The electrophysiologic properties of gonadotropin-releasing hormone neurons

Summary

For about two decades, recordings of identified gonadotropin-releasing hormone (GnRH) neurons have provided a wealth of information on their properties. We describe areas of consensus and debate on the intrinsic electrophysiologic properties of these cells, their response to fast synaptic and neuromodulatory input, Ca²⁺ imaging correlates of action potential firing and signaling pathways regulating these aspects. How steroid feedback and development change these properties, functions of GnRH neuron subcompartments and local networks revealed by chemo- and optogenetic approaches are also considered.

Article Summary:

For about two decades, recordings of identified gonadotropin-releasing hormone (GnRH) neurons have provided a wealth of information on their properties. We describe areas of consensus and debate on the intrinsic electrophysiologic properties of these cells, their response to fast synaptic and neuromodulatory input, Ca2+ imaging correlates of action potential firing and signaling pathways regulating these aspects. How steroid feedback and development change these properties, functions of GnRH neuron subcompartments and local networks revealed by chemo- and optogenetic approaches are also considered.

Introduction

In the half century since GnRH was sequenced, studies using native GnRH decapeptide and antagonist analogs demonstrated the pattern of GnRH release was vital for normal physiology. The hypogonadal (hpg) mouse, a natural GnRH knock-out, established this hormone as the critical final output from the central nervous system to regulate the reproductive system through its effects upon the anterior pituitary gonadotropins luteinizing hormone (LH) and follicle-stimulating hormone. A rich history of physiologic studies on the central control of reproduction is reviewed in this special issue. In this chapter, we focus in on the neurobiological mechanisms revealed by studies of identified GnRH neurons in brain slices, nasal explants and in vivo to help explain how their intrinsic and synaptic properties change with reproductive state, and how these cells are integrated into steroid-responsive networks.

The before time

For the current and next generations of young scientists, for who fluorescently-identified living neurons are a norm, we briefly visit the time before green fluorescent protein (GFP) and other genetically-encoded marker genes. The first reported recordings of putative GnRH neurons were in 1984 by the laboratory of Martin Kelly¹. A few statistics from that paper are worth noting. Guinea pig cells were recorded in the arcuate nucleus using electrodes containing procion-yellow, allowing post hoc identification of recorded cells and, when combined with immunohistochemistry using an anti-GnRH antibody, identification of the recorded cell’s phenotype. Of 102 cells recorded, just four were GnRH-immunopositive. It is important to note that the antibody used in these studies was reported the same year to non-specifically label neurons in the arcuate nucleus in the rat². A later report, using a similar approach with a validated GnRH antibody, revealed that GnRH neurons were sensitive to estradiol and opioid receptor agonists³. Later efforts to identify recorded GnRH neurons post-hoc employed reverse-transcriptase polymerase chain reaction on the cytosol from single neurons collected after patch-clamp recordings to detect GnRH transcripts⁴. These labor-intensive methods of identification were supplanted by promoter transgenic approaches and have since been abandoned.

Identified GnRH neurons

Near the turn of the century, promoter-driven genetic approaches made it possible to identify specific cell types in mammalian tissue. People studying GnRH neurons were among the first to take advantage of this because of the difficulties of recording from this small, anatomically-diffuse population⁵. Early success was likely facilitated by the strength and cell-specificity of the GnRH promoter⁶. GnRH neurons have been identified in living tissue using β-galactosidase substrates⁷, Ca²⁺ indicators^8–13 or variations of GFP derived from Aequorea victoria in mice^14–16, rats¹⁷, and medaka fish¹⁸. GFP identification is the method most used and will be the main source of studies covered in this review. Because of space limitations, the reader is referred to the article on non-mammalian systems for discussion of the many substantial contributions to GnRH neuron electrophysiology from work on teleost fish.

Intrinsic properties of GnRH neurons

For explanations of recording approaches, the reader is referred to these reviews^19–21. Initial studies of GnRH neuron properties suffered from a lack of consistent and rigorous methodology that had been established for patch-clamp investigation; essentially no lab in the field escaped errors ranging from use of inappropriate pipette solutions, to failure to report quality parameters allowing the reader to evaluate data integrity, and to failure to explain how properties (e.g., action potential threshold, resting potential) were defined; these limitations resulted in variable data. We have chosen to focus our review on work done after these initial growing pains.

The majority of GnRH neurons recorded in brain slices exhibit spontaneous action potential firing that is often arranged into short term patterns called bursts, and an array of voltage-gated conductances typical of many neurons, including tetrodotoxin-sensitive Na⁺²², multiple K^{+10, 23–25}, Ca^{2+26, 27}, hyperpolarization-activated²⁸, and transient receptor potential (TRP)^29–31. These conductances enable action potential firing, regulate firing patterns and mediate membrane responses to inputs. From this rather standard list, we highlight four aspects that are important to GnRH neurons: action potential shape, firing patterns, currents regulating action potential initiation, and Cl⁻ homeostasis.

