Control of water intake by a pathway from the nucleus of the solitary tract to the paraventricular hypothalamic nucleus

Abstract

Several brain areas have been shown to participate in thirst and control of fluid intake. An understanding of how these circuits interact, and their roles in the activation, maintenance, and termination of fluid intake remains incomplete. Central glucagon-like peptide-1 (GLP-1) receptor activation appears to be an important part of the termination of drinking, but the site(s) of action for this suppression has not yet been determined. In an attempt to use GLP-1 responsiveness as a means to screen targets of hindbrain cells that participate in the termination of thirst and the resultant water intake, we injected the GLP-1 receptor agonist exendin-4 (Ex-4) into three brain areas known to express GLP-1 receptors, and measured subsequent water intake. Ex-4 reduced water consumption when injected into the paraventricular hypothalamic nucleus (PVH) and nucleus of the solitary tract (NTS), but not when injected into the nucleus accumbens (NAc). Using the effective response after injection into the PVH as a guide, we examined the connection between the NTS – the site of endogenous central GLP-1 production – and the PVH. Retrograde tracing combined with Fos immunohistochemistry suggested intake-induced activity in PVH-projecting NTS cells. To test the hypothesis that this pathway is important in the termination of drinking, we chemogenetically activated PVH-projecting hindbrain cells. Interestingly, activation of this population of cells increased water intake, calling into question the heterogeneity of the pathway with respect to the control of fluid intake. Taken together, we conclude that the PVH is a site of action for GLP-1 receptor activation in the inhibition of water intake, but suspect that endogenous GLP-1 in NTS-to-PVH projections may be counterbalanced by a parallel pathway that either activates or maintains already activated water intake.

1. Introduction

Fluid intake is controlled by several interconnected brain regions. The nature of these connections and how they influence the initiation, maintenance, and termination of drinking are not yet well-understood. This is particularly true for regions and pathways involved in drinking termination.

Glucagon-like peptide-1 (GLP-1) is a centrally-acting peptide that suppresses water intake (Larsen et al., 1997a; McKay et al., 2011; Navarro et al., 1996; Tang-Christensen et al., 1996; Wang et al., 1998), an effect which is not secondary to its effects on food intake (McKay et al., 2011). Although the nucleus of the solitary tract (NTS) is the main source of central GLP-1 (Jin et al., 1988; Larsen et al., 1997b; Merchenthaler et al., 1999), it is unknown where, precisely, in the brain GLP-1 is acting to suppress drinking. Accordingly, testing specific brain regions for fluid-intake-suppressive effects of GLP-1 may be a viable strategy for identifying pathways involved in drinking termination.

A greater understanding of the mechanisms that regulate drinking behavior is important for developing strategies to combat dehydration. Dehydration is surprisingly common in the United States: over half a million people are hospitalized due to dehydration every year (Xiao et al., 2004), and approximately one fifth of American adults have hypertonic plasma (Stookey, 2005). The elderly are especially susceptible to dehydration (Sansevero, 1997; Sheehy et al., 1999), and while young people with high plasma tonicity report drinking more water than those with normal tonicity, older adults do not compensate in a similar manner (Stookey, 2005), which is consistent with findings that older adults have a blunted thirst response (Mack et al., 1994). The adverse health outcomes associated with dehydration are almost certain to impact the well-being of individuals, and the economic consequences affect society as a whole. Almost two decades ago, the economic cost of avoidable hospitalizations due to dehydration in elderly Americans was estimated to be over one billion dollars (Xiao et al., 2004). As the population of North America is expected to continue aging (Lutz et al., 2008), these problems are likely to grow even larger over the coming years. Dehydration can be caused by a failure to drink in response to fluid need, but can also be caused or exacerbated by premature termination of ongoing drinking. The relative lack of knowledge of pathways involved in drinking termination and how these interact with drinking-promoting pathways is therefore concerning.

This report describes work that was presented at the 28th meeting of the Society for the Study of Ingestive Behavior. The overarching goal of the series of experiments that is described was to examine a pathway likely involved in the termination of drinking. Testing specific brain areas for fluid intake suppression after GLP-1 receptor (GLP-1R) agonist treatment was used as a starting point. There is strong evidence that GLP-1R activation suppresses fluid intake and GLP-1 in the brain is mainly produced in one part of the hindbrain, thereby providing an anchor for a pathway that could be further probed for specific relevance to fluid intake control. To this end, we injected the GLP-1R agonist exendin-4 (Ex-4) into the NTS, paraventricular hypothalamic nucleus (PVH) or nucleus accumbens (NAc) of rats and measured water intake over the subsequent 24 h. These areas were chosen because GLP-1R activation in these regions affects food intake (Alhadeff et al., 2012; Dossat et al., 2011; McMahon & Wellman, 1998; Richard et al., 2015), but the effect of GLP-1 in these regions on fluid intake has not yet been investigated. The results of the site-specific GLP-1R agonist injection experiments guided the remainder of the experiments, in which we examined how PVH-projecting hindbrain neurons control drinking. These experiments were conducted using the combination of retrograde tracing and Fos immunohistochemistry to test the hypothesis that cells that project from the NTS to the PVH are activated by drinking, more so than when an animal is simply dehydrated. We also used chemogenetic tools to activate cells in the NTS, or a specific population of cells in the hindbrain that project to the PVH, to test for effects on water intake.

