calcitonin which activate the CTR but not AMYs, have little efficacy to reduce eating Lutz, 2010a).

Caudal hindbrain with the AP and the nucleus of the solitary tract (NTS)
The caudal hindbrain, and in particular the AP seem to play a critical role for the direct mediation of peripheral amylin's effects. The AP which can be directly targeted by circulating amylin Shapiro & Miselis, 1985;van der Kooy & Koda, 1983), has a high density of amylin receptors (Sexton, Paxinos, Kenney, Wookey, & Beaumont, 1994) and the presence of the core receptor plus RAMPs provides the structural basis for a direct action of amylin on AP neurons . Functional evidence from different laboratories support the idea that the AP is critically implicated in mediating amylin's satiating effect, its effect to reduce glucagon secretion and to slow gastric emptying (Edwards et al., 1998;Lutz et al., 1994;Lutz, Senn, et al., 1998;Mack et al., 2010). A recent study suggested that amylin receptors may also be located on presynaptic terminals of AP neurons, and that their activation may trigger glutamate release acting on other AP neurons (Fukuda, Hirai, Maezawa, Kitagawa, & Funahashi, 2013). It is however not clear how the effects of presynaptic terminals may be linked to postsynaptic mechanisms that have been demonstrated convincingly (Riediger, Schmid, Lutz, & Simon, 2002), and whether they are physiologically relevant.
Several recent studies using modern state-of-the-art technology characterized caudal hindbrain neurons and defined specific clusters of cell populations (Dowsett et al., 2021;Ludwig et al., 2021;Zhang et al., 2021). Interestingly, these studies confirmed previous findings that amylin responsive neurons may be a rather specific population that does not co-express receptors for GLP-1 and GDF-15. The latter two peptides (Baraboi, St-Pierre, Shooner, Timofeeva, & Richard, 2011;Borner et al., 2020;Thiele et al., 1997), but not amylin (L. B. Boccia, T; Ghidewon, M.Y.; Kulka, P; Piffaretti, C; Doebley, S.A.; De Jonghe, B.C.; Grill, H.J.; Lutz, T.A.;Le Foll, C., 2022;Lutz, Geary, et al., 1995;Rushing, Seeley, Air, Lutz, & Woods, 2002), are known to induce aversive responses. These studies may therefore provide the explanation why stimuli targeting the AP may induceat least in partnon-overlapping effects. Interestingly and different from the AP, neurons carrying both the CTR and the GLP-1 receptor seem to overlap in the NTS (Ludwig et al., 2021). In some of our own studies, we observed however that the total expression of CTR seems to be low compared to AP neurons ; because NTS receptors are not as easily accessible as AP receptors, the relevance of these receptors was therefore not clear to us.
However, a specific role of CTR activation in the NTS in the control of energy balance has been suggested in a recent study (Cheng et al., 2020); the results suggested a direct NTS -LPBN pathway, leading to the activation of a population of LPBN neurons that mediate a decrease in eating. The potential interaction between these neurons and AP target neurons, and their respective projections to the LPBN needs to be defined in future studies. In addition, the network of CTR-positive neurons throughout the brain (Becskei, Riediger, Zund, Wookey, & Lutz, 2004;Cheng et al., 2020;Gonzalez et al., 2021;Pan et al., 2018a) and their respective role in eating control needs to be defined in the future. In other words, it will be important to understand whether and how the different neuronal populations with CTR interact in the control of eating and other metabolic actions of amylin and its related peptides, and whether peripheral ligands can directly access neurons outside of the circumventricular organs.

