Regulation of food intake by astrocytes in the brainstem dorsal vagal complex.

A role for glial cells in brain circuits controlling feeding has begun to be identified with hypothalamic astrocyte signaling implicated in regulating energy homeostasis. The nucleus of the solitary tract (NTS), within the brainstem dorsal vagal complex (DVC), integrates vagal afferent information from the viscera and plays a role in regulating food intake. We hypothesized that astrocytes in this nucleus respond to, and influence, food intake. Mice fed high-fat chow for 12 hr during the dark phase showed NTS astrocyte activation, reflected in an increase in the number (65%) and morphological complexity of glial-fibrillary acidic protein (GFAP)-immunoreactive cells adjacent to the area postrema (AP), compared to control chow fed mice. To measure the impact of astrocyte activation on food intake, we delivered designer receptors exclusively activated by designer drugs (DREADDs) to DVC astrocytes (encompassing NTS, AP, and dorsal motor nucleus of the vagus) using an adeno-associated viral (AAV) vector (AAV-GFAP-hM3Dq_mCherry). Chemogenetic activation with clozapine-N-oxide (0.3 mg/kg) produced in greater morphological complexity in astrocytes and reduced dark-phase feeding by 84% at 4 hr postinjection compared with vehicle treatment. hM3Dq-activation of DVC astrocytes also reduced refeeding after an overnight fast (71% lower, 4 hr postinjection) when compared to AAV-GFAP-mCherry expressing control mice. DREADD-mediated astrocyte activation did not impact locomotion. hM3Dq activation of DVC astrocytes induced c-FOS in neighboring neuronal feeding circuits (including in the parabrachial nucleus). This indicates that NTS astrocytes respond to acute nutritional excess, are involved in the integration of peripheral satiety signals, and can reduce food intake when activated.

measure the impact of astrocyte activation on food intake, we delivered designer receptors exclusively activated by designer drugs (DREADDs) to DVC astrocytes (encompassing NTS, AP, and dorsal motor nucleus of the vagus) using an adenoassociated viral (AAV) vector (AAV-GFAP-hM3Dq_mCherry). Chemogenetic activation with clozapine-N-oxide (0.3 mg/kg) produced in greater morphological complexity in astrocytes and reduced dark-phase feeding by 84% at 4 hr postinjection compared with vehicle treatment. hM3Dq-activation of DVC astrocytes also reduced refeeding after an overnight fast (71% lower, 4 hr postinjection) when compared to AAV-GFAP-mCherry expressing control mice. DREADD-mediated astrocyte activation did not impact locomotion. hM3Dq activation of DVC astrocytes induced c-FOS in neighboring neuronal feeding circuits (including in the parabrachial nucleus). This indicates that NTS astrocytes respond to acute nutritional excess, are involved in the integration of peripheral satiety signals, and can reduce food intake when activated.

| INTRODUCTION
Food intake is controlled by the coordinated action of numerous brain regions but a complete understanding remains elusive (Andermann & Lowell, 2017). The brainstem dorsal vagal complex (DVC) is the first site for integration of visceral synaptic and hormonal cues that act to inhibit food intake (Grill & Hayes, 2012). The DVC consists of three nuclei: the nucleus of the solitary tract (NTS), area postrema (AP), and dorsal motor nucleus of the vagus. Neurons of the sensory branch of the vagus nerve make glutamatergic synapses onto NTS neurons to relay information from the periphery, including the stomach and upper intestine (Doyle & Andresen, 2001;Williams et al., 2016). Targeted chemogenetic activation of appetite-responsive NTS neuronal populations causes short term decreases in food intake (Cerritelli, Hirschberg, Hill, Balthasar, & Pickering, 2016;D'Agostino et al., 2016;Gaykema et al., 2017;Roman, Derkach, & Palmiter, 2016;Zhan et al., 2013).
Given the similarities between the NTS and the ARC, namely that both nuclei contain neurons that exert powerful effects on feeding behavior and are adjacent to circumventricular organs, we hypothesized that NTS astrocytes may be involved in mediating satiety. In support of this hypothesis NTS astrocytes sense synaptic input from the vagus nerve, which drives intracellular Ca 2+ increases in ex vivo brain slices (McDougal, Hermann, & Rogers, 2011). Furthermore, NTS astrocytes are implicated in the appetite suppressing effects of the glucagon-like peptide 1 (GLP-1) receptor agonist exendin-4 (Reiner et al., 2016). However, what remains unknown is whether NTS astrocytes respond to physiological perturbations in energy homeostasis.
Furthermore, there has not been a causal demonstration of the impact of NTS astrocyte activation on physiological feeding behavior. To address these key questions, we used a 12-hr high-fat feeding paradigm to trigger a feeding binge and examined morphological changes in NTS astrocytes. We then targeted the expression of excitatory designer receptors exclusively activated by designer drugs (DREADDs) to DVC astrocytes to allow specific chemogenetic activation of these cells and determined the impact on feeding behavior in mice.

