Classical and Membrane-Initiated Estrogen Signaling in an In Vitro Model of Anterior Hypothalamic Kisspeptin Neurons

Abstract

The neuropeptide kisspeptin is essential for sexual maturation and reproductive function. In particular, kisspeptin-expressing neurons in the anterior rostral periventricular area of the third ventricle are generally recognized as mediators of estrogen positive feedback for the surge release of LH, which stimulates ovulation. Estradiol induces kisspeptin expression in the neurons of the rostral periventricular area of the third ventricle but suppresses kisspeptin expression in neurons of the arcuate nucleus that regulate estrogen-negative feedback. To focus on the intracellular signaling and response to estradiol underlying positive feedback, we used mHypoA51 cells, an immortalized line of kisspeptin neurons derived from adult female mouse hypothalamus. mHypoA51 neurons express estrogen receptor (ER)-α, classical progesterone receptor (PR), and kisspeptin, all key elements of estrogen-positive feedback. As with kisspeptin neurons in vivo, 17β-estradiol (E2) induced kisspeptin and PR in mHypoA51s. The ERα agonist, 1,3,5-Tris(4-hydroxyphenyl)-4-propyl-1H-pyrazole, produced similar increases in expression, indicating that these events were mediated by ERα. However, E2-induced PR up-regulation required an intracellular ER, whereas kisspeptin expression was stimulated through a membrane ER activated by E2 coupled to BSA. These data suggest that anterior hypothalamic kisspeptin neurons integrate both membrane-initiated and classical nuclear estrogen signaling to up-regulate kisspeptin and PR, which are essential for the LH surge.

Kisspeptin is a neuropeptide inextricably linked to reproductive function across multiple species. The kisspeptin gene encodes a large species-specific precursor of approximately 140 amino acids from which shorter signaling peptides are derived. These biologically active kisspeptin peptides are 10–54 amino acids in length and are highly conserved across species (reviewed in reference ¹). All of the shorter, amidated peptides are biologically active and show similar affinity for kisspeptin receptor in vitro [(2, 3); see reference ⁴ for review], likely because amino acids 6 and 10 of the decapeptide, common to all of these fragments, are critical for binding to its receptor (5). Kisspeptins bind to G protein-coupled receptor 54 (GPR54), a G_q protein-coupled receptor, to elicit an excitatory intracellular signaling cascade in GnRH neurons (eg, reference ⁶). Indeed, GPR54 activation in GnRH neurons specifically is required for fertility in rodents (7, 8). Overall, the evidence of a role for kisspeptin in reproduction is overwhelming, although the specifics of this role continue to be uncovered. Humans with mutations in genes encoding kisspeptin or GPR54 fail to acquire secondary sex characteristics and exhibit low serum gonadotropin levels (9, 10). Other mammals with disruptions in kisspeptin signaling are also infertile or subfertile (eg, references 8 and 11).

Mounting evidence supports a role for kisspeptin in the estrogen modulation of gonadotropin release. Kisspeptin neurons in the arcuate nucleus of the hypothalamus (ARH) have been shown to play an important role in estrogen-mediated pulsatile or tonic release of GnRH/LH, referred to as estrogen-negative feedback (12–14). This negative feedback predominates much of the estrous cycle. However, just prior to ovulation, effects of estrogens on GnRH and gonadotropin release become stimulatory. Hypothalamic kisspeptin neurons in the anterior rostral periventricular area of the third ventricle (RP3V) are generally accepted as mediators of estrogen positive feedback regulating the LH surge. In the RP3V, estradiol (E2) up-regulates kisspeptin, as opposed to the suppressive effects E2 has in the ARH (eg, references 15 and 16). Because kisspeptin is the most potent stimulator of GnRH neurons (17, 18), kisspeptin up-regulation is consistent with a stimulatory influence on the GPR54-expressing GnRH neurons. GnRH released into the portal circuit stimulates a hypophyseal surge release of gonadotropins, LH and FSH, preceding ovulation. Estrogen-positive feedback depends on estrogen receptor (ER)-α (19), but GnRH neurons lack ERα expression, necessitating that another population of cells transduce the estradiol signal (reference ²⁰ but also see references 21 and 22). An overwhelming majority of kisspeptin neurons in the RP3V express ERα (>90%; see references 15 and 23), which makes this population the most likely to receive estrogenic information and transmit it to GnRH neurons through the release of kisspeptin.

Although the roles of the two kisspeptin populations appear to be somewhat characterized in terms of negative (ARH) vs positive estrogen feedback (RP3V), the nature of the shift from negative to positive estrogen feedback preceding ovulation remains uncharacterized. The preovulatory rise in circulating E2 is an important component of this shift, and therefore, E2 has been the focus of many feedback studies. Progesterone also participates in the neural control of ovulation. More recently it has become apparent that local (hypothalamic) synthesis of progesterone is critical for the LH surge (24); however, the cellular target of progesterone action is unknown. Progesterone receptor (PR) message is increased after E2 treatment in the anterior hypothalamus (25, 26), although the specific cell types in which PR is induced are not well defined. We hypothesize that estrogen-positive feedback is achieved through E2 induction of kisspeptin and PR in kisspeptin neurons. Classical PR, which is critical for fertility (27), is expressed in most RP3V kisspeptin neurons (15, 23, 28). We propose that RP3V kisspeptin neurons integrate signals from ovarian steroids and neuroprogesterone to modulate the activity of GnRH neurons.

