Mechanosensation of the heart and gut elicits hypometabolism and vigilance in mice

Interoception broadly refers to awareness of one’s internal milieu. Vagal sensory afferents monitor the internal milieu and maintain homeostasis by engaging brain circuits that alter physiology and behavior. While the importance of the body-to-brain communication that underlies interoception is implicit, the vagal afferents and corresponding brain circuits that shape perception of the viscera are largely unknown. Here, we use mice to parse neural circuits subserving interoception of the heart and gut. We determine vagal sensory afferents expressing the oxytocin receptor, hereafter referred to as NDGOxtr, send projections to the aortic arch or stomach and duodenum with molecular and structural features indicative of mechanosensation. Chemogenetic excitation of NDGOxtr significantly decreases food and water consumption, and remarkably, produces a torpor-like phenotype characterized by reductions in cardiac output, body temperature, and energy expenditure. Chemogenetic excitation of NDGOxtr also creates patterns of brain activity associated with augmented hypothalamic-pituitary-adrenal axis activity and behavioral indices of vigilance. Recurrent excitation of NDGOxtr suppresses food intake and lowers body mass, indicating that mechanosensation of the heart and gut can exert enduring effects on energy balance. These findings suggest that the sensation of vascular stretch and gastrointestinal distention may have profound effects on whole body metabolism and mental health.

We also observed tdTomato-labeled fibers encapsulating a portion of the aortic arch [Fig. 1l]. These fibers formed a claw-like structure positioned between the left subclavian and left common arteries [ Fig. 1m]. A prior study determined that this aortic-claw arises from nodose ganglionic neurons that express mechanically gated ion channels, PIEZO1/2, and function as mechanoreceptors, specifically arterial baroreceptors, that monitor the stretch exerted on the aortic arch 20 . Indeed, our follow-up in situ hybridization experiment revealed expression of PIEZO1/2 mRNAs within tdTomato-labeled nodose ganglionic neurons (Fig. 1n-o). In addition to PIEZO(s), transient receptor potential (TRP) cations channels have been implicated in baroreception 21 and we observed that tdTomato-labeled neurons also expressed mRNAs encoding transient receptor potential ankyrin 1 (Trpa1; see Fig. 1p). These anatomical observations suggest that arterial baroreceptors express Oxtr(s), which conflicts with the results of Bai et al 2019 indicating that NDG Oxtr specifically sense distention of the gastrointestinal tract 10 . It is possible that methodological differences, and in particular, the Cre-driver mice used to direct the expression of tdTomato to the Oxtr gene accounts for this discrepancy. Accordingly, we conducted an additional neuroanatomical experiment that delivered Cre-inducible AAV synthesizing tdTomato into the nodose ganglia of the Oxtr-Cre driver mice used by Bai et al 2019 10 . Using these mice, we again found that tdTomato clearly outlined a claw-like structure on the aortic arch (Fig. 1q). These anatomical results predict that NDG Oxtr function as mechanoreceptors that monitor stretch exerted on the stomach, duodenum, and aortic arch.

NDG Oxtr that project to the gastrointestinal tract and aortic arch are anatomically segregated but each population likely expresses PIEZOs.
Because tdTomato fibers originating from NDG Oxtr were observed in the gastrointestinal tract and aortic arch, it is possible that both tissues are innervated by the same nodose ganglionic neurons. In other words, NDG Oxtr innervating the gastrointestinal tract may also send collateral projections that target the aortic arch. To address this possibility, we delivered Oxtr Cre mice retrograde Cre-inducible AAV synthesizing enhanced yellow fluorescent protein (eYFP) or tdTomato to the gastrointestinal tract or aortic arch, respectively (Fig. 2a). Three-weeks later, mice were perfused, and brain and nodose ganglia were extracted. As expected, a subset of soma in the nodose ganglia were labeled with eYFP or tdTomato; however, co-expression of the fluorophores was not observed (Fig. 2b). Confocal scans through thick coronal sections found that axons labeled with eYFP or tdTomato ascended through the rostral hindbrain (-7.0 mm from Bregma; Fig. 2c) and migrated dorsally and caudally toward the midline where they terminated within discrete regions of the caudal NTS (-7.5 to -7.8 mm from Bregma; Fig 2c); however, like the soma in nodose ganglia, colocalization of eYFP and tdTomato labeled fibers was not observed. These results are consistent with the notion that NDG Oxtr send projections that innervate the stomach, duodenum, or aortic arch and suggest that separate populations of nodose ganglionic neurons target gastrointestinal or cardiovascular tissues.
As mentioned, nodose ganglionic neurons innervating the aortic arch utilize the mechanically gated ion channels, PIEZOs, to transduce vascular stretch into action potentials. Whether NDG Oxtr similarly utilize PIEZO to monitor gastrointestinal distention is unknown. To begin to address this unknown, we delivered Oxtr Cre male mice retrograde Cre-inducible AAV synthesizing tdTomato into the stomach and duodenum (Fig. 2d). Three-weeks later, mice were perfused and nodose ganglia were extracted and processed for in situ hybridization for PIEZO1/2 mRNAs. Figures 2e and 2f depict confocal images of PIEZO1 and PIEZO2 mRNAs and nodose ganglionic neurons retrogradely labeled with tdTomato, and therefore, presumed to project to the gastrointestinal tract and synthesize Oxtr(s). Merged images reveal mRNAs coding for PIEZO1 and PIEZO2 within the confines of tdTomato labeled neurons [ Fig. 2e-f]. Taken together, these results suggest that separate populations of NDG Oxtr innervate the gastrointestinal tract and aortic arch; however, it is likely that subsets of each population utilize PIEZOs to monitor distention.

