Glucagon-like peptide-1 regulates brown adipose tissue thermogenesis via the gut-brain axis in rats

Abstract

Endogenous intestinal glucagon-like peptide-1 (GLP-1) controls satiation and glucose metabolism via vagal afferent neurons (VANs). Recently, VANs have received increasing attention for their role in brown adipose tissue (BAT) thermogenesis. It is, however, unclear whether VAN GLP-1 receptor (GLP-1R) signaling affects BAT thermogenesis and energy expenditure (EE) and whether this VAN mechanism contributes to energy balance. First, we tested the effect of the GLP-1R agonist exendin-4 (Ex4, 0.3 μg/kg ip) on EE and BAT thermogenesis and whether these effects require VAN GLP-1R signaling using a rat model with a selective Glp1r knockdown (kd) in VANs. Second, we examined the role of VAN GLP-1R in energy balance during chronic high-fat diet (HFD) feeding in VAN Glp1r kd rats. Finally, we used viral transsynaptic tracers to identify the possible neuronal substrates of such a gut-BAT interaction. VAN Glp1r kd attenuated the acute suppressive effects of Ex4 on EE and BAT thermogenesis. Consistent with this finding, the VAN Glp1r kd increased EE and BAT activity, diminished body weight gain, and improved insulin sensitivity compared with HFD-fed controls. Anterograde transsynaptic viral tracing of VANs infected major hypothalamic and hindbrain areas involved in BAT sympathetic regulation. Moreover, retrograde tracing from BAT combined with laser capture microdissection revealed that a population of VANs expressing Glp1r is synaptically connected to the BAT. Our findings reveal a novel role of VAN GLP-1R signaling in the regulation of EE and BAT thermogenesis and imply that through this gut-brain-BAT connection, intestinal GLP-1 plays a role in HFD-induced metabolic syndrome.

INTRODUCTION

Glucagon-like peptide-1 (GLP-1) is produced peripherally by enteroendocrine L-cells and centrally by neurons in the nucleus tractus solitarii (NTS). Intestinal GLP-1 enhances glucose-stimulated insulin secretion and reduces food intake (see 19 for review). Yet, beyond its well-documented glucoregulatory and satiating properties, the physiological role of intestinal GLP-1 in energy homeostasis remains elusive (16). Centrally administered GLP-1 or GLP-1 receptor (GLP-1R) agonists acutely increase whole body energy expenditure (EE) (20, 41), presumably via hypothalamic GLP-1R and the subsequent activation of thermoregulatory networks that modulate brown adipose tissue (BAT) thermogenesis (7, 29). On the other hand, whether nutrient-induced intestinal GLP-1 secretion affects BAT thermogenesis, and thereby EE, remains unclear (14). Intestinal GLP-1 is very unlikely to reach hypothalamic GLP-1R because of its rapid enzymatic degradation by dipeptidyl peptidase-4 (40). This means that if intestinal GLP-1 controls BAT thermogenesis, it probably engages sensory afferent neural pathways, which then modulate central thermoregulatory networks.

Vagal afferent neurons (VANs) are key neuronal elements for gut-brain communication. They mediate numerous physiological functions, including satiation, gut motility, and hormone release (8). Their role in nutrient sensing and, more broadly, the regulation of energy metabolism is currently receiving considerable attention (12, 13, 26). We have previously demonstrated that the effects of intestinal GLP-1 on eating, gastric emptying, and glycemia are mediated by GLP-1R signaling in VANs (25). Reducing VAN Glp1r expression, however, did not affect daily food intake and body weight in chow-fed rats. Importantly, several recent studies show that vagal afferent mechanisms mediate the effects of gut-derived signals on BAT thermogenesis in a diet-dependent manner. Manipulating lipid-sensing pathways in VANs alleviated high-fat diet (HFD)-induced obesity in mice, mainly by increasing BAT thermogenesis (28, 34). Another recent study demonstrated that the electrical stimulation of VANs decreased BAT sympathetic nerve activity (SNA) and BAT thermogenesis in rats (33). Furthermore, this vagal afferent mechanism contributes to the reduced sympathetic activation of BAT following chronic HFD exposure (32). However, the gut signals that control BAT thermogenesis together with the neurochemical phenotypes of VANs mediating this effect are unknown.

We performed three sets of experiments to test the hypothesis that activating VAN GLP-1R modulates EE and BAT thermogenesis in rats and to investigate underlying mechanisms and relevant neural circuits. First, we assessed whether the full expression of VAN Glp1r is necessary for the acute reduction of EE and BAT thermogenesis induced by intraperitoneally administered GLP-1R agonist exendin-4 (Ex4) in chow-fed animals. Second, we used viral-mediated knockdown (kd) of Glp1r in VANs to examine the physiological relevance of VAN GLP-1R signaling for EE and BAT thermogenesis during chronic HFD feeding. Third, we performed anterograde viral tracing from nodose ganglia (NG) and retrograde viral tracing from the interscapular BAT (iBAT) to reveal synaptic connections between VANs and the brain regions responsible for the sympathetic control of BAT thermogenesis. Collectively, our results support a novel diet-dependent role of endogenous intestinal GLP-1 that fine-tunes EE and BAT thermogenesis via a gut-brain neuronal pathway.

METHODS

Animals and Housing

Male Sprague-Dawley rats (Charles River Laboratories) were housed in a temperature- and humidity-controlled environment (21 ± 1°C, 55 ± 5% relative humidity) and maintained on a 12 h/12 h light-dark cycle (lights on: 11 PM, lights off: 11 AM). Standard laboratory chow (3.13 kcal/g, 3433, Kliba Nafag) or an HFD (60% energy from fat, 5.16 kcal/g, D12492, ssniff-Spezialdiäten) was provided ad libitum as described below. All procedures were approved by the Zurich Cantonal Veterinary Office or the Animal Care and Use Committee of the Oregon Health and Science University.

Cohort 1

Acute effects of intraperitoneal Ex4 in chow-fed rats on EE, respiratory exchange ratio, and interscapular skin temperature.

Chow-fed rats (370–470 g, n = 16) were housed in an open-circuit indirect calorimetry system (Phenomaster, TSE Systems) and acclimated for 6 days. On experimental days, food was removed 2 h after dark onset. Two hours after removal of the food (i.e., 4 h after dark onset), rats received an injection of Ex4 (Bachem, 0.3 μg/kg ip, n = 9) or vehicle (Veh) solution (PBS, n = 7). EE and respiratory exchange ratio (RER) were measured in undisturbed animals until food was returned 4 h after injection (i.e., 8 h after dark onset). Four days later, an identical protocol was repeated in the same rats to measure interscapular skin temperature before injection (2 h after dark onset, “baseline”) or 2 h after injection (4 h after dark onset) of Veh (n = 8) or Ex4 (0.3 μg/kg; n = 8). Rats were lightly restrained in a stretched position under an infrared camera (E60, FLIR Systems) mounted vertically over the shaved interscapular area (distance: 30 cm). Three pictures were taken at each time point (baseline and 2 h after injection). Final analysis of the infrared pictures was done by defining a standard rectangular area (60 × 40 pixels) over the interscapular region and averaging the mean area temperatures of two pictures (FLIR Tool software for PC).