The GnRH neuron action potential:

GnRH neurons are often spontaneously active from a basal potential between −75 and −60 mV in brain slices^{14, 32}. The GnRH neuron action potential is majestic (Figure 1A). In current-clamp recordings with a pipette solution mimicking intracellular milieu and amplifier settings allowing more precise characterization of these rapid events (e.g., high-frequency (≥10kHz) acquisition and filtering), the amplitude of the first action potential induced by current injection routinely achieves 90 mV from action potential threshold (defined as 1V/s)^{23, 33}. Despite this amplitude, the spikes are over quickly, with full width at half maximum averaging under 0.8ms. Spikes are followed by a pronounced afterhyperpolarization potential (AHP) of about 25mV, and show little evidence of frequency adaptation. Finally, a slow (peaking 1-2s post threshold) afterdepolarizing potential (ADP) follows the AHP in many GnRH neurons and can contribute to ongoing firing^34–36. These characteristics produce a spike profile that distinguishes these cells and is consistent among GnRH neurons (Figure 1). Where these action potentials are initiated is an interesting question. Simultaneous recordings of proximal dendrites and soma indicate that action potentials are initiated in dendrites in some GnRH neurons^{37, 38}. Immunoreactivity for ankyrin G, a protein often linked to the site of action potential initiation, revealed that it is typically located within 150 μm from the soma in GnRH neurons³⁹. Consistent with action potential initiation in the dendrites, ankyrin was located in one of the dendrites of most (75%) GnRH neurons, and in the axon in a small proportion of these cells³⁹. Future work can investigate if the site of initiation is constant for a particular neuron, or if it is regulated by the type of input being received, and/or can be in other regions such as the terminals.

Figure 1: Action potential firing in GnRH neurons.

A. GnRH neuron action potential waveform and its regulation by individual conductances. B. Short-term burst firing and postulated underlying conductances. C. Postulate for generation of long-term firing patterns from bursts. Current abbreviations: I_A, A-type K⁺ current⁵⁸; I_T, T-type Ca current ²⁺²⁷; I_KCa, Ca²⁺-activated K⁺ current, SK small conductance I_KCa¹⁰, I_NaP, persistent Na⁺ current;I_h, hyperpolarization-activated current ²⁸.

GnRH neuron firing patterns:

Spontaneous GnRH firing is not dependent upon fast synaptic transmission⁴⁰, GnRH itself⁴¹ or kisspeptin⁴², a major activator of GnRH neurons. An interesting feature of spontaneous GnRH neuron firing is that it exhibits short-term (burst firing, Figure 1B) and long-term patterning (Figure 1C)^{10, 43–46}. GnRH neuron bursts have a longer intraburst interval (~150-400ms) than neurons in the thalamus and cortex^{47, 48} or even magnocellular neuroendocrine cells⁴⁹, and bursts are short, with typically two to eight spikes per burst⁵⁰. This low spontaneous frequency is curious because GnRH neuron firing can be driven at much higher rates by current injection, suggesting the ionic conductances of these cells are not in and of themselves limiting^{23, 36}. As reviewed ^{20, 21}, GnRH neurons recorded in brain slices may be quiescent or spontaneously firing. While most GnRH neurons exhibit irregular bursting, 1-2% of these cells exhibit parabolic bursting riding on marked slow (~0.05Hz) oscillations in membrane potential⁵¹. Burst firing is linked with peptide release in magnocellular neurons⁵² and the higher frequency activity within bursts is likely important for GnRH release⁵³. Long-term alterations between peaks and nadirs in firing rate have been observed in both sexes; the frequency of these peaks resembles that of LH release in vivo and is likewise modified by gonadal steroid feedback^{45, 46}. Short- and long-term patterns may be related. One investigation demonstrated that bursts within a GnRH neuron maintain consistent characteristics between peaks and nadirs, but the interburst interval increases during nadirs⁴³. Further studies of this relationship and its modulation, the underlying ionic conductances and excitation-secretion coupling in GnRH neurons are required.

Interestingly, despite demonstration of synaptic connections, bundling and even cytoplasmic bridges among GnRH neurons^54–56, there are not convincing data that GnRH neurons in brain slices are coordinated with one another from dual patch-clamp recordings^{51, 57} or calcium imaging⁹. It is important to point out that few attempts have been published despite the fascination of the field with pulsatile release and the presumption that this involves some sort of coordination among these cells. Such studies are technically difficult and the possible caveats of looking for coordination within a reduced slice preparation are many; for example, if connecting fibers are severed during slice preparation, a false negative result will be obtained. This is an important area for future studies, which might take advantage of in vivo methods that can simultaneously monitor multiple cells.

Ionic conductances of GnRH neurons:

GnRH neurons express a typical set of voltage-gated ion channels that enables action potential firing. Channels that can be active at subthreshold membrane potentials help sculpt spike initiation. GnRH neurons have a very large fast-transient (A-type) K⁺ current that fights depolarization^{24, 58}, and may contribute to both the narrow width of the spike and the large amplitude AHP. Ca²⁺-activated K⁺ currents in these cells appear to regulate intra and interburst intervals¹⁰ and the amplitude of the ADP³⁴. Opposing these outward currents are several smaller magnitude inward currents, specifically hyperpolarization-activated or I_h^{24, 28}, low-voltage-activated Ca²⁺ or I_T²⁷, a Ca²⁺-activated mixed cation or CAN⁵⁹, and a tetrodotoxin-sensitive persistent Na⁺ current or I_NaP^{34, 59}. Several of these latter conductances along with Ca²⁺-activated K⁺ currents have been implicated in both burst firing and pacemaker activity in other neurons. Both I_T and I_h contribute to rebound firing after the termination of a hyperpolarizing current injection and CAN and I_NaP augment the ADP in these cells. Interestingly, many of these are regulated by estradiol in a manner that would tend to increase GnRH neuron activity during positive feedback, thus may be important in the achievement of that state.

