Feeding neurons integrate metabolic and reproductive states in mice

Summary Balance between metabolic and reproductive processes is important for survival, particularly in mammals that gestate their young. How the nervous system coordinates this balance is an active area of study. Herein, we demonstrate that somatostatin (SST) neurons of the tuberal hypothalamus alter feeding in a manner sensitive to metabolic and reproductive states in mice. Whereas chemogenetic activation of SST neurons increased food intake across sexes, ablation decreased food intake only in female mice during proestrus. This ablation effect was only apparent in animals with low body mass. Fat transplantation and bioinformatics analysis of SST neuronal transcriptomes revealed white adipose as a key modulator of these effects. These studies indicate that SST hypothalamic neurons integrate metabolic and reproductive cues by responding to varying levels of circulating estrogens to modulate feeding differentially based on energy stores. Thus, gonadal steroid modulation of neuronal circuits can be context dependent and gated by metabolic status.


INTRODUCTION
The homeostatic processes of metabolism and reproduction are mutually dependent on one another.Reproductive milestones, including pregnancy and quiescence, result in major metabolic shifts, [1][2][3] and metabolic status can gate reproductive function in menstrual and estrual mammals.Pubertal onset has been shown to require a critical threshold of body fat, 4,5 and increasing adiposity is associated with a decreasing age of pubertal onset in individuals with ovaries in particular. 6During reproductive years, undernutrition can acutely disrupt menstrual cyclicity 7 via hypogonadotropic hypogonadism 8 and result in fewer successful pregnancies. 9Rodent models have been used to investigate the effects of sex variables on metabolic tissues.Estradiol has been shown to regulate key metabolic processes such as adiposity, 1 thermogenic capacity and locomotion, 1,10-13 and feeding, 14 while sex chromosome complement has been demonstrated to affect fat deposition and energy output. 15However, few studies have investigated the mechanisms by which both metabolic and reproductive status reciprocally interact to modulate behavior.
A candidate region for the seat of this interaction is the tuberal hypothalamus, the central region of the hypothalamus typically thought to be comprised of the mediobasal hypothalamus (arcuate nucleus, ventromedial nucleus, and median eminence), dorsomedial nucleus, lateral hypothalamic area, and the lateral tuberal nucleus (TN) (N.B.: to prevent confusion, references to the specific lateral tuberal nucleus are abbreviated TN, whereas ''tuberal hypothalamus'' and ''tuberal hypothalamic'' refer to the greater anatomical spatial designation -the ''middle'' of the hypothalamus on the rostral-caudal axis -encompassing the listed nuclei.).Positioned near the third ventricle and partially outside of the blood-brain barrier, 16 the mediobasal hypothalamus is situated as a nexus region with the ability to sample circulating homeostatic hormones and relay relevant information to other regions of the brain.Importantly, both this region and the greater tuberal hypothalamus are sensitive to and vital for both reproductive and metabolic homeostasis.For instance, in addition to the role of kisspeptin neurons in the arcuate nucleus as key modulators of estrogen-mediated negative feedback in the hypothalamic-pituitary-ovarian axis, [17][18][19] they also require epigenetic derepression for menarche. 20,21The arcuate nucleus is also home to the canonical homeostatic feeding neurons-appetitive agouti-related peptide (AgRP)/neuropeptide Y (NPY) neurons and satiety-related proopiomelanocortin neurons-which can not only detect and respond to sensory and metabolic cues including leptin, ghrelin, and insulin [22][23][24] but also regulate non-feeding metabolic processes such as bone mass 25 and insulin resistance via modulation of brown adipose. 26The ventromedial nucleus of the hypothalamus, and in particular the ventrolateral subregion, is important for various functions across sexes, including mating behavior (reviewed by Kammel and Correa 27 ).Distinct, estrogen-sensitive cellular populations in the ventrolateral ventromedial nucleus contribute to various aspects of metabolism, including locomotion 10,28,29 and thermoregulation. 30Given the substantive neuronal heterogeneity within a small brain region, it is unsurprising that these populations form complex functional interactions.For example, simulation of starvation via chronic chemogenetic activation of appetitive arcuate AgRP neurons has been demonstrated to acutely disrupt estrous cyclicity, 31 providing a neuronal contributor to the link between metabolic state and reproductive function.
3][34][35][36][37][38][39][40][41] Its expression marks many distinct populations within the tuberal hypothalamus.Two such prominent populations are seen in the arcuate nucleus (ARC) and TN.While the ARC is an uncontested member of the mediobasal hypothalamus, the TN straddles the mediobasal hypothalamus and lateral hypothalamic area.It is an understudied region marked by expression of SST, which has been mostly characterized in rats.Somatostatin neurons in both the ARC (ARC SST ) and TN (TN SST ) regulate food intake, [42][43][44][45] and SST expression within the greater hypothalamus is regulated by circulating gonadal steroids. 46Studies across various animals have found that ARC SST neurons both colocalize with estrogen receptor 47 and seem to be responsive to estrogens. 480][51][52][53] Furthermore, SST signaling in the ARC influences or can be influenced by the hypothalamic-pituitary-gonadal axis across a variety of species. 47,48,54,55Thus, SST neurons in these regions are excellent prospective candidates for integrating metabolic and reproductive cues to affect feeding based on anatomical location, sensitivity to circulating reproductive hormones, detection of metabolic hormones such as ghrelin, 43 and promotion of feeding behavior.Indeed, whole-body knockouts of SST exhibit weight gain that is exacerbated by sex category and high-fat diet. 56ere, we use mice to interrogate the role that SST neurons of tuberal hypothalamus play in integrating metabolic and reproductive cues to affect feeding.SST neurons exhibited differential control of feeding in female and male mice (defined by anogenital distance at weaning and postmortem inspection of the gonads), with neuronal ablation decreasing food intake only in females.This effect was primarily due to a decrease in food intake during proestrus, when circulating ovarian hormones are at high concentrations, and was only observed in animals with a low body mass.To determine whether adiposity could mediate the effect of body mass on food intake, fat transplantation experiments were performed.Increased white adipose tissue was sufficient to alter SST neuronal modulation of food intake.Together, these data reveal a context-dependent role for SST neurons of the tuberal hypothalamic region in the regulation of food intake, by which SST neurons tune feeding behavior in response to metabolic and reproductive states.

