What this is
- Liraglutide, a analog, is effective for weight loss but its mechanisms are not fully understood.
- This research identifies receptor-positive neurons in the lateral septum as key mediators of liraglutide's effects.
- The study demonstrates that activating these neurons suppresses feeding, while their inactivation diminishes liraglutide's efficacy.
Essence
- receptor-positive neurons in the lateral septum mediate the and weight-lowering effects of liraglutide in mice. Their activation suppresses feeding, while their knockdown reduces liraglutide's effectiveness.
Key takeaways
- receptor-positive neurons in the lateral septum are activated by liraglutide, leading to decreased food intake. The study shows that these neurons are crucial for liraglutide's weight-loss effects.
- Knocking down receptors in the lateral septum significantly reduces liraglutide's ability to suppress feeding and promote weight loss, indicating their essential role in mediating these effects.
- Silencing lateral septum neurons increases food consumption and promotes obesity in mice on a high-fat diet, demonstrating their regulatory role in energy homeostasis.
Caveats
- The study primarily uses mouse models, which may not fully represent human physiology. Further research is needed to confirm these findings in humans.
- The effects of liraglutide on feeding behavior may vary with different dietary conditions, limiting the generalizability of the results.
Definitions
- GLP-1: A glucagon-like peptide that regulates appetite and glucose metabolism.
- anorectic: Referring to a substance that suppresses appetite.
AI simplified
Introduction
The obesity pandemic continues to pose a global health threat as safe and effective treatments remain elusive. Among the various pharmacological approaches used, glucagon-like peptide-1 (GLP-1) analogs, including liraglutide, are used for clinically marked weight loss (,). As a synthetic GLP-1 analog, liraglutide diminishes appetite and energy intake in both humans and animals, leading to weight reduction (,). 1 2 2 3
Liraglutide, acting similarly to native GLP-1 (), modulates energy homeostasis by activating G protein–coupled receptors (,–). Studies have shown that GLP-1 receptor–expressing (GLP-1R–expressing) cells within the central nervous system (CNS) are pivotal for the anorectic and body weight–reducing effects of peripherally administered liraglutide in rodent models (–). The hypothalamus, a central governor of energy balance, has consequently emerged as a key locus for GLP-1R research (–). However, genetic deletions of GLP-1Rs in various hypothalamic regions, including the paraventricular nucleus (PVN), arcuate nucleus (Arc), or ventromedial nucleus of the hypothalamus, do not impair the anorectic effects of peripherally administered GLP-1R agonists (,). 4 2 5 8 9 11 12 16 17 18
Beyond the hypothalamus, attention has also been directed toward the hindbrain region, specifically, the area postrema (AP), which has substantial GLP-1R expression (,–). Both endogenous GLP-1 and exogenously administered GLP-1 analogs have been found to exert anorectic effects through GLP-1Rs within the AP (–). However, some findings, such as the fact that AP lesions do not diminish liraglutide-induced reductions in food intake and body weight (), raise questions about the precise brain sites where liraglutide's anorectic effects originate. 8 19 21 22 25 26
Further, GLP-1R is found in the dorsal portion of the lateral septum (LS), a region that projects to key feeding centers in the brain and is believed to regulate gastric emptying after meal consumption (,–). The observed anorectic effects following local microinfusion of GLP-1R agonists into the LS suggest a role of GLP-1Rs in this region in reducing food intake (,). Yet the physiological functions of GLP-1R–positive neurons in the LS (LSneurons), and whether these neurons are necessary for liraglutide's anorectic effects, remain largely unknown. 8 27 29 30 31 GLP-1R
To unravel the above questions, we investigated the physiological and pharmacological roles of GLP-1Rs in the LS. Combining CRISPR/Cas9–mediated targeted knockdown of receptors in a specific group of neurons with neural circuit manipulation techniques, we dissected the function of LS GLP-1 receptors and LSneurons and uncovered an essential role for these neurons in physiological satiation, as well as a role in generating liraglutide-induced anorectic and weight-reducing effects. Our findings delineate the central neural action of the widely used anti-obesity drug liraglutide and provide new insights into the homeostatic regulation of energy and body weight. GLP-1R
Results
GLP-1R–expressing neurons in the LS are activated by liraglutide.