Chemogenetic excitation of NDG Oxtr decreases food intake and energy expenditure
The molecular and structural characteristics of NDG Oxtr predict that these vagal sensory afferents monitor mechanosensation of the gut. To functionally test this prediction, we conducted experiments utilizing in vivo chemogenetics to selectively activate NDG Oxtr while simultaneously recording metabolic parameters in conscious freely moving mice. Cre-inducible AAVs expressing DREADDs (Gq-DREADDs) and mCherry were bilaterally injected into the nodose ganglia of male Oxtr Cre mice (Fig. 3a). This approach allows for selective chemogenetic excitation of NDG Oxtr when CNO is administered. Male Oxtr Cre mice given a Creinducible AAV only producing mCherry served as controls for the expression of Gq-DREADDs. Subsets of these male mice were also implanted with telemetry devices (DSI; HD-X10) and/or housed within a TSE PhenoMaster. After surgical recovery and acclimation to the TSE PhenoMaster, mice were given an injection of the saline vehicle or CNO (0.3 mg/kg, i.p.) at the onset of the dark phase.
Filling of the gastrointestinal tract distends the muscular layers of the stomach and duodenum and this sensation is thought to contribute, at least in part, to meal termination and reduced food intake 22 . We hypothesized that chemogenetic excitation of NDG Oxtr recapitulates the sensation of GI distention, and consistent with this hypothesis, administration of CNO to Oxtr Cre mice given AAV-Gq-DREADDs significantly decreased food intake throughout the 18h testing period (Fig. 3b). Importantly, Oxtr Cre mice given the AAV-mCherry control virus exhibited food intakes that were similar when administered CNO or the saline vehicle (Fig. 3b). These results suggest that chemogenetic excitation of NDG Oxtr at the onset of the dark phase, when most of feeding occurs, potently suppresses food consumption, and it is unlikely that off target effects of CNO or Gq-DREAADs contribute to this effect.
We anticipated that mechanosensation of the gut would suppress feeding behavior but predicting how exciting NDG Oxtr would affect energy expenditure was less clear. In rats, acute gastric filling is followed by post-prandial elevations in energy expenditure 23 ; however, chronically enhancing gastrointestinal distention with inflation of a gastric balloon or bariatric surgery produces the opposite effect 24,25 . When compared to controls, Oxtr Cre mice expressing Gq-DREADDs that were given CNO had significantly decreased, oxygen consumption (VO2; Fig. 3c), respiratory exchange ratio (RER; Fig. 3d), and core body temperature (Fig. 3e). These results suggest that exciting the NDG Oxtr slows metabolic rate and is akin to manipulations that tonically enhance and/or simulate gastrointestinal distention. Moreover, the reduced RER observed shortly after delivery of CNO is indicative of the increased oxidation of fat that occurs during states of negative energy balance. Collectively, these results suggest that chemogenetic activation of NDG Oxtr reduces indices of energy expenditure and promotes fat oxidation with the latter effect likely resulting from the reduced food intake that occurs shortly after Oxtr Cre mice were administered CNO.
Chemogenetic excitation of NDG Oxtr decreases blood pressure, heart rate, and water intake.
Elevated blood pressure activates arterial baroreceptors that provoke baroreflex-mediated reductions in cardiac output and vascular resistance. Subsets of NDG Oxtr formed a claw-like structure on the aortic arch, suggesting that these sensory vagal afferents function as arterial baroreceptors. Here, we combine chemogenetics with telemetric recording of cardiovascular parameters to determine whether excitation of NDG Oxtr mimics mechanical distention of the aortic arch and triggers baroreflex-mediated reductions in blood pressure and heart rate. As shown in Figs. 3f-g, 1 h after administration of CNO, Oxtr Cre mice expressing Gq-DREADDs had significant and robust decreases in systolic blood pressure and heart rate relative to controls. At the nadir, systolic blood pressure and heart rate dropped ≈20 mmHg and ≈200 BPM, respectively. Remarkably, despite these robust effects, all mice fully recovered and were outwardly no different than controls after the effects of CNO had waned. In addition to perfusion pressure, arterial baroreceptors are thought to maintain intravascular blood volume by regulating fluid intake. Specifically, acute elevations in blood pressure or distending the vasculature via inflation of an atrial balloon significantly reduces water and saline consumption in rodents [26][27][28] . Similar to these prior studies, chemogenetic excitation of NDG Oxtr significantly decreased water intake relative to controls (Fig. 3h). Taken together, these results indicate that exciting NDG Oxtr elicits cardiovascular and behavioral responses that are reminiscent of those that follow loading of the arterial baroreceptors.
Mechanosensation of the heart and gut engages neuronal circuits mediating satiety and stress responding.
Next, we discerned the brain circuits that are engaged by chemogenetic excitation of NDG Oxtr . We again bilaterally delivered Cre-inducible AAV-GqDREADDs or AAV-mCherry control into the nodose ganglia of male Oxtr Cre mice. After surgical recovery, mice were administered CNO (0.3 mg/kg, i.p.), and ninetyminutes later were euthanized, and brains were extracted and processed for dual Fos immunohistochemistry (IHC) and RNAscope in situ hybridization. As expected, giving CNO to mice that expressed Gq-DREADDs significantly increased Fos IHC in the area postrema (AP) and NTS relative to control mice expressing only mCherry ( Fig. 3i-j). The AP and NTS mediate satiation 17,29 and emerging work has shone a light on the glucose-dependent insulinotropic polypeptide receptor (GIPR) as a therapeutic target for obesity with agonism of this receptor or activation of GIPR-expressing neurons suppressing food intake 30,31 . Within the AP and NTS, GIPR(s) are primarily distributed on GABAergic neurons 32,33 , and fascinatingly, chemogenetic excitation of NDG Oxtr significantly increased Fos IHC within cells synthesizing mRNAs encoding vesicular GABA transporter (VGAT; GABAergic marker) and GIPR (Fig. 3i-j). These results indicate that NDG Oxtr demarcate vagal sensory afferents that excite neurons in the AP and NTS that synthesize GIPR(s) and GABA.
Neurons within the NTS send excitatory projections to the parabrachial nucleus (PBN), which relays sensory information to forebrain structures that shape the emotional valence of a stimulus 34 . Chemogenetic excitation of NDG Oxtr augmented Fos IHC within the PBN (Fig. 3k-l), and intriguingly, we observed a significant increase in the number of neurons double-labeled for Fos IHC and mRNAs encoding calcitonin gene-related peptide (CGRP; Calca mRNA). The CGRP neurons of the PBN are activated by noxious stimuli 35 and pairing their chemogenetic activation with consumption of a palatable sucrose solution elicits a conditioned taste avoidance 36 . These prior studies, in conjunction with the co-localization of Fos IHC and Calca mRNA, suggest that mechanosensation of the heart and gut may be a noxious stimulus. Towards this end, we paired chemogenetic excitation of NDG Oxtr with consumption of a palatable saccharine solution to determine whether perceived mechanosensation of the heart and gut alters appetition of a hedonic stimulus. After consuming a palatable saccharin solution, Oxtr Cre mice expressing Gq-DREAADD or mCherry were given the saline vehicle. As shown in Fig. 3m (left), 48 h later, both groups of mice exhibit a clear preference for a flavored saccharin solution over water during a twobottle preference test. In contrast, 48 h after pairing saccharin with CNO, the preference for the flavored solution is completely abolished only in mice expressing Gq-DREADD ( Fig. 3m; right). These results indicate that NDG Oxtr recruit activation of Calca neurons residing within the PBN and suppress hedonically driven appetitive behavior.
The paraventricular nucleus of the hypothalamus (PVN) integrates interoceptive and exteroceptive signals and coordinates neuroendocrine responses that mitigate real or perceived threats to homeostasis. We found that chemogenetic activation of NDG Oxtr robustly increased Fos IHC in the PVN (Fig. 3n-o). Dual IHC and RNAscope in situ hybridization revealed that chemogenetic excitation of NDG Oxtr significantly increased Fos IHC within PVN neurons synthesizing mRNAs encoding corticotrophin-releasing-hormone (CRH; Fig. 3n-o). Within the PVN, neurons that synthesize CRH are known to activate the hypothalamicpituitary-adrenal (HPA) axis, which elevates circulating levels of corticosterone (CORT). Corresponding experiments evaluated whether chemogenetic excitation of the NDG Oxtr drives HPA axis activity to increase plasma levels of CORT. Gq-DREADDs or control mice were given CNO (0.3 mg/kg, i.p.) and blood was sampled, and plasma CORT was measured. Relative to controls, Oxtr Cre mice expressing AAV-Gq-DREADD had robust and significant elevations in plasma corticosterone, 1 h after administration of CNO (Fig. 3p). To further characterize HPA axis activity, serial samples were collected 2, 4, 6 and 24 h after administration of CNO. Interestingly, relative to controls, Oxtr Cre mice expressing AAV-Gq-DREADD had elevated plasma CORT 2, 4, and 6 h after administration of CNO with a return to basal levels occurring at 24 h (Fig. 3p). These results indicate that excitation of NDG Oxtr is a stressor that potently activates the HPA axis and elevates plasma levels of CORT.
In mice, activation of CRH neurons in the PVN or CGRP neurons in the PBN is associated with vigilance that manifests as increased anxiety-like behavior 35,37,38 . To determine whether activation of NDG Oxtr is anxiogenic, Gq-DREADDs or control mice were delivered CNO and 1 h later, anxiety-like behavior was assessed using the light-dark box (LDB), elevated plus maze (EPM), and open field arena (OFA). As depicted in Supp. Fig. 1, Oxtr Cre mice expressing AAV-Gq-DREADD that were given CNO had fewer entries in the light-and spent less time on the light-side of the LDB when compared to that of controls. Chemogenetic excitation of NDG Oxtr also reduced entries into the open arms or center area of the EPM or OFA, respectively. The total distance travelled was also significantly reduced by chemogenetic activation of NDG Oxtr . Collectively, these results suggest driving NDG Oxtr engages brain circuits that suppress feeding, alter the valence of hedonic stimuli, elevate circulating stress hormones, and promote behavioral vigilance.

