Chronic androgen excess in female mice does not impact luteinizing hormone pulse frequency or putative GABAergic inputs to GnRH neurons

Abstract Polycystic ovary syndrome (PCOS) is associated with androgen excess and, frequently, hyperactive pulsatile luteinizing hormone (LH) secretion. Although the origins of PCOS are unclear, evidence from pre‐clinical models implicates androgen signalling in the brain in the development of PCOS pathophysiology. Chronic exposure of female mice to dihydrotestosterone (DHT) from 3 weeks of age drives both reproductive and metabolic impairments that are ameliorated by selective androgen receptor (AR) loss from the brain. This suggests centrally driven mechanisms in hyperandrogen‐mediated PCOS‐like pathophysiology that remain to be defined. Acute prenatal DHT exposure can also model the hyperandrogenism of PCOS, and this is accompanied by increased LH pulse frequency and increased GABAergic innervation of gonadotrophin‐releasing hormone (GnRH) neurons. We aimed to determine the impact of chronic exposure of female mice to DHT, which models the hyperandrogenism of PCOS, on pulsatile LH secretion and putative GABAergic input to GnRH neurons. To do this, GnRH‐green fluorescent protein (GFP) female mice received either DHT or blank capsules for 90 days from postnatal day 21 (n = 6 or 7 per group). Serial tail‐tip blood sampling was used to measure LH dynamics and perfusion‐fixed brains were collected and immunolabelled for vesicular GABA transporter (VGAT) to assess putative GABAergic terminals associated with GFP‐labelled GnRH neurons. As expected, chronic DHT resulted in acyclicity and significantly increased body weight. However, no differences in LH pulse frequency or the density of VGAT appositions to GnRH neurons were identified between ovary‐intact DHT‐treated females and controls. Chronic DHT exposure significantly increased the number of AR expressing cells in the hypothalamus, whereas oestrogen receptor α‐expressing neuron number was unchanged. Therefore, although chronic DHT exposure from 3 weeks of age increases AR expressing neurons in the brain, the GnRH neuronal network changes and hyperactive LH secretion associated with prenatal androgen excess are not evident. These findings suggest that unique central mechanisms are involved in the reproductive impairments driven by exposure to androgen excess at different developmental stages.


| INTRODUC TI ON
Polycystic ovary syndrome (PCOS) is a common endocrinopathy, reported to affect approximately 10% of women of reproductive age around the world. 1,2 PCOS is characterised by the presence of at least two of three diagnostic criteria, including clinical and/or biochemical hyperandrogenism, oligo-or anovulation, and a polycystic morphology of the ovary. 3 Approximately 75% of women diagnosed with PCOS present with luteinizing hormone (LH) hypersecretion, 4,5 and more than 90% of patients are likely to have an increased LHto-follicle stimulating hormone ratio. 6 PCOS is also associated with a number of co-morbidities such as metabolic syndrome, which can include obesity, impaired glucose handling and insulin insensitivity. 7 The aetiology for the diverse pathogenesis of PCOS is not clear and may be multifactorial. 8 Although specific PCOS origins remain unclear, exposure to androgen excess is linked to the development of PCOS pathophysiology in women and in several pre-clinical animal models. 9 The key features of PCOS are recapitulated in non-human primates, 10 sheep, 11 and rodents 12-14 exposed to androgen excess. Of interest, different exposure paradigms drive the development of different PCOS-like phenotypes. In the mouse, although chronic androgen exposure from 3 weeks drives anovulation and a metabolic syndrome, 13,15 acute prenatal exposure to androgen excess drives a lean PCOS-like phenotype of anovulation and hyperandrogenism and that lacks a robust metabolic phenotype. 12,14 This suggests that exposure to androgen excess within different developmental windows contributes to the development of different disease phenotypes.
Irrespective of treatment paradigm, there is evidence across several pre-clinical models suggesting that androgen excess-mediated development of PCOS features is likely to involve the brain. [16][17][18][19] Many of the PCOS-like traits observed following chronic exposure to the non-aromatizable androgen dihydrotestosterone (DHT) from postnatal day (PND)21, including anovulation, increased body weight, adiposity, and dyslipidaemia, can be rescued by neuronspecific deletion of the androgen receptor (AR). 15 This suggests that a brain-specific AR-dependent mechanism mediates the impact of chronic DHT exposure on the development of acyclicity and metabolic features. 15 However, the specific mechanisms remain to be determined. Acute, prenatal androgen (PNA) exposure results in increased gonadotrophin-releasing hormone (GnRH) neuron spine density 20 and increased GABA connectivity with GnRH neurons in both mice and sheep. 14,20,21 In the PNA mouse, this reprogramming is associated with impaired negative feedback regulation of GnRH/LH secretion, 20,22 increased GnRH neuron firing, 14,23 and hyperactive LH secretion. 20 The impact of androgen excess from a later developmental timepoint on these features remains to be determined in this common mouse model of PCOS-like hyperandrogenism. Although female hyperandrogenism is associated with blunting negative feedback and a coincident rise in LH, 12,24 female androgen excess delivered in adulthood has been shown to blunt LH secretion in other models. 25 Here, we used transgenic GnRH-green fluorescent protein (GFP) mice, 26 treated with chronic DHT from PND21 to model the hyperandrogenism of PCOS that promotes reproductive impairments and a metabolic syndrome, 13,15,27 to assess LH secretion dynamics and anatomical evidence for GABAergic innervation to GnRH neurons.
We also assessed the number of AR and oestrogen receptor α (ERα) immunoreactive cells within specific hypothalamic nuclei associated with steroid hormone feedback.

