Impaired steroid hormone feedback in polycystic ovary syndrome: Evidence from preclinical models for abnormalities within central circuits controlling fertility

Abstract

Polycystic ovary syndrome (PCOS) is the most common endocrinopathy and cause of infertility in women of reproductive age worldwide. Despite diagnostic features of anovulation, polycystic ovarian morphology, and high androgen secretion indicating the syndrome are the result of ovarian dysfunction, alterations to central neuroendocrine circuits that control reproductive capacity may drive PCOS symptoms. Resistance of gonadotrophin-releasing hormone (GnRH) neurons in the hypothalamus to inhibition by sex steroid hormone-negative feedback leads to a rapid frequency of pulsatile gonadotrophin secretion, which, in turn, drives the ovarian features of the disease. As GnRH neurons do not express steroid hormone receptors, impaired negative feedback is hypothesized to occur within an upstream network that controls GnRH pulse generation. This review will discuss the latest work from preclinical animal models of PCOS used to dissect the specific central mechanisms involved in impaired steroid hormone feedback. In particular, this review will focus on research that indicates neurons in the arcuate nucleus of the hypothalamus that express Kisspeptin, Neurokinin B and Dynorphin (KNDy cells) or γ-aminobutyric acid are targets of androgen-mediated impairment of steroid hormone feedback. Finally, this review will explore the development of therapeutic agents targeting neurons that control LH pulse frequency to resolve PCOS symptoms in the clinic.

1 |. DISRUPTION OF THE HYPOTHALAMIC–PITUITATRY–GONADAL AXIS IN POLYCYSTIC OVARY SYNDROME

Polycystic ovary syndrome (PCOS) is the most common cause of infertility in women of reproductive age, affecting between 8% and 13% of women worldwide.¹ PCOS presents as a heterogenous and complex interaction of reproductive, metabolic and psychological traits that is difficult to diagnose and manage.² A recent international evidence-based consensus on the assessment and management of PCOS endorsed the 2003 Rotterdam Criteria that PCOS diagnosis requires at least two of the following symptoms in the absence of other endocrine disorders; clinical and/or biochemical evidence for hyperandrogenism, irregular or absent menstrual cycles and/or polycystic ovarian morphology.³ Although the above diagnostic criteria focus on peripheral indicators of reproductive dysfunction, regulation of the ovaries and steroid hormone secretion is ultimately dictated by the activity of neurons in the brain. Gonadotrophin-releasing hormone (GnRH) neurons located in a scattered continuum throughout the preoptic area and hypothalamus represent the final common output cell for the central control of reproductive function. The pulsatile secretion of GnRH peptide into the portal vasculature of the anterior pituitary gland stimulates release of the gonadotrophins luteinizing hormone (LH) and follicle-stimulating hormone (FSH) into blood circulation.^4,5 Differential exposure of the ovary to LH and FSH shapes the ovarian cycle and the release of sex steroid hormones. In turn, steroid hormones, such as estradiol and progesterone, act back in the brain to regulate GnRH release.⁶ In the male, and for the majority of the female ovarian cycle, steroid hormones suppress GnRH/LH release through negative feedback. During the midfollicular phase in females, rising levels of estradiol drive a switch from negative to positive central feedback. This leads to a massive and continuous release of GnRH and LH (the preovulatory LH surge) and is necessary for ovulation in mammals. Following ovulation, the corpora lutea formed by the ruptured follicle secretes progesterone that provides negative feedback control and decreases LH pulse frequency.

Although GnRH is difficult to measure in PCOS patients due to release into the portal vasculature of the pituitary gland and rapid decay time, sampling of GnRH in the portal vasculature of nonhuman primates and sheep demonstrates that LH pulses faithfully mirror the episodic release of GnRH.^4,5 Serial blood sampling from women with PCOS revealed that LH pulse frequency was increased when compared to healthy control women,⁷ representing an increase in GnRH pulse frequency. As increased frequencies in GnRH release lead to the preferential release of LH, whereas decreasing frequencies lead to greater FSH release,⁸ over 70% of PCOS women also have an elevated LH:FSH ratio.⁹ Consequentially, elevated LH pulse frequency alters ovarian steroid hormone synthesis, favoring the production of androgens,¹⁰ whereas lower FSH release disrupts follicular development.¹¹ Alterations to intrinsic ovarian steroidogenesis that favour androgen production may also contribute towards hyperandrogenaemia in the syndrome.¹² Potentially, LH hypersecretion in PCOS is the result of an increase in the direct stimulation of GnRH release. For example, anti-Müllerian hormone (AMH) increases GnRH activity and LH release in rodents,¹³ and elevated circulating AMH is a common feature in PCOS patients.¹⁴ In addition, research in PCOS patients and preclinical models suggest LH hypersecretion may result from an impairment in steroid hormone feedback regulation of GnRH release within the brain.

