Deletion of Growth Hormone Secretagogue Receptor in Kisspeptin Neurons in Female Mice Blocks Diet-Induced Obesity

The gut peptide, ghrelin, mediates energy homeostasis and reproduction by acting through its receptor, growth hormone secretagogue receptor (GHSR), expressed in hypothalamic neurons in the arcuate (ARC). We have shown 17β-estradiol (E2) increases Ghsr expression in Kisspeptin/Neurokinin B/Dynorphin (KNDy) neurons, enhancing sensitivity to ghrelin. We hypothesized that E2-induced Ghsr expression augments KNDy sensitivity in a fasting state by elevating ghrelin to disrupt energy expenditure in females. We produced a Kiss1-GHSR knockout to determine the role of GHSR in ARC KNDy neurons. We found that changes in ARC gene expression with estradiol benzoate (EB) treatment were abrogated by the deletion of GHSR and ghrelin abolished these differences. We also observed changes in metabolism and fasting glucose levels. Additionally, knockouts were resistant to body weight gain on a high fat diet (HFD). Behaviorally, we found that knockouts on HFD exhibited reduced anxiety-like behavior. Furthermore, knockouts did not refeed to the same extent as controls after a 24 h fast. Finally, in response to cold stress, knockout females had elevated metabolic parameters compared to controls. These data indicate GHSR in Kiss1 neurons modulate ARC gene expression, metabolism, glucose homeostasis, behavior, and thermoregulation, illustrating a novel mechanism for E2 and ghrelin to control Kiss1 neurons.


Introduction
Rodents have been known to exhibit torpor during periods of food scarcity to compensate for a negative energy balance via suppression of the hypothalamic-pituitary-gonadal (HPG) axis [1]. These periods induce a state of hypothermia and a reduction in metabolism that allow for survival. During these sensitive periods, reproduction becomes secondary to conserving energy. There are numerous neuroendocrine mechanisms that link reproduction with metabolism and thermoregulation; however, they are not well understood in most mammals. Ghrelin is one peripheral hormone that could link the neuromodulation of reproduction and metabolism during times of food scarcity [2][3][4]. Ghrelin is produced

Animals
All animal procedures were completed in compliance with institutional guidelines based on National Institutes of Health standards and were performed with Institutional Animal Care and Use Committee approval at Rutgers University. Adult mice were housed under constant photoperiod conditions (12/12 h light/dark cycle) and maintained at a controlled temperature (25 • C). Animals were given food and water ad libitum, unless noted otherwise. Animals were weaned at postnatal day 21 (PD21). Sexually mature Tac2-EGFP male mice, used previously [15,31,32], were used for cell harvesting experiments ( Figures S1 and S2).

Production of Kiss1 Cre/EGFP ;Ghsr fl/fl Mice
To produce a selective deletion of GHSR in Kiss1 neurons, we mated C57BL/6J-Kiss11tm1.1(cre/EGFP)Stei/J mice from Jackson Laboratory (#017701) to a floxed GHSR (GHSR fl/fl ). The GHSR fl/fl mouse strain provided by Dr. Yuxiang Sun (Texas A&M) were backcrossed over 10 generations. The Kiss1Cre/EGFP allele expresses a Cre-EGFP fusion protein from the Kiss1 promoter and enhancer elements. This Cre line has been used to develop other Kiss1-specific knockouts and knockins, as well as determine the molecular signaling properties of Kiss1 neurons and their role in GnRH, LH, AgRP, and POMC signaling [12,37,[39][40][41][42][43][44][45]. Cre-mediated recombination will result in deletion of GHSR from Kiss1-expressing neurons in offspring (Kiss1 Cre/+ /Ghsr fl/fl ). Littermates that did not express the Kiss1 Cre /EGFP allele (Kiss1 WT /Ghsr fl/fl ) were used as controls. Control and Kiss1-GHSR KO (knockout) females were used for all experiments, unless noted otherwise. We have verified the decrease in GHSR expression in ARC Kiss1 neurons by analyzing pools of Kiss1 neurons from Tac2-GFP females (+Ghsr expression) and Kiss1 Cre/+ /Ghsr fl/fl females (-Ghsr expression) (See Supplemental Methods for Cell Harvesting protocol and Figure S1) [46]. Only females were used in this study as we did not detect Ghsr in KNDy neurons in Tac2-GFP male mice using single cell RT-PCR nor did testosterone propionate increase Ghsr expression in these neurons using quantitative real-time PCR (See Supplemental Methods Cell Harvesting protocol and Figure S2).

Chemicals
For non-energy homeostasis experiments, an estradiol benzoate (EB) injection protocol was used that has previously been shown to alter gene expression in the hypothalamus and mimic a proestrus-like state in mice [20]. Briefly, animals were injected subcutaneously (s.c.) at 10:00 am on post-OVX/GDX day 5 with either 0.25 µg of EB or sesame oil-vehicle. On post-OVX/GDX day 6, a 1.5 µg dose of EB or sesame oil-vehicle was injected at 10:00 am. Ghrelin was purchased from AnaSpec Peptides (Fremont, CA, USA) and was dissolved in saline prior to storage at −20 • C until needed. Estradiol benzoate (EB) was purchased from Steraloids (Newport, RI, USA). Ketamine, Marcaine, and Rimadyl were purchased from Henry Schein Animal Health (Dublin, OH, USA). EB was dissolved in 100% ethanol prior to suspension in sesame oil. Ethanol was allowed to evaporate for 24 h prior to storage at 4 • C until needed. For energy balance experiments, diets were purchased from Research Diets (New Brunswick, NJ, USA): low-fat diet (LFD; 10% kcal fat; D12450B) and high-fat diet (HFD; 45% kcal fat; D12451).

