Mechanisms of glucagon-like-peptide 1 in the brain beyond metabolic effects

Abstract

Ever since its discovery, glucagon-like-peptide 1 (GLP-1) and drugs with similar function (collectively GLP-1s) have been used for type 2 diabetes mellitus and have been effective for obesity. Their profound effect on weight loss has resulted in widespread use of these medicines for treating obesity. Extensive studies have shown that GLP-1s decrease body weight, lean mass, and other metabolic phenotypes. These studies were supported by the mechanism of signaling pathways of GLP-1s in cells and their metabolic effects. Recently, studies have focused on the effect of GLP-1s on the brain, showing that they affect food cognition, depression, drug addiction, and even neurodegenerative diseases. Although recent studies have investigated the impact of GLP-1s on the brain, our understanding of these effects is limited. This review describes how GLP-1s have become the most effective drugs in obesity, such as their known signaling pathway in cells and their pharmaceutical processing over the years. This review covers recent discoveries of the GLP-1 mechanisms in the brain, including their prominent effects on obesity, and discusses discoveries that imply their potential usage in brain disorders.

Highlights

· Discovery of glucagon-like-peptide 1 (GLP-1) has led to effective treatment of obesity.

· GLP-1 mostly works in the brain by GLP-1 receptors in hypothalamic and brainstem regions in the brain.

· GLP-1 is being developed for use in various diseases outside obesity.

Discovery and clinical applications of GLP-1s

Glucagon-like-peptide 1 (GLP-1) was first discovered about 40 years ago in the early 1980s in the Habener lab, stemming from techniques for cloning anglerfish proglucagon cDNA, which was later used to clone mammalian proglucagon cDNA [1-4] (Fig. 1). Discovery of GLP-1 led to elucidation of their conversion to proglucagon by proprotein convertases such as furin [5,6]. Inspection of the amino acid sequences of GLP-1s revealed that they had 2 equally bioactive forms: glycine-extended GLP-1 (GLP17-37) and amidated GLP-1 (GLP17-36). Soon after, GLP-1 was revealed to be insulinotropic, originating from the intestinal L cells and pancreatic β-cells [6,7]. GLP-1s seemed to be highly related to meal consumption, as their plasma concentration levels were elevated during ingestion of food [8]. However, these proteins were in circulation for only a short time, limited to 1-2 minutes, due to their degradation by the enzyme dipeptidyl peptidase-4 (DPP-4), which raised questions regarding use of GLP-1s as pharmaceutical treatments [9]. DPP-4 was first discovered in 1966 and has been shown to circulate to all parts of the body through the plasma [10-12]. As an enzyme, DPP-4 was highly efficient when a substrate had a second amino acid of proline or alanine [12]. As with GLP-1, they had alanine at position 8, which contributed to their short lives during plasma circulation by binding to DPP-4 [9]. This disadvantage was soon mitigated by the discovery of exenatide (Ex-4) in 1992, the first GLP-1 receptor agonist (GLP-1RA), originating from the salivary gland of a lizard, Heloderma suspectum [13,14]. Ex-4 had 53% structural similarity with GLP-1 but could activate GLP-1 receptors (GLP-1R), and its longevity in the plasma was longer than that of GLP-1 [15]. This was due to a second amino acid of glycine rather than alanine, which imparted resistance to DPP-4 [15]. Ex-4 was used in clinical trials and approved in 2005 under the name Byetta as the first GLP-1RA for twice-daily treatment of type 2 diabetes mellitus (T2DM) [15,16]. Although these proteins were present for 2–3 hours longer in the plasma than the original GLP-1, they resulted in fluctuation of blood levels with once-daily or twice-daily administration [17].

Fig. 1.

Developmental timeline of GLP-1RAs. Dates for GLP-1 and exendin-4 are those of discovery. Dates for liraglutide and semaglutide are their approval dates for obesity. GLP-1, glucagon-like-peptide 1; GLP-1RA, GLP-1 receptor agonist; DPP4, dipeptidyl peptidase-4.