Cl⁻ homeostasis:

An interesting feature of GnRH neurons that was once controversial is their maintenance of higher intracellular Cl⁻ levels than is typical of most adult neurons. There is now consensus that the Cl⁻ reversal potential is depolarized relative to the action potential threshold in these cells^{60, 61}. As a result, activation of γ-aminobutyric acid type A (GABA_A) receptors, for which Cl⁻ is the main permeable ion, depolarizes these cells and can trigger action potential firing in preoptic GnRH neurons in rodents and terminal nerve GnRH neurons in teleost fish^60–62. High Cl⁻ levels are likely attributable to continuing activity of the Cl⁻-accumulating cation cotransporter NKCC1 into adulthood, in contrast to most neurons in which the Cl⁻-depleting cotransporter KCC2 become dominant⁶³. At the time of this discovery, GnRH neurons were among a small population of primarily sensory neurons for which excitatory GABA was identified, but this is now recognized in other hypothalamic cells^{64, 65} and throughout the adult brain⁶⁶.

The induction of action potentials in response to GABA, widely recognized as the primary mode of fast inhibition in the brain, raises several interesting questions about GnRH neurons. First, is this a phenomenon limited to the perisomatic region where measurements have been made? Differential distribution of Cl⁻ cotransporters in subcompartments has been observed⁶⁷ and may lead to regional functional changes⁶⁸. While expression of NKCC1 is prominent, GnRH neurons do express KCC2, but subcellular localization requires further investigation^{61, 69}. Second, is regulating Cl⁻ homeostasis a possible mechanism for controlling GnRH activity? In this regard, the excitatory response to GABA appears to persist regardless of age, gonadal status, or sex, even in mice in negative energy balance^{61, 70}. Reduced excitatory response to GABA was observed in prepubertal mice exposed to androgen in utero, a model used to generate a phenotype resembling polycystic ovary syndrome; this blunting was not attributable, however, to a change in the Cl⁻ reversal potential⁷¹. In the models studied, regulating intracellular Cl⁻ thus does not appear to be a major point of physiologic control. Third, is it GABA alone or does the initial depolarization initiated by GABA activate other inward currents that boost the depolarizing response^{8, 72}? Fourth, what inhibits GnRH neurons? While no ‘fast’ synaptic inhibition is known, several neuromodulators, discussed below, may do so.

Fast synaptic transmission to GnRH neurons

In addition to having a non-standard response to activation of the GABA_A receptor, the frequency of spontaneous fast synaptic transmission to GnRH neurons is low compared to, for example, cortex⁷³, and even other hypothalamic neurons regulating reproduction⁷⁴. Fast synaptic transmission detectable at the cell soma is primarily GABAergic and glutamatergic; blocking α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA), N-methyl-D-aspartate (NMDA) and GABA_A receptors eliminates postsynaptic currents (PSCs) and no response is observed to local application of glycine, another Cl⁻-permeable ionotropic receptor⁷⁰. Immunoreactivity for vGLUT2 and vGAT, markers of putative glutamatergic and GABAergic synaptic appositions, respectively, identified the proximal dendrite as the main input target⁷⁵. Of note, knockout of subunits of either ionotropic GABA or glutamate receptors have had little effect on reproduction^{76, 77}. This may indicate fast synaptic transmission is not critically important, or that functional compensatory mechanisms were initiated by the knockout.

Curiously, the density of putative glutamatergic synapses detected with immunostaining methods is twice that of putative GABAergic inputs on GnRH neuron somata and proximal dendrites⁷⁵ . Functional measures of glutamatergic synaptic transmission directly to GnRH neurons made with patch-clamp, however, indicates it is infrequent. A low percentage of cells responding to glutamate receptor agonists, particularly NMDA receptor agonists, was noted in the first recordings of GFP-identified cells¹⁴ and confirmed in subsequent measurements of glutamatergic transmission^{78, 79} and of percent of cells responding to bath application of receptor agonists^{8, 80}. The spontaneous AMPA-mediated PSC frequency in mice is typically under 0.1Hz. Suppression by estradiol negative feedback has been reported^{78, 79}, but should be interpreted with caution as small changes in transmission rate that may not be biologically important can result in large fold changes. Of note, higher rates of AMPA-mediated transmission to GnRH neurons occur in rats, with an additional increase during estradiol positive feedback on proestrus, when Ca²⁺-permeable AMPA receptors are inserted into the membrane⁸¹. Positive feedback is also associated with more spines, considered a site for glutamatergic synaptic input, on murine GnRH neurons⁸² and with changes in expression of genes encoding glutamate receptor subunits in these cells⁸³, but it is important to point out that it is not known if these anatomical and molecular changes are associated with functional differences.

Despite these latter observations, the low rates of glutamatergic transmission observed may call into question its relevance, but two points are worth making. First, the available data on spontaneous transmission is from somatic recordings. Input occurring at a distance from the soma, such as those impinging onto the very distal processes of GnRH neurons⁸⁴, may have either decayed before detection by the somatic recording electrode or been eliminated by brain slice preparation. Second, GnRH neurons may have more glutamatergic input than is appreciated as increasing neurotransmission within a brain slice by blocking K⁺ channels markedly increases glutamatergic transmission⁷⁹; this may indicate low initial probability of release at these synapses under most conditions.