We initiated our examination by performing immunostaining to define the distribution of endogenous GLP-1Rs within the mouse brain (; supplemental material available online with this article;). The specificity of the anti–GLP-1R antibodies was validated in wild-type and GLP-1R–knockout mice (, B and C). We found densely clustered GLP-1R–positive neurons prominently located in the dorsal portion of the LS (dLS) (), a region implicated in energy control according to previous work from our group and other groups (,). A large number of GLP-1R–positive cells were also detected throughout various hypothalamic nuclei, mainly in the PVN, dorsomedial hypothalamic nucleus (DMH), and Arc, and among scattered neurons in the lateral hypothalamus. A considerable GLP-1R–positive presence was also noted in the hindbrain, including the AP and the external cuneate nucleus. Other brain regions that exhibited considerable GLP-1R expression included the ventral pallidum, anterior pretectal nucleus, pontine nuclei, dentate gyrus, and lateral reticular nucleus (). Supplemental Figure 1A Supplemental Figure 1 Supplemental Figure 1A 32 33 Supplemental Figure 1A https://doi.org/10.1172/JCI178239DS1↗
If a brain region mediates the anorectic effect of GLP-1R agonists, its activity should be modulated by liraglutide. Thus, we systematically mapped brain activation induced by intraperitoneal (i.p.) administration of liraglutide with immunohistochemistry for c-Fos, an early marker of neuronal activity. Liraglutide induced a substantial elevation of c-Fos expression compared with saline controls in the LS, the PVN, and the AP, whereas there was no elevation in other GLP-1R–rich regions (, and). As previously noted, ablating GLP-1Rs in the PVN or the AP does not diminish the anorectic or weight-lowering effects of liraglutide (,). This led us to ask whether GLP-1Rs in the LS contribute to the pharmacological effects of liraglutide. Figure 1, A and B Supplemental Figure 1D 17 26
Then we labeled GLP-1R–positive neurons in the LS by crossingmice withreporter mice. Specific expression of red fluorescent protein tdTomato in GLP-1R–positive neurons was observed in the LS (, A–C). Systematic administration of liraglutide (i.p.) markedly upregulated c-Fos expression in LSneurons (, D–F), suggesting activation of these neurons. To exclude potential influences of network effects caused by systematic administration, we cannulated above the dLS for localized infusion (). This local infusion of liraglutide into the LS also markedly upregulated c-Fos expression in LSneurons but not in GLP-1R–negative neurons (). Concurrently, a substantial reduction in food intake and body weight was also observed in mice with local liraglutide infusion (, G–K). GLP-1R-ires-Cre Ai14 Supplemental Figure 2 Supplemental Figure 2 Figure 1C Figure 1, D and E Supplemental Figure 2 GLP-1R GLP-1R
To probe the potential mechanism underlying the activation of LSfollowing liraglutide infusion, we then performed targeted whole-cell patch clamp recording from LSneurons (). Liraglutide dramatically increased the amplitude but not the frequency of excitatory postsynaptic currents in GLP-1–positive neurons. This result is in accordance with a previous report that a GLP-1 analog (exendin-4) enhanced excitatory synaptic transmission through AMPA receptors trafficking in the PVNneurons () (). GLP-1R GLP-1R CRH Figure 1F 16 Figure 1, G and H
Together, these findings demonstrate that LSneurons can be effectively activated by liraglutide, possibly through the enhancement of excitatory synaptic transmission. GLP-1R
GLP-1R knockdown in the LS attenuates the anorectic and body weight–reducing effects of liraglutide.
To elucidate the role of GLP-1Rs in the LS in mediating the anorexigenic effects of liraglutide, we used CRISPR/Cas9 technology to specifically knock down GLP-1R expression in LSneurons. We injected an adeno-associated virus (AAV) containing–targeting single-guide RNA (sgRNA) and Cre-inducible mCherry into the dLS ofmice bred with Cre-inducible Cas9 knockin mice () (). Through this approach, we achieved an approximately 90% reduction in GLP-1R expression in sg–targeted neurons compared with controls (). GLP-1R GLP-1R GLP-1R-ires-Cre GLP-1R 34 Figure 2A Figure 2B
Remarkably, we found that the knockdown of dLS GLP-1Rs did not influence food intake or body weight in mice at the baseline, regardless of dietary conditions (, A–E). As expected, acute i.p. injection of various liraglutide dosages markedly reduced food intake in control mice (, and, F–I). However, the reduction in food intake was dramatically diminished in the GLP-1R–knockdown group (, and, F–I). Furthermore, GLP-1R knockdown in the LS attenuated body weight reduction induced by liraglutide (, and, F–I). A prolonged treatment experiment spanning 3 weeks further substantiated these findings. While liraglutide effectively reduced body weight in wild-type mice, the effect was dramatically blunted in mice with GLP-1R knockdown in the LS (). Supplemental Figure 3 Figure 2, C–H Supplemental Figure 3 Figure 2, C–H Supplemental Figure 3 Figure 2, C–H Supplemental Figure 3 Figure 2, I–K
We also examined the role of GLP-1Rs in other brain regions, such as PVN and Arc, with the same CRISPR/Cas9–mediated knockdown approach. However, neither the targeted knockdown of GLP-1Rs in the PVN nor the Arc altered the anorectic properties following liraglutide (). Collectively, these findings demonstrate that GLP-1Rs in the LS contribute to the anorectic effects of liraglutide. Supplemental Figure 4
GLP-1R overexpression in the dLS leads to decreased food consumption in satiated mice.