Acute effects of intraperitoneal Ex4 in chow-fed rats on locomotor activity.

Ten days after EE and RER measurements, the locomotor activity was monitored during an open field test 90 min after intraperitoneal injection of Ex4 (0.3 μg/kg, n = 9) or Veh (n = 7), which coincides with the period of lowest EE after Ex4. Food was removed and injections were administered with a similar protocol as for EE measurements, except that injections were staggered over 4 h (from 2 to 6 h into the dark phase) to allow for measurements of all rats within 2 days. The open field test (45 min) was conducted in two identical arenas (80 × 80 × 50 cm), filmed by a digital camera and analyzed with a computer running the EthoVision (Noldus, IT) tracking system, as previously described (22). Rearing time was measured by retrospective video analysis of the open field test.

Acute effects of intraperitoneal Ex4 in chow-fed rats on BAT gene expression.

Fourteen days after EE and RER measurements, a similar protocol as for locomotor activity measurements was conducted to collect iBAT 2 h after intraperitoneal injection of Ex4 (0.3 μg/kg; n = 7) or Veh (n = 8). Rats were anesthetized by an intraperitoneal injection of pentobarbital (100 mg/kg; Cantonal Pharmacy of Zurich), and iBAT was immediately collected, cleaned from surrounding tissue on ice, frozen in liquid nitrogen, and stored at −80°C until further processing. RNA was extracted with TRIzol (Life Technologies) using the manufacturer’s protocol for fat tissues. RT-quantitative (q)PCR analysis was performed using SybR Green on a OneStep Plus instrument (Applied Biosystems), and results were analyzed using the 2-ΔΔCT method. Table 1 lists all primers used. Figure 1A summarizes the procedures and timing of experiments performed in cohort 1.

Table 1.

List of gene abbreviations and primers used for qPCR

Gapdh	Glyceraldehyde 3-phosphate dehydrogenase	F: ACAACTTTGGCATCGTGGA
		R: CTTCTGAGTGGCAGTGATGG
Glp1r	Glucagon-like peptide-1 receptor	F: CACTTCCTTCCAGGGCTT
		R: CGAAACTCCATCTGGACCTC
Ucp1	Uncoupling protein-1	F: GCCTGCCTAGCAGACATCAT
		R: TGGCCTTCACCTTGGATCT
Adrb3	β-3 Adrenergic receptor	F: GGGAAGCTGACAGAATTACCTC
		R: GGGGACATTAGGGCTTGG
Cidea	Cell death-inducing DFFA-like effector A	F: AAAGGGACAGAAATGGACACC
		R: TCAGCCTGTATAGGTCGAAGG
Ppargc1a	Peroxisome proliferator-activated receptor γ coactivator 1-α	F: TGTGGAACTCTCTGGAACTGC
		R: GCCTTGAAAGGGTTATCTTGG
Pparg	Peroxisome proliferator-activated receptor γ	F: GGTGAAACTCTGGGAGATCCT
		R: AATGGCATCTCTGTGTCAACC
Dio2	Type 2 iodothyronine deiodinase	F: GTGATGCTGCCCTATGTG
		R: AACAATAGGTTTTCTGGGAACTTTT
Fgf21	Fibroblast growth factor 21	F: CACAGATGACGACCAGGACA
		R: GAATGACCCCTGGCTTCAA
Pnpla2	Patatin-like phospholipase domain containing 2 (adipose triglyceride lipase)	F: GAGAGAATGTCATCATATCGCACT
		R:GGGATAAAAGTGCTGCAAACAT
Lipe	Hormone-sensitive lipase	F: CGAGCACTGGAGGAGTGTTT
		R: GCTCTCCGGTTGAACCAA
Mgll	Monoglyceride lipase	F: CTTCCTCCTGGGCCACTC
		R: AAGTGGGTTGGTCTCTCTGC
Fasn	Fatty acid synthase	F: GGCCACCTCAGTCCTGTTAT
		R: AGGGTCCAGCTAGAGGGTACA
Npy	Neuropeptide Y	F: CCGCTCTGCGACACTACAT
		R: TGTCTCAGGGCTGGATCTCT
Agrp	Agouti-related peptide	F: AGGCCCTGTTCCCAGAGT
		R: CTGTGGATCTAGCACCTCTGC
Pomc	Proopiomelanocortin	F: AGGACCTCACCACGGAAAG
		R: CCGAGAGGTCGAGTCTGC
Trh	Thyrotropin releasing hormone	F: CAGAGTCTCCACTTCGCAGA
		R: CAGGGATACCAGTTAGGGTGAA
Crh	Corticotropin releasing hormone	F: CTCTCTGGATCTCACCTTCCAC
		R: CTAAATGCAGAATCGTTTTGGC
Oxt	Oxytocin	F: CTTGCTTGCTGCCTGCTT
		R: CCGCAGGGAAGACACTTG
Avp	Arginine vasopressin	F: AGGGGAGACACTGTCTCAGCTC
		R: TGCCTGCTACTTCCAGAACTGC
Lepr	Leptin receptor	F: TGTCAGAAATTCTATGTGGTTTTGT
		R: TTGGATAGGCCAGGTTAAGTG
Dbh	Dopamine β-hydroxylase	F: ACTACTGTCGCCACGTGCT
		R: ACCGGCTTCTTCTGGGTAGT
Gcg	Glucagon	F: TGAGATGAACACGATTCTCGAT
		R: AAGATGGTTGTGAATGGTGAAA

F, forward; qPCR, quantitative PCR; R, reverse.

Fig. 1.

Intraperitoneal exendin-4 (Ex4) injection reduces energy expenditure (EE) and brown adipose tissue (BAT) thermogenesis. A: experimental protocol for B–I. B: heat production after intraperitoneal vehicle (Veh) or Ex4 (0.3 μg/kg) injection (n = 7 and 9, respectively; two-way ANOVA; group effect, not significant (ns); time effect, P < 0.0001; interaction, P = 0.1). C: 4-h cumulative heat production after intraperitoneal Veh or Ex4 injection (n = 7 and 9, respectively; Student’s t-test; P < 0.05). D: respiratory exchange ratio after intraperitoneal Veh or Ex4 injection (n = 7 and 9, respectively; two-way ANOVA; group effect, ns; time effect, P < 0.0001; interaction, P < 0.01). E: total distance moved during a 45-min open field test starting 90 min after intraperitoneal Veh or Ex4 injection (n = 8 and 7, respectively; Student’s t-test; ns). F: total distance traveled in the open field arena, shown in 5-min time bins [n = 8 (Veh) and n = 7 (Ex4); two-way ANOVA; drug effect, ns; time effect, P < 0.0001; interaction effect, ns]. G: representative infrared pictures of the interscapular BAT (iBAT) 2 h after intraperitoneal Veh or Ex4 injection. H: skin temperature of the interscapular area before and 2 h after intraperitoneal Veh or Ex4 injection (n = 8; two-way ANOVA; group effect, ns; time effect, ns; interaction, P < 0 0.01; *different from Veh-injected rats; Bonferroni post hoc test; P < 0.05). I: relative mRNA expression in the iBAT [n = 8 (Veh) and n = 7 (Ex4); Holm-Sidak corrected t-tests; P < 0.01 for Adrb3; P < 0.1 for Ucp1, Cidea, Ppargc1a, and Pparg]. Table 1 provides a list of gene abbreviations and primers used. Data are means ± SE.