GABAergic transmission is more prevalent but still rarely exceeds 2Hz, being more typically under 1Hz. Several studies demonstrated that the frequency of GABAergic transmission to these cells correlates with what is expected for GnRH/LH output for that physiologic state. Progesterone negative feedback, estradiol negative feedback and negative energy balance reduce GnRH/LH pulse frequency and reduce the frequency and sometimes amplitude of GABAergic transmission to GnRH neurons^{70, 85–87}. In contrast, states that increase GnRH/LH release, such as mild androgen elevation and estradiol positive feedback typically^85–88, but not always⁸⁹, correlate with increased GABAergic transmission and PSC amplitude. Consistent with these correlational studies, electrical stimulation of the anteroventral periventricular nucleus (AVPV) induces GABAergic and glutamatergic evoked PSCs in GnRH neurons⁹⁰; GABA was predominant and required for AVPV stimulation to elicit firing of most GnRH neurons. Optogenetic stimulation specifically of AVPV kisspeptin neurons, which use GABA as a cotransmitter, evokes GABA_A receptor-dependent firing in GnRH neurons⁹¹. Finally, blocking GABA_A receptors during in vivo recordings of GnRH neurons consistently reduced firing⁹². These findings consistently point to the action of GABA via the GABA_A receptor as being excitatory to GnRH neurons. Activation of GABA_B receptors, in contrast, inhibits GnRH neuron electrical activity^{90, 93, 94}, indicating GABA may bidirectionally control GnRH neuron firing depending on the receptor activated.

It should be noted that the rates and/or patterns of glutamate and GABA synaptic transmission to GnRH neurons in vivo may be substantially different, as brain slice preparation removes many afferent soma. Changes in synaptic transmission detected in brain slices from animals in different physiological or pathological states nevertheless indicate functional synaptic plasticity and are, as such, meaningful. Identification of where fast synaptic inputs originate and the reproductive states and/or phases of a pulse cycle when these inputs are active are interesting future directions.

Recording GnRH neurons in vivo

The brain slice recordings used in work discussed above and below allow targeting of individual cells and mechanistic studies, but also have limitations. Perhaps most prominent is the reduced nature of these preparations, which maintains only local circuitry, removing distal input soma and even some proximal ones. While this can be mitigated to some extent by using different slice orientations^{86, 95}, this is still far from the integrative in vivo situation in which the regulation of reproduction typically occurs. Recording from GnRH neurons in vivo avoids this limitation. This is challenging because the few GnRH neurons are scattered and deep relative to the surface of the brain. One paper has accomplished this by using transpharyngeal approach to expose and record from GnRH neurons under anesthesia⁹². The only area in the mouse brain where GnRH-GFP neurons are superficial enough to be targeted with a recording pipette is dorsal to the optic chiasm at the ventral-most surface of the anterior hypothalamic area, along the posterior cerebral artery. Due to the nearby beating artery, it was only possible to achieve short duration targeted extracellular recordings. Despite these challenges, this heroic effort nonetheless provided important confirmatory evidence of several observations made in brain slices. Specifically, GnRH neuron firing is variable with spontaneous bursts, and responses to GABA, glutamate and kisspeptin-10 are all excitatory.

In sum, there are several aspects of GnRH neuron electrophysiology that are quite consistent among cells. These include basic properties such as a relatively high input resistance (compared to, for example, cortical and hippocampal pyramidal neurons), the observation of spontaneous firing activity, grouping of action potentials into bursts in most cells, action potential properties when recorded with physiologic chloride levels, a predominance of fast GABAergic over fast glutamatergic input in recordings made in the perisomatic region and a depolarizing/excitatory response to GABA. Burst properties can exhibit considerable heterogeneity among cells and the importance of this for neuroendocrine output is something that requires further investigation.

Neuromodulation of GnRH neuron electrical activity

Since the first recordings of GnRH neurons, a large repertoire of neuromodulators altering the electrical activity of GnRH neurons has been compiled. For space reasons, we focused on non-conventional neuromodulators that have emerged as important regulators of GnRH neuron activity, and on RF-amide neuropeptides. For detailed accounts of GnRH neuron regulation by neurotransmitters and neuropeptides, readers are referred to excellent recent reviews^{96, 97}.

Non-traditional neuromodulators

Studies of endocannabinoid, nitric oxide and prostaglandin E₂ regulation of GnRH neuron electrical activity are relatively recent. These neuromodulators are not stored in vesicles but rather produced on-demand, acting directly or indirectly on GnRH neurons, or as retrograde transmitters synthesized by GnRH neurons to regulate synaptic transmission.

Endocannabinoids (eCB):

GnRH neuron membrane potential depolarization or activation of receptors stimulates synthesis and release of 2-arachidonoylglycerol (2-AG), which acts on presynaptic type-1 cannabinoid receptors to decrease GABA release^98–101. eCB-dependent suppression of GABA release occurs in mice of both sexes and may occur constitutively, at least in brain slices^{98, 99, 101}. GnRH neurons also synthesize anandamide, which may signal via TRP vanilloid channels to suppress constitutive eCB signaling from GnRH neurons^{30, 31}. Importantly, estradiol via estrogen receptor β (ERβ) rapidly promotes 2-AG synthesis, subsequent suppression of GABA release and decreased GnRH neuron firing in metestrous females¹⁰⁰, suggesting a role of retrograde eCB suppression of GABA release as one mechanism of estradiol negative feedback.

