Oxytocin-receptor-expressing neurons in the parabrachial nucleus regulate fluid intake

Brain regions that regulate fluid satiation are not well characterized, yet are essential for understanding fluid homeostasis. We found that oxytocin-receptor-expressing neurons in the parabrachial nucleus of mice (OxtrPBN neurons) are key regulators of fluid satiation. Chemogenetic activation of OxtrPBN neurons robustly suppressed noncaloric fluid intake, but did not decrease food intake after fasting or salt intake following salt depletion; inactivation increased saline intake after dehydration and hypertonic saline injection. Under physiological conditions, OxtrPBN neurons were activated by fluid satiation and hypertonic saline injection. OxtrPBN neurons were directly innervated by oxytocin neurons in the paraventricular hypothalamus (OxtPVH neurons), which mildly attenuated fluid intake. Activation of neurons in the nucleus of the solitary tract substantially suppressed fluid intake and activated OxtrPBN neurons. Our results suggest that OxtrPBN neurons act as a key node in the fluid satiation neurocircuitry, which acts to decrease water and/or saline intake to prevent or attenuate hypervolemia and hypernatremia.

Oxytocin-expressing neurons are located in the paraventricular nucleus of the hypothalamus (PVH) and supraoptic nucleus (SON) and project axons to the posterior pituitary to release oxytocin peripherally 1 . The PVH also contains oxytocin-expressing neurons that project within the CNS 1 . The oxytocin receptor (Oxtr) is expressed in several brain regions and is responsible for mediating the central effects of oxytocin 1,13 .
To evaluate the role of oxytocin-related neural circuitry on ingestive behaviors, we generated mice expressing Cre recombinase at the Oxtr locus ( Supplementary Fig. 1). A recent study demonstrated that optogenetic stimulation of Oxtr-expressing neurons in the arcuate nucleus decreases feeding 14 . Here we focused on Oxtr-expressing neurons in the parabrachial nucleus, a brain region that is known to modulate ingestive behaviors [15][16][17][18][19] . We found that Oxtr PBN neurons are important regulators of fluid intake.

Results
Oxtr is expressed in the PBN. To examine where Oxtr is expressed in the mouse brain, we bred Oxtr Cre/+ mice with Cre-dependent TdTomato reporter (Ai14) mice. We observed robust TdTomato expression in the dorsolateral (dl) and external lateral (el) parabrachial nucleus (PBN) (649 ± 39 Oxtr PBN neurons per brain) and several other brain regions ( Fig. 1a and Supplementary Fig. 2a,b). We confirmed adult expression of Oxtr by injecting adenoassociated virus (AAV) carrying Cre-dependent mCherry (AAV1-DIO-mCherry) into the PBN of adult Oxtr Cre/+ mice and by in situ hybridization for Oxtr mRNA in Oxtr Cre/+ ::Ai14 mice ( Fig. 1b and Supplementary Fig. 2c). To characterize the effects of oxytocin on Oxtr PBN neurons, we recorded the activity of fluorescently labeled cells in brain slices and observed that bath application of Thr 4 ,Gly 7 -oxytocin (TGOT), a selective Oxtr agonist, efficiently activated Oxtr PBN neurons, and this activation was blocked by coadministration of atosiban (oxytocin-receptor antagonist), which confirmed the specificity of the activation ( Fig. 1c and Supplementary Fig. 2d).
We also investigated the effect of Oxtr PBN activation on fluid homeostasis. To assess both ECF volume and osmolarity, we used a two-bottle choice task consisting of water and 0.3 M NaCl (saline) 17,19 and fed mice a sodium-depleted diet 21 . The molarity of total fluid intake under control conditions was ~0.15 M NaCl (equivalent to normal saline). Chemogenetic activation of Oxtr PBN neurons after 24-h fluid deprivation suppressed total fluid intake (both water and 0.3 M NaCl) at the start of the dark cycle (Fig. 1f). After ~4 h, there was a rebound such that the cumulative overnight NaCl intake was not significantly different between hM 3 Dq-expressing and control mice (P = 0.5821), and there was only a small difference in cumulative water intake (P = 0.0394; Supplementary Fig. 3a). To assess whether the presence of food affected fluid intake, we stimulated Oxtr PBN neurons after 24 h of fluid deprivation and then denied Oxytocin-receptor-expressing neurons in the parabrachial nucleus regulate fluid intake them food. The mice expressing hM 3 Dq decreased both water and NaCl intake, similarly to when food was present; however, control (mCherry-expressing) mice drank less water but drank more NaCl when food was absent ( Fig. 1g and Supplementary Fig. 3b). We also investigated the effects of Oxtr PBN stimulation at the start of the dark cycle without prior fluid deprivation. We observed a significant ** ** ** ** ** *** *** *** **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** ****  decrease in water intake in hM 3 Dq-expressing mice compared with controls following CNO, but not vehicle, whereas NaCl intake was not significantly different, although intake was low (Fig. 1h). There was no significant difference in cumulative overnight intake of NaCl (P = 0.2038) or water (P = 0.6645) in hM 3 Dq-or mCherry-expressing mice ( Supplementary Fig. 3c). Because Oxtr PBN stimulation decreased fluid but not food intake, we stimulated Oxtr PBN neurons during intake of a liquid diet of Ensure nutrition drink. After 24-h of caloric deprivation, intake of Ensure and water was unaffected (Fig. 1i), indicating that Oxtr PBN stimulation does not decrease the intake of all liquids. Apparently, the highly palatable nature of Ensure overrode the ability of Oxtr PBN stimulation to suppress fluid intake. Because mice drank only minimal water in this experiment, we repeated it, but returned fixed amounts of Ensure (1.6 or 0.7 ml); however, both hM 3 Dq-and mCherry-expressing mice still drank limited amounts of water ( Supplementary Fig. 3d,e), suggesting that the presence of even small amounts of Ensure decreases concurrent water intake. We assessed the effects of activating Oxtr PBN neurons on food intake during mild dehydration and observed decreased food intake (control diet) in hM 3 Dq-injected mice (Fig. 2a), suggesting that Oxtr PBN neurons may be involved in mediating dehydrationinduced anorexia 22 .
