Neuroestrogen, Rapid Action of Estradiol, and GnRH Neurons

Abstract

Estradiol plays a pivotal role in the control of GnRH neuronal function, hence female reproduction. A series of recent studies in our laboratory indicate that rapid excitatory actions of estradiol directly modify GnRH neuronal activity in primate GnRH neurons through GPR30 and STX-sensitive receptors. Similar rapid direct actions of estradiol through estrogen receptor beta are also described in mouse GnRH neurons. In this review, we propose two novel hypotheses as a possible physiological role of estradiol in primates. First, while ovarian estradiol initiates the preovulatory GnRH surge through interneurons expressing estrogen receptor alpha, rapid direct membrane-initiated action of estradiol may play a role in sustaining GnRH surge release for many hours. Second, locally produced neuroestrogens may contribute to pulsatile GnRH release. Either way, estradiol synthesized in interneurons in the hypothalamus may play a significant role in the control of the GnRH surge and/or pulsatility of GnRH release.

Introduction

The classical endocrine effects of estradiol (E₂) have a long history of study (141). A gradual secretion of E₂ from the ovary into the general circulation reaches a variety of cells, binds to nuclear estrogen receptors (ERs), and causes genomic changes over the time course of several hours to days. One typical example is modification of the activity of neurons in the hypothalamus and gonadotrophs in the pituitary gland forming the negative and positive feedback loops of the reproductive cycle. In addition to these classical long-lasting effects of E₂ in feedback mechanisms, investigation of rapid (or acute) actions of E₂ on the uterus, vasculature, and neurons also has a relatively long history (65,72,140,155).

As early as 1971, E₂ production in the hypothalamus by aromatization of androstendione (100) has been reported. However, despite over 40 years of research on acute/rapid E₂ action and its synthesis in the brain, the progress of research in this area has been slow. Just recently, exciting features on the local synthesis of E₂ and the role of rapid action of E₂ in neuronal function have emerged: Locally synthesized E₂ may serve as a neurotransmitter in the bird brain and rat hippocampus (7,57) and E₂ appears to prevent cell death due to acute hypoxia in rodent stroke models (83,95,157).

E₂ action in the brain is quite complex, as E₂ is synthesized and released from the ovary as well as neurons in the brain. E₂ released from the ovary is transported into the brain through the general circulation and causes a long-lasting genomic action in the brain, such as generation of the preovulatory GnRH surge and lordosis behavior. E₂ released from the ovary also induces rapid action in the brain, as a single E₂ injection results in phospho-CREB expression in GnRH neurons within 15 min (3). In contrast, as discussed in section 2, locally synthesized E₂ appears to play a role in neurotransmission. Theoretically, locally synthesized estradiol in the brain may cause genomic action, but presently little direct evidence is available to support this possibility.

Since the discovery of the GnRH molecule (128) and the subsequent identification of GnRH neurons in the preoptic area (POA) and hypothalamus (11,74,131), mechanisms of E₂ action on GnRH neurons have remained elusive. It has been believed for many years that E₂ modifies the activity of GnRH neurons indirectly through interneurons. This is due to an initial study showing that GnRH neurons do not express estrogen receptors (ERs) using combined autoradiography and immunocytochemistry (133), whereas interneurons that synthesize norepinephrine, dopamine, glutamate, GABA, neuropeptide Y (NPY), and kisspeptin do (38,43,81,82,94,127, also reviewed by 49). However, the availability of more sensitive techniques for detection of ERs and the discovery of ERβ (77) have led to additional avenues for the rapid direct action of E₂ in GnRH neurons. Co-localization of ERβ and GnRH has been reported in the POA-hypothalamus of several species (51,58-60,132,134) and novel membrane bound ERs, including the G-protein coupled ER (GPR30, 40), ER-X (149), and a membrane ER sensitive to the diphenylacrylamide compound STX (STX-R, 115) have been described. In fact, direct rapid E₂ action on GnRH neurons has now been reported by several groups including our own (1-3,19,22,103,123,139,142). In this article, we will 1) briefly review recent progress on neurosynthesis of steroids in the brain focusing on neuroestrogen and their potential role as a neurotransmitter, 2) present data of rapid E₂ action on the GnRH neuronal system, highlighting our findings in the nonhuman primate, and 3) discuss potential roles for locally synthesized E₂ in GnRH neuronal function, including the preovulatory surge and pulsatile GnRH release.

1. Synthesis of estradiol in the brain

Discovery of steroid synthesis in the hypothalamus began in the early 1970s. Conversion of progesterone to 5-pregnane-3,20-dione in rat hypothalamic tissue (18,63) and aromatization of androstendione to E₂ in the hypothalamus (101) were both reported within similar time frames. However, despite consistent reports on the presence of neuroestrogens (35,92,130,138), the role of neuroestrogens in brain function has not been studied until recently. This is quite a contrast to that of 3-hydroxy-D5-compounds, such as pregnenolone (PREG) and dehydroepiandrosterone (DHEA), their sulfates, and reduced metabolites such as the tetrahydroderivative of progesterone 3a-hydroxy-5a-pregnane-20-one (3a,5a-TH PROG). The 3-hydroxy-D5-compounds are one of the most prominent allosteric modulators of chloride channels in GABA_A receptors (12,96) and therefore, they became synonymous with the term “Neurosteroids”.

Entering the 21^st century, a hypothesis that locally produced E₂ in the brain modulates neuronal function as a neurotransmitter or neuromodulator has been proposed (7,126). This hypothesis is based on the rapid timing of E₂ synthesis in the brain, the rapid action of E₂ inducing sex behavior in quails and rats (27-30), and the presence of aromatase in the presynaptic boutons in song birds (109).

