The Hypothalamus and Pituitary Gland Regulate Reproduction and Are Involved in the Development of Polycystic Ovary Syndrome

The main features of polycystic ovary syndrome (PCOS) are increased androgen levels, irregular cycles, ovulatory dysfunction, polycystic ovaries, insulin resistance, obesity, and chronic low-grade inflammation [1]. Patients with PCOS have a wide variety of symptoms, the main pathogenesis is unclear, and they respond differently to different treatments. This poses a great challenge for the prevention and treatment of PCOS. It is well known that the role of the hypothalamus and pituitary gland in regulating reproductive function in girls is crucial. In this article, we review articles on hypothalamus regulation of female reproductive function and also review related articles on androgen-induced PCOS in the hope of explaining the mechanism of androgen-induced PCOS from the perspective of the hypothalamic-pituitary-ovarian axis.

The hypothalamic preoptic area (POA) is the main distribution area of gonadotropin-releasing hormone (GnRH) neurons, and afferent neurons of GnRH neurons are located in the anterior ventral periventricular nucleus (AVPV), median preoptic nucleus, periventricular nucleus of the hypothalamus (PeN), and arcuate nucleus (ARC) [2]. The AVPV, PeN, and median preoptic nucleus together are referred to as the region of the anterior ventricle of the third ventricle (RP3V) [3]. GnRH neuron cytosol and dendrites at the POA location are in close contact with a large number of RP3V kisspeptin fibers, but only a small number of ARC kisspeptin fibers [4, 5]; distal dendrites of GnRH neurons at the median eminence (ME) are in close contact with kisspeptin fibers from RP3V and ARC, but kisspeptin fibers from ARC are more numerous [4, 5]; kiss1 neurons in the RP3V receive signals from kisspeptin nerve endings in the ARC, other kiss1 neurons within the RP3V, and neurons in the suprachiasmatic nucleus (SCN) [4, 6]; kiss1 neurons of ARC receive signals from GnRH neuronal nerve endings [4], other kiss1 neurons on both sides of the ARC [4, 7] and kiss1 neurons of the RP3V [7]. GABAergic neurons of the ARC also send extensive projections to GnRH neurons [8]. Both ARC and AVPV/PeN kiss1 neurons project to prodynorphin and Vglut2 neurons in the paraventricular zone of the hypothalamus (PVH) and to CART neurons in the dorsomedial zone (DMH), but they are different in that ARC projects excitatory glutamatergic fibers to the PVH and DMH, and AVPV/PeN projects to the PVH and DMH, and AVPV/PeN to the PVH and DMH are inhibitory GABAergic fibers [5] (Fig. 1).

Kiss1 neurons in the ARC co-express KOR, NK3R, kisspeptin, NKB, and PDyn [9, 10] and are, therefore, referred to as KNDy neurons. KNDy neurons also release glutamate and γ-aminobutyric acid (GABA) [11]. Both AMPAr-mediated glutamatergic signaling and NKB promote the increase of calcium ions in KNDy neurons, and PDyn inhibits the increase of calcium ions [12, 13]. Neuropeptide release requires bursts of discharge or tonic stimulation [14]. Glutamate release can be induced by low-frequency (0.5 Hz) action potential stimulation of KNDy neurons [15]; action potential stimulation of at least 10 Hz for 10 s induces the release of NKB from KNDy neurons; NKB induces a burst of action potentials in KNDy neurons that lasts for 1 min [7, 16]. Action potential stimulation of at least 10 Hz for 30 s induced Kisspeptin release and triggered luteinizing hormone (LH) pulses [17, 18]. PDyn inhibits glutamate release from KNDy neurons, indirectly inhibiting NKB release [19, 20], but does not directly inhibit the burst of action potentials induced by NK3R activation [20, 21]. Knockdown of Slc17a6 reduces NKB release [15].

Both NKB and kisspeptin can induce LH surge [22‒26]. Kisspeptin can be independent of NKB and directly promote LH secretion [27]. NKB stimulates LH secretion mainly by stimulating kisspeptin release [28, 29], but there are still pathways that do not depend on kisspeptin [29, 30].

There is a simultaneous increase in intracellular calcium ions (SE) in KNDy neurons, and these SEs are perfectly synchronized with the LH pulses; intact female rats have a constant frequency of SEs from late estrus to pre-estrus, about once every 40 min or 50 min, during estrus the frequency of SEs decreases considerably, about once every 10 h, the kinetics of SEs are unchanged throughout the cycle, with a total duration of 50 s, the same rapid onset (25 s), with a slower decline (30 s) [31, 32]. Depolarizing currents significantly promote SE, while hyperpolarizing currents inhibit SE [33]. Because KNDy neurons are predominantly distributed caudally and centrally in the ARC, and rarely rostrally [34], SE is mainly found in the middle and end of the ARC [33], another study also showed that only KNDy neurons in the central and caudal parts of the ARC had a role in LH pulses and that non-KNDy neurons expressing only kiss1 but not NKB at other sites had no significance for LH pulses [34] (Table 1).