Here we characterize mHypoA51 cells and use them as a model of RP3V kisspeptin neurons to study responses to E2 and progesterone. This minimizes confounding external influences that, although important, complicate interpretations of RP3V kisspeptin neuronal E2 responses in in vivo experiments. We report that mHypoA51s are sexually mature female hypothalamic neurons that have the characteristics of postpubertal RP3V kisspeptin neurons, expressing ERα, PR, and steroid-regulated kisspeptin.

Materials and Methods

mHypoA51 cultures

mHypoA51 cells (CELLutions Biosystems) are immortalized neurons isolated from adult, female C57BL/6 mouse hypothalamus. Immortalization was achieved using retroviral transfer of Simian virus-40 T-antigen and treatment with ciliary neurotrophic factor to induce neuronal proliferation (29). These cells robustly express neuron-specific enolase (29) and contain Fox-3 mRNA (encoding NeuN), indicating that they are indeed mature neurons. mHypoA51 neurons were thawed from a frozen stock and maintained in high-glucose DMEM media (number 11960; GIBCO/Invitrogen) with 10% qualified fetal bovine serum (FBS; number 26140-079; Invitrogen), and 1% penicillin/streptomycin (number 15140122; Invitrogen). Cells were passaged at 70%–90% confluence and were not used after the 10th passage.

Antibodies

All antibodies used for immunocytochemistry and Western blotting have been extensively characterized in previous publications (Table 1). Immunolabeling with the antibody directed toward kisspeptin was precluded following preadsorption with kisspeptin-10 but not prolactin-releasing peptide (30). In vivo, this antibody has been shown to label cell bodies and fibers in appropriate brain regions consistent with the distribution of kisspeptin-expressing cells (14), and in the current study, the intracellular pattern of staining (ie, cytosolic, negative nuclei) supports labeling of an intracellular signaling peptide. The antibody used for detection of ERα was originally characterized in 1990 (31) and has since been characterized further (eg, references 32 and 33). Staining in mHypoA51 neurons exhibited a distinct nuclear pattern, as expected for ERα labeling. The antibody directed toward PR has been used in many publications and has been well characterized. Blocking with the immunizing peptide prevented immunostaining and visualization of PR-A and PR-B bands in Western blots [see supplemental data reported elsewhere (34)]. In our studies, clear nuclear labeling was observed using this antibody. In addition, some mHypoA51s had cytosolic staining (see Figure 1), which appears to indicate extranuclear receptors (but see references 35 and 36). Only cells with distinct nuclear labeling were counted as PR positive because the C-19 antiserum has cross-reactivity with cyctoskeletal elements (eg, actin). Expression of PR was verified by PCR and Western blotting.

Table 1.

Antibody Table

Peptide/Protein Target	Antigen Sequence (if Known)	Name of Antibody	Manufacturer, Catalog Number, and/or Name of Individual Providing the Antibody	Species Raised (Monoclonal or Polyclonal)	Dilution Used (WB; ICC)	Reference
PR	Epitope mapping at the C terminus of PR of human origin	C-19	Santa Cruz, sc-538	Rabbit; polyclonal	1:200; 1:5000	(34, 71)
ERα	Linear peptide corresponding to the C terminus of rat ERα	C1335	Millipore, 06-385	Rabbit; polyclonal	1:1000; 1:1000	(31, 33)
Kisspeptin	Mouse kisspeptin-10: YNWNSFGLRY	AB9754	Millipore, AB9754	Rabbit; polyclonal	1:500; 1:1000	(30)
GAPDH	Rabbit glyceraldehyde-3-phosphate dehydrogenase	6C5	Millipore, MAB374	Mouse; monoclonal	1:10 000; N/A	(72)

Abbreviations: ICC, immunocytochemistry; N/A, not available; WB, Western blot.

Figure 1.

mHypoA51 neurons express kisspeptin, ERα, and PR. A, Image of representative PCR amplification products electrophoresed on a 2% agarose gel. One band for each amplicon was observed at expected sizes: 120 bp (kisspeptin, PR, and PR-B), 130 bp (ERα), 150 bp (GAPDH), 302 bp (Tac2/NKB). B, Representative images of immunocytochemistry using primary antibodies raised against kisspeptin (top row), ERα (middle row), or classical PR (bottom row) in mHypoA51 neurons. Cells labeled using the antibody directed toward kisspeptin yielded a cytosolic pattern of staining, whereas ERα- and PR-labeled cells exhibited nuclear staining. Some cytosolic staining was visible in PR-labeled neurons (arrow), indicating either extranuclear receptors or cytoskeletal labeling. Cells without nuclear staining were not counted as PR positive. Scale bar in lower right image, 100 μm (and applies to all images). C, Kisspeptin immunoreactivity was observed in nearly all mHypoA51 neurons (817 of 821 cells, 99.51%). ERα-ir was observed in 69% of cells (1354 of 1962 cells), and nuclear PR-ir was observed in 62% of mHypoA51 cells (1765 of 2808).

Immunocytochemistry

mHypoA51 neurons were plated onto poly-D-lysine-coated chambered slides (Millicell EZ slides; EMD Millipore). Approximately 1000 cells were plated in each of the eight chambers for each slide. After the cell attachment and differentiation, cells were fixed with 4% paraformaldehyde and rinsed three times in PBS. Slides were blocked in 1% BSA and 2% normal goat serum (NGS) in PBS for 60 minutes. Primary antibodies were diluted in PBS with 1% NGS and applied to cells for a 48-hour incubation at 4°C. Cells were rinsed twice in PBS, followed by two rinses in Tris-buffered saline (TBS) and a 30-minute incubation in Tris sodium blocking buffer (PerkinElmer). After the TBS rinses, cells were exposed to the appropriate secondary antibody (see Table 1) at the concentration of 1:200 in TBS with 1% NGS and 0.5% Triton X-100 for 60 minutes. Cells were rinsed and exposed to streptavidin-conjugated horseradish peroxidase (1:100 in Tris sodium blocking buffer; PerkinElmer) for 30 minutes. After another set of TBS rinses, the cells were exposed to rhodamine-conjugated tyramide (1:50) for signal amplification (tyramide signal amplification kit; PerkinElmer). After a final set of TBS rinses, chambers were removed and slides were dipped in double-distilled H₂O and coverslipped using Vectashield hard-set mounting medium with 4′,6′-diamino-2-phenylindole (DAPI; Vector Laboratories).