2. Methods

2.1. Animals

Adult male Sprague-Dawley rats (200–349 g) were purchased from Envigo (Indianapolis, IN). Male rats were used because these experiments were designed to build upon previous work that only used male subjects. Rats were housed in a temperature- and humidity-controlled room, maintained on a 12:12 light:dark cycle, and given ad libitum access to standard rodent chow (Teklad 2018; Envigo, Madison, WI) and tap water except where noted. At least four days before behavioral testing, rats were moved from standard plastic cages to stainless steel wire-mesh cages (Unifab Corporation, Kalamazoo, MI). All experimental protocols were approved by the Institutional Animal Care and Use Committee of the State University of New York at Buffalo, and the handling and care of animals was in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

2.2. Surgeries

Rats in Experiments 1, 2, and 4 were implanted with a chronic indwelling cannula aimed at either the lateral ventricle (LV), fourth ventricle (4V), NTS, PVH, or NAc. All cannula were implanted unilaterally. Rats were anesthetized with an intramuscular (IM) injection of ketamine (70 mg/kg; Fort Dodge Animal Health, New York, NY) and xylazine (5 mg/kg; Lloyd Laboratories, Shenandoah, IA) before being secured in a stereotaxic apparatus and receiving a subcutaneous (SC) injection of carprofen (5 mg/kg; Pfizer Animal Health, New York, NY). A small burr hole was made in the skull and a guide cannula (26 gauge; Plastics One Inc., Roanoke, VA) was lowered to the coordinates listed in Table 1. All coordinates were based on a flat skull. The guide cannula was secured to the skull using bone screws and dental cement. LV cannula placement was verified by testing the drinking response to a LV injection of angiotensin II (AngII; 10 ng in 1 μl) a minimum of five days after surgery. Rats that drank a minimum of 6 ml of water were included in the experiments. 4V cannula placement was verified by testing blood glucose response to a 4V injection of 210 μg 5-thio-d-glucose in Experiments 1 and 2. Rats that had at least a 2-fold increase in blood glucose after 5-thio-d-glucose were included in subsequent testing. 4V cannula placement in Experiment 4, and placement of NTS, PVH, and NAc cannula in Experiment 2, were verified by postmortem ink injection.

Table 1.

Surgical coordinates.

Target	Medial-Lateral	Anterior-Posterior	Dorsal-Ventral
LV	1.4 mm lateral to midline	0.9 mm posterior to bregma	1.8 mm ventral to dura
4V	on midline	2.5 mm anterior to the occipital structure	4.8 mm ventral to dura, or 5.8 mm ventral to skull
NTS	0.5 mm lateral to midline	0.05 mm anterior to the posterior skull edge	6.0 mm ventral to skull
PVH	0.35 mm-0.4 mm lateral to midline	1.6 mm-1.8 mm posterior to bregma	6.35 mm-6.5 mm ventral to skull
NAc	1.5 mm lateral to midline	1.2 mm anterior to bregma	5.7 mm ventral to skull

In the same surgery as 4V cannula implantation, rats in Experiment 4 were injected with a virus expressing the excitatory DREADD [hM3D (Gq)] fused to mCherry under the hSyn promoter. The rats were divided into two subsets, one that was given bilateral injections into the PVH and a second that was given a single injection into the NTS. The subset of the rats given PVH injections received the DREADD packaged in a retrograde AAV (0.3 μl/side; #50474-AAVrg; Addgene, Watertown, MA), and the subset injected into the NTS received the DREADD packaged in AAV8 (0.3 μl; #50474-AAV8; Addgene, Watertown, MA). Rats were given a minimum of 4 weeks before behavioral testing. Virus injection location and expression of mCherry were evaluated by immunohistochemistry with antibodies against red fluorescent protein (RFP) as described in section 2.3. Rats in Experiment 3 were given a unilateral injection into the PVH of 0.3 μl of the retrograde tracer cholera toxin β subunit (CTβ; List Biological Laboratories, Campbell, CA).