The ARC is a target of estradiol (E2) negative feedback inhibition of LH secretion [2, 35]. KNDy neurons of ARC co-express ESR1 and ESR2, but predominantly ESR1 [9], and E2 downregulates ESR1 and ESR2 expression in KNDy neurons [36]. E2 upregulates Slc17a6 expression in KNDy neurons [11, 15], and increases glutamate release from KNDy neurons [15, 19]. NKB reduces the amount of AMPAr on the surface of neurons and inhibits glutamate signaling [37].

E2 downregulates NKB, kisspeptin, NK3R, and KOR expression in KNDy neurons [9]. However, the ability of E2 to downregulate KOR expression was weak [38]. The KOR of KNDy neurons was dependent on the presence of E2; after ovariectomy (OVX), the percentage of kiss1 neurons expressing KOR was reduced [20]. Hence, activated KNDy neurons in OVX animals release more kisspeptin than intact females [17]. Both ESR1 deletion >80% in KNDy neurons and OVX result in a significant increase in SE frequency, amplitude, and duration [31, 32]. The more severe the ESR1 deletion in KNDy neurons, the greater the frequency and magnitude of SE, and the increased SE duration would only be seen with >90% ESR1 knockout [32]. The greater the number of KNDy neurons, the higher the frequency and amplitude of the LH pulse [34].

NK1/2/3R and their ligands SP, NKA, and NKB are expressed at the ARC, and E2 downregulates SP and NKA expression [28, 39]. A large number of SP fibers project to KNDy neurons [40] and ME [41, 42]. SP can inhibit PDyn [43]. This is because the receptor NK1R of SP forms a dimer with MOR1 and inhibits the re-sensitization of MOR1 to inhibit the action of PDyn [44]. SP also forms dimers with kiss1r [45]. Activation of NMDAr promotes SP release [46]. PDyn inhibits SP release [47]. SP increases intracellular NO by upregulating NOS1 [48‒50].

SP, NKA, and NKB are necessary for LH pulses [51]. KNDy neurons in OVX animals present more NKB-evoked action potential bursts than intact animals [21]. However, too high-frequency action potential bursts instead inhibit intracellular calcium ion concentration and, thus, inhibit the release of neuropeptides [7, 52]. This may be the reason why activation of NK2/3R decreased LH levels in OVX females but increased LH levels in OVX+E2 females [28, 39, 53]. However, activation of NK1R increased LH secretion in both OVX and OVX+E2 female rats [28]. E2 causes a decrease in presynaptic GABA release in KNDy neurons and also decreases the number of postsynaptic GABAA receptors, but has no effect on the inversion potential of GABAA receptors [54].

In normal female mice and rats, a 10-fold increase in estrogen levels from late estrus to pre-estrus [55‒57]; however, the frequency and amplitude of SE were unchanged, and in contrast, progesterone (P) significantly suppressed SE during estrus [31]. The large decrease in LH pulse frequency during estrus was caused by an increase in P [58‒60]. P increases PDyn signaling and GABAA signaling to suppress LH pulses [61‒63]. Interestingly, P levels also rose during the interoestrus period without affecting the SE frequency of KNDy neurons [31, 55]. E2 upregulates PGr expression in kiss1 neurons [64, 65]. In ARC, E2 downregulation of NK3R and kisspeptin involves an ERE non-dependent pathway, E2 downregulation of PDyn and NKB involves an ERE-dependent pathway, and E2 upregulation of PGr, Cacna1g, and Kcnmb1 involves ERE-dependent pathway [66‒68].

AVPV and PeN are targets of estrogen positive feedback-induced LH surge [2, 35, 69]. Kiss1 neurons of AVPV also express both ESR1 and ESR2, but ESR1 is expressed at a higher level [10]. ESR1 is necessary for positive feedback [2, 70]. The kiss1 neurons of AVPV were less sensitive to E2 than the kiss1 neurons of ARC, possibly related to the higher ESR1/ESR2 ratio of the kiss1 neurons of ARC, high-dose E2 upregulated ESR1, ESR2, kiss1, and PGr expression in kiss1 neurons of AVPV [36]. ESR2 can regulate the transcriptional activity of ESR1 [71, 72] and enhance the function of ESR1 [36]. In AVPV, promotion of kiss1 expression by E2 involves ERE-dependent pathways [66] and ERE-independent pathways [73]. Furthermore ESR2 is essential in gonadotropin promotion of kisspeptin expression in granulosa cells [74]. Significant increase in serum E2 levels in normal females during proestrous [55‒57], so the expression of ESR1, ESR2, kisspeptin, and PGr was upregulated.