Lateral parabrachial nucleus
We had shown early on that an intact LPBN is necessary for peripheral amylin to reduce eating (Becskei et al., 2007), and our follow-up studies indicated an important role for direct projections from amylin activated AP neurons bypassing the NTS to the LPBN (Potes, Lutz, & Riediger, 2010). We recently started to elaborate on the different neuronal populations in the LPBN in mediating amylin's effect on eating. It had been reported that a large proportion of the amylin activated LPBN neurons are CGRP positive, and the large overlap of CGRP neurons with amylin-induced Fos in the LPBN indicated a potential functional role of these neurons (Carter, Soden, Zweifel, & Palmiter, 2013). Despite the role of CGRP neurons to control meal ending satiation (Campos, Bowen, Schwartz, & Palmiter, 2016) and meal size control under normal conditions (Campos et al., 2016) but also e.g. following bariatric surgery (Mumphrey et al., 2016) or in certain disease states (Campos et al., 2017;Carter, Han, & Palmiter, 2015), these neurons seem to play a minor role in mediating amylin's effect to reduce eating. Using various approaches, we recently showed that the silencing of CGRP neurons in the LPBN did not impair the effect of amylin to reduce eating. Interestingly, at least part of the eating inhibitory effect of the amylin agonist salmon calcitonin (sCT) appeared to be mediated by these neurons. Further, specifically knocking down the expression of CGRP in these LPBN neurons did not reduce amylin's or sCT's ability to decrease eating. Finally, the same studies also indicated an important difference between amylin and sCT in that sCT, but not amylin, induced conditioned taste avoidance in rats and induced a vomiting response in the musk shrew. Hence, overall the LPBN plays an important role in mediating the effect of amylin and sCT on eating but the critical neurotransmitter of LPBN neurons needs to be characterized; it has been hypothesized that glutamatergic neurotransmission may be involved but there are no functional data yet supporting this hypothesis.

Other brain areas mediating the actions of amylin
Several other brain areas have been investigated for specific actions of amylin. Despite the widespread expression of an amylin induced cFos signal, the wide distribution of amylin binding sites throughout the brain (Sexton et al., 1994) and the expression of all critical amylin receptor components in many brain areas Ueda, Ugawa, Saishin, & Shimada, 2001), it is in most cases not clear whether peripheral amylin may directly reach these sites to induce biological effects.

Amylin action on specific neuronal populations in the hypothalamus
Based on numerous experiments, we developed a working model of amylin action in the hypothalamus, indicating a direct action of amylin on proopiomelanocortin (POMC) neurons and an indirect effect on neuropeptide Y/Agouti related peptide (NPY; AgRP) neurons in the hypothalamic arcuate nucleus (ARC), the latter effect possibly being mediated by interleukin-6 (IL-6) (Levin & Lutz, 2017;Lutz et al., 2018). Interestingly, other studies indicated that CTR positive and hence amylin sensitive neurons seem to co-localize with NPY rather than POMC in the ARC (Pan et al., 2018b), and some of our own data also indicated that sCT may bind and actually be taken up by NPY/AgRP, but not POMC neurons in the ARC (Zakariassen, John, & Lutz, 2020;Zakariassen, John, Lykkesfeldt, et al., 2020). Finally, amylin has recently been shown to directly suppress AgRP neurons based on a reduction of intracellular calcium signaling (Su, Alhadeff, & Betley, 2017). Hence, while the principal involvement of hypothalamic signaling via POMC and NPY/AgRP neurons in mediating amylin's effects on eating and energy expenditure has been supported by various studies, the exact pathways still need to be defined in future experiments.

Central reward areas including the ventral tegmental area (VTA) and the nucleus accumbens (NAc)
Various groups studied effects of amylin and its agonists on reward mediated actions. These studies have been elegantly summarized in recent reviews (Kern & Mietlicki-Baase, 2020;E.G.;Mietlicki-Baase & Hayes, 2014). Briefly, amylin activation of the VTA and the NAc seems to be linked to a decrease in the rewarding effect of food (Mietlicki-Baase et al., 2013) but also of alcohol intake (Kalafateli, Vallof, & Jerlhag, 2019). The exact pathway how these centers are activated is still a matter of debate because both direct actions (Mietlicki-Baase et al., 2017) and an indirect effect subsequent to AP activation (Whiting, McCutcheon, Boyle, Roitman, & Lutz, 2017) have been claimed.

