Abstract
The anterior lateral bed nucleus of the stria terminalis (alBST) expresses glucagon-like peptide-1 receptors (GLP1Rs) and receives input from caudal brainstem GLP1 neurons. GLP1 administered centrally reduces food intake and increases anxiety-like behavior and plasma corticosterone (cort) levels in rats, whereas central GLP1R antagonism has opposite effects. Anxiogenic threats and other stressors robustly activate c-fos expression in both GLP1-producing neurons and also in neurons within alBST subregions expressing GLP1R. To examine the functional role of GLP1R signaling within the alBST, adult male Sprague Dawley rats received bilateral alBST-targeted injections of an adeno-associated virus (AAV) vector expressing short hairpin RNA (shRNA) to knock down the translation of GLP1R mRNA (GLP1R-KD rats), or similar injections of a control AAV (CTRL rats). In situ hybridization revealed that GLP1R mRNA is expressed in a subset of GABAergic alBST neurons, and quantitative real-time PCR confirmed that GLP1R-KD rats displayed a significant 60% reduction in translatable GLP1R mRNA. Compared with CTRL rats, GLP1R-KD rats gained more body weight over time and displayed less anxiety-like behavior, including a loss of light-enhanced acoustic startle and less stress-induced hypophagia. Conversely, while baseline plasma cort levels were similar in GLP1R-KD and CTRL rats, GLP1R-KD rats displayed a prolonged stress-induced elevation of plasma cort levels. GLP1R-KD and CTRL rats displayed similar home cage food intake and a similar hypophagic response to systemic Exendin-4, a GLP1R agonist that crosses the blood–brain barrier. We conclude that GLP1R expressed within the alBST contributes to multiple behavioral responses to anxiogenic threats, yet also serves to limit the plasma cort response to acute stress.
SIGNIFICANCE STATEMENT Anxiety is an affective and physiological state that supports threat avoidance. Identifying the neural bases of anxiety-like behaviors in animal models is essential for understanding mechanisms that contribute to normative and pathological anxiety in humans. In rats, anxiety/avoidance behaviors can be elicited or enhanced by visceral or cognitive threats that increase glucagon-like peptide-1 (GLP1) signaling from the caudal brainstem to the hypothalamus and limbic forebrain. Data reported here support a role for limbic GLP1 receptor signaling to enhance anxiety-like behavior and to attenuate stress-induced elevations in plasma cort levels in rats. Improved understanding of central GLP1 neural pathways that impact emotional responses to stress could expand potential therapeutic options for anxiety and other stress-related disorders in humans.
Introduction
Affective anxiety is characterized by behavioral inhibition, autonomic arousal, and increased vigilance, reflecting a future-oriented emotional state generated by real and perceived threats to homeostatic well being (Davis et al., 2010). Basic and clinical research has emphasized the importance of the bed nucleus of stria terminalis (BST) in regulating affective and physiological components of anxiety (Lebow and Chen, 2016). Anxiogenic responses to threat can be innate or conditioned through learning (Pacak et al., 1998; Palkovits, 1999; Pacák and Palkovits, 2001), but they always include endocrine and autonomic adjustments that alter physiological functions. Together with the central amygdala (CEA), the BST contributes to this emotional brain–body interface as a conduit through which the cerebral cortex generates somatic, autonomic, and neuroendocrine responses to internal and external threat stimuli (Poulin et al., 2009; Radley et al., 2009; Walker et al., 2009; Lebow and Chen, 2016). The anterolateral BST (alBST) receives interoceptive feedback about the physiological state of the body (Alheid and Heimer, 1988; de Olmos and Heimer, 1999; Walker and Davis, 2008), which can modulate emotional state and even elicit anxiety-like behavior in the absence of consciously perceived threat (Rinaman, 2011). BST neural activity in humans indexes hypervigilant threat monitoring (Somerville et al., 2010), and individual differences in anxiety are determined, at least in part, by individual differences in BST circuit function (Duvarci et al., 2009). In rats and in humans, BST neurons have higher baseline and sensory-driven activity in individuals with more anxious phenotypes (Somerville et al., 2010), and the anxious phenotype in rats is eliminated by alBST lesions (Duvarci et al., 2009). Thus, it is important to elucidate how interoceptive neural signaling pathways alter BST responsiveness.
Interoceptive signals are relayed to the CEA and alBST by axonal projections arising from caudal brainstem noradrenergic and peptidergic neurons (Bienkowski and Rinaman, 2013), including glucagon-like peptide-1 (GLP1) neurons whose cell bodies occupy the caudal nucleus tractus solitarius (cNTS) and adjacent medullary reticular formation (Rinaman, 1999a, 2007, 2010, 2011; Rinaman and Schwartz, 2004). GLP1 neurons are activated and express the immediate-early gene c-fos in rats after stressful and anxiogenic treatments (Rinaman, 1999a,b; Rinaman and Comer, 2000; Maniscalco et al., 2015). Further, we reported that manipulation of the interoceptive state by fasting rats overnight markedly attenuates the ability of anxiogenic threat to activate c-fos expression in GLP1 neurons, and concurrently reduces anxiety-like behavior and threat-induced c-fos activation within the CEA and alBST (Maniscalco et al., 2015).
GLP1 axon terminals target the CEA, alBST, and other hypothalamic and limbic forebrain regions that express the G-protein-coupled GLP1 receptor (GLP1R; Göke et al., 1995; Merchenthaler et al., 1999; Gu et al., 2013). GLP1 delivered intracerebroventricularly (Möller et al., 2002; Gulec et al., 2010) or directly into the CEA increases behavioral signs of anxiety and malaise in rats (Kinzig et al., 2003; Kanoski et al., 2012), whereas anxiety-like behavior is reduced in rats after intracerebroventricular infusion of a GLP1R antagonist (Kinzig et al., 2003) or after targeted deletion of hypothalamic GLP1R in mice (Ghosal et al., 2017). However, the behavioral and physiological roles of endogenous GLP1R signaling within the alBST remain unclear, as does the neurochemical phenotype of alBST neurons that express GLP1R. GABAergic neurons that express corticotropin-releasing hormone (CRH) are concentrated primarily within the oval and fusiform subnuclei of the alBST in rats (Morin et al., 1999; Dabrowska et al., 2013), where GLP1-positive axon terminals are concentrated (Rinaman, 2010). Thus, the present study was designed to test the hypothesis that GLP1R mRNA is expressed by GABAergic neurons within the rat alBST, including neurons that express CRH mRNA, and that GLP1R signaling within this limbic forebrain region contributes to affective and physiological responses to homeostatic threat. Our results support the hypothesized GABAergic phenotype of GLP1R-expressing alBST neurons, and reveal an interesting dissociation between behavioral and stress hormone responses to anxiogenic threat in rats with virally mediated knockdown of GLP1R mRNA within the alBST.
Materials and Methods
Subjects
Adult male Sprague Dawley rats (n = 46; ∼300 g at surgery; Harlan Laboratories) were pair housed in plastic tub cages in an Association for Assessment and Accreditation of Laboratory Animal Care International-accredited facility (20–22°C, 12 h light/dark, lights on at 7:00 A.M.), with ad libitum access to pelleted chow (catalog #5001, Purina) and tap water, except as noted. Rats were acclimated to this environment for at least 1 week before stereotaxic surgery (described below). Experimental protocols were approved by the Florida State University Institutional Animal Care and Use Committee and were consistent with the US Public Health Service Policy on the Humane Care and Use of Laboratory Animals and the Guide for the Care and Use of Laboratory Animals.