We also examined the effect of Oxtr PBN stimulation after salt depletion by injecting mice on a sodium-depleted diet with the diuretic furosemide (40 mg kg -1 ip, twice, 1 d apart) and removing access to saline (0.3 M NaCl) for 48 h. Oxtr PBN stimulation suppressed water, but not NaCl, intake (Fig. 2b), suggesting that it does not suppress salt appetite. We also assessed the effect of Oxtr PBN stimulation after generating thirst by two different mechanisms: increasing osmolarity and inducing volume loss 9,11 . To increase osmolarity, we injected hypertonic 0.5 M NaCl or equiosmolar 1 M mannitol ip 23 ; whereas we induced volume loss by injecting 30% polyethylene glycol (PEG) subcutaneously, which progressively draws out ECF without affecting osmolarity 5 . We observed that Oxtr PBN stimulation suppressed water intake after hypertonic saline injection in hM 3 Dq-versus mCherry-injected mice (Fig. 2c); however, there was no significant difference in overnight water intake (P = 0.2008), suggesting a rebound effect in fluid intake ( Supplementary Fig. 3f). Oxtr PBN stimulation also decreased fluid intake after mannitol and PEG injections (Fig. 2d,e). Throughout the experiments, we observed that mice retained similar daily baseline NaCl and water intake ( Supplementary Fig. 3g). Following experimentation, we confirmed the targeting and activation of Oxtr PBN neurons by injecting CNO 2 h before perfusion, and we observed Fos expression in 72 ± 4% of hM 3 Dq-expressing
Oxtr PBN neurons are necessary for protection against hypernatremia. To ascertain whether Oxtr PBN neurons are necessary for fluid regulation, we chronically inactivated them by bilaterally injecting AAV carrying a Cre-dependent tetanus toxin light chain (AAV1-DIO-GFP:TetTox) 24 or control (AAV1-DIO-YFP) into the PBN of Oxtr Cre/+ mice ( Fig. 3a and Supplementary Fig. 4a,b). We found no significant difference in baseline NaCl preference at 0.075 M (P = 0.6292), 0.3 M (P = 0.3364) or 0.5 M (P = 0.9224) ( Supplementary Fig. 4c), and no substantial difference in baseline fluid intake after vehicle injection, either acutely (2 h) or overnight ( Supplementary Fig. 4d,e). Following 24-h fluid deprivation, however, we observed increased NaCl, but not water, intake in TetToxinjected mice (Fig. 3b), which remained increased overnight, suggesting a longer term effect on fluid intake than that for hM 3 Dqinjected mice ( Supplementary Fig. 4f). We also observed increased 0.3 M NaCl intake after hypertonic (0.5 M) saline ip injection, but not after salt depletion or other tests of fluid intake (Fig. 3c,d and Supplementary Fig. 4g,h). Following experimentation, we observed no significant correlation between TetTox-GFP expression and fluid intake (for NaCl, P = 0.5408; for water, P = 0.3905; Supplementary  Fig. 4i). Overall, these results suggest that Oxtr PBN neuron activity prevents excessive NaCl ingestion following dehydration and hypertonic saline injection.
We also investigated acute inhibition of Oxtr PBN neurons by injecting AAV carrying a Cre-dependent hM 4 Di:mCherry transgene (AAV1-DIO-hM 4 Di:mCherry) 20 bilaterally into the PBN ( Fig. 3e and Supplementary Fig. 5a,b), and observed increased NaCl intake after CNO injection in hM 4 Di-injected mice relative to controls at the start of the dark cycle (Fig. 3f). There was increased fluid intake (NaCl and water) after 24-h fluid deprivation, but the effect was less robust than that observed for TetTox-injected mice ( Supplementary Fig. 5c). There was no difference in fluid intake during the light cycle, suggesting that acute Oxtr PBN inhibition does not spontaneously induce drinking, and there was no difference in feeding ( Supplementary Fig. 5d,e). Following experimentation, we examined the extent of hM4Di expression, but observed no significant correlation of expression with fluid intake (for NaCl, P = 0.2208; for water, P = 0.2013; Supplementary Fig. 5f).  Oxtr PBN neurons are involved in fluid and salt intake under physiological conditions, we examined Fos expression in Oxtr PBN neurons of Oxtr Cre/+ ::Ai14 mice following 48-h salt depletion or 24-h fluid deprivation. We observed significantly increased Fos expression in caudal Oxtr PBN neurons when NaCl and/or water were returned compared with control mice or mice that remained salt or water deprived, suggesting that Oxtr PBN neurons are activated following fluid satiation ( Fig. 4a-e). There were no substantial differences in Fos expression in mid-and rostral-PBN populations, revealing differential activation throughout the PBN (Supplementary Fig. 6a-d). We also examined Fos expression under hypernatremic conditions by injecting 1 M NaCl ip into Oxtr Cre/+ ::Ai14 mice. This treatment increased Fos expression in caudal Oxtr PBN neurons compared with 0.15 M NaCl (normal saline) injection ( Fig. 4a,f,g and Supplementary Fig. 6e,f).

