Glucagon‐like peptide‐1 (GLP‐1) signalling in the brain: From neural circuits and metabolism to therapeutics

Glucagon‐like‐peptide‐1 (GLP‐1) derived from gut enteroendocrine cells and a discrete population of neurons in the caudal medulla acts through humoral and neural pathways to regulate satiety, gastric motility and pancreatic endocrine function. These physiological attributes contribute to GLP‐1 having a potent therapeutic action in glycaemic regulation and chronic weight management. In this review, we provide an overview of the neural circuits targeted by endogenous versus exogenous GLP‐1 and related drugs. We also highlight candidate subpopulations of neurons and cellular mechanisms responsible for the acute and chronic effects of GLP‐1 and GLP‐1 receptor agonists on energy balance and glucose metabolism. Finally, we present potential future directions to translate these findings towards the development of effective therapies for treatment of metabolic disease. LINKED ARTICLES This article is part of a themed issue on GLP1 receptor ligands (BJP 75th Anniversary). To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v179.4/issuetoc


| GLP-1: FROM DISCOVERY TO USE AS AN OBESITY AND DIABETES THERAPEUTIC
The discovery of the first hormone, secretin in 1902, led to the search for other gut hormones that could stimulate pancreatic secretion. However, the concept or existence of such an incretin was met by scepticism and controversy over the following century (reviewed in Holst, 2019;Müller et al., 2019;Rehfeld, 2018). Briefly, the development of radioimmunoassay methods for insulin as well as glucagon greatly advanced the detection of these hormones and allowed for indepth investigations on how they are regulated in normal physiology as well as disease (Scherer & Newgard, 2020). The ability to directly and reliably measure circulating insulin levels led to the description that oral glucose administration results in considerably higher insulin responses when compared with intravenous glucose, heralding in a renewed interest in incretins. The first identified incretin hormone was gastric inhibitory polypeptide (GIP), also known as glucose-dependent insulinotropic polypeptide. Subsequent and parallel interests in the hormone glucagon suggested the existence of another incretin hormone later identified as a pro-hormone, pro-glucagon structure with two glucagon-like-peptides. This ultimately led to the discovery of glucagon-like-peptide 1 (GLP-1).
GLP-1 is a small peptide hormone, which is a post-translational cleavage product of the preproglucagon encoded gene, GCG. GLP-1 is produced by intestinal L-cells and also by a discrete population of neurons in the caudal medulla ( Figure 1) (Drucker, 2018;Larsen, Tang-Christensen, Holst, & Orskov, 1997). The multiple physiological effects of GLP-1 make it a viable candidate for diabetes mellites and obesity therapies. In particular, GLP-1 has potent effects on blood glucose by either stimulating glucose induced insulin release or inhibiting glucagon secretion (Drucker, 2018;Holst, 2019;Müller et al., 2019), both of which limit hepatic glucose production, which has been associated with hyperglycaemia in type 2 diabetic patients. Additionally, GLP-1 suppresses appetite and food intake (Drucker, 2018;Shah & Vella, 2014;van Bloemendaal et al., 2014). However, the beneficial effects of GLP-1 were limited in clinical trials of type 2 diabetic patients due to the very short half-life of GLP-1, which is approximately 2-5 min, via degradation by the enzyme dipeptidyl peptidase 4 (DPP-4) (Holst, 2019;Müller et al., 2019). Orally active inhibitors of DPP-4 and long-acting injectable/oral analogues of GLP-1 (e.g. exendin-4, sitagliptin, liraglutide, semaglutide, albiglutide, dulaglutide and others) were subsequently developed to enhance the efficacy of GLP-1 (Holst, 2019;Kanoski et al., 2016;Müller et al., 2019). Physiological and pharmacological data have shown that activation of the GLP-1 receptors promotes insulin secretion from pancreatic beta cells and also causes weight loss and thus representing a significant pharmacological target for the treatment of type 2 diabetes (Drucker, 2018;Kanoski et al., 2016). Importantly, these effects are shared across species from rodents to humans, as peripheral GLP-1 administration to normal and diabetic human subjects induced satiety and reduced food intake in short term studies (Flint et al., 1998;Gutzwiller et al., 1999;Toft-Nielsen et al., 1999;Verdich et al., 2001). Additionally, chronic GLP-1 or GLP-1 mimetic administration to human diabetic subjects was associated with improvements in glycaemic control and a modest 1.5-5.3 kg weight loss over a period of 0.5-3 years (Klonoff et al., 2008;Riddle et al., 2006). As GLP-1 receptor agonists are effective anti-diabetic/ weight control agents and their use is rapidly expanding. It is critical to understand how GLP-1 mediates beneficial effects on food intake/body weight and glucose homeostasis in order to develop therapies with potential for even greater efficacy and tolerability in patients. This review highlights the emerging findings that illustrate how GLP-1 receptor signalling in the CNS reduces both food intake and body weight (with an emphasis on GLP-1 action within the hypothalamus).