Surgery
Rats were anesthetized by inhalation of isoflurane (1–3% in oxygen; Halocarbon Laboratories) and placed into a stereotaxic frame in the flat-skull position. Rats were treated either with an adeno-associated virus (AAV) that expresses GFP reporter and uses shRNA to chronically knock down translation of GLP1R mRNA (AAV1.shRGLP1r07.CB7.EGFP.SV40; developed by the Hayes laboratory and the University of Pennsylvania Viral Vector Core, as originally described; Schmidt et al., 2016; titer, 5.2 × 1012 vector genomes/ml; GLP1R-KD rats), or with a control AAV1-expressing EGFP but no shRNA (titer, 5.2 × 1012 vector genomes/ml; CTRL rats). In each rat, one of these two AAVs was delivered bilaterally via pressurized microinjection into both the dorsal and ventrolateral subnuclei of the alBST, using the following stereotaxic injection coordinates (from bregma): 0.3 mm posterior, 1.7 mm lateral, and 6.6 mm ventral for the dorsal alBST; and 7.4 mm ventral for the ventral alBST (Paxinos and Watson, 1997). Virus was delivered (300 nl/site; 1.2 μl total/rat) using a glass micropipette (tip diameter, 20–30 μm) connected to a 10 μl Hamilton syringe, with delivery speed (300 nl/min) controlled by a digital stereotaxic microinjector (catalog #QSI 53311, Stoelting). All rats were allowed to recover for at least 2 weeks after AAV injection surgery before the onset of behavioral testing, during which time they were acclimated to handling for periodic recording of body weight (BW).
Experimental cohorts
The AAV injections described above were performed in four separate surgical cohorts, as summarized in Table 1. AAV-injected rats within each cohort were subsequently subjected to a variety of behavioral tests, with at least 3 rest days between consecutive tests. Final group sizes differed across tests (Table 2) because each cohort of rats was not used for every test. Table 1 reports the tests conducted, the order of testing, and the postmortem tissue analyses conducted for rats within each cohort. Table 2 reports group sizes for each behavioral or physiological assay; group sizes also are included in the legends accompanying each figure.
Surgical and Experimental Cohorts
Results of statistical comparisons (ANOVA)
Assessments of anxiety-like behavior
Anxiety-like behavior was measured in CTRL and GLP1R-KD rats using several tests, described below. The face, construct, and predictive validity of these tests for assessing anxiety-like behavior in rats has been reported and reviewed previously (Pellow et al., 1985; Walker and Davis, 1997a; File et al., 2004; Ramos, 2008; Davis et al., 2010; Bourin, 2015; Lezak et al., 2017). Unless otherwise noted, all behavioral assessments were conducted during the early light phase of the photoperiod (9:00 A.M. to 12:00 P.M.) in a procedure room with moderate overhead fluorescent lighting, located adjacent to the animal housing room. Testing apparatuses were cleaned with an odorless mild detergent (Seventh Generation, Grove Collaborative) and allowed to dry completely between rats to reduce olfactory cues.
Open-field test.
The open-field (OF) apparatus comprised a square arena with a smooth dark gray metal floor (0.5 m2) and outwardly sloping gray metal walls (0.5 m high), positioned on the floor. Behavior was recorded using a ceiling-mounted webcam (Logitech Pro 960-001070) positioned above the center of the OF arena, and later analyzed using ANY-maze software (Stoelting). Rats were placed into one corner of the arena with their head pointing toward the center; the investigator then left the room and the exploratory activity of the rat was video recorded for 10 min. During subsequent off-line analysis of video records, the central zone of the OF was defined as the central 0.04 m2 region of the arena floor. The latency of each rat to enter this central zone was defined as the time elapsed before the head of the rat completely entered the zone for the first time. The total time spent within the anxiogenic central zone of the OF also was assessed.
Novelty-suppressed feeding.
Beginning a few hours before lights out, rats were deprived of food (but not water) in their home cage for 21 h before novelty-suppressed feeding (NSF) testing on the following day. The NSF apparatus comprised a standard-sized tub cage (0.2 m wide × 0.2 m deep × 0.45 m long) with clear Plexiglas walls and a square opening on one end connected to an externally mounted chow hopper. To further increase novelty, the tub cage contained no bedding or other floor covering and was topped by a clear Plexiglas lid to permit unobstructed digital recording from a webcam positioned 1.0 m above the cage. Food-deprived rats were placed into the center of this novel tub cage, with familiar chow readily available in the hopper. The investigator left the room, and the activity of the rat was recorded for 10 min. Latency to begin eating from the hopper and time spent eating (i.e., biting, chewing) were quantified manually from video records by an investigator who was blind to treatment group. Mean traveling speed and total distance traveled were analyzed using ANY-maze software. Since rats spent an average of only 1–2 min consuming food during the NSF test (see Results), the amount of food consumed (corrected for spillage) could not be confidently assessed.
Light-enhanced startle.
Acoustic startle responses were measured using an SR-LAB System (San Diego Instruments), as in our previous report (Maniscalco et al., 2015). The system comprised a ventilated sound- and light-attenuated chamber containing an internal stabilimeter beneath a Plexiglas cylinder (diameter, 8.2 cm) that was mounted on a Plexiglas base. Rats were secured within the cylinder, which was large enough for them to turn around, and then the outer chamber door was closed. Inside the chamber, a tweeter mounted 24 cm above the cylinder provided continuous background white noise (50 dB) interrupted by higher-intensity acoustic startle stimuli (50 ms bursts of 90, 95, or 105 dB white noise; 10 bursts/intensity) presented in randomized order over a 15 min period, with noise delivery controlled via the SR-LAB System software. Reflexive acoustic startle responses transduced by a piezoelectric accelerometer mounted on the platform below the cylinder were digitized, rectified, and recorded as 1 ms readings triggered at the onset of each startle stimulus. The peak amplitude response to each startle stimulus (in millivolts) was used as the dependent measure. A standard approach for running the acoustic startle and light-enhanced startle (LES) protocols (Walker and Davis, 1997a) was slightly modified to include interstimulus intervals that varied randomly between 20 and 40 s. Each rat was tested twice on the same day, initially in the dark and subsequently in the light condition, to assess within-subjects LES. After 5 min acclimation within the closed chamber, acoustic startle amplitudes were measured initially in darkness over the 15 min testing period. Each rat was then returned to its home cage for 60–90 min (no food available) and subsequently returned to the acoustic startle chamber and retested under the light condition (500 lux within the chamber; 5 min acclimation followed by 15 min acoustic startle testing). The primary dependent variables for each rat were average peak startle amplitudes at each noise intensity during testing in dark and light conditions. LES was calculated within subjects as the difference in average peak startle amplitude between light and dark testing at each noise intensity level.
Elevated plus maze.
The elevated plus maze (EPMZ) apparatus comprised two open arms (45 × 10 cm) with transparent 1-cm-high edging, and two closed arms (45 × 10 × 48 cm) extending from a common central platform (10 × 10 cm) elevated 0.9 m above the floor of the room. Rats were placed onto the center of the EPMZ with their heads facing an open arm. The investigator then left the room, and the exploratory behavior of the rat was recorded for 5 min using a ceiling-mounted webcam positioned over the maze. Digital video records were analyzed using ANY-maze software to determine the total distance traveled; the mean traveling speed; the time spent within open arms, closed arms, and center; as well as the number of entries into open versus closed arms.