2.3. Immunohistochemistry

In Experiment 3, brain sections were processed by immunohistochemistry for Fos and CTβ. Sections were washed in TBS, endogenous peroxidases blocked with 3% H₂O₂, and following another series of TBS washes, sections were incubated with rabbit anti-c-Fos (1:500; Cell Signaling Technologies, Danvers, MA) and goat anti-CTβ (1:50,000; List Biological Laboratories, Campbell, CA) for 1 h at room temperature and overnight at 4 °C. The next day, sections were washed in TBS, and then incubated for 2 h at room temperature with biotin-SP-conjugated donkey anti-rabbit IgG (1:1000; Jackson ImmunoResearch Laboratories, West Grove, PA). After TBS washes, the sections were incubated for 1 h at room temperature with ABC (1:111; Vectastain Elite ABC Kit, Vector Laboratories, Burlingame, CA). The ABC was washed away with TBS, and the Fos was visualized with DAB (Sigma Aldrich, St Louis, MO) with nickel enhancement. The sections were further washed in TBS and underwent another H₂O₂ wash, before being incubated in biotin-SP-conjugated donkey anti-goat IgG (1:1000; Jackson ImmunoResearch Laboratories, West Grove, PA), and CTβ was visualized using half the concentration of DAB as was used in the first reaction, with the nickel omitted.

In Experiment 4, brain sections were processed for RFP. The procedure was identical to that in Experiment 3 except for the antibodies used. The primary antibody was rabbit anti-RFP (1:2000; Rockland Immunochemicals, Limerick, PA), and the secondary antibody was biotin-SP-conjugated donkey anti-rabbit IgG (1:1000; Jackson ImmunoResearch Laboratories, West Grove, PA).

2.4. Drug injections and intake measures

Injections were made with a 33-gauge injection cannula that was connected to a 2 μl Hamilton syringe via flexible PE-50 tubing. Injections were made manually and injection cannulae were held in place for approximately 30 s after each ventricular injection, and for 1 min after each parenchymal injection. Care was taken to observe movement of an air bubble in the tubing to ensure that the parenchymal injections were successful. Volumes were 1 μl for injections into the ventricle and 500 μl for parenchymal injections. Fluid bottles were weighed immediately before and after testing periods. Total fluid intake was calculated by taking the difference of the pre- and post-bottle weights. Drinking behavior was continuously measured throughout the testing phase by recording licks with a contact lickometer (designed and constructed by the Psychology Electronics Shop, University of Pennsylvania, Philadelphia, PA; for a more thorough description please see (McKay & Daniels, 2013; McKay et al., 2011; McKay et al., 2014)).

2.5. Experimental design

2.5.1. Experiment 1: dose response of Ex-4 injection into the LV or 4V on 24 h water intake

Approximately 30 min before lights-off, food was removed and rats (n = 8–14 per group in a between-subjects design) were injected with vehicle (0.9% NaCl) or Ex-4 (50 ng, 25 ng, 15 ng, 5 ng, 2.5 ng, or 1 ng; American Peptide, Sunnyvale, CA) into the LV. Licks for water were recorded for 24 h using a contact lickometer. Food remained unavailable for the entire 24-h test.

In a separate set of rats, injections were made into the 4V to determine the dose of Ex-4 needed to suppress water intake. Approximately 30 min before lights-off, food was removed and rats (n = 9 in a repeated measures design) were injected with vehicle (0.9% NaCl) or Ex-4 (50 ng, 25 ng, 15 ng, or 5 ng; American Peptide, Sunnyvale, CA) into the 4V. A partial Latin squares design was used with at least 4 d between each dose. A contact lickometer was used to record the licks for water for 24 h after each injection. Food remained unavailable for the entire 24-h test.

2.5.2. Experiment 2: Effect of Ex-4 administered directly into the NTS, PVH, or NAc on 24 h water intake

Approximately 30 min before lights-off, food was removed and rats were injected with Ex-4 or vehicle (0.9% NaCl). Between-subjects designs were used for injections made into the NTS (n = 8–9 per group) or NAc (n = 7 per group) and a repeated measures design was used for injections into the PVH (n = 8). The doses of Ex-4 were based on the results from Experiment 1 and were selected to be below and near the threshold dose needed to reduce intake after injection into the 4V (NTS) or LV (PVH and NAc). Injections into the NTS were made using a single dose (15 ng) of Ex-4, whereas two doses (2.5 or 5 ng) were used for injections into the NAc and PVH. Licks for water were recorded for 24 h after injection. Food was unavailable for the entire 24-h test. One vehicle-injected rat was excluded as an outlier from NTS, and one 2.5 ng-injected rat was excluded as an outlier from the NAc (both more than 2 standard deviations above the mean).

2.5.3. Experiment 3: response of NTS neurons

Eight days after unilateral injection of 0.3 μl of the retrograde tracer CTβ into the PVH, rats were water deprived for 24 h, then either allowed to drink water, or not allowed to drink. Two hours after bottles were returned to some rats (26 h after water was removed), rats in both groups were anesthetized (70 mg/kg ketamine and 5 mg/kg xylazine, IM) and transcardially perfused with 150 ml heparinized isotonic saline followed by 450 ml 4% paraformaldehyde. All drinking and perfusions took place in the first half of the light phase. Brains were removed and post-fixed overnight in 4% paraformaldehyde before being transferred to 20% sucrose for a minimum of 2 days before sectioning on a freezing microtome into 40 μm coronal slices. Sections were stored at −20 °C in cryoprotectant until processed by immunohistochemistry for Fos.