Almost all neurons in AVPV express both Vglut2 and VGAT, suggesting that these neurons are both glutamatergic and GABAergic, but high levels of E2 during the pre-estrus period result in a decrease in VGAT-containing vesicles and an increase in Vglut2-containing vesicles [75, 76]. Thus, during proestrus, AVPV released more glutamate and aspartate and less GABA to GnRH neurons at the POA; however, the amounts of glutamate, aspartate, and GABA released to the distal dendrites of GnRH neurons at the medial basal hypothalamus (MBH) were not affected [77, 78]. This regulation of AVPV neurons by E2 is necessary for the LH surge during proestrus, and the absence of ESR1 in GABAergic and glutamatergic neurons results in a loss of positive feedback [79].

At AVPV, low-frequency action potentials (2 Hz) caused kiss1 neurons to release GABA and glutamate; high-frequency action potentials (10 Hz for 30 s) caused kiss1 neurons to release kisspeptin [18, 80]. Blockade of NMDAr or AMPAr during proestrus results in diminished LH and FSH secretion [81, 82]. E2 induces the synthesis and secretion of P by ESR1-expressing astrocytes in AVPV, which is essential for the proestrus LH surge [83‒89]. Before the LH surge, P stimulates the release of glutamate and aspartate from AVPV to POA [77].

GnRH is required for maximal kiss1 expression in AVPV/PeN in adult women [90]. TH expression in AVPV/PeN is not required for female reproduction [91] (Table 2).

KNDy nerve endings of the ARC release glutamate to kiss1 neurons at RP3V [7, 15, 92],E2 increases glutamate release from KNDy neurons to kiss1 neurons of AVPV/PeN [15, 19]. In proestrus, low-frequency action potentials (5 Hz for 2 h) promote the release of glutamate from KNDy neurons to the RP3V, which can trigger the LH surge a few hours earlier and do not interfere with the subsequent emergence of a spontaneous LH surge [93]. Stimulation of KNDy neurons with high-frequency action potentials (10 Hz for 1 h) also triggered an LH surge a few hours earlier, which was lower in amplitude and shorter in duration, and the subsequent spontaneous LH surge disappeared [92]. ARC releases glutamate and PDyn to AVPV, and glutamate promotes the LH surge in proestrus, but PDyn inhibits the LH surge in proestrus [94].

However, KNDy neurons are not required for LH surge. Experiments showed that female mice with selective ablation of KNDy neurons had reduced LH levels and persisted in diestrus, and that LH surge could be induced with E2+P or E2 alone, and that the magnitude of the LH surge and the expression of kisspeptin in the AVPV were higher than that in positive-feedback OVX mice [95‒97]. Higher LH surge may be associated with reduced PDyn release to AVPV after ablation of KNDy neurons [96].

The SCN is a mammalian circadian pacemaker, and SCN regulates circadian rhythms mainly through vasoactive intestinal polypeptide secreted from its core and arginine vasopressin (AVP) secreted from its shell [98]. Circadian rhythms of kiss1 neurons at AVPV are co-regulated by E2 and SCN [99]. In the presence of E2, AVP secreted by the SCN increases the electrical activity and intracellular calcium ion concentration of neurons in the RP3V [98, 100]. There is no circadian rhythm in LH secretion in OVX female mice, but OVX+E2 females have much more LH secretion in the afternoon than in the morning, as well as more GnRH neuronal activity [101]. Circadian rhythm of AVP release leads to a circadian rhythm of LH secretion [102, 103]. The action of GABA on kiss1 neurons at the AVPV is also regulated by both E2 and circadian rhythms [54]. In the presence of E2, the SCN directly activates the electrical activity of GnRH neurons via vasoactive intestinal polypeptide [98].

GnRH neurons at the POA receive projections mainly from RP3V, and distal dendrites of GnRH neurons at the ME receive projections mainly from the ARC [4, 5]. The synaptic density of distal dendrites of GnRH neurons located at the ME is 2.5 times higher than that of proximal dendrites located at the POA; inhibition of excitability of GnRH nerve endings at the ME eliminated the LH surge and suppressed the LH pulse, and inhibition of excitability of GnRH neurons at the POA only eliminated the LH surge but did not affect the LH pulse [104]. NKB can directly induce GnRH release from GnRH nerve endings in the ME independently of kisspeptin, but NKB cannot induce GnRH release from GnRH neurons at the POA [105]. E2 promotes GnRH release at the ME [106].

