Epigenetic Changes Coincide with in Vitro Primate GnRH Neuronal Maturation

Abstract

Cellular and molecular mechanisms underlying pulsatile GnRH release are not well understood. In the present study, we examined the developmental changes in intracellular calcium dynamics, peptide release, gene expression, and DNA methylation in cultured GnRH neurons derived from the nasal placode of rhesus monkeys. We found that GnRH neurons were functionally immature, exhibiting little fluctuation in intracellular calcium ([Ca²⁺]_i) and sparse pulses of GnRH peptide release in the first 12 d in vitro (div). By 14–18 div, GnRH neurons exhibited periodic [Ca²⁺]_i oscillations, synchronizing at approximately 60-min intervals and GnRH pulses occurred at approximately 60-min intervals. Interestingly, the total GnRH peptide release further increased after 18 div. Measurement of GnRH mRNA and gene CpG methylation status at 0, 14, and 20 div indicated that mRNA levels significantly (P < 0.05) increased between 14 and 20 div, just as maximal decapeptide release was observed. By bisulfite sequencing across a 5′ CpG island of the GnRH gene, we further found that methylation at eight of 14 CpG sites significantly (P < 0.05) decreased between 0 and 20 div. These data indicate that epigenetic differentiation occurs during GnRH neuronal development and suggest that increased GnRH gene expression and decreased CpG methylation status are molecular phenotypes of mature GnRH neurons. To our knowledge, this is the first report that developmental DNA demethylation occurs in postmitotic neurons toward a stable neuronal phenotype.

Primate GnRH neuronal maturation is marked by age-dependent development of spontaneous synchronizing [Ca2⁺]_i oscillations, pulsatile peptide release, increased GnRH gene expression and demethylation of a 5’ CpG-rich region of the GnRH gene.

GnRH is a decapeptide released into the hypothalamic portal circulation, through which it stimulates the synthesis and release of LH and FSH from the anterior pituitary gland. The pulsatile release of GnRH is essential for sexual maturation and maintenance of the ovulatory cycle (1). Consequently, mechanisms governing pulsatile GnRH release are central to reproductive function. However, the investigation of cellular and molecular mechanisms regulating GnRH release is very challenging because the primate brain contains only approximately 2000 GnRH neurons that are widely scattered among other neurons and neuroglia in the preoptic area (POA) and medial basal hypothalamus (MBH) (2,3).

Importantly, the unique origin of GnRH neurons provides a means to investigate the developmental events leading to fully functional neurons. GnRH neurons originate in the nasal epithelium in the mouse (4,5), rhesus monkey (6,7), and human (8). In the rhesus monkey (gestational period 165–168 d), GnRH neurons are found in the nasal pit as early as embryonic day (E) 32 and commonly at E34-E36 (6,7). The GnRH neurons then migrate along the nasal septum and terminal nerve, enter the forebrain at E38 (7), and subsequently migrate into the MBH of the rhesus monkey by E47 (6). While neurons continue to migrate on this tract until the third trimester, the basic distribution pattern of GnRH neurons is already established at E55 (7,9).

Cultured GnRH neurons derived from the nasal pit develop characteristics of functionally mature neurons. Remarkably, the time frame of in vitro maturation of GnRH neuronal function [i.e. spontaneous and synchronizing intracellular calcium ([Ca²⁺]_i) oscillations and pulsatile GnRH release] is similar to the 2- to 3-wk in vivo migratory period outlined above. We previously observed that only few spontaneous [Ca²⁺]_i oscillations occurred in primate GnRH neurons during the first 8–9 d in vitro (div), but by 14 div, [Ca²⁺]_i oscillations appeared in most GnRH neurons (10). Likewise, perifusion experiments with cultured embryonic neurons indicate that GnRH was barely detectable until at least 14 div (11). Recently similar observations were reported in mouse GnRH neurons derived from the nasal placode, i.e. pulsatile GnRH release was detectable at 3 div and the amplitude of GnRH pulses and the secretory rate gradually increased with culture days, reaching the maximum at 14 div (12). These authors also found that synchronization of [Ca²⁺]_i oscillations among GnRH neurons were sparse when cells were young in culture, but they became more frequent when cells matured for 14 div. Although the maturation of intracellular calcium dynamics and pulsatile GnRH release has been the subject of intense investigation in these culture models, the molecular mechanisms of GnRH neuronal maturation remain unclear.

The control of GnRH gene expression during embryonic development appears to be intricately related to mature neuronal function. It has been reported that parallel increases in GnRH mRNA expression and peptide biosynthesis occur along with the development of GnRH peptide release patterns in the mouse (13,14). In GT1 cells, which carry the rat GnRH gene, oscillatory GnRH gene promoter activity and gene expression have been shown in association with pulsatile GnRH release (15). Moreover, a distal 5′ enhancer region of the rat gene, which has high sequence similarity to a distal 5′ portion of the human gene (16), appears to regulate GnRH gene expression (17,18,19).