The number of Fos-positive cells were counted in the NTS, from obex at the caudal end to just rostral of area postrema (from approximately bregma 13.56 mm to 14.52 mm). Every third section in this range was used (120 μm between sections). On average, the analysis included 10.5 sections per animal. The number of Fos-positive cells per section was counted using automated item counts in NIS-Elements (Nikon Instruments Inc, Melville NY) with object detection parameters that were adjusted to match manual counts of Fos-positive cells in a subset of sections before being applied to the remaining images of tissue sections. The condition of the rats was not revealed until after the counts were completed. Fos-positive cells were analyzed between subjects for average number per section in the NTS as a whole, as well as by different levels within the NTS as a repeated measure. For analysis of the NTS by level, we used sections that were just rostral to area postrema (“pre-AP”), sections in which area postrema was visible (“AP”), and sections that were caudal to area postrema but no more caudal than obex (“post-AP”). One subject was excluded as an outlier (more than 2 standard deviations above the group mean). The Fos counts included 10 rats that were given access to water after deprivation and 7 rats that were not allowed to drink after deprivation. One of the rats in the group that was allowed to drink had a misplaced CTβ injection, and was therefore excluded from the analysis of double-labeling.

We chemogenetically activated cells in the NTS to test for an effect on drinking behavior. Rats had excitatory DREADD [AAV8-hSyn-hM3D (Gq)-mCherry] injected into the NTS at least 4 weeks prior. Rats were water deprived overnight and tested approximately 1 h after lights-on. Rats were given a 1 μl 9 mM 4V injection of clozapine-N-oxide (CNO; Tocris, Bristol, UK) or vehicle (32% DMSO in 0.9% saline) in a repeated-measures design with at least 3 d between testing and the next water deprivation. Twenty minutes after injection of vehicle or CNO, water bottles were returned, and drinking behavior was recorded for the subsequent 2 h. Food was removed before injections and returned after testing ended. Only rats with correct 4V cannula placement and virus injection were used for analysis (n = 7).

2.5.4. Experiment 4: response of PVH-projecting hindbrain neurons

Brains from the same rats that were used for Fos counts in Experiment 3 were also processed by immunohistochemistry for Fos and CTβ. The number of CTβ-positive cells, and double-labeled Fos and CTβ cells, were counted. Correct placement of CTβ injection was determined by CTβ staining in the PVH, and only individuals with accurate PVH injections were used for analyses of CTβ. The number of CTβ and number of double-labeled cells were counted manually.

We chemogenetically activated PVH-projecting hindbrain cells to test for an effect on drinking behavior. The rats had excitatory DREADD in a retrograde AAV [AAVrg-hSyn-hM3D(Gq)-mCherry] injected bilaterally into the PVH at least 4 weeks prior. Rats were water deprived overnight and tested approximately 1 h after lights-on. Rats were given a 1 μl 18 mM 4V injection of CNO or vehicle (32% DMSO in 0.9% saline) in a repeated-measures design with at least 3 d between testing and the next water deprivation. Twenty minutes after injection of vehicle or CNO, water bottles were returned, and drinking behavior was recorded for the subsequent 2 h. Food was removed before injections and returned after testing ended. Only rats with correct 4V cannula placement and virus injection were used for analysis (n = 6 for PVH).

2.6. Statistical analyses and data presentation

Data are represented with bars showing means and circles showing individual data points, with lines connecting points in experiments using repeated measures for drug dose; within-subject factor of time is not indicated by lines. Statistical tests were performed using Statistica (StatSoft, Tulsa, OK). Two-way analysis of variance (ANOVA), one-way ANOVA, repeated measures ANOVA, or student’s t-tests were used as appropriate. Statistically significant effects (p < 0.05) were further analyzed using Newman Keuls post-hoc tests. Cumulative licks are shown for Experiments 2 and 4 to provide more insight into the time-course of drinking, but statistical analyses were not conducted on cumulative intake. Images of brain sections were processed to balance brightness and contrast using Adobe Photoshop (Adobe Inc, San Jose, CA). Brightness and contrast settings were applied to entire images and were not used otherwise to alter the appearance of the immunoreactivity.