Cohort 2

Animal preparation.

Fourteen rats (260–315 g) were used to generate VAN Glp1r kd and control animals as previously described (25). Briefly, a midline incision was made with a scalpel, and salivary glands, lymph nodes, sternohyoid, and omohyoid muscles were retracted to expose the trachea and carotid artery. The vagus nerve was separated from the carotid artery by blunt dissection until the NG became accessible. A glass capillary attached to a micromanipulator (Picospritzer III injector; Parker Hannifin) was used to microinject 1.5 μl of lentiviral particles [10¹⁰ plaque-forming units (pfu)/ml] carrying a nontarget (control, n = 7) or a Glp1r targeting shRNA (n = 7) bilaterally into the NG. Efficiency of the kd was verified postmortem; rats received an intraperitoneal injection of pentobarbital (100 mg/kg, Zurich Cantonal Pharmacy), and NG were excised and immediately frozen in liquid nitrogen and stored at −80°C until further processing. RNA was extracted with TRIzol (Life Technologies) and RT-qPCR analysis was performed using SybR Green on a OneStep Plus instrument (Applied Biosystems). Glp1r primers used are listed in Table 1.

Acute effects of intraperitoneal Ex4 on EE, RER, and interscapular skin temperature in chow-fed VAN Glp1r kd rats.

Four weeks after the NG injections, the 14 rats were acclimated to an open circuit indirect calorimetry system (Phenomaster, TSE Systems). Experimental procedures to measure EE, RER, and interscapular skin temperature were identical to those used in cohort 1, with the exception that control (n = 7) and Glp1r kd rats (n = 7) received an intraperitoneal injection of Veh or Ex4 (0.3 μg/kg) in a randomized crossover design with a 1-day washout period.

Cohort 3

Animal preparation.

Twenty-seven rats (280–335 g) were used to generate VAN Glp1r kd (n = 14) and control (n = 13) animals as described for cohort 2. Twenty days after surgery, all rats were switched to an HFD (D12492, ssniff-Spezialdiäten, 60% energy from fat, 5.16 kcal/g).

Effects of VAN Glp1r kd on food intake and body composition during chronic HFD feeding.

Over 10 wk, daily food intake was measured in all animals (n = 13/14) by an in-house automated system piloted by LabX meal analyzer (Mettler Toledo), and body weights were measured every 2–3 days. Analysis of body composition was conducted 10 wk after introduction of the HFD using a computerized tomography system (La Theta LCT-100, Aloka), as validated previously (18).

Effects of VAN Glp1r kd on EE and RER during chronic HFD feeding.

Ten weeks after introduction of the HFD, a subset of the cohort (controls, n = 7; Glp1r kd, n = 8) was housed in an open circuit indirect calorimetry system (Phenomaster, TSE Systems) and acclimated for 3 days. EE, RER, and locomotor activity were measured in undisturbed rats for 3 consecutive days, and data were averaged over these 3 days.

Effects of VAN Glp1r kd on intraperitoneal and BAT temperature during chronic HFD feeding.

Ten weeks after introduction of the HFD, the rest of the animals (control, n = 6; Glp1r kd, n = 6) were implanted with dual telemetry sensors (F40-TT, DSI) to allow for continuous measurements of iBAT and intraperitoneal temperatures. After inhalational anesthesia with isoflurane, a scalpel incision was made unilaterally on the flank of the rats, and the body of the sensor was sutured subcutaneously. The peritoneal cavity was punctured to introduce one temperature probe, and the second probe was led subcutaneously to the interscapular area. The tip of the temperature probe was then placed between and ventral to the iBAT fat pads and caudal to the Sulzer vein, as previously described (38). Both probes were secured by nonabsorbable sutures. Carprofen (Rimadyl; E. Gräb) was injected subcutaneously for analgesia immediately after completion of the surgery and on each of the following 3 days. Intraperitoneal and iBAT temperatures were measured in week 12 after introduction of an HFD for 3 consecutive days in undisturbed animals using Dataquest ART (DSI, version 3.1). Data presented here were averaged over these 3 days.

Effects of VAN Glp1r kd on glucose homeostasis during chronic HFD feeding.

An oral glucose tolerance test was performed 12 wk after introduction of the HFD in all rats of cohort 3 (control, n = 13; Glp1r, kd n = 14). After adaptation to gavage and restraining procedures, rats were deprived of food for 14 h and received an oral bolus of 40% glucose solution (2 g/kg). Tail vein blood glucose was measured using Accucheck (Roche). Insulin was measured from plasma using an immunoassay (single-spot for mouse/rat K152BZC, Meso Scale Diagnostics). In addition, an intraperitoneal insulin sensitivity test was performed 13 wk after the introduction of the HFD in a subset of cohort 3 (controls, n = 6; Glp1r, kd n = 7). After adaptation to restraining and injections, rats were deprived of food for 14 h and received an intraperitoneal injection of 1 IU/kg of insulin (Actrapid HM, Novo Nordisk). Tail vein blood glucose was measured using Accucheck (Roche).

Effects of VAN Glp1r kd on brain, iBAT, and subcutaneous white adipose tissue gene expression and morphology during chronic HFD feeding.

Rats of cohort 3 (controls, n = 8; Glp1r kd, n = 8) received an intraperitoneal injection of pentobarbital (100 mg/kg, Zurich Cantonal Pharmacy). The brain, a sample of subcutaneous white adipose tissue (ScWAT), and iBAT were frozen in liquid nitrogen and stored at −80°C until further processing. The NTS, arcuate nucleus (ARH), paraventricular nucleus of the hypothalamus (PVH), and dorsomedial nucleus of the hypothalamus (DMH) were collected after brain cryosection and micropunching. RNA was extracted with TRIzol (Life Technologies), and RT-qPCR analysis was performed using SybR Green on a OneStep Plus instrument (Applied Biosystems). Table 1 lists the primers used.

For stainings with hematoxylin-eosin, a sample of ScWAT and iBAT were fixed in 4% paraformaldehyde (Sigma-Aldrich) and processed with STP 120 (Microm). Paraffin-embedded samples were cut at 5 μm on a Hyrax M55 (Zeiss), deparaffinized, and stained on a Varistain 24-4 (Shandon). Pictures were obtained with an Axioscope A.1 microscope (Zeiss). The iBAT lipid fraction was measured from hematoxylin-eosin images with Image J (National Institutes of Health, Bethesda, MD) by segmenting out areas of the tissue that remained unstained. The cell-size distribution of white adipocytes was measured using Image J as previously described (42). Western blots were performed to detect the uncoupling protein-1 (UCP1) (rabbit antibody PA1-24894 1:2,000, Thermo Fisher Scientific) using β-actin as reference (1:5,000, mouse antibody AC-74, Sigma-Aldrich).