Images were obtained with an Axioskop 2 microscope with epifluorescent illumination and an Axiocam charge-coupled device camera (Zeiss Inc) and Axiovision version 3.1 software (Zeiss Inc) were used to acquire images. Merged images were coded and counted by an investigator blinded to the experimental conditions. A grid transparency was overlaid on each image, and cells were manually counted, each being classified as ERα or PR positive (clear nuclear signal, also exhibiting DAPI stain) or negative (DAPI staining only). Any apparent staining that was not accompanied by DAPI counterstain was not counted. Percentage of mHypoA51 neurons expressing nuclear ERα or PR was calculated.

Drug and steroid treatments

mHypoA51 neurons were steroid starved for at least 1 hour prior to treatments. Steroid-starvation media consisted of the following: phenol red-free DMEM (number 17-205-CV; Corning), 5% charcoal-stripped FBS (number 100119; Gemini Bio-Products), and 1% penicillin/streptomycin with L-glutamine (number 10378016; Invitrogen). Estradiol (E2, free, number E8875; or cyclodextrin-encapsulated, number E4389; Sigma-Aldrich) was kept as a frozen stock and diluted prior to use (final concentration 1 nM). Progesterone (P4; cyclodextrin encapsulated, number P7556; Sigma-Aldrich) was dissolved in water and kept at −20°C as a frozen stock. P4 was also used at a final concentration of 1 nM. The weights of cyclodextrin-encapsulated steroids were adjusted such that final concentrations accurately reflect the steroid concentrations, excluding the cyclodextrin. 6-Keto-17β-estradiol-6-carboxymethyloxime-bovine serum albumin (E-6-BSA; number E1361; Steraloids) was dissolved in water and stored as a 100 μM stock at −20°C. Stock was used at a 1:100 dilution, for a final concentration of 1 μM E-6-BSA. One μM BSA without estrogen was used as a control for E-6-BSA assays. 1,3,5-Tris(4-hydroxyphenyl)-4-propyl-1H-pyrazole (PPT; number H6036; Sigma-Aldrich) was dissolved in dimethylsulfoxide and stored at −20°C as a 10-μM stock and used at a final concentration of 1–10 nM. 2,3-Bis(4-hydroxyphenyl)-propionitrile (DPN; number 1494; Tocris) was dissolved in EtOH and stored as a frozen stock (1 mM stock; working concentration 1 nM). All drugs and steroids were diluted in phenol red-free DMEM with charcoal-stripped FBS (see above), which also served as control medium. After the treatments, cells were harvested for either protein or RNA isolation (see below).

Western blotting

Harvested cells were lysed in radioimmunoprecipitation assay (RIPA) lysis buffer containing a protease inhibitor cocktail, 1 mM sodium orthovanadate, and 2 mM phenylmethylsulfonyl fluoride (Santa Cruz Biotechnology). Total protein concentration was assessed using the bicinchoninic assay (BCA) method (Thermo Fisher Scientific Inc) on a spectrophotometer (NanoDrop 1000; Thermo Fisher Scientific Inc). Fifteen micrograms of total protein were loaded into each lane of a 10% SDS-PAGE gel and electroblotted onto polyvinyl difluoride membranes (GE Life Sciences). Membranes were blocked in 5% (wt/vol) milk in TBS plus 0.1% Tween 20 (TBST) for 1 hour. Blots incubated in primary antibody (see Table 1) in blocking solution with gentle rocking overnight. The following day, membranes were washed in TBST and then incubated in horseradish peroxidase-conjugated secondary antibody raised in goat (antirabbit or antimouse; 1:10,000 dilution; Santa Cruz Biotechnology) in blocking solution for 1 hour. After additional rinses in TBST, blots were imaged using Clarity chemiluminescent substrate (Thermo Fisher Scientific) and a FluorChem E imager (ProteinSimple). Membranes were stripped of antibody using Restore Western blot stripping buffer (number 21059; Thermo-Fisher Scientific) for 15 minutes. The process of primary and secondary antibody incubation was repeated for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a loading control.

Western blot images were analyzed using AlphaView software (ProteinSimple). A rectangle of consistent size was fitted around each band and above each band for background analysis. The OD was then calculated by the software for both signal and background. Total signal intensity was then calculated by subtracting the pixel density of the background from that of the band of interest. This process was repeated for GAPDH. Final signal was calculated by dividing corrected signal by the corrected value for GAPDH:

$$FinalSigna{l}_{kiss}={\left(bandintensity-background\right)}_{kiss}/{\left(bandintensity-background\right)}_{GAPDH}$$

Multiple lanes per treatment were averaged for each Western blot, and four experiments were done in total. Data points falling outside of the mean ± 2 SD were excluded from analysis. This was necessary for only one result in the 24-hour treatment group, which showed an extremely large and uncharacteristic increase in kisspeptin protein. The other four 24-hour time point values were included in analysis. Groups were analyzed using SigmaPlot for Windows version 12.5 (Systat Software, Inc). Where appropriate, Student-Newman-Keuls (SNK) post hoc tests were used, with a significance level set to P = .05.

