Signaling Responses to Pulsatile Gonadotropin-releasing Hormone in LβT2 Gonadotrope Cells*

The hypothalamic neuropeptide gonadotropin-releasing hormone (GnRH) is secreted in a pulsatile fashion by hypothalamic neurons, and alterations in pulse frequency and amplitude differentially regulate gonadotropin synthesis and release. In this study, we investigated the kinetics of Gs and Gq signaling in response to continuous or pulsatile GnRH using fluorescence resonance energy transfer reporters in live mouse LβT2 gonadotrope cells. cAMP and protein kinase A-dependent reporters showed a rapid but transient increase in fluorescence resonance energy transfer signal with increasing doses of constant GnRH, and in contrast diacylglycerol (DAG) and calcium reporters showed a rapid and sustained signal. Multiple pulses of GnRH caused multiple pulses of cAMP and protein kinase A activation without desensitization, but the DAG and calcium reporters were rapidly desensitized resulting in inhibition of calcium and DAG responses. At the transcriptional level, both a cAMP-dependent cAMP-response element reporter and a DAG/calcium-dependent AP-1 reporter showed a pulse frequency-dependent increase in luciferase activity. However, constant GnRH stimulation gave very little cAMP-response element activation but very strong AP-1 activation. Based on these data, we propose that both the GnRH-R-Gs and Gq pathways are responsive to pulses of GnRH, but only the Gq pathway is responsive to constant GnRH. Furthermore, the Gq pathway is subject to desensitization with multiple GnRH pulses, but the Gs pathway is not.

The hypothalamic hormone gonadotropin-releasing hormone (GnRH) 2 is the central regulator of the mammalian reproduction system. It acts in the anterior pituitary via a spe-cific GnRH receptor (GnRH-R) on the plasma membrane of gonadotrope cells where it triggers the synthesis and secretion of LH and FSH, which in turn regulate production of gonadal steroids and reproduction (1,2). Physiologically, GnRH is secreted in a pulsatile fashion by hypothalamic neurons (2). Gonadotrope responsiveness is modulated by both the GnRH concentration and by the frequency or pattern of its administration. During the female reproductive cycle, estrogen increases the GnRH pulse frequency and amplitude during the pre-ovulatory phase resulting in the LH surge and ovulation. Progesterone then slows and diminishes the hypothalamic GnRH pulses resulting in a preferential increase in FSH to stimulate the next round of follicle development (3). How the gonadotrope responds to the different pulse frequencies and amplitude to differentially produce LH or FSH is poorly understood.
All of GnRH effects are mediated by the GnRH-R, which is a member of the G-protein-coupled receptor family. In primary pituitary cultures, G-GH3, and L␤T2 cells, the GnRH-R couples to G s and G q/11 but not G i . In ␣T3-1 pituitary precursor cells as well as CHO-K1 and COS-7 cells, the receptor seems to couple exclusively to G q/11 (4,5). Coupling to G i and G 12 has also been reported (6). Several reports have also shown increases in second messengers such as cAMP, inositol 1,4,5trisphosphate, Ca 2ϩ , DAG, and PKC with GnRH treatment (7)(8)(9). All of these studies, however, have used acute tonic treatment rather than pulsatile stimulation.
In this study, we have investigated the kinetics of G s and G q/11 signaling in response to GnRH pulses of varying frequency and amplitude in L␤T2-immortalized gonadotrope cells as a model for how gonadotropes decode GnRH pulses. The dynamics of the cAMP-PKA and DAG-Ca 2ϩ responses were monitored in live cells using fluorescence resonance energy transfer (FRET) reporters over Ͼ4 h. The reporters showed a rapid but transient increase in FRET signal with increasing doses of GnRH. In the context of multiple pulses, a strong FRET signal was observed with every pulse with no desensitization for the cAMP pathway, but the DAG-Ca 2ϩ pathway was rapidly desensitized. At the level of transcriptional activation, increasing the pulse frequency caused a strong activation of both CRE-dependent and AP-1-dependent reporters; however, the response to constant GnRH treatment was dramatically different. The AP-1 reporter responded to constant GnRH similar to a high frequency pulse, but the CRE reporter responded like slow GnRH pulses.

EXPERIMENTAL PROCEDURES
Materials-GnRH was purchased from Sigma. Anti-phosphorylated CREB (Ser-133) was purchased from Cell Signaling Technology (Beverly, MA). Rabbit polyclonal anti-CREB-1 and horseradish peroxidase-linked anti-rabbit antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). The FRET plasmid AKAR2 was from Dr. Roger Y. Tsien (University of California, San Diego); plasmids AKAR3 and indicator of cAMP using Epac were from Dr. Jin Zhang (The Johns Hopkins University); plasmid DAGR was from Dr. Alexandra Newton (University of California, San Diego), and GCaMP2 was from Dr. Junichi Nakai (National Institute for Physiological Sciences, Okazaki, Japan). DMEM and fetal bovine serum were purchased from Invitrogen. CRE and AP-1 reporter plasmids were purchased from Stratagene (La Jolla, CA). All other reagents were purchased from either Sigma or Fisher.
