Targeting KNDy neurons to control GnRH pulses

Abstract

Gonadotropin-releasing hormone (GnRH) is the final output of the central nervous system that drives fertility. A characteristic of GnRH secretion is its pulsatility, which is driven by a pulse generator. Each GnRH pulse triggers a luteinizing hormone (LH) pulse. However, the puzzle has been to reconcile the synchronicity of GnRH neurons with the scattered hypothalamic distribution of their cell bodies. A leap toward understanding GnRH pulses was the discovery of kisspeptin neurons near the distal processes of GnRH neurons, which secrete kisspeptins, potent excitatory neuropeptides on GnRH neurons, and equipped with dual, but opposite, self-modulatory neuropeptides, neurokinin B and dynorphin. Over the last decade, this cell-to-cell communication has been dissected in animal models. Today the 50-year quest for the basic mechanism of GnRH pulse generation may be over, but questions about its physiological tuning remain. Here is an overview of recent basic research that frames translational research.

Introduction

Fertility relies on a cascade of events between the hypothalamus, pituitary, and gonads. A small population of hypothalamic neurons releases gonadotropin-releasing hormone (GnRH) into the pituitary portal blood. Subsequently, GnRH stimulates gonadotrophs in the anterior pituitary, which synthesize and secrete two gonadotropins, luteinizing hormone (LH) and follicle stimulating hormone. Gonadotropins in turn initiate gametogenesis and steroidogenesis in gonads [1]. Outside the preovulatory period in females, gonadal steroids exert negative feedback on the hypothalamus and pituitary gland [2].

A key feature of the hypothalamus-pituitary communication is its reliance on episodic GnRH release because gonadotrophs cannot maintain their function under continuous exposure to GnRH (see review in this issue). Periodic coupling between the hypothalamus and the pituitary gland has been revealed during simultaneous central and peripheral blood sampling, showing a perfect correlation between GnRH and LH pulses (review [3]). This initial observation led to the concept of the GnRH pulse generator (historical overview in [4]). However, although early electrophysiological experiments could record the direct manifestation of the GnRH pulse generator as periodic increases in multiunit electrical activity in the arcuate nucleus (ARC) that always preceded GnRH/LH pulses [5,6], its identity was elusive for 35 years.

Although it was apparent that GnRH pulses reflected synchronization of secretory events across the GnRH neuronal population, it was uncertain whether increased multiunit electrical activity reflected synchronized activation of GnRH neurons themselves or other neurons. On the one hand, in vitro experiments using immortalized GnRH cells or GnRH cells derived from olfactory placodes suggested a GnRH pulse generator intrinsic to GnRH neurons, since GnRH pulses were detected in Petri dishes [7,8]. However, communication between GnRH neurons has been difficult to reconcile with their scattered distribution in the brain [9]. On the other hand, in vivo experiments suggested a pulse generator that is extrinsic to GnRH neurons, since GnRH/LH secretion can occur without increased multiunit electrical activity in the ARC [10–14].

Synchronization of GnRH neuronal activity is difficult to reconcile with the anatomy of the GnRH system and its scarcity. For example, the mouse brain contains ~800 GnRH neurons and they are scattered in a continuum extending from the preoptic area to the mediobasal hypothalamus. GnRH cell body synchronization is supported by intertwined processes with common inputs [9]. Nevertheless, most GnRH neurons – regardless of the position of their cell bodies – pass through the ARC and project to the external zone of median eminence where their terminals are regulated [15]. As such, complete deafferentation of the hypothalamus, i.e. disrupted connection between the preotic area and the median eminence, does not prevent LH pulses, unless the anterior ARC is also lesioned [16] and isolated hypothalamic fragments devoid of GnRH cell bodies still generate GnRH pulses [17].

This review will focus on selected steps that led to, and perhaps completed, the search for a GnRH pulse generator (see review for more details [18]), opening a new avenue for clinical applications today.