Cohort 4

Anterograde tracing of VANs using HSV-1 H129-772.

Sixteen adult male Sprague-Dawley rats (280–300 g) were divided into four groups of four animals (24 h, 48 h, 72 h, or 96 h survival). The left NG of each rat was injected with 800–1,200 nl of herpes simplex virus (HSV)-1 H129-772 (original titer: 8.11 × 10⁸ pfu/ml) (51), either undiluted or diluted 1:1 (vol/vol) with PBS buffer. The injectate contained 5% E133 food dye to ensure accurate placement in the NG. After the designated survival times, rats were anesthetized and perfused with pH 9.5 borate-buffered 4% paraformaldehyde. Four series of serial frozen coronal sections (30 µm) were cut through the entire brain and immunohistochemically processed using a chicken primary antibody for green fluorescent protein (1:10K, Abcam) visualized with a secondary conjugated to Alexa 488 (1:2K, The Jackson Laboratory), and a mouse monoclonal dopamine β-hydroxylase (DBH) antibody (1:10K, Cell Signaling Technology) visualized with a secondary conjugated to cyanine 3 (1:2K, The Jackson Laboratory). DBH immunostaining was used to help delineate regions of interest in the hypothalamus and hindbrain. Sections were examined using a Zeiss Axioimager epifluorescence microscope. Images were captured using Volocity (v6.1, PerkinElmer).

Cohort 5

Retrograde labeling from BAT and laser capture microdissection of labeled VANs.

Anesthetized rats (520–570 g; n = 3) were injected with 1 μl of solution containing a PRV-263 pseudorabies virus (PRV) (1.4 × 10⁹ pfu/ml) bilaterally into the iBAT. Six days after PRV-263 injection, the two NGs from PRV-injected rats were removed and immediately cut at ten micrometers in a cryostat (CM 1950, Leica). After dehydration in a series of graded ethanol baths and xylene, PRV-labeled VANs from one NG were collected using an automated laser microdissection instrument (Arcturus XT). An identical number of nonlabeled VANs from the same NG (“TdTomato−”), as well as all VANs from the second NG (“all NG”), were used as comparisons. RNA was extracted using an Arcturus PicoPure kit (Applied Biosystems) and amplified (Ovation One-Direct system, NuGEN Technologies, San Carlos, CA) before qPCR (OneStep Plus, Applied Biosystems).

Cohort 6

Effect of intraperitoneal Ex4 on BAT SNA, BAT temperature, and expired CO₂.

Male rats (300–450 g, n = 6) were anesthetized with isoflurane (2%–3% in 100% O₂), and the femoral artery and vein were cannulated. Rats were transitioned from isoflurane to urethane (750 mg/kg iv) and α-chloralose (60 mg/kg iv) anesthesia. The rats were ventilated with 100% O₂ via a tracheal cannula and paralyzed with d-tubocurarine (0.6 mg/rat iv). A thermocouple inserted in the rectum was used to measure core body temperature, which was maintained at 37.0 ± 0.5°C with a heat lamp and a water-perfused thermal blanket until skin cooling began. BAT temperature and skin temperature were recorded using thermocouples placed in the iBAT and on the hindquarter skin, respectively. A sympathetic nerve innervating the right iBAT pad was recorded using bipolar hook electrodes and processed as described previously (31). The skin was cooled until a stable level of BAT SNA was established, at which time an intraperitoneal injection of Ex4 (30 µg/kg) was made, and variables were recorded for 15 min.

Statistical Analysis

Data are presented as means ± SE. All graphs were generated using GraphPad Prism (versions 6.05 and 7.02), and statistical analyses were performed with GraphPad Prism and IBM SPSS Statistics (version 22). Data normality was verified using the Shapiro-Wilk (when n ≥ 7) and the Kolmogorov-Smirnov (when n ≤ 6) tests, and homoscedasticity was checked by visualizing the distribution of residuals. Nonparametric tests were used otherwise. Tests used and statistical results are given in each figure legend. P values < 0.05 were considered significant.

RESULTS

Intraperitoneal Ex4 acutely decreases EE and BAT thermogenesis.

First, we used Ex4 to test the effects of peripheral GLP-1R activation on whole-body EE (heat production) and BAT thermogenesis in awake and free-moving chow-fed rats. We chose a low dose (0.3 μg/kg ip) previously shown to acutely (<2 h) inhibit eating via a mechanism that depends on VAN GLP-1R (25). Moreover, we used a restricted feeding schedule to control for potential confounders, such as food intake and gastric emptying (Fig. 1A), thereby isolating the direct acute effects of peripheral GLP-1R activation on heat production. We measured heat production after intraperitoneal Ex4 (Fig. 1B) and found a significant reduction of 4-h cumulative heat production (Fig. 1C). The RER was reduced after intraperitoneal Ex4 compared with Veh (Fig. 1D). Because Ex4 can induce visceral malaise (21), it may affect heat production indirectly by reducing locomotor activity. This, however, is unlikely because intraperitoneal Ex4 did not affect the distance traveled (Fig. 1, E and F) or other energy-consuming behaviors, such as rearing (data not shown) in an open field test conducted during the period when heat production and RER were most affected (1 h 30 min to 2 h 15 min postinjection). Using the same feeding paradigm, we found that interscapular skin temperature, an indicator of iBAT thermogenesis, was decreased 2 h after intraperitoneal Ex4 (0.3 μg/kg) compared with Veh (Fig. 1, G and H). β₃-Adrenergic receptor (Adrb3) mRNA expression was downregulated in the iBAT after intraperitoneal Ex4, but the expression of other thermogenic mRNAs was unaffected (Fig. 1I). Together, these data indicate that peripheral GLP-1R activation reduces whole body EE and BAT thermogenesis independent of changes in eating and locomotor activity.

VAN GLP-1R mediates modulation of EE and BAT thermogenesis induced by intraperitoneal Ex4.

To identify the mechanism of peripheral Ex4 action controlling EE and BAT thermogenesis, we used our established rat model with a specific kd of Glp1r (Glp1r kd) in VANs (25). Rats injected with lentiviral particles expressing a Glp1r-targeting shRNA construct showed a significant reduction of Glp1r mRNA in the NG (Fig. 2A) compared with rats injected with a control lentivirus. This reduction (41.3%) was similar to the reduction observed in our previous study [52.5% (25)]. We have previously shown that this reduced Glp1r expression under these conditions is limited to the NG, specific for Glp1r, i.e., the expression of other genes is not affected and the functional integrity of vagal afferents is preserved (25). This model also showed a reduced sensitivity to the satiating and gastric emptying effects of intraperitoneally administered GLP-1 or Ex4 (25). Using the same feeding paradigm described above (Fig. 1A), we measured heat production following intraperitoneal Ex4 (0.3 μg/kg) in control and Glp1r kd rats (Fig. 2B). Intraperitoneal Ex4 reduced the cumulative 4-h heat production in control but not in Glp1r kd rats (Fig. 2C). Similarly, interscapular skin temperature was decreased 2 h after intraperitoneal Ex4 injection compared with Veh injection in control but not in Glp1r kd rats (Fig. 2, E and F). In contrast, intraperitoneal Ex4 reduced RER similarly in both groups (Fig. 2D). Together, these results demonstrate that a full expression of VAN Glp1r is required for the full reduction of EE and BAT thermogenesis by intraperitoneal Ex4.