Repeated chemogenetic excitation of NDG Oxtr decreases food intake and body mass but does not affect anxiety-like behavior.
The hypometabolism and vigilance that we observed during excitation of NDG Oxtr closely resembles torpor, which is a state that heterotherms, like mice, enter when the availability of metabolic fuel is insufficient to meet real or perceived energetic demands 12 . Because torpor slows metabolic rate and alters fuel utilization, methods that artificially induce torpor have been explored for a variety of applications [39][40][41] . In this regard, the reductions in food intake, heart rate and energy expenditure that accompany excitation of NDG Oxtr present the possibility that these sensory vagal afferents may be repeatedly engaged to artificially induced torpor and long-term changes in energy balance. To address this possibility, we again delivered Cre-inducible AAV expressing Gq-DREADDs and mCherry bilaterally into the nodose ganglia of male Oxtr Cre mice. Littermates that only expressed mCherry served as controls. Following surgical recovery, mice were implanted with telemetry devices to enable continuous recording of cardiovascular parameters. After habituation to the TSE PhenoMaster, mice were delivered CNO (0.3 mg/kg, i.p.) at the onset of the dark phase (11:00 h) every-other-day for eleven days and food intake, body weight, indices of energy expenditure, and cardiovascular parameters were recorded. As shown in Fig. 4, on days when CNO was given, mice expressing AAV-Gq-DREADDs had decreased food intake (Fig. 4a), body temperature ( Fig. 4b), VO2 (Fig. 4e), and RER ( Fig. 4f) when compared to baseline and non-injection days. Importantly, these alterations were not observed in control mice expressing AAV-Gq-mCherry (Supp. Fig. 2). That is, food intake, body temperature, VO2, and RER were not altered subsequent to CNO administration in these control mice. These results suggest that the effects of CNO do not linger and the prior reduction in food intake observed during chemogenetic excitation of NDG Oxtr does not elicit compensatory overconsumption of calories or changes in energy expenditure. Regarding cardiovascular parameters, on days when CNO was given, mice expressing AAV-Gq-DREADDs had decreased blood pressure ( Fig. 4i) and heart rate ( Fig. 4j) when compared to non-injection days or baseline measures. However, blood pressure, and heart rate were similar amongst the injection, non-injection and baseline days in the control mice. Intriguingly, repeated chemogenetic excitation of NDG Oxtr produced a 3-5% reduction in body mass that was sustained throughout the experimental paradigm (Fig. 4m). Collectively, these results indicate that repeatedly exciting NDG Oxtr promotes sustained decreases in body mass but chronic reductions in energy expenditure and cardiovascular parameters were not observed.
In mice, chronic unpredictable stress is accompanied by decreased body weight, increased anxiety-like behavior, and augmented HPA axis activity 42,43 . During chemogenetic excitation of NDG Oxtr , mice display an anxiogenic phenotype and greatly elevated plasma CORT, a proxy of HPA axis activity. Therefore, it is possible that the repeated excitation of NDG Oxtr that promotes weight loss is perceived as chronic unpredictable stress. To address this possibility, after the 7 th day, on the even numbered days when CNO was not given (i.e., days 8, 10 and 12), mice were tested for anxiety-like behavior to determine whether a history of chemogenetic excitation of NDG Oxtr alters mood and affect. As depicted in Supp. Fig. 3a-f, control mice and those given AAV-Gq-DREADDs behaved similarly in the LDB and EPM on even numbered days when CNO was not administered. Next, we considered that oxytocin is a potent mediator of social behavior 44 and we used the social interaction paradigm to assess indices of social avoidance or social fear 45 . Controls and Gq-DREADDs mice that were previously treated with CNO spent similar amounts of time investigating novel conspecifics, suggesting that repeated chemogenetic excitation of NDG Oxtr had no effect on social behavior (Supp. Fig. 3g-i). Taken together, these results demonstrate that repeated chemogenetic excitation of NDG Oxtr alters food intake and energy expenditure to reduce body mass; however, these episodes of NDG Oxtr activation are not associated with heightened stress responsivity as measured by anxiety-like behaviors.