Reverse transcription-polymerase chain reaction

Total RNA was isolated using TRIzol reagent (Life Technologies), according to the manufacturer's protocol. After the treatment, media were aspirated and 1 mL of TRIzol was added to each 100-mm plate. Cells incubated in TRIzol for 5 minutes prior to a chloroform extraction of RNA. RNA pellets were washed once with 100% isopropanol and twice in 75% ethanol in diethylpyrocarbonate-treated water. Pellets were allowed to dry for 10 minutes at room temperature and were resuspended in diethylpyrocarbonate (DEPC)-treated water. Concentration and quality of RNA were assessed using a spectrophotometer (NanoDrop 1000; Thermo Fisher Scientific). cDNA was synthesized from 1–2 μg total RNA using the SuperScript III reverse transcriptase kit (number 18080-51; Invitrogen) with oligo(dT)₂₀ primers. The reverse transcription reaction was performed at 50°C for 50 minutes, followed by a 5-minute termination at 85°C. cDNA was used immediately for RT-PCR or stored at −20°C for 1 month or less.

RT-PCR primers for kisspeptin-targeted sites on each of the two exons of the kisspeptin gene (37). Amplification of genomic DNA therefore is unlikely because the product would include an intron and would be too large to efficiently amplify under our PCR conditions. Amplification of a single product was verified using gel electrophoresis (Figure 1C). Mouse PR (also called PR-A/B) primers (38) were designed to target a region downstream of the ATG initiation site common to both PR-A and PR-B gene products (nucleotides 2591–2711). PR-B primers target nucleotides 709–829, a region unique to PR-B (38) (see Table 2 for more PCR primer information).

Table 2.

PCR Primer Information

Primer, Concentration	Gene Accession Number	FW Sequence	RV Sequence	Predicted Size	Annealing Temperature, ºC	Source
Kisspeptin, 500 nM	NM_178260	TGC TGC TTC TCC TCT GT	ACC GCG ATT CCT TTT CC	132	60	(37)
PR, 200 nM	NM_008829	GGT GGG CCT TCC TAA CGA G	GAC CAC ATC AGG CTC AAT GCT	120	57	(38)
PR-B, 200 nM	NM_008829	GGT CCC CCT TGC TTG CA	CAG GAC CGA GGA AAA AGC AG	120	57	(38)
ERα, 200 nM	NM_007956	TGG GCT TAT TGA CCA ACC TAG CA	AGA ATC TCC AGC CAG GCA CAC	130	60	(73)
NKB, 200 nM	NM_009312	TCT GGA AGG ATT GCT GAA AGT G	GTA GGG AAG GGA GCC AAC AG	302	60	(41)
NeuN, 200 nM	NM_001039167	CCA GGC ACT GAG GCC AGC ACA CAG C	CTC CGT GGG GTC GGA AGG GTG G	110	62	(74)
GAPDH, 200 nM	NM_017008	GCA CAG TCA AGG CCG AGA AT	GCC TTC TCC ATG GTG GTG AA	150	57–62	(75)

Abbreviations: FW, forward; RV, reverse.

PCRs were run in duplicate, and at least three duplicates per condition were included in each PCR assay. At least two independent experiments were conducted and combined for each mRNA result. No-template controls (cDNA reverse transcribed with no RNA) were run alongside cDNA for each primer pair in every PCR experiment, and no amplification was observed in the no-template control wells. PCRs were prepared using 2 μL of cDNA in a 20-μL reaction with SYBR GreenER quantitative PCR SuperMix Universal (number 11762100; Invitrogen). Primer information is summarized in Table 2. Reactions were run on an Mx3000p thermal cycler (Agilent). Conditions for amplification were as follows: 2 minutes at 50°C (Uracil-DNA glycosylase [UDG] incubation), 10 minutes at 95°C for UDG inactivation and DNA polymerase activation, and then 40–50 cycles each consisting of 15 seconds at 95°C and 60 seconds at annealing temperature (see Table 2). Reactions were followed by melting curve analysis (55–95°C). Finally, amplified products were electrophoresed on a 2% agarose gel containing ethidium bromide (1.12 g agarose, number A9539; Sigma; 50.4 mL double-distilled H₂O; 5.6 mL 10× Tris-acetate-EDTA (TAE) buffer, number T8280; Sigma; 1 μL ethidium bromide) and visualized using the FluorChem E imager (ProteinSimple). A DNA ladder (GeneRuler^M 100 bp DNA ladder; Thermo-Fisher Scientific) was run alongside samples for verification of amplicon size. All PCR products yielded single peaks in the melting curve analysis and single bands in agarose gel electrophoresis (Figure 1C).

Standard curves for kisspeptin, PR, and GAPDH primer pairs were generated by plotting cycle threshold (Ct) for several serial dilutions of cDNA. Concentrations of cDNA used in experimental assays fell well within this range of cDNA. The slope of the best fit line was then used to calculate primer efficiencies, with the following equation: efficiency = 10 ^ (−1/slope) − 1. Efficiencies for kisspeptin, PR, and GAPDH primer sets were 96.88%, 90.94%, and 83.04%, respectively. Plotting PR Ct-GAPDH Ct (ΔCt) values for serially diluted cDNA revealed a slope of nearly zero (−0.05), indicating comparison of these two targets is valid (39).