A question arises as to whether maturational changes occur at a distal portion of the GnRH gene during neuronal maturation. The distal 5′ region of the rhesus monkey GnRH gene contains a 243-bp segment (−2126 to −1863 bases from the transcription start site) that has 60% GC content and a CpG dinucleotide observed to expected content ratio of 0.65. These characteristics define the region as a CpG island (CGI) (20). CGIs, when associated with gene promoters, are often related to the epigenetic regulation (DNA methylation) of gene expression. Therefore, we hypothesized that epigenetic differentiation of the GnRH gene occurs during neuronal maturation. In the present study, to determine the temporal relationship between functional maturation and GnRH gene expression, we first examined the maturational pattern of primate GnRH neurons monitoring in vitro culture day-related changes in intracellular calcium dynamics and GnRH peptide release. Subsequently we measured GnRH mRNA expression and the methylation status of a 5′ CGI associated with the rhesus monkey GnRH gene at three time points during neuronal maturation.

Materials and Methods

General

All animals used in this study were born and raised in the Wisconsin National Primate Research Center. They were housed in pairs (cages 172 × 86 × 86 cm) in rooms with 12 h light (0600–1800 h) and 12 h dark (1800–0600 h) and controlled temperature (22 C). The animals were fed a standard diet from Harlan Teklad (20% protein primate diet; Madison, WI) twice daily, supplemented with fresh fruit and snacks daily. Water was available ad libitum. The protocol for this study was reviewed and approved by the Animal Care and Use Committee, University of Wisconsin, and all experiments were conducted under the guidelines established by the National Institutes of Health and U.S. Department of Agriculture.

Breeding methods and pregnancy determination

For the time-mated pregnancy, daily sex-skin color changes and menstrual records of adult female monkeys were obtained. Between d 9 and d 17 of the menstrual cycle, blood samples were obtained for assessment of LH levels. A few days before maximum sex-skin color change, female rhesus monkeys were placed with a fertile male until the breakdown of sex-skin color. Pregnancy was determined by ultrasound examination, uterine palpation, and RIA measurement of monkey chorionic gonadotropin. The day after the LH surge was designated as d 0 of gestation (E0). All fetectomies were conducted on either E36 or E37.

Tissue preparations and culture conditions

Before dissection of fetuses, the crown-rump length was measured, the developmental characteristics were observed under stereomicroscope, and the developmental stage was determined as described previously (7). Nasal placode tissues were dissected (a total of 15 fetuses were used in these studies) out into chilled sterile imidazole-buffered L15 medium (Invitrogen, Carlsbad, CA) using a stereomicroscope, under a plexiglas enclosure, with a very fine watchmakers’ forceps, a scalpel, and a fine iris scissors. Tissues were then cut into very small pieces (<0.5 mm³), which were plated onto 2-mm-diameter round glass coverslips (for Ca²⁺ imaging, gene expression and methylation analysis studies; Bellco Glass, Vineland, NJ) or plastic Thermanox coverslips (for perifusion; Nalgene Nunc International, Rochester, NY). Before cell plating, all coverslips were coated with a layer of dried rat-tail collagen and sterilized under a UV light. Cultures on coverslips were maintained in 35-mm tissue culture dishes (Dow Corning, Corning, NY) containing Medium 199 (Sigma, St. Louis, MO) supplemented with 10% fetal bovine serum (HyClone, Logan, UT), 0.6% glucose, 50 mg/ml gentamicin, and l-glutamine, and incubated at 37 C with 1.5% CO₂ and 98.5% air. Serum was stripped with dextran-coated charcoal to remove steroids and other small molecules. On the third day of culture, we exposed cells to the antimitotic agent 5-fluoro-5-deoxyuridine (40 μm) for 72 h to eliminate mitotic cell proliferation. Medium was replaced every 3–4 d at the beginning of cultures and every 1–2 d after the cultures were established. Cultures were maintained up to 4 wk.