Caspase ablation of SST neurons decreases food intake only in females
To determine if permanent inactivation of SST neurons in tuberal hypothalamic regions alters feeding across sexes, an AAV expressing a Credependent modified caspase virus (taCasp3-TEVp 58 ) was stereotaxically delivered to the TN of Sst-Cre mice (Figure 2A).Bilateral elimination of Sst expression was validated by in situ hybridization and pixel intensity quantification (t(5.24)= 5.44, p = 0.00247; Figure 2B).While ablation was largely targeted to the TN, there was also a significant decrease in Sst stain in the ARC (ablation-by-region interaction: F(3,36) = 7.31388, p = 0.001; post hoc TN t(5.24) = 5.44, p adj = 0.00988; post hoc arcuate t(10) = 3.52, p adj = 0.0167; Figure S1A).Mice were subjected to two 96-h food assays along with a battery of other metabolic tests.Final food intake, accounting for spillage, is depicted as an average over 24 h (Figure 2C).ANOVA revealed an overall effect of sex where males consume more food than females, as expected (F(1,38) = 14.1896, p = 0.0006).
The female-specific effect of SST neuron ablation is modulated by estrous cycle stage.There was a significant interaction between SST neuron ablation and estrous stage (F(1,45) = 8.1581, p = 0.0065) which was predominantly due to a decrease in food intake specifically during the night of proestrus (t(5.902)= -2.6044,p = 0.04104, Figure 2D).Within-subjects analysis of mice in the neuronal ablation group suggested that nighttime food intake during proestrus was also slightly lower than consumption during metestrus (t(7) = -1.976,p = 0.0887) and diestrus (t(7) = -2.3276,p = 0.0528), although these effects across the cycle did not reach statistical significance.
Despite effects on food intake, no other metabolic measures were altered by TN SST neuron ablation (see Table S1 for statistical results).SST neuron ablation did not affect telemetry measures of activity/movement (Figure S1C) or core body temperature (Figure S1D), or response to fasting glucose tolerance test (Figure S1E).SST neuron ablation did not affect body mass (Figure S1F), suggesting that the selective decrease in food intake during proestrus was not sufficient to alter body mass.
Ovariectomy to test the influence of ovarian secretions failed to replicate the effects of SST neuron ablation.We did not detect an interaction between ablation and gonad status (F(1,69) = 0.0109, p = 0.9173, Figure S2B), but overall food intake was lower in ovariectomized mice (F(1,69) = 4.4746, p = 0.038, Figure S2B).Body weight was not associated with food intake in either ablated or control animals (24 h: F(1,34) = 0.0076, p = 0.9311; Nightly: F(1,34) = 1.1595, p = 0.289147; Figure S2C), though examining nightly food intake only revealed an effect of ablation status on food intake overall (F(1,34) = 8.4044, p = 0.006512; Figure S2C).Given the influence of body weight on the effect of SST ablation (Figure 3A), it is possible that the lack of an effect in sham mice was at least partially due to higher body weight and altered metabolic profiles in this experiment.