To elucidate the influence of GLP-1R upregulation in LSneurons on food intake, we used a specific AAV vector encoding Cre-inducible mouse GLP-1Rs, complemented with a hemagglutinin (HA) epitope tag, for targeted injection into the dLS ofmice (). Subsequent immunohistochemical analyses confirmed the overexpression of GLP-1R (, B and C). This overexpression in the LS led to a substantial decrease in food consumption under sated conditions, but not in fasted mice (, D–G). The absence of effect in fasted mice may be due to negligible endogenous GLP-1 release in fasted animals (,). Thus, overexpression of GLP-1R in the LS is sufficient to lower food consumption in sated mice. GLP-1R GLP-1R-ires-Cre Supplemental Figure 5A Supplemental Figure 5 Supplemental Figure 5 4 7
Next, we assessed the effect of GLP-1R overexpression in the LS on chronic body weight and metabolic parameters. Overexpression of GLP-1Rs did not change chronic body weight compared with that of EYFP-expressing control mice (). Overexpression of GLP-1Rs in the LS also did not change oxygen uptake (VO), carbon dioxide discharge (VCO), respiratory exchange ratio, and average energy expenditure (, I–P). Supplemental Figure 5H Supplemental Figure 5 2 2
Silencing of LSneurons increases food consumption and promotes obesity induced by high-fat food. GLP-1R
Having identified the vital role of LS GLP-1Rs in mediating the anorectic and weight-reducing effects of liraglutide, we sought to elucidate the specific function of LSneurons in the regulation of energy and body weight homeostasis by conducting loss-of-function experiments. GLP-1R
RNA hybridization has shown that all LS GLP-1R neurons express the vesicular GABA transporter (vGAT), indicative of inhibitory neurons, and none express the excitatory marker vesicular glutamate transporter 2 (vGluT2). Furthermore, among these LSneurons, 39% exhibited somatostatin (Sst) expression, and 23% showed neurotensin (Nts) expression (, A and B). We then used tetanus neurotoxin (TeNT), a specialized protease known to cleave synaptobrevin-2, thereby blocking neurotransmitter release, as a molecular tool for synaptic silencing (). To inactivate LSneurons, we injected AAV expressing double-floxed EGFP and TeNT (AAV-DIO-EGFP-2A-TeNT) into the bilateral LS ofmice to achieve specific expression of EGFP-2A-TeNT in LSneurons (and). On an ad libitum high-fat diet, TeNT mice exhibited greater body weight gain in comparison with EYFP control mice (); however, this difference was not observed under chow-fed condition (). Consistent with this, TeNT-mediated silencing of LSneurons markedly increased the consumption of palatable food but not standard chow (). GLP-1R GLP-1R GLP-1R GLP-1R Supplemental Figure 6 35 Figure 3A Supplemental Figure 6C Figure 3B Supplemental Figure 7A Figure 3, C and D GLP-1R-ires-Cre
Considering that several studies have associated the inhibition of central GLP-1R with aggravated glucose tolerance (), we conducted oral glucose tolerance tests to investigate the role of LSneurons. Interestingly, there was no difference between TeNT and control mice in glucose tolerance (). 36 Figure 3E GLP-1R
We further explored the effect of TeNT expression on liquid food intake during fixed-interval food delivery and operational food self-administration tasks. TeNT expression markedly increased the number of licks and Ensure intake, and the TeNT mice also showed markedly more active poking to obtain Ensure compared with controls (, and). Silencing LSneurons, on the other hand, did not increase water consumption (, C–E), implying that the higher intake of Ensure was not due to the increased need for water. Figure 3, F–H Supplemental Figure 7B Supplemental Figure 7 GLP-1R
We also assessed the effects of LSneuronal silencing on metabolic parameters. Both VOand VCOwere markedly higher in the TeNT mice compared with controls, particularly during the active phase (i.e., the dark cycle) of the day (, F–I). Despite these differences, the respiratory exchange ratio — the ratio of VCOproduced to VOconsumed — remained unchanged (, J and K), indicating no substantial differences in the metabolic utilization of carbohydrates and fats between the TeNT mice and the controls. Interestingly, energy expenditure (EE) was also markedly higher in the TeNT mice (, L and M), possibly because of an increased basal metabolic rate. GLP-1R 2 2 2 2 Supplemental Figure 7 Supplemental Figure 7 Supplemental Figure 7
Mindful of studies implying an anxiogenic role for the LS (–), we explored whether the altered consumption could be attributed to changes in anxiety levels and found that LS:TeNT mice exhibited similar anxiety levels and locomotor activity in an open-field test compared with control mice (, N and O). This suggests that the altered consumption was unlikely to result from changes in anxiety level or locomotion. Together, these data demonstrate that inactivation of LSneurons increases food consumption and promotes obesity induced by high-fat food. 37 39 Supplemental Figure 7 GLP-1R GLP-1R
LSneurons contribute to the anorexigenic and body weight–reducing effect of liraglutide. GLP-1R
To investigate the role of LSneurons in the anorexigenic and weight-reducing effects of liraglutide, we measured the effects of liraglutide on the reduction of food intake and body weight in mice with active (control: AAV-DIO-EYFP) and inactive (TeNT: AAV-DIO-EGFP-2A-TeNT) LSneurons. Liraglutide treatment markedly decreased food consumption and body weight in the control group under both chow-fed and high-sucrose-fed conditions. However, silencing of LSneurons markedly reduced these effects at various doses of liraglutide (, and). GLP-1R GLP-1R GLP-1R Figure 3, I–L Supplemental Figure 8
Nausea is one of the most common adverse side effects reported with GLP-1R agonists, such as liraglutide (–). To investigate whether liraglutide-induced nausea requires the action of LSneurons, we used the standard conditioned taste avoidance (CTA) model () (). Intraperitoneal administration of liraglutide consistently induced a robust CTA across both the TeNT and control groups, with no substantial differences observed between the two (, Q and R). These findings indicated that the LSneurons were not involved in the liraglutide-induced nausea. 40 42 40 Supplemental Figure 7P Supplemental Figure 7 GLP-1R GLP-1R
In summary, these results support the notion that LSneurons play an essential role in mediating the anorexigenic and body weight–reducing effects of liraglutide. GLP-1R