Ca²⁺ imaging

[Ca²⁺]_i levels were assessed as previously described (21,22). Cells cultured on grid-etched round glass coverslips were exposed to the Ca²⁺ indicator fura-2 (Texas Fluorescence Labs Inc., Austin, TX). The coverslips with cells were mounted in a Dvorak-Stotler chamber. Fluorescence imaging of the dye-loaded cultured cells was achieved with a Nikon inverted microscope and superfluor objective lens (Nikon, Melville, NY). Cultured cells were continuously perifused with culture medium under 95% O₂ and 5% CO₂ at a rate of 50 μl/min. All experiments were conducted at room temperature (22–23 C). Under the microscope, GnRH neuron-like cells were chosen. GnRH cells are identifiable by their size, morphology, and unique appearance, forming neuronal bundles. Approximately 50–100 cells, including GnRH neurons, GnRH progenitor cells in the placode, nonneuronal cells, and in rare cases non-GnRH neurons were seen on each coverslip. Using a xenon lamp and ultrahigh speed wavelength changer (DG-4; Sutter Instruments, Navato, CA), cells were excited successively with UV light of 340 and 380 nm wavelengths, and an emission light of 510 nm was captured every 10 sec by a charge-coupled device camera (Photometrics, Tucson, AZ) attached to the microscope. The ratio of the emission from 340 nm excitation to 380 nm excitation, which is proportional to [Ca²⁺]_i concentrations, was calculated by MetaFluor imaging software (Molecular Devices Corp., Downingtown, PA). After the experiment, the recorded region was photographed, and the photoengraved grid location was determined for later histological analysis. Subsequently cells were fixed with 2% paraformaldehyde (pH 7.6) and immunostained for GnRH and neuron-specific enolase as described below. In this way we could confirm that the Ca²⁺ recordings were from GnRH neurons.

Perifusion (GnRH release) experiments

Methods have been previously described in detail (11,22,23). Cells were analyzed at specified time points after the initiation of culture after embryonic tissue dissection. Cultures containing GnRH cells were mounted in Sykes-Moore chambers as described by Martinez de la Escalera et al. (24), i.e. two coverslips were placed, face to face, separated by a rubber O-ring, forming a chamber with a volume of 200 μl. Cells were then perifused with artificial cerebrospinal fluid containing 0.1% glucose (pH 7.4) under 95% O₂ and 5% CO₂ at 37 C. Perifusates were collected at 25 μl/min in 10-min fractions for 6 h using the ACUSYST (Endotronics, Minneapolis, MN) perifusion system. To test the viability of cells, 56 mm K⁺ was perifused for 20 min during the last hour (h 6) of the experiment for most cases including all early cultures when spontaneous GnRH pulsing did not occur. Samples were stored at −80 C until they were assayed for GnRH. After the perifusion experiment, cells were immunostained for GnRH as described below.

Immunostaining

All cultures were immunostained after [Ca²⁺]_i imaging or perifusion experiments, as described previously (22). Briefly, cultured cells were gently rinsed with PBS and fixed with 2% paraformaldehyde in PBS for 30 min at room temperature. They were rinsed thoroughly with PBS and then placed in 10% normal goat serum in PBS for 2 h at room temperature. Cells were exposed to primary antibody for 48 h at 4 C. For GnRH staining, a cocktail of the polyclonal antiserum LR-1, which primarily recognizes mammalian GnRH (a gift of Dr. R. Benoit, University of Montréal, Montréal, Quebec, Canada), and GF6, which recognizes several forms of GnRH (a gift of Dr. N. Sherwood, University of British Victoria, Victoria, British Columbia, Canada), was used at a dilution of 1:20,000 and 1:9,000, respectively. Cultured cells were again rinsed thoroughly with PBS and were incubated with biotinylated antirabbit IgG for 1.5 h at room temperature. After rinsing in PBS, avidin-biotinylated peroxidase complex (Vector Laboratories, Burlingame, CA) in PBS was applied for 1.5 h at room temperature, followed by rinses in Tris-buffered saline. For visualization, 3,3′-diaminobenzidine and H₂O₂, in Tris-buffer, were applied. As a positive control, GnRH neurons in the hypothalamus of rhesus monkeys, at various ages, were also stained. To stain neuron-specific protein, cells were further exposed to antibody against neuron-specific enolase (INCSTAR Corp., Stillwater, MN) at a dilution of 1:1000. Double staining was accomplished using a similar procedure, except for a different chromagen (Vector SG complex; Vector Laboratories).

GnRH RIA

GnRH concentrations from perifusion experiments were measured in 200-μl samples with RIA using antiserum R-1245 (a gift from Dr. T. Nett, Colorado State University, Fort Collins, CO) as described previously (25). Synthetic GnRH (Richelieu Laboratory, Inc., Montréal, Canada) was used for both the trace and the reference standard. The sensitivity of the assay, at 95% binding, was 0.05 pg/tube. Intra- and interassay coefficients of variation were 8.3 and 11.4%, respectively.

Nucleic acid and protein isolation

Total RNA and genomic DNA were isolated from cultured cells using the AllPrep RNA/DNA/protein minikit (catalog no. 80004; QIAGEN, Valencia, CA). These samples were stored at −80 C and were used for the analysis of DNA methylation and mRNA expression.