3. Results

3.1. Experiment 1: dose response of Ex-4 injection into the LV or 4V on 24 h water intake

Experiment 1 was designed to determine a dose of Ex-4 that did not affect fluid intake when injected into the ventricles, thereby allowing effects of parenchymal Ex-4 injections in subsequent experiments to be attributed to site-specific action on GLP-1R rather than the result of efflux into the ventricles. To this end, water intake was measured for 24 h after an injection of vehicle or Ex-4 into either the LV (Fig. 1A; 50 ng, 25 ng, 15 ng, 5 ng, 2.5 ng, or 1 ng) or the 4V (Fig. 1B; 50 ng, 25 ng, 15 ng, or 5 ng). Licks for water were combined into dark phase (initial 12 h after drug injection) and light phase intake (final 12 h after drug injection). Statistical analysis showed a Drug by Time interaction for LV injections (F_{6, 61} = 3.78, p = 0.003), with post-hoc tests revealing that the interaction was driven by the response to 5, 15, and 50 ng Ex-4. The 25 ng Ex-4 group was not different from either vehicle or 5, 15, or 50 ng Ex-4. No differences in intake were detected after treatment with less than 5 ng Ex-4. There were no differences in licks for water during the light phase in any dose group. Ex-4 into the 4V also suppressed intake, but at higher doses than were effective in the LV. There was a Drug by Time interaction (F_{4, 32} = 6.7, p = 0.0005), and post-hoc tests found that it was driven by differences in intake after injection of either 25 or 50 ng of Ex-4. Injections of lower doses did not produce detectable differences in fluid intake.

Fig. 1.

Dose response of Ex-4 injection into the LV (A) or 4V (B) on 24 h water intake. Ex-4 injected into the LV at dark onset suppressed water intake during the dark phase when 5, 15, or 50 ng were injected in the LV, but did not suppress intake at any dose during the following light phase. When injected into the 4V at dark onset, 25 ng and 50 ng Ex-4 suppressed fluid intake during the dark phase, with no differences in intake detected during the following light phase. Bars with different letters are different from one another. Bars with two letters are not different from either letter.

3.2. Experiment 2: Effect of Ex-4 administered directly into the NTS, PVH, or NAc on 24 h water intake

Using doses of Ex-4 that were below or near the threshold for suppression of water intake after injection into a nearby ventricle, we made direct injections of GLP-1R agonist into the NTS, PVH, and NAc to test for anatomical specificity of the effect of GLP-1R in suppression of fluid intake. Rats injected with Ex-4 into the NTS drank less than vehicle-injected controls (Fig. 2A). This difference was detected by ANOVA as a main effect of Drug when data were analyzed in 12 h bins (Fig. 2B; F_{1, 14} = 5.72, p = 0.03). Administration of Ex-4 into the PVH suppressed intake similarly to what was observed after injections into the NTS (Fig. 2C) and there was a Drug by Time interaction when licks were grouped into 12 h bins (Fig. 2D; F_{2, 14} = 6.1, p = 0.01) with post-hoc analysis revealing that rats that received an injection of either 2.5 or 5.0 ng Ex-4 consumed less water 0–12 h after injection. In contrast to the effects observed after direct injections into the NTS or PVH, Ex-4 injected into the NAc did not appear to affect 24-h water intake (Fig. 2E) and we found no significant effect of Ex-4 administration into the NAc on licks for water in either the first or last 12 h of the test (Fig. 2F; F_2, 17 = 2.35, p = 0.13). This was true for the dose that was below threshold for an effect after injection into the lateral ventricle, and for the dose that was effective after injection into the lateral ventricle.

Fig. 2.

Effect of Ex-4 administered directly into the NTS (A, B), PVH (C, D), or NAc (E, F) on 24 h water intake. When injected into the NTS, 15 ng Ex-4 suppressed fluid intake over the next 24 h. Injected in the PVH, both 2.5 ng and 5 ng Ex-4 suppressed intake in the first 12 h after injection, with no differences in the next 12 h. We found no differences in fluid intake in rats given injections of Ex-4 in the NAc, though there was a significant main effect of time. Bars with different letters are different from one another. Asterisks show a significant main effect.

3.3. Experiment 3: response of NTS neurons

We examined the activity of NTS cells after water deprivation and rehydration. Analysis of the number of Fos-positive cells in the NTS (Fig. 3A) revealed that water-deprived rats allowed to drink had more active cells than those not allowed to drink (Fig. 3B; t = 2.26, df = 15, p = 0.04). When we compared the average number of Fos-positive cells in different parts of the NTS (Fig. 3C), we found significant main effects of Condition (F_{1, 15} = 5.59, p = 0.03) and Location (F_{2, 30} = 9.78, p = 0.0005), and a significant Condition by Location interaction (F_{2, 30} = 3.99, p = 0.03). A post-hoc test on the interaction indicated that there were fewer Fos-positive cells in the post-AP portion of the NTS than the AP or pre-AP portions in water-deprived rats allowed to drink. Moreover, there were more Fos-positive cells in the pre-AP portion of the NTS of water-deprived rats that were subsequently allowed to drink than in the post-AP of the NTS in both conditions. Perhaps because of power limitations, this mixed design ANOVA did not find what appeared be obvious differences between experimental conditions within specific parts of the brain. Indeed, t-tests on individual thirds of the NTS found statistically significant differences in number of Fos-positive cells between rats allowed to drink or not in the pre-AP (t = 2.34, df = 15, p = 0.03), AP (t = 2.21, df = 15, p = 0.04), and post-AP (t = 2.55, df = 15, p = 0.02) levels of the NTS. We also found a significant correlation between average number of Fos-positive cells in the NTS and the amount of water consumed by water-deprived rats allowed to drink (Fig. 3D; r² = 0.58, p = 0.01). Representative photos of Fos in the NTS at the level of the area postrema can be seen in Fig. 3E and F.