Nitric oxide (NO):

The gaseous neuromodulator NO may affect GnRH neuron electrical activity through multiple mechanisms with opposing effects. The “NO donor” L-arginine directly suppresses action potential firing in >90% of GnRH neurons from both sexes via neuronal NO synthase (nNOS)-dependent NO synthesis, modulation of soluble guanylyl cyclase (sGC), subsequent suppression of a depolarizing plateau potential and activation of a K⁺ current¹⁰². In addition to these direct inhibitory effects, NO produced in response to activation of specific receptors increases GABA and glutamate release, indirectly increasing GnRH neuron activity^{31, 103, 104}. Of note estradiol, acting at ERβ, rapidly increases neurotransmitter release and action potential firing in GnRH neurons by mobilizing this pathway in proestrous female mice, suggesting a role in estradiol positive feedback¹⁰⁴. NO may also directly alter GnRH neuron excitability by accelerating recovery from prior exposure to kisspeptin-10, thereby enabling repeated responses to this neuropeptide (see below)¹⁰⁵ (Figure 2). Whether or not GnRH neurons synthesize NO is controversial. Neither nicotinamide adenine dinucleotide phosphate (NADP)-diaphorase¹⁰² nor nNOS¹⁰⁶ are detectable in GnRH neurons, arguing against synthesis. In contrast, pharmacological experiments and detection of Nos1 mRNA and nNOS immunoreactivity at the ultrastructural level in GnRH neurons argues for synthesis³¹, opening the possibility NO acts as a retrograde or neuromodulatory signal.

Figure 2: Kisspeptin-Kiss1r signaling in GnRH neurons.

Schematic illustration of the main intracellular pathways identified to date that contribute to kisspeptin-10-evoked stimulation of GnRH neurons. The main effects are to inhibit multiple K⁺ conductances and to activate non-selective cation channels (TRPC4), resulting in membrane depolarization, further activation of voltage-gated currents and, eventually, action potential firing. See text for details and references. AHP: afterhyperpolarization; c-Src: protein-tyrosine kinase Src; DAG: diacyl glycerol; I_A: A-type K⁺ current; I_{KCa slow AHP}: Ca²⁺-activated K⁺ current mediating the slow AHP; I_Kir: inward rectifying K⁺ current; IP3: inositol triphosphate; K_Ca: Ca²⁺-activated K⁺ channel; Kiss1r: kisspeptin receptor; K_v: voltage-gated K⁺ channel; ME: median eminence; NO: nitric oxide; PI4K: phosphatidylinositol 4-kinase; PIP2 phosphatidylinositol 4,5-bisphosphate; PKC: protein kinase C; PLCβ, phospholipase β; POA: preoptic area; sGC: soluble guanylate cyclase; TRPC4: transient receptor potential channel 4.

Prostaglandin E2 (PGE₂):

PGE₂ directly depolarizes the membrane potential and increases action potential firing of most GnRH neurons regardless of sex or estrous cycle stage. These effects are mediated via the prostaglandin EP2 receptor, adenylyl cyclase and protein kinase A to open a non-selective cation channel¹⁰⁷. Ambient PGE₂ may help control GnRH neuron membrane potential and spontaneous firing as blocking its synthesis reduces activity in brain slices¹⁰⁷. Evidence indicates that astrocytes are a primary source of PGE₂ release^{99, 107, 108}.

These nontraditional neuromodulators directly and indirectly regulate GnRH neurons. Interestingly, there may be crosstalk among these pathways as PGE₂ is required for depolarization-induced, eCB-mediated, suppression of GABA transmission to GnRH neurons⁹⁹. Moreover, additional neuropeptides and hormones may recruit these neuromodulators to produce their effects on GnRH neuron firing, as is the case of eCBs and NO^{31, 101, 103}.

RF-amide peptides

Exogenous kisspeptin-10 drives with high potency (low nM EC₅₀) prolonged membrane potential depolarizations, action potential firing and increased intracellular Ca²⁺ ([Ca²⁺]_i) fluctuations in GnRH neurons, as well as GnRH secretion^{29, 92, 105, 109–116}. Kisspeptin receptor (Kiss1r) expression in GnRH neurons is both necessary and sufficient for kisspeptin-10 effects on GnRH neuron firing and LH secretion, for puberty and overall fertility^{42, 117}. Kisspeptin-10 may also have indirect actions by increasing GABA and glutamate release on GnRH neurons, and by modulating NO-synthesizing neurons in the preoptic area^{106, 110}. Kisspeptin-10 actions in GnRH neurons are mediated through inhibition of multiple K⁺ currents, including inward-rectifying, A-type and the K⁺ current mediating the slow AHP^{29, 110, 111, 118–120}, and via activation of a non-selective cationic current likely mediated by TRP canonical (TRPC) channels^{29, 118}. The role of [Ca²⁺]_i in GnRH neuron membrane responses to kisspeptin is somewhat unclear. Although kisspeptin-10 increases [Ca²⁺]_i in GnRH neuron somata, dendrites and terminals^{12, 112, 118}, kisspeptin-10-induced excitation is resistant to Ca²⁺ buffering^{111, 121} and to depletion of Ca²⁺ stores^{112, 121} (Figure 2). Interestingly, kisspeptin-10-evoked increases in [Ca²⁺]_i are required for GnRH secretion at the median eminence (ME)^{116, 122}. Activation of phospholipase Cβ (PLCβ) and c-Src tyrosine kinase are necessary for kisspeptin-10 effects^{29, 118, 121}. PLCβ-mediated hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP2), which inhibits TRPC channels in a subset of GnRH neurons, and the ensuing PIP2 depletion may contribute to prolonged activation of these channels¹²¹ (Figure 2).