Quantitative real-time PCR

Measurements were made using techniques that we have recently published (26). Briefly, concentrations of total RNA isolated from cultured cells was determined using a BioMate 3 spectrophotometer (Thermo Spectronic, Lanham, MD), and cDNA was generated with the StrataScript first-strand synthesis system kit (Stratagene, Cedar Creek, TX). Samples were stored at −80 C until used in quantitative real-time PCR. All reactions were multiplexed with GnRH and the housekeeping gene 18s amplified together. Amplifications were done with platinum quantitative real-time PCR SuperMix-UDG (Invitrogen) and ROX as a reference dye. 18s primers were JOE (6-carboxy-40, 50-dichloro-20, 70-dimethyoxyfluorescein) labeled, whereas GnRH primers were FAM (6-carboxy-fluorescein) labeled. JOE, FAM, and ROX fluorescence was simultaneously measured in each well with optimal absorbance between 535 and 555, 492 and 516, and 584 nm, respectively. Primer and sample concentrations were optimized to ensure linearity of amplification near 15 cycles for 18s and 25 cycles for GnRH. Primer sequences were as follows. GnRH: forward, CGGCTGGAGGAAAGAGAGATGCCG; reverse, GCCAGTTGACCAACCTCTTTGA. 18s primers were supplied by Invitrogen and sequences are proprietary, although they are known to amplify in the 3′ region of the transcript (catalog no. 115HM-02; Invitrogen).

Sodium bisulfite sequencing (DNA methylation analysis)

Genomic DNA isolated from snap-frozen sections was sodium bisulfite treated with the Imprint DNA modification kit (Sigma Aldrich, St. Louis, MO, catalog no. MOD50-1KT). The GnRH 5′ region (accession NW_001122890) was amplified using the following primers: forward, 5′-GAGTTTTGTTTTGTTATTTAGG-3′, and reverse, 5′-CAATAACTCAAACCTATAATCC-3′. This generated a 264-bp product that was then gel purified and ligated into the pCR 4-TOPO vector, which is part of the TOPO TA cloning kit (catalog no. 45-0030; Invitrogen). Ligated vectors were transfected into GC10 competent cells (catalog no. G2794; Sigma Aldrich) by heat shock, and cells were then grown in superoptimal catabolite. Cells were then spread on agar plates containing 100 μg/ml ampicillin and allowed to grow overnight at 37 C. The next day the individual, well-isolated colonies were picked from the plate and allowed to grow further by shaking in 3 ml of Luria-Bertani broth containing 100 μg/ml ampicillin at 37 C for 8 h. Cells were harvested by centrifugation, and cloned vectors were isolated using the Purelink quick plasmid miniprep kit (catalog no. K2100-11; Invitrogen). Vector inserts were sequenced with the supplied reverse sequencing primer in the vector kit. The Big Dye sequencing reactions were cleaned and sequenced at the University of Wisconsin Biotechnology Center Sequencing facility using an 3730xl automated DNA sequencing instrument (Applied Biosystems, Carlsbad, CA).

Data and statistical analyses

Fetuses

A total of 15 fetuses were used. Some of the fetuses contributed to all four experiments (calcium dynamics, peptide release, mRNA expression, and CpG methylation analysis). Each fetus provided cultures for each age group (≤12, 14–18, and ≥ 20 div) studied.

[Ca²⁺]_i imaging experiments

All Ca²⁺ dynamics data were analyzed with the PULSAR algorithm (27) as described previously (22). Intervals between [Ca²⁺]_i oscillations [interpulse interval (IPI)], pulse duration (ascending phase plus descending phase), and pulse amplitude (difference between the baseline and the peak) for each cell were obtained from PULSAR analysis. Group mean (±sem) from all cells was calculated from individual data. Synchronization of [Ca²⁺]_i oscillations were defined as previously described (22). The mean (±sem) interval between the synchronizations was calculated among the synchronized cases. Six cultures for each age group from a total of nine fetuses were studied.

GnRH peptide release/perifusion experiments

GnRH pulse characteristics (pulse number, pulse amplitude, and pulse duration) were determined using the PULSAR algorithm from samples during h 1–5 due to instability of cultures during the first hour of perifusion experiments. The parameters used to characterize the GnRH pulses were similar to those described previously (25). GnRH peaks at less than 0.4 pg/ml, revealed by PULSAR algorithm, were eliminated because they were not reliable due to the proximity to the assay sensitivity of 0.25 pg/ml. The mean release per 4 h was calculated from sum of all values during the 4-h sampling period in each culture. IPI was calculated from the pulse number during that 4-h period. Four to seven cultures for each age group from a total of three fetuses were studied.

Quantitative PCR

Data were compared using the relative expression change in cycle threshold value (ΔCT) method (28) [n = 4 for each developmental time point (0, 14, and 20 div)]. Each of the four samples was a pool of two dishes from a total of three different fetuses.