RNA Extraction and Quantitative Real-Time PCR
To determine if ghrelin alters ARC gene expression (Experiment #1), Kiss1-GHSR KO and control mice were separated into 4 treatment groups: (1) OVX/Oil + ghrelin (12 h  RNA extraction, and qPCR protocols were the same as previously published [47][48][49]. Briefly, brain slices were collected using a brain matrix (Ted Pella, Redding, CA, USA) which allows for the cutting of the brain into 1 mm-thick coronal slices as previously described [50]. Brain slices containing the region of interest were transferred to RNALater (Life Technologies) and stored overnight at 4 • C. For model validation, in addition to the ARC, we also performed RNA extraction using the same technique on tissues including: Anteroventral Paraventricular Nucleus (AVPV), Bed Nucleus of Striata Terminalis (BNST), medial Amygdala (AMG), Ventral Tegmental Area (VTA), and Substantia Nigra (SubN; Figures S6-S8).
Pure RNA was extracted from all brain tissue using Ambion RNAqueous ® Micro Kits (Life Technologies, Carlsbad, CA, USA) according to the manufacturer's protocol, followed by DNase-I treatment to remove contamination by genomic DNA (Life Technologies). Quantity of RNA was assessed by the NanoDrop™ ND-2000 spectrophotometer (ThermoFisher, Inc., Waltham, MA, USA), followed by the RNA 6000 Nano Kit (Agilent Technologies, Inc., Santa Clara, CA, USA) to assess quality. Samples with an RNA integrity number (RIN) greater than 8 were used for quantitative real-time PCR (qPCR). As previously described [47] complementary DNA (cDNA) was synthesized from 250 ng of total RNA using Superscript III reverse transcription (Life Technologies). cDNA was diluted to 1:20 with nuclease-free water (Gene Mate/Bioexpress) for a final cDNA concentration of 0.5 ng/µL and stored at −20 • C. Untreated ARC tissue RNA was used for the calibrator and negative control (no reverse transcription) and processed simultaneously with the Kiss1-GHSR KO samples. All values were normalized and are expressed as relative mRNA expression. Efficiencies were calculated as a percent efficiency and are approximately equal (90-110% or one doubling per cycle, see Table 1 for a list of primer sequences). The reference genes used were Gapdh and Hprt. Positive, negative, and water blank controls were included in the qPCR plate design. The geomean of the Cq values from each reference gene was used to calculate relative gene expression, reference gene Cq values were not affected by the treatments [47,49]. Average Cq values are presented in Table S1.

Surgical Procedures
Adult females were bilaterally ovariectomized (OVX) under isoflurane anesthesia using a previously published sterile no-touch technique according to the NIH Guidelines for Survival Rodent Surgery [31,32,51]. Animals were given a dose of analgesic [4 mg/kg carprofen (Rimadyl ® )] one day following surgery for pain management. Animals typically lost 1-2 g of weight one day after surgery, which was regained by post-surgery day three.
For energy homeostasis experiments (Experiment #2), following OVX, females were separated into two treatment groups-sesame oil vehicle and estradiol benzoate (EB). To reduce constant stressful injections during the 8 weeks of diet, mice were passively orally dosed with sesame oil vehicle or EB (300 µg/kg body weight; suspended in sesame oil) mixed with powdered, peanut butter daily [31]. After sacrifice, uteri were weighed to confirm effect of treatment.

Food Intake, Body Weight, and Body Composition
To determine the influence of E2 (Experiment #2) and high-fat diet (Experiment #3) on the interaction of Kiss1 and GHSR, food intake and body weight were measured weekly. Body composition was measured using a small rodent MRI (EchoMRI, Houston, Texas). For experiment #2, females were OVX and received EB-or oil-treatment. For experiment #3, females remained gonadally intact.

Metabolic Monitoring
To determine the influence of E2 (Experiment #2) and high-fat diet (Experiment #3) on the interaction of Kiss1 and GHSR and metabolic outcomes, indirect calorimetry, food intake, and activity were measured using CLAMS chambers (Comprehensive Lab Animal Monitoring System, Columbus Instruments, Columbus, OH, USA). Mice were in the CLAMS chambers for 4 days total, the first 2 days of which were an acclimation period and the final 48 h were used for data analysis.

Glucose and Insulin Tolerance Tests
To determine the influence of E2 (Experiment #2) and high-fat diet (Experiment #3) on the interaction of Kiss1 and GHSR and glucose homeostasis, a glucose tolerance test (GTT) and insulin tolerance test (ITT) were performed a minimum of 4 days apart in the same animals. Each mouse received an intraperitoneal (IP) injection of glucose (2 g/kg) after a 5 h fast (09:00 am-02:00 pm) or insulin (0.75 U/kg body weight in sterile saline) after a 4 h fast (09:00 am-01:00 pm). Glucose was measured in tail blood using an AlphaTrak glucometer (Zoetis, Parsippany, NJ, USA). Glucose measurements were taken at 0, 15, 30, 60, 90, and 120 min after injection. Fasting blood glucose was determined from the baseline timepoint during GTT, before glucose administration.

Behavioral Tests
To determine the influence of high-fat diet on the interaction of Kiss1 and GHSR and locomotor and anxiety-like behavior (Experiment #4), we performed behavioral tests commonly used to assess avoidance behaviors, protocols used were previously published [51]. Tests were performed in the same mice with a 5 to 7 day rest period in between.

Open Field Test (OFT)
The OFT evaluates anxiety-like behavior in concert with spontaneous exploratory activity [52]. Mice were first subjected to the open field test consisting of a square white opaque arena area (40 cm long × 40 cm wide × 40 cm high, open-top) with a 64-square grid floor (8 × 8 squares, 5 cm/side). Each mouse was placed in the same (10 cm long × 10 cm wide) corner square of the arena and allowed to freely explore for 10 min. After each animal trial, the surface of the arena was cleaned with a sterilizing solution.

Elevated Plus-Maze (EPM)
Mice were next assessed on the EPM. The EPM assesses anxiety-like behavior and is often used in combination with the OFT [53]. The apparatus consisted of two sets of opposing arms (30 cm × 5 cm) extending from a central (5 cm × 5 cm) region. Two arms were enclosed (15 cm high walls) and two arms remained open. Animals were placed in the central area of the maze in the direction of a closed arm and allowed to freely explore both open and closed arms for 5 min. Following each animal trial, the surface of the arena was cleaned with a sterilizing solution.

Light/Dark Box (LDB)
The LDB is another task that evaluates anxiety-like behavior and spontaneous exploratory activity, but through a natural conflict situation between innate exploratory desire of novelty and tendency to avoid the unfamiliar [54]. The LDB assessment took place using a square white opaque arena (40 cm long × 40 cm wide × 40 cm high, open-top), with a black rectangle insert (40 cm long × 20 cm wide × 40 cm high, closed top) that contained an opening (7.5 cm × 7.5 cm) located at floor level in the center of the partition. Each mouse was placed in the same corner of the light side of the apparatus and allowed to freely explore for 10 min. After each animal trial, the surface of the arena was cleaned with a sterilizing solution.