| DISTRIBUTION AND REGULATION OF THE GLP-1 SYSTEM
Although GLP-1 deficiency maybe unlikely to contribute to impaired insulin action in type 2 diabetes , low GLP-1 levels are a potential risk factor in the development of type 2 diabetes (Lastya et al., 2014). Weight gain may also lead to dysregulation of the GLP-1 system and contribute to the maintenance of metabolic dysfunction (Ranganath et al., 1996). Moreover, type 2 (non-insulin-dependent) diabetic patients exhibit a reduced incretin effect (Nauck et al., 1986). Thus, better defining the factors that regulate the production and release of GLP-1, along with the physiological F I G U R E 1 Sagittal view of a murine brain depicting select nucleus tractus solitarius (NTSjGLP-1 projections discussed within the review. Endogenous GLP-1 (left) and GLP-1 receptor analogues (GLP-1As) (right) act on a variety of brain regions. Abbreviations: AMG, amygdala; ARC, arcuate nucleus; BNST, bed nucleus of the stria terminalis; DMH, dorsomedial hypothalamus, DMV, dorsal motor nucleus of the vagus; GLP-1, glucagon like peptide 1; GLP-1RAs, GLP-1 receptor agonists; NAc, nucleus accumbens; NTS, nucleus tractus solitarius; PBN, parabrachial nucleus; PVH, paraventricular hypothalamic nucleus; VLM, ventrolateral medulla and VTA, ventral tegmental area conditions associated with the rise and fall of GLP-1 levels may advance our understanding of GLP-1 physiology in metabolic disease.
Secretion of GLP-1 is regulated by multiple factors involved in homeostasis Müller et al., 2019). Some of these factors overlap with the regulation of both the peripheral and central GLP-1 system. For instance, in a fasted or interprandial state, GLP-1 is secreted from enteroendocrine L-cells at low levels (Orskov et al., 1996). Owing to the incretin actions of GLP-1, GLP-1 secretion is increased and circulating levels are elevated several folds in magnitude in response to a meal, which contributes to increased insulin secretion (Campbell & Drucker, 2013;Drucker & Nauck, 2006). Elevated blood levels of GLP-1 following a meal may depend upon mechanical forces such as gastric distension (Müller et al., 2019;Rowlands et al., 2018).
Within the CNS, hindbrain GLP-1 neurons project to numerous brain regions relevant to metabolic regulation in mice ( Figure 1) (Burcelin et al., 2009;Ghosal et al., 2017;Llewellyn-Smith et al., 2011;Rinaman, 2010). Not surprisingly, these regions also express GLP-1 receptors supporting a potential redundancy of downstream targets for effects of peripheral and/or central GLP-1 ( Figure 2) (Alhadeff et al., 2014;Brierley et al., 2020;Kanoski et al., 2016). However, it is also possible that these systems may play subtle and different roles suggesting that it is not just a matter of simple redundancy (Brierley et al., 2021).
Activation of nucleus tractus solitarius GLP-1 neurons leads to an attenuation of metabolic rate and a reduction of food consumption in both fed and fasted states in mice (Table 1) (Cheng et al., 2020;Gaykema et al., 2017;Holt et al., 2019). Stimulation of nucleus tractus solitarius GLP-1 neurons also suppresses glucose production without effects on glucose uptake, highlighting a potential role for the central GLP-1 system to regulate both energy balance and glucose metabolism. (Table 1) (Gaykema et al., 2017;Shi et al., 2017). Importantly, ablation or inhibition of nucleus tractus solitarius GLP-1 neurons increased refeeding after a fast and inhibits stress-induced hypophagia (Holt et al., 2019). However, constitutive deficiency of pre-proglucagon in nucleus tractus solitarius GLP-1 neurons fails to alter energy balance (Cheng et al., 2020). Thus, there are varied reports of either continuous stimulation or loss of function/inhibition of nucleus tractus solitarius GLP-1 neurons resulting in chronic changes in metabolism (Tables 1 and 2).
