Oestrogen-Dependent Oxytocin Dynamics in the Hypothalamus of Female Rats

Oxytocin (OXT) is produced in the hypothalamic nuclei and is secreted into systemic circulation from the posterior pituitary gland (PP). In the central nervous system, OXT regulates behaviours including maternal and feeding behaviours. Our aim was to evaluate whether oestrogen regulates hypothalamic OXT dynamics. Herein, we provide the rst evidence that OXT dynamics in the hypothalamus vary with sex and that oestrogen may modulate dynamic changes in OXT levels, using OXT-mRFP1 transgenic rats. The uorescence intensity of OXT-mRFP1 in the hypothalamic nuclei and PP was most strongly expressed during the oestrus stage in female rats and decreased signicantly in ovariectomised rats. Oestrogen replacement caused signicant increases in the uorescent intensities in the hypothalamic nuclei and PP in a dose-dependent manner. This was also demonstrated in feeding behaviour and hypothalamic Fos neurons using immunohistochemistry. Hypothalamic OXT expression was oestrogen dependent and could be enhanced centrally by the administration of oestrogen. hormone CCK-8 acts on OXT in the hypothalamus through the NTS to increase blood OXT concentrations and suppress appetite 21 . We demonstrated that food consumption was lowest when CCK-8 was administered alongside oestrogen supplementation. Thus, we demonstrated that oestrogen could further induce OXT production when CCK-8 was administered. Likewise, the strongest levels of appetite suppression corresponded with the highest proportions of OXT + /Fos + neurons. These results suggest that oestrogen administration may enhance hypothalamic OXT production clinically.


Introduction
Oxytocin (OXT) is produced in the paraventricular (PVN) and the supraoptic nuclei (SON) of the hypothalamus 1 . Peripheral OXT is a neurohypophysial hormone that is originally synthesised in the magnocellular PVN (mPVN) and the SON and is secreted from the posterior pituitary gland (PP) into the systemic circulation. In the periphery, OXT regulates parturition and lactation 2 . Recent studies have suggested that in addition to its peripheral effects, hypothalamic OXT produced in the mPVN and the SON acts on the central nervous system to regulate many functions 3 , including social recognition and trust-building 4,5 . In addition, OXT is produced in the dorsal parvocellular PVN (dpPVN) and in this context, is involved in the modulation of stress and pain 6,7 . Interestingly, the OXT pathway from the PVN is involved in the control of feeding 8 . In particular, the hypothalamic OXT has an anorectic action, and therefore, modulating this pathway is anticipated to reduce obesity and high blood glucose levels 9,10 .
However, while the peripheral actions of OXT in pregnant and lactating females are well known, the sex differences in hypothalamic OXT dynamics are still unclear.
Oestrogen is produced in the ovaries and placenta and binds to systemic oestrogen receptors (ER) via the blood to produce oestrogenic activity 11 . Oestrogen plays an important role in maintaining the physiological functions of systemic organs. Additionally, through the oestrus cycle, oestrogen further regulates female reproductive functions. In particular, oestrogen replacement therapy during menopause has been demonstrated to prevent various diseases as well as treat menopausal disorders in women [12][13][14] . Furthermore, the effects of oestrogen on food intake are thought to be mediated through ERs (both the ERα and ERβ) within the central nervous system [15][16][17] . ERβs are located on OXT neurons and are the predominant ER subtype in the PVN, a hypothalamic area involved in eating 18 . However, the details on the correlation between oestrogen and hypothalamic OXT dynamics are as of yet unknown.
In the present study, we used reporter OXT-monomeric red uorescent protein 1 (mRFP1) transgenic rats to visualize OXT expression and clarify the relationship between hypothalamic OXT and the oestrus cycle 19,20 . To con rm the relationship between oestrogen and OXT, we assessed OXT-mRFP1 expression in bilaterally ovariectomised (OVX) rats with or without exogenous oestrogen replacement. We next assessed whether oestrogen replacement and OVX impact the regulation of OXT produced by hypothalamic neurons, thereby regulating central nervous system functions, including satiety. We also assessed food consumption and hypothalamic OXT Fos-neuronal activity in OVX rats with or without oestrogen replacement and with or without intraperitoneal (i.p.) administration of cholecystokinin (CCK)-8, an agent known to selectively activate OXT neurons 21 . Further, we assessed food consumption in OVX and oestrogen replacement rats with i.p. administration of CCK-8 and intracerebroventricular (i.c.v.) administration of OXT receptor antagonist (OXTR-A) [22][23][24] . Thus, we aimed to investigate whether oestrogen could regulate and control hypothalamic OXT dynamics.