Kisspeptin neurons in the spotlight

In 2003, two teams simultaneously discovered the role of the kisspeptin receptor (KISS1R) and its ligands, the kisspeptin isoforms (products of KISS1 gene), in human reproductive endocrinology [19,20]. Two years later, kisspeptin was discovered as a potent GnRH secretagogue in mice and sheep [21], via activation of KISS1R expressed in GnRH neurons [22]. Two subpopulations of kisspeptin-secreting neurons were then identified: one is located in the rostral periventricular region of the third ventricle (RP3V), which is much larger in females than in males [23], and the other is located in the ARC, which is of similar size in males and females [23,24]. Both subpopulations express the estrogen receptor α (gene Esr1), which is critical for gonadal steroid feedback. Kiss1 mRNA in the RP3V subpopulation of neurons is upregulated by estradiol [23,24], and is activated at the time of the preovulatory GnRH/LH surge [25]. In contrast, Kiss1 mRNA in ARC neurons is downregulated by estradiol [23,24]. ARC-kisspeptin neurons coexpress neurokinin B and dynorphin A [26], leading to their nickname KNDy. Therefore, RP3V-kisspeptin and KNDy neurons became prime candidates for the integration of gonadal positive and negative gonadal feedbacks, respectively.

KNDy neurons

Role in the operation of the GnRH pulse generator

Experimentally, negative feedback is evident after gonadectomy as a rise in circulating LH levels and its reversal with steroid supplementation. The downregulation of Kiss1 mRNA by estradiol in the ARC suggested KNDy neurons could transmit negative feedback of steroids to the GnRH pulse generator [23,24]. Finally, when both neurokinin B and dynorphin were proven to be effective modulators of the GnRH pulse generator [27], it was hypothesized that KNDy neurons may form the GnRH pulse generator itself ([28]; historical overview in [29]). In vitro electrophysiology confirmed KNDy neurons are indeed equipped with auto-excitatory (neurokinin B) and auto-inhibitory (dynorphin) neuropeptides [30], which could synchronize KNDy neurons and trigger a periodic kisspeptin output and subsequent GnRH pulses. In vivo calcium imaging has revealed that synchronized calcium episodes initiate by a subset of KNDy neurons then propagate among others and always precede LH pulses [31,32] and that silencing KNDy neurons suppresses LH pulses [32]. Although kisspeptin is not required for synchronization of KNDy neurons themselves [33], it is kisspeptinergic output that triggers GnRH/LH release. Rescue of kisspeptin expression in Kiss1-null rodents restores LH pulses [34] and kisspeptin, but not dynorphin nor neurokinin B, triggers calcium signaling in GnRH nerve terminals [33]. Notably, the coupling efficiency of kisspeptin and GnRH neurons and functional redundancy are such that ~20% of KNDy neurons are sufficient to maintain reproductive function [34]. GnRH/LH pulse frequency can be manipulated via kisspeptin/kisspeptin neuronal communication [35,36]. Alteration of neurokinin B signaling as in tachykinins-deficient mice reduces LH pulse frequency [35]. Conversely, alteration of dynorphin signaling by reducing the numbers of dynorphin receptor-expressing cells increases the frequency of LH pulses [36].

Together, these observations support KNDy neurons as GnRH pulse generator. However, whether KNDy neurons drive or contribute to multiunit electrical activity is unknown because such electrical activity has not been assessed after selective ablation of KNDy neurons in ruminants [37] and cannot be assessed in mouse models that offer the genetic tools.

However, the KNDy model for GnRH/LH pulse generation relies on the following sequence: 1) activation and intra-ARC secretion of neurokinin B from leading KNDy neurons; 2) recruitment of other KNDy neurons via neurokinin B receptor; 3) secretion of kisspeptin around the GnRH dendrons and subsequent activation of GnRH neurons via the kisspeptin receptor; 4) intra-ARC secretion of dynorphin from KNDy neurons; and 5) subsequent deactivation of KNDy neurons via dynorphin receptor (Figure 1). Notably, steroids regulate the genes encoding kisspeptin, neurokinin B and dynorphin, but also the response of KNDy neurons to neurokinin B and dynorphin. In vitro, steroids suppress the response to neurokinin B receptor activation and enhance the response to dynorphin receptor activation [38].

Figure 1:

(A) Diagram representing the connectivity between GnRH neurons and kisspeptin neurons observed in rodents. Kisspeptin neurons from the rostral periventricular region of the third ventricle (RP3V-kisspeptin, orange) contact GnRH cell bodies. In contrast, kisspeptin neurons from the arcuate nucleus (ARC-kisspeptin, magenta) contact GnRH neuron distal processes. While the output of ARC-kisspeptin neurons is kisspeptin, they also coexpress neurokinin B and dynorphin, which act as auto-excitatory and auto-inhibitory signals, respectively, within the network of ARC-kisspeptin neurons.