A rapid decrease of LSneuronal activity during the eating episode. GLP-1R
Next, to explore the intrinsic activity of LSneurons during naturalistic feeding, we recorded Casignals using fiber photometry under freely moving conditions (). We observed a pronounced decrease in Casignal at the initiation of food consumption in food-deprived mice, and this reduction persisted throughout the eating episode, returning to baseline levels once feeding ceased (and). Interestingly, when mice were exposed to more palatable food sources, such as high-sucrose or high-fat items, rather than standard chow, the decrease in neuronal activity was more pronounced (). This result indicates that the dynamics of LSactivity were modulated by food palatability. GLP-1R 2+ 2+ GLP-1R Figure 4A Figure 4B Supplemental Video 1 Figure 4, C–F
To test whether the activity of LSneurons is modulated by energy homeostasis, we presented a pellet of high-fat food to satiated mice. Compared with fasted mice, satiated mice exhibited a markedly smaller inhibitory response during consumption (). To examine this phenomenon more closely, we divided the feeding behavior into 3 stages (initial, middle, and final) and observed a gradually diminishing inhibitory response from the initial to the final stage (). These results demonstrate that the current energy state can actively modulate LSneuronal activity during feeding. GLP-1R GLP-1R Figure 4, G–J Figure 4, K and L
We extended our observation to the food-seeking phase by allowing fasted mice to visually perceive but not physically access a caged food pellet, thus triggering typical seeking behaviors. Despite these actions, no discernible alteration in the Casignal was detected (), demonstrating that food seeking does not modulate dLSneuronal activity. 2+ GLP-1R Figure 4M
Activation of LSneurons reduces food consumption. GLP-1R
According to our fiber photometry data, we propose that this specific cluster of neurons functions analogously to a brake pad in the regulation of feeding behavior. In the reduced activity state, these neurons facilitate the continuation of feeding. Conversely, when activity is heightened, they play a crucial role in suppressing food intake. To examine whether activation of LSneurons is sufficient to suppress food intake independent of endogenous GLP-1R or exogenous liraglutide, we injected an AAV into the dLS ofmice expressing the excitatory Gq-coupled designer receptor exclusively activated by designer drugs (DREADD) hM3D-mCherry in a Cre-dependent fashion (). This method resulted in the selective expression of hM3D-mCherry in LSneurons (). Intraperitoneal injection of the DREADD agonist clozapine-oxide (CNO) induced robust c-Fos expression in LSneurons transduced with hM3D, indicating increased neuronal activity (). GLP-1R GLP-1R GLP-1R GLP-1R-ires-Cre N 43 Figure 5A Figure 5B
Next, we performed feeding assays after i.p. injection of CNO or saline. Compared with saline, DREADD-based activation of LSneurons via CNO reduced a 2-hour consumption period of both standard chow and high-sucrose food during the dark period when mice normally consume food actively (). In an overnight fast condition, activation of LSneurons markedly reduced food intake for both standard chow and high-sucrose food (, A and B). This result suggests that the activity of these neurons can overcome the motivational hunger drive and suppress feeding. Considering prior evidence that GLP-1R activation augments glucose tolerance (,), we probed the potential of LSneurons to modulate glucose homeostasis. Activation of LSneurons did not affect glucose homeostasis (), supporting the role of these neurons in anorectic effects rather than glucose-lowering effects. GLP-1R GLP-1R GLP-1R GLP-1R Figure 5, C and D Supplemental Figure 9 44 45 Figure 5E
We further investigated the long-term effects of chronic LSneuronal activation. CNO was administered bi-daily for 1 week continuously to mice expressing hM3D-mCherry. Chronic CNO administration effectively suppressed both food intake and body weight, which were recovered upon cessation of CNO. In control mice with mCherry expression, the same CNO injection failed to change food intake and body weight (). GLP-1R Figure 5, F–H
To explore whether LSneuronal activity acts as negative or positive valence, we implemented optogenetic activation by expressing channelrhodopsin (ChR2-mCherry) selectively in LSneurons (). In a real-time place preference/avoidance test, LS:ChR2 mice exhibited a substantial avoidance of the photostimulation-paired chamber (, C–E). To examine whether transient activation of LSneurons is sufficient to suppress feeding, we trained LS:ChR2 mice to consume Ensure on a fixed ratio 1 schedule, and photostimulation (20 Hz, 1.5 seconds) was synchronized with Ensure administration (). Photostimulation of LSneurons decreased the nose-poking and Ensure consumption (). Collectively, these experiments suggested that LSneurons encoded negative valence and suppressed appetite. GLP-1R GLP-1R GLP-1R GLP-1R GLP-1R GLP-1R GLP-1R Figure 5I Supplemental Figure 9 Figure 5J Figure 5K
LSneurons project to multiple brain areas involved in feeding control. GLP-1R
Finally, we sought to elucidate the specific projection targets of LSneurons that are integral to food regulation. We injected AAV-FLEX-tdTomato-T2A-synaptophysin-EGFP into the LS ofmice (, A and B). With this tracing strategy, axonal fibers are labeled in red tdTomato, while synapses projecting from LSneurons are labeled in green EGFP (). We found that LSneurons made major synaptic connections with multiple brain regions (, C and D), including the nucleus of the horizontal limb of the diagonal band (HDB), the lateral and medial preoptic area (LPO/MPA), the lateroanterior hypothalamic nucleus (LA), the lateral hypothalamic area (LH), the medial tuberal nucleus (MTu), the lateral and medial habenular nucleus (LHb/MHb), the CA3 field of the hippocampus (CA3), and the pyramidal cell layer of the hippocampus (Py). Previous research has suggested that multiple hypothalamic brain areas, including the LH and MTu mentioned above (,), are involved in feeding regulation. Thus, LSneurons project to multiple brain areas that are implicated in feeding regulation. GLP-1R GLP-1R GLP-1R GLP-1R GLP-1R-ires-cre Supplemental Figure 10 35 Supplemental Figure 10 46 47
Discussion
