8    The GnRH Neuron

Say not the struggle nought availeth,

The labor and the wounds are vain,

The enemy faints not, nor faileth,

And as things have been they remain.

If hopes were dupes, fears may be liars;

It may be, in yon smoke concealed,

Your comrades chase e’en now the fliers,

And, but for you, possess the field.

—Arthur Hugh Clough (1819–1861), “Say Not the Struggle Nought Availeth”¹

When talking about oxytocin neurons, it seems possible to convey an understanding of much of what they do, of how and why they do it. That understanding did not come easily: anomalies, confusions, and misconceptions were gradually resolved; many gaps in our knowledge were first revealed and then filled in by many laboratories. My account has hidden most of this process of excavation, filtering, and refinement, and in so doing it misrepresents the nature of science and discovery.

We have a stable understanding of how oxytocin cells generate the milk-ejection reflex. There are uncertainties, but the shape of our understanding is, for the moment, clear (though it might yet be wrong). For GnRH neurons, our theories are still churning. This chapter is longer than most and more confusing, because here I am trying to convey what our science is like, how we grapple with inconsistencies and uncertainty. We have not answered the fundamental questions about how LH pulses and the LH surge are generated. Perhaps we have broken the problems; we know better what we don’t know and need to know. By the time you read this, perhaps some of those gaps will have been filled.

A complete understanding of any physiological system must embrace many things. We need to know what it does that is important: how it contributes to the fitness of the organism of which it is a part. We also need to know the mechanisms by which it does whatever it does. Both of these require us to know how the system is determined by genetic codes, how it has evolved, and how it develops in early life.

Development is a rule-governed process. Cells born in a particular place at a particular time express just some of all the genes in an organism. That set of expressed genes confers properties that cause cells to develop and migrate in a particular way. As they migrate, they encounter other cells, and the signals that they receive change them further, suppressing some genes and engaging others—and the cells that they encounter are themselves changed by that encounter. By the end of development, we have a brain of immense complexity. If we knew all the rules, perhaps we could build something very like a brain in silicon. We might yet be able to build a brain in this way before we understood anything about its function.

In all mammals, GnRH neurons are essential for reproduction. You might imagine that, being so important, there would be many of them, but in the rat, there are only about 800. In humans, there might be about 10,000; this is how many there are in the fetus, though probably not all of them survive into adulthood.² Of the 10,000 GnRH neurons in the human fetus, about 2,000 are in the hypothalamus and are involved in regulating the pituitary. Other, “extrahypothalamic” GnRH neurons are in the olfactory bulbs, cerebral cortex, hippocampus, and a few other places, but what they do is unknown.

The GnRH neurons are extraordinary. They start life not in the brain, but in the nose. In mice, they are born in the nasal placode on the tenth day of embryonic life, together with olfactory sensory neurons. The sensory cells stay in the nose and form the vomeronasal organ, which is responsible for detecting pheromones—chemical signals that are important in many mammals for social and sexual behaviors. These sensory cells send axons into the olfactory bulbs, at the front of the brain, and the GnRH neurons migrate along them. The axons end at the outermost layer of the bulbs, but the GnRH neurons journey on through the developing forebrain. In rodents, about 800 end up scattered between the olfactory bulbs and the caudal hypothalamus. In vertebrate species from fish to humans, the hypothalamic GnRH neurons project their axons to the median eminence, where they end next to the portal blood vessels that transport releasing factors to the anterior pituitary.