The Effects of Fasting on Meal Patterns
To determine the influence of fasting (Experiment #5) on the interaction of Kiss1 and GHSR and feeding behavior, Kiss1-GHSR KO and control females were placed in the biological data acquisition system (BioDAQ; Biological Data Acquisition, Research Diets, New Brunswick, NJ, USA). Mice were fed a normal chow diet and were allowed to acclimate to single housing for 4 days followed by a 3-day acclimation to the BioDAQ. After acclimation, 1 h before lights out, half of the mice were fasted for 24 h and the other half were fed ad lib. After 24 h, food was returned to the fasted mice and meal patterns were recorded, uninterrupted for the next 48 h.

Core Body Temperature Recording
To determine the influence of mild cold stress (Experiment #6) on the interaction of Kiss1 and GHSR, core body temperature was recorded using temperature loggers (DST nano, Star-Oddi, HerfØlge, Denmark) implanted abdominally, utilizing similar methods as previously published [55]. Loggers were placed in the abdominal cavity using similar surgical technique as OVX and recorded temperature every five minutes until removed. After a 1-week recovery from surgery, mice were placed into the CLAMS and acclimated to the chambers for 3 days. After acclimation, recording began at 25 • C to access baseline. During the second night, the environmental temperature was reduced from 25 • C to 10 • C for a period of 6 h. Mice remained inside the CLAMS for an additional night at 25 • C. At the end of the experiments, the loggers were retrieved, and the data were downloaded and analyzed.

Statistical Analysis
All data were analyzed by a 2-way ANOVA (genotype and steroid/diet) followed by a post hoc Holm-Sidak's except for q-PCR data which were analyzed by multifactorial ANOVA (genotype and steroid) using Statistica (Dell, Round Rock, TX, USA). All gene expression data were normalized to control females for comparison across genotypes. For the behavioral studies, values that exceeded 2 SDs above or below the group mean were considered outliers and dropped. LH values (see supplemental methods) were analyzed by two-tailed unpaired Student t-test to compare LH pulse parameters between Kiss1-GHSR KO females in proestrus and diestrus, Kiss1-GHSR KO and control females in diestrus after a 24 h fast, Kiss1-GHSR KO females in diestrous with and without ghrelin, OVX Kiss1-GHSR KO females treated with EB or oil, OVX Kiss1-GHSR KO females treated with EB and treated with EB/ghrelin, and OVX/EB treated control and Kiss1-GHSR KO females with ghrelin. Results were considered statistically significant at p < 0.05.
To determine the interaction of E2 and ghrelin on ARC gene expression, we repeated the same experiment with the addition of a ghrelin injection (1 mg/kg; IP), given 12 h prior to sacrifice [56]. In Figure 1, expression of neuropeptide KNDy genes Kiss1 ( Figure 1G; steroid: p < 0.0001), Tac2 ( Figure 1H; steroid: p = 0.0002), Pdyn ( Figure 1I; steroid: p = 0.0006), and Tac3r ( Figure 1J; steroid: p < 0.0001; genotype: p = 0.0111, interaction of steroid*genotype: p = 0.0482) were reduced by EB and ghrelin in control females. Furthermore, EB reduced Esr1 expression ( Figure 1K; steroid: p = 0.0006) and abrogated the change in Ghsr expression ( Figure 1L; genotype: p = 0.0048) in controls. In contrast to gene expression without ghrelin ( Figure 1A-F), in Kiss1-GHSR KO females treated with EB and ghrelin, gene expression reduction was expanded beyond Kiss1 ( Figure 1G; steroid: p < 0.0001), and Tac2 ( Figure 1H; steroid: p = 0.0002) to include Pdyn ( Figure 1I; steroid: p = 0.0006), Tac3r ( Figure 1J; steroid: p < 0.0001; genotype: p = 0.0111; interaction of genotype*steroid: p = 0.0482), and Esr1 ( Figure 1K; steroid: p = 0.0006), while Ghsr remained unchanged ( Figure 1L; genotype: p = 0.0048). In addition, there were only genotype differences in oil and ghrelin-treated Kiss1-GHSR KO females for Tac3r ( Figure 1J), while ghrelin injection abrogated genotype differences seen previously with Kiss1, Tac2, Pdyn and Esr1. In addition, we examined ARC Npy, Agrp, Pomc, and Cart gene expression in the same experimental design revealing EB treatment enhanced Npy in KO females, increased Argp gene expression in both genotypes, reduced Pomc and Cart gene expression in control females, and reduced Pomc expression in KO females compared to controls treated with oil. Ghrelin administration caused an increase in Agrp gene expression in EB-treated controls but not KOs, while ghrelin + EB prevented an increase in Pomc and Cart expression in KO females only ( Figure S7). We also analyzed gene expression comparing ghrelin-injected and non-ghrelin injected females within each genotype ( Figure S4) and found a main effect of genotype in reduced Kiss1 expression in both oil and EB-treatment ( Figure S4A,B), Tac3r with oil-treatment ( Figure S4C), Tac2 with oil treatment ( Figure S4E), Pdyn with oil-treatment ( Figure S4I), and Ghsr with both oil and EB treatment ( Figure S4K,L). Furthermore, we found that ghrelin injection reduced Tac2 in oil-treated controls ( Figure S4E). Finally, we found that Tac3r, Tac2, Esr1, and Pdyn were reduced in oil-treated KO females compared to controls without ghrelin ( Figure S4C,E,G,I), while Ghsr was reduced oil-treated KO females with ghrelin compared to controls ( Figure S4K) and in EB-treated KO females without ghrelin compared to controls ( Figure S4L). These data indicate the importance of  To determine the interaction of E2 and ghrelin on ARC gene expression, we repeated the same experiment with the addition of a ghrelin injection (1 mg/kg; IP), given 12 h prior to sacrifice [56]. In Figure 1, expression of neuropeptide KNDy genes Kiss1 ( Figure 1G; steroid: p < 0.0001), Tac2 ( Figure 1H; steroid: p = 0.0002), Pdyn ( Figure 1I; steroid: p = 0.0006), and Tac3r ( Figure 1J; steroid: p < 0.0001; genotype: p = 0.0111, interaction of steroid*genotype: p = 0.0482) were reduced by EB and ghrelin in control females. Furthermore, EB reduced Esr1 expression ( Figure 1K; steroid: p = 0.0006) and abrogated the change in Ghsr expression ( Figure 1L; genotype: p = 0.0048) in controls. In contrast to gene expression without ghrelin ( Figure 1A-F), in Kiss1-GHSR KO females treated with EB and ghrelin,