(B) Schematic time course of the kisspeptin secretion from ARC-kisspeptin neurons (magenta) and RP3V-kisspeptin neurons (orange) that drives two modes of secretion of GnRH neurons (blue): pulses in both males and females, and surge leading to ovulation in females.

(C) Schematic showing how tilting the balance between neurokinin B and dynorphin can affect GnRH pulse frequency.

Many questions remain about this model: How do leading KNDy neurons re/activate? How is it regulated that secretion of neurokinin B and dynorphin from KNDy neurons is sequential? How does estrogen receptor α-mediated signaling regulate this chain of events when it only modestly affects the excitability of KNDy neurons [39,40]? How infancy silences and puberty awakens the KNDy neuron signaling? How does metabolism or stress affect KNDy neuron signaling and consequently fertility?

Role in the negative feedback

Although genes encoding KNDy neuropeptides, neurokinin B, and dynorphin receptors are regulated by gonadal steroid hormones, the role KNDy neurons in negative feedback remains somewhat controversial. On the one hand, the increased calcium activity in KNDy neurons after gonadectomy [41] and the loss of reproductive cyclicity – a proxy for negative feedback and hypothalamic fitness – after deletion of the estrogen receptor α from adult-KNDy neurons [42] support the major role of these neurons in the negative feedback. On the other hand, several studies suggest additional players to negative feedback. For example, while estrogen receptor α undoubtedly transmits negative feedback - not the estrogen receptor β [43], its deletion from neonatal KNDy neurons does not abolish it, despite increased expression of Kiss1 in the ARC [44,45]. Its deletion from adult KNDy neurons does not increase LH pulse frequency or LH levels [42]. However, it blunts the LH response to GnRH, indicating overstimulation of the pituitary gland. Consistent with this, deletion of adult-KNDy neurons reduces LH levels due to cessation of KNDy neuron activity, and the LH response to GnRH is enhanced [46].

However, the data against their role in the negative feedback must be interpreted carefully: 1) deletion of neonatal estrogen receptor α delete can induce developmental adaptations as seen in estrogen receptor β knockouts [43], 2) ablation of adult-KNDy neurons with saponin or estrogen receptor α deletion from adult-KNDy neurons only reduce the number of KNDy neurons or estrogen receptor α in KNDy neurons and 3) if only ~20% of KNDy neurons are sufficient to maintain cyclicity, they may also be sufficient to maintain negative feedback [34,46].

Kisspeptinergic integration by GnRH neurons

GnRH neurons have been described long ago (review: [47]), but their distal dendritic compartment, called the dendron, has recently been identified and exhibits unique characteristics (review: [48]). RP3V-kisspeptin neurons contact GnRH cell bodies and distal processes, while KNDy neurons project to contact GnRH neuron distal processes [49]. Notably these KNDy appositions onto GnRH dendrons are non-synaptic and function through volume transmission [33]. The response of the dendrons to kisspeptin in vitro was indeed a major step forward in the understanding of kisspeptin-regulated GnRH secretion [50,51]. The puzzle with the synchronization of scattered GnRH cell bodies was circumvented by the discovery of the excitability of the GnRH neuron distal processes which converge towards a single hub, the median eminence.

Together, these data led to the concept that two subpopulations of kisspeptin neurons control different dendritic compartments in rodents and two modes of secretion, i.e. pulses and surges [2,52]. Such functional dichotomy is also supported in vivo. Silencing of GnRH cell bodies abolishes the preovulatory LH surge, but not LH pulses, while silencing of GnRH dendron silencing abolishes the preovulatory LH surge and dampened LH pulses [53].