CpG methylation

Big Dye sequencing reaction data were analyzed using PE Biosystems’ version 3.7 of sequencing analysis. Twenty clones containing the expected insert were sequenced for each time point. CpG methylation was averaged across the clones for each of the 14 CpG sites. For each time point, four culture dishes were pooled from two fetuses.

All group differences were compared using Student’s t tests. Differences between groups were considered to be significant when P ≤ 0.05.

Results

Developmental changes in [Ca²⁺]_i oscillations

We assessed changes in [Ca²⁺]_i in GnRH neurons using the Ca²⁺ dye fura 2 starting at 8 div though 28 div (Table 1). In young cultures such as 8–9 div, [Ca²⁺]_i levels were mostly flat (Fig. 1). In young cultures through 12 div, there were infrequent, low-amplitude (2.53 ± 0.48 ΔF/F₀; mean ± sem) [Ca²⁺]_i oscillations. In contrast, by 14 div, [Ca²⁺]_i oscillations were more regular with a higher amplitude (Fig. 1). In fact, the amplitude of [Ca²⁺]_i oscillations increased significantly to 7.18 ± 0.93 ΔF/F₀ (P < 0.001, Table 1). Average IPIs were long with large se (45.3 ± 21.4 min) in young cultures (Table 1) but developed into shorter and more regular IPIs in 14- to 18-div cultures (5.7 ± 1.0 min), although mean values were not statistically significant. These dynamics remained stable through 28 div (6.0 ± 1.3 min, Table 1).

Table 1.

Developmental changes in [Ca²⁺]_i oscillations

Div	Number of cultures, n (cells, n)	Pulse amplitude (ΔF/F0)	Pulse interval (min)	Pulse duration (min)	Synchronization interval (min)
≤12	6 (184)	2.53 ± 0.48	45.3 ± 21.4	1.4 ± 0.4	111.7 ± 22.5
14–18	6 (248)	7.18 ± 0.93a	5.7 ± 1.0	1.3 ± 0.2	61.1 ± 7.0b
≥20	6 (256)	5.99 ± 0.75c	6.0 ± 1.3	1.1 ± 0.2	55.2 ± 4.3b

^aP < 0.001 vs. 12 div or less. 

^bP < 0.05 vs. 12 div or less. 

^cP < 0.01 vs. 12 div or less. 

Figure 1.

Developmental pattern of [Ca²⁺]_i dynamics in monkey GnRH neurons. Representative [Ca²⁺]_i traces from individual GnRH neurons at 8, 9, 12, 13, 14, and 28 div are shown. Note that there is little fluctuation of [Ca²⁺]_i levels in young cultures (8 and 9 div, top panels). At 12–13 div, small fluctuations of [Ca²⁺]_i start to appear (middle panels). After 14 div, most of GnRH neurons exhibit prominent oscillatory patterns of [Ca²⁺]_i. Periodic synchronizations of [Ca²⁺]_i oscillations are seen in older cultures. Synchronization of [Ca²⁺]_i oscillations in the majority GnRH neurons are indicated by vertical lines.

Synchronization of [Ca²⁺]_i oscillations was not often observed through 12 div. As a consequence, average pulse synchronization intervals decreased significantly between 12 div or less and 14–18 div (≤12 div: 111.7 ± 22.5 min vs. 14–18 div: 61.1 ± 7.0 min; P < 0.05, Table 1). Once synchronization of [Ca²⁺]_i oscillations started to appear by 14–18 div, they remained similar, i.e. the [Ca²⁺]_i synchronization interval about 60 min in 14- to 18-div cultures (Fig. 1) was similar to that in cultures of 20 div or greater.

Developmental changes in GnRH peptide release

Using a perifusion system, we examined the developmental pattern of pulsatile GnRH release, starting at 8 div though 28 div (Table 2). As shown in an example of a culture at 11 div, GnRH pulses were rarely seen in young cultures (Fig. 2). However, pulse patterns gradually developed during the first 2–3 wk in vitro (Fig. 2). Peptide release calculated from total release during the 4-h sampling period was low (3.62 ± 0.39 pg/ml per 4 h) in 12 div or less (Table 2). By 14–18 div, total peptide release increased significantly (7.98 ± 0.95 pg/ml per 4 h; P < 0.01) as did pulse duration (≤12 div: 13.3 ± 2.9 min vs. 14–18 div: 21.7 ± 1.9 min; P < 0.05) and number of pulses over 4 h (≤12 div: 1.0 ± 0.4 vs. 14–18 div: 3.6 ± 0.6; P < 0.01). Although pulse amplitude, duration, and number did not change significantly beyond 2 wk in vitro, total release over 4 h continued to rise (≥20 div: 12.76 ± 1.54 pg/ml per 4 h) after 14–18 div (7.98 ± 0.95; P < 0.05). This continued increase in total release is intriguing, considering the time frame for increased GnRH mRNA levels, which is described below.