Results
OXT-mRFP1 uorescence differences between female and male rats We used adult male and female OXT-mRFP1 Wistar transgenic rats that were maintained as described previously 19,20 . We rst aimed to ascertain differences in OXT-mRFP1 uorescence between 10-week-old female rats undergoing a normal oestrus cycle and 10-week-old male rats [ rst experiment (Exp. A)].
Female OXT-mRFP1 transgenic rats were further divided based on the four oestrus stages (prooestrus, oestrus, metoestrus, and dioestrus stages). As in a previous study, we observed the entire hypothalamus [SON, anterior parvocellular PVN (apPVN), dpPVN, and mPVN] and PP in OXT-mRFP1 transgenic rats using high-power uorescence microscopy 25 (Figure 1a). OXT-mRFP1 uorescence in the SON, apPVN, dpPVN, and mPVN was signi cantly different between reproductive males and females in the oestrus stage. OXT-mRFP1 uorescence in the apPVN, dpPVN, and mPVN was signi cantly different among females depending on the oestrus stage. There was a signi cant difference in the OXT-mRFP1 uorescence in the PP between male rats and female rats in the oestrus stage ( Figure 1b). These results suggest that the presence of OXT expression in the hypothalamus and pituitary gland is in uenced by sex.
Effects of OVX on OXT-mRFP1 uorescence OVX and sham operations were performed on the 11th week, and experiments were conducted on the 15th week. In order to further investigate the in uence of sex on OXT expression, OVX (Exp. B) was conducted to induce an oestrogen-de cient state in reproductive female OXT-mRFP1 transgenic rats. These rats were compared to a sham-operated control group, which consisted of both male and female rats. As in Exp. A, OXT-mRFP1 uorescence in the SON, apPVN, dpPVN, and mPVN revealed a signi cant difference between the male rats and female rats in the oestrus stage ( Figure 2a). Moreover, OXT-mRFP1 uorescence in the hypothalamus (SON, apPVN, dpPVN, and mPVN) and PP were signi cantly decreased in OVX rats when compared to females in the oestrus stage and were similar to the levels in male rats ( Figure 2b). Based on these results, we determined that OVX resulted in a decrease in the OXT expression in the hypothalamus and pituitary gland.