Amylin-leptin interactions in the regulation of energy homeostasis
The interest of the scientific community in amylin research was boostered by the finding of a sensitizing effect of amylin towards the action of leptin. Early studies indicated a functionally relevant interaction between amylin and leptin because animals with defective leptin signaling (e.g., ob/ob mice, Zucker fa/fa rats) reacted less to the eating inhibitory effect of sCT (Eiden, Daniel, Steinbrueck, Schmidt, & Simon, 2002). We subsequently showed that central leptin administration increased the effect of peripheral amylin to acutely reduce eating (Osto et al., 2007). A large series of experiments then investigated the type of interaction, mode of interaction and potential site(s) of interaction between amylin and leptin (Chan, Roth, & Weyer, 2009;Roth et al., 2008a;Roth, Trevaskis, Turek, & Parkes, 2010;Trevaskis et al., 2008;Trevaskis, Lei, et al., 2010;Trevaskis et al., 2016;Turek et al., 2010). The relevant findings have recently been reviewed extensively (Hay et al., 2015;Levin & Lutz, 2017). Briefly, an interaction between amylin and leptin seems to occur at several levels, i.e. in the caudal hindbrain Smith et al., 2016), the hypothalamus  and the VTA (Mietlicki-Baase, Olivos, Jeffrey, & Hayes, 2015). In the caudal hindbrain, mice and rats with a defect in the leptin signaling pathway (i.e. leptin receptor deficient db/db mice and Zucker fa/fa rats) had less amylin induced c-Fos and CTR expression in the AP (Duffy, Lutz, & Boyle, 2018). In the mediobasal hypothalamus, the effect of chronic amylin administration to reduce food intake and increase energy expenditure in leptin resistant DIO rats was associated with an increased expression of the POMC gene in the ARC Turek et al., 2010). The enhanced response to leptin by amylin was paralleled by increased leptin receptor activation in the ventromedial nucleus of the hypothalamus (VMN) and the ARC, and increased leptin receptor binding in the ARC and VMN . Using diet-induced obese rats, we showed that the amylin-leptin interaction in the ARC/VMN is probably mediated by IL-6, increasing LepRb expression and potentially also its transport to the cell membrane (Le Foll et al., 2015). Similar synergizing effects on eating and weight loss were reported in overweight humans receiving amylin and leptin analogues, respectively (Roth et al., 2008b).

Role of amylin in pathway development in the caudal hindbrain and the hypothalamus
Two important findings triggered research on a potential role of amylin in the development of neuronal pathways. First, amylin itself was shown to be a growth factor for the development of various organs, including the kidneys (Wookey, Lutz, & Andrikopoulos, 2006). Second, the similarities between amylin and leptin and their synergizing effect in the control of eating raised the question whether amylin, similar to leptin (Bouret, Draper, & Simerly, 2004a, 2004bBouret & Simerly, 2006), may also influence brain development.
In fact, we showed that amylin and amylin signaling are involved in the pathway development between the AP and the NTS (Abegg et al., 2017), and also in the development of projections from the hypothalamic ARC to the PVN (Coester, Koester-Hegmann, Lutz, & Le Foll, 2020;Lutz et al., 2018). Our studies suggested that the effect on the AgRPergic pathway may be indirect, and both this indirect and the direct effect on POMC neurons are mediated by specific AMY (Coester, Koester-Hegmann et al., 2020).
Several studies also suggested that amylin promotes neurogenesis in the hippocampus  and the caudal hindbrain (Liberini, Borner, Boyle, & Lutz, 2016). Interestingly, while amylin was also shown to promote cell proliferation during embryogenesis, the effect appeared to be rather specific on microglia than on neuronal cells (Lutz & Le Foll, 2019). These interesting findings left many questions open. E.g., about the exact timing of the effect (i.e. the critical period pre-or postnatally to influence these effects), the nature and the origin of the endogenous ligand mediating the effects, and potential differences between the action of endogenous versus exogenous amylin that need to be elaborated. In the same context, it needs to be studied whether some of the defects caused by the lack of endogenous amylin, such as e.g. in the caudal hindbrain (Abegg et al., 2017), can be rescued by amylin administration.