Herein, we review various brain regions, including the hypothalamus, which are required and/or sufficient to mediate the acute and chronic effects of GLP-1 and GLP-1 receptor agonists on energy balance and glucose metabolism. However, we acknowledge that there are differences between species (including variability in rodents) in how native peripheral/central GLP-1 and long-acting GLP-1 receptor agonists alter energy balance and glucose metabolism. As our review primarily focuses on studies in mouse models, it is imperative to consider how GLP-1 is metabolized by other species (see Section 4).

| GLP-1 SIGNALLING IN THE PERIPHERY VERSUS CNS
Classical whole animal pharmacological studies established GLP-1 receptor activation promotes glucose tolerance and decreases food intake, thereby inducing weight loss and improving glucose homeostasis (Burcelin et al., 2009;Buse et al., 2009;Richard et al., 2014). Utilization of mouse genetic tools (see technical considerations of rat vs. mouse models) has also supported a physiological role of GLP-1 receptors in the regulation of energy balance and glucose homeostasis.
For example, global deficiency of GLP-1 receptors consistently resulted in decreased glucose excursion in response to an oral glucose tolerance test, which was also accompanied by lowered circulating insulin levels (Table 2) (Hansotia et al., 2007;Scrocchi et al., 1996). Mice globally deficient for GLP-1 receptors also exhibited increased blood glucose levels in response to an intraperitoneal glucose tolerance test (ipGTT) (Scrocchi & Drucker, 1998). These data support a principal action of GLP-1 receptors to properly regulate blood glucose levels. With regard to energy balance, global deficiency of GLP-1 receptors predictably attenuated the GLP-1 induced hypophagia (Table 3) (Scrocchi et al., 1996). However, these mice failed to exhibit altered body weight suggesting that GLP-1 receptors are not a key determinant of body mass under basal conditions (Scrocchi et al., 1996). Subsequent studies showed male and female mice globally deficient for GLP-1 receptors were paradoxically protected against diet-induced weight-gain (Table 2) (Ayala et al., 2010;Hansotia et al., 2007;Scrocchi et al., 1996;Scrocchi & Drucker, 1998). Although not entirely clear, some of the early disparities between global knockout studies on energy balance were attributed to varying dietary components (e.g. % kcal from fat), age of diet exposure, as well as the endpoint chosen for assessment of various metabolic parameters (Ayala et al., 2010).
Specific to the brain, intracerebroventricular (i.c.v.) administration of GLP-1 fails to alter eating in mice globally deficient for GLP-1 receptors (Table 3) (Scrocchi et al., 1996). Inhibition of GLP-1 receptor in the brain also impedes glucose homeostasis and food intake, whereas inactivation of GLP-1 receptor in the gut impairs glucosestimulated insulin secretion, reduces glucose clearance, increases levels of glucagon and increases gastric emptying after disruption of GLP-1 action but not food intake or body weight after 24 h (Table 3) (Knauf et al., 2008;Sandoval et al., 2008;Scrocchi et al., 1996). These reports are consistent with the demonstration that GLP-1 receptors in the brain are capable of controlling food intake and body weight (Drucker, 2018;Knudsen & Lau, 2019;Müller et al., 2019;Shah & Vella, 2014). This is in addition to the hypophagic effects of endogenous GLP-1 via peripheral GLP-1 receptors (Ruttimann et al., 2009). As endogenous GLP-1 is rapidly degraded once it enters circulation, hypothalamic GLP-1 receptors are likely primarily targeted by nucleus tractus solitarius GLP-1 neurons (Kanoski et al., 2016;Richard et al., 2015). However, recently developed long-acting GLP-1 analogues have also been demonstrated to target multiple nuclei within the brain, including the hypothalamus (Gabery et al., 2020;Williams et al., 2009).