2 |. IMPAIRED SEX STEROID HORMONE FEEDBACK IN PCOS

Analysis of circulating sex steroid hormones show that PCOS patients are normoestrogenic with reduced secretion of ovarian progesterone due to anovulation.¹¹ Although reduced progesterone-negative feedback may contribute to elevated LH release in adult PCOS women, the presence of LH hypersecretion before the development of anovulation in adolescent girls suggests an aetiology that precedes low progesterone.¹⁵ In groundbreaking studies, the exogenous delivery of estradiol and progesterone revealed that PCOS patients require higher concentrations of sex steroids to suppress LH pulses to the same frequency measured in control women.^16–18 These data suggest that elevated LH pulse frequency may arise from a reduction in hypothalamic sensitivity to estradiol and progesterone-negative feedback. In addition, studies suggest that hyperandrogenaemia in PCOS patients may play a key role in the development and maintenance of this impaired negative feedback. Discrete periods of hyperandrogenaemia during prenatal and peripubertal development are sufficient to induce PCOS-like neuroendocrine symptoms (discussed further below). The impaired inhibition of LH pulse frequency established by early adolescence is associated with the presence of hyperandrogenaemia.¹⁹ Further, LH levels and the LH:FSH ratio correlates with testosterone levels in PCOS women,²⁰ and treatment with the androgen receptor (AR) antagonist Flutamide is able to restore sensitivity to estradiol and progesterone-negative feedback in PCOS patients.²¹ These results support that (1) androgen exposure during critical developmental windows may play a primary role in establishing impaired steroid hormone feedback and (2) continuous hypothalamic exposure to high androgen levels in adolescent and adult patients may be required to sustain impaired steroid hormone feedback.

3 |. ANIMAL MODELS TO STUDY THE NEUROENDOCRINE PATHOGENESIS OF PCOS: THE ROLE OF PRENATAL ANDROGEN EXPOSURE

Perhaps unsurprising given the heterogenous presentation of PCOS, the aetiology for the syndrome has been linked to polygenic, epigenetic, metabolic, and developmental modifications.²² In line with this, environmental and genetic manipulations can induce PCOS symptoms in animal models. This includes the genetic overexpression or deletion of candidate genes in rodent models (such as the deletion of insulin and leptin receptors in pro-opiomelanocortin cells) to induce symptoms that recapitulate the PCOS phenotype (reviewed in Stener-Victorin et al.²³), manipulating light exposure in rodents to induce anovulation and polycystic ovarian morphology,²⁴ and elevating circulating androgen levels during critical periods of development. In particular, clinical and basic research suggest that a hyperandrogenemic intrauterine environment plays a significant role in programming the development of PCOS.²² Although ~60%–70% of daughters born to PCOS women develop the syndrome themselves,²⁵ under 10% of the syndrome’s heritability is accounted for by risk genes.²⁶ Therefore, changes within the maternal–fetal environment may account for the transgenerational inheritance of the syndrome.²⁵ At birth, daughters of PCOS women display a longer anogenital distance²⁷ and higher levels of facial sebum production,²⁸ both markers of high in utero testosterone exposure. Testosterone measured in amniotic fluid from daughters of hyperandrogenic PCOS women is significantly elevated compared with control women during mid-gestation,²⁹ which represents a critical window for development of the hypothalamus.