Fold change in mRNA vs the control condition was calculated by the 2^−ΔΔCt method (39, 40). Briefly, Ct values for GAPDH were subtracted from Ct values for PR, PR-B, or kisspeptin. This value was defined as ΔCt. Next, the −ΔΔCt was calculated for each treatment condition: −ΔΔCt = − (treatment ΔCt − average control ΔCt). The mRNA fold change was calculated by raising 2 to the power of the −ΔΔCt for each treatment or time point because each cycle represents a doubling of nucleic acid material. In some cases (typically in the non-E2 treated condition), no amplification of kisspeptin was observed, giving a no Ct reading. To calculate a fold change in mRNA, a Ct of 40 was assigned to no Ct reads. Data points falling outside the mean ± 2 SD were excluded from analysis. This was necessary for no more than one data point in any of the experiments reported, and there was always a minimum of six samples per group after exclusions. Data were analyzed with a one-way ANOVA with SNK post hoc tests where appropriate. Due to E2 stimulation of kisspeptin mRNA, E-6-BSA was also hypothesized to increase kisspeptin mRNA. Therefore, a one-way t test was used to evaluate data from the E-6-BSA experiment.

Calcium imaging

Two to four days prior to imaging, mHypoA51 cells (10 000–15 000) were plated onto poly-D-lysine (0.1 mg/mL; Sigma-Aldrich)-coated, 15-mm glass coverslips in 12-well culture plates and grown in DMEM medium with 10% FBS and 1% penicillin/streptomycin at 37°C, 5% CO₂. Eighteen hours prior to imaging experiments, media were changed to phenol red-free DMEM media supplemented with 5% charcoal-stripped, dextran-treated FBS, 1% penicillin/streptomycin+L-glutamine, and sodium bicarbonate. Before imaging, cells were loaded with the calcium indicator Fluo-4 AM (2.5 μM; Life Technologies) dissolved in Hanks' balanced salt solution (HBSS); 0.02% pluronics, and 0.08% dimethylsulfoxide at 37°C for 45 minutes. Cells were washed with HBSS to remove excess dye. Glass coverslips were mounted into a 50-mm chamber insert (Warner Instruments) fixed into a 60- × 15-mm cell culture dish (Dow Corning) and placed into a QE-2 quick exchange platform (Warner Instruments) for imaging on a Zeiss LSM 710 confocal microscope using an IR-Achroplan ×40/0.80 n.a. water-immersion objective (Zeiss). Cells were gravity perfused with HBSS. Drugs or steroids were delivered by gravity perfusion and used in the following concentrations: 1 nM water-soluble E2 or P4 (see Drugs and Steroid Treatments, above), 1 nM PPT, and 10 μM RU486 (mifepristone, number M8046; Sigma). Fluo-4 AM fluorescence was excited with the 488-nm line of an argon laser, and emission signals above 505 nm were collected. Calcium (Ca²⁺) fluorescence was measured as relative fluorescence units (RFU) and was plotted to give baseline fluorescence and peak response to drug or steroid stimulation. Baseline was calculated by averaging 10 RFU readings prior to drug or steroid delivery, during HBSS perfusion. Peak values were defined as the highest RFU value obtained after the perfusion with a particular drug or steroid. Responding cells were those whose peak values were at least 2 times greater than the average baseline value prior to drug/steroid delivery.

Groups were analyzed with SigmaPlot for Windows version 12.5 (Systat Software, Inc). Where appropriate, SNK post hoc tests were used (significance set to P = .05). One-tailed t tests were used to evaluate effects of E2 or PPT on mobilization of intracellular Ca²⁺; both were hypothesized to increase Ca²⁺ release based on results that indicated presence of membrane ER. Similarly, removal of steroids was hypothesized to abrogate P4-induced mobilization of Ca²⁺ stores, and one-tailed t tests were used to evaluate these results as well.

Results

mHypoA51 neurons express kisspeptin, ERα, and PR

mHypoA51 cells displayed morphological characteristics expected for neurons, including visible nuclei and neurites. Primers targeting Fox-3 (the gene for neuronal marker NeuN) mRNA yielded a single PCR amplification product, verifying that mHypoA51 cells are indeed neuronal (data not shown). Expression of ERα, kisspeptin, and PR were also verified by PCR (Figure 1A). Single products were observed for each amplicon, verifying specificity of primer pairs for each gene product. mHypoA51 cells were also tested for presence of neurokinin B (NKB; Tac2 gene) using published primers (41). Although the presence of NKB mRNA was verified in arcuate nucleus samples, no NKB mRNA was observed in mHypoA51 neurons (Figure 1A). Nearly all mHypoA51 neurons were kisspeptin immunoreactive (817 of 821 cells, or 99.51%), as assessed by colocalization with the DAPI nuclear stain (Figure 1B). As expected, mHypoA51 neurons had robust cytoplasmic kisspeptin immunoreactivity with negatively stained nuclei. In distinction, ERα and PR immunoreactivity were localized to the nucleus in most mHypoA51 neurons. ERα-immunoreactive (-ir) nuclei were present in 69.01% of mHypoA51 neurons (1354 of 1962 cells). For PR immunoreactivity, 62.90% (1765 of 2808) of cells had nuclear staining (Figure 1, B and C). Light cytosolic PR staining was also visible in 23% of cells (denoted by arrow in Figure 1B).

E2 induces kisspeptin via activation of membrane-localized ERα

Kisspeptin protein electrophoresed at approximately 65 kDa, verified by a single Western blot band. This value was higher than expected but identical with that observed with a kisspeptin overexpression lysate run alongside samples (Figure 2A). Kisspeptin protein was significantly increased only after a 48-hour E2 treatment (202.9% ± 51.9% of control; one way ANOVA df = 2.5; F = 7.348; P < .05; Figure 2A). A 24-hour treatment with 1 nM E2 significantly up-regulated kisspeptin mRNA (2.47- ± 0.57-fold increase from control levels; one way ANOVA df = 2.21, F = 4.942; P < .05; Figure 2B). By 48 hours of E2 treatment, however, kisspeptin mRNA was no longer significantly elevated (P = .77, different from control).

