Ahmed Ezzat, Alda Pereira, and Iain J. Clarke. Department of Physiology, Monash University, Clayton, Victoria 3800, Australia

ORIGINAL RESEARCH Kisspeptin Is a Component of the Pulse Generator for GnRH Secretion in Female Sheep But Not THE Pulse Generator Ahmed Ezzat, Alda Pereira, and Iain J. Clarke Department of Physiology, Monash University, Clayton, Victoria 3800, Australia We tested the hypothesis that kisspeptin cells constitute the pulse generator for GnRH secretion. In ewes, we determined whether iv administered kisspeptin elicits a secretory pulse of LH in anaesthetized, sex-steroid suppressed ovariectomized ewes. A response was seen in both anaesthetized and conscious animals, which was not associated with induction of c-fos labeling in GnRH cells, supporting the notion that kisspeptin acts on the neurosecretory GnRH terminals. Response was lower in the anaesthetized animals, suggesting that some nonkisspeptin elements may be involved in GnRH responses. Microinjection of kisspeptin (100 nmol) into the median eminence of conscious ewes elicited a pulse of LH, indicating that kisspeptin acts at this level to cause GnRH secretion. To determine which cells are activated at the time of GnRH secretion, we blood sampled 18 ewes during the luteal phase of the estrous cycle and harvested brains after 3 hours. Three of these ewes displayed a pulse of LH within 30 minutes of euthanasia. An increase in c-fos labeling was seen in kisspeptin and glutamate cells of the arcuate nucleus but not in GnRH neurons, preoptic kisspeptin neurons, or preoptic glutamate neurons. Immunohistochemistry in 4 hypothalami showed that 72% of arcuate kisspeptin cells receive glutamatergic input. These data support the concept that the kisspeptin cells of the arcuate nucleus drive pulsatile secretion of GnRH at the level of the median eminence, but this may involve upstream input from glutamate cells. We conclude that the pulse generator for GnRH secretion involves more than 1 element. (Endocrinology 156: 1828 1837, 2015) Pulsatile secretion of GnRH from the median eminence drives the pulsatile secretion of LH from the pituitary gonadotropes (1) and supports the synthesis and secretion of FSH (2). GnRH neurons exhibit inherent phasic patterns of electrical activity (3) and secretion (4, 5), but it has long been held that the generation of secretory episodes is due to a neural pulse generator. Early work indicated phasic patterns of multiunit activity (MUA) in the basal hypothalamus of the nonhuman primate (6), and this was taken to represent the pulse generator. Whether this concept relates to a modulator of the phasic activity of GnRH neurons, or an actual generator of pulses per se, is a matter of debate. It has been suggested that kisspeptin neurons in the arcuate nucleus are pulse generators because phasic MUA in the basal hypothalamus of the goat occurs close to these cells (7), but this does not provide cause-and-effect evidence. Certainly, kisspeptin is a potent stimulus for the secretion of GnRH into the hypophysial portal blood system (8), and a large body of evidence now exists to indicate that kisspeptin neurons play an important role in the control of GnRH secretion as recently reviewed (9). Li et al (10) examined whether kisspeptin is the pulse generator in rats, making injections of kisspeptin or kisspeptin antagonist into the preoptic area (POA) or the arcuate nucleus. Kisspeptin stimulated a pulse secretion of LH when injected into either site, but the antagonist was effective in reducing the pulse frequency when injected into the arcuate nucleus only. It was concluded that kisspeptin signaling in the mediobasal hypothalamus is responsible for GnRH pulse generation. The study had the ISSN Print 0013-7227 ISSN Online 1945-7170 Printed in U.S.A. Copyright 2015 by the Endocrine Society Received September 15, 2014. Accepted February 10, 2015. First Published Online February 24, 2015 Abbreviations: acsf, artificial cerebrospinal fluid; CV, coefficient of variation; DAB, diaminobenzadine; KNDy, kisspeptin/neurokinin B/dynorphin; MUA, multiunit activity; NMDA, N-methyl-D-aspartate; OVX, ovariectomized; PFA, paraformaldehyde; POA, preoptic area; vglut2, vesicular glutamate transporter 2. 1828 endo.endojournals.org Endocrinology, May 2015, 156(5):1828 1837 doi: 10.1210/en.2014-1756