Fertility and LH Pulsatility
Because ghrelin is known to suppress the frequency of pulsatile LH release in male and female rats, in part, due to activation of β-endorphin release from POMC neurons, which do not express GHSR at significant levels [7][8][9][10][11]45], we examined fertility and LH pulsatility in our females (See Supplemental Methods for Vaginal Cytology and Fertility assessments). Fertility was no different between control and KO females, the average number of days from pairing to birth was 23 ± 2 days, average litter size 6 ± 1, male to female pup ratio was equal. Vaginal opening occurred in control females at 34.5 ± 6.5 days of age and KO females were 36.5 ± 5.1 days of age. We also observed no disruption in estrous cycle. Control females were, on average in Diestrous/Metestrus for 3.54 ± 1.48 days, Proestrous for 1.63 ± 0.71 days and Estrous for 1.67 ± 0.96 days while KO females were, on average in Diestrous/Metestrus for 3.33 ± 0.98 days, Proestrous for 1.71 ± 0.58 days and Estrous for 2.04 ± 0.81 days (data not shown). LH dynamics were examined in both Kiss1-GHSR KO and control females while gonadally intact, after OVX with or without E2 replacement, and after fasting or ghrelin injection (1 mg/kg; IP) 30 min prior to the collection (See Supplemental Methods for LH Blood Collection). For intact females, there were no differences in the total number of LH peaks, LH concentration, peak amplitude, or interpulse interval. After OVX, Kiss1-GHSR KO females treated with oil had a significantly higher number of LH peaks, LH concentration and LH peak amplitude compared to EB-treated females ( Figure S3).

Body Weight Gain and Body Composition
EB treatment reduced body weight gain, regardless of genotype ( Figure 2A; steroid: F(3,28) = 9.571, p = 0.0002). At the end of the 8 weeks of treatment, EB-treated females weighed less than oil-treated ( Figure 2B; steroid: F(1,28) = 33.15, p < 0.0001). EB treatment also reduced the percent change in fat mass ( Figure 2C; steroid: F(1,28) = 35.10, p < 0.0001) and increased the percent change in lean mass from week zero to week 8 regardless of genotype ( Figure 2D; steroid: F(1,28) = 53.41, p < 0.0001). There were no differences observed in cumulative food intake in pair-housed females over 8 weeks of diet (data not shown). Serum concentration for Ghrelin and Insulin were measured in control and KO females treated with oil or EB (See Supplemental Methods for Serum Measurements). We found ghrelin was elevated in EB-treated-KO females compared to EB-treated controls and insulin was reduced in KO females treated with EB compared to oil ( Figure S9A,B). Overall, we observed only steroid effects on body weight and body composition.

Glucose and Insulin Tolerance Tests
Kiss1-GHSR KO females consistently exhibited elevated fasting glucose in response to a 5 h fast compared to controls, regardless of treatment ( Figure 2E; F(1,23) = 5.349, p = 0.0300). We observed no differences in GTT or ITT ( Figure 2F; time: F(5,140) = 105.65, p = 0.0000; Figure 2G; time: F(5,115) = 30.860, p = 0.0000). We observed no significant differences in area under the curve analysis of GTT or ITT (data not shown). The observed difference in fasting glucose indicates that Ghsr expression in Kiss1 neurons participates in the regulation of glucose homeostasis.

Feeding, Locomotor and Metabolic Behaviors in CLAMS
EB-treated females consumed more food at night compared to the day. In addition, EBtreated controls consumed less than EB-treated Kiss1-GHSR KO females at night ( Figure 3A and increased the percent change in lean mass from week zero to week 8 regardless of genotype ( Figure 2D; steroid: F(1,28) = 53.41, p < 0.0001). There were no differences observed in cumulative food intake in pair-housed females over 8 weeks of diet (data not shown). Serum concentration for Ghrelin and Insulin were measured in control and KO females treated with oil or EB (See Supplemental Methods for Serum Measurements). We found ghrelin was elevated in EB-treated-KO females compared to EB-treated controls and insulin was reduced in KO females treated with EB compared to oil ( Figure S9A,B). Overall, we observed only steroid effects on body weight and body composition.  All females consumed more oxygen (V.O2) during the night compared to the day. EB-treated females consumed more oxygen during the day and night compared to oiltreated females and EB-treated controls consumed more oxygen compared to experiments during both the day and night ( Figure 3E Hourly V.O2 consumption was elevated in EB-treated controls compared to EB-treated Kiss1-GHSR KOs at 12:00 am, 03:00 am, 07:00 am, 12:00 pm, 04:00 pm, 09:00 pm, 10:00 pm, and 11:00 pm. In addition, oil-treated controls also exhibited an increase in V.O2 compared to oil-treated Kiss1-GHSR KOs at 12:00 am, 06:00 am, 11:00 am, 12:00 pm, 03:00 pm, 08:00 pm, 10:00 pm, and 11:00 pm ( Figure 3F All females produced more carbon dioxide (V.CO2) during the night compared to the day. EB-treated females produced more V.CO2 during the day and night compared to oiltreated females. Furthermore, EB-treated controls produced more V.CO2 compared to EBtreated Kiss1-GHSR KOs during the night ( Figure 3G  All females consumed more oxygen (V.O2) during the night compared to the day. EB-treated females consumed more oxygen during the day and night compared to oiltreated females and EB-treated controls consumed more oxygen compared to experiments Respiratory exchange ratio (RER), a measure of substrate utilization (fat vs. carbohydrates) was higher in EB-treated females at night compared to the day ( Figure 3I; steroid: F(1,51) = 5.0672, p = 0.02872; time F(1,51) = 27.088, p = 0.00000). Hourly RER was increased in EB-treated Kiss1-GHSR KO females compared to oil-treated Kiss1-GHSR KOs at 12:00 am and 07:00 pm, while oil-treated controls revealed a reduced RER compared to EB-treated controls at 02:00 am ( Figure 3J; time: F(23,391) = 9.1637, p = 0.0000). Higher RER values closer to 1.0 indicate a switch to carbohydrate utilization.
All females exhibited elevated heat production, a measure of energy expenditure, at night compared to the day. Additionally, EB-treated females exhibited an increase in nighttime energy expenditure compared to oil-treated females ( Figure 3K Serum concentration for ghrelin and insulin were measured in control and KO females fed on LFD or HFD. We found ghrelin was elevated in KO females compared to controls on both diets and insulin was reduced in KO females on a HFD compared to control females on a HFD ( Figure S9C,D). Overall, deletion of Ghsr in Kiss1 neurons reduced cumulative energy intake and abrogated body weight gain on HFD, indicating a crucial role in regulating feeding behavior and subsequent adiposity.