Figure 2.

E2 induces kisspeptin in mHypoA51 neurons. A, Representative Western blot images of kisspeptin protein. A single band is observed at 65 kDa, visible in both untreated mHypoA51 lysates (left lane) and kisspeptin overexpression lysates (kiss+, right lane, number NBL-12305; Novus). E2 treatment induces kisspeptin protein after 48 hours of 1 nM E2 treatment. At this time point, kisspeptin protein was increased to 202.89% ± 51.9% of control levels (P < .05, greater than control and 24 h time point). B, Kisspeptin mRNA was increased by approximately 2.5-fold after 24 hours of E2 treatment. By 48 hours, mRNA returned to control levels. *, P < .05, different from control and 24 hours of E2; +, P < .05, different from control and 48 hours of E2.

To examine whether the E2-induced increase of kisspeptin mRNA was mediated by ERα, mHypoA51s were incubated with PPT for 24 hours, the peak time of kisspeptin mRNA induction by E2. PPT elicited an increase in kisspeptin mRNA similar to that observed after E2 treatment (PPT: 2.14- ± 0.36-fold increase vs control treatment; one way ANOVA df = 2.24; F = 11.500; P < .01 vs control, Figure 3A). ER agonist DPN had no significant effect on kisspeptin mRNA, although there was a trend toward a decrease (0.39- ± 0.21-fold compared with control; two tailed t test df = 9; t = 2.076; P = .0677). Stimulation with E-6-BSA (membrane impermeable estradiol construct) also significantly increased kisspeptin mRNA after 24 hours of treatment (2.49- ± 0.57-fold; one tailed t test df = 21; t = −2.025; P < .05; Figure 3B).

Figure 3.

ERα-mediated induction of kisspeptin and PR mRNAs. A, Treatment of mHypoA51 cells with ERα agonist PPT (10 nM) for 24 hours caused increases in kisspeptin and PR mRNAs equivalent to that observed with E2, suggesting activation of ERα. DPN (1 nM) treatment did not affect kisspeptin or PR mRNAs. B, Treatment with membrane-impermeable E2, E-6-BSA (1 μM), induced an increase in kisspeptin mRNA similar to that achieved with free E2 but no significant increase in PR mRNA. *, P < .05, different from control; **, P < .01, different from control.

PR is up-regulated by nuclear E2 signaling

E2 (1 nM) treatment for 24 hours increased PR (PR-AB) mRNA 3.47- ± 0.73-fold (df = 2,35; F = 3.923; P < .05; Figure 4). PR-B mRNA was increased approximately 3-fold (2.91- ± 0.43-fold; df = 2.34; F = 8.425; P < .01), indicating that the E2 up-regulation of PR-B mRNA accounted for most of the total PR message induction. After 48 hours of E2 treatment, both PR and PR-B mRNA were no longer different from control levels (P = .175 and P = .76 for PR and PR-B, respectively, vs control). In contrast to kisspeptin, neither PR nor PR-B mRNA were significantly induced by treatment with membrane-impermeable E-6-BSA [1.48- and 1.45-fold increases in mRNA for PR and PR-B, respectively, at 24 h of treatment; one way ANOVA for PR and PR-B: P = .226 (PR) and 0.282 (PR-B); Figure 3B; data shown for PR only]. The ERα agonist, PPT (10 nM), elicited an increase in PR mRNA (PPT: 2.28- ± 0.49-fold vs control; one way ANOVA; df = 2.19; F = 4.216; P < .05), similar to that observed with 1 nM E2 treatment in a parallel assay (2.31- ± 0.35-fold vs control; P < .05; Figure 3A). There was no difference between E2 and PPT treatments (P = .89). The ERβ agonist, DPN, did not affect PR mRNA (0.86- ± 0.40-fold vs control; df = 10; t = 0.509; P = .662).

Figure 4.

E2 induction of PR mRNA. A 24-hour treatment with E2 (1 nM) induced total PR (PR-AB) mRNA (3.5- ± 0.73-fold compared with controls, P < .05). PR-B mRNA increased 2.9- ± 0.43-fold (P < .01) after E2 treatment. After 48 hours of E2 treatment, PR and PR-B mRNA levels returned to control levels. *, P < .05, different from control.

Activation of membrane ERα increases intracellular free calcium concentrations ([Ca²⁺]_i)

To ascertain whether membrane-initiated E2 signaling affected intracellular Ca²⁺ levels, mHypoA51 cells were stimulated with E2, which elicited a rapid increase in [Ca²⁺]_i (baseline: 31.81 ± 2.35 RFU vs E2: 105.64 ± 7.20 RFU; one tailed t test; t = −9.755; df = 174; P < .001; Figure 5A). The selective ERα agonist, PPT (1 nM), induced a similar calcium response (baseline: 41.76 ± 1.84 RFU vs PPT: 126.10 ± 5.25RFU; one tailed t test; t = −15.172; df = 286; P < .001; Figure 5A), indicating that activation of ERα is a critical component of E2-induced increases in [Ca²⁺]_i.

Figure 5.