Fig. 2.

Glucagon-like peptide-1 receptor (Glp1r) expression in vagal afferent neurons (VANs) is necessary for the exendin-4 (Ex4)-induced reduction in energy expenditure (EE) and brown adipose tissue (BAT) thermogenesis. A: relative Glp1r mRNA expression in the nodose ganglia (NG) of control and Glp1r knockdown (kd) rats (n = 7 and 7, respectively; Student’s t-test; P < 0.001). B: heat production following intraperitoneal vehicle (Veh) or Ex4 injection (n = 7; ANOVA; group effect, P = 0.1; time effect, P < 0.0001; drug, P < 0.05; interaction time × drug, P < 0.05; other interactions, ns). C: cumulative heat production during the 4 h following intraperitoneal Veh or Ex4 injection (n = 7; two-way ANOVA; group effect, P = 0.05; drug effect, P < 0.001; interaction, ns; *different from Veh-injected rats, Bonferroni post hoc test; P < 0.05). D: respiratory exchange ratio following intraperitoneal Veh or Ex4 injection (n = 7; ANOVA; group effect, ns; drug, P = 0.1; time effect, P < 0.0001; interaction time × group, P < 0.1; time × drug, P < 0.01; other interactions, ns). E: representative infrared pictures of the interscapular BAT (iBAT) 2 h after intraperitoneal Veh or Ex4 injection. F: average skin temperature of the interscapular area before and 2 h after intraperitoneal Veh or Ex4 injection (n = 7; ANOVA; group effect, ns; drug effect, ns; time effect, P < 0.01; interaction group × drug × time, P < 0.05; other interactions, ns; *different from control rats, Bonferroni post hoc test; P < 0.05). Table 1 provides a list of gene abbreviations and primers used. Data are means ± SE.

VAN Glp1r kd elevates EE and BAT temperature during chronic HFD feeding.

We previously reported that Glp1r kd in VAN does not alter body weight, daily food intake, and EE in rats fed standard chow (25). Figure 2, however, suggests that Ex4 acutely reduces BAT thermogenesis and EE via VAN GLP-1R. Therefore, we tested whether this also occurs in rats chronically fed an HFD, as meal-induced GLP-1 secretion was found to be elevated in the hepatic portal vein and lymph of HFD-fed rats (36, 50). When exposed to a 60% HFD, Glp1r kd rats displayed a small but significant reduction in body-weight gain compared with controls after 9 wk (Fig. 3A), a trend toward a reduced fat mass (Fig. 3B) but no change in daily food intake (Fig. 3C). Consistent with the effects of acute pharmacological VAN GLP-1R activation on BAT thermogenesis displayed in Fig. 2, Glp1r kd rats showed no overall increase in EE (Fig. 3D), but an increased heat production during the early dark phase compared with their controls (Fig. 3E), with no changes in RER (Fig. 3F) and locomotor activity (Fig. 3G). A previous study showed that peripheral GLP-1R activation with intraperitoneal Ex4 results in a long reduction in core temperature (17). Continuous measurements of intraperitoneal and BAT temperatures indicated that dark phase iBAT, but not intraperitoneal, temperature was elevated in Glp1r kd rats compared with controls after 12 wk on the HFD (Fig. 3, H and I). Glp1r kd rats showed a reduced glucose excursion (Fig. 4, A and B) and lower plasma insulin levels than controls during an oral glucose tolerance test (Fig. 4C). Also, an intraperitoneal insulin sensitivity test revealed a more robust reduction in blood glucose 15 min after insulin injection in Glp1r kd rats than in controls (Fig. 4, D and E).

Fig. 3.

Vagal afferent neurons (VANs) Glp1r knockdown (kd) increases energy expenditure (EE) and brown adipose tissue (BAT) temperature and alleviates high-fat diet (HFD)-induced body weight gain. A: body weight gain of control and Glp1r kd rats after introduction of the HFD (n = 13 and 14, respectively; two-way ANOVA; group effect, P < 0.05; time effect, P < 0.0001; interaction effect, P < 0.01; *different from control rats, Bonferroni post hoc comparisons; P < 0.05). B: lean and fat mass of control and Glp1r kd rats fed an HFD (n = 13; Holm-Sidak corrected t-tests; not significant, ns). C: daily HFD intake of control and Glp1r kd rats (n = 13 and 14, respectively; two-way ANOVA; group effect, ns; time effect, P < 0.0001; interaction effect, ns). D: heat production over 24 h in control and Glp1r kd rats (n = 7 and 8, respectively; two-way ANOVA; group effect, P = 0.1; time effect, P < 0.0001; interaction effect, ns). E: cumulative heat production (day split into quarters/6-h time bins) in control and Glp1r kd rats (n = 7 and 8, respectively; two-way ANOVA; group effect, P = 0.1; time effect, P < 0.0001; interaction effect, P < 0.05; *different from control rats Bonferroni post hoc comparisons; P < 0.05). F: respiratory exchange ratio over 24 h in control and Glp1r kd rats fed an HFD (n = 7 and 8, respectively; two-way ANOVA; group effect, ns; time effect, P < 0.0001; interaction effect, ns). G: 24-h locomotor activity (day split into quarters/6-h time bins) in control and Glp1r kd rats fed an HFD (n = 7 and 8, respectively; two-way ANOVA; group effect, ns; time effect, P < 0.0001; interaction effect, ns). H: real-time 24-h monitoring of iBAT temperature (n = 6; two-way ANOVA; group effect, P = 0.06; time, P < 0.0001; interaction, P < 0.001). I: real-time 24-h monitoring of intraperitoneal temperature (n = 6; two-way ANOVA; group effect, ns; time effect, ns; interaction effect, ns). Data are means ± SE.

Fig. 4.

Vagal afferent neurons (VANs) Glp1r knockdown (kd) improves glucose tolerance during chronic high-fat diet (HFD) exposure. A: blood glucose during an oral glucose tolerance test (OGTT, 2 g/kg) in control and Glp1r kd rats (n = 13 and 14, respectively; two-way ANOVA; group effect, ns; time effect, P < 0.0001; interaction effect, P < 0.05; *different from control rats, Bonferroni post hoc comparisons; P < 0.05). B: area under curve of blood glucose during OGTT in control and Glp1r kd rats (n = 13 and 14, respectively; Student’s t-test; P < 0.01). C: plasma insulin during OGTT in control and Glp1r kd rats (n = 12 and 14, respectively; two-way ANOVA; group effect, P < 0.05; time effect, P < 0.0001; interaction effect, ns; *significant main group effect). D: blood glucose during an intraperitoneal insulin sensitivity test (IPIST) in control and Glp1r kd rats (n = 6 and 7, respectively; two-way ANOVA; group effect, ns; time effect, P < 0.0001; interaction effect, P < 0.05; *different from control rats, Bonferroni post hoc comparisons; P < 0.05). E: blood glucose 15 min after insulin injection [n = 6 (control) and n = 7 (Glp1r kd); Student’s t-test; P < 0.05]. Data are means ± SE.