Effects of oestrogen replacement on OXT-mRFP1 uorescence
We performed OVX in 11-week-old female OXT-mRFP1 transgenic rats, conducted hormone replacement on week 15, and performed experiments on week 16. Given that oestrogen levels were expected to be affected by OVX, oestrogen supplementation experiments were performed (Exp. C). Oestrogen was supplemented in female OVX OXT-mRFP1 rats. Among the OVX groups, the rats in the groups with supplementation of low (β-oestradiol) E2 and high E2 elicited signi cant changes in OXT-mRFP1 levels in the hypothalamus (SON, apPVN, dpPVN, and mPVN) and the PP (Figure 3a). Interestingly, the high-dose E2 group demonstrated higher levels of OXT-mRFP1 uorescence than the low-dose group in the hypothalamus (SON, apPVN, dpPVN, and mPVN) (Figure 3b). This represents the rst evidence that oestrogen regulates OXT expression in hypothalamic OXT neurons in a dose-dependent manner.
Relationship between rat body weight and feeding We observed a signi cant change in rat body weight and feeding depending upon sex, OVX, and oestrogen replacement. Wistar rats were divided into ve groups: sham-operated male, sham-operated female, only OVX, OVX plus low E2 replacement, and OVX plus high E2 replacement groups (Exp. D). We observed a signi cant change in rat weight depending upon sex and OVX. Female rats with OVX displayed a signi cant change in body weight compared to sham-operated reproductive female rats, and oestrogen supplementation affected body weight, depending on the dose (Figure 4a). We observed a signi cant change in rat feeding in depending on sex, OVX, and oestrogen replacement ( Figure 4b). This suggests that oestrogen regulates rat body weight and feeding. The difference in body weight is Effect of peripheral administration of CCK-8 on food intake with oestrogen replacement We performed CCK-8 administration experiments to investigate the relationship between oestrogen and food intake. All Wistar female rats with OVX were divided into four groups: oil only and high dose oestrogen in the subcutaneous tube with i.p. administration of saline or CCK-8 (Exp. E). Rats receiving high doses of oestrogen experienced signi cant weight loss (Figure 5a). Rats supplemented with high doses of oestrogen consumed signi cantly less food throughout the day. Cumulative food intake was signi cantly decreased in the OVX/High E2 group compared to the OVX/oil only group (Figure 5b). CCK-8 was administered to examine the amount of food consumed. Cumulative food intake was signi cantly decreased 0.5 h, 1 h, and 1.5 h after i.p. administration of CCK-8. There was a signi cant difference between OVX/oil only group and the OVX/High E2 group at 1.5 h after i.p. administration of saline and 3 h after i.p. administration of CCK-8. After 6 h, there was no signi cant difference in cumulative food intake for all groups (Figure 5c).

Effect of oestrogen on Fos expression in OXT-ir neurons
We conducted immunohistochemistry to assess the levels of Fos and OXT in the hypothalamus (Exp. F). All Wistar female rats with OVX were divided into two groups: oil only and high dose oestrogen tubing.
Tissues were harvested and evaluated for the expression of Fos and OXT through double-uorescence immunohistochemistry (FIHC). We quanti ed immuno uorescently labelled OXT+, Fos+, and OXT+/Fos+ double-labelled cells in the SON and PVN (Figure 5d). The number and percentage of OXT + /Fos + cells were signi cantly higher when CCK-8 was administered than when saline was administered. Among these rats, the number and percentage of OXT + /Fos + cells were signi cantly higher in the high E2 group than in the oil only group (Figure 5e).

Effect of pre-treatment with OXT receptor antagonist (OXTR-A) on food intake
In the previous experiment (Exp. E), there was a signi cant difference between the OVX / oil-only group and the OVX / high E2 group at 3 hours after i.p. administration of CCK-8. Therefore, we assessed food intake for 3 hours after i.p. administration of CCK-8 and i.c.v. administration of OXTR-A. All Wistar female rats with OVX were divided into four groups: oil only and high dose oestrogen in the subcutaneous tube with i.p. administration of CCK-8 and i.c.v. administration of saline or OXTR-A (Exp. G). Rats receiving high doses of oestrogen experienced signi cant weight loss ( Figure 6a). Cumulative food intake for 3 hours was signi cantly increased in i.c.v. administration of OXTR-A group compared to i.c.v. administration of saline group (Figure 6b).