Kisspeptin activates GnRH neurons [107, 108]. Kiss1r regulates GnRH secretion mainly by coupling Gαq/11 [109]. Kisspeptin increases intracellular calcium ions in GnRH neurons in two stages; the first phase of the calcium increase is characterized by an early onset, a short duration, a lack of dependence on extracellular calcium ions, and easily desensitized (a re-injection of kisspeptin after 10 min does not elicit a first-phase response), and the second phase of the calcium increase is characterized by a longer duration, not easily desensitized (dependent on internalization and recirculation of GPR54), dependence on extracellular calcium ions, and dependence on persistence of extracellular kisspeptin (the phase II calcium increase disappeared immediately after kisspeptin removal) [110]. In the first phase, kisspeptin triggers endoplasmic reticulum calcium release directly through activation of the phospholipase C-IP3 pathway [111, 112], so it can directly promote GnRH release independently of action potentials [112‒114]. In the second phase, kisspeptin increases intracellular calcium ions through activation of TPRC channels and inhibition of Kir channels, but the role of TPRC channel activation is greater [115, 116]. E2 upregulates TPRC channel expression in GnRH neurons [117]. Kisspeptin signaling is essential for LH surge, puberty onset, and reproductive development [118, 119]. Blockading kisspeptin can prevent the rise in LH after OVX [120, 121].

GABAB signaling is important for reproductive function [122]. GABAB receptors in the postsynaptic membrane depolarize GnRH neurons by coupling Kir channels and GABAB receptors in the presynaptic membrane depolarize GnRH neurons by coupling calcium channels [123, 124]. GABAB receptor-coupled Kir channels are not regulated by E2 but are inhibited by kisspeptin [123], which inhibits Kir channels [115, 116].

GABA excites 63% of GnRH neurons and inhibits 17% of GnRH neurons via GABAA receptors [18, 80]. NKCC1 ion channels in neurons increase the intracellular concentration of chloride ions, and when exposed to GABA, the cell membrane potential is depolarized as chloride ions leave the cell; KCC2 ion channels decrease the intracellular concentration of chloride ions, and when GABAA receptors are turned on, chloride ions enter the cell, and the neuron is hyperpolarized [125]. GnRH neurons in POA and MBH appear to express only NKCC1 [126]. So, the excitatory effect of GABAA signaling on GnRH neurons was stronger [107]. Sustained glutamate signaling can cause GnRH neurons to generate action potentials with a frequency of 4–9 Hz [127]. Activation of both AMPAr and NMDAr increased intracellular calcium ions, and for AMPAr this appeared to involve direct calcium entry via AMPAr [128]. GABAA, NMDA, AMPA and kisspeptin are all able to increase intracellular calcium ions by activating voltage-gated calcium channels [128, 129]. Voltage-gated calcium channels in GnRH neurons consist mainly of slowly inactivating Cav3.3, and Cav3.3 expression can be upregulated by E2 [130]. Voltage-gated calcium channels in kiss1 neurons consist mainly of fast-inactivating Cav3.1 [94]. E2 upregulates Cav3.1 expression in kiss1 neurons [15, 94, 131]. E2 prevents inactivation of voltage-gated calcium channels in kiss1 neurons [132]. Activation of voltage-gated calcium channels inhibits endocytosis at the presynaptic membrane [133].

GnRH neuronal cell bodies or distal dendrites need to be activated at a frequency of at least 10 Hz for 2 min to elicit an LH pulse [134]. For GnRH neurons in POA, it takes action potentials above a certain frequency to be able to bring the intracellular calcium ion concentration up to the threshold for GnRH release, but too high a frequency of action potentials accelerates intracellular calcium ion depletion, which inhibits GnRH release, depolarization is more likely to elicit action potentials in smaller diameter protrusions, and calcium ion current densities in response to action potentials and depolarization are greater in OVX animals [114]. Calcium-activated potassium channel proteins are classified as large conductance (BK; KCa1.1), small conductance (SK; KCa2.1–2.3, KCa3.1), and molecularly unspecified calcium-activated potassium channel proteins [135, 136]. BK has the fastest onset but short duration, thus promoting action potential repolarization; SK has a faster onset and longer duration, thus controlling the frequency of spike discharges, and inhibition of SK leads to an increase in the number of action potential spikes and a prolongation of the inter-burst interval (IBI); molecularly unspecified calcium-activated potassium channels have a slower onset and longer duration, and inhibition of molecularly unspecified calcium-activated potassium channels leads to a shortening of the IBI, whereas an increase in the IBI <15 s leads to a decrease in the increase of intracellular calcium ions [52]. SK also couples NMDAr [136]. E2 downregulates SK3 expression in GnRH neurons [117].