Glucose and Insulin Tolerance Tests
To determine if Ghsr expression in Kiss1 neurons is involved in the effect of DIO on glucose homeostasis, we conducted glucose and insulin tolerance tests on all mice. Like our previous experiment, Kiss1-GHSR KO females had an elevated fasting glucose in response to a 5 h fast compared to controls, regardless of diet ( Figure 4E; genotype: F(1,28) = 7.806, p = 0.0093). In response to a glucose challenge, we observed a decrease in glucose clearance only in Kiss1-GHSR KO females fed HFD, with glucose levels peaking around minute 30, rather than at 15 min for the other groups ( Figure 4F; time: F(5,140) = 102.03, p = 0.0000). We observed no differences in ITT ( Figure 4G). Additionally, there were no differences in the AUC analysis of GTT or ITT (data not shown). This data is similar to our previous findings in that Ghsr expression in Kiss1 neurons participates in the regulation of glucose homeostasis.

Locomotor Activity and Ingestive Behaviors in CLAMS
Control females fed LFD had greater energy intake (kCal) at night compared to the day ( Figure 5A compared to Kiss1-GHSR KO females. In addition, during weeks 6-8 Kiss1-GHSR KO females on LFD had greater cumulative energy intake compared to Kiss1-GHSR KO females on HFD ( Figure 4D; kCal; genotype: F(1,108) = 110.9, p < 0.0001; diet: F(1,108) = 75.09, p < 0.0001; time: F(8,108) = 999.0, p < 0.000; time*genotype: F(8,108) = 4.974, p < 0.0001; time*diet: F(8,108) = 3.236, p = 0.0024). Serum concentration for ghrelin and insulin were measured in control and KO females fed on LFD or HFD. We found ghrelin was elevated in KO females compared to controls on both diets and insulin was reduced in KO females on a HFD compared to control females on a HFD ( Figure S9C,D). Overall, deletion of Ghsr in Kiss1 neurons reduced cumulative energy intake and abrogated body weight gain on HFD, indicating a crucial role in regulating feeding behavior and subsequent adiposity.

Glucose and Insulin Tolerance Tests
To determine if Ghsr expression in Kiss1 neurons is involved in the effect of DIO on glucose homeostasis, we conducted glucose and insulin tolerance tests on all mice. Like our previous experiment, Kiss1-GHSR KO females had an elevated fasting glucose in response to a 5 h fast compared to controls, regardless of diet ( Figure 4E; genotype: F(1,28) = 7.806, p = 0.0093). In response to a glucose challenge, we observed a decrease in glucose clearance only in Kiss1-GHSR KO females fed HFD, with glucose levels peaking around minute 30, rather than at 15 min for the other groups ( Figure 4F; time: F(5,140) = 102.03, p = 0.0000). We observed no differences in ITT ( Figure 4G). Additionally, there were no differences in the AUC analysis of GTT or ITT (data not shown). This data is similar to our

Elevated Plus Maze
The EPM was also used as a measure of avoidance behavior. The assessment of total open arm entries revealed that HFD reduced the number of entries in controls, but not

Elevated Plus Maze
The EPM was also used as a measure of avoidance behavior. The assessment of total open arm entries revealed that HFD reduced the number of entries in controls, but not Kiss1-GHSR KOs ( Figure 6C

Light Dark Box
We utilized the LDB in conjunction with the OFT and EPM to assess exploratory and avoidance behavior. In the assessment to time spent in the light zone (s; Figure 6E), we did not observe an effect of diet or genotype. The evaluation of latency to first entry to light zone (s) showed Kiss1-GHSR KOs on HFD took longer to enter the light zone compared to controls on HFD ( Figure 6F; genotype: F(1,41) = 5.760, p = 0.0210). The combined results from the OFT, EPM, and LDB reveal that Ghsr expression in Kiss1-expressing neurons plays a role in regulating avoidance behavior on HFD.

Experiment #5: Fasting-Induced Refeeding
Prior to fasting and when fed normally, control and Kiss1-GHSR KO females show no hourly differences in food intake ( Figure 7A,C). During the first hour of refeeding, however, Kiss1-GHSR KO females did not exhibit an increase in food intake to the same extent as the controls ( Figure 7B). Across the 24 h before fasting, 24 h after fasting, and 24 h of ad libitum feeding, total food ingested, meal size, and meal duration was not different between groups (Figure 7D,F,G). Meal frequency was elevated in fed Kiss1-GHSR KO females compared to controls ( Figure 7E). Overall, fasting did not have the same effect in females lacking Ghsr expression in Kiss1 neurons as it did with controls.

Light Dark Box
We utilized the LDB in conjunction with the OFT and EPM to assess exploratory and avoidance behavior. In the assessment to time spent in the light zone (s; Figure 6E), we did not observe an effect of diet or genotype. The evaluation of latency to first entry to light zone (s) showed Kiss1-GHSR KOs on HFD took longer to enter the light zone compared to controls on HFD ( Figure 6F; genotype: F(1,41) = 5.760, p = 0.0210). The combined results from the OFT, EPM, and LDB reveal that Ghsr expression in Kiss1-expressing neurons plays a role in regulating avoidance behavior on HFD.

Experiment #5: Fasting-Induced Refeeding
Prior to fasting and when fed normally, control and Kiss1-GHSR KO females show no hourly differences in food intake ( Figure 7A,C). During the first hour of refeeding, however, Kiss1-GHSR KO females did not exhibit an increase in food intake to the same extent as the controls ( Figure 7B). Across the 24 h before fasting, 24 h after fasting, and 24 h of ad libitum feeding, total food ingested, meal size, and meal duration was not different between groups (Figure 7D,F,G). Meal frequency was elevated in fed Kiss1-GHSR KO females compared to controls ( Figure 7E). Overall, fasting did not have the same effect in females lacking Ghsr expression in Kiss1 neurons as it did with controls.  For Lowercase letters denote significance between control and KO females. Lowercase letters above capped lines denote significance between genotypes withing the same treatment (a = p < 0.05; b = p < 0.01; c = p < 0.001; d = p < 0.0001). Data are represented as mean ± SEM and n = 14 controls and n = 8 Kiss1-GHSR KO mice per group. Data were analyzed with two-way ANOVA with post hoc Holm-Sidak's multiple comparison test. Mice were 8-10 weeks of age during BioDAQ experiments.