Discussion
The present study provides the rst evidence that hypothalamo-neurohypophysial OXT is oestrogendependent and shows dynamic changes during the oestrus cycle. OXT-mRFP1 uorescence intensity in the SON and PVN was expressed most strongly among adult oestrous female rats and was signi cantly reduced in OVX rats. Oestrogen supplementation restored uorescence intensity in the SON and PVN in OVX rats in a dose-dependent manner. Thus, the dynamics of hypothalamic OXT expression is regulated by oestrogen. As one of the physiological meanings, we con rmed that feeding suppression induced by the peripheral administration of CCK-8 resulted in the activation of OXT neurons, and this was enhanced among oestrogen-replaced female rats.
OXT is mainly produced in neurosecretory neurons located in the SON and PVN in the hypothalamus. We successfully generated transgenic rats bearing an OXT-mRFP1 fusion gene, which resulted in the visualization of OXT expression 19,20,[25][26][27][28] . The PVN is divided into regions such as apPVN, dpPVN, and mPVN, and previous studies with the OXT-mRFP1 transgenic rats reported different effects 29 . OXT + neurons in the SON and mPVN project their axons to the PP, where OXT is thereby secreted into the systemic circulation and elicits activity peripherally 30 . There are reports that in males, OXT is involved in sexual behaviour, ejaculation, and transport of spermatozoa. In females, the peripheral effects of OXT are related to labour and lactation 31 . With respect to the central functions, OXT is also somatodendrically released from neurons in the SON and mPVN and acts directly on the brain 32 . Neurons expressing OXT receptors are ubiquitous in the brain and have a wide range of functions 33 . It has been reported that OXT is not only associated with con dence and bond formation but is also strongly associated with autism 34,35 . In the present study, the observed differences related to the sex of OXT-mRFP1 rats suggest that the production of OXT in the hypothalamus differs according to sex. This may further indicate that the central actions of OXT can also vary with sex.
A pathway that projects OXT from the apPVN and dpPVN to the medulla and spinal cord has been identi ed 29 . This is known to affect the autonomic nervous system and has been described to induce analgesic effects and regulate pain 36 , gastrointestinal motility and cardiovascular responses 37 . There are clinical reports that the pain threshold is high during pregnancy and postpartum 38 . Chronic pain after caesarean section is reported to be 1/10 as severe as that after other open surgeries 39 . We speculate that these observations could be explained by hypothalamic OXT dynamics. In our study, OXT-mRFP1 expression was increased in the apPVN and dpPVN throughout the oestrus cycle, and this may, in turn, regulate pain thresholds.
Spontaneous ovulator animals have an ovulation cycle. Humans have an oestrus cycle of approximately 28 days while rats have a shorter cycle of 4-5 days. The blood levels of hormones secreted from the ovaries also uctuate periodically, and these dynamics are common between humans and rats. Oestrogen acts on the mucous membrane epithelium and changes its histology. Therefore the oestrus cycle can be monitored through a vaginal liquid smear examination 40 . Based on this examination, the oestrus cycle in rats can be classi ed into prooestrus, oestrus, metoestrus, and dioestrus cycles. The highest increase in blood oestrogen levels occurs when the LH surge coincides with the oestrus stage.
Ovulation subsequently occurs several hours later 41 . In OXT-mRFP1 transgenic rats, the uorescence intensity of mRFP1 has been shown to be delayed by several hours after stimulation 19,29,42 . It is thought that the uorescence intensity of mRFP1 increases several hours after the point of the highest prooestrus production of oestrogen by the ovary. This, therefore, indicates that uorescence intensity peaks during the oestrus stage.
In females, oestrogen levels are reduced as a result of age-related reductions in ovarian function and when the ovaries are removed due to gynaecological surgery or treatment. In addition to mammary glands and genital organs, oestrogen acts on the liver, cardiovascular system, bones, and the brain 12,43,44 . Therefore, ovarian dysfunction can elicit many symptoms. In this study, we mimicked ovarian dysfunction/menopause through ovariectomies. We observed that OVX reduced mRFP1 uorescence in the SON and PVN (ap, dp, m). Thus, OXT production may have been decreased in all hypothalamic areas.