GnRH cell body and dendron signaling activated by kisspeptin has been thoroughly studied [54]. However, it has been difficult to reconcile the repetitive action of kisspeptin on GnRH neurons in vivo with the rarely repeatable response of GnRH neurons to kisspeptin in vitro [55]. An overlap between kisspeptin- and nitric oxide-induced signaling may explain the return to baseline activity of GnRH neurons after kisspeptin [56]. Briefly, excitation of GnRH neurons by kisspeptin relies on activation of the Gq protein-coupled kisspeptin receptor and initiates the degradation of membrane phosphatidylinositol 4,5-bisphosphate (PIP₂) into inositol 1,4,5-trisphosphate and diacylglycerol. The depletion of PIP₂ activates transient receptor potential canonical channels and initiates calcium signaling. Replenishment of PIP₂ is a time-limiting factor to transient receptor potential canonical channel deactivation and repeated excitation. Nitric oxide, an important signaling player in the median eminence [57], may be the key of replenishing PIP₂, by stimulating phosphatidylinositol 4-kinase, an important enzyme in PIP₂ synthesis. Furthermore, the highly diffusible property of nitric oxide may offer a synchronizing mechanism for GnRH dendrons. However, this nitrergic regulation of kisspeptin signaling in the GnRH cell body must be confirmed at the GnRH dendron.

From rodents to humans

The literature cited above mainly describes rodent models to delineate the control of pulsatile GnRH release by KNDy neurons. Some data exist in ruminants, less in non-human primates. Nevertheless, they largely support the rodent model, with one major difference being the contribution of KNDy neurons to positive feedback in sheep and non-human primates (more details in reviews [58,59]). This section focuses on available human data.

Anatomy

Two kisspeptinergic neuronal populations have also been observed in humans: 1) one in the preoptic/rostral hypothalamus area (rodent RP3V equivalent, but much less defined), 2) one in the infundibular nucleus (rodent ARC equivalent) [60].

In humans, and in contrast to rodents, the population of kisspeptin neurons in the preoptic/rostral hypothalamus is minimally sexually dimorphic, devoid of the neuropeptides found in rodent kisspeptin neurons, virtually absent in aged women, indicating that it is positively regulated by steroids [60]. However, the function of preoptic/rostral kisspeptin neurons remains unknown in primates because non-human primates with a rostrally disconnected mediobasal hypothalamus still respond to positive feedback [61]. The mediobasal hypothalamus may even be bypassed as both GnRH-deficient women and non-human primates whose mediobasal hypothalamus is lesioned still exhibit LH surges and ovulation under constant pulsatile administration of GnRH (review in [59]), indicating that the pituitary contributes greatly to positive feedback. This is in contrast to rodents and ruminants in which a GnRH surge is a prerequisite for ovulation [62].

The population of kisspeptin neurons in the infundibular nucleus coexpresses neurokinin B, substance P, cocaine- and amphetamine-regulated transcripts. Dynorphin is rarely detected, but prodynorphin, especially in younger individuals, indicating a technical reason rather than a species difference. Unlike mice, but similarly to sheep, the KNDy population is sexually dimorphic with a higher number of cells in women [63]. This population appears to be negatively regulated by steroids as postmenopausal women show cellular hypertrophy with increased expression in kisspeptin, neurokinin B and substance P [64]. Unlike rodent KNDy neurons that contact exclusively GnRH dendrons, infundibular kisspeptin neurons contact GnRH cell bodies and this connectivity increases with aging [65]. In fact, in humans and non-human primates, GnRH cell bodies are mainly located in the mediobasal hypothalamus [66] which could eliminate the need for subcellular regulation of GnRH neuron regulation, compared to rodents.

Function

Kisspeptin isoforms are potent LH secretagogues in humans [67,68] and biallelic mutation in the gene encoding KISS1, or its receptor gene KISS1R, results in idiopathic hypothalamic hypogonadism with low LH and FSH levels [19,20]. In rare cases, i.e. two concomitant heterozygous mutations, residual LH pulses are detected, indicating some GnRH secretion might be independent of kisspeptin [19], but the loss of KISS1R function can only be predicted in humans.

In contrast to patients with KISS1/KISS1R mutations, patients with biallelic mutation in the gene encoding neurokinin B (TAC3), or its receptor (TACR3), show idiopathic hypothalamic hypogonadism with low LH levels but normal FSH levels [69]. Indeed, these patients show a low LH pulse frequency and often spontaneous recovery [70], indicating a functional but slow GnRH pulse generator. As such, blocking the kappa opioid receptor, i.e. the dynorphin receptor, with naloxone increases LH pulse frequency in these patients [70]. Similarly, an increase in LH pulse frequency due to dynorphin receptor antagonism can be seen in sheep [71] but not rodents [72]. Naloxone-induced increase in LH secretion has also been observed in men and women during the luteal phase (but not the follicular phase or menopause) [73,74]. Neurokinin B receptor activation has no effect on LH levels in men and pre- or post-menopausal women [73,75,76] [78]. This is in contrast to sheep and rodents, where neurokinin B receptor inhibition reduces LH levels [72,77].