Development of long-lasting GLP-1 drugs

This prompted the development of new GLP-1RAs that could act longer, perhaps for weeks. Even though DPP-4 activity seemed to be the major issue for GLP-1 activity, a breakthrough came from another perspective when Lotte Knudsen from Novo Nordisk developed liraglutide [18]. To achieve a GLP-1RA that can act throughout the day, they made modifications and attached a fatty acid chain in the n-terminal that allows binding to albumin [19,20]. Albumin is abundantly produced from the liver and has multiple binding sites for fatty acids [21-23]. Being one of the most abundant proteins in the plasma, binding with albumin is usually not desirable for drugs because it makes the drugs remain in the body for an unwanted amount of time. However, Knudsen used albumin binding to retain GLP-1RA for as long as 11 hours in plasma [22]. Concurrently, it was becoming clear that T2DM and obesity had a very high correlation, and clinicians started to report a weight-loss effect of liraglutide in T2DM patients [24,25]. Soon, liraglutide was approved for both T2DM (Victoza, Novo Nordisk, Denmark) and obesity under the name of Saxenda (Novo Nordisk) for once-daily use [18,20]. Successful clinical trials and approval for these GLP-1 drugs led to an increase in GLP-1RAs, but once-daily use was still a barrier for some patients with diabetes or obesity because of the frequent use of needles for daily injections. This brought the need for a more convenient once-weekly applicable GLP-1RA drug. The remaining challenge was to change liraglutide to achieve a higher affinity toward GLP-1R even in the presence of albumin, since an improved affinity could accompany problems binding to GLP-1 [26]. Novo Nordisk addressed this and developed semaglutide by first changing the fatty chain acid linker to an "OEG" linker, which was observed to result in high receptor potency and high affinity for albumin [26]. Additionally, the peptide backbone at position 8 of liraglutide was changed from alanine to α-aminoisobutryic acid. In doing so, they kept the structure similar to that of the natural GLP-1 and also produced a new GLP-1RA resistant to DPP-4 [26]. These 2 changes allowed semaglutide durability for up to 1 week and resulted in U.S. Food and Drug Administration (FDA)-approved use for T2DM as the once-weekly drug Ozempic (Novo Nordisk) [27]. After the release of Ozempic, the interest in these drugs skyrocketed due to their high effectiveness in treating obesity, even though it is considered off-label use if not used for T2DM or cardiovascular diseases [28]. In 2021, the FDA approved semaglutide for chronic weight management in adults with obesity or overweight under the name Wegovy (Novo Nordisk) [29]. Due to the extensive amount of interest in these drugs, GLP-1 co-agonists are also being developed around the world to treat T2DM or obesity, and some show greater effects than Ozempic, Saxenda, or Wegovy [30,31].

Toward a comprehensive understanding of GLP-1 mechanisms in the body

Reflecting upon the explosive popularity of these drugs, investigations of the mechanism of GLP-1 and their uses in future clinical applications are being conducted. Clinical and preclinical studies are showing that GLP-1 extends beyond the original use in metabolic diseases to applications in drug addiction, neurodegenerative diseases, and inflammatory diseases [32-34]. This is probably due to the fact that GLP-1R is distributed both in the peripheral organs and in the central nervous system, the system that regulates most body functions [29]. Furthermore, GLP-1RA should not be understood to have the same mechanism as that of the endogenous GLP-1, as marketed GLP-1RAs function on a much longer timescale. Since the action of GLP-1 in the brain involves the activation of GLP-1R by GLP-1RAs to stimulate neuronal activity, it is essential to comprehensively understand the underlying mechanisms, including the signaling pathways and their corresponding neuronal responses. Therefore, in this review, we first cover the conventional signaling pathways of GLP-1 to understand their mechanisms in cells that regulate metabolism. We will then cover previous and recent findings of how GLP-1RA acts on GLP-1R neurons in the brain. Finally, we will discuss how these findings can relate to future applications for neuronal disorders and beyond.