Oxtr
We assessed real-time Oxtr PBN neuron activity by injecting AAV1-DIO-GCaMP6m into the PBN of Oxtr Cre/+ mice and measured calcium fluorescence after returning water to water-deprived mice ( Fig. 4h and Supplementary Videos 1 and 2). We observed low fluorescence in Oxtr PBN neurons during dehydration, which remained low when the water spout was returned; however, we observed a rapid rise in Ca 2+ fluorescence when mice began drinking, which decreased between bouts ( Fig. 4i,l,m). This pattern was observed in ~52% of Oxtr PBN neurons during water intake, but not during Ensure intake or when given an empty bottle, suggesting that Oxtr PBN neurons respond to noncaloric fluid ingestion, but not liquid diet or motor movements associated with licking ( Fig. 4i-m). We co-injected AAV-DIO-hM 3 Dq:mCherry into two mice and observed increased calcium fluorescence within 5 min that peaked at ~15 min and lasted at least 2 h, corresponding to the time at which we observed substantial fluid inhibition behaviorally ( Fig. 4n and Supplementary Video 3).

CGRP PBN neurons have minimal overlap with Oxtr PBN neurons and decrease both food and fluid intake.
We compared the expression pattern of neurons expressing Oxtr or calcitonin-gene-related peptide (CGRP, which is encoded by Calca gene) in the PBN. CGRP PBN -neuron stimulation is known to decrease feeding and mediate many aversive responses 15,24 . We observed minimal coexpression of Oxtr and CGRP in PBN neurons (6 ± 2%; Fig. 5a).
We investigated the effect of CGRP PBN activation on fluid intake by injecting AAV1-DIO-hM 3 Dq:mCherry into the PBN of Calca Cre/+ mice (Fig. 5b). CGRP PBN activation with CNO decreased water, but not NaCl, intake at the start of the dark cycle ( Fig. 5c) and after 24-h dehydration (Fig. 5d). Total fluid intake (combined water and NaCl) after 24-h dehydration was ~56% of that of control mice, whereas Oxtr PBN stimulation resulted in a total fluid intake of ~29% of that of control mice ( Fig. 1f and Supplementary Table 1). After 48-h salt depletion, CGRP PBN activation decreased both NaCl and water intake (Fig. 5e). Overall, these results suggest that CGRP PBN stimulation inhibits all ingestive behaviors (food, fluid and salt), but is less effective than Oxtr PBN stimulation in decreasing fluid intake. We injected CNO 2 h before perfusion and observed Fos expression in 67 ± 7% of the hM 3 Dq-expressing neurons, as well as in non-hM 3 Dq expressing neurons in the dorsolateral PBN, suggesting that CGRP PBN neurons might activate dorsolateral PBN neurons (Fig. 5f).
Oxtr PBN neurons project to brain regions involved in fluid regulation. To investigate downstream projections of Oxtr PBN neurons, we injected AAV1-DIO-synaptophysin:mCherry into the PBN of Oxtr Cre/+ mice, which revealed prominent projections to the  CNQX+D-AP5 central nucleus of the amygdala, bed nucleus of the stria terminalis, organum vasculosum of the lamina terminalis (OVLT), anteroventral periventricular nucleus (AVPV) and median preoptic nucleus (MnPO) (Fig. 6a,b), and less prominent projections to other brain regions ( Supplementary Fig. 7a,b). Many of these regions are activated following excessive fluid intake 25  PBN ( Supplementary Fig. 7d). We also injected AAV-DIO-hM 3 Dq:mCherry into the PVH and AAV-DIO-YFP into the PBN of Oxtr Cre/+ mice and observed minimal Fos expression in Oxtr PBN neurons after CNO (Supplementary Fig. 7e-g).

Activation of Oxt PVH neurons attenuates fluid intake.
To assess the effect of Oxt PVH activation on fluid intake, we injected AAV1-DIO-hM 3 Dq:mCherry bilaterally into the PVH of Oxt Cre/+ mice (Fig. 7a). Oxt PVH stimulation did not significantly attenuate baseline NaCl or water intake following CNO or vehicle injection in hM 3 Dq-injected   or control mice (Fig. 7b); however, Oxt PVH stimulation did attenuate water intake following 24-h fluid deprivation and osmolarityinduced thirst (hypertonic saline and mannitol) (Fig. 7c,d and Supplementary Fig. 7i), but not following salt depletion or volume depletion (PEG) (Supplementary Fig. 7j,k). These results suggest that Oxt PVH stimulation attenuates fluid intake, but does so less robustly than Oxtr PBN stimulation. As previously reported, there was no significant difference in feeding at baseline (P = 0.9325) or after fasting (P = 0.9670) 29 ( Supplementary Fig. 7l). Following CNO, we observed Fos expression in 64 ± 5% of hM 3 Dq-expressing Oxt PVH neurons ( Supplementary Fig. 7m).
We also inactivated Oxt PVH neurons by injecting AAV1-DIO-GFP:TetTox bilaterally to test whether these neurons have a physiological role in fluid intake similar to that of Oxtr PBN neurons ( Fig. 7e and Supplementary Fig. 8a,b). Oxt PVH inactivation revealed no substantial difference in 24-h baseline saline preference ( Supplementary  Fig. 8c); however, TetTox-injected mice increased fluid intake, particularly water intake, following vehicle injection at the start of the dark cycle and during several other tests of fluid intake ( Fig. 7f and Supplementary Fig. 8d-h), suggesting that Oxt PVH neurons have a role in reducing water intake under physiological conditions.
To assess the role of PBN projections from Oxt PVH neurons, we injected AAV1-DIO-hM 3 Dq:mCherry or AAV-DIO-mCherry bilaterally into the PVH of Oxt Cre/+ mice and implanted a bilateral cannula over the PBN (Fig. 7g and Supplementary Fig. 9a). Intra-PBN infusion of CNO (1 mM; 500 nl bilaterally) during the light cycle following 24-h fluid restriction decreased water intake in the hM 3 Dq-injected mice compared with controls (Fig. 7h). For comparison, intraperitoneal CNO also decreased water intake following 24-h dehydration in the light cycle ( Supplementary Fig. 9b). As with previous cohorts, there was no significant difference in baseline NaCl intake (P = 0.3627) or water intake (P = 0.1763) following CNO or vehicle ip injections at the start of the dark cycle ( Supplementary Fig. 9c).
Following experimentation, we confirmed targeting and cannula placement (Supplementary Fig. 9d). To validate that delivery of CNO to the PBN activates Oxtr PBN neurons, we injected Oxt Cre/+ ::Oxtr Cre/+ mice with AAV1-DIO-hM 3 Dq:mCherry in the PVH, AAV1-DIO-YFP in the PBN and a bilateral cannula over the PBN (Supplementary Fig. 9e). When mice were infused with CNO 2 h before perfusion, Fos was induced in 22 ± 2% Oxtr PBN neurons ( Supplementary Fig. 9g), suggesting that the effect may be mediated either by projections of Oxt PVH to Oxtr PBN neurons or by antidromic activation to Oxt PVH cell bodies, which then activate axon collaterals projecting to other brain regions 30 . We observed Fos expression in 26 ± 8% of Oxt PVH neurons ( Supplementary Fig. 9f), but only scattered, low-level Fos expression in the nucleus of the solitary tract (NTS) region, which receives projections from the PVH 31 (Supplementary Fig. 9h). These data suggest that activation of Oxt PVH projections to Oxtr PBN neurons provides a small attenuation of fluid intake, indicating that other neuronal inputs to Oxtr PBN neurons are required for more robust fluid suppression.
NTS neurons also suppress fluid intake and activate Oxtr PBN neurons. Previous research has suggested that the NTS and the adjacent area postrema modulate fluid intake and provide substantial input to the PBN 9,11 . To investigate the effect of the medial NTS on fluid intake, we injected nonselective AAV-hM 3 Dq:mCherry or control nonselective AAV-DsRed into the NTS of Oxtr Cre/+ mice and AAV-DIO-YFP into the PBN to fluorescently label Oxtr PBN neurons ( Fig. 8a and Supplementary Fig. 9i). Following 24-h dehydration, we activated NTS neurons with CNO and observed a large decrease in fluid (water and NaCl) and food intake in hM 3 Dq-injected mice relative to controls (Fig. 8b,c). Mice did not display overt signs of distress or impaired movement. We injected CNO 2 h before perfusion and observed robust Fos expression in hM 3 Dq-expressing NTS neurons (Supplementary Fig. 9j,k) and in 33 ± 6% of Oxtr PBN neurons and other neurons in the dorsolateral and external lateral rostral PBN (Fig. 8d).
Previous studies have identified cholecystokinin (CCK) as a neuropeptide that decreases fluid intake following infusion into the PBN 32 . CCK is expressed in NTS neurons (CCK NTS ) that project to the PBN and decrease food intake by directly activating CGRP PBN neurons 33 . We investigated whether CCK NTS neurons also decrease fluid intake by bilaterally injecting AAV1-DIO-hM 3 Dq:mCherry into the NTS of Cck Cre/+ mice ( Fig. 8e and Supplementary Fig. 9l). Following 24-h dehydration, hM 3 Dq-injected mice significantly decreased both NaCl and water intake (Fig. 8f). This decrease was not as robust as that evoked by Oxtr PBN stimulation, but was larger than that evoked by CGRP PBN stimulation (Supplementary Table 1), suggesting that it is not merely mediated via CCK NTS projections to CGRP PBN neurons. We observed Fos expression in the NTS and in both the dorsolateral and external lateral PBN regions ( Supplementary Fig. 9m,n). To investigate whether Oxtr PBN neurons were specifically activated, we performed in situ hybridization for Fos and Oxtr mRNA in hM 3 Dq-injected mice and found that 19 ± 2% of Oxtr PBN neurons expressed Fos (Fig. 8g). We also observed that bath application of a CCK agonist, CCK-8 (30 nM), increased the firing rate in both Oxtr PBN and CGRP PBN neurons (Fig. 8h), which is indicative of functional CCK receptors in both PBN populations.
A comparison of 2-h and 4-h fluid intake after stimulating each of the neuronal populations following 24-h dehydration revealed a range of effects on fluid consumption. From largest to smallest, the effects on total fluid intake were nonspecific NTS > Oxtr PBN > CC K NTS > CGRP PBN > Oxt PVH (Supplementary Table 1). Of these, only Oxtr PBN and Oxt PVH stimulation selectively decreased noncaloric fluid intake.