Table 2.

Developmental changes in GnRH pulsatility

Div	Cultures, n	Release/4 h (pg/ml)	Pulse amplitude (pg/ml)	Pulse durations (min)	Pulse no./4 h	IPI (min)a
≤12	4	3.62 ± 0.39	0.38 ± 0.04	13.3 ± 2.9	1.0 ± 0.4	210.0 ± 30.0
14–18	7	7.98 ± 0.95b	0.54 ± 0.12	21.7 ± 1.9c	3.6 ± 0.6b	66.7 ± 10.3b
≥20	6	12.76 ± 1.54b,d	0.85 ± 0.25	27.2 ± 2.7b	5.2 ± 0.5b	49.7 ± 6.5b

^aCalculated from pulse number per 4 h. 

^bP < 0.01 vs. 12 div or less. 

^cP < 0.05 vs. 12 div or less. 

^dP < 0.05 vs. 14–18 div. 

Figure 2.

Developmental pattern of GnRH peptide release in monkey GnRH neurons. Representative cases of in vitro GnRH release from cultures at 11, 17, and 20 div, obtained using a perifusion system, are shown. GnRH release was low with infrequent, low-amplitude pulses in young cultures (top panel). After 2 wk in culture (middle panel), pulse duration and frequency increased significantly, reaching maximal release levels by 20 div (bottom panel). GnRH peaks (indicated by arrows) were identified by the PULSAR algorithm. Fine lines at 0.25 pg/ml indicate the assay sensitivity.

Developmental changes in GnRH mRNA expression and GnRH gene methylation status

Total RNA and DNA from cells plated on the coverslips were extracted at 0, 14, and 20 div. Using quantitative RT-PCR, GnRH and 18s mRNA levels were measured from total RNA. Relative GnRH mRNA expression (mean ± sem) was calculated using the ΔCT method (28) (Fig. 3). We found that expression was low at the beginning of cultures (0 div: 6.84 × 10⁻⁷ ± 6.29 × 10⁻⁸) and remained low through 14 div (1.08 × 10⁻⁶ ± 3.18 × 10⁻⁷), but by 20 div, GnRH mRNA levels rose significantly (1.26 × 10⁻⁵ ± 4.75 × 10⁻⁶, P ≤ 0.05 vs. 0 and 14 div). Remarkably, CpG methylation status in a 5′ region of the GnRH gene reciprocally mirrored the mRNA expression pattern. Using bisulfite sequencing, we analyzed the CpG methylation status at 14 CpG sites. We found that methylation at eight of these sites significantly (P < 0.05) decreased during development, with the lowest CpG methylation status found at 20 div (Fig. 4).

Figure 3.

Changes in GnRH mRNA levels during GnRH neuronal development. Total RNA was extracted from cultures at 0, 14, and 20 div (n = 4 in all age groups). GnRH mRNA levels started to increase after 14 div, reaching the highest level at 20 div. *, P < 0.05 vs. 0 div; #, P = 0.05 vs. 14 div. GnRH mRNA levels are relative to 18s in each sample and analyzed using the ΔCT method (28).

Figure 4.

GnRH gene methylation during GnRH neuronal development. A, A schematic representation of the rhesus monkey GnRH gene depicting the location of the 5′ CGI and nucleotide sequence of this region. CpG sites are indicated with numbers corresponding to the CpG sites in the bar chart below. Underlined CpG sites exhibit higher methylation status at 0 compared with 20 div. B, GnRH neurons dissected from the nasal placode region of two rhesus monkey embryos at E36 and E37 were plated and then harvested on 0, 14, or 20 div. DNA extracted from pooled samples (four cultures) at each time point was bisulfite sequenced. Percent changes in methylation at each CpG site on 0 div (black bars), 14 div (gray bars), and 20 div (white bars) are shown. CpG methylation status was significantly higher at 0 div compared with 20 div at sites 4, 5, 6, 8, 9, 11, 12, and 14. *, P ≤ 0.01). CpG methylation status was significantly higher at 0 div compared with 14 div at site 5 (#, P < 0.05) and at 0 div compared with 14 div but not 20 div at site 10 (##, P < 0.05).