Decreased OXT production suggests that broad-ranging central OXT actions may be attenuated by OVX.
There are three known variants of oestrogen -oestrone (E1), oestradiol (E2), and oestriol (E3), and three known subtypes of ERs -ERα, ERβ, and G protein-coupled receptor 30 (GPR30) 45 . ERs are expressed systemically, and ERα and ERβ are localized in the brain. However, only ERβ has been reported to be expressed in the PVN and SON in the hypothalamus 11 . E2, which elicits the strongest effects, is often used as an experimental or therapeutic drug. In this experiment, the type of oestrogen used was E2. Oestrogen replacement has various effects on the liver, cardiovascular system, bones, and the brain. It is further used to treat menopausal symptoms [46][47][48] . The subcutaneous administration of oestrogen has fewer side effects compared to intravenous and oral administration and can reproduce the systemic effects of oestrogen 49 . In our work, the subcutaneous administration of E2 to OVX female rats restored mRFP1 uorescence intensity in the SON and PVN (ap, dp, m) in a dose-dependent manner. Thus, the subcutaneous administration of E2 may increase OXT production in all hypothalamic regions.
Oestrogen has further been reported to suppress feeding. Mechanistically, oestrogen has been shown to increase pro-opiomelanocortin (POMC) gene expression, which induces appetite suppression through signal transducers and activates transcription 3 (STAT-3) in the hypothalamus 50 . Oestrogen also regulates hypothalamic OXT activity in the NTS of the solitary tract and thereby suppresses food intake 18 . Similar to the previous reports, we demonstrated that oestrogen administration suppressed feeding and promoted weight loss. Furthermore, we con rmed that hypothalamic OXT production increased depending on oestrogen levels. This suggests that food consumption may differ depending upon oestrogen levels.
Oestrogen and OXT share a common antifeedant activity 20 . Food consumption was measured following the administration of CCK-8, which is known to selectively activate OXT neurons 21 . We demonstrated the link between OXT and oestrogen by counting Fos + /OXT + neurons. Fos is an early expression prooncogene expressed at low levels in most cell types 51 and can be activated by various second messenger signals. However, Fos is an indicator of cell activation. Accordingly, the evaluation of Fos expression cannot be used to determine any direct effects related to feeding versus other secondary effects, which may likewise induce cellular activation. The gastrointestinal hormone CCK-8 acts on OXT in the hypothalamus through the NTS to increase blood OXT concentrations and suppress appetite 21 . We demonstrated that food consumption was lowest when CCK-8 was administered alongside oestrogen supplementation. Thus, we demonstrated that oestrogen could further induce OXT production when CCK-8 was administered. Likewise, the strongest levels of appetite suppression corresponded with the highest proportions of OXT + /Fos + neurons. These results suggest that oestrogen administration may enhance hypothalamic OXT production clinically.
Many studies have demonstrated the e ciency of oestrogen and OXT as anti-obesity peptides 52 . Intracerebroventricular or peripheral (i.p. and subcutaneous) injection of OXT decreases food intake, body weight, and fat mass in rats and mice 37,53 . Fat was divided into visceral fat and subcutaneous fat, with differences between sexes. The percentage of visceral fat was higher in males, and subcutaneous fat was altered in females. However, in this experiment, it is unclear whether OXT was directly involved in fat composition.
To the best of our knowledge, this is the rst report demonstrating a difference in the dynamics of the hypothalamic OXT based on the sex of rats. Additionally, we demonstrated that hypothalamic OXT expression is speci cally dependent upon oestrogen, as E2 administration increased central OXT production in OVX rats. However, our study has several limitations. We did not assess any association of hypothalamic OXT with sex hormones (e.g. progesterone) other than oestrogen. All hypothalamic examinations were conducted in rats and not clinically in humans. No adverse effects of oestrogen replacement have been considered.