The effects of exogenous kisspeptin-10 last many minutes and cannot be repeated at the GnRH neuron soma in brain slice recordings^{111, 118}. The prolonged nature of this response is likely the result of PIP2 depletion because inhibition of phosphatidylinositol 4-kinase (PI4K), which synthesizes PIP2, prolongs kisspeptin-10 excitation of GnRH neurons^{105, 121}. Of interest in this regard, NO increases PI4K activity¹²³ and thus PIP2 synthesis and thereby both shortens kisspeptin-10-mediated excitation and enables subsequent kisspeptin-10 stimulations of GnRH neurons¹⁰⁵ (Figure 2). NO-mediated recovery from kisspeptin excitation may provide an “off” signal terminating kisspeptin-10-induced GnRH neuron activity/secretion and might facilitate periodic kisspeptin signaling to promote pulsatile GnRH secretion. Of note, because NO diffusion can potentially affect many GnRH neuronal elements, NO-modulation of kisspeptin signals might help coordinate GnRH secretion between many terminals¹⁰⁵. It should be noted that endogenous kisspeptin actions may be repeatable as consecutive optogenetic stimulations of kisspeptin neurons at ~10-minute intervals repeatedly evoke action potential firing in GnRH neurons⁹¹ and LH secretion in vivo¹²⁴, suggesting that “on-off” kisspeptin effects might require sustained activation of the kisspeptin receptor and associated signaling pathways. The impact of NO on kisspeptin signaling has only been assessed so far at the GnRH neuron somata¹⁰⁵ and it remains to be seen if kisspeptin signaling in the ME is similarly regulated by NO. This is particularly important as GnRH neuron subcompartments may operate independently from one another in vivo¹²⁵ (see below). Lastly, the source of NO remains unknown. Bearing in mind the controversy around nNOS expression discussed above, GnRH neurons remain possible candidates. Alternatively, nNOS-expressing neurons are found in the preoptic area, some of which, interestingly, express Kiss1r, and arcuate nucleus¹⁰⁶.

RF amide-related peptide 3 (RFRP-3), the mammalian counterpart to avian gonadotropin-inhibiting hormone¹²⁶, inhibits GnRH neuron [Ca²⁺]_i oscillations in nasal explants and decreases the firing rate of 50-70% of prepubertal and adult male and female GnRH neurons via binding to neuropeptide FF receptor 1 and activation of K⁺ currents^127–129. This inhibition transiently opposes the prolonged actions of kisspeptin-10^{128, 129}, in keeping with the inhibitory effects of RFRP-3 on GnRH secretion in brain slices^{122, 130} and LH secretion in vivo^{131, 132}, making RFRP-3 a candidate “off” signal. It should be noted, however, that ≈10% of GnRH neurons increase their firing in response to RFRP-3¹²⁷ and that central administration of RFRP-3 may stimulate LH secretion in males¹³².

Functional interrogation of native circuits

While the studies reviewed above have been very valuable, at least some of these results should be interpreted with caution. Bath or local agonist and antagonist applications provide important information but have limited interpretative value due to off-target actions at all binding sites present within the preparation. Moreover, many neuromodulators are expressed in several brain areas and these neuronal populations may express multiple cotransmitters and/or be involved in completely different physiological functions. Inferring the functional impact of specific neuromodulatory populations thus requires a combination of agonist-induced changes in electrical activity/[Ca²⁺]_i, neuronal tract-tracing and functional circuit interrogation with opto- and/or chemogenetics.

Applying these approaches to the control of GnRH neurons by kisspeptin revealed that kisspeptin neurons in the preotic area in females potently stimulate GnRH neuron firing and that a subset of preoptic kisspeptin neurons further excite GnRH neurons by co-releasing GABA⁹¹. These types of studies also unveiled an unsuspected indirect mode of communication between arcuate kisspeptin neurons and GnRH neurons, via projections to and stimulation of preoptic kisspeptin neurons; whether this pathway substantially contributes to GnRH and LH secretion in vivo is not known¹³³. It is interesting to note that direct interrogation of the arcuate kisspeptin-to-GnRH pathway in brain slices still eludes researchers, probably because, in rodents, the former neurons project predominantly to the GnRH neuron distal processes near and in the ME¹³⁴, a compartment almost inaccessible for conventional electrophysiology. Further studies aimed at determining direct inputs to the GnRH neurons, and their impact on these cells’ activity are needed to gain deeper understanding of the regulation of GnRH release patterns.

Function of GnRH neuron compartments

Another area that has progressed over the past 10-15 years has been the functional definition of GnRH neuron compartments. Most patch-clamp recordings have been made of the soma, but dual recordings of soma and proximal processes with dendrite-like morphology revealed that these processes can initiate and actively propagate action potentials in some GnRH neurons^{37, 38, 135}. In rodents, most GnRH neuron projections to the median eminence exhibit features such as presence of spines, synaptic input, larger diameter and ultrastructural profiles together reminiscent of dendrites^{84, 135, 136}. Because of these attributes, which collectively suggest these projections share dendritic and axonal properties, they have been referred to as “dendrons”¹³⁵. Near the median eminence (ME), GnRH neuron processes often branch out into multiple axon-like profiles (i.e., smaller diameter) with putative secretory terminals in proximity to blood vessels^{84, 135}. Determining if these unusual features of GnRH neuron projections exist beyond mice and rats will be an important future research direction¹³⁷.