At the National Institutes of Health in Bethesda, Susan Wray has long studied this migration, and in elegant and sophisticated experiments she has revealed many of the mechanisms involved.³ The migration is guided by “chemotactic” cues from the cells that GnRH neurons encounter along the way, and defects in any of many different cues can disrupt it, resulting in a failure to enter puberty, and infertility. This is associated with anosmia, a defect in the sense of smell, and the combination of symptoms is known as Kallmann syndrome, after Franz Josef Kallmann, who recognized its hereditary nature in 1944.⁴

Before GnRH was discovered, the first significant attempt to locate the source of the releasing factor regulating LH secretion was made by Halász and Pupp in 1965.⁵ When the pituitary is separated from the hypothalamus, LH secretion is abolished; in males the testes shrink, and females stop ovulating. Using a cleverly designed microscalpel, Halász and Pupp cut all around the mediobasal hypothalamus of rats, isolating it from the rest of the brain. Surprisingly, the drastic surgery had no effect at all on testis weight or structure, but it blocked ovulation in females. This suggested that the LH pulses needed to maintain the testes were regulated by the mediobasal hypothalamus, but the LH surge needed for ovulation was probably regulated elsewhere. But we now know, as Halász and Pupp did not, that the mediobasal hypothalamus of rats contains the nerve terminals of GnRH neurons but few if any of the cell bodies. However shaky the foundations of their hypothesis were, it is a hypothesis that now looks correct. But before we come to that, we need to understand more about the observations that guided our present understanding, and particularly the studies of Ernst Knobil at the University of Texas.

To generate a pulse of GnRH secretion, many of the GnRH neurons must be activated at about the same time, probably firing a burst of spikes. In conscious rhesus monkeys, Knobil and his coworkers began recording multiunit activity—signals from the combined spike activity of many neurons. In the mediobasal hypothalamus, they observed occasional “volleys” of multiunit activity, bursts of spikes that lasted for several minutes. These volleys were invariably associated with LH pulses.^6,7

This finding was astonishing. There are a few GnRH neurons in the mediobasal hypothalamus of the monkey, but other neurons in the same region regulate pulsatile secretion of growth hormone, prolactin, and thyroid-stimulating hormone, and no volleys were associated with any of those pulses. Equally unexpectedly, just before the LH surge there was a reduction in the frequency of volleys, followed sometimes by a complete cessation for a day or two—nothing was seen concurrent with the surge itself.

Similar observations were soon made in goats and in rats.^8,9 Although anesthetics generally block LH pulses, Kevin O’Byrne and his colleagues in King’s College London showed that the pulses are intact in ketamine-anesthetized rats. As in monkeys, electrodes in the mediobasal hypothalamus could record volleys of multiunit activity that coincided with LH pulses. The volleys were very like those in rhesus monkeys: they lasted between one and four minutes, and occurred every twenty minutes, in accord with the higher frequency of LH pulses in rats.¹⁰

Knobil had thought it likely that the volleys came from GnRH neurons themselves; if not from the cell bodies, then from bundles of their axons. The volleys had a shape reminiscent of the bursts in vasopressin cells, and the mechanisms underlying those bursts were well understood. Did similar mechanisms underlie what Knobil called the “LH pulse generator”?

Many neurons generate bursts of spikes, including dopamine and NPY cells of the arcuate nucleus and mitral cells of the olfactory bulb; and so do many endocrine cells, including corticotrophs and gonadotrophs of the anterior pituitary and insulin-producing cells of the pancreas. In these different cells bursts are generated by different mechanisms, and the bursts have distinctive temporal features. The first bursting neuron to be studied extensively was neuron R15, a neuron in the abdominal ganglion of the mollusc Aplysia californica that is important for egg-laying.¹¹ R15 displays parabolic bursting, its bursts wax and wane, and bursts can develop without any synaptic input. Its membrane potential oscillates between a depolarized “up state” and a hyperpolarized “down state” because of a calcium channel that is activated when R15 is depolarized and inactivated when the intracellular calcium concentration is high. Bursts of spikes “ride” on the up state, and synaptic input modulates their intensity.¹²

Do GnRH neurons generate bursts? The ability to record their electrical activity was advanced by transgenic mouse lines in which the GnRH neurons had been engineered to express a fluorescent “reporter.” In brain slices, kept alive in oxygenated medium, the GnRH neurons fluoresce, and microelectrodes can be introduced into them to study their properties. A few GnRH neurons showed regular parabolic bursting: the bursts were shorter and more frequent than in R15, but, as in R15, they waxed and waned. However, while most neurons showed some irregular bursting, parabolic bursting was rare. None of the observed bursting conspicuously helped to understand LH pulses: the bursts were too frequent and mostly contained just a few spikes.