GLP-1R signaling pathway in cells

In natural conditions, GLP-1 is produced by proteolytic cleavage of the protein proglucagon in a few body regions including the nucleus solitary tract (NTS) of the brainstem, enteroendocrine L cells, and the α cells of the pancreas [29,34,35]. GLP-1 functions by binding to GLP-1R, which is largely distributed throughout the peripheral and various regions in the brain [35,36]. In the peripheral, GLP-1R is distributed in the heart, pancreas, kidney, lung, and gastrointestinal (GI) tract. In the brain, it is distributed in the hypothalamic regions of the arcuate nucleus (ARC), dorsomedial hypothalamus (DMH), ventromedial hypothalamus (VMH), paraventricular hypothalamus (PVH), lateral hypothalamus (LH), and the subfornical organ; as well as in the brainstem regions of NTS, area postrema (AP), and dorsal vagal complex; and limbic system regions of the nucleus accumbens (NAc), ventral tegmental area (VTA), central amygdala (CeA), lateral septum (LS), hippocampus (HP), caudate putamen, globus pallidum, and other regions [37,38]. When GLP-1 reaches these regions, it binds to GLP-1R to produce a signaling cascade of the g-protein coupled receptor (GPCR) pathway. GLP-1R is a B-class GPCR that acts on both Gq and Gs signaling, comparable to the other incretin hormone glucose-dependent insulinotropic polypeptides (GIPs), though they only activate the Gs signaling pathway. Although the signaling pathways for pancreatic α, β, and δ cells are different as each cell type ultimately secretes different neuropeptides, they share similar cAMP-PKA pathways [29]. Gq/Gs signaling stimulates adenylate cyclase and increases cyclic AMP (cAMP) levels [39], which activates protein kinase A (PKA)-dependent intracellular signaling to stimulate different cascading pathways in different cells. The pathways modulate secretion of glucagon in α cells by regulating K-ATP channels, of insulin in β cells by regulating the PI3K/Akt signaling pathway, and of somatostatin in δ cells [2,29,40,41]. Inhibition of glucagon release by inhibition of α cell activity ultimately prevents an increase in the blood level, while β cell activation induces insulin release to lower blood levels. In addition, an increase of somatostatin by δ cells increases the activity of GI motility, reducing the secretion of glucagon from α cells and promoting insulin release from β cells [34]. In neurons, cAMP and PKA signaling pathways regulate synaptic plasticity by regulating pathways related to N-Methyl-D-aspartic acid (NMDA) and aminomethylphosphonic acid receptor signaling such as calcium influx into intracellular membranes to change synaptic plasticity for long-term potentiation related to memory function [42-45]. Indeed, GLP-1 was able to change synaptic plasticity in the PVH [46], hippocampus [47], LS [48], and the DMH [49,50]. In addition, co-treatment of obese mice with GLP-1RA and an NMDA receptor antagonist can induce greater change compared to GLP-1RA alone [51].

Activity of GLP-1R neurons in the brain

As stated above, GLP-1 modulates signal pathways in cells to induce changes in metabolism and neurophysiology. Recently, the role of GLP-1R neurons in feeding behaviors and neuropathology has been elucidated. Clinical studies have shown that the major effects of GLP-1RA in humans are regulation of blood levels or gut motility. Humans also report increased satiation and side effects such as nausea or vomiting [4]. These behaviors are modulated by regions in the brain such as the hypothalamus, hindbrain, and other areas near the circumventricular organ (CVO) that contain GLP-1R neurons [36].