Importantly, symptoms that both closely resemble that of PCOS in women and manifest a similar trajectory of disease can be induced by discrete periods of prenatal androgen excess in a number of preclinical animal models. A PCOS phenotype can be induced by androgen excess during midgestation in the nonhuman primate and sheep, models with a comparable study of fetal development to that of human patients.²³ High androgens during late gestation in female rodents, an equivalent developmental timepoint to midgestation in primates, generates PCOS-like reproductive and neuroendocrine features.²³ More recently, it was identified that AMH, which is high in adult PCOS women, remains elevated during mid to late gestation in PCOS women who achieve pregnancy.¹⁴ The exogenous delivery of AMH into the circulation of pregnant mice induces maternal GnRH/LH stimulation of testosterone secretion which, subsequently, masculinizes the brain of female prenatal AMH-treated (pAMH) offspring.¹⁴ In addition, aromatase expression (Cyp19a1) is reduced in the placenta of AMH-treated dams, indicating placental protection of the female fetus from elevated maternal circulating androgens is reduced.¹⁴ As a result, pAMH adult female mice exhibit PCOS-like reproductive and neuroendocrine symptoms that mirror that of prenatal androgen-treated mice, specifically those exposed to prenatal dihydrotestosterone (pDHT mice).¹⁴ Currently, prenatal androgen exposure in the above species is most widely used to model at least two of the three Rotterdam Criteria for PCOS. Depending on the developmental period of exposure and the form of androgen manipulation used (testosterone (T), DHT, dehydroepiandrosterone or AMH), these models recapitulate PCOS features of irregular oestrus cycles, impaired estradiol and progesterone feedback on LH pulsatility, hyperandrogenaemia and varied metabolic changes (reviewed in detail in Stener-Victorin et al.²³). Reproductive and metabolic PCOS-like symptoms can also be induced by manipulation of androgen levels during peripubertal and adult life, such as though the chronic administration of DHT or an aromatase inhibitor (Letrozole) that prevents the conversion of testosterone to oestrogen.²³ However, the persistence of PCOS symptoms after adult androgen manipulations cease has not yet been demonstrated.

Importantly, the ability to transgenically label and manipulate neurons in a cell-specific manner in rodent models permits researchers to directly assess the impact of prenatal androgen exposure on specific populations within the GnRH neuronal network. Direct electrophysiological recordings of GnRH neurons expressing green fluorescent protein (GnRH-GFP) in mouse brain slices validated that GnRH neuron activity is higher in pDHT and pAMH adult female mice compared with prenatal vehicle-treated controls,^14,30 and in pAMH mice, reproductive symptoms could be ameliorated through antagonism of GnRH release either during the prenatal AMH insult or in adult life.¹⁴ These data support that androgen-mediated reproductive symptoms are driven by prenatal programming of excessive GnRH release and high GnRH release maintains the adult neuroendocrine pathophysiology. As GnRH neurons do not express AR and, further, do not express steroid hormone receptors required for estradiol and progesterone feedback, androgen interference with steroid hormone-negative feedback control of GnRH neuron activity is hypothesized to occur within a complex system of upstream cells. In support of this, the neuron-specific knockout of the AR in mice revealed that central neuronal AR signalling is required to generate acyclicity and anovulation that is induced by peripubertal DHT treatment in wildtype controls.³¹ In contrast, the knockout of AR from ovarian granulosa cells has no effect on the ability of DHT to induce acyclicity and measures of anovulation.³¹ Although this has not yet been demonstrated in a model of prenatal androgen exposure, these data highlight the importance of AR signalling in the brain for induction of PCOS symptoms. Currently, identifying the location and phenotype of cells with impaired steroid hormone sensitivity is under intense exploration in PCOS animal models. To date, investigators have reported alterations within a number of cellular phenotypes upstream to GnRH neurons, including γ-aminobutyric acid (GABA) neurons, kisspeptin cells and nonneuronal populations. Consistently, two main mechanisms of disruption have been reported within the GnRH neuronal network of PCOS models: (1) changes in steroid hormone sensitivity and peptide expression consistent with impaired negative feedback and (2) rewiring of synaptic connections within the GnRH neuronal network that may impair the communication of steroid hormone feedback signals to GnRH neurons.

4 |. ANDROGEN-MEDIATED INTERFERENCE WITH STEROID HORMONE SENSITIVITY IN THE GnRH NEURONAL NETWORK

It has been well documented that androgens interfere with central progesterone-negative feedback in females. Testosterone infusion reduces the expression of progesterone receptors (PRs), the classical receptor required for progesterone-negative feedback³² and blocks progesterone feedback effects on GnRH release in rodents.³³ This generated the hypothesis that high testosterone in PCOS women reduces estradiol-mediated PR expression within the hypothalamus. Supporting this, a reduction in PR expression was identified in hypothalamic regions controlling GnRH release in rodent and sheep models exposed to prenatal androgen excess.^34–37 Characterization of cells with changes in PR signalling upstream from GnRH neurons is under investigation, with cells expressing either the fast amino acid GABA or the neuropeptide kisspeptin identified as the most likely candidates (Figure 1).