Experiment #6: Thermoregulation
Kiss1-GHSR KO females consumed more oxygen (V.O2) during the night at both 10 and 25 • C compared to the day. In addition, Kiss1-GHSR KO females consumed more V.O2 at 10 • C compared to 25 • C at night, while control females only consumed more V.O2 at 10 • C at night compared to the day ( Figure 8A; time: F(2,30) = 29.932, p = 0.00000). Hourly V.O2 consumption was no different between groups ( Figure 8B; time: F(47,470) = 29.534, p = 0.0000). Kiss1-GHSR KO females produced more carbon dioxide (V.CO2) during the night at both 10 and 25 • C compared to the day. Furthermore, Kiss1-GHSR KO females produced more V.CO2 at 10 • C compared to 25 • C at night, while control females only produced more V.CO2 at 10 • C at night compared to 25 • C at night ( Figure 8C; time: F(2,30) = 27.231, p = 0.00000). Like V.O2 consumption, hourly V.CO2 production consumption was no different between groups ( Figure 8D
X-activity was not significantly different between controls and Kiss1-GHSR KOs; however, there was a main effect of genotype in that Kiss1-GHSR KO females moved more than controls ( Figure 8I Figure 8M). These data, taken together reveal the importance of Ghsr expression in Kiss1 neurons regarding thermoregulation. Kiss1-GHSR KO females experience elevated metabolic activity and ambulatory activity to maintain the core body temperature during cold stress, indicating a dysregulation in thermoregulation.

Extra-Arcuate Gene Expression
To determine if the deletion of GHSR in Kiss1 neurons altered the expression of Ghsr or Kiss1 in other brain regions where both genes are expressed, we explored Kiss1 and Ghsr gene expression in OVX EB-or Oil-treated females in the BNST and AMG ( Figure S6). In addition, we explored intact control and KO females Kiss1 expression in AVPV, BNST and AMG and Ghsr expression in AVPV, BNST, AMG, VTA, and SubN ( Figure S8). Kiss1 was detected in both control and Kiss1-GHSR KO females in all regions without any genotype differences. Ghsr expression was reduced in Kiss1-GHSR KO females compared to controls in the BNST and AMG, but no reduction in Ghsr was found in the AVPV, substantia nigra, or VTA ( Figure S8). These findings indicate the potential for Kiss1 and Ghsr co-expression in regions outside of the hypothalamus with potential influences on mood, motivation, and reward, which currently remain unexplored. Additional experiments would be necessary to make that determination.

Liver Gene Expression
Because Kiss1-GHSR KO females exhibited a higher fasting glucose compared to controls, we determined if we had inadvertently deleted GHSR from kisspeptin expressing cells in the liver (See Supplemental Methods for Liver RNA Extraction and qPCR). Ghsr expression was not detectable in either controls or Kiss1-GHSR KO females (data not shown), these findings are corroborated in multiple publications [57,58]. More recent publications have found that liver Ghsr expression does increase to detectable values after a 24 h fast, which we did not do [59,60]. These findings indicate that the differences in glucose levels after a 5 h fast are due, in part, to the central deletion of GHSR from Kiss1 neurons and not a peripheral deletion.