E2 and P4 rapidly mobilize [Ca²⁺]_i via activation of classical, membrane-localized receptors. A, Representative traces in RFUs from a single mHypoA51 neuron during calcium imaging. The cell was stimulated with E2 or PPT (1 nM), each of which elicited a significant increase in [Ca²⁺]_i (P < .001 for each). B, mHypoA51 cells are responsive to acute stimulation with P4. Steroid starvation decreases the percentage of mHypoA51 cells that respond to P4 (from 37.4% ± 5.6% to 12.4% ± 5.6%). C, Representative tracing from a cell stimulated with 1 nM P4, before and after exposure to RU486. Cells displayed a peak in RFUs after stimulation with P4. After RU486 treatment, P4 response was abrogated. D, P4 response was decreased by 91% with RU486 pretreatment, suggesting that P4 mobilization of calcium is achieved through the activation of classical PR. #, P < .001, different from control baseline, RU486 baseline, and P4+RU486.

P4 rapidly mobilizes calcium via activation of classical PR

As an assay of membrane-initiated P4 signaling in mHypoA51 neurons, Ca²⁺ imaging was performed. In response to P4 stimulation, 37.4% ± 5.6% of mHypoA51 cells rapidly increased [Ca²⁺]_i (Figure 5B). This response was attenuated when mHypoA51s were preincubated in a steroid-free media (phenol red free DMEM, charcoal stripped FBS) for 18 hours and prior to stimulation with P4. After removal of steroids, only 12.4% ± 5.6% of cells responded (one tailed t test; t = −3.029; df = 8; P < .01; Figure 5B).

Treatment of mHypoA51s with the PR antagonist RU486 (10 μM) prior to P4 stimulation blocked the [Ca²⁺]_i response (baseline: 17.9 ± 2.0 RFU; P4: 61.8 ± 7.0 RFU; RU486 + P4: 25.9 ± 2.5 RFU; two way ANOVA; F = 25.427; df = 220; P < .001 overall effect; P < .001 P4 vs baseline; P = .504 RU486 + P4 vs RU486 alone; Figure 5, C and D). RU486 itself did not affect baseline fluorescent readings (22.2 ± 2.1 RFU; P = .453).

Discussion

The major findings of these studies are that mHypoA51 neurons, immortalized kisspeptin neurons isolated from adult, female mouse hypothalamus, exhibit both direct nuclear action and membrane-initiated E2 signaling and are a model of RP3V kisspeptin neurons. We report that these cells express kisspeptin, ERα, and classical PR, similar to their in vivo counterparts. Importantly, E2 induced both PR and kisspeptin in mHypoA51 cells, a cardinal feature of RP3V neurons. PR induction required free (as opposed to membrane impermeable) E2, suggesting the involvement of nuclear signaling. In contrast, kisspeptin expression was stimulated by E-6-BSA, implicating membrane-initiated E2 signaling. Finally, activation of membrane-associated steroid receptors, ERα and PR, increased [Ca²⁺]_i. This demonstrates an important pathway through which E2 and P4 can activate kisspeptin neurons, which have been suggested to integrate and transduce steroid information to GnRH neurons.

Immunocytochemistry revealed that nearly all mHypoA51 neurons (99.51%) express kisspeptin. Approximately 70% expressed nuclear ERα, and approximately 62% expressed nuclear PR. These percentages are similar to those reported for kisspeptin neurons in vivo (15, 23, 28). In female mouse anteroventral periventricular nucleus (AVPV) kisspeptin neurons, most cells have been reported to express ERα, although this number varies greatly [64% (28) to more than 90% (15, 23)]. This variability may arise from several factors, including but not limited to complexities of the subdivision of RP3V, age and hormonal profile of animals, and ERα detection technique (immunohistochemistry vs in situ hybridization). For example, in the study by Clarkson et al (28), ERα was found in 36%-64% of RP3V kisspeptin neurons using double-label immunohistochemistry, whereas Smith et al (15) observed that nearly all kisspeptin neurons in the RP3V coexpressed ERα, using double in situ hybridization. Furthermore, Zhang et al (23) found that ERα-immunopositive kisspeptin cells decreased significantly from the 3- to 4 month to the 9- to 10 month groups, suggesting that the age of the animals influences expression levels. In any case, it seems clear that most RP3V kisspeptin neurons express ERα, findings that are mirrored by this study in mHypoA51 neurons.

Approximately two-thirds of anterior hypothalamic kisspeptin neurons (60%-67%) express PR (23, 28), a finding that is very consistent between studies: no age-related differences in PR expression were found in anterior hypothalamic kisspeptin neurons (23), which suggests that PR expression may be more stable than that of ERα. Here we report that 62% of mHypoA51 neurons express classical, nuclear PR, further evidence that mHypoA51 neurons are an appropriate model for RP3V kisspeptin neurons.

The E2-induced expression of kisspeptin that we report in mHypoA51 neurons is also consistent with in vivo studies in which increases in kisspeptin mRNA (15, 23) and immunoreactivity were measured (42). However, the time course of E2-induced kisspeptin (often 24–48 h) in the in vivo studies has been interpreted to reflect a direct, nuclear action of estradiol. We also observed increases in kisspeptin mRNA and protein after 24–48 hours of treatment, respectively, which is comparable with the timing of in vivo studies. However, our findings with membrane-impermeable E2 (E-6-BSA) indicate that, despite this time course, membrane-initiated signaling underlies E2-induction of kisspeptin.