Obesity is a common feature of PCOS [137, 138]. Obesity is also a characteristic of OVX females [139, 140]. Proopiomelanocortin (POMC) and corticotropin-releasing hormone inhibit food intake, and pro-neuropeptide Y (NPY) promotes food intake, and E2 upregulates the expression of POMC and corticotropin-releasing hormone mRNAs, and downregulates the expression of NPY mRNAs [141]. Low-frequency action potentials stimulated glutamate released from KNDy neurons excited POMC neurons and NPY/agouti-related protein (AgRP) neurons, and high-frequency action potentials stimulated glutamate and kisspeptin released from KNDy neurons excited POMC neurons but inhibited NPY/AgRP neurons [15]. Activation of NPY/AgRP neurons rapidly increases food intake, activation of POMC neurons rapidly inhibits food intake [142‒144]. Ablation of KNDy neurons or TAC1 deletion resulted in weight loss and reduced insulin resistance [145, 146]. E2 promotes glutamate release from KNDy neurons [15]. E2 inhibits food intake and weight gain [147, 148]. Women with kiss1r knockouts are obese and insulin resistant [149]. Leptin is secreted by adipocytes, and obesity leads to an increase in leptin, which attenuates inhibition of POMC neurons primarily through leptin receptors in presynaptic GABAergic neurons [150].

E2 protects female POMC neurons from insulin resistance by enhancing excitability of POMC neurons and coupling of insulin receptors to TRPC5 channels [151]. E2 also protects NPY/AgRP neurons from insulin resistance [152]. In the periphery, E2 can inhibit pancreatic β-cell apoptosis and promote insulin secretion via ESR1/2 [153, 154], glucagon and insulin stimulate and inhibit hepatic secretion of kisspeptin, respectively, and kisspeptin inhibits pancreatic β-cell insulin secretion via kiss1r [155, 156].

NPY/AgRP neurons release GABA to kiss1 neurons in RP3V and ARC, inhibiting kiss1 neurons, thereby leading to prolonged estrous cycle and impaired fertility [157]. α-Melanocyte stimulating hormone (α-MSH) released by POMC neurons upregulates kiss1 expression at the ARC thereby stimulating LH secretion in adolescent rats [158]. POMC neurons also release OFQ to inhibit GnRH neurons [159].

Mkrn3 expression in the hypothalamic ARC is high early in life and declines before the onset of puberty; Mkrn3 is the true controller of the onset of puberty [160]. Mkrn3 inhibits the promoter activity of kiss1 and TAC3 possibly through ubiquitination [161]. With puberty, the expression of TAC1(SP/NKA) and TAC3(NKB) gradually increased [162, 163]. SP, NKA, and NKB are essential for sexual maturation and the onset of puberty, and specific activation of NK1/2/3R leads to an earlier onset of puberty [39, 162‒164]. Leptin does not directly regulate kiss1 neurons [165, 166]. Lack of leptin signaling in GABAergic neurons in the hypothalamus leads to delayed onset of puberty and also downregulates kisspeptin expression in ARC and AVPV [167]. Leptin receptors neurons and PACAP neurons in the ventral premammillary nucleus are associated with sexual maturation and the onset of puberty [166, 168, 169].

Two pathways exist in the medial posterior dorsal amygdala that can promote LH release, one through the release of kisspeptin and the other through the kisspeptin-independent NKB signaling pathway [30]. The amygdala regulates LH secretion through kisspeptin, and injection of kisspeptin into the amygdala resulted in an increase in LH secretion, whereas blockade of endogenous kisspeptin signaling in the amygdala by injection of kisspeptin antagonists into the amygdala decreased LH secretion and the LH pulse rate [170]. Ucn3 neurons of the medial posterior dorsal amygdala release GABA and glutamate into the hypothalamic paraventricular nucleus, which ultimately inhibits the frequency of the GnRH pulser [171].

Purinergic/noradrenergic neurons from areas A1 and A2 of the hindbrain project to kiss1 neurons but not GnRH neurons, and E2 can stimulate their release of ATP that acts on KNDy neurons to promote kiss1 expression and trigger LH release [172]. Norepinephrine released from these noradrenergic neurons promotes kiss1 expression in ARC but inhibits kiss1 expression in AVPV [173].