One of the first studies that delivered GLP-1 into the brain was published in 1995, and showed that injection of intracerebroventricular (ICV) GLP-1 into hypothalamic regions reduced feeding and increased c-fos activity in the PVH and CeA [52]. Later studies delivered GLP-1 to specific regions in the hypothalamus and observed decreased feeding behaviors. Micro ICV injection of GLP-1 into the LH, a region associated with reward-related behaviors, decreased lever pressing in operant conditioning reward reinforcement tasks and showed decreased food intake and body weight [53]. ICV injection of GLP-1 into the VMH also induced anorexia and glucose tolerance [54]. Importantly, GLP-1 injection into the ARC was shown to decrease the activity of agouti-related peptide (AgRP) neurons and their strong orexigenic effect and to activate the pro-opiomelanincortin (POMC) neurons, which are prominent in satiety via melanocortin pathways [50,55,56]. AgRP and POMC neurons have reciprocal mechanisms; active AgRP neurons evoke hunger and inhibit POMC neurons, while POMC neurons activate satiety and inhibit AgRP neurons [57]. GLP-1R is not expressed in AgRP neurons but is expressed in POMC neurons [55], indicating that its effect on AgRP neurons is indirect and originates from other regions [50,55,56]. Recent studies have shown that GLP-1R also exists in the DMH while co-expressing leptin receptors (LepR) [58]. Furthermore, DMH LepR neurons respond to sensory information and send inhibitory projections to AgRP neurons, contributing to a decrease in anticipatory hunger caused by AgRP neurons [57,59,60]. This suggests that GLP-1 could send anticipatory signals of inhibited hunger via DMH GLP-1R neurons. Recently, we showed that GLP-1RA regulates AgRP neurons via DMH GLP-1R neurons to cause pre-ingestive satiation to control eating behavior [61] (Fig. 2). Other studies also highlighted the crucial role of DMH GLP-1R neurons in metabolism [49,62]. Although there has been some debate as to how the GLP-1R neurons in the hypothalamus are activated by GLP-1RA (when administered intraperitoneally), GLP-1 itself is degraded in the plasma in few minutes.

Fig. 2.

Mechanism of GLP-1RA activity in the brain. GLP-1RA travels to the brain by CVOs and to neighboring regions of the BBB. Regions in the brain that are colored blue indicate regions close to the CVO, while green indicates regions relatively far from the CVO. Red words indicate the effect of GLP-1RA in the brain or the effects of receptors. Arcuate nucleus of the hypothalamus (ARC), dorsomedial hypothalamus (DMH), ventromedial hypothalamus (VMH), paraventricular hypothalamus (PVH), lateral hypothalamus (LH), and the subfornical organ (SFO). Brainstem regions: nucleus tractus solitarius (NTS), area postrema (AP), dorsal vagal complex (DVC). Limbic system areas: nucleus accumbens (NAc), ventral tegmental area (VTA), central amygdala (CeA), lateral septum (LS), hippocampus (HP), and circumventricular organs (CVOs). GLP-1R, glucagon-like-peptide 1 receptor; GLP-1RA, GLP-1 receptor agonist; ATP, adenosine triphosphate; cAMP, cyclic AMP; PKA, protein kinase A.

The blood-brain barrier (BBB) is a major obstacle for drug entry into the brain. It is thought that GLP-1RA enters the hypothalamus through the third ventricle from circulation in the CVO by binding to tanycytes near the third ventricle [63]. The type of GLP-1RA affects the ability to cross the BBB [64]. Ex-4 is high in lipophilicity but is needed in high doses for penetration into the BBB [65]. Fluorescent-tagged Ex-4 has also been reported in regions deeper than the hypothalamus such as the VTA or the NAc, known for their primary role in reward processing [66,67]. Fluorescent-tagged liraglutide and semaglutide have also been shown to penetrate the hypothalamus and hindbrain but not further to the limbic systems [55,68]. These results show that the hypothalamus has essential roles in GLP-1RA-induced regulation of feeding behaviors, specifically by controlling meal termination.

Studies have also revealed the role of GLP-1RA in modulating the hindbrain for controlling feeding behavior. The hindbrain receives satiety signals from the body via GLP-1R and preproglucagon (PPG) neurons that express the gene gcg, which sends endogenous GLP-1 throughout the brain [69]. One of the regions that is most vulnerable to GLP-1 is the AP of the hindbrain ,which is located outside the BBB [70]. Administering GLP-1 ICV into the hindbrain reduced feeding [71] and increased nausea [72]. GLP-1R exists both in the AP, a region mainly known to cause nausea [72] and the NTS, a region mainly known to cause satiation effects [73]. Although it was difficult to determine the role of GLP-1R in either AP or NTS due to the spread of GLP-1RA drugs, a recent study elucidated the separate role of AP GLP-1R neurons and NTS GLP-1R neurons to encode aversion and satiation, respectively [74]. Furthermore, the study showed that AP GLP-1R neurons were not required for reducing food intake, while NTS GLP-1R neurons could evoke satiation without aversive effects, highlighting future directions for a non-aversive GLP-1R drug [74] (Fig. 2). Studies have also shown that AP GLP-1R neurons do not project to NTS PPG neurons, and NTS PPG neurons are not required for the effect of GLP-1RA [75]. These results show that GLP-1RA action on the hindbrain can be thought of as an independent effect from the endogenous GLP-1 activity induced by NTS PPG neurons. The role of endogenous GLP-1 signaling by NTS PPG neurons is currently under investigation. Studies have shown that activation of these neurons increases anorexia [76]; projects to form synapses with the PVN, a major controller of satiety [46,57]; and projects to the DMH to control glucose metabolism [49]. As NTS PPG neurons send multiple projections to other regions related to feeding, such as the ARC, CeA, and VMH, the mechanism of these circuits in controlling appetite suppression should be elucidated [75].

