Suppression of ovarian hormones in adolescent rats has no effect on anxiety‐like behaviour or c‐fos activation in the amygdala

In humans, sex differences in mood disorders emerge during adolescence, with prevalence rates being consistently higher in females than males. It has been hypothesised that exposure to endogenous ovarian hormones during adolescence enhances the susceptibility of females to mood disorders from this stage of life onwards. However, experimental evidence in favour of this hypothesis is lacking. In the present study, we examined the long‐term effects of suppressing adolescent gonadal hormone levels in a group of female Lister‐hooded rats via administration of a gonadotrophin‐releasing hormone antagonist (Antide; administered on postnatal day [PND] 28 and 42) compared to control females and males (n = 14 per group). We predicted that, in adulthood, Antide‐treated female rats would exhibit more male‐like behaviour than control females in novel environments (elevated‐plus maze, open field and light‐dark box), in response to novel objects and novel social partners, and in an acoustic startle task. Progesterone and luteinising hormone assays (which were conducted on blood samples collected on PND 55/56 and 69/70) confirmed that the hypothalamic‐pituitary‐gonadal axis was temporarily suppressed by Antide treatment. In addition, Antide‐treated females were found to exhibit a modest pubertal delay, as measured by vaginal opening, which was comparable in length to the pubertal delay that has been induced by adolescent exposure to alcohol or stress in previous studies of female rats. However, Antide‐treated females did not substantially differ from control females on any of the behavioural tests, despite the evidence for predicted sex differences in some measures. Following the acoustic startle response task, all subjects were culled and perfused, and c‐Fos staining was conducted in the medial and basolateral amygdala, with the results showing no significant differences in cell counts between the groups. These findings suggest that ovarian hormone exposure during adolescence does not have long‐term effects on anxiety‐related responses in female rats.