Cell Culture-L␤T2 cells were maintained in monolayer cultures in DMEM supplemented with 10% fetal bovine serum in a humidified 10% CO 2 atmosphere at 37°C. For FRET assay, cells were plated onto sterilized glass coverslips in 35-mm dishes coated with poly-lysine and grown to 50 -90% confluency in DMEM with 10% fetal bovine serum. Cells were then transfected with the FRET reporter plasmids with FuGENE-HD transfection reagent (Roche Applied Science) or electroporated using a Microporator at 1300 V, 20-ms pulse width, and 2 pulses (BTX/Harvard Apparatus, Holliston, MA) and allowed to grow for 24 -48 h before imaging.
cAMP Immunoassay-L␤T2 cells were placed in serumfree media with 0.1% bovine serum albumin for 16 h. GnRH (1-100 nM) was added for 5 min and then removed and incubated in DMEM with 0.1% bovine serum albumin for the indicated times. The cells were rapidly washed with ice-cold phosphate-buffered saline, and the intracellular cAMP content was determined using the cAMP direct Biotrack EIA kit (GE Healthcare).
Fluorescence Imaging-Cells on coverslips were washed twice with Hanks' balanced salt solution buffer with 25 mM HEPES and 1% glucose and were maintained in the dark at 37°C. Coverslips were mounted in a temperature-controlled perfusion cell on a Zeiss Axiovert microscope with a 40ϫ/1.3 NA oil-immersion objective lens. Typically, 3-8 fluorescent cells were analyzed in a single field. The intensity within selected cell regions of interest was measured in both cyan direct (excitation 440 and emission 480 nm) and FRET (excitation 440 and emission 535 nm) channels. Because yellow fluorescent protein is more photobleachable, the intensity within regions of interest was also measured in yellow direct channel (excitation 495 and emission 535 nm) to monitor photobleaching. Prolonged illumination was avoided to prevent photobleaching during the measurement. Exposure time was 30 -240 ms, and images were taken every 10 s and processed using the SimplePCI software. The ratios of cyan-to-yellow emission for ICUE or yellow-to-cyan emissions for AKAR3, AKAR2, and DAGR were calculated at different time points and normalized by dividing all ratios by the emission ratio just 5 min before stimulation.
PKA Kinase Assay-Samples were prepared as for the cAMP immunoassay. PKA kinase activity was determined by PepTag assay for nonradioactive detection of cAMP-dependent protein kinase assay kit as recommended by the manufacturer (Promega, Madison, WI).
CRE and AP-1 Reporter Assays-L␤T2 cells were maintained in 10-cm diameter dishes in DMEM-supplemented cell lysates with 10% fetal bovine serum at 37°C with 10% CO 2 . On the day before the transfection experiment, L␤T2 cells (3 ϫ 10 5 cells per well) were plated in 12-well plates (BD Biosciences). Each well was transfected with 500 ng of CRE or AP-1 reporter plasmid and 50 ng of tk-lacZ. The following day, the cells were switched to serum-free DMEM supplemented with 0.1% bovine serum albumin. After incubation for 24 h, the cells were treated with 1, 10, or 100 nM GnRH for 5 min every 30, 60, or 120 min for 6 h. Cell lysates were assayed directly for luciferase (Luciferin, Sigma) and ␤-galactosidase (Galacto-Light Plus, Tropix, Bedford, MA) activity according to the manufacturer's instructions in a 96-well plate using a Veritas microplate luminometer (Turner Biosystems, Sunnyvale, CA).
Western Blotting-L␤T2 cells were grown to confluence in 6-well plates, washed once with phosphate-buffered saline, and incubated in serum-free medium overnight. Cells were stimulated with agonists for various times at 37°C. Thereafter, cells were washed with ice-cold phosphate-buffered saline, lysed on ice in SDS sample buffer (50 mM Tris, 5% glycerol, 2% SDS, 0.005% bromphenol blue, 84 mM dithiothreitol, 100 mM sodium fluoride, 10 mM sodium pyrophosphate, and 2 mM sodium orthovanadate, pH 6.8), boiled for 5 min to denature proteins, and sonicated for 5 min to shear the chromosomal DNA. Equal volumes (30 -40 l) of these lysates were separated by SDS-PAGE on 10% gels and electrotransferred to polyvinylidene difluoride membranes (Immobilon-P, Millipore, Bedford, MA). The membranes were blocked with 5% nonfat dried milk in TBS/Tween (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% Tween 20). Blots were incubated with primary antibodies in blocking buffer for 60 min at room temperature and then incubated with horseradish peroxidase-linked secondary antibodies followed by chemiluminescent detection. For the phospho-specific antibodies, the polyvinylidene difluoride membranes were immediately stripped by placing the membrane in stripping buffer (0.5 M NaCl and 0.5 M acetic acid) for 10 min at room temperature. The membrane was then washed once for 10 min in TBS/Tween, re-blocked, and blotted with antibodies to the unphosphorylated form of the enzyme to control for equal protein loading.
Statistics and Mathematical Modeling-Statistics were analyzed by analysis of variance with Tukey post-tests, and FRET responses were fit to bell-shaped or exponential curves using Prism 4 (GraphPad Software, La Jolla, CA).