Home cage feeding after elevated open platform stress.
Brief elevated open platform (EOP) exposure was used as an acute stressor, as in previous reports (Kinzig et al., 2003; Maniscalco et al., 2015). The EOP comprised an open arm of the above-described EPMZ, with access to the center blocked to thereby restrict the rat to the EOP. Rats were placed onto the EOP for 10 min, 30–60 min before dark onset. Rats were then returned to a clean home cage, and their overnight chow and water intake was recorded 16 h later (3 h after lights on).
Plasma corticosterone response to EOP stress
After a baseline tail blood sample (100 μl) was collected at time 0 (immediately after removing rats from their home cage), rats were placed onto the EOP for 10 min, and then returned to their home cage. Additional tail blood samples were collected 30, 60, and 90 min after the onset of EOP exposure (within-subjects). Blood was collected into SAFE-T-FILL capillary collection tubes (catalog #07 6013, RAM Scientific) on ice, and subsequently centrifuged at 4°C for 15 min at 14,000 rpm (Centrifuge 5804R, Eppendorf). Plasma was aliquoted into Eppendorf tubes and stored at −80°C. Plasma corticosterone (cort) concentrations were later analyzed in duplicate from thawed samples using a Corticosterone EIA Kit (catalog #AC-14F1, Immunodiagnostic Systems) according to the manufacturer instructions. Briefly, plasma was diluted 1:10 with Sample Diluent, and 100 μl each of diluted plasma and Enzyme Conjugate Solution was added into each well of the microplate. After overnight incubation at 4°C, the plasma-enzyme solution was discarded, the microplate was rinsed three times with Wash Solution and then reacted with 200 μl of tetramethylbenzidine substrate for 15–20 min at room temperature (RT). The reaction was terminated by adding 100 μl of Stop Solution, and the microplate wells were read immediately using an ELx800 NB Universal Microplate Reader (BIO-TEK Instruments). Assay sensitivity was 0.55 ng/ml. Cort levels are expressed as nanograms per milliliter plasma.
Hypophagic response to systemically administered Exendin-4
After overnight (21 h) food deprivation (water was available), rats received an intraperitoneal injection of either vehicle (sterile 0.15 m NaCl) or the GLP1R agonist Exendin-4 (Ex4; 2.0 μg/kg), then were returned to their home cage, where premeasured amounts of chow and water were available. Chow (corrected for spillage) and water intake were measured at 1, 4, and 21 h after intraperitoneal injection. Injections were counterbalanced such that each rat received both saline and Ex4 injections after overnight food deprivation, with the two treatments separated by 5 d.
Quantitative analysis of GLP1R mRNA knockdown
Translatable levels of GLP1R mRNA within BST-enriched forebrain tissue samples were measured using quantitative real-time PCR (qRT-PCR) in a subset of rats injected bilaterally with either the AAV-shRNA construct (GLP1-KD rats, n = 8) or with the control AAV (CTRL rats, n = 8). For this purpose, after completing behavioral assessments, rats were anesthetized and decapitated, and brains were rapidly removed, flash frozen in isopentane, and stored at −80°C. A forebrain tissue block containing the BST was mounted onto the cryostat chuck, and thin sections were examined under a fluorescence microscope (model 80i, Nikon) until GFP-expressing cells were visualized within the alBST injection site. Bilateral BST micropunches (1 mm3) of GFP-labeled tissue were collected and kept frozen for subsequent qRT-PCR analysis, using previously reported primers and procedures (Alhadeff et al., 2017). Total RNA was extracted from micropunches using TRIzol (Thermo Fisher Scientific) and the RNeasy kit (Qiagen). cDNA was synthesized using a High Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). TaqMan gene expression kits and PCR reagents (Applied Biosystems) were used to quantify the relative levels of translatable mRNA for GLP1R (GLP1r, Rn00562406) relative to rat GAPDH mRNA (Gapdh, Rn01775763_g1). Relative mRNA expression was calculated using the comparative Ct method, as previously described (Hayes et al., 2010; Alhadeff et al., 2017). qRT-PCR data from 2 of the original 16 rats prepared for analysis (n = 1/AAV injection group) were subsequently discarded due to technical errors during RNA isolation and cDNA synthesis.
Immunohistochemistry and in situ hybridization
Rats not used for qRT-PCR were anesthetized with sodium pentobarbital (100 mg/kg, i.p.; Nembutal) and transcardially perfused with physiological saline followed by 4% paraformaldehyde in 0.1 m sodium phosphate buffer (PB). Perfused brains were removed, postfixed overnight, blocked, and cryoprotected in a 20% sucrose solution for 24–72 h. Brains were cut into six adjacent series of coronal sections [25 μm for immunohistochemistry (IHC), 35 μm for in situ hybridization (ISH)] using a freezing stage sliding microtome (Leica). Free-floating sections were collected into cryopreservant solution (Watson et al., 1986) and stored at 20°C before processing.
IHC to localize GFP.
To evaluate the placement and extent of AAV transfection in GLP1R-KD and CTRL rats, sections were processed for immunofluorescent localization of GFP to enhance native EGFP reporter gene fluorescence. Sections were rinsed for 1 h in several changes of 0.1 m PB, pH 7.2, then incubated in chicken anti-EGFP (catalog #ab13970, Abcam; RRID:AB_300798; 1:2K) in 0.1 m PB, pH 7.3, containing 0.3% Triton X-100 and 1% normal donkey serum for 22–24 h at RT, rinsed for 45 min in several changes of PB, incubated in Alexa Fluor 488-conjugated donkey anti-chicken IgG (1:400; Jackson ImmunoResearch) for 2 h at RT followed by overnight incubation at 4°C, and finally rinsed with PB for 1 h. Immunolabeled tissue sections were mounted onto charged microscope slides (Diamond White Glass, Globe Scientific), dehydrated in ascending ethanol solutions, defatted with xylene, and coverslipped using Cytoseal 60 [Thermo Fisher Scientific (via VWR)].
RNAscope to visualize alBST expression of mRNA for GLP1R, GAD1, and CRH.
Free-floating forebrain sections containing the alBST were rinsed for 1 h in several changes of PB, treated with H2O2 (RNAscope reagent 322335, Advanced Cell Diagnostics) for 30 min at RT, and rinsed for 45 min in PB at RT. Sections were mounted out of 0.01 m Tris buffer, pH 7.3, onto Gold Seal UltraStick Adhesion Microscope Slides (catalog #3039-002, Thermo Fisher Scientific), air dried for 1 h, dipped into 100% ethanol (1 s), and air dried for 30 min. After a hydrophobic barrier was created around each section, slides were air dried overnight at RT. Sections were treated with Protease IV (RNAscope reagent 322336, Advanced Cell Diagnostics) for 20–30 min at RT followed by 3 × 1 min rinses with distilled water. Pretreated tissue sections were processed immediately using probes designed by Advanced Cell Diagnostics to label GLP1R mRNA (Rn-GLP1r; catalog #315221) or to colocalize mRNAs for GLP1R and GAD1(Rn-GAD1; catalog #316401-C2) or GLP1R and CRH (Rn-Crh; catalog #318931-C3), as described below.