It is important to note that although subdiaphragmatic vagal afferent deafferentation in rats inhibits the effects of liraglutide and exendin-4 to suppress food intake and body weight at low doses, subdiaphragmatic vagal deafferentation fails to attenuate the food intake and body weight lowering effects of these GLP-1 receptor agonists at high doses (Table 3) (Kanoski et al., 2011). Similarly, peripheral administration of exendin-4 suppressed food intake and body weight equally in rats that underwent F I G U R E 2 (a) Four nuclei are shown here: arcuate nucleus (ARC), lateral parabrachial nucleus neurons (L-PBN), paraventricular hypothalamic nucleus (PVH) and nucleus tractus solitarius (NTS). Each neuronal structure has input and output projections represented by a coloured line. Each colour represents a specific cell population, that is, pro-opiomelanocortin (POMC) is pink and projects from the ARC to the L-PBN, nucleus tractus solitarius (NT)S and PVH. The coloured lines represent cell-specific projections to different brain structures. Projections to a specific region are cell-type dependent projections to that region and not to a specific cell type in that region. Projections are labelled with either F, A or O-fluid intake, anorexigenic or orexigenic pathways, respectively. Cell populations expressing GLP-1 receptor (R)s are highlighted within a yellow box in the figure legend at the bottom right as well as with a yellow triangle on top of the GLP-1 receptor expressing cells (coloured circle). Yellow triangles in any nuclei signify the presence of GLP-1 receptors but not assigned to a specific cell type. Central GLP-1 originates from the NTS (yellow) whereas peripheral GLP-1 from the intestine (bottom right; yellow-brown). The black arrows represent non-cell-type specific projections from the NTS, PVH and L-PBN to other hypothalamic and extra-hypothalamic sites. (b) The highlighted region focuses on the PVH single-minded 1 (Sim1)+, pituitary adenylate-cyclase-activating polypeptide (PACAP)+, thyrotropin releasing hormone (TRH)+ to ARC agouti-related protein (AGRP) orexigenic pathway (Krashes et al., 2014). Positions of cells do not represent hemispheric segregation nor exact location. Abbreviation: SCP, superior cerebellar peduncle T A B L E 1 The effect of activating or inhibiting GLP-1 and GLP-1 receptor expressing neurons in the brain on energy balance and glucose metabolism   lesioning of the chemoreceptor trigger zone (CTS) for emesis -the area postrema (Baraboi et al., 2010). Moreover, the hypophagic effects of the GLP-1 agonists were blunted by administration of GLP-1 receptor antagonists into the brain ventricular system (i.c.v. injection into the 3 rd ventricle; Table 3) (Kanoski et al., 2011). Together, these data suggest the potential requirement of GLP-1 receptors in both the vagal afferents as well as the CNS for the full effects of long-acting designer GLP-1 receptor analogues on energy balance.
The aforementioned studies relied upon pharmacology and broad spectrum genetic tools (e.g. whole animal knockouts) to elucidate the effects of GLP-1 on energy balance and glucose metabolism. Although these studies were greatly informative in demonstrating among other things that direct microinjection of GLP-1 into the brain (either via administration into the ventricles or hypothalamic as well as extrahypothalamic nuclei-discussed below) may result in decreased food intake/weight gain and improved blood glucose control, it may be difficult to determine the sites of action in intact animals. Examining cellular activity directly linked or associated with physiology might also be disconnected. As a means of circumventing this dilemma, researchers used cyclization recombination-locus of X over P1 (Cre-loxP) technology alone or in combination with pharmacological approaches to (1) interfer with circuits, guided by neuroanatomic information coupled with the power of mouse genetics and (2) assess effects on energy and glucose homeostasis in awake, unrestrained mice.
Deficiency of GLP-1 receptors in the CNS (both neuronal and glial cells) or in the visceral nerves failed to alter food intake of rodents when fed either a chow or a high fat diet (Table 2) (Sisley et al., 2014). Similar to global knockout studies, these data suggest that CNS and vagal GLP-1 receptors may not be necessary for the control of food intake or body weight (Sisley et al., 2014). However, separate cohorts of mice null for GLP-1 receptors in the CNS did reveal an increase in food intake (Sisley et al., 2014). Speculatively, these effects may be highly transient or potentially compensated for with time, as no differences in cumulative body weight or body composition were observed (Sisley et al., 2014). Knockdown of GLP-1 receptors within the CNS also failed to alter baseline blood glucose levels (Table 2), supporting a principal action for GLP-1 in the periphery for proper basal glycaemic control (Sisley et al., 2014). Although there was a non-significant impairment of glucose changes in response to either an intraperitoneal glucose tolerance test or oral glucose tolerance test, lack of GLP-1 receptors in the CNS also failed to alter glucose tolerances (Table 2) (Sisley et al., 2014). In support of these data, chronic inhibition of GLP-1 receptors potently increased food intake, while failing to alter long-term body weight of high fat diet fed mice (Knauf et al., 2008). However, these mice were also protected from hyperinsulinaemia and insulin resistance suggesting a potential central action of GLP-1 receptors in regulating glucose metabolism (Knauf et al., 2008). reports which suggested a disparity between diet composition-with utilization of standard chow (Varin et al., 2019) or high fat diets of 58% from fat in the former study and 72% from fat in the later (Knauf et al., 2008;Sisley et al., 2014).