Other evidence came from immortalized GnRH cells (GT1 cells).¹³ These were established by introducing, into mice, a transgene containing the promotor region of the GnRH gene coupled to the coding region of a tumor-promoting oncogene. This produced a tumor of the anterior hypothalamus, and the cells were dissociated and cloned. Some of these cell lines produced lots of GnRH, and GT1 cells secreted GnRH in regular pulses. Many of the genes that these tumor cells express are not the same as in “normal” GnRH neurons, but it was intriguing that these cells could generate pulses. In one of the first electrophysiological studies of GT1 cells, Charles and Hales saw that many of them showed regular rises in intracellular calcium at intervals that varied between 3 and 120 seconds in different cells, but the rises happened at different times in different cells.¹⁴ Some cells showed occasional bursts of four or five spikes, but most spiked apparently randomly. However, longer recordings indicated that, although the bursts were not synchronous, the mean activity level of the cells fluctuated with a period of about 20 minutes, close to the period of GnRH pulses measured from similar cultures. These episodes seemed to arise from weak, spike-dependent signals.¹⁵ Thus GT1 cells had some ability to generate regular rhythmic behavior, although the calcium oscillations could not explain GnRH pulses.

In 1999, Ei Terasawa and colleagues at the University of Wisconsin cultured neurons from the olfactory placode of rhesus monkey embryos, and saw that they secreted GnRH in pulses every hour or so—similar to the period of LH pulses in monkeys.^16,17 Many of the cultured GnRH neurons also showed oscillations in intracellular calcium but with a much shorter period, of about 8 minutes. This was longer than that of the calcium oscillations in GT1 cells but still too short to explain GnRH pulses. However, the oscillations were intermittently synchronized: every hour or so, many of the GnRH neurons showed synchronous rises of calcium. Similar reports followed from GnRH neurons cultured from embryos of mice, rats, and sheep.¹⁸

The calcium oscillations in GT1 cells had seemed to be only weakly linked with spike activity, and the same seemed to be true of the GnRH neurons from monkey embryos. Abe and Terasawa found that most fired very irregularly, and the only bursts were rather unimpressive, comprising clusters of at most ten spikes.¹⁹ These are in marked contrast to the intense milk-ejection bursts of oxytocin cells, which typically contain more than a hundred spikes. Moreover, the timing of the bursts in GnRH neurons appeared to be random. In just 2 of 20 neurons was there a regular rhythmic pattern of bursts, but even these had at most seven spikes in each burst.

In many cells, oscillations in calcium arise not from spike-triggered calcium entry but from intracellular signaling pathways.²⁰ In embryonic mouse GnRH neurons at early stages of culture, calcium oscillations depend on spike activity: they are blocked by tetrodotoxin, which blocks the sodium channels on which spikes depend.²¹ In older cultures, tetrodotoxin does not completely block the oscillations; these now involve mobilization of intracellular calcium. Perhaps only some calcium oscillations reflect spiking activity, and only these trigger the release of signals that affect other GnRH cells. It’s not easy to tell: it is possible to measure intracellular calcium simultaneously in many neurons in culture, but not possible to measure spiking activity in many neurons simultaneously.

Nevertheless, these studies suggested that GnRH neurons interact in a way that leads to intermittent synchronization. GnRH cell bodies are not clustered together, but this does not mean that they do not contact each other. Rachel Campbell in Otago filled mouse GnRH neurons with a dye, biocytin, and discovered that the neurons are not anatomically isolated: they have long dendrites that come into close contact with each other,²² and where the dendrites of two neurons intertwined, afferent axons often made synapses with both neurons.²³ Campbell speculated that these “shared synapses” might be important for synchronizing the electrical activity.