Single labeling of GLP1R mRNA was detected using RNAscope 2.5 HD Detection Reagent-BROWN (reagent 322310, Advanced Cell Diagnostics) according to the manufacturer instructions and using kit components. Sections were incubated with a GLP1R mRNA probe for 2 h at 40°C in a HybEZTM oven (Advanced Cell Diagnostics), followed by amplification steps 1–6 and a diaminobenzidine (DAB) reaction to generate brown labeling. Double labeling of mRNA for GLP1R/GAD1 or for GLP1R/CRH was detected using RNAscope Multiplex Fluorescent Reagent Kit version 2 (catalog #323100). Sections were incubated in a mix containing probes for GLP1R/GAD1 or for GLP1R/CRH for 2 h at 40°C in the HybEZTM oven, followed by amplification steps 1–3, and by sequential labeling of each probe with a fluorophore-conjugated TSAP (Tyramine Signal Amplification Plus, PerkinElmer), according to the Advanced Cell Diagnostics protocol, which includes DAPI nuclear counterstaining as the final step before coverslipping. GLP1R was labeled with TSAP-Cy3, while GAD1 and CRH were labeled with TSAP-Cy5. For both DAB- and fluorescent-labeling methods, sections were washed 3 × 3 min in wash buffer (catalog #310091) at RT between each step. After the labeling reaction, sections were dehydrated in ascending ethanol solutions, defatted with xylene, and coverslipped using Cytoseal 60.
Image acquisition
Images of GFP-labeled BST injection sites, DAB peroxidase-labeled GLP1R mRNA, and fluorescent DAPI plus dual GLP1R/GAD1 mRNA were acquired with a bright-field and epifluorescent microscope (catalog #BZ-X700, KEYENCE). Higher-resolution images of GLP1R/GAD1 and GLP1R/CRH double-fluorescent mRNA labeling were acquired using a Leica TCS SP8 Confocal Microscope with a 20× air-objective and a 100× oil-objective. Cy3 was excited using 552 nm OPSL laser, and Cy5 using a 638 nm Diode laser. Nuclear DAPI could not be visualized using the confocal microscope, which lacked the required laser. Confocal images were collected sequentially using Leica LAS version 4.0 image collection software, and were combined using the 3D Visualization module for rendering/reconstructing tissue volumes.
For quantification of the extent of GLP1R/GAD1 mRNA colocalization in two CTRL rats, multiple confocal tile images centered on the ventrolateral alBST were collected using a 20× air-objective and merged. Each tile consisted of 14 0.69-μm-thick optical sections. For imaging GLP1R/GAD1 and GLP1R/CRH double labeling at higher magnification, 10–12 0.28-μm-thick optical sections were collected through selected neurons using a 100× oil-objective. Leica image software was used to generate maximum intensity Z-stacks and 3D rotatable maximum intensity projections.
Experimental design and statistical analysis
The experimental design across four surgical cohorts is summarized in Table 1, and the results of statistical comparisons are presented in Table 2. Data were analyzed using one-way, two-way, or mixed repeated-measures ANOVA (with Bonferroni correction for repeated comparisons), as specified for different experiments in the Results section. Post hoc comparisons were conducted when ANOVA revealed significant (p < 0.05) main effects and/or interactions of experimental variables on dependent measures. The within-subjects relationship between GLP1R mRNA within the BST (assessed by qRT-PCR) and change of body weight over a 5 week period after AAV microinjection in GLP1R-KD and CTRL rats was analyzed by computing the Pearson correlation coefficient between these variables. The 5 week time point for body weight gain was selected for this correlational analysis because it occurred before rats were exposed to behavioral assays involving food deprivation. Combined group data are presented in figures as the mean ± SEM.
For each assay, an a priori decision was made to remove data points from analysis if (1) they met an outlier criterion of lying ≥2 SDs away from the group mean, (2) if postmortem qRT-PCR failed to confirm expected effects on GLP1R mRNA within the alBST, or (3) if postmortem histology indicated that stereotaxically guided AAV injection sites were centered in regions other than the dorsal and ventral alBST. This resulted in the elimination of all data from two rats (one CTRL, one GLP1R-KD) for which technical errors in qRT-PCR prevented the interpretation of GLP1R mRNA levels, and one GLP1R-KD rat with misplaced AAV injection sites (histologically defined). Data from one CTRL rat were eliminated as outlier data within the OF test. The group sizes and statistical outcomes presented in Tables 1 and 2 do not include these eliminated cases.
Results
shRNA-induced silencing of GLP1R mRNA expression in the alBST
The accuracy of bilateral alBST injection sites was confirmed by examining GFP immunofluorescence in tissue sections from all GLP1R-KD and CTRL rats that were perfused with fixative after completing behavioral assessments. Injection site localization and spread of GFP reporter is depicted in a rostrocaudal series of sections taken from a representative GLP1R-KD rat with the largest bilateral injection sites within the dorsal and ventral alBST (Fig. 1A–D), as evidenced by the largest regions of robust EGFP labeling. The EGFP reporter expressed by both AAV constructs anterogradely labels the axonal projections of transfected neurons (including labeling of individual axons and large fiber bundles within the BST, hypothalamus, and other regions), but this did not interfere with the ability to confirm that AAV injection sites were centered within the dorsal and ventral alBST.
A–L, Rostrocaudal distribution of EGFP reporter immunofluorescence in a GLP1R-KD rat with the largest bilateral AAV1-shRNA injection sites targeting the alBST dorsal and ventral to the anterior commissure. Black and white images were inverted to enhance visualization of tissue landmarks; thus, EGFP-positive profiles are black. Distance from bregma (in mm) is indicated in the bottom right corner of each image. A is the most rostral; L is the most caudal. A few tissue landmarks are labeled for orientation. ac, anterior commissure; f, fornix; fi, fimbria of hippocampus; LH, lateral hypothalamus; LS, lateral septum; LV, lateral ventricle; mfb, medial forebrain bundle; ox, optic chiasm; sfo, subfornical organ; 3, third ventricle.
When tissue sections from a subset of rats were processed for single-label RNAscope ISH to localize GLP1R mRNA within the alBST, visual inspection revealed less labeling in sections from GLP1R-KD rats (n = 3) compared with sections from CTRL rats (n = 3). The apparent reduction in labeling in GLP1R-KD rats (Fig. 2, compare A, B) was not quantified, due to the incomplete sampling of matched sections containing the same GLP1R-expressing BST subregions in both AAV injection groups. However, in fresh-frozen brains that were processed using qRT-PCR to detect translatable GLP1R mRNA in BST-enriched tissue, ANOVA revealed a significant 60% reduction in GLP1R-KD rats compared with CTRL rats (Fig. 2C; Table 2).
GLP1R mRNA detected using RNAscope and qRT-PCR. A, Single in situ hybridization (DAB peroxidase label) localizing GLP1R mRNA within the ventral alBST in a CTRL rat. B, Single in situ hybridization (DAB peroxidase label) localizing GLP1R mRNA within the ventral alBST in a GLP1RKD rat, in which reduced labeling is evident compared with the CTRL. The box within the schematic inset (lower left) indicates the lateral ventral BST region (BSTLV) depicted in A and B. C, Results of qRT-PCR analysis of translatable GLP1R mRNA levels within the alBST in CTRL (n = 7) vs GLP1R-KD (n = 7) rats. Levels were significantly reduced by ∼60% in rats injected with the shRNA-expressing AAV (GLP1R-KD) compared with rats injected with the control AAV (CTRL). #p = 0.046.