Opposite to loss of function, peripheral administration of liraglutide provides beneficial effects on blood glucose control in both chow-and high fat diet fed mice (Table 3) (He et al., 2019;Secher et al., 2014;Sisley et al., 2014). Liraglutide also suppresses food intake and body weight of high fat diet fed mice (Table 3) (Table 3) (Sandoval et al., 2008). Similarly, a bolus of GLP-1 or GLP-1 receptor agonist infusion directly into the arcuate nucleus reduces hepatic glucose production and improves glucose changes in response to a glucose challenge, supporting a link with GLP-1 receptors in the arcuate nucleus to regulate blood glucose levels (Sandoval et al., 2008). These effects occur independent of changes in feeding behaviour on a chow diet (Sandoval et al., 2008).
In agreement with these data, knockdown of GLP-1 receptor expression in arcuate pro-opiomelancortin neurons fails to alter basal food intake or energy expenditure of mice on a chow diet (Table 2) (Burmeister et al., 2017). Although the pharmacological effects of the GLP-1 receptor agonist, exendin-4, to reduce food intake was also similar between knockdown and control mice (Table 3) (Table 2) (Burmeister et al., 2017). This effect may be due to distinct cell-type and/or brain region specific roles for GLP-1 receptor signalling as well as a potential for pro-opiomelanocortin GLP-1 receptors to alter handling of nutrient stores independent of changing food intake or energy expenditure (Burmeister et al., 2017).  (Table 1) (Li, Navarrete, et al., 2019). The activation of paraventricular hypothalamic nucleus GLP-1 receptor neurons occurs in response to both chow and high-energy diets (Table 1)  In particular, cell specific loss of GLP-1 receptor-expressing paraventricular hypothalamic nucleus neurons in adult mice increased body weight as well as elevated fasted blood glucose levels and impaired insulin sensitivity (  (Abtahi et al., 2016). Additionally, exendin-4 pretreatment attenuates the combined effects of NPY and ghrelin coinfusion into this same nucleus (Dalvi et al., 2012). Therefore, GLP-1 in the paraventricular hypothalamic nucleus and the arcuate nucleus can spur a metabolic shift towards lipid utilization and contribute to peripheral substrate utilization.

| Candidate paraventricular hypothalamic nucleus neurons in GLP-1 induced hypophagia
The As GLP-1 activates neurons within the paraventricular hypothalamic nucleus to suppress food intake (Liu et al., 2017), it is unlikely that GLP-1 would activate this orexigenic circuit.
Oxytocin neurons in the paraventricular hypothalamic nucleus have classically been associated with feeding behaviour (Sabatier et al., 2013). This was recently supported by evidence that AGRP neurons (which when activated potently drive feeding behaviour) project to paraventricular hypothalamic nucleus oxytocin neurons ( Figure 2) (Atasoy et al., 2012). Moreover, simultaneous activation of paraventricular hypothalamic nucleus oxytocin neurons and arcuate AGRP neurons blunts AGRP-induced food intake (Atasoy et al., 2012).

| The parabrachial nucleus (PBN)
The parabrachial nucleus is implicated in various aspects of energy balance including feeding behaviour and visceral satiety as well as visceral malaise, taste, temperature, pain and itch (Kanoski et al., 2016;Palmiter, 2018;Richard et al., 2014;Rinaman, 2010). Central nucleus tractus solitarius GLP-1 expressing neurons send projections to the lateral and medial lateral parabrachial nucleus (lPBN and mPBN) (Alhadeff et al., 2014;Richard et al., 2014). Lateral ventricle injection of exendin-4 results in the activation of lateral parabrachial nucleus neurons which can be blocked by exendin-9 (a GLP-1 receptor antagonist), suggesting the involvement of the central GLP-1 system on energy balance via the parabrachial nucleus (Richard et al., 2014). This is further supported by the demonstration that pharmacological activation of GLP-1 receptors in the lateral parabrachial nucleus neuron results in a reduction of food intake and body weight (Table 3) (Richard et al., 2014). Accordingly, blockade of GLP-1 recetors in the lateral parabrachial nucleus neurons increases body weight and chow intake demonstrating that GLP-1 receptors in the lateral parabrachial nucleus neurons are both sufficient and required to control food intake (Table 3) (Alhadeff et al., 2014;Richard et al., 2014). Moreover, consumption of palatable foods such as chocolate pellets and the motivation to eat palatable foods is reduced by intra-parabrachial nucleus injection of GLP-1 agonists (Richard et al., 2014). Caloric density and hedonic properties of food may be interacting with GLP-1 receptor signalling within the lateral parabrachial nucleus neurons (Richard et al., 2014). Ablation of the parabrachial nucleus may also play a role in blunting the ability of the nucleus accumbens to raise dopamine levels in response to appetizing food (Richard et al., 2014).