Bell-shaped curves were fit to Equation 1, where y max is the peak of the curve; y 1min and y 2min are the initial and final plateaus; x is the time; t1 ⁄ 2 1 and t1 ⁄ 2 2 are the midpoint of the curves; and n H 1 and n H 2 are the Hill slope factors. Singlephase exponential increases were fit to Equation 2, where y min and y max are the initial and final plateaus; x is the time, and k is the rate constant. Single-phase exponential decreases after an exponential increase were fit to Equation 3, where y min is the final plateau; x is the time; x 0 is the time of switching from an increase to a decrease; S is the difference between y(x 0 ) and y min ; k on is the on-rate constant, and k off is the off-rate constant. Two-phase exponential decreases were fit to Equation 4,

RESULTS
Pulsatile GnRH Stimulation Induces Intracellular Pulses of cAMP-We initially measured the cAMP response to a single pulse of GnRH. L␤T2 cells were starved for 16 h and then stimulated with increasing doses of GnRH (0.1, 1, 10, and 100 nM) for 5 min. The GnRH was removed; the cells were then harvested at different times, and cAMP was measured by radioimmunoassay. GnRH caused a dose-and time-dependent increase in cAMP (Fig. 1A). The peak GnRH effect was observed at 10 min, and cAMP levels returned to basal levels by 60 min. No increases in cAMP were observed at the lowest dose of 0.1 nM, but a significant increase was observed at 1 nM GnRH at 10 min. Higher amplitude GnRH pulses allowed increases in cAMP to be observed earlier, and the elevations in cAMP were maintained longer. To measure the cAMP response to GnRH in real time, we used an Epac-based FRET reporter ICUE (10). This reporter contains the cAMP-binding domain from Epac (amino acids 1-881) fused between enhanced cyan fluorescent protein (N-terminal) and the citrine variant of the yellow fluorescent protein (YFP, C-terminal). Binding of cAMP reorients the enhanced cyan fluorescent protein and citrine domains causing a change in FRET. L␤T2 cells were transfected with ICUE and then imaged over a period of 30 min using a temperature-and pH-controlled perfusion cell mounted on the fluorescence microscope. Cells were given increasing doses of constant GnRH (1, 10 and 100 nM) or a 5-min pulse of GnRH at the same concentrations. Constant GnRH treatment induced a dose-dependent transient increase in FRET that was maximal at 5 min and had returned to basal levels by 15 min (Fig. 1B). A similar response was obtained with a single pulse of GnRH (Fig. 1C). The dose response and time course are consistent with the measurement of cAMP levels by radioimmunoassay (Fig. 1A). Interestingly, with constant GnRH treatment the elevation of cAMP was only transient despite the continued presence of GnRH and, furthermore, did not to respond to a second GnRH pulse (data not shown). Cells transfected with ICUE were then imaged using multiple pulses of GnRH over 4 h. Five-minute pulses of 1, 10, or 100 nM GnRH were administered at intervals of 30, 60, or 120 min. A strong transient FRET signal was observed with each pulse irrespective of the pulse frequency, and the amplitude of the FRET signal did not diminish with subsequent pulses indicating that this pathway does not desensitize ( Fig. 1, D-F). The FRET signals for all the pulses were averaged and modeled mathematically to derive kinetic parameters for the response. The data were fit to bell-shaped curves, which were significantly different for each dose of GnRH (p Ͻ 0.0001). All GnRH doses gave maximal responses at 3 min despite the GnRH pulse continuing for 5 min (Fig. 1G). The higher doses of GnRH resulted in a faster activation of FRET (1 nM GnRH, half-life t1 ⁄ 2(on) ϭ 116 and t1 ⁄ 2(off) ϭ 288 s, slope factors n H(on) ϭ 0.011 and n H(off) ϭ 0.0056; 10 nM GnRH, t1 ⁄ 2(on) ϭ 69 and t1 ⁄ 2(off) ϭ 315 s, n H(on) ϭ 0.021 and n H(off) ϭ 0.0067; 100 nM GnRH, t1 ⁄ 2(on) ϭ 85 and t1 ⁄ 2(off) ϭ 199 s, n H(on) ϭ 0.012 and n H(off) ϭ 0.0027, respectively). Integration of the area under the FRET curve showed a significant dose-dependent increase in FRET (Fig. 1H). To investigate the origin of the rapid extinction of the cAMP signal, we tested whether phosphodiesterase inhibition would alter the response of the ICUE reporter. Cells expressing the ICUE reporter were imaged and stimulated initially with a pulse of GnRH and then with a pulse of GnRH in the presence of isobutylmethylxanthine to block cAMP hydrolysis (Fig. 1I). Phosphodiesterase inhibition caused an augmentation in the response but did not alter the kinetics, suggesting that phosphodiesterase activation is not responsible for the deactivation.