Discussion

The ontogeny of GnRH neurons in mammalian species is unique. They differentiate from progenitor cells in the embryonic nasal pit and migrate over a long distance to permanently reside in the POA-MBH during the first 2 wk after differentiation. The results of the current studies with cultured monkey GnRH neurons derived from the nasal placode indicate that characteristics of mature GnRH neurons develop gradually in vitro during the first 2 wk in culture. Furthermore, we found that patterns of spontaneous [Ca²⁺]_i oscillations and pulsatile GnRH release developed along a tightly coordinated time frame during this period. Specifically, the amplitude of spontaneous [Ca²⁺]_i oscillations and synchronization interval of [Ca²⁺]_i oscillations reach mature levels by 14–18 div. Similarly, the duration and IPI of GnRH pulses also reached a mature state by 14–18 div. In fact, by 14–18 div, both the synchronization of spontaneous [Ca²⁺]_i oscillations and GnRH pulse intervals were similar to the IPI of in vivo pulsatile GnRH release observed in adult monkeys (25). In contrast, GnRH mRNA expression remained low through 14 div, suggesting functional maturation of GnRH neurons in terms of ability to exhibit mature [Ca²⁺]_i dynamics and capacity to release the decapeptide is independent of a significant increase in GnRH mRNA expression. However, mRNA did increase significantly between 14 and 20 div, when the maximum peptide release was observed. Remarkably, the DNA methylation status in a 5′ region of the GnRH gene reciprocally mirrored GnRH mRNA expression, In fact, decreasing DNA methylation status coincided with increasing GnRH mRNA levels during neuronal maturation.

Similar to our observations in these studies, using cultured mouse embryonic placode cells, Constantin et al. (12) reported that GnRH pulses became detectable at 3 div (cultures started E11.5), and the amplitude of GnRH pulses and the secretory rate gradually increased with culture days, reaching the maximum at 14 div (12). These authors also found that synchronization of [Ca²⁺]_i oscillations among GnRH neurons were sparse when cells were young in culture but became more frequent and occurred in parallel with GnRH release in static cultures after 14 div. Additionally, at 1 wk in culture, the interval between spontaneous [Ca²⁺]_i spikes in mouse embryonic GnRH neurons is approximately 13 min but decreases to approximately 7 min by 3 wk in culture (29). This shorter interval is similar to the approximately 8-min interval between spontaneous [Ca²⁺]_i spikes reported for adult mouse pericam GnRH neurons (30). Although there are subtle differences in monkey and mouse studies, the results are consistent with a concept that GnRH neurons undergo functional maturation during the first 2 wk after differentiation from their progenitor cells with a close coincidence of GnRH release and oscillatory [Ca²⁺]_i events.

The 2-wk period for functional maturation is not likely an artifact because of in vitro cultures. It has been consistently shown that in vitro maturation recapitulates in vivo neuronal maturation. For example, protein expression of metabotropic glutamate receptors in cultured rat hypothalamic neurons started at birth, increased with culture days in parallel to the postnatal (P) ages (31), and γ-aminobutyric acid neurotransmission was reversed from excitatory to inhibitory between 4 and 18 div in cultured cells, similar to those observed in slice preparations obtained from the hypothalami of P4 and P18 rats (32). In mouse GnRH neurons, Moore and Wray (14) reported that GnRH peptide contents gradually increase with a similar developmental time course in vivo and in vitro: GnRH contents in mouse nasal placode cultures (started at E11.5) significantly increase between 7 and 10 div, which parallels the in vivo increase between E14.5 and P1, and the GnRH peptide contents at 10 div are indistinguishable from peptide contents at a comparable time point in the mouse brain at P1.

The pulsatile release of GnRH is indispensable for normal reproductive function and is assumed to be due to the synchronous activity among widely scattered GnRH neurons in the POA/MBH. Although the concept that multiple GnRH neurons release the decapeptide synchronously in vivo is yet to be proven, findings of dendrodendritic bundling and shared synapses between two mouse GnRH neurons (33), and the presence of gap junctions in the median eminence (34) indicate interactions among GnRH neurons and/or their neuroterminals, which would enable synchronous activity. In vitro studies in cultured GnRH neurons derived from the embryonic nasal placode provide additional evidence for coordinated activity among GnRH neurons. For example, GnRH neurons exhibit synchronization of spontaneous [Ca²⁺]_i oscillations with species-specific intervals, such as approximately 60 min in rhesus monkeys (10) and approximately 20 min in mice (29). Moreover, direct measurements from mature GnRH neurons in vitro indicate that GnRH is released with intervals at approximately 60 min in monkeys (11), approximately 20 min in mice (12), and approximately 40 min in sheep (35). Not surprising, the species-specific pulse intervals of GnRH release in vitro are consistent with IPI of in vivo GnRH/LH release in adult monkeys (25,36), mice (37), and sheep (38). Importantly, synchronization of [Ca²⁺]_i oscillations was often seen as a calcium wave in monkey GnRH neurons (39), and synchronization of [Ca²⁺]_i oscillations among 30–50% of GnRH neurons was associated with increases in GnRH peptide release in mouse neurons (12).