FIGURE 1.

Interference of progesterone-negative feedback in the hypothalamus by androgens in polycystic ovary syndrome (PCOS). (A) In women with intact negative feedback, progesterone acts through the progesterone receptor (PR) at cells in the arcuate nucleus (ARC), likely γ-aminobutyric acid (GABA) and Kisspeptin, Neurokinin B (NKB) and Dynorphin (KNDy) neurons, to reduce the production of excitatory input to gonadotrophin-releasing hormone (GnRH) neurons (GABA, kisspeptin). At KNDy cells, GnRH pulse generation is regulated by NKB and dynorphin exciting and inhibiting kisspeptin release onto GnRH neurons, respectively. (B) High testosterone levels in PCOS women or PCOS-like animal models may interfere with the transcription of PR in GABA or KNDy cells within the ARC. At GABA cells, reduced progesterone-negative feedback may increase the number of excitatory GABAergic synaptic inputs to GnRH neurons (as represented by an increase in arrow number), leading to an increase in GnRH neuron excitability. At KNDy cells, impaired progesterone-negative feedback may reduce production of inhibitory dynorphin, which may lead to an increase in the release of excitatory kisspeptin (as represented by changes in arrow width). Potentially, an increase in NKB may also contribute to the release of excitatory kisspeptin. Changes in KNDy peptide release may result in an increase in the frequency of GnRH pulse generation in PCOS

A number of studies in rodent models indicate that GABA plays a critical role in mediating impaired progesterone-negative feedback to GnRH neurons. Contrary to its role as an inhibitory neurotransmitter in other neural systems, there is evidence that GABA can excite GnRH neurons. This unique response is hypothesized to result from GnRH neurons containing a high level of intracellular chloride, leading to depolarization (excitation) of the cell when GABA^A receptors are activated.^38,39 Accordingly, progesterone reduces GABAergic transmission to GnRH neurons in the brain slice preparation of mice collected under normal physiological conditions, which may result in reduced GnRH neuron activity.⁴⁰ The exogenous delivery of androgens in adult mice interferes with progesterone-negative feedback by increasing GABAergic drive to GnRH neurons,⁴¹ and, in the pDHT mouse, GABAergic drive to GnRH neurons is increased.⁴² To assess the specific GABA population in which androgen interference with progesterone feedback occurs, the expression of PR within hypothalamic regions has been mapped. Cells in the arcuate nucleus (ARC) of the hypothalamus are a likely site of progesterone-negative feedback, as microimplants of the PR antagonist RU468 into the ARC of the ewe blocks the inhibitory actions of progesterone upon LH secretion.⁴³ Consistent with this, PR expression is reduced in the ARC of mice, rat and sheep following prenatal androgen excess,^34,37 and this reduction has further been detected within ARC GABA cells that directly innervate GnRH neurons in pDHT mice³⁶ (Figure 1).

Alternatively, cells in the ARC that express kisspeptin may be involved. As less than 2% of ARC GABA cells express kisspeptin,⁴⁴ it is likely that kisspeptin neurons are intermingled with ARC GABA cells but do not represent the same population. Kisspeptin is a peptide expressed by cells in the brain that potently stimulates GnRH secretion and mutations to genes encoding kisspeptin or its receptor predispose the subject to hypogonadotropic hypogonadism.⁴⁵ Multi-labelling experiments revealed that kisspeptin cells in the ARC coexpress two other important mediators of GnRH release: the tachykinin neurokinin B (NKB) and the endogenous opioid peptide dynorphin.⁴⁵ As an abbreviation, these cells were termed Kisspeptin/NKB/Dynorphin (KNDy) neurons. The coexpression of inhibitory and excitatory peptides, and evidence for interconnected KNDy cells first drove forward the hypothesis that pulsatile secretion of GnRH is formed by NKB and dynorphin acting on reciprocally connected KNDy neurons to stimulate and inhibit kisspeptin stimulation of GnRH neurons, respectively.⁴⁵ Visualization of KNDy neuron activity using in vivo recordings of intracellular calcium, a proxy for neuron activity, revealed that this population exhibits synchronized episodic activity immediately before LH pulse secretion,^46,47 strongly indicating their role as the GnRH pulse generator. In addition, it is postulated that estradiol and progesterone may act directly at KNDy cells to mediate pulse frequency and amplitude. In brief, KNDy neurons are highly steroid hormone sensitive,⁴⁵ estradiol and progesterone alter KNDy peptide levels in a manner consistent with negative feedback in the rodent and sheep,⁴⁵ and can modulate the response of KNDy cells to NKB and dynorphin,⁴⁸ and the neurotransmitters glutamate and GABA.^49,50 In the ewe, progesterone-negative feedback on GnRH release depends on dynorphin signalling through its receptor, the κ-opioid receptor (KOR), in the ARC.⁵¹ The knockout of oestrogen receptor α (ERα) from kisspeptin cells in mice disrupts positive feedback.⁵² In some studies, negative feedback is not affected by ERα knockout from kisspeptin cells;⁵² however, a recent report demonstrated an elevation in LH pulse frequency.⁴⁹ The knockout of PR from kisspeptin cells in mice disrupts fertility via eliminating the LH surge and ovulation,⁵³ although the effect of absent progesterone signalling in kisspeptin neurons on LH pulse generation was not assessed.