Fig. 3.

Response of NTS neurons. We quantified the number of Fos-positive cells in the NTS (A) and found that water-deprived rats allowed to drink had more active cells than those not allowed to drink (B). This appeared to be true at all levels of NTS (C), though see text for caveats. Within the group of water-deprived rats allowed to drink, there was a correlation between the number of active cells in the NTS and the amount of water consumed (D). Representative micrographs of the NTS at the level of AP of water-deprived rats not allowed to drink (E) and allowed to drink (F). Panel G shows the experimental strategy that used injections of an excitatory DREADD into the dorsal hindbrain. A representative micrograph of the injection site and subsequent spread of virus in and around the NTS at the level of the AP is shown in H. When CNO was injected in the 4V of rats with excitatory DREADD expressed in the NTS, we found no differences in number of licks for water (I, J), burst number (K), or burst size (L). Asterisks denote significant difference between groups. Scale bars in E and F are 100 μm and the scale bar in H is 1 mm.

We used chemogenetics to activate the NTS and tested for an effect on drinking behavior (Fig. 3G–L). In this experiment, a DREADD-mCherry fusion protein packaged in AAV8 was injected into the NTS. Fig. 3H shows a representative micrograph of an injection site and the expression of the fusion protein in the hindbrain. Rats were water deprived overnight before being injected with vehicle or CNO into the 4V and subsequently allowed to drink. The time-course of licking appeared similar between rats given 4V injections of vehicle or CNO in rats that had excitatory DREADD in the NTS (Fig. 3I), and there were no differences in total licks for water (Fig. 3J; t = 1.83, df = 6, p = 0.12), burst number (Fig. 3K; t = 0.83, df = 6, p = 0.43), or burst size (Figure 3L; t = 2.03, df = 5, p = 0.10).

3.4. Experiment 4: response of PVH-projecting hindbrain neurons

We used the same brain sections as those in Experiment 3 to look at the activity of PVH-projecting NTS neurons. CTβ was unaffected by treatment condition, but there was an effect of drinking on the number of CTβ-positive cells that were co-localized with Fos. We found no difference between groups in average number of CTβ-positive cells per section (t = 0.06, df = 15, p = 0.95), nor in average number of double-labeled cells per section (t = 1.70, df = 15, p = 0.11), but the percent of CTβ-positive cells that were double-labeled was significantly higher in water-deprived rats allowed to drink than those not allowed to drink (Fig. 4A; t = 2.57, df = 14, p = 0.02). We tested for, but did not find, a correlation between average number of double-labeled cells and amount of water consumed by rats provided with a bottle (Fig. 4B; r² = 0.004, p = 0.86). A photo of representative double-labeled cells in the NTS is shown in Fig. 4C.

Fig. 4.

Response of PVH-projecting hindbrain neurons. We counted the number of CTβ-positive cells, and the number of CTβ-positive cells that were also Fos-positive, and found that a greater number of CTβ cells were double-labeled in water-deprived rats allowed to drink than in those not allowed to drink (A). In the group allowed to drink, there was no correlation between number of double-labeled cells and amount of water consumed (B). Micrograph (C) showing cells positive only for CTβ (white arrows) and cells that are double-labeled (black arrows). Asterisks denote significant difference between groups and the scale bar is 100 μm.

Next, we used chemogenetics to activate a subset of hindbrain cells that project to the PVH (Fig. 5) and tested for an effect on drinking behavior. A representative micrograph of DREADD expression in the dorsal vagal complex after injection into the PVH is shown in Fig. 5B. Rats were water deprived and given injections of CNO or vehicle into the 4V before being allowed to drink. All rats drank during the test, but unlike rats in the vehicle group, in which drinking plateaued after the first hour, rats given CNO continued to drink throughout the 2-h test (Fig. 5C) and total intake was greater in rats treated with CNO than it was in rats given vehicle (Fig. 5D; t = 5.04, df = 5, p = 0.003). The difference in intake was largely driven by the number of bursts of licking (Fig. 5E; t = 3.91, df = 5, p = 0.01), in spite of a lower average number of licks per burst (Fig. 5F; t = 4.45, df = 5, p = 0.007).

Fig. 5.