| Dark-phase high-fat feeding studies and histology
Two independent cohorts of mice (aged 16 weeks) were individually housed for 4-5 days. At lights-off on the test day, in the experimental group, standard diet was substituted for high-fat chow (D12492, TestDiet) while control mice remained on standard chow (EURodent diet). Then, 12-14 hr later mice were euthanized (with sodium pentobarbital) and transcardially perfused with heparinized 0.9% saline then 4% paraformaldehyde. Brains were postfixed in 4% paraformaldehyde for 4-6 hr before being transferred to 30% sucrose in phosphate buffered saline (PBS). Then, 30 μm coronal sections of the brainstem were taken with a freezing sledge microtome (Bright instruments, UK) in four series. One series was stained with mouse anti-GFAP (MAB360; Millipore, UK) followed by donkey anti-mouse Alexa Fluor 568 (A10037; Invitrogen, UK) which allowed for GFAPimmunoreactivity to be visualized (Buckman, Thompson, Moreno, & Ellacott, 2013). Omission of the primary antibody was used to confirm that the immunoreactivity seen was not due to nonspecific binding of the secondary antibody to the tissue (Supplementary Figure S1a). Stained sections were mounted onto glass slides and coverslipped with fluoroshield mounting medium with DAPI (ab104139; Abcam, UK). Images for cell counting were acquired at 10× magnification on a fluorescence microscope (DM 4000; Leica, Germany). The NTS was subdivided (Figure 1e) as follows: rostral = Bregma −7.08 to −7.2 mm, postremal = Bregma −7.32 to −7.64 mm, caudal = Bregma −7.76 to −8 mm (Paxinos & Franklin, 2001). Images for morphological analysis were acquired at 20× magnification on a confocal microscope (DMi8; Leica). Images were analyzed in FIJI (Schindelin et al., 2012) (National Institutes of Health) using the cell counter, simple neurite tracer, and Sholl analysis plugins (Ferriera et al., 2014). For tracing, five cells per z-stack from two z-stacks per mouse were randomly selected for tracing and morphological analysis. The investigator was blinded to the diet of the mice during immunohistochemical staining, image acquisition, and analysis.