Understanding GnRH neuron subcompartments is particularly relevant because several studies suggest functional differences may exist among GnRH somata and/or proximal processes in the preoptic area and GnRH neuronal elements near the ME. In vivo opto- and chemogenetic suppression of GnRH neuron activity in behaving mice revealed that distal GnRH neuron projections may operate independently of activity in their somata and proximal dendrites, and, further, that these latter regions may be sufficient to control pulsatile, but not surge LH secretion¹²⁵. In brain slice studies to look at mechanisms, GnRH release from the perisomatic and terminal regions is differently regulated¹³⁸. In agreement with this, distal projections near the ME are the overwhelming site of arcuate kisspeptin neuron inputs to GnRH neurons in rodents¹³⁴. This region responds to local kisspeptin-10 application in an action potential-independent manner with increases in [Ca²⁺]_i and GnRH release^{12, 13, 138}. Neurokinin B (NKB), dynorphin A and glutamate, co-expressed in arcuate kisspeptin neurons, do not alter [Ca²⁺]_i dynamics in the distal processes of GnRH neurons, but NKB application to the ME induces GnRH release in a kisspeptin-independent manner^{12, 13, 138}. Differences in kisspeptin-10-induced signaling might also exist between GnRH somata and distal processes^{12, 122}. These observations are consistent with those made in medial basal hypothalamic preparations from rodent and sheep, which contain arcuate kisspeptin but very few GnRH neuron somata. These preparations release GnRH in response to kisspeptin^114–116 and, in rodents, can release GnRH in a pattern reminiscent of pulsatile LH secretion in vivo¹³⁹. This suggests arcuate kisspeptin neurons and their projections to the distal processes of GnRH neurons might be sufficient to support pulsatile secretion. An even more reduced preparation of isolated rat ME exhibited episodic release¹⁴⁰, perhaps indicating only the very distal processes of both GnRH and kisspeptin neurons are needed. Together these observations suggest specialization of different regions of the GnRH neurons; understanding how various neuromodulators influence these functional compartments is a rich area for future studies.

Development of GnRH neuron properties

Early GnRH secretion:

The ability to release GnRH in a pulsatile manner is a core function of these neurons. GnRH is detectable before these cells migrate from the olfactory placode and neurosecretion appears to arise early in their development¹³⁰. Sexual differentiation of the brain requires a postnatal testosterone surge; this may rely upon GnRH secretion immediately after birth¹⁴¹, although other work suggests this may be GnRH independent¹⁴². In mice, connections from the putative pulse generator arcuate kisspeptin neuron and GnRH neurons are established before birth^{143, 144}. Consistent with this, fetal MBH isolated from humans¹⁴⁵ and non-human primate¹⁴⁶ release GnRH pulses, as do GnRH neurons derived from primate and rodent olfactory placodes^147–150. Prenatal preoptic tissue containing GnRH neurons restores LH pulses and/or gonadal function models lacking endogenous GnRH including lesioned adult female monkeys¹⁵¹ and rats¹⁵² and hpg male mice^{153, 154}, suggesting prenatal tissue either has episodic release as an inherent function or can mature into this role.

Olfactory placode-derived GnRH neurons were an early model used to decipher the mechanisms underlying pulsatility and other aspects of GnRH physiology. An important advantage of this preparation is that the whole GnRH neurons is present, unlike in most brain slice preparations, which cut processes. Ca²⁺ imaging was used to assess both GnRH neuron function and coordination. Bursts of action potentials occur concomitantly with [Ca²⁺]_i oscillations in placodal GnRH neurons and GnRH neurons in brain slices from adults^{10, 12}. Increases in [Ca²⁺]_i are often equated to action potential firing but, while often related^{10, 12}, this is not technically accurate. Fluctuations in [Ca²⁺]_i can reflect any of several phenomena, including changes in action potential-dependent or subthreshold Ca²⁺ entry, receptor-mediated Ca²⁺ influx and/or Ca²⁺ release from internal stores. An important advantage to Ca²⁺ imaging is the ability to monitor simultaneously several cells. In addition, Ca²⁺ imaging makes possible the recording of subcompartments, such as GnRH neuron terminal regions, that are not readily accessible with electrophysiology.

Ca²⁺ imaging of primate and mouse GnRH neurons in placode cultures revealed GnRH neurons exhibit [Ca²⁺]_i oscillations at a higher frequency than is typical for LH release. These high frequency oscillations are typically not coordinated among cells. Intriguingly, however, periodic coordination of these higher frequency oscillations occurred at a frequency similar to LH release; further work showed this coordination is correlated with GnRH pulses^{150, 155, 156}. These cultures are devoid of neuronal CNS inputs, suggesting pulsatile GnRH secretion is an endogenous property. Supporting this postulate, the coordination of [Ca²⁺]_i oscillations in primate placodal cultures is followed by a delay in resumption of [Ca²⁺]_i oscillations. But there are several things to consider before this postulate is accepted. First, increased frequency of the high-frequency oscillation can be driven by neuromodulators, e.g., GABA^{156, 157} and estradiol^{158, 159}; this increased frequency is associated with increased coordination of oscillations, raising the question of whether this is merely an increase in mathematical probability. Second, nonneuronal cells in placodal cultures are also coordinated with GnRH neuron [Ca²⁺]_i fluctuations and GnRH release¹⁶⁰. Third, glial cells in placodal cultures express gap junctions and blocking these reduces coordination and GnRH release¹⁶¹. Consistent with this, the putative gliotransmitter ATP induces coordinated elevations in [Ca²⁺]_i in both GnRH neurons and nonneuronal cells¹⁶². Fourth, GnRH pulses in placodal cultures develop over time¹⁵⁰, in parallel with the network surrounding GnRH neurons¹⁶³. While these mechanistic questions remain to be resolved, it is clear that GnRH neurons are equipped with the exocytotic machinery for peptidergic secretion and can sustain pulsatile secretion early on.