The reduced food intake and body mass that accompanies repeated chemogenetic activation of NDG Oxtr does not require temporally coupling torpor with eating behavior.
Finally, we contemplated that torpor shares several commonalities with sleep and when faced with energetic challenges, mice will tend to enter torpor at the onset of the light-phase, during their circadian nadir of activity 46 . Taking this into account, we evaluated whether repeatedly delivering CNO at the onset of the light-phase would recapitulate the cardiometabolic changes that were previously observed. Mice were delivered AAV-Gq-DREADDs or AAV-mCherry, implanted with telemetry devices and housed in the TSE Phenomaster as described. Following surgical recovery and habituation to the TSE PhenoMaster, mice were delivered CNO (0.3 mg/kg i.p.) at the onset of the light phase (ZT 23h), every-other-day for 11 days. As with our previous experiments, administration of CNO to mice expressing AAV-Gq-DREADDs decreased food intake (Fig. 4c), body temperature (Fig. 4d), VO2 (Fig. 4g), RER (Fig. 4h), blood pressure (Fig. 4k) and heart rate (Fig. 4l) when compared to days that CNO wasn't injected and to baseline measurements; however, many of these effects were now constrained to the light-phase. These alterations were not observed in controls (Supp. Fig. 3). Remarkably, on the days that CNO was delivered, chemogenetic excitation of NDG Oxtr significantly suppressed food intake and this effect persisted throughout the darkphase when measures of cardiovascular function were largely similar amongst the conditions (Fig. 4). Similar to results observed during the Chronic PM administration of CNO (Fig. 4m), alterations in energy balance during Chronic AM CNO administration led to a 3-4% reduction in body mass (Fig. 4n). Collectively, these results indicate that the reductions in food intake that follow excitation of NDG Oxtr cannot be attributed to decreased blood pressure and heart rate. Moreover, chemogenetic excitation of NDG Oxtr does not need to be concurrent with ingestive behavior in order to observe suppressed food intake. Rather, the excitation of NDG Oxtr may engage brain circuits that mediate the long-term regulation of energy balance.