Discussion
It is important to understand the interplay of the neural circuits and hormonal influences that control reproduction, energy balance, and thermoregulation, especially as reproductive complications can result from dietary imbalances in humans. One such interplay in female rodents is the hormonal influence of ghrelin and E2 in the hypothalamic kisspeptin networks to regulate the energetic demands of reproduction. A recent study from our lab found that both negative (24 h fast and 30% caloric restriction) and positive (diet induced obesity) states of energy balance differentially impact the expression of ARC KNDy neuropeptides and their receptors. We demonstrated that E2 can both augment and oppose the effects of positive or negative energy states on KNDy neuropeptides and receptors [31]. Furthermore, our lab has determined that preovulatory levels of E2 increases ARC Ghsr expression, the gene that expresses the ghrelin receptor [31], which we found does not occur in males in response to testosterone ( Figure S2). Although we did not rule out the actions of testosterone on Ghsr expression in other brain regions like DMH, AVPV, AMG and BNST. The increase in Ghsr expression in the ARC in females is due to a 6-fold increase in KNDy neurons (Tac2), not NPY neurons [49]. In a follow-up study, we found that ghrelin inhibits the M-current, a KCNQ potassium channel current, in KNDy (Tac2-GFP) neurons and that E2 induces greater excitability by ghrelin [32]. To elucidate the importance of the interactions of E2 and ghrelin in ARC Kiss1 neurons and their control of energy homeostasis, we developed a novel Kiss1-specific GHSR knockout.
Oil-treated Kiss1-GHSR KO females exhibited reduced Kiss1, Tac2, Pdyn, Tac3r, and Esr1 when compared to oil-treated controls. Furthermore, EB treatment reduced Kiss1 and Tac2 expression in both controls and Kiss1-GHSR KOs. However, we only observed a reduction in Pdyn, Tac3r, and Esr1 and an increase in Ghsr expression in control females with EB treatment. Thus, we confirmed that the E2-induced increase in ARC Ghsr expression is indeed due to the increased gene expression in Kiss1 neurons. Another limitation to our model is that the reduction in KNDy-associated genes may be due to the effect of Kiss1 Cre heterozygous expression which reduces Kiss1 expression [61] in the Cre mouse model we used. In ghrelin injected OVX females without EB replacement, the differences we observed between control and KO females in expression of Kiss1, Tac2, Pdyn, and Esr1 were eliminated. This could indicate that the neuropeptides essential for reproduction and LH pulsatility are regulated by an interaction of ghrelin and E2 and that GHSR provides a critical link between nutritional state and reproduction.
KNDy neurons control energy homeostasis by interacting with the melanocortin circuitry. KNDy neurons directly depolarize POMC neurons through kisspeptin and/or glutamate release and directly or indirectly hyperpolarize NPY/AgRP neurons through glutamate release and/or through an enhancement of inhibitory GABAergic tone [24,25]. Furthermore, POMC and NPY/AgRP neurons modulate KNDy activity [24,26,28,43,44,62], thus producing a neuronal network of three cell types that respond to steroids, nutrients, and peptide hormones to modulate energy homeostasis and reproduction. Based on this information, we conducted a study to examine the importance of E2-induced Ghsr expression in Kiss1 neurons in the maintenance of energy balance. We hypothesized that E2 increases Kiss1 neuronal sensitivity to ghrelin in females by augmenting Ghsr expression. Greater KNDy activity would enhance the POMC excitability, via glutamate release, while diminishing the orexigenic actions of ghrelin induced NPY neuronal activity. The overall melanocortin tone (more α-MSH release) would increase output from the paraventricular hypothalamus (PVH) and downstream sympathetic tone (see Figure 9).
We found that deleting Ghsr in Kiss1 neurons, regardless of steroid treatment or diet, resulted an elevated blood glucose level in response to a 5 h fast compared to control females. This in conjunction with the lack of Ghsr expression in the liver indicates that the result of elevated fasting glucose in Kiss1-GHSR KO females is due to a central effect, not a peripheral artifact. In ghrelin or GHSR null mice, studies reveal they have similar body weight compared to littermates and are not resistant to diet-induced obesity but have been shown that under caloric-restriction there is impaired maintenance of glucose homeostasis [62,63]. POMC and NPY/AgRP neurons play a critical role in the control of glucose homeostasis and hepatic glucose production, particularly in females [64,65]. Glucose tolerance tests performed in POMC knockout female mice found that these mice were glucose intolerant while the same could not be said for males [66]. Ghrelin is a critical mediator of glucose homeostasis [67]. Ghrelin administration increases blood glucose levels and reduces insulin levels, although the opposite effects have been described [68]. The differential effects of ghrelin on glucose homeostasis are dependent on the heteromerization between GHSR and somatostatin receptor 5, both G-protein-coupled receptors (GPCRs) [69]. By deleting Ghsr in Kiss1 neurons, the heteromerization between GHSR and somatostatin receptors in Kiss1 neurons is potentially disrupted. The lack of heteromerization in addition to decreasing KNDy glutamatergic release, reduces POMC excitability and results in the observed increase in fasting blood glucose. Therefore, activation of the KNDy-POMC circuit by ghrelin in females with high E2 (proestrus) controls hepatic glucose production to ensure sufficient energy is available for elevated locomotor activity and energy expenditure necessary for reproductive behavior.
Biomolecules 2022, 12, x 21 of 27 Figure 9. Schematic of hypothesis that E2 increases KNDy sensitivity to ghrelin in females by augmenting Ghsr expression to suppress the LH production during states of elevated ghrelin (caloric restriction), and to diminish the orexigenic input of ghrelin from NPY neurons by activating POMC neurons, via glutamate, and to control thermogenesis and heat dissipation via activation of sympathetic tone from POMC innervation into the PVH and via the release of neurokinin B in the Preoptic Area (POA), respectively. During states of elevated E2 (proestrus or OVX with E2 replacement), Ghsr expression is increased in KNDy neurons augmenting KNDy sensitivity to ghrelin (1). The increase in KNDy sensitivity to ghrelin augments KNDy neuronal output via glutamate onto ARC POMC and NPY/AgRP neurons. Activation of POMC neurons would increase the release of β-endorphin on GnRH neurons thereby suppressing LH (2). The ghrelin increase during fasting enhances KNDy inhibition of NPY/AgRP neurons via glutamate (3) and POMC activation would also increase the release of α-MSH on MC-4 receptor (MC4R)-expressing neurons in the PVH to reduce food intake in opposition to ghrelin's activation of the feeding drive through NPY neurons (4). Finally, ghrelin-activated KNDy neurons projecting to the POA release NKB to control thermoregulatory output from the POA and heat dissipation (5).
We found that deleting Ghsr in Kiss1 neurons, regardless of steroid treatment or diet, resulted an elevated blood glucose level in response to a 5 h fast compared to control females. This in conjunction with the lack of Ghsr expression in the liver indicates that the result of elevated fasting glucose in Kiss1-GHSR KO females is due to a central effect, not a peripheral artifact. In ghrelin or GHSR null mice, studies reveal they have similar body weight compared to littermates and are not resistant to diet-induced obesity but have been shown that under caloric-restriction there is impaired maintenance of glucose homeostasis [63,62]. POMC and NPY/AgRP neurons play a critical role in the control of glucose homeostasis and hepatic glucose production, particularly in females [64,65]. Glucose tolerance tests performed in POMC knockout female mice found that these mice were glucose intolerant while the same could not be said for males [66]. Ghrelin is a critical mediator of glucose homeostasis [67]. Ghrelin administration increases blood glucose levels Figure 9. Schematic of hypothesis that E2 increases KNDy sensitivity to ghrelin in females by augmenting Ghsr expression to suppress the LH production during states of elevated ghrelin (caloric restriction), and to diminish the orexigenic input of ghrelin from NPY neurons by activating POMC neurons, via glutamate, and to control thermogenesis and heat dissipation via activation of sympathetic tone from POMC innervation into the PVH and via the release of neurokinin B in the Preoptic Area (POA), respectively. During states of elevated E2 (proestrus or OVX with E2 replacement), Ghsr expression is increased in KNDy neurons augmenting KNDy sensitivity to ghrelin (1). The increase in KNDy sensitivity to ghrelin augments KNDy neuronal output via glutamate onto ARC POMC and NPY/AgRP neurons. Activation of POMC neurons would increase the release of β-endorphin on GnRH neurons thereby suppressing LH (2). The ghrelin increase during fasting enhances KNDy inhibition of NPY/AgRP neurons via glutamate (3) and POMC activation would also increase the release of α-MSH on MC-4 receptor (MC4R)-expressing neurons in the PVH to reduce food intake in opposition to ghrelin's activation of the feeding drive through NPY neurons (4). Finally, ghrelinactivated KNDy neurons projecting to the POA release NKB to control thermoregulatory output from the POA and heat dissipation (5).