Aromatase is expressed in many different brain regions, including the hypothalamus, in most vertebrate species (14,100,124). Not only is aromatase expressed in various regions of the brain in multiple species, but all of the enzymes necessary to synthesize E₂ de novo are also expressed in specific brain regions (54,55,84,87-90,152). Aromatase is expressed not only as a cluster of neurons but also as scattered individual neurons in the hypothalamus (62,100,129), high vocal centers, and the caudomedial nidopallium of the bird brain (109), and in the rat hippocampus (55,122). Importantly, aromatase is found in neuronal cell bodies and neuroterminals (100,109,122,129). In zebra finches, 50% of aromatase expressed in the brain is at neuroterminals and the expression and activity of aromatase at the nerve terminal is differentially regulated from that in other cellular compartments (122).

E₂ synthesis in some brain regions is independent of E₂ synthesis in the gonads, whereas other brain regions are dependent upon E₂ from a gonadal origin (see further discussion on this issue in section 4). For example, aromatase activity in the male monkey ventromedial hypothalamic nucleus (VMH), cortical amygdala, and basal nucleus of the stria terminalis (BNST) are unaffected by castration, while aromatase activity in the POA, anterior hypothalamus, and medial amygdala are abrogated (124). In addition to genomic seasonal regulation of aromatase activity in birds and fish (45,48,86,108,111,137,150), evidence suggests that aromatase activity is also rapidly regulated within a time frame of a couple of minutes (13,26,28). In the quail hypothalamus, aromatase activity is inhibited by depolarization with high K⁺ as well as by treatment with the glutamate receptor agonists NMDA, AMPA, and kainate within 5 minutes and this rapid inhibition appears to require phosphorylation of the aromatase enzyme (7-10,17). Additionally, in response to acute stress, aromatase activity is rapidly increased within 15 min in the male quail hypothalamus (32). Collectively, these aromatase studies indicate possible rapid changes in de novo synthesis of E₂ in the brain.

Indeed, in the past 10 years, the concentration of E₂ in the brain has been directly assessed by RIA, ELISA, and liquid chromatography with single or tandem mass spectrometry (LS/MS or LC/MS/MS analysis). As summarized in Table 1, to date, five groups have reported E₂ levels of homogenized rat brain tissue samples, spent/conditioned culture media from organotypic or dispersed cell cultures of the rat hippocampus, and microdialysate samples obtained from the bird cerebral cortex. Reviewing the published data, we have first noticed that there are distinct differences in E₂ concentration due to differences in species, sampling method, or detection methodology. In fact, Hojo et al. (56) highlights methodological differences between LC/MS/MS and RIA for assessment of E₂ measurements, showing that there are ~10 fold differences in the levels of E₂ detected in brain tissues (LC/MS/MS values are higher than RIA values). However, within the data obtained from individual labs, E₂ concentration differs among sex, age, and brain region studied (4,73): E₂ concentrations in the hypothalamus and hippocampus are highest at late embryonic ages and decrease after birth on postnatal day 4 (P4) to adulthood levels in rats of both sexes (4,54,73). Importantly, E₂ concentrations in the homogenized samples from various brain regions are in the high picomolar (pM) to low nanomolar (nM) range (4,55,64,73), about 10 to 100-fold higher than E₂ concentrations in plasma of male rats (56,64). Apparently, E₂ concentrations equivalent to the levels at the proestrous morning to afternoon are found in spent media of cultured hippocampal neurons (39,76), indicating that locally synthesized E₂ is released into extracellular space. Strikingly, E₂ release from cultured hippocampal neurons is enhanced by GnRH challenge (112).

Table 1.