DISCUSSION
Here, we reveal that Oxtr(s) are expressed on nodose ganglionic neurons that monitor mechanosensation of the heart and gut. Our anatomical experiments found that NDG Oxtr comprise a subset (≈20%) of nodose ganglionic neurons innervating the aortic arch or gastrointestinal tract that express molecular markers and structural endings indicative of mechanosensation. Fascinatingly, chemogenetic excitation of NDG Oxtr produced a torpor-like phenotype that includes suppression of eating and drinking as well as robust decreases in blood pressure, heart rate, body temperature, and energy expenditure. Activation of NDG Oxtr was coupled to augmented Fos expression within brain circuits mediating stress responding and this expression was predictive of conditioned taste avoidance, augmented HPA axis activity, and increased anxiety-like behavior. Repeatedly coupling excitation of NDG Oxtr with ingestive behavior by delivering CNO at the onset of the dark-phase produced sustained reductions in body mass without affecting anxiety-like behavior. Remarkably, delivering CNO at the onset of the light-phase, when mice are inactive or sleeping, suppressed food intake during the subsequent dark-phase and this also produced sustained reductions in body mass. Collectively, these results demonstrate that vagal afferents transducing mechanosensation of the heart and gut have a profound influence over brain circuits mediating cardiometabolic function and stress responding. The broad implication is that NDG Oxtr govern interoceptive pathways that can be leveraged to understand and alleviate cardiometabolic diseases and mental health disorders. The brain monitors the moment-to-moment status of blood volume and perfusion pressure, in part, via arterial baroreceptors whose soma reside within the NDG 47 . Arterial baroreceptors form a claw-like structure that encapsulates the aortic arch and transduces stretch exerted on the vessel wall into action potentials that are carried to the NTS 20, 48,49 . While still an active area of research 21,50 , recent evidence points to mechanosensitive ion channels, PIEZO(s), as molecular markers for arterial baroreceptors 51 . We found that NDG Oxtr express PIEZO1/2 mRNAs and send projections to the aortic arch that form a claw-like structure. Chemogenetic excitation of NDG Oxtr significantly reduced blood pressure and heart rate, which is reminiscent of baroreflex activation elicited by hypertensive stimuli. We also found that chemogenetic excitation of NDG Oxtr suppressed water intake. This presents somewhat of a conundrum because the unloading of arterial baroreceptors during hypotension drives 52 , but the loading that occurs with hypertension abrogates, fluid intake 53 . Thus, the suppressed water intake that we observed may appear paradoxical because chemogenetic excitation of NDG Oxtr produced hypotension; however, our interpretation of these results is that activation of NDG Oxtr recapitulates the sensation of baroreceptor loading and this blunts thirst and drinking behavior.
Studies using electrophysiological recordings found that distention of the GI tract elicited firing of vagal afferents that persisted after ablation of the mucosal surface [54][55][56][57] . This suggested that vagal mechanoreceptors were not located in the mucosa, but rather, resided within the muscle wall of the gastrointestinal tract. Indeed, anatomical studies revealed the presence of IGLE(s) and IMA(s) within the circular and longitudinal muscle layers 16,58 ; though, the strongest evidence linking IGLE(s) to gastrointestinal mechanosensation comes from experiments that found that stretch sensitive vagal afferents were associated with receptive fields on the stomach that were innervated by IGLE(s) 19,59,60 . Consistent with the results of Bai et al. 10 , we report that NDG Oxtr give rise to IGLE(s) and IMA(s) that populate the stomach and densely occupy portions of the duodenum that are proximal to the pylorus. Additionally, NDG Oxtr innervating the gastrointestinal tract synthesize PIEZO1/2 mRNAs, which were recently found to transduce stomach distention and mediate feeding in Drosophila 61,62 . Our results, in conjunction with those from prior electrophysiological and anatomical studies, support the notion that Oxtr(s) demarcate vagal afferents subserving mechanosensation of the gastrointestinal tract. Interestingly, we found that, relative to males, IGLE(s) originating from NDG Oxtr were less dense in females.
Peripherally administered oxytocin appears to be more efficacious at reducing food intake in males relative to females 63,64 and the scant number of IGLE(s) that we observed in the proximal duodenum of females may contribute to this effect.
Vagal sensory afferents form excitatory synapses onto cells within the NTS 48 and these cells, deemed 2 nd order neurons, relay interoceptive signals to nuclei orchestrating body-to-brain communication. Prior work from our group revealed that NDG Oxtr synapse onto 2 nd order neurons that synthesize preproglucagon (PPG) 65 . These PPG neurons are thought to release glucagon-like peptide 1 (GLP-1), which stimulates glucagon-like peptide 1 receptors (GLP-1R) expressed throughout the brain 66 . Relevant to the current study, peripherally administered oxytocin suppresses feeding by acting on vagal afferents 14 and PPG neurons are required for this anorexigenic effect 65 . Interestingly, PPG neurons are activated by gastric distention as well as psychogenic stressors. Moreover, GLP-1 labeled fibers and GLP-1R(s) are distributed throughout brain regions governing physiological and behavioral response to stress 66 . Central administration of GLP1-R agonists lower core body temperature 67 , which is similar to the reduction in core body temperature that we observed when NDG Oxtr are excited. Therefore, it is possible that the anorexigenic and stress responsive phenotype that accompanies chemogenetic excitation of NDG Oxtr can be attributed, in part, to the engagement of PPG neurons. However, the bradycardia that we observed is contrary to what occurs with engagement of the central GLP-1 system 68 , and therefore, it is likely that NDG Oxtr recruit diverse populations of 2 nd order neurons to produce the complex phenotype that follows their excitation.
The hypometabolism and vigilance that we observed during excitation of NDG Oxtr closely resembles torpor, which is a state that heterotherms, like mice, enter when the availability of metabolic fuel is insufficient to meet real or perceived energetic demands. Reductions in body temperature 69 , reduced cardiac output and perfusion pressure 70 , decreased respiratory rate and oxygen consumption 71,72 , suppression of hunger 73 or thirst 74 , and behavioral vigilance 73 have all been reported as torpor-induced effects. Excitation of NDG Oxtr similarly produced these effects and, like torpor 75,76 , our effects persisted for 4-6 hours. Food scarcity, cold and/or threat of predation are conditions that promote torpor, and by convention, nervous system connections that sense the external environment are studied as mediators. Here, we demonstrate that exciting vagal afferents relaying mechanosensation of the heart and gut potently elicits a torpor-like phenotype. Recently, Matsuo and colleagues reported that vagal afferents partly contribute to the torpor-like phenotype that accompanies systemic delivery of Trpa1 agonists 71 , and intriguingly, we determined that ≈30% of NDG Oxtr expressed Trpa1 mRNA. Taken together, these results indicate that excitation of NDG Oxtr is sufficient to drive physiological and behavioral responses that are hallmarks of torpor and further suggest that interoceptive pathways potently alter cardiometabolic function and behavior in response to real or perceived threats to the internal milieu.
Artificially-induced torpor has been proposed for applications ranging from medical therapies to space travel [39][40][41] . These applications often require prolonged bouts of torpor, and as proof of concept, we evaluated the effects that repeated activation of NDG Oxtr had on cardiometabolic function and stress responding. Artificially inducing torpor by repeatedly exciting NDG Oxtr promoted sustained weight loss but transiently lowered energy expenditure, body temperature and heart rate. These results suggest the weight loss can likely be attributed to the blunted food intake that occurs on the day of NDG Oxtr excitation. The repeated bouts of 'artificial torpor' were not associated with exaggerated indices of stress responding (e.g., anxiety-like behavior), suggesting that the intervention is well-tolerated. Intriguingly, weight loss was also observed when the activation of NDG Oxtr and the circadian rhythm of ingestive behavior were temporally uncoupled. That is, exciting NDG Oxtr at the onset of the light-phase, when mice are inactive or sleeping but not eating, promoted sustained weight loss. Decreased food intake also contributed to this weight loss; however, this anorectic effect occurred 12 hours after CNO administration when the hypometabolism that follows excitation of NDG Oxtr had subsided. Taken together, these results establish that repeated activation of NDG Oxtr is well-tolerated and promotes decreased food intake that contributes to sustained reductions in body mass. The implication is that NDG Oxtr may be a promising therapeutic target for the treatment of cardiometabolic disorders, like obesity. However, it should be noted that the suppressed feeding behavior and weight loss was observed in lean mice maintained on a standard diet. It is possible that artificially inducing torpor with activation of NDG Oxtr may produce larger reductions in body mass when mice are rendered obese on a high-fat diet. Alternatively, maintenance on a palatable diet may override the mechanically gated satiation signals supplied by NDG Oxtr but still engage mechanosensitive pathways that promote hypometabolism and vigilance. In this way, overconsumption of palatable and calorically dense foods and fluids may promote states of positive energy balance and affective disorders, like obesity and anxiety, by simultaneously overwhelming satiety-signals supplied by mechanosensation of the heart and gut while also recruiting neuronal pathways promoting energy conservation and stress responding.

Animals
All procedures were approved by the Institutional Animal Care and Use Committee at the University of Florida. Mice were on a C57BL/6J background, housed individually, and maintained on a 12:12 h light/dark cycle, in temperature (20-26°C) and humidity (30-70%)-controlled rooms. Additionally, subjects were given ad libitum access to standard rodent chow (Envigo Teklad [7912]) and water, unless stated otherwise. The majority of the studies were performed using Oxtr-Cre mice (Mutant Mouse Resource Research Centers, Stock #036545-UCD). This bacterial artificial chromosome transgenic mouse line expresses Cre recombinase under control of the Oxtr-specific promoter and has been described previously 77,78 . In addition, a cohort of Oxtr-T2A-Cre-D (Jax Strain #031303) was utilized for anatomical tracing studies 79 .