| INTRODUC TI ON
In humans, the prevalence of mood disorders is consistently reported to be higher in women than in men, [1][2][3] and this sex difference in prevalence emerges during adolescence. [4][5][6] These observations have led to hypothesis that the fluctuations in ovarian hormone levels that accompany puberty increase the susceptibility to mood disorders amongst adolescent girls. 7 This hypothesis is supported by evidence that pubertal status, rather than age, predicts the incidence of mood disorders 8,9 and evidence that supplementation with synthetic ovarian hormones during adolescence (via hormonal contraceptive use) is associated with a raised incidence of depression and altered physiological stress responses. [10][11][12] However, because these human studies are correlational, caution is required when inferring a direct link between endogenous ovarian hormone exposure during adolescence and long-term mental health and stress reactivity. 13 Studies on laboratory rodents provide the opportunity to experimentally investigate the immediate and long-term effects of adolescent gonadal hormone alterations on behavioural and brain development. In rodents, adolescence is generally defined as the period that encompasses pubertal sexual maturation 14 and is characterised in female rodents by vaginal opening and onset of ovarian cycling and in male rodents by rising testicular hormone levels and sperm production. 15,16 Sex differences in behaviour emerge during adolescence in rodents; for example, adolescent male rats (Rattus norvegicus) exhibit a higher preference for novel objects compared to adolescent females 17 and locomote less than aged-matched females in novel environments. 18,19 The hypothalamic-pituitary-adrenal (HPA) axis also undergoes significant changes during adolescence in rodents and humans, 20 which means that any factors affecting the developing HPA axis could have long-term effects on response to stressors. 21 Previous research on adolescent rodents has shown that removing gonadal hormones during this stage of life has significant, longterm effects on behavioural development. 22,23 The majority of these studies have been conducted on male rodents and have compared the behaviour of males that were castrated before pubertal onset with males that were castrated at the end of adolescence (ie, only the post-pubertally castrated males experienced normal gonadal hormone exposure during adolescence). Post-pubertally castrated male rodents are reported to exhibit higher levels of aggression and sexual proficiency, as well as spend less time in the aversive areas of novel environments, in adulthood than pre-pubertally castrated male rats, 24 Syrian hamsters (Mesocricetus auratus) [25][26][27] and mice (Mus musculus). 28 These findings indicate that exposure to gonadal hormones during adolescence influences behavioural development in male rodents, consistent with the hypothesis that adolescence is a 'sensitive period' of development. 29 Fewer studies have investigated the immediate and long-term behavioural effects of manipulating adolescent gonadal hormone levels in female rodents; these studies have reported that exposure to gonadal hormones during adolescence increases maternal behaviour in adulthood (in mice 30 ) and has organisational effects on subtle aspects of movement during social and sexual interactions (in rats [31][32][33] ). However, little is known about the effects of manipulating adolescent gonadal hormones on anxiety-like behaviour or fear responses. A recent study reported that pre-pubertally ovariectomised Siberian hamsters (Phodopus sungorus) spend more time than controls in the light area of a light-dark box, 34 whereas another study reported that pre-pubertally ovariectomised female rats spend less time in the centre of a novel open field in adulthood than sham-operated females, 35 which is consistent with data from adult rodents showing that low oestradiol levels are associated with enhanced anxiety-like behaviour (eg, in rats 36 ). However, the later study 35 did not include a separate control group of females ovariectomised after puberty, and so any apparent long-term behavioural differences could reflect the activational effects of adult hormones; such effects are not necessarily consistent with the organisational effects of the same hormone at earlier life stages. 24 Gonadal hormone exposure during adolescence could impact upon the development of brain regions that are involved in emotional processing, such as the amygdala, ventral striatum and prefrontal cortex, which undergo reorganisation in adolescent rodents and humans. [37][38][39][40][41] Gonadal hormone receptors are located within these brain regions, 41,42 and these regions exhibit sex differences in structure and function. 43,44 For example, the basolateral amygdala, which is highly sensitive to stress, 45 shows substantial sex differences associated with emotion-related behaviours corresponding to greater excitatory synaptic input in female rats. 46,47 In addition, the posterior medial amygdala is sensitive to ovarian sex hormones, 48 showing robust post-pubertal differences in soma morphology and size. 49,50 In humans, the developmental trajectories of the amygdala and prefrontal cortex have been reported to correlate with pubertal status. [51][52][53][54] Given that gonadal hormones are known to influence a range of neurodevelopmental processes, 55 adolescent hormone exposure could thus potentially have long-term, 'organisational' effects on later affective behaviour by influencing the development of these brain area.
The present study aimed to examine the effects of suppressing ovarian hormones during adolescence on later anxiety-like behaviour, including responses to novelty, and amygdala function in female rats. Previous rodent studies have used ovariectomies, an approach that involves invasive surgery and results in the permanent loss of gonadal function, alongside significant elevation in circulating gonadotrophin levels as a result of the permanent disruption of negative-feedback loops. An alternative method for temporarily reducing gonadal hormone levels involves treatment with gonadotropin-releasing hormone (GnRH) peptide antagonists, which can be delivered via subcutaneous injection. GnRH peptide antagonists competitively bind to GnRH receptors in the pituitary, without activating these receptors (in contrast to GnRH agonists, which produce an initial phase of hyperstimulation), and reversibly suppress gonadotrophin and gonadal hormone production. 56 These antagonists have a long half-life and, once metabolised, the hypothalamic-pituitary-gonadal (HPG) axis is reactivated. Therefore, using a GnRH antagonist to suppress gonadal hormone levels has several advantages compared to ovariectomy, including that negative-feedback is maintained in the HPG axis and that the treatment avoids the use of stressful surgical procedures and is reversible. This experimental design, involving juvenile treatment with a GnRH antagonist, delays the hormonal changes that normally accompany puberty, allowing for hormone-independent developmental changes to be dissociated from hormone-dependent processes.
We examined the effects of suppressing ovarian hormones during adolescence using a GnRH peptide antagonist (Antide), and the goal was to delay pubertal gonadal hormone exposure to a similar extent as seen with adolescent exposure to alcohol or stress, both of which can interfere with HPG functioning during this period of life. 20,57 We predicted that, in adulthood, Antide-treated female rats would exhibit more male-like behaviour than control females in novel environments (elevated-plus maze, open field and light-dark box), in response to novel objects and novel social partners, and in an acoustic startle task. Previous studies have reported behavioural sex differences in these tasks 58 ; for example, on average, female rats usually spend more time than males in the exposed sections of the elevated-plus maze and light-dark box, [18][19][20] and male rats exhibit stronger acoustic startle responses than females. 59, 60 We also postulated that c-fos activation in the amygdala following the acoustic startle task would be greater in Antide-treated females than in control females.