Estradiol (E₂) levels in the brain and plasma

Publications	Specie s	Brain Regions	Age	Conditions or Challenges	Male E2 (nM)*	Femal e E2	Sex unknow n E2	Sample Collection	E2 Assay
Kawato et al., 2002 Group A	Rat	Hippocampus Hippocampus	Adult	Baseline NMDA (100 μM)	0.67 1.35	- -	- -	Brain slices treated then homogenized tiss	sRIA
		Plasma	Adult	Plasma	0.10	-	-
Prange-Kiel et al., 2003 Group B	Rat	Hippocampus Hippocampus	Adult, 8 div	Control Letrozole (1 nM)	- -	0.11 0.02	- -	Conditioned medium from primary culture	RIA
Kretz et al., 2004 Group B	Rat	Hippocampus Hippocampus Hippocampus Hippocampus	P4-7, 8 div	Control Letrozole (10 pM) Letrozole (1 nM) Letrozole (100 nM)	- - - - -	- - - - -	0.09 0.06 0.03 0.03	Conditioned medium from brain slice cultur	RIA
		Hippocampus Hippocampus	P4-7, 8 div	Control Letrozole (100 nM)	- -	- -	0.07 0.06	Conditioned medium from primary culture
Hojo et al., 2004 Group A	Rat	Hippocampus Hippocampus	Adult	Baseline NMDA (100 μM)	0.59 1.27	- -	- -	Brain slices treated then homogenized tissues	RIA
Amateau et al., 2004 Group C	Rat	Frontal cortex Hippocampus Hypothalamus POA Cerebellum Brainstem	P 2h	Intact	6.55 6.26 5.79 5.40 4.39 3.35	4.25 6.37 3.20 3.74 3.60 3.09	- - - - - -	Homogenized tissues	RIA
		Frontal cortex Hippocampus Hypothalamus POA Cerebellum Brainstem	P 32h	Intact	4.71 3.74 5.11 2.84 2.91 2.77	4.07 5.25 2.09 1.73 2.59 2.77	- - - - - -
Fester et al., 2006 Group B	Rat	Hippocampus Hippocampus	P5, 7 div	Control Letrozole (100 nM)	- -	- -	0.32 0.21	Conditioned medium from primary culture	RIA
		Hippocampus Hippocampus	P5, 7 div	Control siRNA StAR siRNA	- -	- -	0.26 0.20
Remage-Healey et al., Group D	Bird	NCM NCM NCM NCM	Adult	Baseline Female present Male song Glutamate (10 mM)	1.34** 2.57** 2.20** 0.73**	- - - -	- - - -	Microdialysis	ELISA
		Plasma Plasma Plasma	Adult	Baseline Female present Male song	0.16 0.21 0.15	- - -	- - -
Prange-Kiel et al., 2008 Group B	Rat	Hippocampus Hippocampus	P5, 8 div	Control GnRH (10 nM)	- -	- -	0.73 0.88	Conditioned medium from brain slice culture	RIA
		Hippocampus Hippocampus Hippocampus	P5, 8 div	Control GnRH (10 nM) GnRH (100 nM)	- - -	- - -	0.73 1.69 2.50	Conditioned medium from primary culture
Fester et al., 2009 Group B	Rat	Hippocampus Hippocampus Hippocampus Hippocampus Hippocampus	P5, 7 div	Control Letrozole (100 nM) Testosterone (100 nM) Cholesterol (100 nM) Chol (100nM) +Letr (100 nM)	- - - - -	- - - - -	0.07 0.04 0.31 0.64 0.51	Conditioned medium from primary culture	RIA
Munetsuna et al., 2009 Group A	Rat	Hippocampus Hippocampus	Adult	Pair housed Single housed	8.27 12.38	- -	- -	Homogenized tissues	EIA
Hojo et al., 2009 Group A	Rat	Hippocampus Hippocampus	Adult	Intact Castrated	8.44 6.98	- -	- -	Brain slices treated then homogenized tissu	LeCs/MS/MS
		Plasma Plasma	Adult	Intact Castrated	0.01 0.01	- -	- -
Konkle and McCarthy, Group C	Rat	Hypothalamus Hypothalamus Hypothalamus	E19 E21 P2-4 P60	Intact	14.39 5.40 7.20 3.60	19.79 1.08 7.20 1.80	- - - -	Homogenized tissues	RIA
		Hippocampus Hippocampus Hippocampus	E19 E21 P2-4 P60	Intact	1.73 0.40 0.54 0.05	1.44 0.36 0.45 0.05	- - - -
		POA	P 2h	Intact	9.90	1.13	-
		Hypothalamus Hypothalamus	P3 P3	Intact ADX/GDX	9.00 11.69	8.64 7.20	- -
		Hippocampus Hippocampus	P3 P3	Intact ADX/GDX	0.40 0.40	0.36 0.36	- -
		Plasma Plasma	P3 P3	Intact ADX/GDX	0.09 0.12	0.07 0.09	- -
Remage-Healey et al., Group D	Bird	NCM NCM NCM NCM	Adult	Baseline Male song Female visual stimuli Male visual stimuli	1.70** - 3.06** 1.59**	1.50* 2.32* 1.71* 1.47*	- - - -	Microdialysis	ELISA
		Plasma Plasma	Adult	Baseline Male Song	- -	0.12 0.11	- -

Group A=Kawato lab; Group B=Rune lab; Group C=McCarthy lab; Group D=Schlinger and his former student’s lab

POA: preoptic area, NCM: caudomedial nidopallium

E: embryonic day, P: postnatal day, div: days in vitro

ADX/GDX: adrenalectomized/gonadectomized

RIA: radio immuno assay, ELISA: enzyme-linked immunosorbent assay, EIA: enzyme immunoassay, LC/MS/MS: liquid chromatagraphy/ tandem mass spectrometry

^*E₂ concentrations converted to nM using the estimation that 1 ml≈ 1 g wet weight and 1 nM≈ 0.1 fmol/mg wet weight≈ 10 fmol/mg protein (see Hojo et al., 2004 sup

In an earlier study, aromatase activity in some brain regions, such the VMH, amygdala, and BNST, is not affected by castration (124). A recent direct E₂ measurement study in neonatal rats further indicates that E₂ levels in the hypothalamus are independent of circulating gonadal steroids, as E₂ levels in the hypothalamus are approximately 100-fold higher than circulating E₂ and developmental changes in E₂ concentrations (reduction in E₂ occurs between P0 and P2) are unaffected by neonatal gonadectomy/adrenalectomy in both sexes (73). However, it is also possible that a small portion of E₂ synthesis in these brain areas is attributable to the conversion of circulating testosterone. In fact, in the neonatal rat hippocampus castration reduced E₂ levels by 17% (56), although this small decrease by castration again suggests that the majority of E₂ synthesis (nM concentrations) in the hippocampus is independent of circulating gonadal steroids. Collectively, locally synthesized E₂ in the brain at high pM to low nM levels appears to be maintained regardless of circulating steroid levels, although there may be subtle modifications by peripheral steroids depending on brain region, sex, or age. Nonetheless, presently, we do not have any information regarding 1) whether the total E₂ concentration in the hypothalamus is important or 2) whether the source of the E₂ (peripheral vs. central) is important for control of GnRH release. While to date abundant data show that peripheral E₂ modifies neuronal activity in the hypothalamus, the concept of neuroestrogen is newly born. Therefore, coordinated interactions caused between central and peripheral E₂ remain to be investigated.