Other G s -coupled Receptors Do Not Show the Same Pulse Sensitivity-To test whether other G s -coupled receptors show the same kinetics, ICUE-expressing cells were imaged and stimulated with continuous GnRH, isoproterenol, or pituitary adenylate cyclase-activating polypeptide (PACAP) to activate GnRH and ␤-adrenergic or PAC1-R receptors, respectively, over 30 min. As a control, we stimulated cells with forskolin to activate adenylate cyclase directly. As before, GnRH caused a transient increase in FRET that was maximal at 3 min and then decreased to basal over 20 min ( Fig. 2A). In contrast, PACAP, isoproterenol, and forskolin gave a sustained FRET signal over 30 min ( Fig. 2A). We modeled the change in FRET mathematically. The GnRH response fit a bell-shaped curve (t1 ⁄ 2(on) ϭ 56 and t1 ⁄ 2(off) ϭ 231 s, n H(on) ϭ 0.011 and n H(off) ϭ 0.0025) similar to our earlier analysis. The isoproterenol data fit to a one-phase exponential increase (t1 ⁄ 2 ϭ 1.5 min, k on ϭ 0.0062), but the forskolin and PACAP data fit better to bell-shaped curves (forskolin, t1 ⁄ 2(on) ϭ 82 and t1 ⁄ 2(off) ϭ 842 s, n H(on) ϭ 0.0054 and n H(off) ϭ 0.0011, p Ͻ 0.0001; PACAP, t1 ⁄ 2(on) ϭ 129 and t1 ⁄ 2(off) ϭ 4933 s, n H(on) ϭ 0.0079 and n H(off) ϭ 1 ϫ 10 Ϫ4 , p Ͻ 0.0001). The onrates were similar for all agonists and likely reflect the diffusionlimited mixing of the medium containing the agonist in the chamber. With a flow rate of 0.05 ml/min and a chamber volume of 100 l, the expected t1 ⁄ 2(on) would be ϳ1 min. We also tested the FRET response to a 5-min pulse of the same agonists (Fig. 2B). All agonists gave transient increases in FRET that decayed with different off-rates. GnRH gave the expected pulse signal with a maximum at 3 min that fit to a bell-shaped curve (t1 ⁄ 2(on) ϭ 49 and t1 ⁄ 2(off) ϭ 311 s, n H(on) ϭ 0.0089 and n H(off) ϭ 0.0095), but the off-rate (n H(off) ϭ 0.0095 versus 0.0025 for constant GnRH) was significantly faster. As with the constant GnRH treatment, the FRET signal decreases after 3 min despite the continued presence of GnRH. The responses to isoproterenol and forskolin gave FRET signals that increased exponentially to a maximum at 5 min then decreased exponentially when the agonist was removed (isoproterenol, t1 ⁄ 2(on) ϭ 1.6 min and k on ϭ 0.007, t1 ⁄ 2(off) ϭ 0.96 min and k off ϭ 0.012; forskolin, t1 ⁄ 2(on) ϭ 0.96 min and k on ϭ 0.012, t1 ⁄ 2(off) ϭ 0.61 min and k off ϭ 0.019). The on-and off-rates are again consistent with the diffusion-limited mixing and suggest that these targets rapidly activate and deactivate as the agonist is applied and then removed. It also implies that activation of ␤-adrenergic signaling is only determined by the availability of ligand, unlike GnRH. The FRET response to a 5-min pulse of PACAP is more complicated (Fig. 2B). The increase in FRET is consistent with the data from constant PACAP treatment, but the decrease in FRET signal upon removal of PACAP followed a two-phase exponential decay (t1 ⁄ 2(on) ϭ 1.5 min and k on ϭ 0.0078, t1 ⁄ 2(off) 1 ϭ 0.64 min and k off 1 ϭ 0.018, t1 ⁄ 2(off) 2 ϭ 27 min and k off 2 ϭ 0.00043) with a rapid partial loss of signal followed by a much slower loss of the remaining signal.
Pulsatile GnRH Stimulation Induces Intracellular Pulses of PKA-We then investigated signaling downstream of cAMP by measuring protein kinase A activation. Initially, cells were given a single 5-min pulse of 100 nM GnRH and harvested at various times, and PKA enzyme activity was measured in cell extracts. A single pulse of GnRH caused a transient increase in PKA activity that was maximal by 15 min and returned to basal by 90 min (Fig. 3A). We then performed FRET using a novel protein kinase A activity reporter AKAR3 (11). This reporter contains an N-terminal enhanced cyan fluorescent protein followed by the phosphoamino acid binding domain from FHA1, a PKA substrate peptide sequence, and a circularly permuted variant of Venus YFP. Phosphorylation of the PKA target sequence allows the FHA1 domain to bind intramolecularly, and changes the protein conformation and increasing FRET. Cells were stimulated with increasing concentrations of constant GnRH (1, 10, and 100 nM) or a single 5-min pulse of GnRH at the same concentrations. Constant GnRH treatment induced a dose-dependent transient increase in FRET that was maximal at 5 min and had returned to basal levels by 15 min (Fig. 3B). A similar response was obtained with a single pulse of GnRH (Fig. 3C). As before, the FRET signal with tonic GnRH rapidly returned to basal levels despite the continued presence of GnRH. Cells transfected with AKAR3 were imaged over 4 h, and 5-min pulses of GnRH were administered at intervals of 30, 60, or 120 min as before (Fig. 3, D-F). A strong transient FRET signal was observed with each pulse, and the amplitude of the FRET signal did not diminish with subsequent pulses in agreement with the cAMP reporter ICUE (Fig. 1, D-F). The FRET signals for all the pulses were averaged and modeled mathematically to derive kinetic parameters for the response. The data were fit to bell-  shaped curves, which were significantly different for each GnRH dose (p Ͻ 0.001). All GnRH doses gave maximal responses at 3 min despite the GnRH pulse continuing for 5 min (Fig. 3G). The FRET signal showed similar kinetics with the three GnRH doses (1 nM GnRH, half-life t1 ⁄ 2(on) ϭ 116 and t1 ⁄ 2(off) ϭ 220 s, slope factors n H(on) ϭ 0.008 and n H(off) ϭ 0.0055; 10 nM GnRH, t1 ⁄ 2(on) ϭ 78 and t1 ⁄ 2(off) ϭ 220 s, n H(on) ϭ 0.012 and n H(off) ϭ 0.0046; 100 nM GnRH, t1 ⁄ 2(on) ϭ 87 and t1 ⁄ 2(off) ϭ 218 s, n H(on) ϭ 0.0099 and n H(off) ϭ 0.0027, respectively). Integration of the area under the FRET curve again showed a significant dose-dependent increase in FRET (Fig. 3H). These results indicated that activation of PKA is tightly coupled to cAMP second messenger generation.