Surgical procedures
Pre-and post-operative procedures. For all survival surgeries, mice were anesthetized using 2% isoflurane and then administered analgesic (Buprenorphine SR-LAB (Zoo Pharm), 1 mg/kg, s.c.; meloxicam (Pivetal Alloxate), 20 mg/kg, s.c.). Body temperatures were then maintained on a heating pad throughout the duration of the surgery and ophthalmic ointment was applied for corneal protection. AAV delivery to the Aortic Arch. After performing pre-operative procedures, a subset of Oxtr-Cre mice was placed in dorsal recumbency and then delivered AAV2/retro-FLEX-tdTomato (Addgene, Catalog #28306-AAVrg), onto the aortic arch following surgical procedures previously described 5 . Briefly, following intubation for artificial ventilation (tidal vol; 0.2 ml, min vol; 26 ml min -1 , Pressure 21 cmH2O), a longitudinal midline cervical incision was made from the sternal notch to mid-chest and pre-trachea muscles bluntly separated. A sternotomy was then performed (approximately 3-4 mm) to expose the aorta and pericardial cavity. A pulled glass micropipet was then used to apply 1 ul of virus to the aortic arch. Viral constructs were left in situ for 5 minutes to allow for the diffusion of AAVrg into the aortic arch. The incision site was then sutured closed, and the mouse either returned to a recovery cage or then delivered virus to the GI tract.
AAV delivery to the GI tract. After performing pre-operative procedures, a longitudinal laparotomy was performed and the stomach and proximal duodenum exposed and isolated. Cre-dependent AAV [5 µl; AAV2/retro-FLEX-tdTomato (Addgene, Catalog #28306-AAVrg) or AAV2/retro-pCAG-FLEX-EGFP-WPRE (Addgene, Catalog #51502-AAVrg)] was then applied to the surface of the antrum of the stomach, the pylorus and the proximal duodenum using a disposable pipet. The incisions were sutured and mice were returned to home cages subsequently for recovery.
Telemetry device implantation. In order to evaluate cardiovascular parameters during chemogenetic excitation of NDG Oxtr , a subset of mice (n = 24) that were previously administered AAVs into the NDG, underwent a further surgery to implant radiotelemetry devices (HDX10; Data Sciences International, St. Paul, MN). These surgeries were performed as previously-described 37 , using the anesthesia and analgesia procedures that are described above. Briefly, radiotelemetry transmitters were placed intraperitoneally and secured to the abdominal wall using suture. The abdominal muscles were then closed with absorbable suture. Using the same procedure as for insertion of the Millar catheter, the fluid-filled catheter was inserted into the distal left carotid artery and secured in place with suture. The skin was closed with nonabsorbable 5-0 monofilament suture.

Indirect calorimetry and ingestive behavior
Three weeks after surgical procedures, mice were transferred into an automated indirect calorimetric system (PhenoMaster, TSE systems) that recorded oxygen consumption (VO2), respiratory exchange ratio (RER) and continuous food/water intake. RER is the ratio of carbon dioxide production to oxygen consumption (VCO2 / VO2). The environmental temperature (22-25℃), light (12:12 h light/dark cycle) and humidity (50%) were maintained by way of the climate-control chamber and TSE PhenoMaster software. Mice were habituated in the chamber for at least three days prior to any experimental data collection.

Telemetric cardiovascular recording
During cardiovascular recordings, mice were also housed in the TSE PhenoMaster (as described above), with their cages placed on PhysioTel receivers (Data Sciences International, St. Paul, MN). For the duration of the cardiovascular studies, systolic blood pressure (SBP), diastolic blood pressure (DBP), mean arterial pressure (MAP), heart rate (HR) and core body temperature were continuously acquired and hourly bin data were calculated using Ponemah Software (Data Sciences International, St. Paul, MN).

Acute CNO administration
For acute experiments, food access was restricted for 2 h prior to dark phase. After at least 3 days habituation to i.p. injections (0.1 ml saline), mice received one injection of 0.1 ml CNO (0.3 mg/kg, i.p.) 1h before onset of the dark phase (zeitgeber time, ZT, 11). Access to food was given at the onset of the dark phase, and food intake, VO 2 and RER recorded. A subset of mice (n= 4 CON-NDG Oxtr and 5 Gq-NDG Oxtr ) had telemetry devices implanted for assessment of cardiovascular parameters.

Chronic CNO administration
Mice were given at least 3 days to habituate to the Phenomaster cages and saline injections (0.1 ml i.p.). CNO (0.3 mg/kg i.p.) was injected every other day for 11 days (6 CNO injections), 1h before onset of the dark phase (ZT 11, for chronic PM CNO administration) or 1h before onset of the light phase (ZT 23, for chronic AM CNO administration). VO2, RER and food intakes were measured during this time. A subset of these mice (n=15; 7 CON-NDG Oxtr and 8 Gq-NDG Oxtr ) had telemetry implants in order to record cardiovascular parameters.

Conditioned taste avoidance
To determine whether acute activation of NDG Oxtr was aversive, a conditioned taste avoidance (CTA) test was performed using a 2-bottle intake chamber (52 cm x 10 cm x 30.5 cm, l x w x h) with lickometry devices (Noldus) using a protocol adapted from previously published CTA experiments 80,81 . Prior to the experiment, mice were habituated to the intake chamber for 1h per day for 3 consecutive days with ad libitum access to a water bottle. Mice were not water deprived during this habituation phase. This was then followed by a training period; in order to motivate the mice to drink from the bottles during a short period of time, mice were water restricted overnight (<16h) every other night, prior to the intake session. During the training session, a water bottle was placed in one of the 2 bottle ports and mice were given 30 min to drink. The location of the bottle was alternated and counterbalanced between sessions to limit side preferences. Mice were then returned to their home cage. Two hours following the end of the session, food and water were returned. This habituation period took 4-7 days; conditioning did not proceed until water intakes stabilized over the 30 min session. For conditioning, 2 different palatable solutions of 0.15% saccharine flavored with 0.05% Kool-Aid were used. On the first day of conditioning (following overnight water deprivation), the mice were given 30 min to consume flavored solution 1, and then injected with vehicle (saline) and returned to the home cage. The 2nd day was a rest day, and the mouse was then water deprived overnight and tested for preference on the 3 rd day, where the mouse was given 30 min access to both a water bottle and a bottle containing flavor 1. Mice were given another rest day, and then underwent conditioning to flavor 2 in the same manner as flavor 1; however, the 2 nd flavor was paired with CNO (0.3 mg/kg), rather than saline. They were given a day off after flavor 2 conditioning, and were tested on the following day with access to bottles containing water and flavor 2. Flavors 1 and 2 were counterbalanced to account for any differences in palatability. Preference was calculated as licks for flavored solution divided by the total licks for water and flavored solution combined.