The roles of GLP-1R neurons in other regions of the brain are also being elucidated. A recent study has shown that active GLP-1RA neurons in the LS suppress feeding [48]. Interestingly, these neurons were deactivated during consumption, comparable with the activation of DMH GLP-1R neurons during consumption to encode pre-ingestive satiation [48,61]. These studies imply a complex network of GLP-1RA mechanisms in the brain. Other studies have shown that ICV delivery of GLP-1 into the CeA can evoke conditioned taste aversion, and the electrophysiological properties of these neurons have been investigated [77,78]. Systemic injection of Ex-4 was also able to evoke reduced cocaine seeking by activating GLP-1R in the NAc and the VTA [66,67]. Increased activity of HP GLP-1R neurons followed by Ex-4 injection reduced food intake, possibly due to volume transmission from NTS PPG neurons [79].

Collectively, these results imply that the action of GLP-1R induced by GLP-1RAs cannot be attributed to one specific region of the brain but must be understood from a brain-wide perspective. Modulating specific GLP-1R neurons in the brain might potentially enable control of certain aspects of feeding behavior, such as pre-ingestive satiation or non-aversive satiation.

Recent advancements in GLP-1RAs

The success of GLP-1 drugs has been mostly based on long-acting mono-agonists such as liraglutide and semaglutide. A randomized clinical trial comparing the effect of once-daily treatment of liraglutide and once-weekly treatment of semaglutide showed that adult patients treated with semaglutide for 68 weeks showed more than twice the decrease in body weight compared to people treated with liraglutide [80]. Furthermore, compared to liraglutide, semaglutide had fewer cases of side effects such as nausea, vomiting, diarrhea, and constipation [81]. Although both are GLP-1-based agonists, these 2 drugs have different effects on body weight and side effects.

Studies have further shown that greater body weight-loss effects can be achieved by combining GLP-1RA with other ligands as dual agonists. One of the first studies that confirmed the efficacy of GLP-1 dual agonism was on the synergetic effects of GLP-1 and GIP. Preclinical studies have shown that dual agonism of GIP and GLP-1RA could have a synergetic insulinotropic and glucose-lowering effect on beta cells in the pancreas [82]. We have further shown that this effect could originate from the hypothalamus, as a dual agonist of GLP-1RA with GIP had a synergetic effect in decreasing feeding behavior compared to GLP-1 or GIP alone when delivered to the hypothalamus in an ICV manner [83]. Recent studies have shown a strong effect of tirzepatide in inhibiting activity of AgRP neurons [84]. Notably, the effects of GIP on AgRP neurons were focused in nutrient response, while GLP-1R had an effect on response to sensory cues for food [84]. These GLP-RA and GIP dual agonists were approved by the U.S. FDA in November 2023 under the name Mounjaro, which shows a much larger body weight decrease than semaglutide and similar side effects [81,85].