Pulsatile GnRH Stimulation Induces Phosphorylation of CREB-We then tested a second protein kinase A activity reporter, AKAR2, that shows rapid activation but slower deactivation kinetics as the phosphorylated site is less accessible to phosphatases (12). The AKAR2 reporter is similar to AKAR3 but contains a citrine YFP instead of the Venus domain. Constant GnRH (1, 10, and 100 nM) produced a dose-dependent increase in FRET (Fig. 4A). The response was rapid and sustained over 30 min. A single pulse of GnRH at the same concentrations caused the same rapid transient rise in FRET, but the signal decreased more slowly and did not return to basal within 30 min (Fig. 4B). The response to multiple pulses of GnRH was also noticeably different from AKAR3. At a pulse interval of 120 min, a rapid rise in FRET and a slow decay were observed (Fig. 4C). The maximal signal was dependent on the GnRH dose, as was observed for the constant stimulation, but the signal decayed at a rate that was independent of GnRH dose (t1 ⁄ 2 ϳ45 min). At 1 nM GnRH, the FRET signal had returned to basal by 90 min and by 120 min at 10 nM GnRH, but it did not reach basal before the next pulse at 100 nM GnRH. At higher pulse frequencies, this sawtoothed pattern became compressed (Fig.  4D), and at the highest frequency with a pulse interval of 30 min, the stimulation approximated to constant stimulation (Fig. 4E). Interestingly, the multiple submaximal pulses did not cause a stepwise increase in signal suggesting that the pulses are not additive.
The finding of very different responses with the FRET reporters raised the question of the response of endogenous PKA targets. Many transcriptional responses to cAMP are mediated by PKA phosphorylation of the CREB at Ser-133 (14). Therefore, we examined the kinetics of CREB phosphorylation following a single 5-min pulse of GnRH. Cells were stimulated with increasing doses of GnRH (0, 0.1, 1, 10, and 100 nM) for 5 min and then washed and harvested at different times. CREB Ser-133 phosphorylation was measured by immunoblotting. CREB was rapidly phosphorylated at Ser-133 upon 5 min of GnRH pulse stimulation even at the lowest 0.1 nM dose of GnRH (Fig. 4F). At 1 nM GnRH, the rate of CREB dephosphorylation (t1 ⁄ 2 ϳ15 min) was intermediate between the two A-kinase activity reporters (Fig. 4G). At higher GnRH doses, the dephosphorylation was similar to the slow AKAR2 reporter (t1 ⁄ 2 ϳ60 min). Thus, we observed a rapid dose-independent phosphorylation of CREB with a slower GnRH dose-dependent dephosphorylation.
Pulsatile GnRH Stimulation Induces Intracellular Pulses of DAG-The GnRH-R also signals via the G q/11 family of G-proteins, so we measured the DAG response to a single pulse of GnRH. To measure the response in real time, we used a PKC-␤II-based FRET reporter DAGR (13). This reporter contains the DAG-binding domain from PKC-␤II between cyan fluorescent protein and YFP. The FRET signal with this reporter is weaker than with the previous reporters as the conformational change is caused by binding of the PKC-␤II C1 domain to the plasma membrane rather than an intramolecular conformational change. L␤T2 cells were transfected with DAGR and then imaged over a period of 30 min. Cells were given increasing doses of constant GnRH (1, 10, and 100 nM) or a 5-min pulse of GnRH at the same concentrations. Constant GnRH treatment induced a dose-dependent increase in FRET that was maximal at 5 min and was maintained for the 30-min perfusion (Fig. 5A). The response to a single pulse of GnRH was transient, reaching a maximum at 5 min and then returning rapidly to basal levels. This shows not only that the DAGR reporter responds quickly to changes in DAG, but also that binding of the reporter to DAG does not protect the DAG from being metabolized when GnRH is removed (Fig.  5B). Cells transfected with DAGR were then imaged using multiple pulses of GnRH over 4 h. Five-minute pulses of 10 nM GnRH were administered at intervals of 30, 60, or 120 min. A strong transient FRET signal was observed with the first pulse, but the amplitude of the FRET signal diminished with each subsequent pulse indicating that this pathway desensitizes (Fig. 5,  C-E). Desensitization was evident by 60 min after the first pulse, and no further pulses were observed after 120 -180 min.