Kiss1r-positive cells in the pars tuberalis of the adenopituitary gland as well [113]. Kisspeptin can act directly on the pituitary gland to stimulate LH and FSH secretion [174]. Kisspeptin enhances the pituitary response to GnRH [17, 29, 175]. Non-E2 components secreted by the ovary act to enhance pituitary sensitivity [176]. Musashi family mRNA-binding proteins inhibit translation of GnRHr and FSHb mRNA in pituitary cells [177].

Prenatal androgen exposure (PNA), postnatal androgen exposure (PAE), and letrozole (LET) are commonly used to induce PCOS in animals. However, these PCOS models are not characterized in exactly the same way.

Cellular studies have shown that dihydrotestosterone upregulates the expression of kisspeptin and NKB and downregulates the expression of PDyn in KNDy neurons [178]. But animal experiments are different.

For PNA females, NKB and kisspeptin expression is upregulated and PDyn expression is downregulated at the ARC in adulthood [179‒181], and increased number of KNDy neurons [182]. These promote LH pulses. There was a decrease in the number of KNDy neurons, a decrease in synapses between KNDy neurons, a decrease in projections from KNDy neurons to the MBH, an upregulation of KOR expression in KNDy neurons, and an increase in the proportion of glutamatergic neurons expressing KOR. These inhibit LH pulses. However, the promoting effect was stronger than the inhibiting effect; thus, PNA females have elevated LH levels and increased LH pulse frequency in adulthood [8, 179, 180, 182‒185]. Paradoxically, PNA does not affect the firing activity of KNDy neurons in female mice as adults [186]. Prolonged activation of KOR partially normalizes the peak LH pulse, LH levels, androgen levels and reproductive cycle in PNA mice [187]. NK3R agonists, kisspeptin and leptin all stimulate LH surge in adult PNA female mice [179, 186, 188].

PAE is the exposure of females to high levels of dihydrotestosterone shortly after birth or during puberty. For PAE females, kisspeptin and NKB expression in the ARC is downregulated, and the number of KNDy neurons is reduced in adulthood [182, 189‒192]. Therefore, LH levels and LH pulse frequency are reduced in adulthood [182, 188‒191]. However, some studies have suggested that LH levels and LH pulse frequency are normal in adulthood [193, 194]. Blocking NK3R cannot improve abnormal reproductive function [195]. Kisspeptin cannot cause LH surge [188, 189].

For LET females, the expression of kisspeptin, kiss1r, NKB, and PDyn was upregulated at ARC, the number of KNDy neurons increased, and the proportion of KNDy neurons activated increased [196‒199]. Therefore, LH levels were elevated, LH pulse frequency increased, and LH pulse amplitude increased [196‒200]. Inhibition of KNDy neurons resulted in decreased LH pulse frequency and LH levels in LET animals [201].

10% of kiss1 neurons, 33% of NPY neurons, 15% of tyrosine hydroxylase neurons, and 10% of neuronal nitric oxide synthase neurons in the ARC are GABAergic, and PNA does not alter these proportions [202]. Prolonged activation of GABAergic neurons in the ARC resulted in a reduction in the number of entries into proestrus and the number of luteum, but had no effect on the number of follicles prior to ovulation or on the restoration of follicular morphology [203]. GABAergic inputs to GnRH neurons from ARC in PNA animals are increased before puberty [185, 204‒206]. Increased GABAergic input to POA in female rats decreased GABAergic input to POA and increased GABAergic input to MBH in sheep [8, 126]. The use of androgen receptor (AR) antagonists normalized GABAergic input to GnRH neurons in PNA animals but had no effect on normal females [185]. However, these GABAergic synapses may activate GnRH neurons by releasing neuropeptides rather than GABA. One piece of evidence is that low-frequency stimulation (action potentials at 2 or 10 Hz) of nerve endings of GABAergic neurons has no effect on LH secretion, but that high-frequency stimulation (20 Hz) results in a long-lasting increase in LH secretion [203], and that classical neurotransmitters require only low-frequency stimulation for their release whereas neuropeptides require high-frequency stimulation for their release [14]. Another piece of evidence is that the frequency of GnRH neuron firing in pubertal PNA female mice does not differ from that of normal animals [207], but by this time, GABAergic inputs have increased. A third piece of evidence is that AR on GABAergic neurons is not necessary for PNA to induce PCOS [208]. Unlike PNA females, GABAergic inputs to GnRH neurons from ARC in PAE females were unchanged [193].

For PNA females, there was an increase in ESR1-positive cells in the PeN, no change in the number of ESR1-positive cells in the AVPV [8], normal or downregulated kisspeptin expression in the AVPV [180, 182], normal number of kiss1 neurons, upregulated AR expression in the kiss1 neurons, and upregulated AR expression in the neurons at the SCN [8, 209]. Kiss1 neurons were unchanged in their response to AVP, but SCN released less AVP to AVPV [209]. So PNA females have prolonged estrus, lack of proestrus, altered ovarian weights, ovarian polycysticity, and lack of luteum [180, 182, 185, 205, 210]. However, E2 can still stimulate LH surge [211].