doi: 10.1210/en.2014-1756 endo.endojournals.org 1829 limitation that large (400 500 L) injection volumes were used, and the site of action of kisspeptin was not defined. In other work (11), kisspeptin elicited secretion of GnRH from mediobasal hypothalamic explants, which was blocked by tetrodotoxin, suggesting kisspeptin action on secretory terminals. In another study (12), systemic injection of kisspeptin to mice stimulated LH secretion without c-fos protein being detectable in GnRH neurons, reinforcing the notion that pulsatile secretion of GnRH was effected by kisspeptin at the level of the GnRH secretory terminals. Further support for this concept came from in vivo studies, where dialysis of kisspeptin into the median eminence of nonhuman primates elicited GnRH secretion (13). We found that kisspeptin elicits GnRH secretion when administered to the isolated ovine median eminence in vitro (8). In sheep (14), iv injections of kisspeptin evoked a discharge of GnRH into the hypophysial portal blood within 2 minutes, arguing very strongly that the stimulatory peptide acts at the level of the GnRH neurosecretory terminals in the median eminence, outside the blood-brain barrier. Most recently, Choe et al (15) used a destabilized luciferase reporter sequence driven by the GnRH promoter and showed, in brain slices, that kisspeptin may synchronize the activity of GnRH neurons at the level of the cell bodies. These data suggest that kisspeptin acts as a pulse generator by coordinating the phasic pattern of activity in sets of GnRH neurons, but how this relates to secretion from the terminals in the median eminence is not clear. In sheep, kisspeptin cells are up-regulated by gonadectomy, and this is corrected by steroid replacement (16). The effect of ovariectomy is seen after 24 hours (17) and, although the increased activity of the kisspeptin neurons suggests an association with GnRH/LH secretion in the ovariectomized (OVX) female, this does not necessarily indicate that the kisspeptin cells are the pulse generators. Accordingly, we sought to ascertain whether the function of kisspeptin neurons is temporally associated with GnRH/LH pulses in the ewe. Using anesthesia to block GnRH secretion, we administered kisspeptin iv to provide in vivo evidence of the action of kisspeptin on neurosecretory GnRH terminals. In addition, we obtained further evidence that kisspeptin acts to generate pulses of GnRH secretion at the level of GnRH secretory terminals within the median eminence. Materials and Methods Animals and experimental procedures The experiments were performed on 3- to 4-year-old Corriedale ewes with a mean ( SEM) body weight of 62 1.0 kg, maintained under natural lighting and temperature. The experiments conformed to the National Health and Medical Research Council/Commonwealth Scientific and Industrial Research Organization/Australian Animal Commission Code of Practice for the Care and Use of Animals for Experimental Purposes and were approved by the Animal Ethics Committee of Monash University. Ovariectomies were carried out as previously described (18). Experiment 1. Pilot study of the effect of ketamine/thiopentone anesthesia on pulsatile LH secretion Six OVX ewes received a jugular venous cannula (Dwellcath; Tuta Laboratories), and anesthesia was induced with 15 mg/kg iv of either ketamine or thiopentone (Jurox Pty Ltd) or both anesthetics in combination (n 2/group). Anesthesia was maintained for 3 hours, with monitoring of heart rate, respiration rate, eye reflex, and digital reflex, and deep anesthesia was sustained with 3-mg injections of the anesthetic as required. Thiopentone is a barbiturate that has multiple mechanisms of action, including those via glutamate and gamma-aminobutyric acid a receptors, whereas ketamine acts via N-methyl-D-aspartate (NMDA) and nicotinic acetylcholine receptors (19). Blood samples (5 ml) were taken at 10-minute intervals for 3 hours before anesthesia and for 3 hours during anesthesia. Plasma was harvested and held at 20 C for LH assay using an established RIA (20). The pulsatile secretion of LH was taken as a reflection of the pulsatile secretion of GnRH (1). Pulse analysis was as previously described (21). Experiment 2. LH responses to kisspeptin-10 in anesthetized animals This experiment aimed to determine whether inhibition of neural function with a combination of ketamine and thiopentone anesthesia and sex steroid suppression of GnRH/LH secretion prevents stimulation of GnRH/LH secretion when kisspeptin is administered iv. All animals were OVX. Progesterone treatment was via intravaginal controlled internal drug release devices containing 0.3 g of progesterone (Artificial Breeders Ltd). Sustained estradiol-17 treatment was achieved with 3-cm sc implants of crystalline estradiol-17 (Sigma) packed into SILASTIC tubing (inside diameter 3.35 mm, outside diameter 4.65 mm, SILAS- TIC; Dow Corning Corp) implanted in the axillary region for 1 week. These implants produce circulating levels in adult ewes of 2 3 pg/ml (22). Progesterone levels in animals treated with a controlled internal drug release devices were 1.8 0.1 ng/ml (n 13), measured in a single RIA (23) with a sensitivity of 0.3 ng/ml. Intraassay coefficient of variation (CV) for this assay averaged 8.5% (n 10 assays) and was less than 10% between 4.7 and 15.1 ng/ml. Four groups (n 4) were included as follows: 1) anesthetized (as in experiment 1, above), administered 2-mL saline iv (vehicle); 2) anesthetized, administered 10 nmol of human kisspeptin iv (YNWNSFGLRY-NH 2 ; Phoenix Pharmaceuticals Ltd); 3) conscious, administered vehicle; and 4) conscious, administered kisspeptin. The animals received a jugular venous cannula, and blood samples (5 ml) were collected at 10-minute intervals for 3 hours and were then either anesthetized or remained conscious until the end of the experiment. After 4 hours, kisspeptin/vehicle was injected iv, and the animals were killed 40 minutes later. For col-