Pulsatile GnRH Stimulation Induces Pulses of Intracellular Calcium-We also measured the calcium response in real time using a calmodulin-M13-based fluorescence reporter GCaMP2 (14). This reporter consists of a nonfluorescent split circularly permuted EGFP protein that folds to form a fluorescent protein in the presence of calcium. We and others have previously shown calcium increases, both acute spike and extended plateau phase, in L␤T2 cells using calcium dyes, but these dyes are not suitable for the extended perfusion studies to measure the response to multiple pulses. Cells were transfected with GCaMP2 and then imaged over a period of 30 min. Cells were given increasing doses of constant GnRH (1, 10, and 100 nM). Constant GnRH treatment induced a dosedependent increase in EGFP fluorescence that was maximal at 5 min and was maintained for the 30-min perfusion (Fig. 6A). The perfusion system is not rapid enough to see the acute spikephase of calcium release, so the GCaMP2 signal likely reflects the plateau-phase calcium increase. We do not see a calcium increase at the lowest GnRH concentration, but both 10 and 100 nM GnRH give robust calcium increases. Cells transfected with GCaMP2 were then imaged using multiple pulses of GnRH over 4 h. Five-minute pulses of 100 nM GnRH were administered at intervals of 30 or 60 min. A strong transient EGFP signal was observed with the first pulse, but the amplitude of the EGFP signal diminished rapidly with each subsequent pulse (Fig. 6, B and C). Desensitization was faster than for the DAG response, being evident by 30 min after the first pulse and complete by 60 min. The multiple pulses were repeated at 10 nM GnRH. At the lower GnRH concentration, only a single C-E, PKA activity in transfected cells was monitored by FRET in response to multiple GnRH pulses over a 4-h perfusion. Graphs show normalized FRET from cells perfused with GnRH at 1 nM (blue), 10 nM (green), or 100 nM (red) GnRH was for 5 min every 120 min (C), every 60 min (D), or every 30 min (E). Arrows indicate GnRH pulses. Data are the means of three to four independent assays. F, single GnRH pulse induces phosphorylation of CREB. Cells were stimulated with 0.1, 1, 10, or 100 nM GnRH for 5 min, then washed, and incubated for increasing times as indicated. Phosphorylation of CREB in whole cell extracts was assessed by immunoblotting with anti-phospho-CREB (Ser-133). G, qualification of phosphorylated CREB after a single pulse. Data are expressed as fold over basal (mean Ϯ S.D.) for four samples in each condition. Exponential decay curves were fit using Prism. Asterisks indicate significant differences in fitted curves (p Ͻ 0.05).
pulse of calcium was observed. As G q/11 is subject to proteasomal degradation, we measured the calcium response in the presence of MG-132 to inhibit the proteosome. Cells were imaged with pulses of 10 nM GnRH at 30-min intervals in the presence of MG-132 (Fig. 6F). Multiple calcium increases are now observed, but the pulses are still subject to desensitization,  suggesting that loss of G q/11 is only partially responsible for the desensitization.
Pulsatile GnRH Stimulation Induces Pulse-dependent CRE and AP-1-dependent Transcription-Having observed differences in the signaling response to multiple GnRH pulses, we then tested whether these differences would cause different transcriptional responses. CREB binds to the CRE that is found in many promoters. Therefore, we measured CRE-dependent transcriptional activation to assess the transcriptional response to pulsatile GnRH. CRE-luciferase was transfected into L␤T2 cells that were stimulated with 5-min pulses of GnRH (0, 1, 10, and 100 nM) every 30, 60, or 120 min over 6 h, a single GnRH pulse given at the start, or constant GnRH. At 10 and 100 nM GnRH, we observed a pulse frequency-dependent increase in luciferase activity (Fig. 7A). Activation of CRE-dependent transcription was greatest at the 30-min inter-pulse interval. At the lowest dose of GnRH, only the 30-min pulse interval caused a significant increase in luciferase activity. Interestingly, the response to constant GnRH was identical to the response to a single pulse in agreement with our signaling data.
Calcium-DAG signaling activates PKC to induce transcription via TPA/AP-1-response elements. So we measured AP-1dependent transcriptional activation. AP-1-luciferase was transfected into L␤T2 cells, which were stimulated with pulses of GnRH as before. At 10 and 100 nM GnRH, we observed a pulse frequency-dependent increase in luciferase activity (Fig. 7B). Activation of AP-1-dependent transcription was greatest at the 30-min inter-pulse interval. At the lowest dose of GnRH, we did not observe a significant increase in luciferase activity nor did we observe significant increases with a single pulse of GnRH or the lowest pulse frequency of 120 min at any concentration. Interestingly and unlike the CRE reporter, the response to constant GnRH was as great as the highest pulse frequency, which agrees with the FRET signaling data.