| NTS viral vector injection
Mice (8-12 weeks) were injected with adeno-associated viral (AAV) vectors in the NTS as described previously (Cerritelli et al., 2016). The vectors used were AAV5/2-hGFAP-hM3Dq_mCherry (v97-5, titer ≥7.5 × 10 12 viral genomes/ml; University of Zurich Viral Vector Facility, Switzerland) and AAV5/2-hGFAP-mCherry (custom preparation, titer 3.89 × 10 13 viral genomes/ml; ViGene Biosciences) diluted 1:1 in sterile PBS prior to injection. In brief, mice were deeply anaesthetized with an intraperitoneal (i.p.) injection of ketamine (70 mg/kg) and medetomidine (0.5 mg/kg) and placed in a stereotaxic frame (David Kopf Instruments) with the nose angled down by 20 . An incision from the occiput to the nape of the neck was made, the muscles were parted, and the atlantooccipital membrane incised to expose the surface of the brainstem. Injections were made from a pulled glass pipette attached to an injection system (Neurostar, Germany) mounted on the stereotaxic frame at an angle of 35 to the vertical, tip facing rostral.
The pipette was inserted 400 μm lateral to the midline at the level of calamus scriptorius to a depth of 1 mm. Four AAV injections of 180 nl were made bilaterally at angled depths of 1 mm, 750 μm, 500 μm, and 250 μm, respectively, at a rate of 100 nl/min. The pipette was left in place for 1 min after each injection. A second, independent cohort of mice underwent the same procedure but with a single 180 nl viral injection per side at 200 μm lateral to the midline at a depth of 500 μm (Supplementary Figure S5). Mice had atipamezole (1 mg/kg i.p.) and buprenorphine (0.1 mg/kg subcutaneous) for anesthetic reversal and analgesia, respectively, and were transferred to a heated cage to recover. Following surgery mice were individually housed for the duration of the experiment. The investigator was blinded to mouse group allocations for all the subsequent experiments.

| Feeding assays
At 4-8 weeks following surgery, mice were acclimatized to experimenter handling and i.p. injections of 0.9% NaCl (saline) daily for 4 days. For dark-phase feeding studies, mice were given an i.p. injection of saline 15-30 min prior to the beginning of the dark phase. Food intake was manually measured by the investigator (under red-light illumination where necessary) at 2, 4, 6, 12, and 24 hr after lights-off. Body weight was measured 8 hr after lights-off. The following day mice were given an injection of clozapine-N-oxide (CNO, 0.3 mg/kg i.p., Tocris, UK; diluted in saline) and food intake and body weight were measured at the same time points. For fast-induced refeeding experiments, food was removed from cages at the beginning of the dark phase. Subsequently, 15-30 min prior to the onset of light-phase (11.5-11.75 hr later), mice received an injection of CNO (0.3 mg/kg i.p.). After injection, food was returned to cages at the onset of the light phase and intake was measured 1, 2, 4, 6, 8, 12, and 24 hr after lights-on. Body weight was measured twice: once prior to food being removed and once at 8 hr after the reintroduction of food.

| Conditioned place aversion assay
For conditioned place aversion (CPA) testing, an apparatus consisting of two distinct chambers joined by a clear plastic external corridor (see Figure 4b) was used. The left chamber had horizontal black and white stripes on the walls and a perforated floor, and the right chamber had vertical black and white stripes on the walls and a floor with horizontal grating. A counterbalanced paradigm was used to test CPA (see Figure 4a). On the first day, mice were given 20 min of free access to explore the whole apparatus. This session was recorded using a video camera and used to determine initial location preferences. On the second day (Conditioning 1), mice were assigned to receive either CNO (0.3 mg/kg i.p.) or an equivalent volume of saline 15 min prior to being placed in either the left or right chamber for 45 min, with the access to the external corridor and second chamber blocked. On the third day (Conditioning 2), mice received the opposite treatment with their access restricted to the alternate chamber, compared to the second day. On the fourth/final day, mice were given free access to the whole apparatus for 30 min. Again, this session was recorded and used to determine conditioned preference. Recorded sessions were analyzed offline using Ethovision XT (Noldus, Netherlands) tracking software. An independent cohort of C57BL6/J mice (n = 7) underwent the same conditioning protocol with the known aversive agent lithium chloride (LiCl, 150 mg/kg i.p. diluted in saline) replacing CNO as the experimental stimulus.