Rats Models
All rats were treated after 10 weeks of age once the oestrus cycle was established. Adult male and female OXT-mRFP1 Wistar transgenic rats (aged 10-16 weeks and weighing 223-474 g) were bred and maintained as described previously 19,20 . The OXT-mRFP1 transgenic rat was created by inserting the mRFP1gene into the OXT gene. This reporter strain facilitates the visualization of OXT dynamics in the hypothalamus through the quanti cation of OXT uorescence intensity changes under various stimulation loads 28,29 . All rats were genotypically screened through PCR analysis of their genomic DNA extracted via ear biopsies 19  For hormone replacement, hormone-containing tubes were subcutaneously implanted into in the midback region of the rats. Rats were anesthetized with iso urane (3% iso urane with a ow rate of 5.0 L/min). Silastic tubing (1.57 mm inner diameter; 3.18 mm outer diameter; 37.0 mm in length; Dow Corning, Midland, MI, USA) was lled with fat-soluble E2 (17β-oestradiol ≥ 98%, Sigma-Aldrich, Tokyo, Japan) dissolved in sesame oil (Sigma-Aldrich) 54,55 .
For i.c.v. administration, animals were implanted with stainless steel canulae targeting the lateral ventricle. They were anaesthetized (i.p. injection of a cocktail of three different anaesthetic agents (0.3 mg/kg of medetomidine, 4.0 mg/kg of midazolam, and 5.0 mg/kg of butorphanol)) and placed in a stereotaxic frame. Stainless steel guide canulae (550 μm outer diameter and 10 mm length) were stereotaxically implanted at the following coordinates: 0.8 mm posterior to the Bregma, 1.4 mm lateral to the midline, and 2.0 mm below the surface of the left cortex, such that canula tips were 1.0 mm above the left cerebral ventricle 56 . Two stainless steel anchoring screws and acrylic dental cement were used to secure cannulae in place. After the surgical procedure, animals were handled daily, individually housed in a plastic cage, and allowed to recover for at least 10 days.
Experimental procedure 1. Assessing differences in OXT-mRFP1 uorescence in male and female transgenic rats We rst aimed to ascertain differences in OXT-mRFP1 uorescence between 10-week-old female rats undergoing a normal oestrus cycle and 10-week-old male rats [ rst experiment (Exp. A), n = 30]. Female OXT-mRFP1 transgenic rats were further divided based on the four oestrus stages (prooestrus, oestrus, metoestrus, and dioestrus stages). In total, we assessed ve groups (1 male, 4 females; n = 5-9 per group). The oestrus cycle was con rmed by examination of the vaginal smear from the rats collected every morning by two researchers. In brief, the prooestrus stage is identi ed mainly by nucleated cells, oestrus by all keratinocyte cells, metoestrus by the presence of many white blood cells, and dioestrus by the presence of few white blood cell and various cells. Rats with irregular oestrus cycles were excluded from the experiment.