VAN Glp1r kd increase thermogenic capacity of both BAT and ScWAT during chronic HFD feeding.

Molecular analysis of the iBAT after 12 wk on an HFD revealed increased Ucp1 mRNA and protein expression in Glp1r kd rats compared with controls (Fig. 5, A and B). Consistent with these findings, Glp1r kd rats exhibited a lower lipid deposition (Fig. 5C). We observed an increase in the mRNA expression of Ucp1 and Adrb3 in the ScWAT of VAN Glp1r kd rats compared with controls (Fig. 5D). A robust increase in UCP1 protein expression was also detected in the ScWAT of Glp1r kd rats (Fig. 5E). Consistent with the changes in gene and protein expression, the Glp1r kd group showed a shift toward smaller subcutaneous adipocytes compared with controls (Fig. 5F). Together, these results are consistent with an increase in the thermogenic capacity of the iBAT and the browning of the ScWAT as a result of VAN Glp1r kd. We further investigated whether VAN Glp1r kd altered the expression of central nervous system (CNS) genes involved in energy balance during HFD exposure. We found no difference in the expression of eating-related neuropeptides or genes involved in the neuroendocrine control of energy metabolism in the ARH, the PVH, and the DMH (Fig. 6, A–C). In the NTS, VAN Glp1r kd resulted in an increase in the preproglucagon gene (Gcg, the precursor gene for GLP-1) expression compared with control rats (Fig. 6D), suggesting that an enhanced action of central endogenous GLP-1 in response to the VAN Glp1r kd could link VAN Glp1r kd and BAT enhanced thermogenesis.

Fig. 5.

Vagal afferent neurons (VANs) Glp1r knockdown (kd) increases brown adipose tissue (BAT) thermogenic capacity and subcutaneous white adipose tissue (ScWAT) browning during chronic high-fat diet (HFD) exposure. A: relative mRNA expression of thermogenic markers in the interscapular BAT (iBAT) (n = 8; Holm-Sidak corrected t-tests; P < 0.05 for Ucp1; not significant, ns for other genes). B: relative protein expression of uncoupling protein-1 (UCP1) in the iBAT, with representative bands [n = 6 (control) and n = 7 (Glp1r kd); Student’s t-test; P < 0.05]. C: representative pictures of hematoxylin-eosin (H&E) staining (scale bar: 100 μm) and lipid area fraction in iBAT of control and Glp1r kd rats fed an HFD (n = 8; Student’s t-test; P < 0.05). D: relative mRNA expression in the ScWAT [n = 6 (control) and n = 7 (Glp1r kd); Holm-Sidak corrected t-tests; P < 0.01 for Adrb3 and Ucp1; ns for other genes]. E: relative protein expression of UCP1 in the ScWAT, with representative bands [n = 6 (control) and n = 7 (Glp1r kd); Student’s t-test]. F: representative pictures of H&E staining (scale bar: 100 μm) and distribution of measured subcutaneous adipocyte area in control and Glp1r kd rats fed an HFD (n = 6 and 7, respectively; Student’s t-test on average cell area after logarithmic transformation; P < 0.05). Table 1 provides a list of gene abbreviations and primers. ARH, arcuate nucleus; PVH, paraventricular nucleus of the hypothalamus. Data are means ± SE.

Fig. 6.

Vagal afferent neurons (VANs) Glp1r knockdown (kd) alters Gcg mRNA expression in the nucleus tractus solitarii (NTS) during chronic high-fat diet (HFD) exposure. Relative mRNA expression in (A) ARH [n = 7 (control) and n = 8 (Glp1r kd)], (B) PVH (n = 8 for control and Glp1r kd), (C) dorsomedial nucleus of the hypothalamus (DMH, n = 8 for control and Glp1r kd), and (D) NTS (n = 7 for control and Glp1r kd) of HFD-fed control and Glp1r kd rats (Holm-Sidak corrected t-tests; *different from control rats; P < 0.05). Table 1 presents a list of gene abbreviations and primers. ARH, arcuate nucleus; PVH, paraventricular nucleus of the hypothalamus. Data are means ± SE.

VANs project to a central neural network controlling BAT thermogenesis.

Glp1r mRNA was not detectable in the iBAT (data not shown), thereby excluding any possible direct action of Ex4 on brown adipocytes. To test whether neuronal pathways can link VAN GLP-1R activation to the modulation of BAT activity, we assessed whether VANs are synaptically connected to central BAT sympathetic control networks. We injected the anterogradely transported neurotropic virus, HSV-1 H129-772 [H129 (51)], into the left NG to label downstream neurons receiving VAN information. We then examined the extent of H129 infections in those regions previously identified as providing sympathetic control to BAT by injecting PRV into BAT (10, 37). After 24 h, H129 infections were evident in the medial part of the NTS (data not shown) but nowhere else in the brain, indicating the specificity of VAN projections. H129 infections increased dramatically by 48 h in the NTS and area postrema (Fig. 7A,a). There were also H129-labeled neurons in the parapyramidal groups and rostral ventrolateral medulla (RVLM) by 72 h but relatively fewer in the rostral raphe pallidus area (rRPa) in these same animals (Fig. 7A, b and c). Although VANs can activate catecholamine neurons in the NTS (2), we did not see H129-labeled neurons colocalized with DBH-positive neurons in the NTS at the level of the area postrema (Fig. 7A,a). Some double-labeled DBH-H129 neurons were evident in the RVLM (Fig. 7A,c). The DMH and PVH, particularly its ventral parvicellular part, showed the earliest forebrain infections (Fig. 7A, d and e). These were well developed by 72 and 96 h, at which time neurons were also seen in the median preoptic nucleus, medial preoptic area, the parastrial nucleus, and parts of the bed nuclei of the stria terminalis (Fig. 7A,f). In contrast, only occasional H129-labeled neurons were seen in the ARH after 72 h (Fig. 7A,d). These results are in accordance with the idea that VANs can modulate the central sympathetic activation to BAT.

Fig. 7.