An important question arises as to whether E₂ synthesized in the brain influences circulating E₂ levels. The result that neonatal gonadectomy/adrenalectomy in male and female rats does not alter plasma E₂ levels (73) support this possibility, at least during an early developmental stage. However, the answer to this question requires further investigations, as the authors of that study (73) did not measure the effects of gonadectomy/adrenalectomy on plasma testosterone and neonatal castration of male rats is well known to eliminate sexual differentiation of the brain,

Emerging evidence suggests that synthesis of neuroestrogens is also regulated by neurotransmitters and neuromodulators. In rat hippocampal slices, application of NMDA results in an increase in the production of E₂ within 30 min in vitro (55,64). E₂ concentration in the rat hippocampus also increases in response to stress (99). Exposure of male songbirds to females increases E₂ levels in the forebrain cortex within 30 min in vivo without changes in testosterone levels (120). Infusion of glutamate inhibits E₂ neurosynthesis, whereas infusion of GABA stimulates testosterone neurosynthesis in the forebrain cortex of birds within 30 min in vivo (120). These time resolutions can be refined, whenever a more sensitive method for detecting E₂ becomes available. Nonetheless, these results indicate that E₂ synthesis occurs rapidly when neurons receive excitatory and inhibitory signals from other neurons, such as glutamatergic and/or GABAergic input.

2. E₂ induces a rapid excitatory action on primate GnRH neurons

There are several models to examine the effects of E₂ on GnRH neurons (145). Among them we have been using cultured GnRH neurons derived from the nasal placode of monkeys at embryonic age (E) 35-37, which are obtained from time-mated pregnancies. Because transgenic monkeys with GFP-labeled GnRH neurons are not yet available, this approach is quite useful for studying the cellular physiology of GnRH neurons in primates. Earlier, we have demonstrated that 1) cultured GnRH neurons from monkeys contain almost no non-GnRH neurons or glia, as they are not present in the nasal placode at this specific developmental stage (146), 2) cultured GnRH neurons undergo maturational changes in vitro (79,146), and 3) placode derived GnRH neurons are functional, as transplantation of fetal placode into the infundibular recess of the third ventricle of adult female monkeys, in which the GnRH final common pathway has been lesioned, restores cyclic ovulation (125).

We first examined the effects of E₂ on firing activity. Using cell-attached patch clamp recording we found that application of E₂ (1 nM) to cultured primate GnRH neurons induces a ~250% increase in action potential firing frequency within a minute (Figure 1A). E₂ also increases the number of action potentials per burst and burst duration. However, E₂ does not change the timing of bursts (interburst interval) nor the cluster pattern in primate GnRH neurons (2). Similar rapid stimulatory E₂ effects on action potential firing rate have also been reported in GFP-labeled GnRH neurons of ovariectomized mice: E₂ (100 pM-100 nM) directly enhances the firing rate of GnRH neurons in a dose responsive manner, whereas 10 pM E₂ indirectly reduces firing activity of GnRH neurons via suppressing the excitatory GABA neurotransmission (22). The authors of that study state that the E₂ effects started within 5 min and were completed between 10 and 15 min after the initiation of E₂ treatment (22).

Figure 1.

Application of E₂ induces excitatory firing activity (A), intracellular calcium, [Ca²⁺]_i, oscillations (B) and GnRH release (C) in cultured primate GnRH neurons. E₂ (1nM) was applied to GnRH neurons for 10 or 20 min, as indicated by the black bar (A) and (B) and gray bar (C). Note that in all cases, excitatory effects are induced within 10 min. Gray bars (B) indicate synchronization induced by the E₂ treatment. Modified from 1,2,103, permission pending.

Next, we examined the effects of E₂ on [Ca²⁺]_i oscillations (1,67,69,103). Exposure of GnRH neurons to 1 nM E₂ for 10 min causes an increase in the frequency of [Ca²⁺]_i oscillations starting during the E₂ application and lasting for 40-50 min (Figure 1B). E₂ also increases the number of activated GnRH neurons. The E₂ effects on [Ca²⁺]_i oscillations are dose dependent, as 10 nM E₂ induces [Ca²⁺]_i oscillations with a higher frequency and a longer period as well as a higher percentage of activated cells (69). In addition, E₂ increases the number of synchronized [Ca²⁺]_i oscillations/hour from ~1 event/hour to ~2.7 events/hour after the initiation of E₂ treatment (1), although the frequency of synchronized [Ca²⁺]_i oscillations is not dose dependent (69). Importantly, the E₂-induced increase in [Ca²⁺]_i oscillations is not blocked by TTX (1), indicating that E₂ causes rapid action directly on GnRH neurons (see more discussion in section 4). Additionally, treatment with the membrane impermeable E₂, E₂-BSA, and the nuclear impermeable estrogen dendrimer conjugate (EDC) also increases the frequency of [Ca²⁺]_i oscillations, suggesting that E₂-induced [Ca²⁺]_i oscillations are a membrane initiated event (1,103). Similar stimulatory effects of E₂, including E₂-BSA conjugates, on [Ca²⁺]_i oscillations have been reported in cultured mouse GnRH neurons. However, because these neurons are exposed to E₂ for 30 min before recording (142,143) the latency of the response is unclear. Another study of [Ca²⁺]_i changes in Pericam expressing mouse GnRH neurons shows that at minimum 15 min is required for direct stimulatory E₂ action as well as indirect inhibitory E₂ action transsynaptically mediated through GABA neurons (123). The data in Pericam expressing mouse GnRH neurons indicate that the latency in murine GnRH neurons appears to be longer than that in primate GnRH neurons.

Henceforth, as E₂ induces a rapid excitatory action in primate GnRH neurons, E₂ should also stimulate GnRH release. Indeed, exposure of cultures to 1 nM E₂ for 20 min results in a rapid increase of GnRH peptide release (103). Again, the increase occurs within 10 min of E₂ application and lasts for 40 min (Figure 1C). Moreover, the plasma membrane impermeable E₂-BSA and the nuclear membrane impermeable form of E₂, EDC, both stimulated GnRH release within 10 min, although the duration and amplitude is smaller than E₂ alone (103). A rapid release of GnRH from mouse cultured or GFP-labeled GnRH neurons, comparable to ours, has not been reported. Although the suppressed frequency of GnRH release with a 4-hour treatment of E₂ (17 pM) in GT1-7 cells (102) has been reported, a 4-hour treatment period is difficult to categorize as a “rapid” E₂ action, because most “rapid” E₂ actions occur within minutes. In addition, E₂ application to rat median eminence explants rapidly stimulates GnRH release (33), suggesting that E₂ can cause excitatory action at the GnRH neuroterminals and this observation from nearly 30 years ago is consistent with our observation. Importantly, we recently found that this rapid stimulatory E₂ action on GnRH release was seen in vivo, when we directly applied E₂ to the MBH in both ovarian intact or ovariectomized adult female monkeys (66).