Importantly, recent studies have demonstrated that KNDy cells are targeted by androgens in the PCOS-like state. In pDHT female mice, AR coexpression with KNDy cells is significantly elevated to levels seen in males, demonstrating greater androgen sensitivity by the population in the PCOS state.³⁵ In the pAMH mouse model, the conditional knockout of AR from kisspeptin cells (KARKO pAMH mice) prevented the development of delayed puberty, disrupted cyclicity and reduced fecundity in this model.⁵⁴ These data provide the tantalizing prospect that the reversal of elevated androgen signalling to kisspeptin cells is sufficient to prevent PCOS development, supporting treatment of PCOS women with therapeutic agents targeting KNDy peptide (discussed below). Although the impact of steroid hormones on LH pulse secretion in the KARKO pAMH mice were not assessed, a modest reduction in PR mRNA expression has been recorded within KNDy cells with elevated AR in pDHT mice,³⁵ supporting androgen interference with PR transcription in these cells (Figure 1). Downstream consequences from altered PR may include changes in the expression of KNDy peptide levels, which may, in turn, alter GnRH regulation. In rodents with postnatal treatment of letrozole or DHT, excitatory ARC kisspeptin and NKB gene expression is dramatically upregulated.^55,56 A recent study in the pDHT mouse did not identify the same changes in kisspeptin and NKB gene expression, but did detect a significant reduction in inhibitory dynorphin gene expression.³⁵ Although results vary between models, in general, an increase in GnRH/LH pulse frequency in PCOS-like models are associated with changes in KNDy peptide expression that is consistent with elevated excitatory output by KNDy cells.

5 |. ANDROGEN-MEDIATED CHANGES IN SYNAPTIC CONNECTIVITY WITHIN THE GnRH NEURONAL NETWORK

The organization of synaptic connections between cells in the brain, including the innervation of GnRH neurons, is reliant upon appropriate androgen exposure during male and female development.⁵⁷ Androgens also modulate synaptic plasticity in adulthood. Changes in the ontogeny of regulatory input to GnRH neurons and amongst upstream steroid hormone-sensitive networks may provide a mechanism through which impairments in the afferent control of GnRH neurons occurs in PCOS women. It is probable that the synaptic organization of multiple steroid hormone-sensitive populations are altered by prenatal androgen exposure, as a number of cell types express AR during the initial prenatal androgen exposure and during adult life.

Recordings from brain slices collected using pDHT GnRH-GFP mice revealed GnRH neurons receive increased postsynaptic current frequency, indicating an increase in the number of synaptic vesicles apposing GnRH neurons.⁴² This was verified by anatomical investigation of GABAergic synaptic connections to GnRH neurons using immunolabelling for the vesicular GABA transporter in prenatal testosterone-treated (pT) sheep,⁵⁸ pAMH mice and pDHT mice,^14,36 which identified a robust increase in the innervation of GnRH cell bodies and proximal dendrites by GABA cells. Viral-mediated tract-tracing in pDHT mice demonstrated this elevated input originated from GABA neurons located within the hypothalamic ARC.³⁶ Selective activation of projections from these cells to GnRH neurons increased LH release in control mice and was, surprisingly, blunted in pDHT mice,³⁹ potentially due to a reduction in available LH at the pituitary gland due to the past high frequency of release. In pDHT adult mice, elevated synaptic connectivity appears dependent on high circulating T, as treatment with the androgen-receptor antagonist Flutamide is able to reduce GABAergic connectivity and the activity of GnRH neurons to control levels.^42,59 However, GABAergic input was also elevated to GnRH neurons in pDHT mice before puberty and the development of hyperandrogenaemia.⁵⁹ Together, these data indicate that the prenatal exposure may directly programme changes within the hypothalamus before puberty that drives downstream hyperandrogenaemia. Hyperandrogenaemia in adult life may then contribute to the foundation of altered steroid hormone feedback at the hypothalamus.