| Mesolimbic reward system (MRS) nuclei and GLP-1 signalling
GLP-1 receptor signalling may also play an important role in the nonhomeostatic control of eating via activity within the ventral tegmental area (VTA) and nucleus accumbens (NAc). In particular, GLP-1 expressing neurons in the caudal medulla project directly to the VTA and nucleus accumbens (Alhadeff et al., 2012). Administration of GLP-1 receptor agonists to the ventromedial hypothalamus, nucleus accumbens core and shell of rats on a high-energy diet results in the reduction of food intake (Table 3) (Alhadeff et al., 2012). This may also involve satiety promoting and food intake reducing effects by GLP-1 via a reduction in the reward value of food by direct action in the mesolimbic reward system (MRS) (Alhadeff et al., 2012). It is also important to note that administration of exendin-4 to the nucleus accumbens core and not the shell displayed a reduction of sucrose intake (Table 3) (Alhadeff et al., 2012). Conversely, blockade of GLP-1 receptor in the ventromedial hypothalamus and nucleus accumbens core increased food intake (Table 3) (Alhadeff et al., 2012). This suggests that a physiological role of GLP-1 signalling in the mesolimbic reward system to regulate energy balance. Moreover, the nucleus accumbens core may play a more significant role in carbohydrate intake and preference under food deprivation conditions (Alhadeff et al., 2012). Although these data highlight nutrient dependent effects of GLP-1 in the nucleus accumbens, specific macronutrient selection or orosensory processing remains undefined (Alhadeff et al., 2012). Collectively, these data indicate that GLP-1 receptor signalling in the mesolimbic reward system reduces food intake; however, the signalling cascade and downstream targets mediating this effect are not well established.
The hypothalamic, hindbrain and mesolimbic reward pathways discussed herein are reciprocally connected as well as with various other brain regions (including but not limited to the hippocampus, lateral dorsal tegmental area and the lateral septum). Many of these brain regions are also involved in the GLP-1 receptor dependent regulation of energy balance (Reiner et al., 2018). Collectively, these data highlight a multi-nodal neural circuit in the brain which is necessary and sufficient (and in some cases redundant or compensatory) for the full effects of GLP-1 on energy balance and glucose metabolism.

| TECHNICAL CONSIDERATIONS
As with all studies, there are several technical considerations that should be considered in context with the conclusions presented. First, there are species differences with regard to GLP-1 dependent changes in food intake (Tornehave et al., 2008). For example, central GLP-1 is a physiologically important signal in the control of eating and energy balance in rats Hayes et al., 2009). However, brain derived GLP-1 in the mouse may be more responsible for stress-induced hypophagia, limiting unusually large intake of food and remaining relatively irrelevant for the control of normal meals (Cheng et al., 2020;Holt et al., 2019). This suggests a context specific control of food intake in a physiological setting by central GLP-1 receptors between rodent models. In a pharmacological context, food intake is minimally impacted by interference with endogenous GLP-1/GLP-1 receptor, while eating is decreased by activation of GLP-1 neurons within the nucleus tractus solitarius (Cheng et al., 2020;Gaykema et al., 2017;Holt et al., 2019). These differences suggest that activa-  (Fortin et al., 2020). In rats, several studies have shown that native GLP-1 administered intraperitoneally (i.p.) requires intact vagal afferents to cause satiation, that is, to reduce meal size (Ruttimann et al., 2009). Consistent with these findings, inhibition of eating in response to i.p. administered native GLP-1 in rats can be fully blocked by peripheral, but not by central administration of a GLP-1 receptor antagonist (Williams et al., 2009). Together, these findings indicate that i.p. administered native GLP-1 reduces food intake by acting on peripheral, most likely vagal afferent GLP-1 receptors in rats. Thus, it becomes easy to see that translating results between species (even between rodents) is very challenging. The divergence of GLP-1 neural circuitry between rats and mice are at least one likely culprit .
Together, this highlights a need to further delineate differences in GLP-1/GLP-1 receptor signalling between species in future investigations in order to better understand the physiology versus pharmacology of the GLP-1 system.
Another distinction that should be considered is the reporting of short versus long term effects with respect to changes in feeding behaviour. In particular, many of the studies mentioned herein refer to the measurement of food intake at 24-h intervals. It is important to note that 24-h food intake is not feeding behaviour and examining shorter intervals may reveal important insights. For instance, vagal afferent GLP-1 receptor knockdown by RNA interference in rats increased meal size and was compensated for by a decrease in meal frequency such that 24-h food intake was not affected .