TN SST neurons are sensitive to estrogens and adiposity signals
To test if TN SST neurons, specifically, can respond to estrogens and/or signals from white adipose tissue, we profiled the transcriptome of fluorescently labeled TN SST neurons using flow cytometry followed by bulk RNA sequencing (Flow-Seq; Figure 4A).Transcriptomic analysis of isolated TN SST neurons uncovered numerous differentially expressed genes between females and males, with the gene for estrogen receptor alpha, Esr1, being more highly expressed in females (Wc 2 = 6.736, adj p = 2.356 3 10 À8 ; Figure 4B).However, we were unable to detect ERa immunoreactivity in the TN using standard antibodies (Millipore Sigma 06-935).Instead, we confirmed expression from the Esr1 locus through the injection of a Cre-dependent tdTomato reporter into the TN of female and male Esr1-Cre mice.Subsequent colocalization of tdTomato with Sst via in situ hybridization revealed co-expression of Esr1 in approximately 10% Sst-expressing cells across sexes (Figures 4C and 4D), confirming that at least a subset of TN SST neurons is sensitive to estrogens.
To determine if TN SST neurons communicate with adipose tissue or vice versa, we used a co-correlation analysis method based on genetic variation.High-expressing genes from TN SST neurons (based on counts > glial fibrillary acidic protein expression) were used as the ''target'' pathways for human orthologs within the GTEx database 60 and subjected to cross-tissue genetic co-correlational analyses 61,62 (Figure 5A).As previous data indicated that TN SST neuron responsivity to metabolic cues might be localized to periods of higher circulating estradiol (Figure 2D) and TN SST may be able to directly sense circulating estradiol levels (Figure 4), individuals in the GTEx database were binned into groups with either ''high'' or ''low'' circulating estradiol levels via weighted aggregation of pan-tissue Z scores corresponding to estrogenresponsive gene expression (Figures S3A-S3D).Binning individuals into groups with indicators of ''low'' or ''high'' estrogen signaling revealed weakly associated groups (as per biweight midcorrelation 63 ; bicor coefficient = À0.32,p = 0.0023), suggesting that strong cross-tissue interactions differed depending on estrogen signaling status.Next, genetic co-correlation analyses from adipose (subcutaneous & omental), skeletal muscle, stomach, and small intestine to hypothalamic highly expressed genes were conducted.A lack of significant co-correlations with small intestine and several other tissues resulted in these tissues being omitted from the rest of analyses.A contributing factor for this lack of significance may be the limited number of individuals with matching tissue expression (Figure S3E).Given that subsets of strong cross-tissue correlations remained between the other tissues and highly expressed hypothalamic gene orthologs, relevant pathways which might contribute to signaling were examined accordingly (Figures 5B-5D).Significant interactions for co-correlations between estrogen signaling group and tissue were observed across tissues for all secreted proteins (Kruskal-Wallis test for interaction between estrogen category + tissue (D) Co-correlations across tissues showed differential impact on genes associated with feeding pathways.Individuals with higher estradiol showed increased cocorrelations within these pathways with adipose tissue and decreased co-correlations with stomach as compared to those with lower estradiol levels.(B)-(D) depict violin density plots with superimposed box-and-whisker plots denoting medium, first and third quartiles, and minimum and maximum values.Nonparametric ANOVA and post hoc t test p values reported on graphs.p = 1.7 3 10 À218 ; Figure 5B), known ligands (Kruskal-Wallis interaction p = 9.9 3 10 À20 ; Figure 5C), peptide hormones (Kruskal-Wallis interaction p = 1.8 3 10 À13 , data not shown), and feeding behavior pathways (Kruskal-Wallis interaction p = 0.0023; Figure 5D).In addition, several pathways showed specificity in strength of co-correlations from one tissue to another.For example, individuals in the higher inferred estrogen signaling group exhibited higher co-correlations between TN SST and adipose within secreted proteins (p < 2.2 3 10 À16 ), ligand (p = 0.025), and feeding behavior pathways (p = 0.042) as compared to individuals in the low-estrogen signaling group.Interestingly, this relationship was reversed for all secreted proteins in skeletal muscle, with individuals in the low-estrogen signaling group exhibiting higher co-correlations (p = 4.2 3 10 À12 ).These results indicate increased communication between adipose and TN SST neurons during periods of high estrogen signaling, and a switch to skeletal muscle communication when estrogen signaling is low.Across all gene sets, individuals with inferred low estrogen signaling exhibited higher co-correlations with stomach as compared to individuals with high estrogen signaling (all secreted proteins: p < 2.2 3 10 À16 ; ligands: p = 1.2 3 10 À5 ; peptide hormones: p = 1.8 3 10 À4 ; and feeding behavior: p = 0.029).Together, these human genetic co-correlation data indicate that TN SST neurons modulate feeding pathways through preferential communication with adipose when estrogen signaling is high and stomach hormones when estrogen signaling is decreased.These observations are in line with known responsivity of TN SST neurons to stomach peptide hormone and known regulator of feeding ghrelin, 43 but indicate that this communication pathway may be more salient when estradiol levels or estrogen signaling is low.Further, these analyses of human data may suggest a similar integration of metabolic cues alongside reproductive hormones in humans as well as mice.
Fat transplant also increased food intake in general (F(1,25) = 52.524,p < 0.0001), and the effect of SST neuronal ablation was affected by fat transplant (F(1,25) = 26.660,p < 0.0001; Figure 6D).Post hoc t tests revealed that SST neuronal ablation significantly decreased food intake in sham transplant animals (t(11.646)= -2.917,p = 0.01327) but significantly increased food intake in animals receiving fat transplant (t(4.8.2536) = 4.8427, p = 0.001175), similar to the previous relationship with body mass in proestrus (Figure 3A).However, we were unable to detect a significant interaction with fat transplantation and ablation status over the estrous cycle (Figure 6E), possibly due to high variability in nighttime food intake in sham-transplanted mice.Together, these findings indicate that fat transplantation masks the effect of SST neuron ablation and reveal a role for fat mass in modulating the function of tuberal hypothalamic SST neurons within the feeding circuit.