| Animals
Female GnRH-GFP mice were used for all experiments and were bred and housed at the University of Otago Biomedical Research Facility (Dunedin, New Zealand). All protocols and procedures were approved by the University of Otago Animal Ethics Committee (Dunedin, New Zealand) and performed in accordance with the regulations of the Australasian and New Zealand Council for the Care of Animals in Research and Teaching. All animals were housed under a 12:12 h light/dark photocycle (lights on 6.00 AM) with food and water available ad libitum.
The hyperandrogenism of PCOS was modelled with chronic exposure to excess androgens from a peri-pubertal timepoint. 13,15,27 SILASTIC capsules (inner diameter, 1.47 mm; outer diameter, 1.95 mm; 1 cm; Dow Corning; 508-006) filled with approximately 10 mg of dihydrotestosterone (DHT) (n = 6) or left empty (blank; n = 7) were placed at PND 21 under isofluorane anaesthesia (2%). A small intrascapular incision was made and a s.c. pocket was created down the length of the back towards the tail with sterile forceps. The capsule was then placed into this pocket and the incision was sutured closed to keep the capsule in place for 90 days (13 weeks; experimental outline in Figure 1). Mice were weighed weekly, and daily handling and habituation training began from week 8 to prepare the mice for serial tail-tip blood collection. Vaginal cytology was assessed from for the final 3 weeks of the experiment to assess oestrous cyclicity. At 13 weeks, when central mechanisms are involved in the reproductive impairments driven by exposure to androgen excess at different developmental stages.

K E Y W O R D S
androgens, GABA, GnRH, LH, membrane/nuclear, receptors animals were in the dioestrous stage of the cycle, animals were killed humanely by pentobarbital overdose (3 mg per 100 μL), and then transcardially perfused with ice-cold 4% paraformaldehyde (PFA) in 0.1 m phosphate buffer. Brains were dissected from the skull and post-fixed for 1 h at room temperature (RT) in 4% PFA before being transferred to 30% sucrose in Tris-buffered saline (TBS) overnight for cryopreservation. Once saturated, brains were sectioned coronally at a thickness of 30 μm on a freezing stage microtome (Leica 2400; Leica Biosystems) and collected in three series.