| CONCLUSION AND FUTURE DIRECTIONS
In summary, there are multiple brain regions and neural circuits by which GLP-1 controls food intake and body weight. GLP-1 activity within these circuits may also contribute to proper blood glucose con- Although these data highlight a growing understanding of GLP-1 action in the brain, they also raise several questions for future investigation.
First, how do GLP-1 neurons and their projections develop/form and what is required to maintain them throughout life? The brain regions and neural circuits involved in the central actions of GLP-1 are highly plastic or malleable (Lieu et al., 2020). This plasticity may be triggered in response to a variety of nutrient, humoral and metabolic challenges (Lieu et al., 2020). The melanocortin circuit within the hypothalamus is a classic example of this plasticity (Bouret et al., 2004a;Bouret et al., 2004b;Bouret & Simerly, 2004;Lieu et al., 2020). Early studies suggested that trophic factors, such as leptin, guided the axon neurite outgrowth from arcuate NPY/AGRP and pro-opiomelancortin neurons influencing the connections to downstream targets (Bouret et al., 2004a;Bouret et al., 2004b;Bouret & Simerly, 2004 (Considine et al., 1996), and weight gain itself may result in impaired GLP-1 signalling (Ranganath et al., 1996).
Understanding the leptin induced suppression of GLP-1 inputs to the hypothalamus during weight gain and/or in the obese state may provide an additional pathology of obesity and diabetes. Ultimately, understanding how the GLP-1 to hypothalamic/extrahypothalamic circuits develop and are maintained may be a critical aspect in understanding the development of obesity and diabetes.
Second, there are several pharmaceuticals that have emerged for use in chronic weight management and the treatment of diabetes that act on metabolically relevant brain regions (Gautron et al., 2015;Yanovski & Yanovski, 2014). However, most of the weight loss from these medications takes place within the first 6 months of usage and rarely exceeds 5%-10% weight loss (Yanovski & Yanovski, 2014).
Importantly, even moderate weight loss results in measurable improvements in blood sugar control, BP regulation and triglyceride levels (Magkos et al., 2016;Yanovski & Yanovski, 2014). However, work has continued in an attempt to achieve effects that are even more robust. Combination drug therapy was introduced as a way to surpass the weight loss barrier of a single medication (Gautron et al., 2015;Muller et al., 2018;Yanovski & Yanovski, 2014). This concept has recently been extended to the incretin system and has shown real promise. In particular, the beneficial effects of GLP-1 have been combined in dual agonists (for both GLP-1 and GIP receptors or GLP-1 and glucagon receptors) or tri-agonist (for GLP-1, GIP and glucagon receptors) strategies (Capozzi et al., 2018;Tschöp et al., 2016).
There has been a preference towards a single-molecule multi-agonist strategy due to a variety of competing complications when using a combination of mono-agonists (Capozzi et al., 2018;Tschöp et al., 2016). These can include differing bioavailabilities, half-lives, tissue specificity and pharmacokinetics (Capozzi et al., 2018). There are several clinical trials investigating the utility of these multi-agonists strategies in humans and showing significant benefits on body weight, postprandial glucose levels, insulin sensitivity and cholesterol levels (Bastin & Andreelli, 2019;Mathiesen et al., 2019). In particular, dual agonists improve metabolic control by promoting weight loss and improving glucose tolerance (Thomas et al., 2020;Willard et al., 2020). Tri-agonists in preclinical/clinical research have also demonstrated some superior beneficial effects compared to dual agonists, which include reduction of body weight, enhancement of glycaemic control and the reversal of nonalcoholic steatohepatitis (NASH) (Capozzi et al., 2018). Although the beneficial effects of these compounds likely involve activity at cognate receptors, better understanding any biases within these compounds for varying receptors and their effects on metabolism may advance development of future therapeutic interventions. Moreover, there is a need to understand the ability of these compounds to target receptors in the periphery versus the CNS (with an emphasis on specific brain regions) in order to maximize metabolic benefits. In addition to GIP and glucagon, GLP-1 has also been implicated to interact with other hormones involved in metabolism. In particular, ghrelin, a potent orexigenic peptide, has also been implicated in appetitive motivation, energy metabolism, homeostasis and respiratory exchange ratio/respiratory quotient. Importantly, long-acting GLP-1 receptor agonists antagonize the metabolic effects of acylated ghrelin signalling within the paraventricular hypothalamic nucleus (Abtahi et al., 2019). These effects may also be apparent in leptin and 5-HT systems (Holt et al., 2017), illustrating the neural integration of what was once thought of as separate neural systems regulating eating and metabolic-related disorders. Together, this highlights a broad action across multiple nuclei and a need to simultaneously target multiple systems or receptor mechanisms in order to provide maximum pharmacological and therapeutic benefit for eating and metabolic-related disorders.