DISCUSSION
These data suggest that tuberal hypothalamic SST neurons are a locus in the brain that mediates metabolic and reproductive crosstalk.While activation of TN SST neurons increases food intake across sexes, permanent inactivation of ARC SST and TN SST neurons by ablation during adulthood results in decreased food intake only in females during the proestrus phase.This effect depends on body mass, as it is apparent only in lighter animals.In wild-type mice, body mass inversely correlates with food intake on the night of proestrus, but SST neuron ablation uncouples this relationship.Further analysis reveals that white adipose tissue abundance is a significant contributing factor.Not only does postmortem adipocyte size correlate with body mass in neuron ablation experiments, but also fat transplantation studies confirm that SST neuron ablation only decreases food intake in lean animals compared to their fat-transplanted counterparts.An interaction between cycling adipokines and gonadal hormones may be mediated by the direct effects of these circulating molecules on SST neurons, as these cells show some estrogen sensitivity via co-expression analyses.Furthermore, co-correlations between the hypothalamus and adipose tissue in humans and fat transplantation experiments in mice point to the importance of secreted proteins and ligands, suggesting SST neurons may detect and respond to adipokines.Future studies are needed to confirm and dissect the mechanisms of these cellular effects.
What adipokine factor is possibly being detected by the SST neurons (and specifically by the TN SST population) remains to be determined.6][67] Adiponectin, an adipokine that negatively correlates with visceral fat mass in mammals, 64 has long been shown to downregulate reproduction through direct impacts on the hypothalamus. 68Resistin is correlated with higher overall adiposity, 69 exhibits numerous interactions with the hypothalamic-pituitary-gonadal axis, [70][71][72] and is downregulated during the fertile periods of the mouse estrous cycle. 73egardless of adipokine contributor, this crosstalk paradigm provides a plausible explanation for the varied effects of estradiol on food intake in mice.9][80][81][82][83] Indeed, in our hands, ovariectomy decreased food intake across ablation groups, as has been seen previously in mice. 81,83Instead, the more consistent phenotype in mice is a decrease in energy expenditure following estradiol depletion. 10,12,13,83In light of this study, it is possible that the effects of estradiol on feeding across mouse studies, as observed by either endogenous estrous cycle fluctuations or ovariectomy manipulation, could be confounded by body mass and adiposity.Thus, factors like age at time of experiment, differences in fat distributions between species or strains, diet, or ovariectomy, and time from ovariectomy to estradiol replacement might present confounds based on changes to fat and/or lean mass.
How circulating estrogen levels contribute to this circuit also requires further investigation.Our human GTEx analyses show that co-correlations between genes expressed in TN SST and that in peripheral tissue shift from predominantly adipose-based in high estrogen conditions to skeletal muscle-and stomach-based in low estrogen signaling conditions (Figure 5).This suggests that higher estrogen levels may increase communication between TN SST neurons and white adipocyte depots, particularly as it relates to regulation of feeding behavior (Figure 5D).While this could be due to the actions of circulating estrogens on white adipose tissue itself (reviewed by Palmer and Clegg and Hevener et al. 1,84 ), it is also possible that estrogens directly act on TN SST (and/or ARC SST ) neurons to increase their sensitivity and/or responsivity to adipokines.Indeed, a subset of TN SST neurons exhibit estrogen sensitivity (Figure 4), and ARC SST neurons have been found to be estrogen sensitive 47 and responsive 48 in non-rodent models.Future studies would be needed to test for a possible direct effect on either of these populations in mice.Previous studies of the TN include both female and male mice but do not report Esr1 expression or estrogen sensitivity in Sst+ neurons. 42,43However, the idea that estrogen receptor is expressed in the TN and that the function of TN neurons is sensitive to estrous cycle stage is consistent with spatial mapping of Esr1 + neurons to a continuous region that spans the ventrolateral region of the ventromedial hypothalamus and the TN. 85It is also plausible that the concomitant decrease in effective communication with the stomach in high-estrogen signaling states is due to an interaction of estradiol with the hunger hormone ghrelin.TN SST neurons are directly responsive to ghrelin, 43 while estradiol can blunt the orexigenic effects of ghrelin via reduced ghrelin treatment efficacy 77 and reduced release of the hormone. 86Thus, estradiol may indirectly influence TN SST activity and subsequent food intake via regulation of ghrelin release and/or signaling.
Estradiol might also be detected elsewhere in the brain and impact TN SST and ARC SST neuronal modulation of feeding through integration at the circuit level.TN SST neurons project to many estrogen-sensitive nodes or nodes receiving direct input from estrogen-responsive regions, including the bed nucleus of the stria terminalis, parabrachial nucleus, and central amygdala. 43Importantly, bifurcated TN projections to the paraventricular nucleus and bed nucleus of the stria terminalis have been shown to mediate the feeding effects of TN SST neurons. 43As a known target for arcuate AgRP neurons-cells which have been shown to integrate reproductive and metabolic cues 31 -the paraventricular nucleus may contribute to the integration of metabolic and reproductive cues as a downstream integrative node.However, given that ablation of TN SST neurons eliminated the changes in the relationship between body weight and food intake over the estrous cycle, it is plausible that nodes of estradiol detection are TN SST neurons and/or neurons upstream of the TN.This circuit-wide integration of estradiol is a known mechanism of action for the gonadal hormone, with estrogens acting on many circuit nodes to coordinate behavioral output in a variety of cases, including reward/addiction 87 and thermoregulation. 88It is therefore probable that the effects of estradiol on feeding function similarly, as the anorexigenic effects of estradiol have been localized to numerous feeding nodes such as the hypothalamic arcuate nucleus [89][90][91][92][93] and the nucleus of the solitary tract of the brainstem. 94,95Retrograde tracing studies would help determine if these or estrogen-sensing regions (including the ventral subiculum, an upstream region targeting TN SST neurons 44 known to express estrogen receptors 96,97 ) are candidate nodes mediating the effects of estradiol seen in this study.