Central amylin synthesis
In recent years, progress has been made in respect to the presence of brain-derived amylin and its potential metabolic role. Peptides with similarity to amylin had been described a long time ago in the brain (Fischer, Tobler, Henke, & Tschopp, 1983) but it was not until about 10 years ago that the first convincing findings about central amylin expression were published (Dobolyi, 2009;Szabo, Cservenak, & Dobolyi, 2012). This was supported by a recent study with an extensive characterization of neurons in various hypothalamic areas (Li, Kelly, Heiman, Greengard, & Friedman, 2015). Pertinent findings have been summarized recently (Boccia, Gamakharia, et al., 2020).
These data indicate that amylin is expressed in various nuclei and that there may be a marked sexual dimorphism (Li et al., 2015). Amylin expression seems to be particularly strong in the medial preoptic area (MPOA) in lactating rat dams (Dobolyi, 2009). This was recently confirmed in our own studies. The physiological role of centrally synthesized amylin in general and of amylin in the MPOA is only partly understood (Dobolyi, 2009;Szabo et al., 2012;Szabó et al., 2015). Some studies indicate that CTR signaling in the MPOA influences maternal behavior (Yoshihara et al., 2021), but it is unclear whether this effect is based on an autocrine effect within this brain area. Future experiments will have to define how central amylin expression is regulated, whether central amylin is secreted to act as a neurotransmitter, where the pertinent neurons project to, and what their role in physiology or pathophysiology is.

AMY subtypes
All AMY components are expressed in numerous brain regions. However, the heterodimeric nature of the AMY, the lack of receptor subtype specific antibodies and the lack of specific RAMP antibodies are some hurdles which may explain why the exact cellular mediation of amylin-induced effects is still rather unclear. A further complicating factor is the fact that RAMPs can bind other peptides when combined with other GPCR (Barbash, Lorenzen, Persson, Huber, & Sakmar, 2017;Barbash et al., 2019), in addition to the CTR and the calcitonin receptor-like receptor (CLR). Finally, the expression of some receptor components seems to be rather dynamic. In the AP, e.g. exogenous amylin reduced the expression of RAMP1 and RAMP3 mRNA, but not CTR, and increased RAMP2 expression . Hence, it cannot be excluded that the relevance for specific AMY subtypes may vary depending on the experimental conditions. Nonetheless, some progress has been made recently and the data support previous assumptions that the AMY3 receptor may be particularly relevant for amylin's effects on eating. Using RAMP knockout animals, we recently provided evidence that the AMY3 receptor seems to be mainly responsible for the eating inhibitory and glucose regulatory effects of amylin, and that the AMY1 may be particularly relevant for the control of fat utilization (Coester, Pence, et al., 2020). The latter findings add to previous reports that RAMP1 overexpressionpresumably by increasing endogenous amylin action at the AMY1 receptorincreases energy expenditure in mice (Fernandes-Santos et al., 2013;Zhang et al., 2011).