| Ethical statement
Ethical guidelines, as set out in the Principles of Laboratory Animal Care (NIH, Publication No. 85-23, revised 1985) and the UK Home Office Animals (Scientific Procedures) Act 1986, were adhered to throughout the study (Home Office Project Licence PC33CDA1C; Personal Licences IF9BFD0E9 and I09B3A36E).

| Subjects and housing
The subjects comprised 42 Lister-hooded rats (28 females and 14 males) that were bred in-house (stock animals were obtained from Charles River, Margate, UK). The subjects were derived from five litters, and an additional six males from these litters were used as social stimuli in one of the behavioural tests. Breeding females were individually housed in plastic and mesh cages (52 × 40 × 26 cm, length × depth × height) with access to water and pellet food (DBM Food Hygiene Supplies Ltd, Scotland) available ad lib. Pups were weaned at postnatal day (PND) 23 and housed in single-sex sibling groups until PND 27, then re-housed as same-sex pairs (cage dimensions same as above) with access to water and food available ad lib.
All animals were housed in a holding room under a 12:12 hour light/ dark photocycle (lights on 7.00 am) at 20 ± 1°C and 55 ± 5% relative humidity.

| Experimental design
Three sets of subjects were produced: (i) females (n = 14) that were treated on PND 28 and 42 with a gonadotrophin-releasing hormone 3 weeks, as in previous studies, 61,62 and so the regime of two injections 14 days apart was designed to suppress gonadal hormone levels from the late juvenile phase through to the end of late adolescence (age ranges based on previous studies 18,63 ). No more than five animals from each litter were assigned to a single experimental group, and, in the majority of cases, no more than four females from a single litter were used as subjects in the experiment (ie, two Antide-treated females and two control females), with all subjects paired with a littermate in the same experimental group.
All subjects were weighed once per week from PND 28 onwards.
Vaginal opening (VO) was assessed in female subjects daily by visual inspection from PND 28, with VO recoded as the first day when the vagina was either partially or fully opened, and males received similar handling each day. Previous studies have shown that vaginal opening typically occurs around PND 32-34 in rats. 64,65 Anogenital distance (AGD) was measured for all subjects on PND 35 and 49 using electronic calipers, with care being taken not to touch the genital area with the caliper tips, given that artificial genital contact can accelerate sexual maturation in adolescent rats. 66 Behavioural testing was undertaken during adolescence (PND the acoustic startle response (ASR) task (PND 98-103, followed by perfusion 90 minutes after the ASR task). The same order of testing was maintained for all subjects, as in previous studies in our laboratory, 18,19 because some tests had to be conducted at specific subject ages (adolescent social behaviour) and the ASR test had to precede perfusion for all subjects. Although this design does not allow the estimation of any carry-over effects that might result from the order of behavioural testing, between-group comparisons remain valid. All behaviour tests took place in a single testing room with the relevant piece of apparatus set-up directly below the ceiling-mounted camera, and a black curtain separated the apparatus from the rest of the testing room. Individual subjects were transported to the testing room in a carrying box and returned to the home-cage immediately after completing the test.
Information on ovarian cyclicity was not collected for the female subjects in this experiment for three reasons. First, our research question did not involve investigating cycle-dependent effects, and it has been suggested previously that experimental designs do not need to take ovarian stage into account unless the study is specifically investigating cycle-dependent effects, 67,68 particularly because the behaviour of unstaged females is not more variable than that of males. 69 Second, we did not require cyclicity data to confirm that Antide treatment had been successful because hormonal and vaginal opening data were considered to be sufficient. Third, cyclicity monitoring would have required swab sampling to be undertaken alongside the behavioural testing, which could have induced stress in both of the female groups (but not in the control male group) and thereby impacted upon the behaviour of subjects and influenced between-group comparisons.
Blood samples were taken from tail veins on PND 55/56 and 69/70, under gas anaesthesia (isofluorane), to confirm that treatment with Antide successfully suppressed the HPG axis, and serum was stored at −70°C prior to assaying of progesterone and luteinising hormone (LH) levels. All subjects were culled via perfusion following the final behavioural test (ie, 90 minutes after the acoustic startle response task) on PND 98-103, and the brain tissue was sectioned, mounted and frozen at −20°C prior to c-Fos immunohistochemistry and cell counting.