Alternatively, ARC SST neurons may detect fluctuating hormone levels to modulate the observed phenotype.In ewes, a subset of ARC SST neurons colocalize with ESR1 and increase their activity in response to estradiol treatment. 47Should this pattern of activation be confirmed in mice, it would suggest a mechanism whereby activation of ARC SST neurons during high periods of estradiol induces feeding under low adiposity conditions.
In all, this study adds to the growing literature interrogating the contributions of TN SST neurons to feeding behavior.Central SST (originally named growth hormone inhibiting hormone in the central nervous system) 98 had long been known to affect food intake through somatostatin receptor 2. 33,[37][38][39][40][41]45,[99][100][101][102] This effect was seemingly localized to the lateral tuberal nucleus, where TN SST neurons were found to integrate into the melanocortin feeding system, though the effect of these neurons on feeding was attributed to g-aminobutyric acid (GABA) release as opposed to direct SST effects. 43 Howeve, given that dynamics of GABA, NPY, and AgRP release from arcuate neurons in the temporal regulation of food intake, 103 it is possible that similar dynamics are apparent between GABA and SST in the TN.In line with previous studies, 42,43 chemogenetic activation of TN SST increased food intake across animals.Lack of excess crumbs in subsequent caspase ablation experiments confirmed that this is not a side effect of gnawing; a possibility also excluded in a previous study looking at wood stick gnawing.43 However, a subsequent study did see an increase in gnawing behavior with chemogenetic activation, probably due to a larger target area encompassing large portions of the lateral hypothalamus.42 Furthermore, whereas caspase ablation did decrease food intake, previous studies pooled across the sexes.43 Similar pooling in our hands also resulted in an overall decrease in food intake with SST neuron ablation, indicating that females possibly carry the effect when there is low power to detect sex differences.Alternatively, the inclusion of ARC SST neurons in our ablation may have also resulted in the observed sex difference and integrative capacities of tuberal hypothalamic SST neurons at large.
In our hands, this ablation-mediated decrease in food intake did not result in any alteration to body weight.This is in line with a previous study demonstrating that while the decrease in food intake may be due to ablated animals gaining weight more slowly, there are no raw body weight differences that result. 43Previous studies have also been conflicted on the role of TN SST neurons on movement and locomotion.While one study demonstrated a decrease in total locomotion and increase in vertical movement with TN SST neuron ablation, 43 another study found no change in walking and showed an increase in vertical movement with TN SST neuron activation. 42Our telemetry results suggest a lack of change in movement in general with SST neuron ablation.The differences among these studies may be due to the methods of movement quantification and/or viral spread of ablation.Similarly, while our study did not measure drinking behavior, it is unclear whether and how TN SST neuron ablation is involved in thirst processes, with one study demonstrating the necessity of TN SST neurons to maintain baseline drinking levels 43 but another failing to show TN SST neuron activation as sufficient to induce increases in drinking. 42N SST neurons were also found to contribute to food context learning in males (N.B.: in all external papers discussed, no definitions for sex category were ever provided.In mice, we assume that sexes were defined using anogenital distance.), 44indicating that the TN may straddle homeostatic and hedonic feeding mechanisms. 14And although many of these previous studies include female and male mice, they do not report Esr1 expression or estrogen sensitivity in TN SST neurons.Thus, the current paper adds to this growing literature by not only delineating an apparent sex difference but also context dependence in tuberal hypothalamic SST neuronal modulation of food intake.
We further speculate that tuberal hypothalamic SST neurons serve as a nexus of integration and a mediator of reproductive and metabolic interactions within the feeding circuit.In cycling rodents, fertile periods during the estrous cycle are accompanied by alterations to metabolic output, including a decrease in food intake, [74][75][76]104 increase in locomotion, [104][105][106][107] and increased core body temperature.105,107 These changes are hypothesized to suppress energy intake and promote active mate-seeking behavior and sexual receptivity. This stdy identifies tuberal hypothalamic SST neurons as possible mediators of such interactions, actively promoting energy intake during fertile periods when metabolic reserves may be insufficient to support reproduction.
for the Care and Use of Laboratory Animals of the National Institutes of Health.UCLA is AALAS accredited, and the UCLA Institutional Animal Care and Use Committee (IACUC) approved all animal procedures.