1830 Ezzat et al The GnRH Pulse Generator Endocrinology, May 2015, 156(5):1828 1837 lection of the brains, the animals received 25 000-U heparin iv and an overdose (2 g) iv of sodium pentobarbital (Lethabarb; Virbarc). Heads were perfused bilaterally via the carotid arteries with1lof0.9% sodium chloride followed by 1.5 L of 4% paraformaldehyde (PFA) in 0.1M phosphate buffer (ph 7.3), followed by 0.5 L of 0.5% PFA containing 30% sucrose. After removal of the brains, the hypothalami were dissected. The margins of the dissection were 5 mm rostral to the optic chiasma, 3 mm lateral to the longitudinal fissures and caudal to the mammillary bodies, and the tissue was taken to a depth of 10 mm. The tissues were stored for 7 days in PFA/30% sucrose in 0.1M phosphate buffer at 4 C and then at 20 C. Coronal sections (20 m) were cut on a freezing microtome and stored in a cryopreservative solution (30% ethylene glycol, 1% polyvinylpyrrolidone, and 30% sucrose in sodium phosphate buffer) at 20 C until processed for immunohistochemistry (see below). Plasma was harvested from the blood samples and stored at 20 C for LH assay. For each sheep, 3 sections were selected to represent the POA using the organum vasculosum of the lamina terminalis, the anterior commissure and the supraoptic recess of the third ventricle as landmarks. These were processed to visualize GnRH neurons and c-fos immunoreactivity in the same sections. Immunohistochemistry was as follows: free-floating sections were washed in 0.1M PBS and incubated in 0.1% sodium borohydride, 3% H 2 O 2, and blocking serum (PBS, 10% normal goat serum, and 0.3% Triton X-100). A rabbit polyclonal antibody raised against c-fos (1:20 000, 3 d at 4 C, catalog number sc-52; Santa Cruz Biotechnology, Inc) was applied to the sections and labeling was visualized with a biotinylated goat anti-rabbit secondary antibody (1:400; Vector) and the avidin-biotin complex streptavidin-horseradish peroxidase (Vector) and nickel-enhanced diaminobenzadine (DAB) (Sigma-Aldrich). Then GnRH immunolabeling was performed with the Benoit LR1 rabbit antibody (1:8000 dilution, 3 d at 4 C). Biotinylated goat antirabbit secondary antibody, avidin-biotin complex streptavidin horseradish peroxidase, and DAB were used to reveal brown cytoplasmic staining. Sections were mounted onto gel-coated slides, left to dry overnight and then coverslipped. GnRH cells that were either negative or positive for c-fos immunoreactivity were counted by an individual blind to the condition of the animal. Between 20 and 63 GnRH cells were counted per animal in the 3 POA sections. To obtain an index of the LH response to vehicle or kisspeptin, the average level of LH in plasma in the 40-minute period before injection was calculated for a baseline, and the average levels in the 40 minutes after injection were taken as the response. The response was determined statistically by comparing response and basal levels by paired t test. Experiment 3. In vivo kisspeptin administration to the median eminence This experiment was designed to test the hypothesis that kisspeptin acts within the median eminence in vivo, to stimulate GnRH/LH secretion in sheep as in other species (see the introductory section). Sheep (n 5) received guide tubes (19 gauge) for microinjection that were placed 2 mm above the floor of the third ventricle (Supplemental Figure 1A). Surgical placement was essentially as previously described (24, 25), but x-ray images were processed with a digitized CR30-X Agfa platform (Agfa Healthcare) that allowed accurate measurement of the distance from the tip of the guide tube to the surface of the median eminence (Figure 1A). The animals were allowed to recover for one month and then received a jugular venous cannula for blood sampling; samples (5 ml) were taken at 10-minute intervals for 360 minutes. Microinjection (100 nl) of artificial cerebrospinal fluid (acsf) (vehicle) was given after 120 minutes and of 100 Figure 1. Examples of the effect of thiopentone (Thio) (A) or ketamine (Ket) (B) or a combination of the 2 anesthetics (C) on plasma LH concentrations in OVX ewes., identified pulses.