Glp1r expressing vagal afferent neurons (VANs) are synaptically connected to the central thermoregulatory network and interscapular brown adipose tissue (iBAT). Representative pictures of hindbrain (A, a–c) and hypothalamus (d–f) brain sections showing infected neurons 2–4 days after injection of herpes simplex virus (HSV) 129 into the nodose ganglia (NG). Green represents HSV 129 and red represents dopamine-β-hydroxylase. Rectangle in the top represents the rostral raphe pallidus area (rRPa), shown at the bottom (b). Dotted line in bottom (b) shows the rRPa (scale bar: 200 μm). Representative pictures of the NG (B) 6 days after injection of pseudorabies virus (PRV)-263 into the iBAT (scale bars: 25 μm). Representative pictures of laser capture microdissection (C) of VANs labeled by PRV-263 (scale bar: 25 μm). Relative Glp1r mRNA expression (D) in laser-captured cell bodies of the NG labeled by PRV-263 (tdTomato+), unlabeled (tdTomato−), or in all cell bodies of the NG (all NG) (n = 3; Friedman test; *different from all NG, Dunn’s post hoc test; P < 0.05). Representative example (E) demonstrating that intraperitoneal Ex4 (30 μg/kg) injection (dashed line) inhibited cold-evoked BAT sympathetic nerve activity (SNA) and decreased BAT temperature (T_BAT) and expired CO₂. ac, anterior commissure; AHN, anterior hypothalamic nucleus; AP, area postrema; ARH, arcuate nucleus; BST, bed nuclei of the stria terminalis; dm, dorsomedial nucleus; DMH, dorsomedial nucleus of the hypothalamus; DMX, dorsal motor nucleus of the vagus; fu, fusiform nucleus of the BST; fx, fornix; Glp1r, glucagon-like peptide-1 receptor; MnPO, median preoptic nucleus; MPA, medial preoptic area; oc, optic chiasm; ov, oval nucleus of the BST; PPY, parapyramidal nucleus; PS, parastrial nucleus; PVH, paraventricular nucleus of the hypothalamus; PVH_lp, lateral parvicellular part of the PVH; PVH_mpv, medial parvicellular part ventral zone of the PVH; Py, pyramidaltract; RVLM, rostral ventrolateral medulla; VMH, ventromedial nucleus of the hypothalamus.

Glp1r-expressing VANs are synaptically connected to BAT.

To further test the plausibility of the gut-brain-BAT pathway downstream of VAN GLP-1R activation, we determined whether iBAT- and VAN Glp1r-expressing neurons are synaptically linked by injecting PRV-expressing tdTomato (PRV-263) bilaterally into iBAT. Six days later, we found tdTomato-labeled neurons in all CNS areas previously identified as constituents of a BAT thermogenesis control network (data not shown) (35). Importantly, a subset of neurons in the NG also contained tdTomato labeling (Fig. 7B), indicating that some VANs are synaptically connected to the sympathetic efferent innervation of the iBAT. To examine whether PRV-263-infected VANs can mediate the GLP-1-induced modulation of BAT thermogenesis, the cell bodies of tdTomato-labeled VANs were isolated using laser capture microdissection (Fig. 7C). RT-qPCR revealed a highly enriched expression of the Glp1r mRNA in tdTomato-labeled VANs compared with nonlabeled VANs (Fig. 7D). These results indicate that Glp1r-expressing VANs are structurally positioned to modulate sympathetic inputs to BAT.

Peripheral Ex4 administration inhibits cold-evoked BAT SNA.

Our data suggest that VAN GLP-1R signaling inhibits BAT thermogenesis via the gut-brain-BAT connection. Finally, to test whether peripheral GLP-1R activation decreases BAT thermogenesis by modulating sympathetic neural pathways, we used cold exposure as a model to activate the central thermoregulatory network regulating BAT SNA (35). Ex4 (30 µg/kg) was administered intraperitoneally during cold exposure (skin temperature: 34.1 ± 0.7°C, core body temperature: 35.6 ± 0.5°C) under anesthesia. A previous dose-response study (data not shown) revealed that a higher Ex4 dose was necessary to see an effect in anesthetized rats compared with awake, free-moving animals (Figs. 1 and 2). During cold exposure, BAT SNA was characterized by large bursts of activity (Fig. 7E), reflecting the summed action potentials of postganglionic axons in the recorded nerve bundle. The cold-evoked increase in BAT SNA was completely reversed within 10 min of administration of Ex4 (Fig. 7E, from a cold-evoked level of 889 ± 157% baseline to 110 ± 7% baseline, representing a 99 ± 1% inhibition, P < 0.01). Ex4 administration also decreased BAT temperature (from 35.0 ± 0.6°C to 33.6 ± 0.5°C, P < 0.05) and expired CO₂ (from 4.7 ± 0.2% to 4.2 ± 0.2%, P < 0.05) (Fig. 7E). Mean arterial pressure and heart rate were not significantly altered by Ex4 (data not shown).

DISCUSSION

Endogenous intestinal GLP-1 contributes to the physiological control of meal size and postprandial glycemia, but its role in the control of EE and BAT thermogenesis is unexplored. In contrast to brain GLP-1R activation, which stimulates BAT thermogenesis (7, 29), we show that peripheral GLP-1R activation reduces EE and BAT thermogenesis. Using viral mediated Glp1r kd in VANs, we then demonstrate that VAN GLP-1R activation is necessary for intraperitoneal Ex4 to acutely inhibit EE and BAT thermogenesis and for the reduced BAT thermogenesis and EE evident during chronic HFD feeding. The VAN dependence of this acute low dose Ex4 effect is consistent with the VAN dependence of Ex4’s acute eating-inhibitory effect (25) and contrasts with longer-term and higher-dose Ex4 effects that are primarily mediated by central GLP-1R. Furthermore, using viral transsynaptic tracers (4, 37, 43, 47), we demonstrate for the first time, to our knowledge, that VANs can provide central thermoregulatory sites with viscerosensory information relevant for controlling BAT thermogenesis. Importantly, we also found that some Glp1r-expressing VANs are synaptically connected to the brain networks that control iBAT, thereby providing the critical neural link for these neurons to mediate the effects of endogenous peripheral GLP-1 on BAT thermogenesis.

Our findings provide compelling evidence that nutrient-induced signals modulate BAT thermogenic activity by activating VANs. This idea is consistent with several other findings. First, lesions of VANs blunt the modulation of BAT activity induced by duodenally infused lipid (9) and hepatic glucokinase overexpression (49). In addition, disrupting VAN lipid sensing by knocking out peroxisome proliferator-activated receptor γ (Pparg) and liver X receptor α and β (Nr1h2 and Nr1h3) increases the expression of BAT thermogenic markers and EE during HFD exposure in mice (28, 34). These studies indicate that nutrient-induced signals are integrated by VANs and can modulate BAT thermogenesis via neural mechanisms. Notably, vagally mediated signals appear to be physiologically relevant for BAT thermogenesis only when high caloric food is consumed. Indeed, a recent study confirmed that HFD feeding alters VAN signaling to the NTS and impairs cooling-evoked BAT activation in rats (32). Altogether, this suggests that BAT thermogenesis is modulated by nutritional status and that this mechanism may contribute to HFD-induced obesity, which is in accordance with our results.