Estrous cycle staging
Vaginal lavages were performed on females daily, between ZT 0 and ZT 4, using 30 mL of standard phosphate buffered saline (PBS).Samples were deposited onto slides allowed to dry prior to staining.Males were subjected to similar handling during this time to ensure roughly equivalent handling stress.Giemsa staining was carried out to visualize cellular composition of the vaginal cavity.Stock Giemsa stain was prepared at least one week in advance of use.An 18.5% solution of Giemsa powder (Fisher G146-10) in glycerin was heated to 60 C and cooled before diluting 9:14 with 100% methanol.Stock was diluted 1:30 in PBS before use, shaking vigorously before stain.Slides were incubated for one hour at room temperature.Prior to staining, slides were briefly fixed in 100% methanol.Staging was assessed via light microscopy to determine the relative abundance of leukocytes, nucleated epithelial cells, and cornified epithelial cells 108 and stages were assigned with morning samples indicating the prior night's estrous stage, to align the cytology stage to the behavioral changes that occur in the dark (night) of estrus. 109This staging method was confirmed by examining patterns of core body temperature during the dark phase across the estrous cycle, which also changes during estrus. 105

Surgical procedures
Mice received analgesics (0.074 mg/kg buprenorphine two times daily, 7.11 mg/kg carprofen one time daily) on the day of and one day postsurgery.Mice were anaesthetized with 3% isoflurane and maintained within a range of 1.25-2.5%.AAVs were bilaterally injected into the TN of adult mice (coordinates relative to Bregma: A-P -1.65 mm, lateral G0.75, D-V -5.45; scaled when Bregma-Lambda distance was not equivalent to 4.2 mm) at a rate of 5 nL/s using a glass-pulled needle.See below table for titers and injection volumes.Controls consisted of both wildtype animals injected with the experimental virus (virus controls) and Cre positive animals injected with cell-filling GFP (genotype controls).Ovariectomy surgeries included complete removal of gonads from adult mice.Gonadectomies occurred immediately prior to stereotaxic viral injections within the same surgical period.In telemetry experiments, G2 eMitters (Starr Life Sciences) were implanted intraperitoneally on the same day as viral injection.Experiments were conducted following at least two weeks recovery from surgical proceedings.

Caspase ablation experiments
Gross movement and core body temperature were passively measured every other week for eight weeks using VitalView software (Starr Life Sciences).Body weight was measured every week.Food assay was performed when mice were not on telemetry pads.At ZT 0.5 on the start day of the experiment, 2/3 of the non-caloric paper bedding was removed.A pre-measured amount of food was delivered, and mouse body weight measured.Food in hopper was weighed at ZT 0.5 and ZT 11.5 every day until experiment conclusion.After 96 hours, food and all bedding, including interspersed crumbs, were collected.Measurements for total food were obtained by subtracting the final weight of food pellets and crumbs from the original weight of food provided.Measurements for crumbs were obtained by subtracting the final weight of food pellets and crumbs from the final weight of food in the cage hopper.Both measurements were divided by 4 to obtain the average food eaten or crumbs produced per day (24 h food intake).For estrous cycle experiments, 13-hour nightly food intake (ZT 11.5 -ZT 0.5) was utilized instead as we were more confident in nightly staging based on the behavioral changes that occur during the night of estrus 109 and our ability to confirm estrus staging with core body temperature in the dark phase of estrus. 105For some experiments, 4-5 hour fasted glucose tolerance tests were performed prior to sacrifice.In ovariectomy experiments, two food assays were performed back-to-back, non-fasted resting glucose levels were collected, body composition was measured via NMR, and indirect calorimetry was performed in Oxymax metabolic chambers (Columbus Instruments) at room temperature.Upon experiment completion, all brains were collected using RNase-free conditions.Inguinal white adipose tissue (iWAT) and gonadal white adipose tissue (gWAT) were collected for histology analyses.