In PCOS animal models, KNDy cells, as previously described, display changes in steroid hormone sensitivity in conjunction with robust changes in synaptic connectivity. In pT sheep and pDHT mice, synaptic input from cells expressing the neurotransmitter glutamate (identified by vesicular glutamate transporter 2-positive terminals) is significantly reduced to KNDy cells.^35,60 In pT sheep, at least part of this reduced glutamate input was colocalized with kisspeptin and NKB axon immunolabelling, suggesting the reciprocal connectivity between KNDy neurons is reduced.⁶⁰ Potentially, these changes provide a mechanism through which a reduction in inhibitory dynorphin signalling impairs the progesterone break on pulse frequency. pT sheep also display an increase in GABAergic input onto KNDy cells, whereas in pDHT mice, GABAergic input to KNDy cells is reduced.³⁵ These morphological differences may be related to differences between the two species in the responses of KNDy cells to GABA. In mice, there is evidence for both excitatory⁶¹ and inhibitory⁶² effects of GABA on KNDy cells; similar electrophysiological studies have not been carried out in sheep primarily because of the lack of transgenic animals bearing KNDy-specific markers. Alternatively, GABA synapses may originate from cells with a different neuropeptide phenotype with correspondingly differential effects on KNDy cell activity. In mice, rabies-mediated tract-tracing revealed that reduced synaptic input to KNDy cells arises from other steroid-hormone sensitive nuclei, including the preoptic area and anteroventral periventricular nucleus.³⁵ Intra-ARC connections have not yet been assessed in PCOS mice, but it is possible KNDy cells receive modified synaptic input from ARC GABA cells that also have elevated innervation of GnRH neurons, reciprocally connected KNDy neurons, or, other cell populations in the ARC responsible for integrating reproduction with other internal signals, such as metabolism.⁴⁵ Finally, in pT sheep, synaptic input from KNDy cells to GnRH neurons is reduced. Although initially somewhat contradictory to the idea of a quickened pulse generator, subpopulations of KNDy cells in sheep, and likely also in humans, project to GnRH neuron cell bodies in the POA and are activated during the preovulatory LH surge.⁶³ Therefore, reduced KNDy input to GnRH neuron cell bodies may reflect a mechanism through which estradiol-positive feedback mediated generation of the GnRH/LH surge is disrupted in pT sheep. Together, KNDy cells exhibit changes in: (1) reciprocal connections between KNDy cells, which may alter the transmission of feedback signals within the population; (2) inputs from afferent populations that may alter the regulation of KNDy cell activity from cells relaying external and internal cues to the pulse generator; and (3) synaptic input to GnRH neurons from KNDy cells that may impair preovulatory surge generation. It remains to be determined whether an increase in GnRH/LH pulse frequency correlates with synchronized episodic activity in KNDy cells. A recent study using mouse brain slices did not find any difference in the activity of ARC NKB (KNDy) neurons in brain slices from control and pDHT mice,⁶⁴ although peripubertal letrozole treatment in mice drives an increase in cFos expression (a marker of cell activation) within KNDy neurons.⁵⁶ However, in vivo studies will be required to determine whether KNDy neurons exhibit an increased frequency of episodic activity that is in sync with hypersecretion of LH pulses in PCOS-like models. Although it is hypothesized these architectural changes are induced by neuronal AR activation with the subsequent disruption of oestrogen and PRs, it is also possible that other mechanisms, such as microglial sculpting of synapse architecture during development, may mediate these findings. In support of this, a recent study in pDHT mice identified changes in microglia number and morphology and reduced evidence for microglial engulfment of GABAergic synapses within the proximity of GnRH neuron cell bodies at different stages of development, which may suggest microglia are not available to prune GABAergic synapses in PCOS.⁶⁵ Although it is possible that microglial sculpting also alters KNDy synaptic structure, the above study did not report changes in microglial morphology or number in the ARC.