Although young (≤12 div) GnRH neurons released a small amount of GnRH and mature patterns of [Ca²⁺]_i oscillations and GnRH release were observed by 14–18 div, the highest amount of GnRH release during the 4-h period did not occur until 20 div. Interestingly, this maximal peptide release was observed when GnRH mRNA levels significantly increased. These results indicate that low GnRH mRNA expression is sufficient for peptide release and that maximal peptide release from GnRH neurons requires increased gene expression. Similar observations were reported in mouse GnRH neurons, in which total peptide contents increased between 14 and 21 div, well after pulsatile GnRH release was first detected at 3 div (12). Taken together, the neurosecretory process does not require a high level of GnRH biosynthesis.

In the present study, we found a tight relationship between GnRH mRNA levels and CpG methylation status in a 5′ region of the GnRH gene. Perhaps it is reasonable to speculate that transcriptional efficiency, influenced by DNA methylation status, is responsible for differences in GnRH mRNA. It is, however, possible that methylation status is not a determinant of mRNA levels at this stage of neuronal development because GnRH biosynthesis requires several steps, including posttranscriptional processing of the primary GnRH gene transcript to a mature mRNA (40). Although, presently, no information is available for posttranscriptional control of GnRH mRNA levels in primates, especially during development, measurements of mRNA levels during mouse embryonic development (E16 to P0) indicate that mRNA levels are correlated with levels of transcription rather than differences in posttranscriptional processing (40). Given the close resemblances of GnRH neuronal developmental patterns in mouse and monkey discussed above, it is likely that posttranscriptional events are not primarily responsible for developmental changes in mRNA levels in the monkey. Rather, different levels of transcription likely control GnRH mRNA levels at this point of development.

Reviewing literature of studies conducted in GT1, GN, and NLT cells, it became clear that sequences within the 5′ region of the GnRH gene and trans-acting factors that associate with these sequences are well characterized for the rat but not humans. Specifically, the rat gene contains proximal and distal 5′ regions that are responsible for neuron specific (41,42,43) and pulsatile gene expression (44). The trans-acting factors that associate with these regions and appear to be responsible for episodic gene expression include octamer transcription factor-1 (15,18,45,46), GATA-4, and GATA-5 (19,47,48). Other factors including the homeobox protein Otx-2 (49), Msh homeobox 1 (50,51), and CCAAT/enhancer-binding protein-β (17) also interact with these regions, and corepressors from the Groucho-related-gene family (51,52) and GnRH enhancer region GATA-B site binding factor (48) are known to reduce gene expression through direct interactions with activators or competitive binding at cis sequences in the 5′ region of the GnRH gene. In contrast, few reports on human GnRH gene structure (53) and function are available. Investigations have identified 5′ flanking regions of the human gene that are necessary for migration into the brain (54), and other 5′ flanking regions are necessary for gene expression (16) or tissue-specific expression (55). Kepa et al. (16) also identified sequence similarities between rat and human genes in the 5′ region. Although it is currently not known whether methylation in the region we analyzed directly alters the activity of transcription factors listed above, it is clear that CpG methylation can indirectly inhibit trans-activating factor activity across a wide range of promoter sequences (at least 2 kb) through interaction with methyl-CpG-binding proteins (56). Because the region we analyzed contains the highest CpG density in the 5′ region of the rhesus GnRH gene and resides within 2 kb of all the known regulatory portions of the human GnRH gene, it is reasonable to speculate that CpG methylation in this region is associated with the epigenetic regulation of GnRH gene expression.

The apparent GnRH gene demethylation during neuronal development is quite remarkable, even though a causative relationship between CpG methylation status and mRNA expression is yet to be clarified. Until recently CpG demethylation was generally considered a passive process related to loss of methylation during DNA replication. Consequently, DNA demethylation would not be expected in nondividing cells, such as neurons. Recent studies, however, found that CpG methylation status decreases in neurons during context dependent memory consolidation (57) and electroconvulsive treatment induced neurogenesis (58), suggesting a demethylase is present and responsive to activity or context in mature neurons. The results of this study also suggest active demethylation occurs in postmitotic neurons. To our knowledge, our studies are the first to report developmental changes in DNA demethylation in a postmitotic stable neuronal phenotype. Perhaps DNA demethylation is a mechanism common to maturing peptide hormone secreting cells. Quite recently demethylation of the Ins2 gene promoter during development was reported in association with increased insulin gene expression in murine pancreatic β-cells (59).

In summary, we have found that GnRH neuronal development is characterized by an intricate coordination of Ca²⁺ events, peptide release, mRNA expression, and gene demethylation. Future studies remain to investigate the causal relationship between the DNA demethylation and GnRH neuronal maturation and the functional role of GpG demethylation in mature GnRH neuronal activity. Nonetheless, the results from these studies suggest that GnRH neurons derived from the nasal placode will be an important model for investigating the mechanisms of epigenetic regulation in neuroendocrine neurons.