Open Field Test
The open field (OF) test was conducted in a 45 x 45 x 30 cm (l x w x h) square arena constructed of white plastic floor and walls. Each mouse was brought into the procedure room immediately before the test and placed in the center of the open field arena 30 min after i.p. CNO (0.3 mg/kg) injection. Results including total distance travelled and entries into center (located in the center of arena with 50% of total area) or edge were recorded and analyzed using Ethovision XT 13 software (Noldus Information Technology, Netherlands). The total duration of open field test was 5 min.

Light Dark Box Test-Acute and chronic activation
Approximately 2h after the start of the light phase, mice were placed in a light dark box (LDB) test arena in which one chamber (20 x 40 x 35 cm, l x w x h) is kept dark, with black polycarbonate walls and lid, and the light side (20 x 40 x 35 cm) is made of clear polycarbonate walls with no lid; the floor of both sides is grey. The two chambers are separated by a black wall with a small central opening (7 x 7 cm), which the test mouse could freely cross through. Mice were initially placed into the middle of the dark chamber, facing the back wall, and allowed 5 min of free exploration. Tests were digitally-recorded and handscored. Acute administration: 30 min after i.p. CNO (0.3 mg/kg) injection, the test mouse was placed in the center of the dark chamber and allowed free exploration of the arena. Chronic administration: the experimental mouse was tested in the LDB on the morning following PM CNO (0.3 mg/kg, i.p.) administration (approximately 14h after administration). LDB was performed on day 10 of chronic CNO treatment.

Three-chambered Social Interaction Test
To determine whether chronic activation of NDG Oxtr affected social behavior, a 3-chamber social interaction test (3CSIT) was run on D12 of chronic stimulation. The polycarbonate arena (60 cm x 40.5 cm x 22cm, l x w x h) is divided into 3-19 cm wide chambers with doors between them. During the first phase of the test, the mouse was given 5 min to investigate the empty arena. For the social preference test, the mouse was returned to the center chamber and a clear acrylic cage (10 cm diameter, 20 cm tall) was placed in each end chamber. One chamber had an empty cage, and a novel C57BL6/J mouse was placed in the other cage. The experimental mouse was then allowed to explore the chambers for 5 minutes. Both phases were recorded and scored using Ethovision XT13 (Noldus, the Netherlands). Time spent in proximity to the cage was measured, and social preference was calculated as time spent near the mouse divided by total time spent near the mouse and near the empty cage. The test was performed 2 h after onset of the light cycle, approximately 14 h after the final (D11) PM administration of CNO.

Assessment of Plasma CORT
One hour after i.p. CNO (0.3 mg/kg) injection, blood samples from the tail tip were collected from CON-NDG Oxtr and Gq-NDG Oxtr mice; additional samples were collected 2, 4, 6, and 24 h following injection. Blood samples were collected into EDTA-treated tubes and centrifuged at 2500 x G for 15 min at 4°C, and isolated plasma samples stored at -80°C. Plasma corticosterone (CORT) levels were measured using an 125 I radioimmunoassay kit (MP Biomedicals, Orangeburg, NY) as previously described 82 .

Tissue collection and sectioning
At the end of the studies, mice were injected with 390 mg sodium pentobarbital/50 mg phenytoin per ml solution (Euthanasia-III Solution, Med-Pharmex Inc., Pomona, CA), 0.1 ml i.p.). Once breathing ceased, transcardial perfusions were performed by clearing with 0.15M NaCl followed by fixation with 4% paraformaldehyde (Sigma Aldrich). Brains were extracted and post-fixed in 4% paraformaldehyde for 4 h. After post-fixation, brains were transferred into 30% sucrose and stored at 4°C, until sectioned for either in situ hybridization or immunohistochemistry. The left and right NDG were dissected and placed in in 4% paraformaldehyde for 5 min before they were stored in 20% RNase-free sucrose at 4 °C, until being whole-mount imaged before being sectioned for in situ hybridization. The aortic arch and stomach were also carefully dissected, collected and stored in RNase-free saline at 4 °C, until being whole-mount imaged. Brains and NDG for in situ hybridization were sectioned at a thickness of 20 µm or 10 µm respectively, using a Leica cryostat (Leica CM3050S, Leica). The cryostat and other tools were cleaned with RNase remover (Fisher Scientific) and ethanol before use. Tissues were first collected in RNase-free PBS and then mounted on slides (Tissue Path Superfrost Plus Gold Microscope Slides, Fisher Scientific). After airdrying at room temperature for 1 h, slides were rinsed with PBS, incubated in 4% PFA for 15 min, and then dehydrated for 5 min in 50%, 75%, and 100% EtOH at room temperature. Following that, slides were incubated in H 2O2 for 10 min, rinsed with PBS, and allowed to dry for 10-15 min before being stored at -80°C for further processing. Brains for immunohistochemistry experiments were sectioned at 30 µm and stored in cryoprotective solution (1L 0.1M PBS, 20g PVP-40, 600ml ethylene glycol, 600g sucrose) at -20°C.