E2 increases activity levels in female mice; however, in our Kiss1-GHSR KO females, the effect of EB treatment was magnified compared to controls, indicating that Ghsr expression in Kiss1 neurons does play a key role in locomotor activity. We know that POMC and NPY/AgRP neurons project to the ventromedial hypothalamus (VMH) and act through melanocortin-4-receptor (MC4R) to control locomotor behavior. When E2 is high, input to the MC4R in the VMH increases activity in female mice [35]. Thus, during periods of food scarcity (high ghrelin), a female in proestrus (high E2) would have more ghrelinsensitive Kiss1 neurons, enhancing the activation of POMC neurons and α-MSH release in the VMH [11]. Furthermore, EB-treated KO females exhibited a reduced V.O2 consumption and V.CO2 production compared to EB-treated controls. The modest reduction in metabolism in the KOs is independent of locomotor activity suggesting that the lack of GHSR suppresses sympathetic output directed at metabolic tissues. Collectively, these data indicate that E2's influence on metabolism, substrate utilization, and energy expenditure is disrupted in Kiss1-GHSR KOs.
We also found that EB-treated controls consumed less than EB-treated Kiss1-GHSR KO females at night, this is a clear indication of the reduced input to POMC neurons in KO females compared to controls. In the HFD study, cumulative food intake was lower in the KO compared to controls in both LFD-fed and HFD-fed mice. Lower HFD intake underlies, in part, the resistance to diet-induced obesity as KOs exhibited reduced body weight and adiposity compared to HFD-fed controls, as previous studies have shown that HFD does not change total caloric intake [70]. While E2-treated KO females consumed more than controls, intact KO females did not respond to a 24 h fast with an increase in food intake as compared to controls. Upon refeeding, controls have a clear surge in food intake immediately upon the return of food, while KO females do not exhibit the same robust spike in food intake, which again is an indication of the importance of ghrelin signaling in Kiss1 neurons in controlling food intake in females. Furthermore, in exploratory behavioral studies, we observed a reduction in avoidance behaviors in HFD-fed KO females. It is important to note that while body weight can alter avoidance behavior in mice, we saw no differences in locomotor activity in metabolic chambers, between controls and KO females, therefore the differences seen in avoidance behaviors are likely due to the Ghsr knockout. These data in conjunction with the reduction of GHSR in both the BNST and amygdala of KO females ( Figure S6) indicates that the co-expression of kisspeptin and Ghsr in these two regions, along with hypothalamic co-expression, contributes to approach/avoidance behaviors in response to HFD and offers a novel role of kisspeptin neurons in these behaviors [71].
Unlike in the E2-treated OVX females, activity in intact Kiss1-GHSR KO females was not elevated compared to controls and metabolic parameters were no different. Only HFD reduced RER, which indicates the utilization of fat over carbohydrates for energy, a common effect of HFD. Our data is supported by previous findings in which aged GHSR null mice had lower body weight and reduced adiposity; however, they also found elevated metabolic rates (V.O2 and V.CO2), which we did not. As GHSR is only deleted in Kiss1 neurons in our model, our hypothesis is that the differences observed in body weight and adiposity were due to the reduced food intake on HFD due to GHSR deletion in Kiss1 neurons, as activity and metabolic rates were not different. Furthermore, our hypothesis is supported by recent data suggesting a role for POMC and NPY/AgRP neurons in mediating the central control of adiposity by leptin [72]. Leptin receptor activation in both POMC and NPY neurons plays a critical role in the sympathetic innervation of adipose tissue. When leptin receptors are deleted from POMC and NPY neurons, innervation of brown adipose tissue is reduced by 50% and in white adipose tissue is reduced by 30% [72]. While leptin receptor activation in ARC neurons is not the focus of our study, we can postulate that the lack of ghrelin induced Kiss1 activation may interfere with the sympathetic activation by melanocortin neurons leading to changes in adipose sympathetic innervation. Further HFD experiments measuring brown adipose tissue thermogenesis and sympathetic innervation would be necessary to determine if this is true.
Kisspeptin neurons, in conjunction with E2 and ghrelin signaling, have been found to be involved in other homeostatic processes including thermoregulation and activity [26,35]. In mice, which can undergo torpor due to food scarcity and low ambient temperatures, ghrelin administration lowers core body temperature and enhances torpor. Interestingly, the ghrelin-induced torpor response is eliminated by ablation of the ARC [62]. Fasting typically stimulates ghrelin and induces torpor in mice and rats housed in cool environments [73,74]. Our thermoregulation study reveals that deletion of GHSR in Kiss1 neurons modestly increases energy expenditure and induces greater activity in Kiss1-GHSR KO females experiencing a short-term cold stress while maintaining core body temperature. These data indicate that KO females moved more and produced more heat compared to controls to maintain the same core body temperature during cold stress. We hypothesize this is due to ghrelin-induced Kiss1 neuronal activation which results in increased release of neurokinin B in the Preoptic Area to modulate thermoregulation (see Figure 9), as neurokinin B release regulates body temperature in mice [26].
In conclusion, by deleting Ghsr expression from ARC Kiss1 neurons, we have revealed a novel central pathway for the control of reproduction and metabolism in female mice. E2 upregulates Ghsr expression in ARC Kiss1 neurons, which, in turn, increases Kiss1 neuronal sensitivity to ghrelin [28]. This increases input (potentially glutamate) to POMC neurons resulting in an elevated β-endorphin released near GnRH axon terminals in the median eminence to inhibit GnRH release and subsequently LH release from the anterior pituitary, suppressing the HPG axis, until food is available again and ghrelin levels decrease. The activation of the Kiss1-melanocortin circuit also reduces food intake and increases activity and metabolism. New insights from our studies show that Kiss1-GHSR KO females exhibited reduced metabolic rates when E2-treated, decreased adiposity due to hypophagia and delayed glucose clearance on HFD, and elevated activity and energy expenditure when cold stressed, we also conclude that the interaction of E2 and ghrelin in Kiss1 neurons regulates metabolism by altering the activation of the downstream circuits that control sympathetic tone, glucose homeostasis, food intake, and adiposity. This conclusion is supported by our previous findings [28] and by other studies characterizing the Kiss1-POMC-NPY/AgRP circuit [14,22,30,75]. Ghsr expression clearly plays a major role in adiposity and glucose homeostasis, without it, these aspects of energy balance will be disrupted. Future studies will investigate both reproductive and metabolic parameters during long-term negative energy balance, i.e., chronic caloric restriction and during times of temperature stress (cold-induced torpor). Collectively, these data suggest that GHSR activation in Kiss1 neurons modulates metabolism, activity, glucose homeostasis and thermoregulation illustrating one of many novel mechanisms for E2 and ghrelin to control Kiss1 neurons and their physiological functions.