This seemed to put the responsibility for coordinating GnRH neurons onto neurons that were the origin of the shared synapses. Did the close contacts between GnRH neurons allow some signals to pass between them? The contacts are not synapses, nor do they involve gap junctions that might mediate electrical coupling. Could there be any chemically mediated interactions? Magnocellular oxytocin neurons also have dendrites that are bundled together. In nonlactating rats, these dendrites are sheathed in glial processes that insulate each from its neighbors, and the processes retract during lactation, bringing the dendrites close together. Accompanying this retraction, shared synapses appear, as in GnRH neurons.²⁴ However, these are mostly inhibitory, and while their presence means that there is some common input to adjacent neurons, that might still be only a very small part of the input. What is more important for oxytocin cells is that the dendritic contacts facilitate chemical interactions: the dendrites contain many vesicles, some of which are released during suckling, and the oxytocin that is released acts on oxytocin receptors expressed by oxytocin cells.

Both GT1 cells and embryonic GnRH neurons express GnRH receptors, and in these and in adult GnRH neurons GnRH is excitatory,²⁵ but the dendrites have few large vesicles. Thus, although it seems that GnRH neurons must intercommunicate, how much of this is direct and how much is mediated by other neurons is unclear. Linked to this is the issue of how the bursts of activity that underlie pulsatile GnRH secretion arise—whether GnRH neurons generate these bursts themselves, or respond to a bursting input from another cell population.

Neuroscientists prefer to study neurons in vitro or in cell cultures rather than in living animals, in part from ethical concerns, in part to take advantage of simpler preparations and the opportunities for more precise control of experimental conditions. However, there are problems with interpreting electrophysiological recordings made in vitro. In brain slice preparations, axons and dendrites are severed, so most synaptic inputs are lost, and this can radically alter the behavior of neurons. Unfortunately, the scattered distribution of GnRH cells has made it almost impossible to study them in living animals. The only recordings so far, made by Stephanie Constantin,²⁶ were in conditions incompatible with simultaneously measuring LH. Those recordings showed that GnRH neurons are very heterogeneous, but they revealed little in the way of behavior that could be related either to the multiunit activity recorded by Knobil and others, or to the activity of GT1 neurons, or to that of GnRH neurons in vitro.

The best that electrophysiological studies in vitro can do is display the repertoire of mechanisms available to GnRH neurons that might contribute to their behavior in the living animal. However, the repertoire is vast, and different elements of it are available to different GnRH neurons. This is not unique to GnRH neurons. Even though we might be studying cells of a tightly defined population isolated from many factors that, in their normal environments, cause differences in their behavior, we still see great heterogeneity. Sometimes, we see more heterogeneity when we isolate neurons, because cells adapt to their environment. Neurons that receive more or less of a particular chemical signal typically respond by down-regulating or up-regulating expression of receptors for that chemical. Neurons that receive a stronger or weaker excitatory drive often alter their expression of ion channels to restore a “normal” level of excitability. When we isolate neurons, we see differences that in part reflect adaptations to their previous individual histories.

If bursts are important, there might be many different mechanistic pathways to that bursting. They don’t necessarily arise from intrinsic properties: growth hormone is secreted from the anterior pituitary in large pulses every few hours. These pulses arise because the somatotrophs of the anterior pituitary are alternately stimulated by GHRH from the arcuate nucleus and inhibited by somatostatin from the periventricular nucleus: the somatostatin neurons are activated by delayed effects of growth hormone secreted into the blood, and they inhibit the GHRH neurons as well as inhibiting growth hormone release. There are many means by which bursts can be generated and many more by which pulses arise, and often, it seems, bursts and pulses are generated by multiple, overlying mechanisms, involving both intrinsic properties and network mechanisms.

In any complex biological system subject to natural selection, many mutations will have accumulated over the long and erratic course of evolution, a course that takes no straight path from sufficiency to excellence, but which weaves through an ever-changing landscape. Mutations that impair a behavior that at some place in that landscape is critical may be eliminated, but any that duplicate or mimic some existing facet are likely to persist. This type of redundancy—better known as degeneracy—arises spontaneously and inevitably, and it supports the robustness of biological systems. To the researcher, it presents difficulties. It’s hard to design experiments to test the importance of a mechanism when neurons can adapt to its failure.