More details on the genetics of the probands can be found in the review [79]. Nevertheless, overall, all three neuropeptides coexpressed in infundibular kisspeptin neurons function broadly in humans as predicted from sheep and rodent data. However, species differences exist and more data from non-human primates are needed to fully understand the control of GnRH neurons by KNDy neurons. For example, the primate menstrual cycle is marked by changes in LH pulse frequency [80,81]. These changes do not exist in rodents [82], indicating that the tuning – probably by steroids – of KNDy neuron output is different. Neither neurokinin B nor its receptor mutation in mice mimics the human phenotype due to the redundancy of the tachykinin system in rodents [83]. The anatomy of the GnRH system is unique to primates, which affects the GnRH/KNDy neuron relationship outlined in rodents [49,65].

From bench to bedside

Most patients with idiopathic hypothalamic hypogonadism – even those with mutations that impair GnRHR efficiency - can initiate folliculogenesis and spermatogenesis after pulsatile delivery of exogenous GnRH [79,84]. In contrast, kisspeptin stimulation does not work with patients whose GnRH neuron development or signaling is impaired [79] and its efficacy cannot be sustained over time [85]. Nevertheless, the KNDy neuronal population appears to be a strategical target for conditions with increased GnRH pulsatility, such as postmenopausal hot flushes and polycystic ovary syndrome (PCOS) [79,86].

Hot flashes (also referred to as vasomotor symptoms) are highly disruptive symptoms of the menopausal transition, occurring at least daily for ~4 years and affecting 85% of menopausal women. It is described as a sudden feeling of heat in the upper body, most intense over the face, neck and chest, for few minutes, followed by feeling of cold, [87]. Postmenopausal women show hypertrophied hypothalamic neurons that overexpressing kisspeptin/neurokinin B [64]. Ovariectomy-induced neuronal hypertrophy is reversed by estradiol replacement in non-human primates [88]. Although hot flushes do not systematically synchronize GnRH/LH pulses as previously thought [89], a role of KNDy neurons in thermoregulation is evident [90] and hot flash-like events can be induced in mice by stimulation of KNDy neurons [91]. Clinically, the use of fezolinetant, an orally active small-molecule and selective neurokinin-3 receptor antagonist, has become the method of choice to alleviate the frequency and severity of hot flushes [92].

PCOS is the most common of anovulatory subfertility in women of reproductive age associated with abnormal production of androgens by the ovaries; the name describes the numerous small cysts that form in the ovaries. The exact cause of PCOS is unknown [93]. Increased GnRH pulsatility in such patients favors LH secretion, therefore leads to anovulation [65]. As with hot flashed, the efficacy of fezolinetant, in women with PCOS was evaluated. While the treatment showed promising reduction in LH and testosterone levels, it did not improve follicular development or the menstrual cycle in the time frame of the treatment that was probably too short [92]. Alternatively, one might predict that increasing dynorphin inhibition with an opioid agonist might normalize GnRH/LH pulses. A partial agonist could indeed reduce LH levels, restore pituitary GnRH response and improve the outcome of induced ovulation in women with PCOS [86], opioid regulation might offer an alternative route for the treatment of PCOS treatment. However, more needs to be known about the regulation of GnRH pulses in humans and its dysregulation in women with PCOS to find an effective therapeutical option.

Conclusions

A large body of data supports a role for KNDy neurons in the genesis of GnRH/LH pulses across species, and communication between KNDy neurons is central to GnRH/LH rhythmogenesis. For obvious convenience, most of the data deciphering the KNDy-to-KNDy cellular communication come from rodents, which presents a challenge for translational research. The menstrual cycle is significantly different from the estrous cycle. Major changes in GnRH/LH pulse frequency occur between the follicular and luteal phases of the menstrual cycle, while the GnRH/LH pulse frequency remains relatively constant during the estrous cycle. Thus, a better understanding of the function of KNDy neurons in non-human primates appears to be a necessary step to address pathological GnRH/LH rhythmogenesis in the clinical setting.