Discussion
Our results identify a population of Oxtr-expressing neurons in the PBN that regulate noncaloric fluid intake. These neurons receive direct inputs from hypothalamic Oxt PVH neurons, are activated by NTS neurons and project to several forebrain regions ( Supplementary Fig. 10a). Oxtr PBN neurons are predominantly separate from CGRP PBN neurons, which decrease both food 15 and fluid intake (see Fig. 5), suggesting that distinct neuronal populations mediate different aspects of ingestive behaviors.
Under physiological conditions, we observed increased Fos expression in a subset of Oxtr PBN neurons (~20%) after rehydration, suggesting that these neurons are engaged in fluid satiation. Using calcium fluorescence, we observed real-time activity in Oxtr PBN neurons, which was low during dehydration and remained low before consumption, suggesting that Oxtr PBN neurons do not respond to anticipatory cues, unlike neurons that express agouti-related protein or vasopressin 34,35 . Drinking water precipitated a rapid rise in Oxtr PBN calcium activity, which decreased between bouts, suggesting that these neurons interact with thirst-related neural circuits to adjust the overall level of fluid intake. We observed no difference in calcium activity for Ensure or for an empty bottle. One limitation of our study was that photobleaching of Oxtr PBN neurons occurred when the recording was extended, leaving us unable to quantitatively assess overall calcium activity changes over a prolonged period; however, qualitatively, calcium activity appeared to increase over the course of rehydration. We observed that ~52% of Oxtr PBN neurons responded to water, as measured by GCaMP florescence, although this is likely to be an underestimate because some neurons with low levels of GCaMP6 expression became photobleached. These results also suggest that Fos reveals only the subset of Oxtr PBN neurons with the greatest activity.
Chemogenetic activation of ~72% of Oxtr PBN neurons suppressed both NaCl and water intake, but activation of ~20% of caudal Oxtr PBN neurons following Oxt PVH activation only mildly decreased water intake. Inactivating Oxt PVH neurons predominantly increased NaCl intake. Taken together, these results suggest that Oxtr PBN neurons provide an overall inhibitory effect on total fluid intake, with the amount of fluid intake and the type of fluid intake (NaCl and/or water) being dependent on the percentage of Oxtr PBN neurons that are activated.
Results from inactivation studies suggest that Oxtr PBN neurons are necessary for preventing excessive NaCl intake. In the case of dehydration, the body engages physiological mechanisms to reduce NaCl and retain water to prevent hypernatremia-for example, by inducing a dehydration-induced natriuresis 36,37 . Our results suggest that Oxtr PBN neurons are activated following rehydration to prevent excessive NaCl ingestion. In addition, Oxtr PBN activation decreased food intake during dehydration, but not after fasting, suggesting that the decrease may be a result of salt (0.25%) in the food rather than its caloric content 38 . In the case of hypertonic saline injection, hypernatremia is known to increase thirst and water intake 9,12,39 , as well as to stimulate renal NaCl excretion 40 . Our results suggest that Oxtr PBN activation also decreases NaCl ingestion. Activation and inactivation studies revealed that Oxtr PBN neurons did not alter NaCl intake after salt depletion and did not alter food intake except during dehydration, suggesting they are not involved in salt appetite or caloric-containing food consumption. Overall, our results suggest that activation of Oxtr PBN neurons is essential for maintaining fluid homeostasis, as they decrease NaCl and/or water intake to prevent or attenuate hypervolemia and/or hypernatremia ( Supplementary Fig. 10b).
The hypothalamus is an integrative center of the brain that coordinates responses to maintain homeostatic setpoints 41 . Whole-cell patch-clamp recordings have revealed increased firing of oxytocinexpressing neurons under hypertonic conditions and decreased firing in response to angiotensin II, which signals hypovolemic conditions 42 , suggesting that hypothalamic oxytocin-related neurons may be activated predominantly by hypervolemic hypertonic conditions. We expected that activating Oxt PVH neurons would decrease NaCl intake; instead, we observed only a mild attenuation of water intake. A possible explanation is that Oxtr PBN neurons are already engaged in preventing excessive NaCl intake, and further activation via Oxt PVH stimulation provides only a small extra attenuation of total fluid intake. In addition, activating Oxt PVH may release peripheral oxytocin and induce renal sodium excretion [6][7][8] , which, in conjunction with a mild attenuation of water intake, may decrease ECF volume.
The NTS receives peripheral signals, including baroceptor input, via inputs from cranial nerves IX and X, and is known to suppress fluid intake 9,11 . We observed that activating NTS neurons, including CCK NTS neurons, substantially reduced fluid intake, suggesting that NTS neurons may provide a large functional input to the PBN to reduce fluid intake. Although we did not demonstrate direct projections to Oxtr PBN neurons, we observed Fos expression in ~33% of Oxtr PBN neurons after nonspecific activation of the NTS and in ~19% of Oxtr PBN neurons after activation of CCK neurons in the NTS, suggesting that Oxtr PBN neurons are regulated by several distinct inputs from the NTS.
Overall, our results identify Oxtr as a marker for PBN neurons that are involved in regulating fluid intake. Studying the neural circuitry of fluid satiation may help to increase our understanding of body fluid homeostasis.