DISCUSSION
Even though pulsatility of GnRH is recognized as a major determinant for differential gonadotropin subunit gene expression and gonadotropin secretion, very little is known about the kinetics of the signaling circuits governing GnRH action in the pituitary. Most studies of GnRH signaling use maximal continual treatment and biochemical end points such as changes in second messengers or phosphorylation of downstream targets. Although some limited temporal information can be gleaned by taking multiple time points, this approach completely ignores the very dynamic changes in signaling following pulsatile GnRH stimulation. To address this issue, we have investigated GnRH signaling in real time in immortalized gonadotropes. Of the multiple signaling pathways activated downstream of the GnRH receptor, we studied activation of the G s -cAMP-PKA and G q/11 -DAG-calcium signaling pathways using fluorescence reporters. To our knowledge, this is the first study to document the dynamics of these responses to pulsatile GnRH in live gonadotrope cells over an extended period. It is important to note that our imaging system allows us to see the effects of pulsatile GnRH at physiological concentrations.
We observed distinct differences in the response of the two pathways to continual GnRH. Stimulation of the G s -cAMP-PKA pathway only resulted in a transient increase in cAMP and PKA activation that had returned to basal levels within 10 min, but the G q/11 -DAG-calcium pathway maintained the elevated second messenger levels for the entire period of perfusion (Fig.  8A). The signaling responses to multiple pulses of GnRH were also distinct. Multiple pulses of GnRH caused corresponding pulses of cAMP and PKA activation that did not diminish over the 4-h perfusion. In contrast, both DAG and calcium activation rapidly desensitized within a few pulses (Fig. 8B).
We were intrigued by the transience of the cAMP response. Inhibition of phosphodiesterases enhanced the cAMP increase but did alter the kinetics, suggesting that the transience was not due to induction of phosphodiesterase activity. This transience was only observed with GnRH stimulation as PACAP and isoproterenol gave the expected tonic increase in cAMP levels. The kinetics of the FRET response to a pulse of isoproterenol suggested that mixing of the agonist in the imaging chamber is the limiting factor in the response with on-and off-rates similar to direct forskolin stimulation of adenylate cyclase. The kinetics of the PACAP response were unexpected, but it probably reflects that PACAP has unusual binding properties as it associates nonspecifically with membranes through an ␣-helical domain (15). The very rapid extinction of the cAMP response when an agonist is removed is consistent with the measured in vitro GTPase activity (4 min Ϫ1 ) of G s (16).
We have previously shown that the cAMP response in the L␤T2 gonadotrope cell line is blocked by small peptides that uncouple G s , so we believe the observed cAMP signaling is mediated via G s (17). The observed transient kinetics for elevation of cAMP is consistent with a model in which the G s protein functionally re-engages only the unliganded GnRH-R. Studies on other receptors are consistent with this model. G s has been shown to pre-couple to the unliganded prostacyclin receptor, a predominantly G s -coupled receptor, but not with ␣ 2A -adrenergic receptors (18). Another study showed an increase in FRET between the ␣ 2A -adrenergic receptor and G␥ upon ligand stimulation with a decrease in FRET between G s and G␥ consistent with activation and release of G s (19).
The desensitization of the G q signal that we observed in L␤T2 cells has also been observed in ␣T3-1 cells. Pretreatment of ␣T3-1 cells with 100 nM GnRH for 1 or 2 h completely eliminated the subsequent inositol 1,4,5-trisphosphate response to increasing concentrations of GnRH (20). The GnRH-R only couples to G q/11 in ␣T3-1 cells, so the observed desensitization must be the result of G q/11 rather than G s signaling (4). We initially thought that this desensitization was related to the known proteolytic degradation of G q/11 (21,22); however, the desensitization is only partially prevented by MG-132, so other mechanisms must also be involved. Two independent reports have noted that the GnRH-R is not phosphorylated following GnRH stimulation and does not bind ␤-arrestins and G-protein-coupled receptor kinases (20,23), so classical G-proteincoupled receptor desensitization can be eliminated. A number of other negative feedback loops have been demonstrated for G q/11 signaling, including the induction of repressors of G q/11 signaling, including the RGS family of proteins or the phosphorylation and inhibition of phospholipase ␤1. For example, the inhibition of a standing outward K ϩ current by G q signaling in HEK293 cells is relieved by endogenous RGS proteins, and an RGS-insensitive G q protein impairs this recovery (24). Alternatively, phospholipase C␤1 and GDP-bound G q are stably associated in unstimulated PC12 and HEK293 cells (25). Activation and GTP loading of G q increase the affinity of binding by 2 orders of magnitude and activate PLC␤1 but do not change the association or localization. Interestingly, PKC␣ phosphorylation of PLC␤1 on serine 887 inhibits enzyme activity, but it is not known if it alters complex assembly with G q (26).