GLP1R mRNA is expressed by a subset of GABAergic alBST neurons
To determine the neurochemical phenotype of alBST neurons expressing GLP1R mRNA, tissue sections from two CTRL rats were processed using RNAscope for dual fluorescent in situ hybridization. When DAPI-labeled tissue sections were viewed using an epifluorescence microscope, many DAPI-positive nuclei were surrounded by fluorescently labeled puncta that suggested colocalized mRNA for GLP1R and GAD1 (Fig. 3A,B). While only a subset of GAD1-positive neurons within the alBST also expressed GLP1R, most GLP1R-positive neurons appeared to coexpress GAD1, identifying them as GABAergic. To quantify this observation, 60 GLP1R mRNA-expressing neurons within the alBST were photographed at high magnification using a confocal microscope, with dual labeling inspected in flattened z-stack images. The large majority of these GLP1R mRNA-positive neurons (i.e., 31 of 33 neurons in one rat, and 26 of 27 neurons in the other) also appeared to express mRNA labeling for GAD1 (Fig. 3C). While these counts may overestimate the actual extent of dual labeling (due to our inability to image nuclear DAPI on the confocal microscope), dual labeling was confirmed in selected neurons using a 100× oil-objective and rotated views of individual cells, in which GLP1R and GAD1 mRNA transcripts were intermixed (Fig. 3C, inset). In addition, a subset of GLP1R mRNA-expressing neurons in the dorsal alBST appeared to express CRH mRNA (Fig. 3D), whereas no examples of colocalized GLP1R mRNA and CRH mRNA were found within the ventral alBST.
A, B, Epifluorescence images of dual in situ hybridization (RNAscope) with DAPI counterstaining (A, B depict the same field) to localize mRNA for GLP1R (red) and GAD1 (green) within the ventral alBST in a CTRL rat. The majority of DAPI-positive nuclei that are surrounded by GLP1R mRNA labeling also are surrounded by GAD1 mRNA labeling; eight dual-labeled cells visible within this field are marked. Many additional GAD1 mRNA-positive neurons do not express GLP1R mRNA. In the confocal image shown in C, small arrows point out cells that appear to be double labeled. The larger arrow with a different orientation (C, lower left) points out one double-labeled neuron shown at higher magnification in the inset. The inset depicts a 3D maximum intensity projection of this same neuron, slightly rotated to demonstrate intracellular colocalization of both GLP1R and GAD1 mRNA transcripts within the cell. D, Dual in situ hybridization localizing mRNA for GLP1R (red) and CRH (green) within the dorsal alBST in a CTRL rat. A subset of GLP1R mRNA-expressing cells also express CRH mRNA. The inset shows a 3D maximum intensity projection, slightly rotated to demonstrate intracellular colocalization of both mRNA transcripts within a single neuron (indicated by large white arrow in the top right region in B). Smaller arrows in D point out three additional cells that appear to be double labeled. E, Dual in situ hybridization localizing mRNA for GLP1R (red) and CRH (green) within the left hypothalamic PVN in a GLP1R-KD rat in which AAV-shRNA was microinjected bilaterally into the dorsal and ventral alBST. F, The distribution of CRH mRNA-expressing neurons within the medial parvocellular PVN in this tissue section overlaps the distribution of anterogradely labeled, EGFP-immunofluorescent fibers (green; photographed using a different fluorescent emission channel) in the same tissue section. The same blood vessel (bv) is labeled in E and F as a landmark. Scale bar labels are in micrometers.
Using RNAscope, we also visualized CRH mRNA within the medial parvocellular subdivision of the hypothalamic paraventricular nucleus of the hypothalamus (PVN), where GLP1R mRNA is expressed (Fig. 3E). The same PVN sections contained anterogradely labeled, EGFP-positive fibers originating from the alBST AAV injection sites (Fig. 3F); however, no EGFP-positive neural cell bodies were observed within the PVN, evidence that the AAV construct did not anterogradely transfect postsynaptic PVN neurons. However, it is possible that GLP1R is transported to the axon terminals of alBST neurons that express GLP1R mRNA and innervate the PVN and other central regions, such that GLP1R knockdown within the alBST alters functional GLP1 signaling in these regions.
GLP1R knockdown increased body weight gain but not home cage food intake
Repeated-measures ANOVA (with AAV postinjection day as the repeated measure) was used to compare body weight gain over time in GLP1R-KD versus CTRL rats. There was a significant between-subjects effect of the AAV injection group, a significant within-subjects effect of postinjection time, and a significant interaction between injection group and time on body weight gain (Table 2). Post hoc t tests demonstrated that GLP1R-KD rats gained more weight than CTRL rats by 15 d postinjection, before behavioral tests began, and that this difference in BW gain persisted through 61 d postinjection (Fig. 4A). Conversely, when 22 h home cage water and chow intake was assessed 2 and 6 weeks after surgery in rats from both AAV injection groups, there were no group differences in intake (Fig. 4B; Table 2).
Body weight, home cage chow and water intake, and correlation of BST GLP1R mRNA with body weight change over time. A, Body weight gained by rats after alBST injection of AAV-shRNA (GLP1R-KD; n = 11) or control AAV (CTRL; n = 10). GLP1R-KD rats gained more body weight than CTRL rats, with the difference evident at 9 d and reaching significance 15 d after AAV injection. #p < 0.001 to p = 0.03. B, Baseline home cage daily (22 h) chow and water intake assessed 2 and 6 weeks after AAV injections. At both time points, GLP1R-KD (n = 18) and CTRL rats (n = 14) consumed similar daily amounts of chow and water. C, Within-subjects correlation of GLP1R mRNA levels within the BST and change of body weight 5 weeks after AAV injections (n = 7/group). Change of body weight was significantly and negatively correlated with GLP1R mRNA, such that lower levels of mRNA were associated with higher body weight gain over the 5 week period. Open symbols, GLP1R-KD rats; closed symbols, CTRL rats.
In the subset of CTRL and GLP1R-KD rats in which qRT-PCR data were obtained for GLP1R mRNA levels in BST-enriched tissue, there was a significant negative correlation between GLP1R mRNA levels and body weight gain during a 5 week period after AAV microinjection, such that lower levels of GLP1R mRNA corresponded to a higher percentage body weight gain during this time (Fig. 4C; Table 2).
GLP1R knockdown reduced anxiety-like behavior in the OF test
Separate one-way ANOVAs were used to compare rats in both AAV injection groups for latency to enter the central zone of the novel OF, total time spent in the central zone, total distance traveled in the OF, and average traveling speed during the 10 min (600 s) test. There was a main effect of AAV injection group on latency to enter the central zone, and on total time spent within the central zone (Table 2), such that GLP1R-KD rats entered the central zone more quickly and spent more total time there compared with CTRL rats (Fig. 5A). There was no effect of AAV injection group on the total distance traveled (Fig. 5B) or average speed of traveling (Fig. 5C) within the OF. Figure 5D shows computerized traces of two-dimensional exploration of the OF by a representative CTRL rat and a GLP1R-KD rat; the GLP1R-KD rat can be seen to traverse the center of the OF more often than the CTRL rat.