Methods
Methods, including statements of data availability and any associated accession codes and references, are available at https://doi. org/10.1038/s41593-017-0014-z.

Methods
Mice. Experiments were approved by the University of Washington Animal Care and Use Committee and performed in accordance with the US National Institutes of Health Guide for the Care and Use of Laboratory Animals. Mouse lines were bred onto a C57Bl/6 J background and backcrossed > 6 generations. The following mouse lines were used in these experiments: heterozygous Oxtr Cre/+ and Calca Cre/+ (developed in our laboratory 15 ); heterozygous Cck Cre/+ and Oxt Cre/+ mice (Jackson Laboratory), and Oxt Cre/+ ::Oxtr Cre/+ mice. The Oxt Cre and Oxtr Cre mice lines were crossed to Cre-dependent Gt(ROSA)26Sor tm14(CAG-tdTomato)Hze/J mice (Allen Institute, Ai14) to reveal expression patterns.
Most behavioral experiments were performed on male mice (7-14 weeks old at start of experimentation). Combined cohorts of male and female mice were used in the following experiments: Fos study in Oxtr Cre/+ mice following fluid deprivation; Oxtr Cre/+ mice injected with non-specific hM 3 Dq or DsRed in the NTS; behavioral studies on Oxt Cre/+ mice, including Oxt Cre/+ mice implanted with cannulas; and Calca Cre/+ mice. In these experiments, female mice were distributed evenly between experimental and control groups and data were combined, as we found no significant difference in water or saline intake, as previously reported in the literature for rats 43 .
Animals in each litter were randomly assigned to either experimental or control groups. Before stereotaxic surgery, mice were group housed and maintained on a rodent diet (Picolab, number 5053) with water available ad libitum in a 12-h light:dark cycle at 22 °C. Mice were at least 7 weeks old before surgery. Following surgery, mice were single housed and allowed to recover for at least 1 week before experimentation. All fluid and food experiments were performed with at least two cohorts of mice and data were combined, unless otherwise stated.
During salt-related experiments, mice were fed on a sodium-deficient diet (< 0.02% NaCl, Diet #99091603, Research Diet) and had access to saline and water in tubes, unless otherwise noted. At other times, such as during feeding experiments in BioDAQ chambers, mice were fed a control rodent diet (Diet D12450, Research Diets). Some mice were also tested on a liquid diet, Ensure (Abbott Laboratories).
Generation of Oxtr Cre:GFP mice. The 5′ arm (~5 kb with PacI and SalI sites at each end) and 3′ arm (~5 kb with PmeI and NotI sites at each end) of Oxtr gene were amplified from a C57Bl/6 BAC clone by PCR using Phusion Polymerase (New England Biolabs) and cloned into polylinkers of a targeting construct that contained ires-mnCre:GFP, a frt-flanked Sv40Neo gene for positive selection, and HSV thymidine kinase and Pgk-diphtheria toxin A chain genes for negative selection (see Supplementary Fig. 1). The ires-mnCre:GFP cassette has an internal ribosome entry sites followed by a myc-tag and nuclear localization signal at the N terminus of Cre recombinase fused to green fluorescent protein.
The construct was electroporated into G4 ES cells (C57Bl/6 × 129 Sv hybrid) and correct targeting was determined by Southern blot of DNA digested with HindIII using a 32 P-labeled probe downstream of the 3′ arm of the targeting construct. Thirteen of 84 clones analyzed were correctly targeted. One clone that was injected into blastocysts resulted in good chimeras that transmitted the targeted allele throughout the germline. Progeny were bred with Gt(Rosa)26Sor-FLP FLP recombinase mice to remove the SV-Neo gene. Mice were then continuously backcrossed to C57Bl/6 mice.
Stereotaxic surgery. Mice were anesthetized with isoflurane, placed on a stereotaxic frame (David Kopf Instruments) and injected with ketoprofen (5 mg/kg, ip) for analgesia. Stereotaxic coordinates were normalized for the anterior-posterior plane by a correction factor (F = distance between bregma and lambda/4.21) 45 . Viruses were injected (500 nl bilaterally) via glass pipettes using the Nanoject II Auto-Nanoliter Injector Infusion into the PBN was performed via bilateral cannulas (double cannula guide; 26 gauge, 2.7 mm apart, cut 5 mm below pedestal; PlasticsOne), which were inserted according to coordinates (AP: -5.2; ML: ± 1.35; DV: -3.1) and affixed to the skull with C&B Metabond (Parkell) and dental acrylic. Prior to and between infusions, cannulas were kept patent by inserting a dummy double cannula (0.2 mm width; 2.7 mm apart; fit 5 mm) and small round-top dust caps. Infusions were performed using an internal double injector (33 gauge; 2.7 mm apart; 0.2 mm width) for a cannula guide.
Tissue processing. Following experimentation, mice were anesthetized with Beuthanasia 0.2 ml ip (effective content 320 mg/kg pentobarbital; Schering-Plough) and perfused transcardially with phosphate-buffered saline (PBS), followed by 4% paraformaldehyde (Electron Microscopy Sciences) in PBS. Brains were removed and placed in paraformaldehyde at 4 °C overnight, and then cryoprotected in 30% sucrose in PBS overnight, then OCT (Fisher HealthCare) and kept at -80 °C before processing. Coronal sections of 30 μ m thickness were cut from the brain on a cryostat. For all Fos studies, every third section was collected (90 μ m apart) for quantification. For confirmation of viral targeting, every fifth brain section was mounted onto glass slides (Fisherbrand Superfrost Plus microscope slides, Fisher Scientific) and coverslipped with Dapi Fluoromount G (Southern Biotech), and remaining sections were collected in PBS for further processing as required.
Although Oxtr Cre/+ mice have GFP fused to Cre, we were unable to visualize GFP even after anti-GFP antibody staining; hence Oxtr PBN neurons were visualized by genetic crosses with Ai14 mice, which allows expression of TdTomato in Oxtr neurons.
Slice optogenetics and electrophysiology. Acute 250-μ m coronal slices that included the PBN were prepared from 2-6-month-old mice, which were deeply anesthetized in their housing room before decapitation. Slice procedure and recordings were performed essentially as previously described 33 . Neurons of interest were identified by GFP or mCherry expression (CGRP PBN and Oxtr PBN neurons, respectively). To observe neuronal activity under minimally invasive conditions, endogenous spike activity and optogenetically evoked spikes were recorded in loose-patch cell-attached configuration with patch pipettes filled with the same artificial cerebrospinal fluid used for slice perfusion (ACSF, in mM: NaCl 115, KCl 3, CaCl 2 2, MgCl 2 1, NaH 2 PO 4 1, NaHCO 3 25 and d-glucose 11; mOsm 295, pH 7.4 when aerated with carbogen, 33 °C) and 0 holding current. Synaptic currents were recorded under voltage-clamp conditions and patch pipettes filled with intracellular solution (in mM: 100 cesium methanosulfonate, 25 CsCl, 2 MgCl 2 , 10 HEPES, 0.4 EGTA, 4 ATP, 0.4 GTP, 10 phosphocreatine, mOsm 295, pH 7.3). Light-evoked action potentials and synaptic currents were initiated by optical activation of ChR2 with 2-5-ms pulses of blue light (up to 3 mW optic power) either from a 473-nm laser (Laserglow) via an optic fiber positioned over the slice, or from a high power 460 nm LED (UHP-LED, Prizmatix) via a microscope objective. Synaptic latency was measured as the time between onset of the blue-light pulse and the onset of synaptic current. Synaptic jitter was calculated as the s.d. in variation of synaptic latency over multiple trials  in a particular neuron. Currents were elicited at 10-s interval between successive trials, recorded at 2-5-kHz bandwidth, digitized at 20 kHz and acquired by pCLAMP10 software (Molecular Devices). Drugs were dissolved in ACSF and delivered to slices by local perfusion via a three-barrel system positioned over the slice. The drug application was gravity driven and on-off timing was manually controlled. Where indicated, AMPA and NMDA glutamate receptors were inhibited by a cocktail of CNQX (20 μ M, Tocris) and d-AP5 (50 μ Μ , Tocris) added to ACSF, and GABA A receptors were inhibited by picrotoxin (100 μ M, Tocris). Other drugs used include the following: Oxtr agonist, TGOT (200 nM; Bachem, H-7710) 46 ; specific Oxtr antagonist, atosiban (1 μ M; Sigma-Aldrich, A3480) 47 ; and CCK-8 (30 nM; Bachem; H-2080) 48 . Overall, recordings were made from 45 Oxtr PBN neurons from 47 PBN slices derived from 26 Oxtr Cre/+ mice (some crossed with Oxt Cre/+ mice), of which 11 were male and 15 were female; and from 7 CGRP PBN neurons from 7 PBN slices derived from 7 Calca Cre/+ mice (all male).