Liraglutide, a long-acting synthetic GLP-1 analog, suppresses body weight in both humans and preclinical models via a reduction in appetite and energy metabolism (,,). Since its approval for obesity treatment in 2014, liraglutide has been widely prescribed worldwide and has generated sales of billions of dollars (). However, the underlying mechanisms and specific sites of action contributing to its anorectic properties remain obscure. Our findings illustrate that liraglutide activates GLP-1R–expressing neurons in the dorsal LS (dLS) (), and knockdown of GLP-1Rs in the dLS, but not the Arc or PVN, diminishes liraglutide's weight-reduction effects (). The silencing of LSneurons enhances food consumption and obesity and diminishes liraglutide's anorectic effects (). During food consumption, there is a decrease in the activity of LSneurons (), and chemogenetic or optogenetic activation mimics liraglutide's anorectic effects (). These results emphasize the critical role of LSneurons and LS GLP-1Rs in feeding regulation, not only in orchestrating energy metabolism via endogenous GLP-1 but also in mediating anorectic and body weight–lowering effects of liraglutide. 41 48 49 50 Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 GLP-1R GLP-1R GLP-1R
As is the case with native GLP-1, the anorectic and weight-lowering effects of liraglutide require central GLP-1R activation (,). GLP-1Rs are distributed across the CNS, notably in areas critical for the regulation of food intake and body weight, such as the Arc and the PVN. Other regions that have GLP-1Rs, such as the dLS, have received little attention. Despite prior evidence linking endogenous GLP-1 in the LS with increased satiety (), the contribution of these GLP-1Rs to the pharmacological effects of liraglutide was unknown. Hence, we employed a sophisticated approach using viral strategies in concert with transgenic animal models to determine the physiological function of GLP-1Rs in the dLS. This allowed us to elucidate with high temporal and spatial precision the essential role these receptors play in the food intake and body weight reduction effects of liraglutide. 4 9 30
Our data reveal that while GLP-1R knockdown in the LS does not alter basal food intake or body weight, TeNT-mediated inactivation of these neurons promotes food intake and weight gain on a high-fat diet. These likely stem from subtle but important methodological differences. Selective knockdown of GLP-1Rs preserves other neuronal functions, allowing continued engagement of LSneurons in neural circuits via alternative receptors and signaling pathways. Conversely, TeNT-mediated synaptic silencing completely blocks neurotransmitter release from LSneurons and shuts down all physiological functions mediated by these neurons (,). Thus, our results suggest that the synaptic activity of LSneurons plays a crucial role in the regulation of food intake and body weight. Upon GLP-1R knockdown, additional signaling pathways beyond GLP-1 are involved to regulate LSneuronal activity and energy homeostasis. GLP-1R GLP-1R GLP-1R GLP-1R 33 35
Our data show that GLP-1Rs in the dLS are a crucial contributor of liraglutide's anorectic and body weight–reducing effects. These receptors likely collaborate with GLP-1Rs in other brain regions, collectively mediating the anorectic and body weight–lowering effects of liraglutide (,,). Evidence supporting this concept is provided by the binding of fluorescently labeled liraglutide to neurons in the Arc and specific hypothalamic regions in mice (). Direct intra-hypothalamic GLP-1R agonism also induces anorexia (,,–). Despite these findings, our data revealed that the knockdown of GLP-1Rs in the Arc and PVN did not alter liraglutide's anorectic effects. Other studies corroborate this view, showing preserved anorectic effects of peripheral liraglutide after hypothalamic GLP-1R knockdown (,), with no subsequent hypothalamic neuronal activation (,,). Thus, while the hypothalamus may play a role, it is likely not the primary locus of action mediating the anorectic effects of GLP-1 and GLP-1 derivatives. The vital contribution of GLP-1Rs in the dLS is clear and should not be overlooked in our understanding of these multifaceted mechanisms. 9 26 49 49 15 17 51 53 17 18 17 54 55
Liraglutide, along with other GLP-1R agonists, not only diminishes food intake but also frequently elicits some side effects, such as nausea (,,). The anorectic and nausea-inducing responses to these agents critically depend on the activation of GLP-1Rs within the central nervous system (CNS) (–,,,,,); whether these effects arise from different brain regions remains an open question. Our data suggest that silencing of LSneurons diminishes the anorectic effect but preserves the nausea response induced by liraglutide (and, P–R), implying the involvement of other brain regions in nausea. Supporting this, GLP-1Rs in the nucleus tractus solitarius (), locus caeruleus, and central amygdala have been reported to mediate nausea induced by GLP-1R agonists (,). Thus, our results suggest that targeting GLP-1R in the LS or LSneurons could preserve the anorectic and body weight–lowering effects of liraglutide while avoiding side effects such as nausea. 41 42 56 9 11 23 25 26 40 57 Figure 3 Supplemental Figure 7 40 52 58 GLP-1R GLP-1R
Our fiber photometry experiments revealed a larger decrease in LSneuronal Cadynamics when mice consumed palatable food as opposed to standard chow. This observation prompted us to hypothesize that alterations in LSneuronal activity exert a more profound influence on palatable food consumption. This potentially explains why the silencing of LSneurons impacted food intake and body weight only when mice were exposed to palatable food, leaving the effects on standard chow unaltered. In the context of Ensure, identified as highly palatable, more conspicuous phenotypes were observed, resonating with the above findings. GLP-1R 2+ GLP-1R GLP-1R
Collectively, these data not only underscore the complexity of the underlying mechanisms but also demonstrate the integral involvement of LSneurons in mediating the effect that liraglutide has on food intake and body weight. Such insights reaffirm the importance of GLP-1Rs in the dLS and pave the way for targeted interventions in the future. GLP-1R
Methods
Sex as a biological variable
Our study examined male and female animals, and similar findings are reported for both sexes.