3. Rapid E₂ action is mediated through multiple membrane receptors

The consequence of E₂ action is dependent upon several factors including dosage, timing, spatial aspects, and receptor subtypes or signaling molecules involved. A fundamental question to understanding rapid direct E₂ action on GnRH neurons is which ER(s) mediate(s) E₂ effects. The most commonly studied ERs are ERα and ERβ. Initially, we too expected to find a role of ERα and/or ERβ in the rapid E₂ action in primate GnRH neurons. Surprisingly, however, the ER inhibitor, ICI182,780, fails to block the E₂-induced increase in [Ca²⁺]_i oscillations and GnRH release (1,103), indicating that neither ERα, nor ERβ, is involved. In a follow-up study, we transfected primate GnRH neurons with siRNA specific to human ERα or ERβ and tested the effects of E₂. Again, exposure to siRNA for ERα or ERβ fails to block the E₂-induced [Ca²⁺]_i oscillations (67), confirming our previous results. By contrast, in mouse GnRH neurons ERβ appears to mediate the majority of rapid direct E₂ effects (3,19,22,139,142) or alternatively, indirect E₂ effects are mediated through ERα involving presynaptic GABA inputs (19,22,123). Although Sun et al. (139) report that in mouse GnRH neurons E₂ causes a minor direct effect through GPR30, in general, receptors involved in E₂ action in mouse GnRH neurons are significantly different from those in primate GnRH neurons.

To date three nonclassical membrane ERs have been proposed: ER-X (149), GPR30 (148), and STX-R (115). We first investigated the role of GPR30 in rapid E₂ action in primate GnRH neurons. To our surprise, treatment of the GPR30 agonist G1 at 10 nM, but not 1 nM, stimulates [Ca²⁺]_i oscillations similar to E₂ (103). Additionally, GPR30 knockdown with siRNA blocks both E₂- and EDC-induced [Ca²⁺]_i oscillations (103). Interestingly, a high dose of ICI182,780 (1 μM) alone elicits an increase in [Ca²⁺]_i oscillations (103). It has been shown that a high dose (1 μM) of ICI182,780 is an agonist for GPR30 in cancer cells (40). Finally GPR30 is expressed in a subset of adult GnRH neurons in the monkey hypothalamus (~30%, 103).

There are two reasons to believe that the rapid excitatory E₂ action in primate GnRH neurons is mediated by more than one receptor subtype. First, 10 nM E₂ effects on [Ca²⁺]_i oscillations are larger than those of 1 nM E₂ and GPR30 siRNA reduces but does not completely block 10 nM E 2+ 2 effects on [Ca ]_i oscillations (69). Second, a higher dose of G1 is required to elicit a response similar to E₂ (103). Because STX has been shown to elicit changes in hypothalamic neurons through a phospholipase C (PLC) mechanism in mutant mice lacking ERα, ERβ, ERα/ERβ, and GPR30 (115-117), we examined the role of STX-R in 10 nM E₂ action in primate GnRH neurons. To our surprise, STX (10 nM) treatment of primate GnRH neurons elicits changes in [Ca²⁺ ]_i oscillations, similar to those with 1 nM E₂ and STX (10 or 100 nM) treatment also stimulates GnRH release in a dose dependent manner (69). Moreover, GPR30 siRNA transfection of GnRH neurons fails to block effects of STX, whereas treatment with ICI182,780, an antagonist for STX-R (115), blocks STX-induced [Ca²⁺]_i oscillations, suggesting that E₂ action through STX-R in primate GnRH neurons is independent of GPR30 (69). Therefore, multiple receptor mechanisms are clearly involved in mediating rapid E₂ action on primate GnRH neurons (Figure 2).

Figure 2.

Schematic illustration showing two modes of rapid E₂ action in GnRH neurons (middle) with two synapses (Top and Bottom). Assuming E₂ is released at the synaptic cleft, there are direct (top) and indirect (bottom) rapid E₂ actions. Top: E₂ released at the synaptic cleft directly activates GPR30 and STX-R in the primate hypothalamus (and ERβ in the rodent hypothalamus) and modulates activity of GnRH neurons. Exposure of primate GnRH neurons to E₂ rapidly induces [Ca²⁺]_i oscillations and GnRH peptide release (vesicles in blue color) within 10 min (69,103). Two possible mechanisms for the rapid E₂ action through GPR30 and STX-R are discussed in this article, although many details are yet to be clarified, which are noted by question marks in the scheme. First, E₂ binding to GPR30 may induce activation of two intracellular pathways: 1) E₂ activation through GPR30 depolarizes the GnRH neuronal membrane via VGCCs (139), which allows [Ca²⁺]_e entry, resulting in CICR (68) and 2) E₂ transactivates AC and/or ErbB pathways (41), which also results in CICR. Second, E₂ binding to STX-R appears to cause 1) activation of CICR through a PLC and IP3-R mechanism leading to a [Ca²⁺]_i increase (69) and 2) activation of a PKCδ-AC-PKA mechanism resulting in hyperpolarization of the GnRH neuronal membrane through KATP channels (159), which are essential for burst firing of GnRH neurons, hence neurosecretion. Direct E₂ action through ERß has also been reported in mice GnRH neurons (19,139). Bottom: Rapid action of E₂ may be indirect, transsynaptically mediated by other neural input, as shown by Romano et al. (123). In this case, E₂ rapidly stimulates GABA release (vesicles in black color), which causes IPSPs in mouse GnRH neurons, resulting in suppression of GnRH release. Alternatively, it is possible that E₂ rapidly stimulates excitatory interneurons (vesicles in magenta color), which would cause EPSPs in GnRH neurons, resulting in stimulation of GnRH release. Abbreviations: AC: adenylyl cyclase; Ca²⁺: calcium; [Ca²⁺]_e: extracellular Ca²⁺; [Ca²⁺]_i: intracellular Ca²⁺; CamKII: Ca²⁺ calmodulin kinase II; CICR: calcium induced calcium release; DAG: diacylglycerol; E₂: estradiol; E.R.: endoplasmic reticulum; ErbB: epidermal growth factor receptor; ERK1/2: Extracellular signal regulated kinase 1 and 2; GnRH: gonadotropin-releasing hormone neuron; GPR30: G protein coupled estrogen receptor; IP3: inositol triphosphate; IP3-R: inositol triphosphate receptor; KATP: ATP sensitive potassium channel; mGluR: metabotropic glutamate receptor; PIP2: phosphatidylinositol biphosphate; PKA: protein kinase A; PKCδ: Protein kinase C delta; PLC: phospholipase C; RyR: ryanodine receptor; SERCA: Sarco/endoplasmic reticulum Ca²⁺ ATPase; STX-R: membrane estrogen receptor sensitive to STX; VGCC: voltage gated calcium channel.