6 |. STRATEGIES TO TARGET THE BRAIN IN PCOS PATIENTS

Major strategies for the treatment of PCOS have involved either hormonal or lifestyle interventions. Lifestyle interventions, such as diet and exercise to induce weight loss, ameliorate PCOS symptoms in women with an obese or metabolic phenotype.⁶⁶ However, 30%–50% of PCOS women with hyperandrogenaemia are lean. Hormonal interventions include anti-androgen therapies and hormonal contraception. These treatments are used to mitigate the effects of hyperandrogenaemia and oligomenorrhea but have varied success and need to be continually altered according to the patients reproductive needs.⁶⁶ Importantly, these interventions follow the establishment of programmed PCOS symptoms and do not address the underlying pathogenesis of the disorder. Although prenatal androgen excess has been linked to the pathogenesis of PCOS, reducing testosterone in utero in human patients is an unlikely strategy due to potential adverse effects on normal female developmental processes that are influenced by androgens and androgen metabolites. Further, the identification of reliable biomarkers before PCOS onset at puberty will be required to safely predict the need for interventional strategies before menarche is established in adolescent PCOS patients.

Alternatively, direct manipulation of hypothalamic circuits regulating GnRH pulse generation may prevent development of the downstream PCOS ovarian phenotype. Promising strides have been made in testing antagonism of receptor for NKB, the neurokinin 3 receptor (NK3R). In male and female patients, NK3R antagonism reduces LH release⁶⁷ and in PCOS patients, NK3R antagonism reduces LH pulse frequency and total testosterone levels.⁶⁸ Future studies will require evaluation of long-term treatments to determine whether antagonism of NK3R is a viable clinical alternative to current hormonal therapy. Of note, the median eminence, where GnRH and other hypophysiotropic hormones are released into the portal vasculature, contain leaky fenestrated capillaries, which permit entrance of large molecules into nearby brain regions, including areas of the ARC.⁶⁹ This provides a unique anatomical advantage, which may permit researchers and clinicians to specifically modulate neuronal systems controlling fertility with simple peripheral intravenous delivery of peripherally restricted drugs. In a recent example of this strategy, Navarro and colleagues⁷⁰ demonstrated that peripherally restricted KOR agonists reduce the magnitude of LH pulse secretion in pDHT mice, while reducing testosterone levels and restoring cyclicity and ovulation.⁷⁰ The development of peripherally restricted drugs may therefore specifically target cells controlling fertility without adversely affecting other brain systems that control cognition, mood and other higher-order functions. This provides a promising avenue for the development of future therapeutic treatments in PCOS.

7 |. CONCLUSIONS

Clinical and animal studies support the hypothesis that alterations to neurons in the brain that mediate progesterone-negative feedback to GnRH neurons may underlie the PCOS neuroendocrine phenotype. These changes may be programmed in utero by AMH or androgen excess and/or established by androgens over puberty. Work in preclinical models have pinpointed two progesterone-sensitive neuronal populations upstream from GnRH neurons, GABA and KNDy cells, as probable mediators of impaired negative feedback in PCOS (Figure 1). Within these cells, AR-mediated interference of PR transcription may initiate modifications that increase GnRH/LH release, including altering neurochemical expression for increased excitation of GnRH neurons and reorganizing the synaptic architecture to either support excitatory transmission between cells or disrupt steroid hormone feedback signalling.

A number of unanswered questions require continued investigation by both clinical and basic researchers, including, but not limited to, what are reliable predictors of neuroendocrine dysfunction that can be identified before pubertal maturation in PCOS? What is the developmental timeline and mechanistic sequence of changes in the brain that permanently establish neuroendocrine dysfunction? Can reversing abnormal androgen signalling specifically within the brain following the pubertal establishment of PCOS reverse PCOS symptoms? Despite these remaining questions, recent encouraging approaches slowing GnRH/LH pulses in PCOS patients by antagonizing KNDy signalling support the continued investigation of neuronal mechanisms underlying GnRH hypersecretion to develop treatments that are tailored towards PCOS patients with neuroendocrine dysfunction.

DATA AVAILABILITY STATEMENT

Data sharing is not applicable to this article as no data sets were generated or analysed during the current study.