| Home cage food seeking
Individually housed mice were injected with CNO (0.3 mg/kg i.p.) 15 min prior to the beginning of the dark phase and their home cage placed under a video camera with food removed from the hopper and two pieces of food placed on the cage floor in the far corner from the nest. The activity was then recorded for 3 hr and food intake at 3 hr was measured. Recorded sessions were analyzed offline using Ethovision XT tracking software with a 6 cm 2 zone centered around the pellets termed the "food zone."

| Validation of DREADD transduction of astrocytes and evaluation of DREADD-induced c-FOS expression
Mice received CNO (0.3 mg/kg i.p.) during the first 4 hr of the light phase and had food removed from the cage. After 2-3 hr, mice underwent transcardial perfusion fixation as described above. Then, 30 μm coronal sections were taken and stained with primary antibodies against GFAP, mCherry, NeuN, and c-FOS (Table 1) followed by the appropriate secondary antibodies: donkey anti-mouse Alexa Fluor488, donkey anti-goat Alexa Fluor594, and donkey anti-rabbit Alexa Fluor488 (Invitrogen). Omission of primary antibody was used to confirm that the immunoreactivity seen was not due to non-specific bind-

| Chemogenetic activation induced structural changes in NTS astrocytes
It is established that Gq-coupled DREADDs can be used to selectively activate astrocytes by driving increases in intracellular Ca 2+ (Adamsky et al., 2018;Bonder & McCarthy, 2014;Chen et al., 2016;Durkee et al., 2019;Martin-Fernandez et al., 2017). We bilaterally injected AAV vectors containing hM3Dq_mCherry (DVC::  Figure S3). This indicates that in DVC::GFAP hM3Dq mice CNO treatment was sufficient to induce changes in the structure of the DVC GFAP-immunoreactive astrocytes, indicative of activation.

| Chemogenetic activation of DVC astrocytes reduced food intake
Mice typically eat most of their food during the dark-phase.
Chemogenetic activation of the transduced astrocytes in DVC:: GFAP hM3Dq mice (CNO; 0.3.mg/kg, i.p. 15-30 min prior to onset of the dark-phase) produced an 84% reduction in food intake (4 hr after injection) compared to food consumption by the same mice following saline control treatment (Figure 3a). This anorectic effect lasted for approximately 6 hr after which the rate of food intake returned to that seen following saline treatment (Figure 3b). Body weight, measured 8 hr after lights-off, was reduced by 4.4% on the day of CNO injection compared with the day of saline injection ( Figure 3c).
In DVC::GFAP mCherry mice, the same CNO injection protocol had no effect on dark-phase food intake, feeding rate, or body weight (Figure 3d-f) intake and food seeking behavior were lower in DVC::GFAP hM3Dq mice compared with DVC::GFAP mCherry mice (Supplementary Figure S4a-c). This suggests that the reduced food intake is a result of suppressed drive to feed rather than reflecting a motor impairment of feeding, such as could be anticipated as a result of activation of astrocytes in the hypoglossal nucleus (HGN) disrupting tongue movement.
To evaluate whether chemogenetic activation of DVC astrocytes was sufficient to reduce feeding when there was an increased drive to eat, we utilized a fast-induced refeeding paradigm. CNO injection  Figure S5). These findings indicate that it is the glia within the DVC that mediate the suppression of feeding.