4. The role of rapid E₂ action in the mechanism of GnRH release

A “rapid” timing of E₂ action is a membrane initiated phenomenon, rather than “long term” E₂ action, which requires nuclear transcription after E₂ binding to ERs. Then a question arises as to what is the physiological significance of rapid E₂ action in the hypothalamus, and more specifically within the GnRH system? It has been proposed that locally synthesized E₂ contributes to acute synaptic formation in the hippocampus, as E₂ increases spine density and enhances long-term potentiation (LTP) and long-term depression (LTD) in hippocampal neurons within 30 min (42,91,98), whereas spine density decreases by treatment with an aromatase inhibitor (76). Whether there is a similar function of E₂ in GnRH neurons remains unknown.

Is there any role of rapid E₂ action in the negative and positive feedback control of GnRH release? In ovariectomized female rhesus monkeys, injection of E₂ induces suppression of LH/GnRH release with a latency of 2-4 hours (the negative feedback phase, 20,53,97,156), followed by stimulation of LH/GnRH release with a latency of 24-36 hours, lasting for 36-48 hours in a positive feedback phase (53,80,156). Because in primate GnRH neurons E₂ induces membrane initiated excitatory, not inhibitory, action within 10 min, it is unlikely that E₂ is involved in the negative feedback mechanism. Is it then involved in the positive feedback mechanism? We will discuss this further in section 5.

Alternatively, is there any possible role in GnRH pulse generation? As discussed earlier, our observations with [Ca²⁺]_i dynamics in vitro consistently show that E₂ is a potent frequency modulator of GnRH neurons. Before we present our view, however, we need to discuss several issues.

Species differences

There are clear species differences in the preovulatory and E₂-induced GnRH and gonadotropin surges. As discussed in a recent review by Plant (110), in rodents the preovulatory GnRH surge is controlled by the AVPV and circadian signals, whereas in highly evolved primates (old world monkeys and humans) the medial basal hypothalamus is sufficient for cyclic ovulations and the preovulatory LH surge is independent of circadian signals. Moreover, the male rodent brain is sterilized by the perinatal elevation of estrogens aromatized from androgens and thus E₂ is not able to induce a surge in castrated males. In contrast, the capacity of the male primate hypothalamus for the preovulatory LH surge remains, as the GnRH neuronal system is not sterilized by prenatal/perinatal gonadal steroids (104). Finally, while a neuronal signal for GnRH/LH surges in rodents is limited to the pentobarbital sensitive critical period of 2h (36,37) and surges last for 4-6 hours (136), in primates there is no critical period for the GnRH/LH surges and surges last for over 36-48 hours (53,80,153,156). Importantly, as discussed above, direct rapid excitatory E₂ effects on GnRH neurons in non-human primates are mediated by GPR30 and STX-R, whereas in mice ERβ appears to be responsible (19). It is also important to point out the fact that the promoter region of the GnRH gene in humans contain an ERE, whereas the rodent GnRH gene does not (118), indicating that genomic E₂ actions through classical ER in rodent GnRH neurons are likely through interneurons. In primates, however, genomic E₂ actions through classical ERs in GnRH neurons can occur both directly or indirectly.

Concentration

Concern has been cast regarding the doses of 1-100 nM E₂ used in in vitro studies examining the effects of E₂ on GnRH neurons (50). In our studies in non-human primates the doses at 1-10 nM E₂ were routinely used (1,2,67,69,103) and in electrophysiological studies with acute brain slices or embryonic GnRH neurons in rodents the doses of E₂ as high as 100-1000 nM were used (22,139,142). Certainly, these doses are higher than circulating E₂ levels during the ovulatory cycle, when E₂ levels fluctuate from pM to low nM (0.1-0.2 nM in the follicular phase and 0.7-1.4 nM at the preovulatory surge) in female monkeys (106,107) and 20 pM in diestrus and 200 pM at the preovulatory surge in female mice and rats (136, also see 25). However, as discussed in section 1 and Table 1, the adult rat hypothalamus, hippocampus, and other brain regions of both sexes contain E₂ at high pM to low nM levels and our preliminary data indicate similar E₂ levels in the female rhesus monkey hypothalamus as well as in microdialysate samples obtained from the stalk-median eminence in vivo (B.P. Kenealy and E. Terasawa, unpublished data). Therefore, E₂ at 1-10 nM in the hypothalamus is not likely “supraphysiological”. How about 100-1000 nM used in rodent electrophysiology and culture studies? Peak concentration of E₂ in the synaptic cleft could reach 100 nM or even higher, when E₂ is released as a neurotransmitter. In fact, it has been shown that calculated peak concentrations of the neurotransmitters glutamate and GABA are much higher than those in tissue concentrations measured by a microdialysis method. That is, in the synaptic cleft, glutamate and GABA concentrations reach 1.1 mM (24) and 30-100 mM (47,61), respectively, whereas concentrations of glutamate and GABA in the hippocampus and striatum measured with in vivo microdialysis methods are 1-4 μM (6,85,105) and 25 nM in hippocampal brain slice preparation (52), respectively.