Studies are further developing novel dual agonists to mitigate the side effects of mono-GLP-1RA or provide synergetic effects by stimulating other signaling pathways. A recent study developed a new dual agonist combination of semaglutide and MK-801 (an NMDA antagonist-based antidepressant) to mitigate the known yo-yo effect of GLP-1RA [51]. This study considered the limited ability of semaglutide to penetrate regions deeper than the hypothalamus and the internalization of GLP-1RA with binding to GLP-1R. They showed that MK-801 was cleaved after internalization in neurons, limiting its effect on synaptic plasticity to only GLP-1R neurons in the hypothalamus. Although the effects of these drugs were studied only for 2 weeks, they produced an average 20% body weight decrease, twice the effect of semaglutide alone. Furthermore, they confirmed that these drugs could change the expression levels of RNA related to synaptic plasticity, suggesting that the yo-yo effect could be mitigated by a change in long-term potentiation in hypothalamic neural circuitries [51]. Another study described a dual agonist to stimulate DMH neurons that co-express GLP-1R and LepR for decreased feeding behavior. This study further confirmed the validity of this effect in non-human primates, which shows that these dual agonists have high potential of success in humans [86]. Other studies have shown that GLP-1RA with melanocortin-4-receptor (MC4R) has a synergetic effect on body weight decrease [87,88].

As leptin-based drugs or MC4R agonist drugs alone have a limited effect on obesity, the co-agonists might treat broader conditions in obesity with better effects than mono-agonists. In combination with state-of-the-art technologies in neuroscience, recent studies are elucidating the impact of the mechanisms of synergetic compounds in neurons to shed light on the mechanism of GLP-1 co-agonists [84,89-91].

Another major point in the advancements of the usage of GLP-1 is its use in adolescents. This is important as metabolism rates such as total daily energy expenditure are greatly increased in children compared to adults [92]. Recent studies report that youths with obesity or T2DM (age<18 years) treated with lirglutide or semaglutide along with lifestyle interventions have better metabolic levels and experience greater body weight loss when treated with GLP-1 drugs but showed primary side effects such as nausea and vomiting [93-95]. Interestingly, the number of youths in the United States who was prescribed GLP-1RA increased from approximately 8,722 to 60,567 (594.4%) (ages 12–25 years) from January 2020 to December 2023, with most of this population being female (about 70%) [96]. The female preponderance may be because females are more prone to be interested in their weight. One preclinical study showed that exendin-4 injection in prenatal mice had no effect on body weight but did impact behaviors related to locomotion or anxiety but not learning [97]. Further studies are necessary to verify the impact of GLP-1RA treatment in early life, including effects on development, long-term consequences, and usage in adolescents younger than 12 years.

Future applications beyond obesity

As noted above, GLP-1RA affects cells in various organs in the peripheral through cAMP signaling pathways, changing activity in neurons that have GLP-1R. Their effect in the body varies depending on the drug used. To date, preclinical and clinical studies are reporting various uses for GLP-1 medicines [33]. Studies have shown that GLP-1 can be used for treatment of cardiac diseases, possibly independent of their effect in obesity [33] (Fig. 3). Furthermore, GLP-1 drugs have also been shown to affect inflammation in the heart, kidney, and liver [98,99]. Their effectiveness in addressing inflammation might be the reason for GLP-1RA effects in neurodegenerative diseases [100,101]. Investigation is ongoing to validate their effects in neurodegenerative diseases in preclinical and clinical trials [102]. As briefly mentioned above, GLP-1RA also affects drug addiction. Ex-4 can penetrate deep into the limbic system [66,67], well-known for its role in drug addiction due to changes in synaptic plasticity [103]. Furthermore, GLP-1RAs have been reported to modulate synaptic plasticity in neurons, which shows the possibility of using these drugs as a drug addiction treatment [45]. Indeed, clinical studies have shown such effects of GLP-1RA [104,105]. Given the eminent global opioid crisis, development of GLP-1RA is an important research goal [106].

Fig. 3.

Potential applications and development of GLP-1 drugs. GLP-1, glucagon-like-peptide 1; NTS, nucleus tractus solitarius; DMH, dorsomedial hypothalamus; VTA, ventral tegmental area; NAc, nucleus accumbens.

Conclusion

The future looks bright for GLP-1-based disease treatment. Given the large body of literature on GLP-1 mechanisms ranging from endocrinology to neuroscience and corresponding preclinical and clinical data, GLP-1-based treatment could offer a new pathway for treatment in diverse fields.