Degeneracy in biological systems can also lead naturally to differences between species. Perhaps the only “law” in biology is that there are (almost) always exceptions. In horses, it is possible to sample the blood that enters the pituitary gland from the hypothalamus by introducing a cannula into the facial vein and pushing it into the cavernous sinus that lies above the pituitary gland. Samples taken this way show, as expected, that pulses of LH secretion are invariably preceded by GnRH pulses. However, in mares there is no preovulatory LH surge, only a modest increase in the frequency of LH pulses—but this is enough to trigger ovulation, given the dramatic increase in LH receptor expression in the dominant follicle. The increase in LH pulse frequency is not a stimulatory effect of estrogen on LH secretion; in the mare, estrogen levels begin to fall before ovulation, not as a result of it, and it is this that leads to the progressive acceleration of GnRH and LH secretion.²⁷

What is currently known about the properties of the GnRH neurons in the mammalian hypothalamus seems of little help in understanding the generation of pulses. Their observed electrical activity has at best a tenuous relationship with the pulse-generator volleys. GnRH neurons are heterogeneous, and are not tightly coupled together. Some signals probably pass among them; those signals might include GnRH, but GnRH signaling seems not to be essential. However, looking for the explanation of pulses in the properties of GnRH neurons was always a long shot. These neurons are embedded in a network of many different cell types. Nearby GABA neurons appeared to be important, as did distant noradrenergic inputs from the caudal brainstem, but GABA and noradrenaline seem to be important for everything and nothing, depending on how you look. Further progress in understanding how the hypothalamus regulates LH secretion required some new discovery: that came, as described in the next chapter, with the discovery of a new peptide—kisspeptin.

Notes

1.  Clough, Arthur Hugh (1888) Poems of Arthur Hugh Clough. Macmillan.

2.  Casoni F, Malone SA, Belle M, Luzzati F, Collier F, Allet C, Hrabovszky E et al. (2016) Development of the neurons controlling fertility in humans: new insights from 3D imaging and transparent fetal brains. Development 143:3969–3981.

3.  Forni PE, Wray S (2015) GnRH, anosmia and hypogonadotropic hypogonadism: where are we? Frontiers in Neuroendocrinology 36:165–177.

4.  Kallmann F, Schoenfeld WA, Barrera SE (1944) The genetic aspects of primary eunuchoidism. American Journal of Mental Deficiency 48:203–236.

5.  Halász B, Pupp L (1965) Hormone secretion of the anterior pituitary gland after physical interruption of all nervous pathways to the hypophysiotrophic area. Endocrinology 77:553–562.

6.  Knobil E (1989) The electrophysiology of the GnRH pulse generator in the rhesus monkey. Journal of Steroid Biochemistry 33:669–671.

7.  O’Byrne K, Knobil E (1993) Electrophysiological approaches to gonadotrophin releasing hormone pulse generator activity in the rhesus monkey. Human Reproduction 8 Suppl 2:37–40.

8.  Tanaka T, Mori Y, Hoshino K (1992) Hypothalamic GnRH pulse generator activity during the estradiol-induced LH surge in ovariectomized goats. Neuroendocrinology 56:641–645.

9.  Nishihara M, Sano A, Kimura F (1994) Cessation of the electrical activity of gonadotropin-releasing hormone pulse generator during the steroid-induced surge of luteinizing hormone in the rat. Neuroendocrinology 59:513–519.

10.  Goubillon ML, Strutton PH, O’Byrne KT, Thalabard JC, Coen CW (1999) Ketamine-induced general anesthesia is compatible with gonadotropin-releasing hormone pulse generator activity in gonadectomized rats: prospects for detailed electrophysiological studies in vivo. Brain Research 841:197–201.