Anxiety-like behavior assessed in the OF. A, Compared with CTRL rats (n = 9), GLP1R-KD rats (n = 11) displayed reduced latency to enter the “anxiogenic” center of the OF (i.e., they entered earlier during the test), and spent more time within the center. B, C, Neither the total distance traveled within the OF (B) nor the mean traveling speed (C) differed between CTRL and GLP1R-KD rats. D, Computerized tracing of exploratory activity recorded within the OF in representative rats from each group, illustrating more center avoidance in the CTRL vs GLP1R-KD rat. #p = 0.004 for latency, #p = 0.045 for time in center.
GLP1R knockdown blunted the effect of novelty to suppress feeding in the NSF test
Separate one-way ANOVAs were used to compare rats in both AAV injection groups for latency to begin eating familiar chow in a novel environment after overnight food deprivation, time spent eating during the 10 min (600 s) test, total distance traveled, and average traveling speed within the novel environment. There were significant main effects of AAV injection group on latency to begin eating and also on time spent eating (Table 2). Compared with CTRL rats, rats injected with GLP1R-KD began eating familiar chow in the novel cage sooner, and spent twice as much time engaged in eating (Fig. 6A). There were no between-group differences in total distance traveled (Fig. 6B) or average speed of traveling (Fig. 6C; Table 2) within the NSF arena.
NSF in CTRL (n = 9) and GLP1R-KD (n = 11) rats deprived of chow for 22 h. A, Compared with CTRL rats, GLP1R-KD rats displayed reduced latency to begin eating familiar chow in the novel cage (i.e., they began eating sooner) and more time spent eating during the test. #p < 0.01. B, C, Neither the total distance traveled within the novel cage (B) nor the mean traveling speed (C) differed between CTRL and GLP1R-KD rats.
GLP1R knockdown did not alter anxiety-like behavior in the EPMZ
Separate mixed ANOVAs were performed to compare rats in both AAV injection groups for time spent within the open arm, closed arm, and center of the EPMZ, the total number of entries to these three zones, and the average traveling speed and distance traveled during the 5 min (300 s) test. AAV injection group had no significant effect on any of these parameters (Fig. 7; Table 2). The lack of between-group differences in assessed EPMZ behaviors held true regardless of whether rats had been previously tested in other anxiety assays (i.e., OF and NSF for cohorts 1 and 2; Table 1) before the EPMZ test, or whether the EPMZ was the first behavioral test of the rats (i.e., cohorts 3 and 4; Table 1).
Anxiety-like behavior in the EPMZ. A, There were no significant differences between CTRL (n = 21) and GLP1R-KD (n = 22) rats in time spent within the open arms, closed arms, or center of the maze. B–D, There also were no group differences in the number of entries to different arms of the maze (B), the average traveling speed within the maze (C), or the total distance traveled (D).
GLP1R knockdown abolished LES
Acoustic startle amplitudes were first analyzed separately within each AAV injection group using repeated-measures ANOVA, with lighting condition (dark vs light) and noise intensity (in decibels) as within-subjects factors. Under both lighting conditions, there was a significant within-subjects effect of noise intensity on average acoustic startle amplitude in both the CTRL and GLP1R-KD groups (Table 2). There also was a significant main effect of lighting condition and a significant interaction between lighting condition and noise intensity in CTRL rats, but not in GLP1R-KD rats (Table 2). As evident in the group data depicted in Figure 8A, startle amplitudes were increased in CTRL rats tested in the light compared with the dark condition, whereas the anxiogenic effect of light to increase startle amplitude was absent in GLP1R-KD rats. For statistical comparisons of LES within subjects, data were analyzed using repeated-measures ANOVA, with AAV injection group as the between-subjects factor, and noise intensity as the repeated within-subjects factor. There were significant main effects of both AAV injection group and noise intensity on LES, and a significant interaction between these factors (Table 2). Post hoc paired t tests revealed significant within-subjects LES at the two highest noise levels (95 and 105 dB) in CTRL rats, but a complete absence of LES in GLP1R-KD rats (Fig. 8B).
Acoustic startle and LES. A, Average startle amplitude increased with increasing sound intensity (in dB) in both GLP1R-KD (n = 10) and CTRL (n = 12) rats tested under the dark or light condition. In CTRL rats, ANOVA revealed a significant effect of lighting condition, such that acoustic startle response amplitudes were significantly increased under the “anxiogenic” light compared with the dark condition (*). Conversely, there was no significant effect of lighting condition on acoustic startle responses in GLP1R-KD rats (ns, not significant). This between-group difference is also evident in B, where within-subjects LES at each noise intensity level is plotted in CTRL and GLP1R-KD rats. In CTRL rats, significant LES was evident at the 95 and 105 dB levels (#p = 0.02), whereas LES was absent at all noise intensities in GLP1-KD rats. In A, bars marked with different letters (a–c) are significantly different (p ≤ 0.02) within the same AAV group and lighting condition, demonstrating the effect of noise level to increase acoustic startle amplitude in both groups.
To examine whether differences existed between CTRL and GLP1R-KD rats in within-session habituation to noise stimuli, data were further analyzed to evaluate whether startle amplitudes differed during the first 10 versus the final 10 (of 30) noise presentations at a given decibel level during each 20 min testing period. This analysis was conducted separately for rats within each surgical group (CTRL vs GLP1R-KD) under each lighting condition. The results confirmed that startle responses did not differ at the beginning versus the end of each 20 min session, regardless of surgical group, lighting condition, or noise level (CTRL: dark, F(1,11) = 0.33, p = 0.58; light, F(1,11) = 1.17, p = 0.30; GLP1R-KD: dark, F(1,9) = 0.15, p = 0.71; light, F(1,9) = 0.02, p = 0.88).
GLP1R knockdown prevented acute stress-induced hypophagia
Overnight home-cage chow intake and body weight changes in non-food-deprived rats after acute stress (i.e., 10 min exposure to EOP) were analyzed separately using two-way ANOVA, with AAV injection group and EOP exposure as between-subjects factors. Results revealed significant between-subjects effects of AAV injection and EOP exposure, and a significant interaction between these variables on overnight (16 h) chow intake and loss of body weight (Table 2). Post hoc t tests indicated that after EOP exposure, CTRL rats consumed less chow overnight compared with GLP1R-KD rats, whereas baseline overnight chow intake in the absence of acute stress did not differ between groups (Fig. 9A). Similar results were obtained when body weight was assessed, such that CTRL rats lost weight overnight after EOP stress, but GLP1R-KD rats did not (Fig. 9B).
Home cage chow intake and overnight change of BW in non-food-deprived rats tested under baseline conditions (BL; no stress) or stressed conditions (after 10 min exposure to an EOP). A, Chow intake was reduced in CTRL rats (n = 15) after EOP stress, but not in GLP1R-KD rats (n = 16). B, CTRL rats lost BW overnight after EOP stress, whereas GLP1R-KD rats gained BW, similar to BW gain in rats from both AAV injection groups under BL conditions. In both panels, *p < 0.01 for EOP vs BL within the CTRL group. For CTRL vs GLP1R-KD rats after EOP stress, #p < 0.01 for chow intake; and #p = 0.024 for change of BW.