For PAE females, kiss1 neuron numbers are decreased and kisspeptin expression is downregulated in AVPV and PeN [189, 191]; high levels of E2 can upregulate kisspeptin expression in AVPV but cannot activate kiss1 neurons and cannot induce LH surge [189]. So PAE females are unable to enter pre-estrus, have reduced ovarian weight, polycystic ovaries, and lack luteum [182, 188, 189, 193].

For LET females, the number of kiss1 neurons at the AVPV of LET animals was similar to that of normal animals during the interoestrus period, but the number of kiss1 neurons at the AVPV of LET animals was significantly lower than that of normal animals during the preoestrus period [198, 199]. The number of kiss1 neurons expressing ESR1 was decreased [197]. So LET females have prolonged estrus, lack of pre-estrus, enlarged ovaries, polycystic ovaries and lack of corpus luteum [197‒200, 212].

PNA females have unchanged firing of GnRH neurons during puberty and increased firing of GnRH neurons during adulthood [207]. The activation and inactivation potentials of transient potassium currents in GnRH neurons are depolarizing [213]. Unlike normal animals, PNA animals have reduced GnRH neuron firing rates after OVX and increased GnRH neuron firing rates after dihydrotestosterone injection [207].

For PNA females, NKB and leptin expression were upregulated in the hypothalamus prior to the onset of puberty [179], but NKB failed to increase the firing of KNDy neurons in pubertal [186], and the onset of puberty was normal in PNA females [184].

PNA females had normal or reduced serum P [8, 182], a significant reduction in the number of PGr-positive cells in AVPV, PeN, and ARC, and a reduction in the number of PGr-positive GABAergic neurons [8, 181], resulting in a loss of the negative feedback effect of P on LH [8, 183, 210]. For LET females, PGr expression was unchanged at POA and downregulated at MBH [200].

Dihydrotestosterone upregulates pituitary voltage-dependent calcium channel inhibitor expression leading to decreased sensitivity to GnRH [214]. For PNA females, LH pulse amplitude may be decreased, normal, or increased [8, 179, 183], and pituitary response to exogenous GnRH may be increased or decreased [183, 203].

For PAE females, pituitary response to GnRH was attenuated, and LHβ and FSHβ expression in the pituitary was downregulated [188‒190]. For LET females, LHb and GnRHr expression was upregulated, Fshb expression was downregulated, and kiss1r and PGr expression was unchanged in the pituitary [200].

PNA females have normal androgen levels before and after puberty [205, 207], and increased androgen levels in adulthood [8, 180, 183, 185, 205], except for one case in which the androgen levels were normal in adulthood [182]. AR expression is increased in KNDy neurons [181]. The use of AR antagonists normalized the estrous cycle, preovulatory follicle number, corpus luteum number, and follicular morphology but had no effect on normal females [185, 205].

For PAE females, the number of AR-positive cells in the hypothalamus is increased [193]. AR of neurons is associated with abnormal reproductive function, obesity, and dyslipidemia, and AR of granule cells is associated with reduced thickness of the granule cell layer [215]. However, it has also been suggested that AR of neurons is not associated with abnormal reproductive function [216]. AR of hepatocytes was not associated with abnormal reproductive function but was associated with insulin resistance [217, 218]. AR of neurons and adipocytes was not associated with insulin resistance but was associated with obesity [219]. AR of AgRP neurons was not associated with reproductive and metabolic abnormalities [220].

For LET females, androgens were significantly increased in both the pre-estrus and inter-estrus periods [196‒200, 212]. Inhibition of KNDy neurons reduces androgen levels in LET animals [201]. The number of kiss1 neurons expressing AR was increased [197]. Specific knockdown of AR in kiss1 neurons did not completely prevent LET from inducing PCOS [221].

The alterations of hyperandrogenism on the hypothalamic-pituitary-ovarian axis are long-term. Acute injection of androgens does not alter LH pulses and LH response to GnRH in PCOS patients [222], and prolonged AR blockade has no effect on LH pulse frequency and amplitude, LH levels, or LH sensitivity to exogenous GnRH in PCOS patients but restores the inhibitory effects of E2 and P on LH pulses [223]. Hyperandrogenism inhibits nitric oxide synthase activity [224]. Hyperandrogenism inhibits granulosa cell proliferation and causes follicular abnormalities by inducing H3K27 deacetylation [225, 226]. Hyperandrogenism promotes granulosa cell apoptosis [227]. Hyperandrogenism impairs ovarian function [228‒230].