Pre-clinical studies with amylin as an anti-obesity target
Various amylin analogues have been developed and tested in preclinical models for their effect on body weight and the control of glucose metabolism (Andreassen et al., 2014;Andreassen et al., 2021;Baggio & Drucker, 2020;Fletcher et al., 2021;Gydesen et al., 2016;Gydesen et al., 2017;Hay et al., 2015;Henriksen et al., 2021;Jorsal, Rungby, Knop, & Vilsboll, 2016;Thomas Kruse et al., 2020). Many of the compounds tested can be classified as dual agonists at the calcitonin and the amylin receptors because their in vitro characteristics imply the potential to activate both receptors. This is interesting because CTR activation by mammalian (human or rat) calcitonin per se seem to have a minor effect on eating. This is in contrast to fish (salmon or eel) calcitonin which inhibit eating and which strongly activate the AMY(Le Lutz, 2010b). How exactly, i.e. by which cellular mechanisms, CTR plus AMY activation by the dual receptor agonists leads to a strong eating and weight reducing effect is currently unknown. It is also unknown whether the effects of the dual receptor agonists on eating depend on the activation of CTR plus AMY in the same target cells or whether CTR and AMY in different brain areas need to be activated simultaneously to achieve the strong biological effects.
Some of the most recent studies tested the involvement of AMY receptor subtypes in the mediation of long acting amylin analogues. These studies indicated that the relatively AMY specific agonist NN1213 also requires the AMY3 for its eating inhibitory and weight reducing effect, similar to native amylin (Arrigoni et al., 2021). Interestingly, the same study showed the CTR and AMY agonist sCT did not lead to sustained weight loss in mice, which is different from rats. Hence, in addition to the unknown cellular mechanisms and interactions that contribute to the effects of dual CTR and AMY agonists, species differences may also need to be considered and need to be elaborated in future studies.

Clinical studies using amylin agonists in obesity therapy
The only approved amylin-based drug is the amylin analogue pramlintide which has similar characteristics and a similar pharmacodynamic and pharmacokinetic profile than (rat) amylin. Early studies had indicated a weight lowering profile of pramlintide but the absolute effect was modest (Aronne et al., 2007;Chapman et al., 2005;Fineman, Weyer, Maggs, Strobel, & Kolterman, 2002;Fineman, Koda, et al., 2002;Hollander et al., 2004). However, pramlintide given together with the leptin analogue metreleptin produced a very strong decrease in body weight, presumably by enhancing leptin sensitivity similar to what had been observed in preclinical studies (Ravussin et al., 2009).
More recent studies indicated that several amylin-based drugs are safe and highly effective in people. The rather unspecific CTR and AMY agonist AM833 (cagrilintide) was shown to produce sustained and marked weight loss, and that its effect when combined with the GLP-1 agonist semaglutide was even stronger and bypassed the effect of most other pharmacotherapies (Becerril & Frühbeck, 2021;David et al., 2020;Enebo et al., 2021;Thomas Kruse et al., 2020). Amylin analogues are therefore considered an interesting strategy for the further development of highly effect weight lowering agents.

Summary
Most research focused on amylin's effect on eating, glucagon secretion and gastric emptying. Based on these findings, amylin-based drugs are in clinical use or in clinical testing and have shown promise to be a new generation of weight lowering compounds. Interestingly, some biological effects of amylin described early after its discovery (inhibition of insulin secretion, reduction of insulin sensitivity) have attracted very little attention in the last 20 years, even though antagonist studies had suggested effects of potential physiological relevance (Young, 2005a;2005b;2005c. However, it needs to be mentioned that the role of the AMY, as we know it since the discovery of the heterodimers CTR/RAMPs, has never been tested for these effects so that it is unclear whether such effects are truly amylin mediated. Next to these metabolic effects, amylin in primates and cats has a second "face" in that it can aggregate and form insoluble oligomers and fibrils that contribute to beta-cell loss in T2D and feline diabetes. Recent research efforts tried to diminish the load of amylin aggregates in an attempt to reverse this factor that causes beta-cell dysfunction in T2D (Roesti et al., 2020).
In summary, this review gave an admittedly personal view on the amylin story which started in 1986 with its (partial) sequence analysis. Since then, we tremendously increased our knowledge on the physiology and pathophysiology of amylin.

Acknowledgement
I gratefully acknowledge many of my colleagues and friends who helped me perform my research on the physiology and pathophysiology of amylin. I also gratefully acknowledge the financial support from many funding sources that helped me perform my research, in particular the Swiss National Science Foundation, the National Institutes of Health, the EU Frame Program 7, and the University of Zurich. And finally, my sincere gratitude to the Society for the Study of Ingestive Behavior for awarding me the Hoebel Prize of Creativity in 2021.