GLP1R knockdown did not alter the ability of systemic Ex4 to suppress food intake
The hypophagic effect of a systemically administered GLP1R agonist (i.e., Ex4, 2.0 μg/kg, i.p.) was examined in food-deprived GLP1R-KD and CTRL rats, using a within-subjects crossover design (Ex4 vs intraperitoneal saline vehicle). Separate mixed and repeated-measures ANOVAs were performed to reveal the potential effects of AAV injection groups and drug treatment on chow and water intake over time. There was a significant within-subjects effect of intraperitoneal drug treatment on chow and water intake, a significant within-subjects effect of time, and a significant interaction between these variables (Table 2). However, there was no between-subjects effect of AAV injection group on either food or water intake. Post hoc t tests confirmed that intraperitoneal Ex4 treatment suppressed chow intake (Fig. 10A) and water intake (Fig. 10B) similarly in CTRL and GLP1R-KD rats at each time point examined (i.e., 1, 4, and 21 h).
Home cage chow and water intake in CTRL (n = 6) and GLP1R-KD (n = 5) rats deprived of food for 21 h after intraperitoneal injection of saline (0.15 m NaCl) or Ex4 (2.0 μg/kg BW; within-subjects comparisons). A, B, Compared with saline, Ex4 treatment suppressed chow intake (A) and water intake (B) to a similar extent in GLP1R-KD and CTRL rats. #p ≤ 0.02, Ex4 vs saline within the same AAV injection group at that same time point.
GLP1R knockdown prolongs the plasma cort response to acute stress
Plasma cort levels were assessed in GLP1R-KD and CTRL rats at baseline (before stress, time 0) and at three time points after the onset of EOP stress exposure. Repeated-measures ANOVA (time as repeated measure) revealed a significant between-subjects effect of AAV injection, a significant within-subjects effect of time, and a significant interaction between these variables on plasma cort concentration. In both AAV injection groups, post hoc t tests indicated that cort levels were significantly elevated 30 min after the onset of EOP exposure compared with prestress baseline levels, which did not differ between AAV injection groups (Fig. 11A; Table 2). However, the duration of the cort response to EOP stress was significantly prolonged in GLP1R-KD rats compared with CTRL rats, as was the total cort response over 90 min [assessed as area under the curve (AUC); Fig. 11B]. In CTRL rats, plasma cort levels peaked at 30 min and declined back toward baseline at 60 and 90 min after EOP stress, whereas in GLP1R-KD rats, cort remained at peak levels at the 60 min time point and showed significantly less decline by 90 min compared with CTRL rats (Fig. 11A).
Plasma cort levels in CTRL (n = 6) and GLP1R-KD (n = 6) rats at baseline (time 0) and 30, 60, and 90 min after the onset of 10 min EOP stress. A, Plasma cort levels increased significantly after EOP stress in both AAV injection groups. In CTRL rats, plasma cort peaked at 30 min and then returned toward baseline at 60 and 90 min. In GLP1-KD rats, peak cort levels apparent at 30 min persisted through 60 min, and remained elevated above baseline at 90 min. In A, data points marked with different letters (a–c) indicate significant differences (p < 0.01 to p = 0.02) across time within each AAV injection group. Data points marked with the same letters within each AAV group are not significantly different. B, Total AUC values for plasma cort levels were significantly higher in GLP1R-KD vs CTRL rats, #p = 0.02.
Discussion
GLP1R mRNA within BST-enriched tissue was significantly reduced by ∼60% in GLP1R-KD rats compared with CTRL rats, similar to other studies in which the same AAV construct was used to reduce translatable GLP1R mRNA in different brain regions (Schmidt et al., 2016; Alhadeff et al., 2017; Tuesta et al., 2017; Lee et al., 2018; López-Ferreras et al., 2018). We refer to this as “GLP1R knockdown,” because we assume that reduced levels of detectable GLP1R mRNA are accompanied by reduced levels of functional GLP1R protein; however, the current lack of specific antibodies with which to detect GLP1R protein via immunocytochemistry or Western blot precludes a direct test of this assumption. Despite the incomplete reduction in translatable mRNA for GLP1R, significant effects were revealed across several distinct behavioral and physiological assays (discussed further, below). We also report novel evidence that GLP1R mRNA within the alBST is expressed by a subset of GABAergic alBST neurons, and that some GLP1R-expressing neurons within the dorsal (but not the ventral) alBST express CRH mRNA, consistent with known interactions between central GLP1 and CRH signaling pathways (Gülpinar et al., 2000; Malendowicz et al., 2003; Sarkar et al., 2003; Gotoh et al., 2005; Kageyama et al., 2012; Ghosal et al., 2013). Growing evidence supports the view that GABAergic CRH neurons within the alBST participate in emotional and behavioral responses to stress (Dabrowska et al., 2013; Gafford and Ressler, 2015), and dysfunctional regulation of limbic CRH systems may contribute to human anxiety disorders (Walker et al., 2009).
Interestingly, we found that GLP1R mRNA within BST tissue punches was inversely correlated with body weight gain over time, despite no effect of mRNA knockdown on baseline food intake; thus, alBST GLP1R knockdown may reduce energy expenditure. Additional work is needed to determine whether chronic alBST knockdown of GLP1R alters basal metabolism, caloric efficiency, core body temperature, sympathetic outflow, and/or home cage locomotor activity. We considered the possibility that the mild to moderate stress associated with behavioral testing suppressed body weight gain in CTRL rats (José Jaime et al., 2016; Herrera-Perez et al., 2017) more than in GLP1R-KD rats. However, body weight gain divergence was apparent within 2 weeks after AAV injection, before behavioral testing began. The total distance traveled and the speed of traveling within the OF, EPMZ, and NSF testing arenas did not differ between CTRL and GLP1R-KD rats, evidence that GLP1R knockdown did not suppress locomotor activity during these tests. Consistent with evidence that forebrain GLP1Rs are unnecessary for systemic Ex4-induced hypophagia in rats (Hayes et al., 2008), we also detected no difference between CTRL and GLP1R-KD rats in their hypophagic response to systemic Ex4, which suppresses food intake via vagal and central GLP1Rs (Kanoski et al., 2011). However, since alBST knockdown of GLP1R was incomplete in the present study (∼60%), we cannot conclude that these receptors play no role in the feeding-suppressive effects of GLP1 released from the intestine or of systemic Ex4 or other GLP1R agonists that cross the blood–brain barrier.
In the present study, GLP1R mRNA knockdown (1) reduced anxiety-like behavior in the OF test, (2) blunted the effect of novelty to suppress feeding in the NSF test, (3) abolished acoustic LES, and (4) prevented the hypophagic effect of acute EOP stress. These results extend previous reports that alBST circuits contribute to LES and other anxiety-like behaviors in rats (Walker and Davis, 1997b; Merali et al., 2003; Sajdyk et al., 2008; Walker et al., 2009; Calhoon and Tye, 2015). The assays of anxiety-like behavior used in our study are ethologically based on spontaneous innate responses of rats (i.e., avoidance, startle, hypophagia) to “anxiogenic” environmental stimuli such as illumination, novelty, and open spaces, which do not explicitly evoke pain or discomfort. These assays do not depend on learning and are thought to induce a state of uncertainty about potential danger that places the animal into a state of sustained apprehension, perhaps similar to anxiety in humans (Davis et al., 2010). For example, in the acoustic startle assay, unexpected noise bursts elicit reflexive startle responses both in humans and laboratory rats (Walker and Davis, 1997a). In diurnal humans, darkness is anxiogenic and enhances startle amplitudes, whereas light enhances startle (i.e., LES) in nocturnal rats. Hypervigilance and arousal are reflected by elevations in the acoustic startle response, which is a core feature of human anxiety (Grillon, 2002). Since the activation of BST GABAergic neurons promotes behavioral arousal in mice (Kodani et al., 2017), GLP1R knockdown in GABAergic BST neurons may attenuate stress-related arousal, perhaps contributing to the lack of LES in GLP1R-KD rats.