Animals
All animal husbandry and experimental procedures were approved by the Animal Care and Use Committees at the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences (CAS). Adult male C57BL/6J mice (3–5 months old; Guangdong Medical Laboratory Animal Center),mice (Shanghai Model Organisms, NM-KI-200134), GLP-1R-KO mice (Shanghai Model Organisms, NM-KO-190803),mice (The Jackson Laboratory, 007908), andmice (The Jackson Laboratory, 024857) were used for this study. Mice were maintained at a constant temperature of 22°–25°C, following a 12-hour light/12-hour dark cycle. GLP-1R-ires-Cre Ai14:Rosa26-loxP-STOP-loxP-tdTomato Rosa26-LSL-Cas9
Stereotaxic surgery
Virus injections.
Mice were anesthetized using pentobarbital (80 mg/kg). Stereotaxic injections delivered 200–300 nL of the respective virus into the dLS (coordinates: anteroposterior [AP], +0.75 mm; mediolateral [ML], +/– 0.30 mm; dorsoventral [DV], –2.30 mm), the Arc (coordinates: AP, +1.94 mm; ML, +/– 0.30 mm; DV, –5.75 mm), and the PVN (coordinates: AP, +0.75 mm; ML, +/– 0.25 mm; DV, –4.50 mm) at a rate of 60 nL/min. After injection, mice were given a minimum recovery period of 3 weeks before initiating behavioral experiments. All the virus information is summarized in. Supplemental Table 1
Optic fiber and infusion cannula implantation.
Twenty minutes after the unilateral injection of AAV2/9-hEF1a-DIO-hChR2(H134R)-mCherry-WPRE-pA into the dLS, optic fibers (200 μm diameter; numerical aperture 0.37) were implanted. These fibers were positioned above the right dLS with the following coordinates: 0.75 mm anterior, 0.30 mm lateral, and 2.10 mm ventral from the dura. The ferrule was then securely attached to the skull using light-curable resin. After implantation, the incision was sutured, and antibiotic ointment was applied to the surgical site. Optogenetic and fiber photometry studies commenced after a 3-week post-surgery recovery period.
Infusion cannulas were implanted over the unilateral LS (coordinates: AP, +0.75 mm; ML, –0.30 mm; DV, –2.10 mm) and were anchored to the skull using dental cement. After the procedure, a dummy cannula was inserted, topped with a cap, ensuring that the guide cannula remained unobstructed. A recovery period of a minimum of 2 weeks was provided to the mice before initiation of behavioral tests.
At the study's completion, the locations of virus expression and optic fiber placement were verified post hoc in all animals, and this information is summarized in. Supplemental Figure 11
Fiber photometry
Fiber photometry recordings were initiated at least 3 weeks after surgery using the fiber photometry system ThinkerTech. A 470 nm blue LED light (30 μW) was channeled through a singular implanted ferrule. The resulting signals were analyzed using a custom MATLAB script. Thescore was determined as (− μ)/σ, where the mean and standard deviation were derived from the signal 5 seconds before stimulation. z x
For recordings during free-moving feeding, mice — either food-deprived for 24 hours or fed normally — were introduced to a transparent acrylic enclosure measuring 15 × 15 × 30 cm and then connected to the fiber photometry apparatus. A 5-minute habituation period was followed by the gentle placement of a food pellet (choices included chow or high-sucrose or high-fat content) in the center of the box. The system simultaneously captured the free-feeding behavior and the GLP-1R:GCaMP signals for 15 minutes. All obtained data were segmented and time-aligned according to individual behavioral bout onsets. Fluorescence signals were normalized usingscores, taking into account the mean and standard deviation from time windows of −5 to 0 seconds before the first bite and −10 to 0 seconds leading up to the last bite. Any behavioral bouts lasting fewer than 10 seconds were disregarded. z
Behavioral assays
Drug administration.
Clozapine-oxide (CNO; Enzo Life Sciences, BML-NS105) was injected i.p. at 2 mg/kg. This was a within-subjects design, and all mice were treated with both CNO and saline vehicles in a counterbalanced approach. Liraglutide (GL Biochem) was administered i.p. 30 minutes before dark onset at doses of 200, 100, or 50 μg/kg. During our study on the effects of LSneuron inactivation and lateral septal GLP-1R knockdown, both test and control groups were given drugs and vehicles in a counterbalanced method in a between-subjects design. N GLP-1R
Solid food intake assays.
Mice were individually housed for a minimum of 3 days before assay. Daily food (standard chow or high-sucrose) allocations of about 20 g were provided, and cages were refreshed daily. Food consumption was manually determined early in the dark phase (8–10 pm), with caloric intake deduced from a burden sheet. Diet shifts were introduced according to experimental needs, and each shift was given a 3-day acclimation period before intake measurements. For liraglutide treatment, intake was measured 30 minutes after administration of liraglutide. In chemogenetic activation experiments to study acute activation effects, CNO was administered i.p. 30 minutes before the dark phase, and food intake during the 2 hours prior to the dark phase was measured. To study chronic long-term effects, CNO was administered i.p. every 12 hours for a continuous period of 7 days, with daily measurements of food intake and body weight. All the food ingredients are summarized in. Supplemental Table 2
Liquid food intake assays.