11.  Rinzel J, Lee YS (1987) Dissection of a model for neuronal parabolic bursting. Journal of Mathematical Biology 25:653–675.

12.  Canavier CC, Clark JW, Byrne JH (1991) Simulation of the bursting activity of neuron R15 in Aplysia: role of ionic currents, calcium balance, and modulatory transmitters. Journal of Neurophysiology 66:2107–2124.

13.  Weiner RI, Wetsel W, Goldsmith P, Martinez de la Escalera G, Windle J, Padula C, Choi A, Negro-Vilar A, Mellon P (1992) Gonadotropin-releasing hormone neuronal cell lines. Frontiers in Neuroendocrinology 13:95–119.

14.  Charles AC, Hales TG (1995) Mechanisms of spontaneous calcium oscillations and action potentials in immortalized hypothalamic (GT1–7) neurons. Journal of Neurophysiology 73:56–64.

15.  Nunemaker CS, DeFazio RA, Geusz ME, Herzog ED, Pitts GR, Moenter SM (2001) Long-term recordings of networks of immortalized GnRH neurons reveal episodic patterns of electrical activity. Journal of Neurophysiology 86:86–93.

16.  Terasawa E, Schanhofer WK, Keen KL, Luchansky L (1999) Intracellular Ca²+ oscillations in luteinizing hormone-releasing hormone neurons derived from the embryonic olfactory placode of the rhesus monkey. Journal of Neuroscience 19:5898–5909.

17.  Terasawa E, Keen KL, Mogi K, Claude P (1999) Pulsatile release of luteinizing hormone-releasing hormone (LHRH) in cultured LHRH neurons derived from the embryonic olfactory placode of the rhesus monkey. Endocrinology 140:1432–1441.

18.  Constantin S, Caraty A, Wray S, Duittoz AH (2009) Development of gonadotropin-releasing hormone-1 secretion in mouse nasal explants. Endocrinology 150:3221–3227.

19.  Abe H, Terasawa E (2005) Firing pattern and rapid modulation of activity by estrogen in primate luteinizing hormone releasing hormone-1 neurons. Endocrinology 146:4312–4320.

20.  Tsien RW, Tsien RY (1990) Calcium channels, stores, and oscillations. Annual Reviews of Cell Biology 6:715–760.

21.  Constantin S, Klenke U, Wray S (2010) The calcium oscillator of GnRH-1 neurons is developmentally regulated. Endocrinology 151: 3863–3873.

22.  Campbell RE, Han SK, Herbison AE (2005) Biocytin filling of adult gonadotropin-releasing hormone neurons in situ reveals extensive, spiny, dendritic processes. Endocrinology 146:1163–1169.

23.  Campbell RE, Gaidamaka G, Han SK, Herbison AE (2009) Dendro-dendritic bundling and shared synapses between gonadotropin-releasing hormone neurons. Proceedings of the National Academy of Sciences USA 106:10835–10840.

24.  Theodosis DT, Chapman DB, Montagnese C, Poulain DA, Morris JF (1986) Structural plasticity in the hypothalamic supraoptic nucleus at lactation affects oxytocin-, but not vasopressin-secreting neurones. Neuroscience 17:661–678.

25.  Han SK, Lee K, Bhattarai JP, Herbison AE (2010) Gonadotrophin-releasing hormone (GnRH) exerts stimulatory effects on GnRH neurons in intact adult male and female mice. Journal of Neuroendocrinology 22:188–195.

26.  Constantin S, Iremonger KJ, Herbison AE (2013) In vivo recordings of GnRH neuron firing reveal heterogeneity and dependence upon GABA_A receptor signaling. Journal of Neuroscience 33:9394–9401.

27.  Duittoz A, Cognié J, Decourt C, Derouin F, Forestier A, Lecomte F, Bouakkaz A, Reigner F (2018) The horse: an unexpected animal model for (unexpected) neuroendocrinology. In Ludwig M and Levkowitz G (eds) Masterclasses in Neuroendocrinology. Wiley. In press.