Receptors

The classical positive and negative feedback effects of E₂ are mediated by mechanisms through ERα, requiring nuclear transcription. First, E₂ fails to induce LH surges in ERα knockout mice as well as in mutant mice lacking estrogen response element (ERE)-dependent ERα signaling (21,46). By contrast, in ERβ knockout mice E₂ induces LH surges (154), although these mice had ERβ splice variants (75), therefore, reexamination of the role of ERβ in positive feedback in mice with complete elimination of ERβ splice variants (5) is still needed. Second, whereas convincing evidence for the presence of ERα in GnRH neurons has not been shown, the presence of ERα in interneurons that innervate GnRH neurons, such as kisspeptin, have been consistently reported (43,94,135). Taken together, while it is possible that membrane ERα signaling may be, in part, responsible for negative feedback effects of E₂ on LH release, as E₂ can suppress LH levels in mice lacking ERE-dependent ERα signaling (46), ERα expression in interneurons, such as kisspeptin neurons (94) that directly innervate GnRH neurons, is indispensable for LH/GnRH surges. Little is known about ERs involved in the feedback actions of E₂ in primates. Nonetheless, this does not preclude the role of rapid E₂ action in the positive feedback mechanism, as discussed in section 5.

Role of non-neuronal cells

Because our GnRH cultures contain non-neuronal cells (146) and they are involved in the propagation of the [Ca²⁺]_i wave of the GnRH neuronal network (121), it is possible that E₂ action to GnRH neurons are also mediated through non-neuronal cells. In fact, E₂ rapidly induces a [Ca²⁺]_i release from astrocytes, stimulates progesterone synthesis within 5 min in astrocytes of rat hypothalamus which are dependent upon both nonclassical and classical receptors (16,78), and modifies tanycyte morphology (70,113). E₂ also rapidly alters function of epithelial cells in the median eminence (31,114). Moreover, E₂ stimulates release of prostaglandin E₂ (PGE₂) from astroglia (107), which directly depolarizes the membrane of GnRH neurons (23). Although we have previously shown that TTX does not block the E₂-induced rapid changes in [Ca²⁺]_i oscillations (1), this observation does not exclude input from non-neuronal cells. Thus, we need to address this issue in the future.

Time domain

It is important to point out that the negative feedback effect of E₂ on LH/GnRH release precedes its positive feedback effect (144,156). Also, as we have already discussed, the membrane-initiated E₂ action occurs within min, whereas E₂ induced genomic changes require at least several hours to days. Based on this major difference in the time domain of the two modes of E₂ action, it is clear that the positive feedback effects of E₂ are initiated by “genomic action” of E₂. However, this does not preclude the possible role of the membrane-initiated rapid excitatory E₂ action in augmenting the preovulatory GnRH surge. To extend this issue, while the rapid E₂ action on GnRH release in both in vitro and in vivo are “brief” (5-20 min exposure of E₂ inducing a brief GnRH increase), E₂ increases from the ovary or by peripheral administration last several hours to days resulting in suppression (negative feedback) for at least several hours and stimulation (positive feedback) for 24-48 hours of GnRH release (Figure 3A).

Figure 3.

Schematic illustration showing the effect of ovarian estradiol (E₂) on GnRH release (A) and possible roles of neuroestrogens in control of GnRH surge (B) and/or pulsatility of GnRH release (C) in primates. A. E₂ from the ovary suppresses GnRH release (negative feedback phase) with a latency of a couple of hours lasting 24-48 hours. Subsequently, E₂ facilitates GnRH release (positive feedback phase) with a latency of 24-48 hours lasting 36-48 hours. B. A possible role of E₂ synthesized in the MBH in the preovulatory GnRH surge. An increase in preovulatory E₂ from the ovary causes action in three types of interneurons to GnRH neurons in the hypothalamus: (1) inhibitory interneurons, such as GABA neurons, (2) excitatory interneurons, such as kisspeptin neurons, and (3) E₂ synthesizing neurons. Exposure of inhibitory interneurons (1) to E₂ inhibits GnRH release (4), initially through a membrane-initiated E₂ mechanism and subsequently through a genomic E₂ mechanism. Over 24 hours after E₂ exposure, activity of excitatory interneurons is stimulated by a genomic action of E₂ (2), leading to the initial increase in GnRH release into the portal circulation (5) as well as at the synaptic junction between GnRH neurons and E₂-synthesizing interneurons (6). Although it is possible that neurotransmitter E₂ released from E₂-synthesizing interneurons starts to stimulate GnRH release via the membrane-initiated mechanism, synaptic GnRH release may also feedback to stimulate neurotransmitter E₂ release from E₂-synthesizing interneurons (7). This positive feedback loop between GnRH neurons and E₂-synthesizing interneurons sustains GnRH release to the portal circulation (8). Although this figure illustrates input of interneurons (e.g., GABA and kisspeptin expressing neurons and E₂ synthesizing neurons), to GnRH cell bodies, interaction between GnRH neurons and interneurons could also occur at the neuroterminals. C. A possible role of E₂ synthesized in the MBH in pulsatility of GnRH release in primates. E₂ synthesized in the MBH may modulate GnRH neurons directly or indirectly through KNDy (kisspeptin, neurokinin B, ß-dynorphine) neurons and/or glia.