For PNA females, co-expression of IRβ in KNDy neurons and AgRP neurons is reduced in the ARC in adulthood, and the use of anti-androgen drugs restores IRβ expression in AgRP neurons, but not in KNDy neurons [231]. LET animals have increased insulin levels and are insulin resistant [200, 212]. Improved insulin resistance downregulates kisspeptin expression in the hypothalamus and improves reproduction [232].

For PNA females, body weight always seems to be normal in adulthood [182‒184, 188]. NPY, NPYr, AgRP, and POMC expression is normal in adulthood [179]. There was no increase in GABAergic signaling from NPY neurons to GnRH neurons [233].

PAE females are obese in adulthood [182, 189, 190, 193]. Obesity is associated with upregulation of hypothalamic NPY and AgRP expression and decreased leptin concentration in cerebrospinal fluid [234]. Blocking NK3R reduces body weight and fat content and makes fat cells smaller [195]. Kisspeptin expression is upregulated in adipose tissue [191]. In adulthood, LET females are obese and have increased leptin [196‒200, 212] (Table 3).

ARC is associated with LH pulses, and glutamate and NKB can excite KNDy neurons and promote kisspeptin secretion. E2 and P regulate glutamate, NKB and kisspeptin secretion from KNDy neurons by regulating the expression of genes such as kisspeptin, NKB, NK3R, PDyn in ARC. RP3V is associated with LH surges. Signals from regions such as ARC and SCN can excite RP3V neurons, while E2 and P enhance the excitability of RP3V neurons by regulating gene expression. These two processes cause RP3V to release more glutamate, aspartate, and kisspeptin to GnRH neurons. Glutamate, aspartate, and kisspeptin excite GnRH neurons and stimulate the secretion of GnRH, which stimulates the pituitary to release LH and FSH.

PNA, PAE and LET can all cause PCOS-like symptoms in animals. All PCOS animals showed disorganized estrous cycle, lack of proestrus, anovulation, polycystic follicles and lack of corpus luteum. In preoestrus, the number of kiss1 neurons at RP3V was significantly lower in PCOS animals than in normal animals, and exogenous E2 increased the number of kiss1 neurons and LH secretion. This suggests that the function of the RP3V is not completely lost in PCOS animals. LH pulses and LH levels were altered in all PCOS animals. The expression of NKB and kisspeptin of ARC was upregulated in PNA and LET animals, the frequency of LH pulses was increased, and LH levels were elevated. The expression of NKB and kisspeptin of ARC was downregulated in PAE animals, the frequency of LH pulses was decreased, and LH levels were reduced. PCOS animals are deficient in P negative feedback, which is associated with downregulation of hypothalamic PGr expression. AR in neuronal cells was associated with abnormal reproductive function, obesity, and dyslipidemia, AR in hepatocytes was associated with insulin resistance, and AR in adipocytes was associated with obesity. PAE and LET animals were obese in adulthood, but PNA animals had normal body weight in adulthood.

However, there are still unresolved questions. GnRH neurons, SCN, and pituitary gland are critical for female reproduction, but the number of studies on this is small. PCOS models are simpler, but the patient profile is often more complex, with a possible coexistence of prenatal androgen exposure, pubertal hyperandrogenism, and adult hyperandrogenism, for which there is a dearth of studies. E2 is critical for regulating female reproductive function and is essential for maintaining a normal estrous cycle, but there are no studies on how E2 levels change throughout the estrous cycle in animals with PCOS. There are more studies on the pathogenesis of PCOS and fewer studies on the treatment of PCOS. Low concentrations of kisspeptin increased ovarian granulosa cell viability, inhibited granulosa cell apoptosis, promoted FSHr expression, and facilitated the release of P, E2, and IGF-I, but high concentrations of kisspeptin had the exact opposite effect on granulosa cells [235‒237]. Many studies have shown altered serum kisspeptin levels in PCOS patients [137, 138, 238‒246]. But these have not been validated in animal models of PCOS.

The authors declare no competing interests.

This study was supported by the Sichuan Provincial Administration of Traditional Chinese Medicine Project (2023MS270), Sichuan Medical Research Topics Program (S22014), and Opening fund of NHC Key Laboratory of Chronobiology (SichuanUniversity) (NHCC-2023-02).

Bin-Yang Long: conceptualization, data curation, formal analysis, investigation, software, and writing – original draft. Xipeng Liao: investigation, methodology, software, and writing – original draft. Xin Liang: funding acquisition, project administration, resources, and supervision.