Assessing the effects of OVX on OXT-mRFP1 uorescence
In the second experiment (Exp. B, n = 33), all OXT-mRFP1 transgenic rats were divided into six groups, including ve sham-operated groups (consisting of both males and females at all oestrus cycle stages) and one OVX group (n = 5-6 per group). We con rmed the oestrogen-de cient state (like dioestrus stage) according to the vaginal smear of OVX rats.
3. Assessing OXT-mRFP1 uorescence after oestrogen replacement For the third experiment (Exp. C, n = 24), we conducted hormone replacement experiments involving subcutaneous implantation of oestrogen-containing tubes in the mid-back region of the rat. Female OXT-mRFP1 transgenic rats with OVX were divided into four groups (n = 5-6 in each group): control (sham back operation), vehicle (subcutaneous sesame oil only), low E2 replacement (20 μg β-oestradiol/ml sesame oil), and high E2 replacement (400 μg β-oestradiol/ml sesame oil) 54,55 . Low doses of E2 have previously been shown to negatively affect luteinizing hormone (LH) pulses but did not induce LH surges in OVX rats. High E2 levels were shown to induce an LH surge and resulted in an oestrus-like state in OVX rats 54 . We con rmed the effect of low E2 and high E2 in the rats through vaginal smears examinations.

Measurement of body weight and food intake
For the fourth experiment (Exp. D, n = 30), in male and female Wistar rats we assessed body weight and food intake. Wistar rats were divided into 5 groups (n=6 in each group): sham-operated male, shamoperated female, only OVX, OVX plus low E2 replacement, and OVX plus high E2 replacement groups. We performed sham operations and OVX in 10-week-old male and female rats, conducted hormone replacement (tube implantation) with OVX rats at week 14, and assessed cumulative food intake for one day at 9, 13, and 16 weeks.

Assessing the effects on administration of CCK-8
For the fth experiment (Exp. E, n = 24), in female Wistar rats, we assessed food intake. Ovariectomised Wistar rats were divided into 4 groups: vehicle (subcutaneous sesame oil only) with i.p. administration of saline, vehicle (subcutaneous sesame oil only) with i.p. administration of CCK-8 (50 μg/kg body weight), high E2 (400 μg β-oestradiol/ml sesame oil) with saline injection, and high E2 with CCK-8 injection.
Ovariectomies were performed at 10 weeks and tube implantation was performed at 14 weeks of age; saline and CCK-8 were injected at 16 weeks. On week 16, cumulative food intake was measured at 0.5 hours (h), 1 h, 1.5 h, 3h and 6 h after i.p. administration of saline and CCK-8. Prior to injections, all rats were fasted for 24 h (n = 6 in each sub-group). CCK-8 was administered, and the effect on food intake was assessed.
For the sixth experiment (Exp. F, n = 24), in female Wistar rats, we conducted immunohistochemistry to assess the levels of Fos and OXT in the hypothalamus. Ovariectomised Wistar rats were divided into four groups: vehicle (subcutaneous sesame oil only) with i.p. administration of saline, vehicle (subcutaneous sesame oil only) with i.p. administration of CCK-8 (50 μg/kg body weight), high E2 (400 μg βoestradiol/ml sesame oil) with saline injection, and high E2 with CCK-8 injection. Ovariectomies were performed at 10 weeks and tube implantation was performed at 14 weeks of age; saline and CCK-8 were injected at 16 weeks. All rats were anesthetized and sacri ced at 1.5 h after i.p. administration of saline and CCK-8. Slices of brain tissues were harvested and evaluated for the expression of Fos and OXT through double-FIHC.  Fluor 488 donkey anti-rabbit IgG; Molecular Probes, OR, USA; 1:2,000 in PBS containing 0.3% Triton X-100) 58 . Sections were washed twice in PBS and then mounted on the slides and coverslipped using vectashield (Vector Laboratories Co. Ltd., CA, USA) 59 . Images of Fos + , OXT + , and Fos + /OXT + double-labelled cells were counted manually by two researchers who were blinded to avoid bias. The number and percentage of Fos + , OXT + , and Fos + /OXT + cells in the SON and PVN were estimated.

Statistical analysis
All data points are presented as the mean ± standard error of the mean. Statistical signi cances were calculated based on one-way analysis of variance (ANOVA) and repeated-measures ANOVA, using a Tukey-Kramer-type adjustment for multiple comparisons. A P-value <0.05 was considered statistically signi cant.

Figure 4
Relationship between rat body weight and feeding a. Changes in body weight after treatment. Shamoperated male, sham-operated female, and OVX female rats at 10 weeks of age. Tube implantation (oil only, low E2 and high E2) was performed in OVX female rats at 14 weeks of age. The data are presented as the mean±SEM (repeated-measures ANOVA) (**P<0.01, compared with the treated rats; † †P<0.01, compared with all female rats). b. Cumulative food intake for one day at 9, 13, and 16 weeks of age. The data are presented as the mean±SEM (one-way ANOVA) ( †, ‡, **P<0.01, *P<0.05, compared with oil only and high E2 groups).