Before experiments, mice were individually housed with unrestricted access to food and water. For the duration of the study, mice were placed in an operant conditioning chamber outfitted with a nose-poke system (22 × 16 × 15 cm; AniLab). The time stamps of licks and nose pokes were collected, and subsequent data analysis was performed using a custom MATLAB script.
During the fixed-interval delivery assay, a 10 μL droplet of Ensure solution (Abbott) was offered every 10 seconds, totaling 180 presentations. Mice underwent training to lick the spout from day 1 to day 3. Data collection commenced on day 4.
During the self-administration food intake (fixed ratio 1) assay, one port was randomly selected as the "active" port. Poking this active port yielded an Ensure solution reward, while pokes to inactive ports went unrewarded. The test spanned 3 consecutive 45-minute daily sessions, with data from the third day being the focus of analysis. When this test was paired with optogenetic manipulation experiments, mice received intracranial stimulation (473 mm, 10–15 mW, 20 Hz, 1.5 seconds) concurrent with the nose poke on day 4, and data were subsequently captured.
Real-time place preference/avoidance.
In the real-time place preference/avoidance (RTPP/A) assay, a specially designed 2-chamber setup was used. Before testing, mice were carefully connected to the optical fiber. On the first day, mice roamed freely within the apparatus for 15 minutes, in the absence of any light stimulation. The following day, one side of the apparatus, chosen in a counterbalanced manner, was designated for stimulation. Upon entry to this side, the mouse received a 20 Hz laser pulse (473 nm, with a 20-millisecond pulse duration and intensity of 5–10 mW/mmper side). This stimulation ceased once the mouse moved back to the opposite chamber. Place preference scores were computed by measuring the time spent in the stimulated chamber and then contrasting this with the baseline duration. 2
Open-field test.
For behavioral testing, mice were moved to the testing environment 3 hours before the test, where they acclimated to the new surroundings for the duration. Before each test, the equipment was meticulously cleaned using a 20% ethanol solution to ensure the removal of residual odors from previous test mice. Mice were then introduced to a square open-field arena crafted from plastic, measuring 40 cm by 40 cm. Their movements within this space were continuously tracked from above by a camera and using custom MATLAB code, operating at a 30 Hz capture rate. For analysis, the arena was conceptually bisected into 2 distinct zones: a central square of 20 cm by 20 cm, and the surrounding peripheral space. Both the overall movement patterns and the proportion of time the mice allocated to the central region were subjects of our analytical focus.
Conditioned taste avoidance.
The experimental procedure was adapted from a previous report (). In brief, mice were individually housed and deprived of water for 24 hours on day 1. Then, for 3 consecutive days, the mice were simultaneously presented with 2 water bottles for 90 minutes at 6 pm, followed by an i.p. injection of saline. On day 5, each mouse was introduced to 2 bottles of a new beverage (0.2% saccharin mixed with 0.15% cherry-flavored or grape-flavored Kool-Aid) during the regular water presentation time, followed by an i.p. injection of saline. On day 6, the procedure was repeated but with a different Kool-Aid flavor (if cherry was used on day 5, grape was used on day 6, and vice versa), followed by an i.p. injection of liraglutide (200 μg/kg). Days 7 and 8 replicated the procedures of days 5 and 6, respectively. To counteract potential taste preference bias, the order of cherry-flavored and grape-flavored beverages was balanced within each group. Finally, on day 9, mice were presented with both flavors simultaneously, with the water bottles being switched every 30 minutes. The total liquid consumption from both bottles was recorded over a 90-minute period. 40
Body weight measurement
Body weight was consistently measured manually at the same daily interval. In experiments involving TeNT-induced synaptic inactivation and GLP-1R overexpression, mice were provided with either standard chow or high-fat food. For experiments focused on the knockdown of LS GLP-1Rs, mice received daily i.p. injections of liraglutide at a dose of 200 μg/kg of body weight. For chemogenetic activation experiments, mice received semidiurnal i.p. injection of CNO at a dose of 2 mg/kg body weight.
Energy expenditure measurement
Mice were individually housed and acclimated to metabolic cages (TOW-INT Tech.) for at least 3 days before testing under a 12-hour light/12-hour dark cycle. Energy expenditure, VO, VCO, respiratory exchange ratio, food intake, and water intake were measured continuously and simultaneously at a sampling frequency of 5 times per minute. Data were exported to Excel and analyzed using MATLAB (MathWorks). Light and dark cycles were indicated by white (6 am to 6 pm) and black (6 pm to 6 am), respectively. Food and water were freely available during testing. The gas sensors for COand Owere calibrated before each test. 2 2 2 2
Oral glucose tolerance test
Blood glucose was measured by making a small horizontal incision on the mouse's tail end using a razor blade. A drop of blood was gently massaged out, and its glucose concentration was determined using a Bayer blood glucose meter. Throughout this procedure, food was withheld. In the TeNT-induced synaptic inactivation experiments, blood glucose was gauged after an overnight fast of 16 hours. Subsequently, these fasting mice were given an oral dose of 1.5 g/kg 30%-glucose solution (Sigma-Aldrich, catalog G6125). Blood samples were taken at baseline, and then at intervals of 10, 30, 60, 90, and 120 minutes after glucose administration. For chemogenetic activation experiments, fasting blood glucose levels (the baseline at time zero) were measured 30 minutes after administration of CNO injections. d
Histological procedures
Mice were euthanized using a pentobarbital sodium overdose and then transcardially perfused with phosphate-buffered saline (PBS; pH 7.4), followed by 4% paraformaldehyde (PFA). The brains underwent postfixation in 4% PFA for 8–12 hours and were subsequently dehydrated in 30% sucrose for 24–48 hours or until fully submerged. The tissue was then embedded in Tissue-Tek OCT compound (Sakura) and frozen with dry ice. Using a cryostat (Leica), 40-μm sections were prepared, with free-floating sections stored in PBS.