RNAscope. We performed in situ hybridization (ISH) using the RNAscope assay 49 on brain tissue collected from Oxtr Cre/+ ::Ai14 and Cck Cre/+ mice. Mice were perfused as described above then embedded gradually through a series of 10%, 20% and 30% sucrose followed by OCT (Fisher HealthCare). Brains were cut coronally at 10 μ m thickness, mounted and stored at -80 °C before processing. On the day of processing, sections were thawed at 20-22 °C for ~30 min before performing ISH according to the manufacturer's protocol (Advanced Cell Diagnostics). We used probes for Oxtr (Mm-Oxtr-C3 probe, 1:50 dilution) and Fos (Mm-Fos-C2 probe, 1:50 dilution). Following ISH, slides were coverslipped and imaged as described.
Microscopy. For Fos, in situ and coexpression studies, brain sections were imaged on an Olympus Fluoview FV1200 confocal microscope (Shijuku). To confirm targeted injections, brain sections were imaged using the Nikon upright epifluorescent Eclipse E600 microscope (Minato) or Keyence Fluorescence Microscope BZ-X700. ImageJ, a program for optimizing brightness and contrast, was used to color images-for example, Oxtr Cre ::Ai14 cells red and AAV1-DIO-YFP-injected neurons green. Images were minimally processed to enhance brightness and contrast for optimal representation. Following imaging, any mouse whose targeted injection site was missed or demonstrated very sparse expression (≤ 6 fluorescent neurons/section), suggesting inadequate injection, was excluded from experimental analysis. In addition, any mouse that was unilaterally injected was included in experimental analyses for stimulatory (one unilateral injection in each group of hM 3 Dq-injected Oxtr PBN and Oxt PVH mice) and control groups, but was excluded from inhibitory (TetTox-and hM 4 Di-injected) groups. For infusion experiments, mice whose targeted injection was missed were excluded (two mice).
Fos experiments. For salt-depletion and fluid-deprivation experiments, Oxtr Cre ::Ai14 mice were single-housed and placed on a sodium-deficient diet and given access to both 0.5 M NaCl and water for 5 d of habituation. For salt depletion, mice were divided into three groups: (1) salt returned, (2) salt depleted and (3) control (n = 3 or 4 per group). Mice in group 1 were injected with furosemide (5 mg ml −1 ) daily for 2 d and denied access to saline (but not water), to develop a salt appetite 50 . On day 3, saline was returned and mice were perfused 2 h later (during the light cycle). Mice in group 2 underwent a similar protocol; however, saline was not returned before perfusion. Mice in group 3 had no furosemide challenge and were perfused on day 3.
For fluid deprivation, mice were divided into two groups: (1) fluid returned and (2) fluid deprived (n = 3 or 4 per group). Mice in group 1 were deprived of saline and water for ~24 h, and were perfused 2 h after fluid was returned (during the light cycle). Mice in group 2 were also deprived of saline and water for ~24 h, which were not returned before perfusion.
For hypertonic saline injections, mice were single-housed for 5 d with ad libitum access to food and water, and divided into two groups: (1) 1 M saline (10 µ l g −1 body weight) injection and (2) normal saline injection (n = 3 or 4 per group). Each mouse was injected with saline and perfused 2 h later (during the light cycle).
Quantification of Fos expression was performed on six 30-µ m-thick coronal brain sections 90 µ m apart, ranging from bregma -5.1 to -5.5 (encompassing the major portion of the PBN where Oxtr is expressed). The investigator who quantified Fos was blinded to the identity of the conditions. Given that there was differential activation throughout the PBN, the sections were evenly divided into caudal, middle and rostral regions. An estimate of total Oxtr-expressing neurons was made by multiplying by 3.
To examine Fos expression in Oxtr PBN , CGRP PBN , Oxt PVH , CCK NTS or NTS neurons following experimentation, mice were injected with CNO and perfused 2 h later.
Calcium imaging. Mice were prepared for calcium imaging as described 51 . Briefly, 3 weeks after AAV1-DIO-GCaMP6m and AAV1-DIO-hM 3 Dq:mCherry viral injection, mice were anesthetized (as described above) and implanted with a microendoscope lens (6.1 mm length, 0.5 mm diameter; Inscopix, catalog #100-000588) with assistance of a ProView implant kit (Inscopix, catalog #100-000754) that allowed visualization of the fluorescent activity during the lens implantation. Because basal fluorescence from Oxtr neurons was low, we treated the mice with CNO before lens implantation to facilitate visualization and lens placement. The lens was targeted to be ~200-300 µ m above the neurons using the following coordinates: -4.80 mm posterior to bregma, -1.40 mm lateral from midline, and -3.00 mm ventral to skull surface. One week after lens implantation, mice were anesthetized and a baseplate (Inscopix, catalog #100-000279) was implanted above the lens. The baseplate provides an interface for attaching the miniature microscope during calcium imaging experiments, but at other times a baseplate cover (Inscopix, catalog #100-000241) was attached to prevent damage to the microendoscope lens. Calcium fluorescence was recorded at five frames per second, 200-ms exposure time, and 50% LED power using a miniature microscope from Inscopix (nVista). The recording parameters were based on pilot studies that demonstrated the least amount of photobleaching while allowing sufficient detection of fluorescent activity. We used Ethovision (XT 10, Noldus Technology) to trigger and synchronize calcium recordings with behavioral video recordings.
Water deprivation and rehydration. Mice were water-deprived for 24 h. On the experimental day, mice were briefly anesthetized with isoflurane to attach the microscope, and then allowed to acclimate in their home cage for 1 h. After the acclimation period, baseline fluorescence was recorded for 30 s, a water bottle was placed in the cage, and fluorescence was continuously recorded for 10 min. On separate test days, we placed an empty bottle or a bottle with Ensure in the cage. We adjusted the focus of the microscope between tests, accounting for slightly different total numbers of neurons between different test days. Calcium recording files were spatially downsampled (factor of 4), motion-corrected (Inscopix, Mosaic v1.2), and fluorescent traces from individual neurons were extracted using constrained non-negative matrix factorization for microendoscopic data (CNMF-E) 52 . The key parameters used for CNMF-E were: l, 7; l n , 14; P min , 10; L min , 0.85 (ref. 52 ). We then normalized (0 to 1) individual fluorescence traces taken from the entire 10-min recording period. Photobleaching was apparent during the recording session, so we were unable to adequately measure changes in activity over extended periods of time (for example, 10-min session); instead, we analyzed the change in fluorescence activity for 5 s before and 5 s during each bout. Bouts were defined as uninterrupted mouth contact with the water spout, and the beginning of an inter-bout interval was > 10 s of not contacting the spout. The data from every bout were averaged for each neuron across the 10-min recording session. ' Activated neurons' were defined as neurons that had an average fluorescence (after time point 0) of > 2 s.d. above baseline (baseline = average fluorescence from -5 to 0 s). Following the recording session, we injected mice with CNO, which aided in identifying all the neurons. These rehydration studies were conducted during the first hour of the dark cycle.