| Chemogenetic activation of DVC astrocytes did not alter locomotion
Reductions in food intake in mice can be indicative of malaise and/or aversion (Maniscalco & Rinaman, 2018). To test whether DVC F I G U R E 3 Legend on next page. astrocyte activation was aversive, we used a CPA assay (Figure 4a).
During initial assessment, prior to conditioning, neither DVC:: GFAP hM3Dq nor DVC::GFAP mCherry mice showed a preference for either chamber (Figure 4c,e). After conditioning (pairing each side of the apparatus with either saline or CNO injection [0.3 mg/kg i.p.]), mice again showed no preference or aversion (Figure 4b,c,e). However, following the same conditioning protocol using LiCl (150 mg/kg i.p.), a known aversive stimulus, an independent cohort of mice showed no avoidance of the LiCl chamber ( Figure 4g); as such, from our experiment, the effect of DVC astrocyte activation on conditioned aversion is unclear.
The conditioning protocol did however allow measurement of locomotion following chemogenetic DVC astrocyte activation. There was no statistically significant difference in the total distance travelled during the 45 min conditioning trial following saline or CNO treatment the DVC::GFAP hM3Dq or DVC::GFAP mCherry groups (Figure 4d,f). This suggests that CNO-mediated DREADD activation did not impact locomotion. This is in contrast to mice treated with LiCl which displayed a statistically significant reduction in total distance travelled, when compared with the saline conditioning trial (Figure 4h). This indicates that at the dose used LiCl is sufficient to acutely attenuate locomotion in mice, potentially indicative of malaise, while either chemogenetic activation of DVC astrocytes or CNO treatment in DVC::GFAP mCherry mice is not. However, in the PVH, another projection target of satiety signaling NTS neurons (D'Agostino et al., 2016;Roman, Sloat, & Palmiter, 2017), there was no difference in the number of c-FOS expressing cells between groups (Figure 5j-m). This is consistent with DVC astrocyte activation signaling to neuronal circuitry implicated in regulation of energy homeostasis (Atasoy, Betley, Su, & Sternson, 2012;Carter, Soden, Zweifel, & Palmiter, 2013).