Chemogenetic activation of PVH-projecting hindbrain neurons. An excitatory DREADD packaged in a retrogradely transported AAV was injected into the PVH (A). Representative immunohistochemistry for DREADD expression in the dorsal vagal complex is shown in B. When the rats were water deprived and given an injection of CNO into the 4V, we found a significant potentiation of water intake (C, D) that was a function of an increased number of bursts (E) that increased intake in spite of a reduced burst size (H). Asterisks indicate statistical significance (p < 0.05) and the scale bar in B is 100 μm.

4. Discussion

The present studies yielded several novel findings about the role of GLP-1 in the control of fluid intake, as well as the role of PVH-projecting hindbrain cells in fluid intake. The findings related to fluid intake control by GLP-1 were two-fold. First, we demonstrated that GLP-1R agonist application to either the forebrain or hindbrain ventricle affects fluid intake, but that the dose needed to affect drinking is lower when the drug is applied to the lateral ventricle than it is when applied to the hindbrain ventricle. This is different from our previous findings showing no difference between drinking after 4V or LV injection, but the previous studies did not include the doses needed to reveal the present differences (McKay et al., 2011). Perhaps more interesting, this is markedly different from previous studies showing that feeding effects of GLP-1 require lower doses of drug in the hindbrain ventricle than are required after forebrain ventricle application (Kinzig et al., 2002). Second, we provide evidence that GLP-1R activation in discrete nuclei within the brain affects water intake. Specifically, we found that GLP-1R agonist injected into the NTS and PVH reduced water intake, but the same effect was not observed after injection into the NAc. Although the NAc and PVH are bilateral structures, we made unilateral injections of Ex-4. The lack of an effect on drinking from NAc injections could, therefore, be due to the necessity of GLP-1R activation in both hemispheres, and this possibility is worthy of further exploration. While it is important to consider this caveat when drawing conclusions about the sufficiency of GLP-1R actions in the NAc in the impact on drinking, our results show that there is, at least, a clear difference in sensitivity to Ex4 between the PVH and the NAc. Moreover, it is noteworthy that food intake-suppressive effects of Ex4 are found after unilateral NAc injections of Ex4 (Alhadeff et al., 2012). Should bilateral activation be necessary for drinking effects but unilateral activation sufficient for eating effects, this would be an interesting example of how GLP-1 function in this region of the brain appears to be differentially sensitive to food and water.

The lack of an effect of Ex4 in the NAc on drinking is especially interesting considering previous studies show that GLP-1R activation in the NTS, PVH, and NAc reduces food intake (Alhadeff et al., 2012; Dossat et al., 2011; McMahon & Wellman, 1998; Richard et al., 2015), suggesting that NAc GLP-1 is more relevant for food intake than for water intake. Moreover, these findings, in conjunction with previous work from our lab demonstrating that the GLP-1 system responds differently to food deprivation and fluid deprivation (McKay et al., 2014), demonstrate that there are at least partially separable portions of the GLP-1 system for the control of food intake and water intake. The apparent separability of feeding and drinking effects of GLP-1 add to previous findings that suppression of eating and visceral illness are mediated by different GLP-1 subsystems (Kinzig et al., 2002), highlighting the diversity in the GLP-1 system and the need to better understand the different subsystems in which GLP-1 acts. This is especially important for feeding and drinking, because the behaviors are intertwined. Indeed, eating and drinking commonly co-occur (Fitzsimons & Le Magnen, 1969), and dehydration suppresses eating (Watts, 1999), but there also are clearly occasions in which an animal needs to drink but not eat. Having separate but overlapping systems for intake termination is therefore likely advantageous. A more thorough knowledge of how GLP-1 functions could lead to improvements in drug therapies based on GLP-1 analogs. Presently the use of GLP-1R agonists for treatment of type 2 diabetes mellitus or obesity is not targeted to the particular subsystems responsible for the desired effect of the drugs. One of the most common side effects of these drugs is nausea (for example (Astrup et al., 2012; Lean et al., 2014),), and it is easy to see how the ability to retain the incretin or food-suppressive effects of the drug while eliminating nausea would be beneficial. Similarly, if food- and water-suppressive effects could be separated, the risk of dehydration could be reduced. A first step towards this is identifying which brain regions and pathways are common and which are distinct between GLP-1-mediated behaviors and physiological actions.