CNO-induced activation. Mice were briefly anesthetized with isoflurane to attach the microscope, then allowed to acclimate in their home cage for 1 h. After acclimation, we recorded baseline fluorescence activity for 30 s, followed by an ip injection of CNO (1 mg/kg). We non-continuously recorded (20 s for each time point) fluorescence at 5,10,15,20,25,30,45,60 and 120 min after CNO injection. The calcium recording files were spatially downsampled (factor of 2), motion-corrected, and fluorescent traces were extracted using individual component analysis and principal component analysis (Inscopix, Mosaic v1.2). The change in fluorescence (Δ F/F) was calculated as (F -F 0 )/F 0 , where F 0 was the average fluorescence 30 s before injection and F was the average fluorescence during a particular time point. These studies were conducted during the light cycle.
Food, water and saline intake. For acute feeding assays, mice were placed in BioDAQ chambers (Research Diets) which measured food intake electronically. Mice were individually housed, and water and body weight were measured manually. Mice were supplied with a rodent diet (D12450B, Research Diets), habituated to cages for 5-7 d and habituated to saline injections for 2-3 d. To measure food intake during the dark cycle, mice were injected with CNO 30 min before lights out and had access to food removed until lights out. To measure refeeding after a fast, food was removed for 24 h, then mice were injected with CNO 30 min before the return of food (in the morning). To measure food intake during the light cycle, mice were injected with CNO in the morning, then denied access to food for 30 min to allow time for the CNO to take effect. The experimenter was not blinded to the identity of the experimental versus control groups.
For drinking and salt-related studies, mice had access to both water and saline, and were placed on a sodium-depleted diet (diet #99091603, Research Diets), unless otherwise stated, to allow accurate measurement of both fluid and salt intake. Tubes were placed into angled ports at the front of a cage and weighed manually (in grams; converted to milliliters), taking care to prevent dripping.
For TetTox-injected mice, baseline salt intake was monitored at three different concentrations (0.075, 0.3 and 0.5 M) for 3 d each (after 1 d of habituation); for subsequent experiments, a saline concentration of 0.3 M was used, unless otherwise stated. The order of experiments was baseline food and/or saline studies, 48-h salt depletion, 24-h fluid deprivation, hypertonic (0.5 M) saline ip, mannitol (1 M) ip and PEG 30% subcutaneous injections (these last three were randomly ordered). All these experiments were performed at the start of the dark cycle, unless otherwise noted. Recovery time was at least 7 d following salt depletion, and at least 3 d following other experiments.
For baseline fluid experiments in Oxtr Cre/+ and Oxt Cre/+ mice, we compared CNO versus vehicle injections in hM 3 Dq-and mCherry-injected mice, and found no significant difference in fluid intake following vehicle injection. Given that there were multiple fluid intake experiments, we limited the number of experiments performed on individual mice by injecting only CNO and comparing fluid intake in hM 3 Dq-versus mCherry-injected mice. We observed a difference in NaCl intake between Oxt Cre/+ and Oxtr Cre/+ mice, which is likely due to a difference in mouse strains obtained from difference sources, so we ensured matched littermates were used for controls in all experiments. In Oxtr Cre/+ mice, one hM 3 Dq-injected mouse appeared sick 2 d before PEG injection, so it was perfused and not included in that experiment.
Given that mice underwent a series of fluid intake experiments, we also compared daily water and 0.3 M NaCl intake from the start of experimentation (just before the first study) to the end of experimentation (in final week of fluid intake studies). Salt and fluid deprivation were performed as described in the "Fos experiments" section. For hM 3 Dq-, hM 4 Di-and mCherry-injected mice, CNO (1 mg kg −1 ) ip was injected 30 min before returning fluid. Fluid was returned ~15 min after the start of the dark cycle. For all deprivation experiments, we observed that mice would typically drink a large proportion in the first 15-30 min of the experiment, so we reported the first 2 h in 15-min intervals, unless otherwise stated. Baseline CNO and vehicle (normal saline) injections were also performed 30 min before returning fluid.
Hypertonic (0.5 M) saline and 1 M mannitol were injected ip ~10 min before CNO injection, whereas 30% PEG was injected subcutaneously ~30 min before CNO injection to allow extra time to diffuse. In general, these experiments showed increased differentiation of fluid intake over a longer time period, so we reported 4 h in 1-h intervals, unless otherwise stated. For liquid diet (Ensure) experiments, mice were habituated to Ensure and water for 2 d and then deprived of Ensure for 24 h before experimentation.
A separate cohort of mice was used to test limited Ensure intake and fluid intake in the absence of food. For the limited Ensure intake protocol, we initially returned a small quantity of Ensure (1.6 ml or 0.7 ml) after 24-h deprivation; at the end of experimentation, we returned ad libitum Ensure. For fluid intake in the absence of food, we repeated the protocol for 24-h dehydration, but removed food during the experimental period.
A separate cohort of mice was also used to test dehydration-induced food intake. Mice were habituated to BioDAQ chambers and supplied with a rodent diet (D12450B). On the day of experimentation, water and food were removed 7 h before lights out, then food (but not water) was returned at the start of the dark cycle. CNO was injected 30 min before experiments.
Intra-PBN infusions were performed during the light cycle. Infusions were performed using CMA-100 microinjection pump (Bioanalytical Systems) over 5 min. Fluid was returned 5 min after infusion and intake measured for 2 h.
For Calca Cre/+ mice, we performed baseline fluid, rehydration and salt depletion studies, as described above. Rehydration was performed with 0.3 M NaCl and water; baseline fluid and salt depletion experiments were performed with 0.5 M NaCl and water. For Cck Cre/+ mice and nonspecific hM 3 Dq-injected mice, we performed rehydration experiments with 0.3 M NaCl and water in hM 3 Dq versus control mice.
Statistics and figure preparation. Data analysis and generation of histograms were performed using GraphPad Prism Version 6.01 for Windows (GraphPad Software). Results are expressed as mean ± s.e.m. Statistical significance was determined by different tests appropriate for each dataset: for comparing the means of two groups, unpaired two-tailed Student's t test; for comparing the means of three or more groups, a one-way ANOVA with Tukey's post hoc tests; for correlation studies, a Pearson product-moment correlation; for comparing food or fluid intake over time, a repeated measures two-way ANOVA with Sidak's post hoc tests, unless otherwise noted. Three-way mixed design ANOVAs were performed using IBM SPSS Statistics for Windows v.20 (IBM). If Mauchly's test of sphericity was violated, a Greenhouse-Geisser correction was applied; if significant interaction was determined, a simple two-way interaction was performed with statistical significance accepted at a Bonferroni-adjusted α level of 0.025. We also tested for equality of variance (Brown-Forsythe test for one-way ANOVAs; or Levene's test of equality of variances). Data distribution was assumed to be normal, but this was not formally tested. ****P < 0.0001; ***P < 0.001; **P < 0.01; *P < 0.05. Calcium fluorescence data were analyzed using OriginPro v.2016 (OriginLab).
A power analysis was performed for an effective sample size using http://powerandsamplesize.com. Based on a pilot study for fluid deprivation, we used a mean of 1.8 and s.d. of 0.4. Assuming a significance level of 0.05 and power of 0.8, we calculated a sample size of 7 per group if means were 1.5-fold different with a two-tailed Student's t test. Data and figures were exported into Adobe Illustrator CS6 (Adobe Systems) to prepare figures. Images of sagittal mouse brain were taken from the Motifolio toolkit.
Supplementary statistical analyses and methods checklist are provided in Supplementary Table 2  Life Sciences Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form is intended for publication with all accepted life science papers and provides structure for consistency and transparency in reporting. Every life science submission will use this form; some list items might not apply to an individual manuscript, but all fields must be completed for clarity.