Despite the inability of chronic VAN Glp1r kd to alter EE during chow feeding (25), we demonstrate that VAN Glp1r kd increases EE and BAT thermogenesis with HFD feeding. In parallel, HFD-induced weight gain was reduced and glucose homeostasis was markedly improved in Glp1r kd rats compared with controls. These unexpected outcomes of reducing VAN GLP-1R signaling during HFD feeding might be considered counter-intuitive to current models of GLP-1’s actions and the known anorectic and weight loss effects of GLP-1R agonists. Indeed, a pharmacological blockade of central GLP-1R signaling increased food intake and fat accumulation in HFD-fed rats (5). Also, a high dose and/or chronic peripheral administration of GLP-1R agonists decreases food intake, but this effect is mediated by central rather than peripheral GLP-1R (45). Recently, chronic exenatide administration was shown to reduce energy intake but not 24-h EE in a randomized controlled trial (6). Yet, two sets of studies indicate that endogenous GLP-1 contributes to HFD-induced obesity and insulin resistance during HFD exposure. First, Glp1r knockout mice were protected from HFD-induced obesity and insulin resistance (3, 15), an effect that the authors attributed to an unrecognized role of endogenous GLP-1 in the control of EE and adipose tissue functions. Importantly and consistent with the effects of Glp1r deletion, mice lacking proglucagon-derived peptides are also resistant to HFD-induced obesity, mainly because of increased EE and BAT activity (48). A second study used one month of intraperitoneal infusions of the GLP-1R antagonist exendin 9–39 in wild-type mice exposed to an HFD (23). Exendin 9–39 improved fasting hyperglycemia and glucose intolerance in these animals. However, body weights were unchanged compared with controls, whereas energy intake was increased, suggesting elevated EE. Therefore, our study adds to a coherent set of studies indicating that endogenous GLP-1 secretion and signaling during HFD feeding have an unfavorable effect on BAT thermogenesis and EE, which could contribute to weight gain and reduced insulin sensitivity.

While several studies have indicated that nutrient-induced signals can modulate BAT thermogenesis via VANs, the neuronal effector pathways were unknown. Using the transsynaptic retrograde viral tracer PRV-263, we now show that Glp1r-expressing VANs are synaptically linked to iBAT. VANs containing GLP-1R are therefore structurally well positioned to control BAT thermogenesis, but whether they project to the intestine or other organs remains unclear from these data. Similarly, the central circuits relaying VAN signals to BAT are still unclear. Using H129 as a transsynaptic anterograde viral tracer to label the downstream targets of VANs, we show a robust H129 infection in several brain areas that contribute to the neural control of sympathetic outflow to BAT. First, the NTS showed robust labeling by H129; VAN synapse on second-order viscerosensory neurons in the NTS and primarily use glutamate as their neurotransmitter (1). Importantly, activation of neurons in the NTS is sympathoinhibitory to BAT (11), and blockade of glutamate receptors in the NTS reverses VAN-mediated inhibitions of BAT SNA in rats (32, 33). Other CNS regions labeled by H129 following NG injection include major hindbrain- and hypothalamic nuclei-controlling SNA in response to metabolic signals (11, 30, 31, 46). Notable among these areas are the PVH and RVLM, both of which contain neurons whose activation inhibit BAT SNA and decrease BAT thermogenesis (35), supporting the role of VANs in mediating a GLP-1-evoked decrease in BAT SNA and BAT thermogenesis. We also found H129-infected neurons in the rRPa, indicating that VAN information could influence BAT thermogenesis by modulating sympathetic premotor neurons via the NTS projection. There is, however, a lack of evidence for the direct regulation of rRPa via monosynaptic projections from the NTS. Therefore, VANs must engage integrative mechanisms upstream of the rRPa, most likely in the RVLM in the hindbrain and the PVH and DMH in the hypothalamus. Interestingly, during HFD exposure, NTS preproglucagon expression (Gcg) was increased in Glp1r kd rats compared with controls. Because reduced central GLP-1 levels have been linked to HFD-induced hyperphagia and fat accumulation (5), and because central GLP-1R activation modulates BAT thermogenesis (7), central GLP-1 signaling may contribute to the metabolic benefits in the VAN Glp1r kd model. Further systematic investigation of the central circuits controlling BAT thermogenesis linked to VANs will provide a more comprehensive understanding of this gut-to-BAT neuronal pathway.

Perspectives and Significance

From a physiological point of view, the existence of a gut-to-BAT neuronal pathway tuning down BAT thermogenesis implies that, in the postprandial state, heat dissipation of nutrient-derived energy in the BAT is being limited. This mechanism probably favors the uptake (and subsequent oxidation or storage) of incoming nutrients by peripheral organs over their dissipation as heat in the BAT. This idea is supported by recent findings indicating that diet-induced thermogenesis is limited in the BAT (see Refs. 24 and 27 for review). In summary, the gut-to-BAT neuronal pathway we identify here would have the ability to preferentially allocate incoming nutrients to energy-conserving mechanisms rather than heat dissipation. This may provide an evolutionary advantage. During high-caloric diet exposure, however, this mechanism would favor a positive energy balance, thereby contributing to the development of obesity.

In conclusion, our study reveals a novel role for endogenous peripheral GLP-1 in BAT thermogenesis beyond its well-described satiating and incretin effects. It shows for the first time, to our knowledge, that GLP-1 is involved in a crosstalk between the gut and BAT that is mediated through VANs and the brain. Importantly, this newly identified role for intestinal GLP-1 has significant implications for the development of the metabolic syndrome during HFD feeding. Because metabolically active BAT is reduced in obese humans (39, 44), manipulating the sympathoinhibitory gut-to-BAT pathway we describe here may be a successful strategy to improve BAT activity and glucose metabolism. A deeper understanding of the role of gut peptides in the vagal control of BAT thermogenesis should help to develop better treatment options for obesity and diabetes.

GRANTS

This work was supported by the Swiss National Science Foundation, Marie Heim-Vögtlin Grant PMPDP3_151360 (to S. J. Lee), Eidgenössische Technische Hochschule Zürich research Grant 47 12-2 (to W. Langhans and S. J. Lee), NIH Grant R01-NS-091066 (to S. F. Morrison), and NIH Grant NS-029728 (to A.G. Watts).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.-P.K., W.L., and S.J.L. conceived and designed research; J.-P.K., E.P.S.d.C., G.S.-W., M.A., K.G.P., M.M., S.M., P.L., and S.J.L. performed experiments; J.-P.K., E.P.S.d.C., G.S.-W., C.J.M., A.G.W., and S.J.L. analyzed data; J.-P.K., E.P.S.d.C., G.S.-W., C.J.M., A.G.W., W.L., and S.J.L. interpreted results of experiments; J.-P.K., E.P.S.d.C., G.S.-W., C.J.M., A.G.W., and S.J.L. prepared figures; J.-P.K. and S.J.L. drafted manuscript; J.-P.K., E.P.S.d.C., G.S.-W., S.F.M., C.J.M., A.G.W., W.L., and S.J.L. edited and revised manuscript; J.-P.K., E.P.S.d.C., G.S.-W., M.A., K.G.P., M.M., S.M., P.L., S.F.M., C.J.M., A.G.W., W.L., and S.J.L. approved final version of manuscript.