| Adolescent social behaviour
On PND 40, 41, 43 and 44, subjects were paired with a same-aged, same-sex partner from the same experimental group (ie, not the cage-mate [insufficient non-experimental animals of the same age were available to conduct the social behaviour tests with fully naïve animals]). Social interaction sessions took place in a perspex arena

Novel object (NO) test
All subjects underwent the NO test on PND 89. The test was conducted in a painted, wooden arena measuring 67 × 67 × 45 cm (length × width × height). The arena floor was divided visually into four equal quadrants, and the light illuminance was approximately 20 lux at floor level. Three novel objects were used that were similar in size but differed in texture (glass or plastic) and colour (blue, pink or clear). During the first stage of the test, two objects were placed into the arena, and one in each of two adjacent quadrants. The subject was then placed in the empty half of the arena, facing away from the novel objects, and allowed to explore for 3 minutes. The subject was then removed and placed into a carrying box for 2 minutes, during which time the objects were removed and both the arena and the objects were cleaned with liquid disinfectant. One of the original objects was then placed back into the arena in its original position, and the third object was placed into the position previously occupied by the removed object. The subject was returned to the arena for a 3-minute period, then returned to the home-cage and the apparatus and objects cleaned. The objects were counterbalanced across subjects.
In the NO test, time spent exploring the novel objects can be used as a measure of anxiety-like behaviour, whereas the relative time spent exploring the novel and familiar object can either reflect novelty preference or short-term memory. 72 The following specific behavioural

| Acoustic startle response (ASR) task
All subjects were tested once in the ASR task between PND 98 and an electronic startle monitor, we were unable to evaluate additional measures, such as startle amplitude; however, 'freezing' behaviour has previously been shown to correlate with startle amplitude in female rats. 74 At the end of the ASR task, the subject was returned to the home-cage for 90 minutes before being culled for brain tissue collection.
Serum samples were diluted with assay buffer (sera from females were diluted 1:100, sera from males were diluted 1:25) and 100 µL of each diluted sample was assayed in duplicate. The assay has a detection limit of 8.57 pg mL -1 , an intra-assay coefficient of variation of 5.4% and an inter-assay coefficient of variation of 8.3%.

| LH assay
Serum LH levels were measured using established in-house enzymelinked immunosorbent assays as described previously. 75 Serum samples were assayed neat or diluted (1:10) when necessary and 20 µL of each sample was assayed in duplicate. The 3,3',5,5'-tetramethylbenzidine substrate (34021; Thermo Scientific, Waltham, MA, USA) was developed for 30 minutes at room temperature and stopped by the addition of 2 mol L -1 H 2 SO 4 . The absorbance in each well was then read at 450 nm. The assay had a detection limit of 0.05 ng mL -1 , an intra-assay coefficient of variation of 2.2% and an inter-assay coefficient of variation of < 10%.

| Immunohistochemistry and cell counting
At the end of the behavioural testing (ie, 90 minutes after the ASR task), rats were transcardially perfused with 4% paraformaldehyde in 0.1 mol L -1 phosphate buffer after anaesthesia with Euthatal®