Transient activation food intake assay
Clozapine-n-oxide (CNO; MilliporeSigma #0832) was used to activate TN SST neurons in Sst-Cre animals expressing hM3Dq-mCherry.Stock solution of 20 mg/mL in DMSO was stored at -20 C and diluted to a working solution of 0.03 mg/mL in sterile saline also stored at -20 C. Saline control (0.15% DMSO) or CNO (10 mL/g body weight, dose of 0.3 mg/kg) solution were administered IP in a counterbalanced design.Experiments were completed in duplicate replicate trials.Mice were transferred to experimental room at least 15 minutes prior to experimentation.Experiments were begun between ZT 2-3 and terminated between ZT 6-7.Following injection, food intake was measured at 0.5, 1, 2, and 4 hr.Vaginal lavage was performed on female mice after experiment conclusion to prevent stress interference with food intake.All mice were injected with CNO 90 minutes prior to sacrifice to enable neuronal activation validation via cFOS immunohistochemistry.

Fat transplantations
Donor fat was taken from various visceral (i.e., periuterine perigonadal, retroperitoneal, and omental) depots of wildtype female C57BL/6J mice and implanted into female mice recently stereotaxically injected under standard surgical conditions.Four depots of 0.15-0.25gwere placed subcutaneously on the dorsal surface through a single incision mid-back, for a final transplantation total of 0.6-0.9g of white adipose.Fat for each depot was divided into at least three individual pieces to promote vascularization.The visceral-to-subcutaneous paradigm was used due to the deleterious metabolic effects of this graft. 110Food intake was assayed 2-3 weeks following transplantation to allow for sufficient angiogenesis 111 and graft stabilization without endogenous fat depot compensation. 112Upon sacrifice, grafts were examined to confirm tissue was not necrotic.

Histology
In situ hybridization (ISH) and immunostaining (IHC) Sst sense and antisense probes were transcribed using a DIG or FITC RNA labeling kit (Roche) and purified with RNA Clean & Concentrator (Zymo Research).PCR products were amplified using Allen Brain Institute-derived reference primer sequences and cloned into pCR

DREADD viral injection mapping
Sections were anatomically matched using the Allen Mouse Brain Atlas (mouse.brain-map.org), and the anatomical areas with hM3Dq-mCherry positive cells were manually identified and outlined in Illustrator.Areas containing dense hM3Dq-mCherry expression were outlined with 5% opacity, and areas containing sparse hM3Dq-mCherry expression were outlined with 3% opacity.Thus, areas with more hM3Dq-mCherry expression across animals appear more opaque than areas with less hM3Dq-mCherry.

Caspase ablation quantification
Sections were anatomically matched using the Allen Mouse Brain Atlas (mouse.brain-map.org), and then the TN, ARC, lateral hypothalamic area (LHA), and ventromedial nucleus (VMH) were manually outlined in ImageJ.The mean pixel intensity was then measured for each region and normalized to the local background.The mean normalized Sst signal in each region was then compared using a 2-way linear mixed-effects model with ablation status (control vs ablated) and region (TN, ARC, LHA, and VMH) as factors and animal as a nested random effect in R (version 4.2.1) using the lme() function (nlme package, version 3.1-157).Following a significant (p<0.05)ablation status x region interaction, we performed multiple t-tests to quantitatively compare Sst signal in each region individually.T-tests were performed and p-values calculated using the t_test() function (rstatix package, version 0.7.0), using the Holm method to correct for multiple comparisons.

Figure 2 .
Figure 2. Caspase ablation of SST neurons decreased food intake only in females (A) Schematic of experimental paradigm.AAVs encoding taCasp3-TEVp or GFP within flip-excision (FLEX) cassettes were injected bilaterally into the TN of Sst-Cre or wild-type mice.Created with BioRender.com.(B) Representative bright-field images and quantification of Sst transcript expression in the TN of caspase-ablated and control animals.Dotted line indicates the boundary of the TN.Scale bar = 100 mm.Viral spread to ARC is quantified in Figure S1A.(C) Permanent SST neuronal ablation decreases average daily food intake in females but not males.F Control n = 10; M Control n = 11; F Ablated n = 11; M Ablated n = 10.(D) This decrease in food intake is detected only in the night of proestrus.Proestrus: Control n = 6, Ablated n = 8; Estrus: Control n = 7, Ablated n = 7; Metestrus: Control n = 10, Ablated n = 9; Diestrus: Control n = 10, Ablated n = 9.Mean G SEM; ANOVA and post hoc t tests where applicable; between subjects: *p < 0.05, **p < 0.01; within subjects: # p < 0.10.