For staining, sections were initially rinsed with PBS 3 times, 10 minutes each. They were then blocked at room temperature using a solution of 10% normal goat serum and 0.3% Triton X-100, followed by an overnight incubation at 4°C with the primary antibody. After three 10-minute washes with PBST, sections were incubated with a fluorophore-linked secondary antibody for 2 hours and then counterstained with DAPI (1:3,000 dilution). Images were captured using an Olympus Virtual Slide Microscope (VS120-S6-W) and later analyzed by an evaluator unaware of the experimental group identities. Antibody information is summarized in. Supplemental Table 3
In situ hybridization
Mouse brains were sectioned into 18-μm slices using a Leica cryostat and affixed to SuperFrost Plus microscope slides. Probes targeting GLP-1R (catalog 418851-C3), vGAT (catalog 319191-C3), vGluT2 (catalog 319171), somatostatin (Sst; catalog 404631), and neurotensin (Nts; catalog 420441) were specifically designed and validated by Advanced Cell Diagnostics. All fluorescence in situ hybridization experiments were performed using the RNAscope v2 Assay (catalog 323100) following the manufacturer's guidelines. Briefly, brain slices were air-dried at 39°C for 2 hours, washed with 1× PBS, treated with 3% hydrogen peroxide in methanol and Target Retrieval buffer (Advanced Cell Diagnostics [ACD]) for 15 minutes, and subjected to RNAscope Proteinase III incubation at 40°C for 15 minutes. The brain slices were subsequently incubated with mRNA probes at 40°C for 2 hours. Specific signals were then amplified using a multiplexing amplification solution, and detection was carried out using the TSA Plus Fluorescence Kit (Advanced Cell Diagnostics, 322809). Subsequent quantitative analyses were performed using ImageJ software (NIH).
Cell counting
For cell counting of c-Fos, tdTomato, and GLP-1Rmarkers, we obtained 40 μm coronal sections from the target brain areas of each mouse. Images were captured using an Olympus Virtual Slide Microscope (VS120-S6-W). A custom MATLAB script was used for cell counting, while colocalization was visually determined. + + +
Electrophysiological recordings
Coronal 250- to 300-μm slices containing the LS were prepared using a vibratome (Leica, VT-1000S) in an ice-cold choline-based solution containing (in mM) 110 choline chloride, 2.5 KCl, 0.5 CaCl, 7 MgCl, 1.3 NaHPO, 1.3 Na-ascorbate, 0.6 Na-pyruvate, 25 glucose, and 25 NaHCO, saturated with 95% Oand 5% CO. Slices were incubated in 32°C oxygenated artificial cerebrospinal fluid (in mM: 125 NaCl, 2.5 KCl, 2 CaCl, 1.3 MgCl, 1.3NaHPO, 1.3 Na-ascorbate, 0.6 Na-pyruvate, 25 glucose, and 25 NaHCO) for at least 1 hour before recording. Slices were transferred to a recording chamber and superfused with 2 mL/min artificial cerebrospinal fluid. Patch pipettes (2–5 MΩ) pulled from borosilicate glass (World Precision Instruments, PG10150-4) were filled with a Cs-based low-Clinternal solution containing (in mM) 135 CsMeSO, 10 HEPES, 1 EGTA, 3.3 QX-314, 4 Mg-ATP, 0.3 Na-GTP, 8 Na-phosphocreatine, 290 mOsm/kg, adjusted to pH 7.3 with CsOH. The whole-cell voltage-clamp recording was performed at room temperature with a Multiclamp 700B amplifier and a Digidata 1440A (Molecular Devices). Data were sampled at 10 kHz and analyzed with Clampfit (Molecular Devices) or MATLAB. The excitatory postsynaptic currents were recorded at a holding potential of –70 mV. 2 2 2 4 3 2 2 2 2 2 4 3 3 2 –
Statistics
All experiments were conducted blinded. All statistical data can be found in the figure legends. Statistical significance was set at *< 0.05, **< 0.01, ***< 0.001, ****< 0.0001. Data are presented as means ± SEM. Sample sizes and sex distribution for figures are listed in. P P P P Supplemental Table 4
Study approval
All procedures performed were approved by the Animal Care and Use Committees at the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences (CAS).
Data availability
All data sets generated or analyzed during this study are included in the published article and its supplemental material. The supporting data values are available in thefile. Any additional information required to reanalyze the data reported in this paper is available upon request. Supporting Data Values
Author contributions
ZC, XD, CS, YT, JZ, GC, HJ, and YX performed the experiments. YL and YZ supervised the analysis. ZC and YZ designed all the experiments and wrote the manuscript with input from the coauthors.