5. Two possible hypotheses

As discussed in above, we propose two hypotheses for a physiological role of rapid E₂ action: Hypothesis 1 is a potential role for local E₂ that augments the preovulatory GnRH surge. Hypothesis 2 is that local E₂ contribute to GnRH pulse frequency. At this time, evidence for both hypotheses is circumstantial and direct evidence needs to be shown.

Augmentation of the preovulatory GnRH surge (Figure 3B)

During the preovulatory phase, ovarian E₂ binds interneurons, such as kisspeptin neurons and GABA neurons, in the hypothalamus/ POA, which express ERα (and also ERβ to a certain extent). At the negative feedback phase, the suppression of GnRH release is caused by interneurons. In fact, involvement of both membrane initiated and genomic E₂ action through interneurons during the negative E₂ feedback phase has been reported (46,94). In the initial phase, inhibitory interneurons, such as GABA neurons, may rapidly reduce the activity of GnRH neurons via increasing inhibitory postsynaptic potentials (IPSP, 123) resulting in reduction of GnRH release. E₂ activation of nuclear transcription of GABA neurons leads to an increase in GABAergic activity resulting in a subsequent decrease of GnRH release. At the positive feedback phase, excitatory interneurons, such as kisspeptin neurons, are stimulated by genomic action of ovarian E₂ resulting in facilitation of GnRH neuronal activity and an increase in GnRH release.

Ovarian E₂ may also stimulate E₂ synthesis in interneurons, the phenotypes of which are yet to be determined. Evidence for this stimulation of synthesis is based on E₂ effects on aromatase expression. For example, the promoter region of the aromatase gene contains an ERE and aromatase expression is up-regulated in an in vitro culture system (158). In addition, aromatase expression in the rat BST and medial amygdala decreases after castration with adrenalectomy, while treatment with E₂ reverses this decrease (160). Assuming that an E₂-induced increase in E₂ production leads to elevated E₂ release from the neuroterminal/ synaptic junction between an aromatase expressing interneuron and GnRH neuron, the released E₂ could rapidly stimulate GnRH release through GPR30 and/or STX-R in primates and ERβ and/or STX-R in mice. Finally, if the GnRH neuron makes a synaptic connection with the aromatase expressing interneuron, GnRH release at synapses could stimulate E₂ release. Because in the hippocampus GnRH stimulates E₂ release (112), it is not difficult to speculate that a similar GnRH stimulation of E₂ release occurs in the hypothalamus. This hypothetical perpetuating positive feedback loop within the hypothalamus would be quite advantageous in the primate brain, as the preovulatory GnRH surge needs to be sustained for 24-48 hours until the pituitary and ovary properly respond for successful ovulation. Importantly, the termination of the GnRH surge would also be prompt due to the rapid nature of E₂ action. Currently, the mechanism of GnRH surge termination is unknown.

GnRH pulse frequency modulation (Figure 3C)

Considering our knowledge that the frequency modulation of GnRH pulses is essential for proper maintenance of the female reproductive cycle (71), we can extend our hypothesis that local E₂ may modulate pulsatility of GnRH release in a subtle manner. This hypothesis is derived from our observations that E₂ modulates frequency of [Ca²⁺]_i oscillations and synchronization of [Ca²⁺]_i oscillations in GnRH neurons in vitro (1,69,103). It is also known that E₂ is a potent frequency modulator of LH/GnRH release. Although presently, the mechanism of GnRH pulse generation is unclear, coordinated periodical release of GnRH is regarded as synchronized activity among GnRH neurons. In fact, GnRH neurons appear to communicate through dendro-dendritic interactions (15) and recently, we have shown that GnRH is released from the cell body and dendrites (44). Local E₂ could cause its effect through KNDy neurons, kisspeptin, neurokinin B, and ß-dynorphin co-expressing neurons, in the ARC (34,119), a concept that has been proposed as the source of GnRH pulse-generation (93,151). Either way (directly or indirectly), it is possible that locally synthesized E₂ may contribute to the frequency modulation of pulsatile GnRH release in vivo.

Conclusion

Estradiol causes direct excitatory action within min in GnRH neurons. This rapid E₂ action is a membrane-initiated phenomenon mediated by non-classical receptors, GPR30 and STX-sensitive receptors in primate GnRH neurons. Surprisingly, preliminary data suggest that the hypothalamus of female monkeys release high pM to low nM levels of E₂, which is significantly higher than circulating E₂ levels and infusion of E₂ at a similar concentration into the medial basal hypothalamus results in stimulation of GnRH release in vivo (66). Based on the time course and stimulatory nature, this rapid action of E₂ is not likely involved in the negative feedback phase of the ovulatory cycle. Consequently, we propose two hypotheses that the rapid membrane-initiated E₂ action augments and sustains the positive feedback phase of GnRH release or pulsatility of GnRH release. Specifically, ovarian E₂ initiates the preovulatory GnRH surge through indirect stimulation of interneurons, but neuroestrogens in the hypothalamus augment and sustain GnRH release for a prolonged period by a positive feedback loop between GnRH neurons and E₂ synthesizing neurons in the hypothalamus. Alternatively, neuroestrogens may be profoundly involved in GnRH pulse generation. Certainly, these hypotheses are quite provocative. Nonetheless, if future studies validate these hypotheses, we can make a significant advancement in the field of Reproductive Neuroendocrinology.

Highlights.

* Estradiol induces a rapid excitatory action in primate GnRH neurons

* The rapid estradiol action is mediated by non-classical estrogen receptors

* Possible role of neuroestrogens in control of GnRH release is discussed

* Historical perspective of neuroestrogen in the hypothalamus is reviewed