| Statistical analysis
Analyses were conducted using spss, version 23 (IBM Corp., Armonk, NY, USA) and r, version 3.5.3 (R Foundation for Statisitical Computing, Vienna, Austria). Body weight and AGD data were analysed using repeated-measures ANOVAs (for one control female, AGD data were not included as a result of a missing value) and VO data were analysed using an independent samples t test. Hormone data were analysed using non-parametric statistics (Kruskal-Wallis test and Dunn's posthoc test with Bonferroni correction) as a result of a lack of normality in the data. One Antide-treated female did not appear to exhibit suppression of the HPG axis (ie, both the progesterone and LH serum levels fell above of the 95% percentile on PND 55/56), which strongly suggests that the Antide treatment was not successful for this subject, and the data for this female were therefore excluded from all analyses (whether this subject was included or not had minimal effects on the alpha values throughout). Adolescent social interaction data were analysed using a repeated-measures ANCOVA, with PND as the repeated measure and the identity of the play partner as a covariate. The EPM and SN data were analysed using repeated-measures ANOVA with Tukey's post-hoc test (in the EPM analyses, one Antide-treated female was excluded because the animal left the apparatus during the test) and the remaining behavioural data were analysed using one-way ANOVAs. Cell count data were analysed using two repeated-measures ANOVAs. All data are presented as the mean ± SD, except hormone data, which are presented as medians and interquartile ranges.

| Body weight
Antide-treated and control females did not differ in body weight (F 1,25 = 0.39, P = 0.540) (Figure 1) and the interaction between group and age was also not significant (F 9,225 = 1.65, P = 0.103). The main effect of age was significant (F 9,225 = 1843.89, P < 0.001).

| Hormone levels
Serum progesterone levels differed significantly between groups at PND 55/56 (P < 0.001) ( Table 1), with Antide-treated females having lower average progesterone levels than control females and higher average progesterone levels than control males (all pairwise comparisons, P < 0.05). At PND 69/70, serum progesterone levels differed between groups (P < 0.001) as a result of control males having lower average progesterone levels than both control females (P < 0.001) and Antide-treated females (P < 0.001), with no difference between the two groups of females (P = 1.000).

Serum LH levels differed between groups at PND 55/56
(P = 0.001) ( Table 1) as a result of Antide-treated females having lower average LH levels than both control males (P = 0.001) and control females (P = 0.017), whereas control males had similar LH levels to control females (P = 1.000). At PND 69/70, serum LH levels differed between groups (P = .0007) as a result of control males having higher average LH levels than Antide-treated females (P = 0.015) and control females (P = 0.028), with no difference between the two female groups (P = 1.000).

| Adolescent social behaviour
The total number of social interactions did not differ between groups (F 2,15 = 1.07, P = 0.367) and did not differ across postnatal days (F 3,45 = 1.92, P = 0.140) (Figure 2). Although Antide-treated females exhibited a reduction in social interactions on PND 43 (ie, 1 day after the second Antide injection), the interaction between group and day was not significant (F 6,45 = 1.36, P = 0.262).

| LDB
The total number of transitions between sections of the LDB dif-

| SN test
The

| ASR
The total number of immobilisation responses differed between groups (F 2,38 = 4.34, P = 0.020) ( Figure 6A), with control males, on average, exhibiting a higher number of immobilisations than control females (P = 0.022), but not Antide-treated females (P = 0.081), with no difference between the female groups (P = 0.868). The time spent immobile following a startle noise also differed significantly between groups (F 2,38 = 7.34, P = .002) ( Figure 6B), with control males, on average, spending longer immobile than both control females (P = 0.004) and Antide-treated females (P = 0.009), with no difference between the two female groups (P = 0.960).

| Cell counts
Cell counting, which was carried out on the tissue sections that were collected 90 minutes after the ASR task, revealed that the