It was surprising that GLP1R knockdown attenuated behavioral stress/anxiety-like responses in the four assays listed above (i.e., OF, NSF, LES, and stress-induced hypophagia), but had no discernable effect on EPMZ behavior. The EPMZ is a conflict paradigm that is widely used to assess anxiety-like behavior in rodents (Walf and Frye, 2007). An earlier study (Kinzig et al., 2003) reported anxiolytic effects in the EPMZ in rats after intracerebroventricular administration of a GLP1R antagonist, and anxiogenesis after pharmacological stimulation of GLP1R specifically within the CEA. Since the CEA and alBST are highly interconnected, and share many common inputs and outputs (Bienkowski and Rinaman, 2013), we expected GLP1R knockdown within the alBST to reduce anxiety-like behavior in the EPMZ. Instead, the collective evidence suggests that GLP1R signaling within the CEA selectively modulates anxiety-like behavior in this test. We also considered the possibility that the prior exposure of rats to other behavioral tests (i.e., OF, NSF) affected subsequent EPMZ outcomes in our first two experimental cohorts (Table 1), but there also were no EPMZ behavioral differences in our final two cohorts, in which the EPMZ test was conducted first. The differential effect of GLP1R knockdown across tests within subjects is evidence that GLP1R signaling within the alBST exerts a differential influence on behaviors measured in these assays. These results reinforce the value of using more than one behavioral test when seeking to resolve circuits that underlie anxiety-like behavior, for which different features and components likely have different underlying neural substrates (Bourin, 2015; Calhoon and Tye, 2015; Lezak et al., 2017).
In rats, GLP1 neurons are glutamatergic (Zheng et al., 2015b), which is consistent with an overall excitatory/stimulatory effect of GLP1R signaling within the CNS (Susini et al., 2000; Perry and Greig, 2003; Acuna-Goycolea and van den Pol, 2004; Hayes, 2012). GLP1 enhances both excitatory and inhibitory synaptic inputs to brainstem and hypothalamic neurons through presynaptic stimulation of transmitter release (Acuna-Goycolea and van den Pol, 2004; Wan et al., 2007), and also has direct postsynaptic effects on neuronal activity in rats (Acuna-Goycolea and van den Pol, 2004; Riediger et al., 2010). Conversely, in mice, GLP1 has been reported to excite or to inhibit equivalent proportions of BST neurons identified as expressing GLP1R (Williams et al., 2018), and another electrophysiological study using mice reported inhibitory effects of GLP1R signaling within the thalamus (Ong et al., 2017). In addition to possible species differences, the effect of GLP1 signaling on different populations of GLP1R-expressing BST neurons may differ based on their phenotypic profile, including differential connectivity and neurochemical identities. For example, only a subset of alBST neurons identified as expressing GLP1R mRNA in the present study were observed to also express CRH mRNA, and these were observed only within the dorsal alBST, identifying them as a phenotypically distinct subpopulation (Dabrowska et al., 2013).
A recent publication in which hindbrain GLP1 neurons were chemogenetically activated in mice reported a significant effect to reduce food intake, but no effect on assessed anxiety-like behaviors, and no effect on plasma cort levels (Gaykema et al., 2017). However, it is unclear whether chemogenetic stimulation in that study was sufficient to induce “enough” presynaptic release of GLP1 to increase anxiety-like behavior and activate the hypothalamic–pituitary–adrenal (HPA) stress axis. In another study (Ghosal et al., 2017), targeted deletion of GLP1R within the hypothalamic PVN in mice was sufficient to suppress autonomic and hormonal responses to acute stress, and also attenuated anxiety-like behavior and loss of body weight after chronic stress. Thus, hypothalamic GLP1R signaling appears necessary for both behavioral and physiological stress responses in mice. It will be important to determine whether these apparently discrepant results in mice reflect differences in the amount or central location of GLP1 signaling required to affect food intake versus anxiety behavior and stress responses, and whether stimulation of GLP1R within the hypothalamus, BST, and other brain regions is necessary versus sufficient to evoke these responses. Future studies also should determine whether differences exist between rats and mice in the physiological and behavioral effects of central GLP1R signaling.
Finally, in the present study, GLP1R knockdown within the alBST did not alter baseline cort levels, but significantly prolonged the plasma cort response to acute stress. We interpret these results as evidence for GLP1R-mediated activation of GABAergic neurons within the alBST that project to the hypothalamic PVN to put a “brake” on stress-induced activation of the HPA axis, and that GLP1R knockdown in the alBST effectively releases this brake to promote a larger, more prolonged endocrine response to stress. This interpretation is partly based on our preliminary studies using a rat ex vivo slice preparation, in which GLP1R agonists depolarized neurons within the ventral alBST that project to the PVN (Povysheva et al., 2016), and also on evidence that the projection pathway from ventral alBST to PVN is GABAergic and serves to restrain stress-induced activation of the HPA axis (Radley et al., 2009; Johnson et al., 2016). Although baseline (i.e., prestress) cort levels did not differ between GLP1R-KD and CTRL rats, it is possible that GLP1R-KD rats in our study had more pronounced cort responses to behavioral testing and day-to-day disruptions (e.g., body weight assessments), such that they experienced more total exposure to circulating cort than CTRL rats, despite their generally “less anxious” phenotype. Published studies focused on projection pathways arising from the alBST (Johnson et al., 2016) and on GLP1R signaling in other stress-related circuits that also have documented dissociations between behavioral and endocrine responses to acute stress, including differential effects of manipulating GLP1R signaling within different forebrain regions (Kinzig et al., 2003; Helmreich et al., 2012; Ghosal et al., 2017). Such studies support the dissociability of central neural circuits controlling physiological and behavioral responses to stress-inducing stimuli and treatments.
The novel results reported here have potential translational value. In rats, alBST stimulation is potently anxiogenic (Herman et al., 2005; Waddell et al., 2006; Choi et al., 2007, 2008), and the activation of BST outflow is thought to model many aspects of generalized anxiety disorder in humans (Shekhar et al., 2001; Walker et al., 2003; Sajdyk et al., 2008). GLP1R mRNA and protein are expressed in the human forebrain (Alvarez et al., 2005), and GLP1-positive neurons occupy the human cNTS and medullary reticular formation (Zheng et al., 2015a), similar to rodents. Future studies can determine whether GLP1R also is expressed by GABAergic and CRH-expressing neurons within the human alBST, as in rats, and whether this represents a potential target to improve therapies to treat anxiety and other stress-related clinical disorders.
Footnotes
This research was supported by National Institutes of Health Grants MH-059911 (to L.R.), DK-115762 (to M.R.H.), and F31 award DK-105858 (to D.J.R.).
The authors declare no competing financial interests.
- Correspondence should be addressed to Linda Rinaman at Rinaman{at}psy.fsu.edu
References
In this issue