Figure 4 .
Figure 4.A subset of TN SST neurons is sensitive to estrogens (A) Schematic of Flow-Seq analysis (also applies to Figure 5 genetic input).AAV-flex-tdTomato was stereotaxically injected into the TN of Sst-Cre mice.Lateral hypothalamic region was grossly dissected and enzymatically dissociated before segregating Sst + red neurons via flow cytometry.Resultant cells were sent for bulk RNA sequencing.Created using BioRender.com.(B) Numerous genes are more differentially expressed between females and males (black dots), including canonical Y-associated genes (Ddx2y, Eif2s3y, Uty, and Kdm5d; gray text) and Esr1 (purple dot).(C) Representative fluorescent images from Esr1-Cre mice showing colocalization of Sst (magenta), Esr1::tdTomato (yellow), and DAPI nuclear stain (cyan).White arrows indicate Sst-Esr1 double-labeled cells.Scale bar = 100 mm.(D) Quantification of colocalization confirms Esr1/Sst co-expression but reveals no sex difference.Mean G SEM.

Figure 5 .
Figure 5. TN SST neurons display increased hormonal pathway co-correlations with white adipose tissue in individuals with inferred high estrogen signaling (A) Schematic overview of co-correlation analysis.High-expressing TN SST genes from mouse Flow-Seq experiments were co-correlated across various peripheral metabolic tissues across high-and low-estradiol groups identified in the GTEx database.Created with BioRender.com.(B) Inferred estradiol levels affected co-correlation pathways relevant to all secreted proteins.High-estradiol individuals showed increased co-correlations in adipose tissue whereas those with lower estradiol showed increases in skeletal muscle and stomach communication.(C) Similar trends in adipose and stomach co-correlations across estradiol groupings were seen for ligand pathways.(D)Co-correlations across tissues showed differential impact on genes associated with feeding pathways.Individuals with higher estradiol showed increased cocorrelations within these pathways with adipose tissue and decreased co-correlations with stomach as compared to those with lower estradiol levels.(B)-(D) depict violin density plots with superimposed box-and-whisker plots denoting medium, first and third quartiles, and minimum and maximum values.Nonparametric ANOVA and post hoc t test p values reported on graphs.
2.1 TOPO (Invitrogen).Plasmid DNA was then isolated from bacterial cultures (ZymoPURE II Plasmid Midiprep kit), linearized, and purified (Zymo DNA Clean & Concentrator).Validation of caspase ablation was carried out on 35mm-thick coronal slices via chromogen ISH using BCIP (5-bromo-4chloro-3-indoyl phosphate) and INT (iodonitrotetrazolium).Validation of hM3Dq targeting and activation was accomplished by visualization of native mCherry expression and IHC stain for cFOS.Briefly, slides were blocked and incubated with rabbit anti-cFOS (1:200, Synaptic Systems # 226003, RRID: 2231974) primary antibody overnight at 4 C.The next day, sections were incubated for 1 hour at room temperature with goat anti-rabbit Alexa Fluor 488 secondary (1:500, Thermo Fisher Scientific # A11034, RRID: AB_2576217) and counterstained with DAPI.For colocalization experiments, Esr1-Cre mice were bilaterally injected with 400 ml AAV2-flex-tdTomato into the TN coordinates.Native tdTomato fluorescence destroyed by combined ISH protocol was recovered by rabbit anti-DsRed (1:1000, Takara Bio Clontech # 632496, RRID: AB_10013483) antibody and switched to the green channel using an Alexa Fluor 488 secondary.Dual Sst ISH & tdTomato IHC protocol was accomplished via TSA amplification.Briefly, 35 mm sections were fixed, permeabilized with Triton X-100, and acetylated before overnight ISH probe incubation at 65 C. The next day, tissue was then washed, blocked with Blocking Reagent (MilliporeSigma 11096176001Roche) and heat inactivated sheep serum, and incubated with anti-DsRed overnight at 4 C.The final day, tissue was washed before ISH signal was developed with the TSA Plus Cyanine 5 System (Akoya Biosciences # NEL745001KT).Slides were then stripped of horseradish peroxidase and blocked with normal goat serum before incubating with goat anti-rabbit Alexa Fluor 488 (1:400) for 2 hours at room temperature.