| DISCUSSION
In this study, we have shown for the first time that astrocytes in the NTS react dynamically to excess intake of an energy dense food by upregulating GFAP expression and showing morphological plasticity.
Furthermore, DREADD-mediated activation of astrocytes in the DVC caused a potent but reversible decrease in food intake, associated changes in NTS astrocyte morphology, and activation of local and distal neuronal circuits. This evidence supports the hypothesis that DVC astrocytes are involved in a homeostatic or allostatic response to nutritional excess, and their activation may serve to drive a counteracting decrease in food intake to restore energy balance.
Astrocytes of the ARC have previously been implicated in the regulation of food intake (Buckman et al., 2015). In the hypothalamus, these glia are responsive to short-term energy imbalance, both fasting and acute high-fat feeding (Buckman et al., 2015;Fuente-Martín et al., 2012), and direct manipulation of their activity using DREADDs alters food intake Yang et al., 2015). Both studies using activating DREADDs in ARC astrocytes show a modest increase in food intake during the light phase, when mice are not typically eating Yang et al., 2015). In contrast, it appears from our observations that chemogenetic DVC astrocyte activation has more pronounced effects on feeding with respect to the magnitude, direction, and onset/duration of the response. The less pronounced impact of chemogenetic modulation of ARC astrocytes on feeding behavior in mice may be due, in part, to the fact that the ARC contains at least two neurochemically unique neuronal populations: AgRP and proopiomelanocortin (POMC) neurons; the activation of which drives and inhibits feeding behavior, respectively (Aponte, Atasoy, & Sternson, 2011;Krashes et al., 2011). As the two neuronal to date, neurons of the NTS have principally been implicated in satiety rather than hunger (Grill & Hayes, 2012), which may account for the more pronounced physiological effect consequent on local astrocytic activation.
Our study builds upon evidence showing that astrocytes of the NTS sense hormonal satiety signals (Reiner et al., 2016)  In this study, we used morphological complexity as a marker of altered astrocyte signaling to demonstrate that NTS astrocytes were responsive to both acute high-fat feeding and chemogenetic stimulation. A higher baseline of GFAP expression was observed in animals that had undergone AAV injection surgery. The morphological analysis of NTS astrocytes presented in this study encompassed the proximal processes of GFAP-expressing astrocytes. While this may represent only a small percentage (~10%) of the total astrocyte surface area (Shigetomi et al., 2013), there was a significant difference in the Sholl profile and number of processes between groups supporting a reactive change in astrocyte structure under these experimental conditions. Other strategies which label a greater proportion of the astrocyte cell volume (e.g., Golgi staining or membrane targeted green fluorescent protein [lck-GFP]) could be used for future studies but each has their own limitations: Golgi staining does not positively differentiate between astrocytes and neurons and sparsely labels cells, while lck-GFP (which labels the entire astrocyte surface area) yields a "bushy" signal (Shigetomi et al., 2013), that is not resolvable by standard confocal microscopy.
Critically, glutamatergic signaling is the principal mechanism for communication between the vagus and second-order NTS neurons (Doyle & Andresen, 2001) and NTS astrocytes directly sense vagal glutamate release via Ca 2+ -permeable AMPA receptors expressed on the cell membrane (McDougal et al., 2011). In the NTS, synaptic clearing of glutamate by astrocytes has been shown to restrain NTS neuronal firing and vagal outflow to cardiorespiratory organs (Matott et al., 2017). As such, astrocyte glutamate transport can manipulate the firing rates of NTS neurons and alter output from the NTS. In our study, it is unclear how (and indeed if) astrocyte Gq-GPCR activation may be linked to altered glutamate uptake; however, previous reports (Oliet, Piet, & Poulain, 2001) have shown that altered morphology of astrocytes results in corresponding changes in glutamate uptake and neuronal excitability. An additional potential mechanism of communication that could be important in mediating the effects seen is active gliotransmission (Araque et al., 2014). Activation of a GPCR expressed on NTS astrocytes (protease-activated receptors) leads to increased activation of neurons by glutamate, possibly exocytosed by astrocytes (Vance, Rogers, & Hermann, 2015). Given that antagonism of NMDA receptors in the NTS increases meal size (Treece, Covasa, Ritter, & Burns, 1998), it is possible that activation of these receptors by astrocyte-derived glutamate or D-serine would reduce food intake, but further study would be required to specifically test this.
Our data support the conclusions of an earlier study which showed that DVC astrocytes are critical for the appetite suppressive effect of GLP-1 receptor (GLP-1R) activation. In rats, NTS astrocytes take up exendin-4 (an agonist of the GLP-1R) and metabolic inhibition of astrocytes in the DVC using fluorocitrate abolishes the anorexigenic actions of this compound. Together, this suggests that astrocytes contribute to the suppressive effect of activation of GLP1-R signaling on feeding and supports their role in DVC-mediated satiety signaling (Reiner et al., 2016). What remains unclear is whether this is due to removal of GLP-1R-mediated astrocyte activation or alternatively astrocyte inhibition causing disruption to glutamatergic signaling stimulated by neuronal GLP-1R activation. While fluorocitrate has historically been widely used to inhibit glial cell activity, it would be beneficial to repeat these studies using genetically targeted approaches to modulate astrocyte activity. Whether GLP-1R is expressed in astrocytes is also not clear as GFP expression was not apparent in DVC astrocytes of GLP-1R-GFP mice (Cork et al., 2015). However, in vitro and in ex vivo rat brainstem slices astrocytes increase intracellular Ca 2+ in response to exendin-4 providing evidence that these cells are directly sensitive to GLP-1R ligands (Marina et al., 2017;Reiner et al., 2016).