| D ISCUSS I ON
The results of the present study indicate that treatment of adoles- example, on average, control males exhibited less locomotion in the OF, as well as a higher preference for novel objects in the NOR task, than females, with Antide-treated females not differing from control females on these tasks. In addition, no group differences in c-Fos cell counts were found in nuclei of the medial and basolateral amygdala.
Overall, these data suggest that a moderate delay in exposure to gonadal hormones during adolescence does not significantly impact upon anxiety-related behavioural and brain development in female rats, which contrasts with previous studies suggesting that adolescent exposure to testicular hormones has subtle, long-term effects on behaviour in male rodents. [24][25][26][27][28] One of the benefits of using GnRH antagonists, rather than ovariectomies, to manipulate gonadal hormone levels is that, once the antagonist has been metabolised, the HPG axis is reactivated. The dose of Antide used in the present study was identical to the dose used in a previous study in our laboratory, 61  ble delay in VO (2-3 days) has been found in previous studies that exposed female rats to either alcohol or stress during the adolescent period. 81,82 The delay in VO in Antide-treated female rats did not appear to result from a more general developmental delay, given that average body weight did not differ between the two groups of females. If anything, Antide-treated females gained slightly more weight than control females, which is consistent with evidence that pre-pubertal ovariectomy leads to weight gain in female rats. 33,35 Average AGD did not differ between the two female groups and thus rodents where AGD can show some plasticity during adolescence and adulthood in response to circulating androgen levels. 28,83,84 During adolescence, Antide-treated and control females exhibited similar levels of social interactions with novel partners, which suggests that suppressing the HPG axis did not have immediate activational effects on social behaviour. This finding is consistent with a previous study showing that the total amount of adolescent social play was not influenced by pre-pubertal ovariectomy in female rats. 33 By contrast, a recent study reported increased frequencies of play with a familiar partner following pre-pubertal gonadectomy in female Siberian hamsters. 85 In addition, subtle aspects of posture during social interactions have been shown to be influenced by early gonadal hormone exposure in rats, 31 leaves open the possibility that peri-pubertal gonadal hormones have short-term effects on locomotor exploration. In the present study, control male and control female subjects did not differ in the amount of time spent in the exposed sections of each apparatus, which could potentially be explained by the handling that subjects received during adolescent testing. Adolescent handling has been shown to reduce later anxiety-like behaviour, 87 whereas adolescent social interactions per se do not alter behavioural responses on these tasks. 88 The lack of sex differences in control subjects leaves open the possibility that the effects of suppressing the HPG axis might only be revealed following specific adolescent experiences.
In the NO task, control males exhibited a higher preference than control females for the novel object, which is consistent with the sex difference that we have previously reported in adolescent rats. 17 In this task, Antide-treated females had an average preference score that was intermediate between that of control males and control females. Similarly, in the ASR task, the number of immobilisations shown by Antide-treated females was intermediate between the scores for control males and control females. By contrast, the total time spent immobile did not follow the same pattern and instead the standard sex difference was replicated, with males on average spending more time immobile than both groups of females. 60 Although these data suggest that Antidetreated females might differ slightly from control females on some measures of responsiveness to novelty and fear-inducing stimuli, this interpretation of the results was not supported by the underlying brain activity because no differences in c-Fos labelling were evident between groups in the basolateral or medial amygdaloid nuclei. This lack of between-group differences in c-Fos measures was unexpected, given the sensitivity of the BLA to stressors, including aversive acoustic noises. 89 A potential explanation is that measures of immobilisation in the ASR task do not strongly reflect aversive responses, and future studies could benefit from employing startle amplitude measures instead. Overall, we conclude that suppressing the HPG axis during adolescence did not result in marked behavioural effects or differences in c-fos activity in the basolateral or medial amygdaloid nuclei.
In summary, although previous research has suggested that adolescent exposure to ovarian hormones has long-term effects on subtle aspects of social behaviour, [30][31][32] the results of the present study do not provide strong evidence for similar effects on anxiety-like Abbreviations: MeAD, anterodorsal medial; MeAV, anteroventral medial; MePD, posterodorsal medial; MePV, posteroventral medial; BLA, Basolateral anterior; BLP, Basolateral posterior; BLV, Basolateral ventral.
TA B L E 2 c-Fos + cell counts in the medial and basolateral amygdaloid nuclei in control males (n = 14), Antide-treated females (n = 13) and control females (n = 14) (per mm 2 ; data are presented as the mean ± SD) behaviour. One of the limitations of previous experimental studies that have used ovariectomy techniques is the possibility that the surgical procedure itself acts as a stressor and thus introduces a potential confounding factor. 90 The use of injectable GnRH antagonists to suppress the HPG axis provides a useful, alternative approach. Given that a range of factors can influence circulating gonadal hormone levels, including alcohol and stress exposure, 91,92 the long-term effects of manipulating the HPG axis during adolescence deserve continued investigation.

DATA AVA I L A B I L I T Y
The data that support the findings of this study are available from the corresponding author upon reasonable request.