| Measuring pulsatile LH
Mice were handled daily for habituation 4 weeks prior to serial tailtip blood collection. Sampling was carried out twice, 2 weeks apart on dioestrus. As previously described, 28 4 μL of whole blood was collected every 6 min for 2 h (10.00 AM to 12.00 PM) from a small incision in the tail-tip. Blood was then immediately suspended in 50 μL of phosphate-buffered saline with 0.05% Tween-20 before being snap frozen on powdered dry ice. Samples were stored at −20°C until LH was measured by an ultrasensitive sandwich enzyme-linked immunosorbent assay (ELISA), as described previously. 20,28,29 Briefly, 96-well high affinity binding plates were  Pulses/hr

| Microscopy and image analysis
GnRH neurons were imaged using an inverted Nikon A1R confocal microscope (Nikon Instruments Inc.), with 488 and 543 nm lasers. GnRH neurons (8-10 per animal) were randomly selected across two representative rPOA sections from each animal (DHT, n = 6; blank, n = 6) and images were captured at 40× magnification (0.5μm Z-step, 1 AU pinhole, digital magnification 2×). Using NIS-Elements AR 4.5 (Nikon Instruments Inc.), images were analysed for VGAT appositions and GnRH spines on the GnRH neuron soma and at intervals of 15 μm along the primary dendrite as previously described. 20,31,32 VGAT puncta were considered in close apposition to GnRH neurons when there was an absence of black pixels between the cyan and magenta signals. Spines were defined as spiny protrusions from the cell body measuring greater than 1 μm and less than 5 μm. 33,34 AR and ERα stained brain sections were imaged using bright-

| Statistical analysis
All statistical analyses were carried out in Prism, version 9.0 (GraphPad Software Inc.

| Recapitulation of the chronic androgen excess model
Mice with DHT capsules inserted at PND21 ( Figure 1A) exhibited significant weight gain compared to blank capsule controls (F 1,11 = 11.1; p mean effect of treatment = .0067) ( Figure 1B;) as observed previously. 13 DHT treated mice also exhibited a pronounced loss of oestrous cyclicity ( Figure 1C,D), with the majority of mice remaining completely acyclic. DHT treated mice spent significantly more time in dioestrus compared to controls (two-way ANOVA; F 2,33 = 403.8; p mean effect of oestrous cycle < .0001) ( Figure 1D) and a reduced amount of time in both oestrus and pro-oestrus (p < .0001) ( Figure 1D).

| Chronic androgen excess does not alter LH pulsatility
LH secretion dynamics over time were measured with serial tail-tip blood sampling from intact female animals in dioestrus (Figure 2A,B).
The total amount of LH released over the 2-h sampling window (calculated as area under the curve) was not different between blank and DHT-treated mice (t 11 = 0.4987; p = .6288) ( Figure 2C).

| Chronic androgen excess does not affect GnRH neuron spine density or putative GABA appositions
To determine whether the chronic DHT model exhibits altered GnRH neuron morphology or evidence of enhanced GABAergic input, the number of VGAT puncta (magenta) in close apposition with the GnRH neuron soma and primary dendrite (cyan) was quantified ( Figure 3A,B). Immunolabelling of GFP in brain tissue from GnRH-GFP mice enabled robust visualisation of GnRH neuron morphology, including lengthy dendrites and spiny protrusions. Labelling of VGAT revealed widespread punctate labelling and VGAT puncta were frequently found closely apposed with the GnRH neuron soma and dendrite ( Figure 3A,B, solid arrowheads).
No differences in the total density of VGAT appositions to GnRH neurons were detected between DHT treated animals and blank controls (t 10 = 0.02653; p = .9794) ( Figure 3C), nor were any differences evident in any specific region of the neuron (repeated measures two-way ANOVA; F 1,10 = 0.2460, p mean effect of treatment = .6306; p = .99 for all post-hoc tests) ( Figure 3C). Total GnRH neuron spine density was also found to be similar between DHT and blank animals (t 10 = 0.3643; p = .7232) ( Figure 3D). Likewise, the spine density in specific neuronal sub-compartments, including the soma and 15μm intervals along the primary dendrite, was also not different between groups (repeated measures two-way ANOVA;    Table 1). By contrast, the number of AR positive cells was significantly elevated in all regions examined from mice chronically exposed to DHT (p < .0001) ( Figure 5 and Table 1). Figure 6 shows lower magnification views of DHT labelling at a more anterior and more posterior level of the hypothalamus to demonstrate the general upregulation of AR expression.
Semi-quantitative analysis of AR and ERα positive cells throughout hypothalamic and some limbic regions shows that, although ERα