It is likely that DVC astrocytes are important for integration of hormonal and nutritional cues. Hypothalamic astrocytes express receptors for ghrelin, insulin, and leptin (Fuente-Martín et al., 2016;García-Cáceres et al., 2016;Kim et al., 2014). In common with the ARC, the DVC contains a circumventricular site, the AP, making it likely that DVC astrocytes also share the same hormone sensing capabilities, although this remains to be conclusively demonstrated.
Reexpression of the glucose transporter GLUT2 in astrocytes is sufficient to restore brainstem hypoglycemia detection in GLUT2-knockout mice providing further evidence for a critical role of these cells in energy homeostasis (Marty et al., 2005).
Of note, three studies that inhibit inflammatory nuclear factor kappa b (NF-κB) signaling in all GFAP-expressing astrocytes (including work from our group) report feeding phenotypes, namely, increased initial intake of a high-fat diet following initial exposure, resistance to the obesity phenotype when already on a high-fat diet and protection from metabolic dysfunction and weight gain on a high fat diet (Buckman et al., 2015;Douglass, Dorfman, Fasnacht, Shaffer, & Thaler, 2017;Zhang et al., 2017). Although these studies attribute the observed effects to astrocytes of the hypothalamus the contribution of NTS astrocytes, identified here as responsive to high-fat diet intake, cannot be ruled out.
The induction of c-FOS-immunoreactivity in a distal target, namely, the lPBN, following CNO injection in DVC::GFAP hM3Dq mice strongly suggests the recruitment of long-range neuronal circuitry following DVC astrocyte activation. Since there are a number of distinct neuronal populations in the NTS whose activation induces a similar reduction in feeding, namely, POMC (Cerritelli et al., 2016;Zhan et al., 2013), cholecystokinin (CCK) (D'Agostino et al., 2016;Roman et al., 2016), tyrosine hydroxylase (Roman et al., 2016), and preproglucagon (Gaykema et al., 2017), it is likely that these are among the neurons recruited by chemogenetic DVC astrocyte activation.
Given the reduction in food intake we observed, it was important to evaluate whether induction of malaise and/or aversion could be contributing to the effect seen, particularly as the DVC is implicated in mediating these responses in addition to satiety (Maniscalco & Rinaman, 2018). This particularly important in light of the increased in c-FOS immunoreactivity observed in the lPBN of DVC::GFAP hM3Dq mice after CNO treatment, which may indicate the activation of a NTS-lPBN pathway previously reported to reduce food intake through aversion/negative salience (Roman et al., 2017). To do this, we used a CPA assay utilizing a protocol previously used by our group to indicate aversive responses/negative salience in response to chemogenetic activation of prefrontal cortex-projecting locus coeruleus neurons (LC PFC ) in rats (Hirschberg, Li, Randall, Kremer, & Pickering, 2017). We found that after one conditioning session per agent CNO treatment in DVC::GFAP hM3Dq or DVC::GFAP mCherry mice did not induce CPA. However, since the known aversive LiCl agent was unable to induce CPA in this paradigm it suggests our assay was likely not sufficiently sensitive to detect/induce CPA in this context. As such, the affective properties of DREADD-mediated DVC astrocyte activation remain unclear. A longer protocol may have revealed an effect since three or four conditioning sessions with LiCl does induce CPA in mice (Le Merrer et al., 2011;Longoni, Spina, Vinci, & Acquas, 2011;Sanjakdar et al., 2015;Zhang et al., 2019). We did, however, find that LiCl injection reduced locomotion during the conditioning trial while CNO treatment did not. This suggests LiCl acutely induces malaise; an effect that was not seen in the CNO-treated mice. Evidence suggests that satiety and aversion exist on a continuum, where in some cases aversion may be a manifestation of extreme satiety, for example, associated with excessive gastrointestinal distention (Maniscalco & Rinaman, 2018). As such, ability of an agent to induce CPA or conditioned taste aversion does not necessarily indicate that an effect is non-physiological. For example, exogenous CCK can induce satiety or aversion dependent on dose (Moran, 2000;Swerdlow, van der Kooy, Koob, & Wenger, 1983;West, Greenwood, Marshall, & Woods, 1987).
Our results suggest that that DREADD-mediated DVC astrocyte activation likely does not produce acute malaise, but its emotional salience remains unclear and further studies would be required to explore this.
In conclusion, our experiments show that astrocytes may be a previously overlooked element of DVC-mediated satiety and provide the first causal link between their activity and the regulation of feeding. We propose that NTS/DVC astrocytes are a key component in satiety signaling that act to modulate the integration of vagal and hormonal inputs to the brainstem and as such could represent a potential target for intervention in obesity.