A recent publication reported the rapid translocation of NFAT-EFP to the nucleus in response to multiple pulses of GnRH in real time in transfected HeLa and L␤T2 cells (27). Although NFAT activation is thought to be dependent on the calcium activation of calcineurin, and indeed complete removal of calcium did inhibit the translocation, Armstrong et al. (27) did not observe desensitization of NFAT translocation. NFAT can also be activated by cAMP signaling in cardiomyocytes and osteoclasts, however, so the observed NFAT translocation may reflect cAMP rather than calcium signaling (28,29). The observed desensitization of G q/11 signaling is consistent with a published mathematical model for pulsatile LH secretion (30). The original model predicted pulses of LH secretion that did not desensitize but, allowing for calcium channel desensitization and receptor internalization, produced a model in which LH release steadily declined with each pulse (30). Our results would argue that receptor internalization is not occurring as cAMP/PKA responses do not desensitize, and moreover, we would argue that the mechanism of negative feedback is at the level of G q/11 -PLC as both DAG and calcium responses desensitize.
Many studies have shown the role of the phosphorylation rate for the initiation of signaling, but the importance of the dephosphorylation rate on the subsequent signal profile is underscored by comparing the two PKA-dependent reporters. One reporter has a very rapid dephosphorylation rate (t1 ⁄ 2 ϳ 1 min) resulting in a signal that matches the acute cAMP response, whereas the other has a slower dephosphorylation rate (t1 ⁄ 2 ϳ 45 min). This slower rate results in a sawtooth pattern of signaling that approximates to a constant signal at higher pulse frequencies. Dephosphorylation of kinase substrates depends on many factors, including the strength of the initial signal, the amount and location of the relevant phosphatase activity, the intramolecular accessibility of the phosphorylated residue, and the sequestration of the phosphorylated residues by binding proteins. All these factors will modulate the final response and determine the pattern of the propagated signal. For example, CREB dephosphorylation was very rapid following a pulse of 1 nM GnRH but slowed significantly with pulses of GnRH at 10 and 100 nM. This suggests that CREB phosphorylation will follow each GnRH pulse at low pulse amplitudes but will convert to a constant activation at higher pulse frequencies and amplitudes.
The significance of these signal responses for downstream transcriptional events is exemplified by the induction of the CRE-and AP-1-dependent reporters. Both CRE-and AP-1-dependent transcriptions are sensitive to GnRH pulse frequency with maximal induction seen at a 30-min pulse interval, but the response to continual GnRH is distinct. For the CRE-dependent reporter, tonic stimulation with GnRH gives the same response as a single GnRH pulse as would be expected from the FRET data. In contrast, the AP-1-dependent reporter showed a strong response to continual GnRH. Notably, the transcriptional response to low dose GnRH (1 nM) is only observed with cAMP-dependent transcription at the highest GnRH pulse frequency. These findings would suggest that the response of purely cAMP-dependent genes should be sensitive to pulse frequency and much greater than tonic GnRH. We analyzed a previous microarray dataset of pulse-regulated genes to identify any whose profile correlated with our CRE reporter (31). Using Pearson correlation (p Ͼ 0.9), the expression profiles of 23 genes correlated with the CRE reporter. Among these genes, two are particularly noteworthy. The Ngfi-A-binding protein Nab1 has been implicated in pulse sensing of the LH␤ promoter, and the MAD homolog Smad7 has been implicated in Fsh␤ promoter activity (31,32). We performed a similar analysis using the profile of the AP-1 reporter gene. Again using Pearson correlation (p Ͼ 0.9), 353 genes correlated with the AP-1 reporter. Among these genes, a number are known to, or have the potential to, regulate the gonadotropin genes, including Egr2, Crem, Ets2, Inh␤A, and Egr1 as well as the AP-1 subunits c-jun, JunB, Fra1, c-fos, and FosB (31,33). Most gene promoters are targets for multiple signaling pathways, so the final regulation may represent more than these two inputs, and it remains to be determined whether these genes are targets for DAG/Ca 2ϩ and cAMP signaling.
Pulsatile GnRH differentially regulates LH and FSH subunit genes, with faster frequencies favoring Lh␤ transcription and slower frequencies favoring Fsh␤. There is evidence that some of these transcriptional effects may be mediated by the G s and G q/11 pathways. cAMP increases GnRH stimulation of a rat Lhb-luciferase transgene in mouse pituitaries and the Lhb promoter in L␤T2 cells (34). Egr1 is essential for GnRH induction of Lh␤ and female fertility in vivo (35), and the Egr1 gene expression requires several distinct serum-response elements/ Ets elements and a cAMP-response element and is induced downstream of ERK (36). We have shown previously that both G s and G q/11 signals contribute to ERK activation in L␤T2 cells (17). There is also a connection between G s and G q/11 signaling and Fsh␤ expression. CREB serves to integrate signals for basal-and GnRH-stimulated transcription of the rat FSH␤ gene (37), and the GnRH-responsive element contains a partial CRE site that binds CREB. Others have shown that GnRH induces binding of the AP-1 complex to the Fsh␤ promoter, and mutation of the AP-1 site reduces GnRH induction (33).
We conclude that the G s -cAMP-PKA and G q/11 -DAG-calcium pathways downstream of the GnRH-R are very sensitive to GnRH pulse frequency and amplitude and may underlie some of the known differential effects of GnRH on the pituitary gonadotrope. Further studies will be needed to unravel how the distinct patterns of second messenger generation observed here transduce signals to regulate gonadotropin expression and secretion.