In situ hybridization
In situ hybridization with various probes was performed to examine Cre recombinase expression in Oxtr-Cre mice and the phenotype of Oxtr-expressing neurons using the RNAscope ® V2 Multiplex Fluorescent Reagent Kit (Advanced Cell Diagnostics, Newark, CA). Detailed procedures and the reagents are provided as per the manufacturer's instructions (Advanced Cell Diagnostics, Inc., Hayward, CA) and as previously described 83 . Briefly, brain and NDG slides were first treated with protease IV or III respectively (Advanced Cell Diagnostics, Lot#2002337) for 20 min in the dark at room temperature. After 3 consecutive washes with RNase-free water, brain slides were incubated in probes for the mRNA(s) of vGat (Mm-Slc32a1, Cat. to target RNA, and thus decreasing nonselective binding. After incubation, slides were rinsed 3 times with wash buffer reagents followed by Amplification steps I, II, and III. After Amplification III, slides were incubated in HRP, before they were incubated in TSA Plus Cy5, Cy3 or fluorescein at 1:1000 in TSA buffer (TSA TM Plus System, Akoya Bioosciences). Following that slides were incubated in HRP blocker for 15 min and then rinsed 3 times with wash buffer reagent. Upon completion of the in-situ hybridization, slides immediately underwent immunohistochemistry for c-Fos (brain sections), NeuN and m-cherry (NDG sections).

Immunohistochemistry
Immunohistochemistry was conducted immediately after in situ hybridization. Brain and NDG slides were rinsed 5 times in 50 mM potassium phosphate buffered saline (KPBS) for 5 min, and then blocked in 2% normal donkey serum and 0.2% Triton X-100 in KPBS to minimize nonspecific binding, before being incubated in the primary antibodies. For the NDG, slides were either incubated in Anti-NeuN or Anti-mCherry. For NeuN labelling, slides were incubated in the conjugated antibody A60 mouse anti-NeuN-AlexaFluor 488 (1:1000; MAB377X, EMD Millipore) for 2 h at room temperature. Slides were then rinsed in 50 mM KPBS allowed to air-dry, then cover-slipped. For mCherry labelling, to amplify the tdTomato signal, slides were incubated in primary antibody rabbit anti-mCherry (1:500; PA5-34974, Thermo Scientific) overnight at 4°C. Brain slides were incubated in the polyclonal rabbit anti-cFos (1:2000; RPCAc-Fos, EnCor Biotechnology) overnight at 4°C. On the next day, slides were rinsed in rinsed in KPBS three times at room temperature. Following that, slides were incubated for 2 h in the secondary antibody donkey anti-rabbit Cy3 (711-165-152) purchased from Jackson ImmunoResearch and used at a 1:500 dilution. Slides were then rinsed with KPBS, incubated in DAPI (Advanced Cell Diagnostics, Newark, CA) for 30 s, rinsed again in KPBS, and allowed to air dried overnight. Slides were then cover slipped with ProLong TM Gold antifade mounting medium before being imaged.

Image capture
Images were captured and processed using a laser-scanning confocal microscope (Nikon Instruments Inc. Melville, NY, USA). Large coronal scans captured at 10 x magnification throughout the aortic arch and stomach (whole-mount), NDG (whole-mount LNG and RNG), and hindbrain (coronal sections) were acquired to assess the expression of tdTomato and GFP 3 . Regions with high expression of tdTomato and GFP were then imaged at 20 x magnification and z-stacks through sections were acquired. For IHC and in situ hybridization in the brain and NDG, z-stacks of the proteins and mRNAs of interest were captured at 20 x or 40 x magnification. An average of 20 optical sections were collected per z-stack (0.5 µm between z-steps). For in situ hybridization, sections hybridized with the probes of interest were used to determine the exposure time and image processing required to provide optimal visualization of RNA signal. As described in detail previously 8 , these same parameters were then used to assess background fluorescence in sections hybridized with the negative control probe (DapB). This allows for distinguishing bright positive control puncta (indicative of individual mRNA transcripts) from low-intensity background puncta. For evaluation of c-Fos-positive nuclei, the exposure time for the c-Fos channel was kept the same for all samples within a cohort. All representative photomicrographs were then prepared using FIJI 84 and any adjustments to settings were kept consistent within a cohort.

Quantification of RNAscope in situ hybridization and immunohistochemistry
Quantification of mRNA(s), proteins (c-Fos, NeuN and mCherry) and their colocalization was done on QuPath software. Maximum projection images from z-stacks of regions of interest (ROIs) of brain sections and NDG were quantified for mRNA expression and co-localizations with the different protein markers. For evaluation of percentage of c-Fos, NeuN, and m-Cherry cells containing mRNAs of interest, cells were considered to contain or colocalize to mRNA if at least three visible transcripts or puncta dots were observed with the volume of the protein markers. The mean number/percentages per section were determined by averaging the values across sections for each individual animal and then by calculating the mean across the group.

Quantification of duodenal intraganglionic laminar endings (IGLEs)
Whole mount duodenum samples were prepared and analyzed as previously described 85 . Briefly, maximum intensity projections of confocal z-stack images were acquired from both tdTomato positive sensory afferent fibers and a separate background channel identifying muscle layers, ganglia, villi, and other tissue types. Quantitative analysis was performed using a 1 mm 2 grid to determine IGLE density. IGLEs were identified as 1) a distinct cluster of terminal puncta, 2) the terminal portion of a single axon, and 3) overlapping ganglia identified on the background channel.

Statistics
Data were analyzed and graphed using Prism 9 for Windows (GraphPad Software, Inc. La Jolla, CA). The number of intraganglionic laminar endings (IGLEs) per mm duodenum were analyzed using a 2-way analysis of variance (ANOVA) with Šídák's multiple comparisons tests. Patterns of neuronal activation were analyzed using unpaired one-tailed t-tests. Acute and chronic CNO administration calorimetry, telemetry and food intake data were analyzed using either 2-way repeated measures ANOVAs or mixed effects models where appropriate. For acute administration, comparisons between Gq-NDG Oxtr -CNO and Gq-saline, Con-CNO and Con-Saline data were made. For chronic CNO administration, CNO injection data were compared to Baseline and No Injection data within treatment group. When significant treatment or treatment x time interactions were observed, Dunnett's multiple comparisons tests were performed. Anxiety-like behaviors (EPM, OF and LDB tests) and sociability were evaluated using unpaired two-tailed student t-tests. Conditioned taste avoidance was analyzed using 2-way RM ANOVA with Šídák's multiple comparisons tests. Corticosterone responses to CNO administration were analyzed using 2-way RM ANOVA with Fisher's LSD post-hoc tests. Differences in patterns of neuronal activation were analyzed using unpaired one-tailed t-tests. Detailed statistical results are provided as a supplementary table.