| DISCUSS ION
We show that females chronically exposed to the non-aromatisable androgen DHT from a peri-pubertal timepoint for 3 months develop the PCOS-like traits of impaired oestrous cyclicity and increased body weight, as shown previously. 13,15,36,37,38 Although this mouse DHT-treated mice with a neuron specific deletion of AR exhibit partially rescued ovulatory function, protection from adipocyte hypertrophy and ameliorated dyslipidemia. 15 Of interest, knockout of AR from both neurons and adipose tissue protects DHT-treated mice to an even greater extent, with the majority of mice resuming regular oestrous cyclicity. 39 The present data suggest that, although chronic DHT exposure initiated from 3 weeks of age in mice is likely acting in the brain given the robust increase in the number of AR-expressing cells in the hypothalamus, it has no significant effect on LH pulse frequency.
Androgen excess is associated with an impaired ability for gonadal steroid hormones, particularly progesterone, to exert negative feedback effects on GnRH/LH pulse generation. In healthy women, Chronically elevated androgens in mice may instead generate an "androgen clamp" that interferes with the preovulatory surge and inhibits LH secretion, as has been shown in rats. 45,46 Acute exposure of female rats to elevated testosterone abolishes the endogenous and the oestrogen-induced LH surge. 45 Similarly, chronic DHT exposure in rats from 3 weeks of age abolishes the oestrogen-primed surge. 46 By contrast to the present study, however, LH pulses were not detectable in ovariectomised DHT-treated treated rats. 46 In a more recent study, LH pulses were measured in serial tail-tip blood samples from ovariectomised mice with DHT capsules inserted at 5 weeks of age. DHT was found to dose-dependently reduce LH pulse frequency and amplitude and increase the inter-pulse interval. 25  Although acute prenatal androgen excess is reported to result in a small but significant increase in ERα and AR expression in the RP3V of adult females, 15 we identified that chronic DHT dramatically increases AR expression throughout the RP3V and ARN and has no effect on ERα expression in any region investigated.
Although ER expression is relatively stable in the brain, there is evidence to suggest that elevated oestradiol down-regulates ERs 52 and that ERs are upregulated in the absence of circulating oestradiol. 53 The absence of ERα expression changes here is in line with the absence of differences in circulating oestradiol in this model. 13 AR expression in the hypothalamus is well known to be autoregulated by androgens in both males and females. 53 Here, the number of AR immunoreactive cells in the brain was increased by 4-6-fold in females chronically exposed to DHT. Although the phenotypes of those neurons with upregulated AR remain to be determined, increased AR expression in the PNA model has been identified in agouti-related protein neurons, 54  Challenging this notion, administration of testosterone as part of a gender affirming hormone therapy to female-to-male transgender individuals does not promote insulin resistance or a metabolic phenotype. 60 Additionally, genetic causes of extreme insulin resistance in women result in high circulating testosterone, suggesting that hyperinsulinaemia promotes hyperandrogenism, not necessarily the reverse. 61 In any case, female hyperandrogenism is associated with both blunting negative feedback control of the HPG axis, as well as clamping down HPG axis output. The outcome may be dependent upon the developmental window of first exposure, the relative level of the androgen exposure, and the aromatisable nature of the elevated androgen. Although artificially increasing DHT enables the specific investigation of AR-mediated effects, endogenous or exogenous testosterone will result in both ER-and AR-mediated effects following the metabolism of testosterone to oestradiol in the brain.
Together with previous reports on this model, our findings suggest that chronic DHT exposure from 3 weeks of age recapitulates an obese, acyclic, hyperandrogenic phenotype that is not associated with a hyperactive HPG axis. Although there are no perfect models of PCOS, which only naturally occurs in humans and non-human primates, 9 preclinical models remain critical with respect to determining the pathogenesis of androgen excess in driving PCOS features in females. Reductionist models such as this that highlight the consequences of female androgen excess can identify novel targets for intervention in a disease that is currently only treated symptomatically as a result of a lack of understanding of its aetiology and pathogenesis.

ACK N OWLED G M ENTS
Open access publishing facilitated by University of Otago, as part of the Wiley -University of Otago agreement via the Council of Australian University Librarians.

CO N FLI C T O F I NTE R E S T S
The authors declare that they have no conflicts of interest.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/jne.13110.

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