Various genetic tools have been useful at identifying the glucoregulatory action of GLP-1. However, most of these same genetic models-in the absence of pharmacological administration of GLP-1 or GLP-1 receptor mimetics-failed to identify a role for GLP-1 receptors to alter food intake or body weight (Ayala et al., 2010;Ghosal et al., 2017;Scrocchi et al., 1996;Scrocchi & Drucker, 1998;Sisley et al., 2014). This might be due to several reasons: First, GLP-1 receptor agonists are similar to other monoagonist therapies for chronic weight management, as they result in a modest 5%-15% weight loss in adults with obesity and/or diabetes (Klonoff et al., 2008;Riddle et al., 2006;Wilding et al., 2021). This modest change in body weight might be difficult to delineate in genetic models that examine changes in energy balance in only a few weeks of weight gain on either a standard chow or high-energy diet. Second, most of the genetic models employed to investigate GLP-1 receptors in energy balance were constitutive deletion models resulting in the ablation of GLP-1 receptors in development. The hypothalamus (particularly the melanocortin circuit) is a classic model that is susceptible to up-regulation of developmental compensatory pathways that mask food intake and body weight phenotypes (Wu & Palmiter, 2011;Xu et al., 2018). It is possible that an analogous compensation occurs in models with constitutive deficiency of GLP-1 receptors, which prevents the identification of GLP-1 receptors as required in the regulation of energy balance at basal levels. In support of this idea, ablation of GLP-1 receptors in the adult (using AAV-Cre injected directly into the paraventricular hypothalamic nucleus) resulted in marked obesity whereas constitutive deletion of GLP-1 receptors in the paraventricular hypothalamic nucleus (using Sim1-Cre) failed to alter energy balance (Table 2) (Burmeister et al., 2017;Ghosal et al., 2017;Liu et al., 2017). Perhaps, further examination of loss or gain of function of GLP-1 receptors in the adult might be necessary for a better understanding of the requirements for GLP-1 receptors in energy homeostasis. Another possibility is that GLP-1 receptor signalling simply is not required for proper energy balance regulation in the basal state. Rather the GLP-1 system and downstream GLP-1 receptors must be activated-as occurs after a meal or in response to GLP-1 receptor mimetics-in order to alter energy balance. However, this suggests that if GLP-1 functions at least in part as a satiety factor then other satiety systems must compensate for the loss of GLP-1 receptors to control food intake and body weight in times which GLP-1 signalling would be activated (e.g. in response to a meal). The end result of these studies demonstrate that conventional genetic screens alone (in the absence of pharmacology) fail to realize the potential for improvements in food intake and body weight regulation of GLP-1 signalling and highlight a need to pair pharmacology with genetic models.
Finally, GLP-1 and designer GLP-1 receptor agonists elicit changes in cellular activity within the brain that is localized to both pre-and post-synaptic sites (He et al., 2019;Liu et al., 2017;Secher et al., 2014). Much of this activity has been assessed acutely and there are some discrepancies between ex vivo and in vivo measurements of this activity that warrant further investigation (Beutler et al., 2017;Su et al., 2017). However, less is also known about the effects of chronic GLP-1 receptor agonist administration on the activity of these brain regions or neural circuits. This is an important area of investigation as clinically, patients are likely to undergo long-term therapy with these compounds as opposed to a single administration. Better understanding the development/maintenance of GLP-1 neurons, combinatorial therapy involving GLP-1 designer agonists, as well as the acute versus chronic effects of GLP-1 on the brain will undoubtedly be growing areas of study to better understand the therapeutic benefits of this system in metabolic disease.

| Nomenclature of targets and ligands
Key protein targets and ligands in this article are hyperlinked to corresponding entries in the IUPHAR/BPS Guide to PHARMACOL-OGY http://www.guidetopharmacology.org and are permanently archived in the Concise Guide to PHARMACOLOGY 2021/22 (Alexander et al., 2021).

ACKNOWLEDGEMENT
This work was supported by National Institutes of Health grants to K.W.W. (R01 DK119169 and DK119130-5830).

CONFLICT OF INTERESTS
No competing interests.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.