The first two experiments yielded findings relevant to our understanding of GLP-1 and fluid intake control, and the next two tested the role of a particular pathway in fluid intake control more generally. The results of Experiments 1 and 2 informed the choice of pathway, but the methods employed were intended to be broad and not exclusively focused on the functions played by a particular neuropeptide. Because the first two experiments indicated that GLP-1R activation in the NTS or PVH is sufficient to suppress drinking in rats, the next two experiments were intended to probe the pathway between the NTS and PVH and its role in mediating fluid intake. Although the experiments were not designed in a way that allows for direct conclusions to be drawn about the GLP-1 system and fluid intake, the NTS is the main source of central GLP-1 (Jin et al., 1988; Larsen et al., 1997b; Merchenthaler et al., 1999), and GLP-1 expressing cells in the NTS have been shown to project to the PVH (Jin et al., 1988; Rinaman, 1999; Vrang et al., 2007). Thus, it is tempting to speculate that at least some of the cells targeted by the experiment are part of the GLP-1 system and represent a means by which endogenous GLP-1 can act in the PVH. Our results show that the act of drinking by water deprived rats was associated with increased Fos expression in the NTS. This is consistent with what others have shown (Dalmasso et al., 2015; Gottlieb et al., 2006, 2011), and extends those findings with a more anatomically precise analysis of different levels of the NTS. Moreover, our experiments found an interesting correlation between the average number of Fos-positive cells and the amount of water consumed by rats allowed to drink. Although the regression analysis found that intake was mathematically predicted by the average number of Fos-positive cells, the experimental design does not allow us to draw conclusions about the causal direction. Future experiments are needed to test for causality and the direction of that effect, but the observation of the correlation is nevertheless of interest. Consistent with this correlation, we found an increased amount of Fos in the subset of PVH-projecting NTS cells of rats allowed to drink. In this population of cells there was, however, no correlation between the number of Fos-positive cells and the amount of water consumed, suggesting that PVH-projecting NTS neurons are fundamentally different from NTS neurons generally. Future experiments are needed to tease apart these differences and to better understand the phenotype and function of these populations of cells.

The results of the DREADD experiments provide additional support for the notion that PVH-projecting cells in the NTS are a distinct subset of the NTS. We found no effect on fluid intake after activating the NTS as a whole. We urge caution when interpreting this result, however, because others have found a suppression of intake after NTS activation (Ryan et al., 2017), and despite a lack of significance in our experiment, the direction of the effect was consistent with previous reports. Perhaps more compelling was the difference between the impact of CNO that depended on the target of the DREADD. Specifically, when the DREADD was injected into the NTS, 4V application of CNO had, if anything, an inhibitory effect on intake (without being statistically significant), but when the DREADD was injected into the PVH, application of CNO to the 4V caused an enhancement of water intake. Microstructural analysis of licking patterns showed that this enhancement of intake was driven by an increase in the number of bursts, to the point that the additional bursts compensated for a reduced burst size. Changes in burst number are typically thought to reflect changes in postingestive feedback, while changes in burst size are thought to reflect changes in orosensory feedback (reviewed in (Smith, 2001)). Therefore, the present data suggest that activation of PVH-projecting hindbrain neurons drastically diminishes satiation from drinking water, but at the same time reduces any orosensory hedonic value of the water that is being consumed. Because GLP-1R activation in the PVH suppresses water intake and because endogenous GLP-1 is produced in the NTS, we hypothesized that activating PVH-projections in the hindbrain would similarly reduce intake. Instead, we found the opposite effect on intake (GLP-1 decreases intake whereas activation of these cells increased intake) and on the microstructure of the intake. Specifically, we have reported previously that icv GLP-1R agonist and antagonist treatment, respectively, reduce and increase burst number (McKay & Daniels, 2013; McKay et al., 2014). Reconciling these differences will require specific targeting of GLP-1 in the NTS-PVH pathway to test the possibility that although the pathway contains GLP-1 neurons, activation of this subset is not sufficient to outweigh the effect of activation of other subsets of cells in the pathway. How many such subsets exist remains unknown, but it seems likely that some signal with GLP-1 and suppress drinking, whereas others promote drinking, and that a non-selective activation of all projections is weighted more towards the drinking-promoting population. Although our data found increased Fos in a subset of PVH-projecting NTS neurons in rats that were allowed to drink after deprivation, the function and role of those cells remains to be determined. These cells could be activated in promotion of drinking, in response to drinking and act to attenuate subsequent drinking, or a combination of the two. The low temporal resolution of the Fos approach prevents us from knowing if the Fos was expressed at the onset or at the termination of drinking. The difference between rats that were allowed to drink after deprivation and rats that were not allowed to drink suggests, however, that the activation does not precede the onset of drinking, because that would likely be found in rats that were similarly deprived, but not allowed to drink. Overall, however, the last two experiments call into question whether endogenous GLP-1 acts at the PVH to decrease drinking, with further experiments required to answer this question.

In conclusion, the current experiments provide additional information to expand the understanding of how GLP-1 and hindbrain-forebrain interactions control fluid intake. Central GLP-1R activation suppresses water and saline intake (McKay & Daniels, 2013; McKay et al., 2011), and this is not merely a pharmacological effect (McKay et al., 2014). The present studies identified the NTS and PVH as potential targets of GLP-1 in the brain, and suggest that the NAc is likely not a site at which GLP-1 acts to control fluid intake. In contrast to the suppression of intake observed after GLP-1R agonist treatment in the PVH, activation of the hindbrain-to-PVH pathway increased water intake, leaving open the question of whether endogenous GLP-1 acting in the PVH is part of normal drinking termination.