Data exclusions
Describe any data exclusions. See Methods -Microscopy section. Following imaging, any mouse whose targeted injection site was missed or demonstrated very sparse expression (≤6 fluorescent neurons/section), suggesting inadequate injection, was excluded from experimental analysis. In addition, any mouse that was unilaterally injected was included in experimental analyses for stimulatory (1 unilateral injection in each group of hM3Dq-injected OxtrPBN and OxtPVH mice) and control groups, but was excluded from inhibitory (TetTox-and hM4Di-injected) groups. For infusion experiments, mice whose targeted injection was missed were excluded (2 mice).

Replication
Describe whether the experimental findings were reliably reproduced.
See Methods -Mice section -paragraph 3. All fluid and food experiments were performed with at least two cohorts of mice and data were combined, unless otherwise stated.

Randomization
Describe how samples/organisms/participants were allocated into experimental groups.
See Methods -Mice section -paragraph 3. Animals in each litter were randomly assigned to either experimental or control groups.

Blinding
Describe whether the investigators were blinded to group allocation during data collection and/or analysis.
1) The investigator who quantified Fos was blinded to the identity of the conditions.
2) The experimenter was not blinded to the identity of the experimental versus control groups.
nature research | life sciences reporting summary June 2017

Statistical parameters
For all figures and tables that use statistical methods, confirm that the following items are present in relevant figure legends (or in the Methods section if additional space is needed).

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement (animals, litters, cultures, etc.) A description of how samples were collected, noting whether measurements were taken from distinct samples or whether the same sample was measured repeatedly A statement indicating how many times each experiment was replicated The statistical test(s) used and whether they are one-or two-sided (note: only common tests should be described solely by name; more complex techniques should be described in the Methods section) A description of any assumptions or corrections, such as an adjustment for multiple comparisons The test results (e.g. P values) given as exact values whenever possible and with confidence intervals noted A clear description of statistics including central tendency (e.g. median, mean) and variation (e.g. standard deviation, interquartile range)

Clearly defined error bars
See the web collection on statistics for biologists for further resources and guidance.

Software
Policy information about availability of computer code

Software
Describe the software used to analyze the data in this study.
See Methods -Statistics and figure preparation -paragraph 1 Data analysis and generation of histograms were performed using GraphPad Prism Version 6.01 for Windows. Three-way mixed design ANOVA were performed using IBM SPSS Statistics for Windows v.20. Calcium fluorescence data was analyzed using OriginPro v. 2016.
For manuscripts utilizing custom algorithms or software that are central to the paper but not yet described in the published literature, software must be made available to editors and reviewers upon request. We strongly encourage code deposition in a community repository (e.g. GitHub). Nature Methods guidance for providing algorithms and software for publication provides further information on this topic.

Materials and reagents
Policy information about availability of materials

Materials availability
Indicate whether there are restrictions on availability of unique materials or if these materials are only available for distribution by a for-profit company.
No unique materials were used.

Antibodies
Describe the antibodies used and how they were validated for use in the system under study (i.e. assay and species).
See Methods -immunohistochemistry section. The ordering information for the antibodies used are all referenced. The companies provide validation data for histological staining of fixed tissue.