Although kisspeptin mRNA was significantly elevated at 24 hours, we did not observe an increase in protein until 48 hours of E2 treatment. This may reflect a delay between kisspeptin transcription, translation, and processing. Kisspeptin is expressed as a large preproprotein that is posttransationally processed and cleaved into smaller, bioactive fragments, which may account for the delay. Alternatively, we may be able to detect a change in kisspeptin mRNA more easily than in protein because PCR is more sensitive than Western blotting. In any case, it is clear that estrogen membrane-initiated signaling (EMS) increased kisspeptin expression in these assays. EMS is known to mediate rapid, nonnuclear events in neurons and astrocytes, such as the release of internal Ca²⁺ stores and synthesis of P4 (43, 44), but EMS can also signal from the membrane to the nucleus [(45) and reviewed in reference ⁴⁶]. The latter pathway leads to the activation of cAMP response element-binding (CREB) protein-mediated gene transcription. In vivo, the time course of E2 stimulation of kisspeptin has suggested the activation of estrogen response element (ERE) for both RP3V kisspeptin expression and estrogen feedback underlying the LH surge (19, 47, 48). Indeed, the present results with PR strongly imply an ERE-mediated expression. However, the use of mHypoA51 cells has allowed us to segregate stimulatory E2 actions on PR vs kisspeptin expression. Based on these in vitro results, we suggest a reinterpretation of the in vivo results. Whole-animal experiments measure kisspeptin expression as a cumulative response of a complex neuronal-glial network 48 hours after E2 treatment (eg, reference ⁴⁹). During this time, E2 induces kisspeptin and neuroprogesterone synthesis through EMS (43), whereas E2 induces PR through what appears to be a nuclear/ERE action [estrogen binding loci have been identified on or close to PR genes in multiple species (50–55)]. Therefore, ERE action may be an important feature of regulating kisspeptin through an indirect mechanism: ERE-mediated induction of PR could, in turn, influence kisspeptin expression. Indeed, we have preliminary data that show P4 augmentation of the estrogenic induction of kisspeptin. Because PR expression is dependent on nuclear E2 action, preventing ERE activation would produce an attenuated response to E2 that would also affect kisspeptin expression.

Although published reports show increased PR message in the AVPV (25, 26), to date, the E2 regulation of PR specifically in kisspeptin neurons has not been formally examined. PR immunoreactivity in kisspeptin neurons was reportedly too low for quantification at diestrus (although quantifiable on proestrus), suggesting that E2 does increase PR in kisspeptin neurons (23). In mHypoA51 cells, we demonstrated PR up-regulation specific to kisspeptin neurons. Treatment with E-6-BSA did not alter PR mRNA levels, suggesting this up-regulation is mediated by intracellular ER and probably ERE-dependent. Although E2 pathways that mediate kisspeptin and PR induction may be different, it is likely that they both involve activation of ERα, as evidenced by the ability of PPT, but not DPN, to elicit increases in expression of both PR and kisspeptin. Indeed, ERα has been shown to be necessary for E2 positive feedback (19), and a very recent report indicates ERα in kisspeptin neurons is critical for this positive feedback and for the up-regulation of RP3V kisspeptin by E2 (56). Preliminary experiments using ICI 182780 to block E2 effects reported in this study were equivocal. We have observed membrane estrogen receptor G-protein-coupled estrogen receptor (GPER)/GPR30 in mHypoA51 neurons, which complicates interpretation of ICI experiments because ICI is a potent GPER/GPR30 agonist. Further studies are being pursued aimed at elucidating the role of GPER/GPR30 in mHypoA51 neurons.

ERα and PR appear to be physiologically active because mHypoA51 neurons responded to acute stimulation with 1 nM E2 or 1 nM P4 with a rapid increase in [Ca²⁺]_i. We have identified ERα and PR on mHypoA51 neuronal membranes, supporting the idea that increases in [Ca²⁺]_i occur via activation of classical steroid receptors localized to the plasma membrane. The P4-induced cytoplasmic free Ca²⁺ ([Ca²⁺]_i) increase was abrogated when cells were steroid starved, which is consistent with the E2-induced PR mRNA results. PR, like ERα, ERβ, and the androgen receptor, has a highly conserved nine-amino acid sequence in the ligand binding (E) domain that can be palmitoylated, mediating binding to scaffolding proteins that allow for trafficking to the cell membrane (57, 58). At this point, we have not characterized the proximal signaling of membrane PR. It is possible that PR is itself coupled to a G protein or that PR, like ERα, transactivates another membrane receptor (eg, mGluR1a). Elucidating PR downstream signaling pathways will require additional experiments.

In addition to PRs, mHypoA51 cells express several of the membrane PRs [mPR; also referred to as progestin and adipoQ receptors (PAQRs)]. mHypA51 cells express PAQR7, -8, and -5 (mPRα, -β, and –γ, respectively; data not shown), discovered and characterized in fish and in other vertebrates including humans (reviewed in reference ⁵⁹). PAQR8/mPRβ is the most prominent in regions regulating GnRH [rostral hypothalamus and diagonal band of Broca (60)]. PAQRs can inhibit or excite neurons and are known to signal through MAPK activation and increasing [Ca²⁺]_i [(59, 61–66; but see reference ⁶⁷)]. Although there is some controversy, the weight of evidence suggests that classical PRs, but not PAQRs (mPRs), are antagonized by RU486 (68–70). In the present experiments, RU486 dramatically attenuated the P4 response, suggesting that the main receptor contributing to the rapid Ca²⁺ effects produced by P4 in mHypoA51 neurons appears to be classical PR trafficked to the cell membrane. In summary, the present results are consistent with a model of estrogen-positive feedback in which RP3V kisspeptin neurons integrate steroid information and activate GnRH neurons. PR and kisspeptin are induced in mHypoA51 neurons after E2 stimulation, similar to increases in these proteins in the RP3V in response to increasing levels of ovarian E2. Peaking E2 levels on proestrus facilitate P4 synthesis in hypothalamic astrocytes (reviewed in reference ⁴⁹). Thus, estradiol's simultaneous induction of PR in kisspeptin neurons and the initiation of local progesterone synthesis in astrocytes may constitute the switch from negative to positive estrogen feedback in the hypothalamic control of LH release.