Development of GnRH firing and signaling properties:

The literature on the intrinsic properties of GnRH neurons before adulthood is sparse but the firing activity of GnRH neurons before weaning might shape their adult properties¹⁶⁴. Placodal GnRH neurons display action potentials and are equipped with Na⁺, K⁺ (delayed rectifier and A-type), and Ca²⁺ (high- and low-voltage activated) conductances^{165, 166}. Recordings of GnRH neurons from prepubertal GnRH-GFP mice demonstrate adult-like bursting characteristics develop early and that the mean firing rate is developmentally regulated^{44, 50}. Placodal and adult GnRH neurons exhibit other consistent features including excitation by activating GABA_A receptors^{156, 167}, and less influence of glutamatergic signaling^{8, 9}. Early on, GnRH neurons also express G-protein-coupled receptors (GPCRs) and their coupling partners (G_i/o, G_q/11, G_s). G_i/o-coupled receptors provide a robust inhibition, directly via G-protein-gated inwardly rectifying K⁺ channels^{94, 168, 169}. G_q/11-coupled receptors provide a robust excitation, via PLC and downstream effectors^{29, 112, 118}. In contrast, G_s-coupled receptors provides a mild excitation, via protein kinase A and downstream effectors^{11, 120, 170}. Kisspeptin-10 potently elicits GnRH release from placodal cultures¹⁶⁸, neonatal to juvenile¹⁷¹ and adult preparations¹¹⁴. Importantly, the response of GnRH neurons to kisspeptin-10 increases during development^{109, 130}. The complexity of GnRH neuron cell signaling described above can be appreciated with the activation of G_q/11-coupled kisspeptin receptor in both placodal and adult GnRH neurons^{29, 110, 112, 118–120}. Many GPCRs identified in adult GnRH neurons¹⁷² have been found in placodal GnRH neurons and linked to a signaling pathway modulating the frequency of GnRH [Ca²⁺]_i oscillations^{162, 168, 173}. The close signaling parallel between placodal and adult GnRH neuron highlights the precocity of GnRH neurons.

Conclusions and perspectives

The above and continued characterization of GnRH neuron intrinsic properties is prerequisite to understanding physiologic and pathophysiologic regulation of their output, for generating computational models and for understanding the entire circuitry underlying GnRH release. There remain several critical and unanswered questions about the electrophysiologic properties of GnRH neurons including how is coordination achieved, how does action potential firing relate to hormone release and is this a point of feedback regulation, does signaling and its functional outcomes differ depending on which domain of the GnRH neuron receives the input, and how do changes in ionic conductances in GnRH neurons contribute to long-term patterns of activity in these cells that correlates with hormone release? Pairing electrophysiologic approaches with opto- and chemogenetic interrogation of brain circuits in vivo is a powerful approach to determine the role of specific cell populations in GnRH secretion and will expand our understanding of the functional GnRH network, generation of the secretory pattern, and the regulation of these elements throughout postnatal development, by physiologic steroid feedback, by internal and external cues and under pathophysiologic conditions.

Abbreviations:

ADP

afterdepolarization

AHP

afterhyperpolarization

AMPA

α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid

AVPV

anteroventral periventricular nucleus

c-Src

protein-tyrosine kinase Src

[Ca²⁺]_i

intracellular Ca²⁺ concentration

CAN

Ca²⁺-activated mixed cation channel

eCB

endocannabinoid

ERβ

estrogen receptor β

GABA

gamma-aminobutyric acid

GPCR

G protein-coupled receptor

GFP

green fluorescent protein

GnRH

gonadotropin-releasing hormone

hpg

hypogonadotropic mutant mouse

I_A

A-type K⁺ current

I_h

hyperpolarization-activated current

I_KCa

Ca²⁺-activated K⁺ current

I_{KCa slow AHP}

Ca²⁺-activated K⁺ current mediating the slow AHP

I_Kir

inward rectifying K⁺ current

I_NaP

persistent Na⁺ current

I_T

transient low-voltage-activated Ca²⁺

KCC2

K⁺ Cl⁻ cotransporter 2

Kiss1r

kisspeptin receptor

LH

luteinizing hormone

ME

median eminence

NKB

neurokinin B

NKCC1

Na⁺ K⁺ Cl⁻ cotransporter 1

NMDA

N-methyl-D-aspartate

nNOS

neuronal nitric oxide synthase

NO

nitric oxide

PGE₂

prostaglandin E₂

PIP2

phosphatidylinositol 4,5-bisphosphate

PI4K

phosphatidylinositol 4-kinase

PLCβ

phospholipase β

PSC

postsynaptic current

RFRP-3

RF amide-related peptide 3

sGC

soluble guanylate cyclase

TRP

transient receptor potential channel

Data Availability Statement:

Data sharing not applicable – no new data generated