doi: 10.1210/en.2014-1756 endo.endojournals.org 1831 nmol of kisspeptin was injected at 320 minutes. The injections were made 3 mm past the tip of the guide tubes (1 mm into the body of the median eminence), using a 1- L syringe (SGE) with a 23-guage needle of 115-mm length, with a needle plunger and a beveled tip. In pilot trials, it was found that injection of saline caused a pulse secretory episode of LH, which was presumed to be due to disruption of the GnRH terminals in the external zone of the median eminence. When repeated 2 weeks later, this was not seen and the experiment proceeded on the understanding that scar tissue had been created in the area of the injections (see Supplemental Figure 1B). No responses to vehicle injection were seen in the animals of the main experiment (see Results). Plasma was harvested from the blood samples and stored at 20 C for LH assay. After the experiment, injections of 50 nl of dye (Chicago Sky Blue) was injected at the same depth as for acsf/kisspeptin, and the brains of the animals were perfused (as above) 1 day later. The tissues were processed and stored frozen (as above) until cut at 40 m. The injection sites were visualized by locating the dye in coverslipped unstained sections. Experiment 4. Activation of kisspeptin and/or glutamate cells at the time of a natural GnRH/LH pulse and the relationship between kisspeptin and glutamate cells This experiment tested the hypothesis that kisspeptin cells and/or glutamate cells are activated at the time of a natural GnRH/LH pulse. Eighteen sheep were sampled during the midluteal phase of the estrous cycle. This phase of the cycle was chosen because the frequency of GnRH/LH pulses is lowest at this time and distinct pulses are clearly defined (26). Each sheep received a jugular venous cannula, and blood samples (5 ml) were taken every 10 minutes for approximately 3 hours, at which time the animals were killed and the brains perfused to recover hypothalami (as above). Plasma samples were assayed for LH, and 3 animals were identified as having an LH pulse within 30 minutes of the time of brain collection. Three other animals, which showed no evidence of an LH pulse within 1.5 hours of brain collection, were also taken for analysis. The remaining brains were used for other purposes. Hypothalamic blocks were processed for immunohistochemistry (as above). For kisspeptin/c-fos immunocytochemistry, sections representing the rostral, middle, and caudal regions of the arcuate nucleus and the lateral POA (1 for each division of the nucleus) were chosen from each ewe and mounted on SuperFrost slides. Immunocytochemistry was performed as previously described (27). The c-fos-labeled cells (as above) were visualized with nickel-enhanced DAB (Vector). A rabbit polyclonal antibody against ovine kisspeptin-10 (AC566; gift from Prof A Caraty, Institut National de la Recherche Agronomique, France) was used at a dilution of 1:20 000 for 72 hours at 4 C. This antibody was validated previously for use on sheep tissues (28, 29). Kisspeptin cells were visualized with DAB and counted by an individual blind to the condition of the animal. The total number of kisspeptin cells was recorded as well as those containing c-fos, and percentage of colabeled cells was calculated for each ewe. Between 68 and 438 kisspeptin cells were counted per animal in the arcuate nucleus and 91 and 179 cells were counted per animal in the POA. The percentage of kisspeptin/c-fos cells per ewe in each region was averaged to produce a mean ( SEM). Glutamate cells were immunostained using a guinea pig antivesicular glutamate transporter 2 (vglut2) antibody (a gift from Dr Kaneko, Kyoto University, Japan) (30). This stains glutamatergic neurons and nerve terminals in the ovine brain as previously described (31). Three sections representing the rostral, middle, and caudal regions of the arcuate nucleus were chosen from each ewe and mounted on SuperFrost slides. Antigen retrieval was performed with 0.01M citrate buffer (ph 6) in a microwave oven at 1000 W (2 5 min). Sections were then washed in Tris-buffered saline and incubated with blocking solution (as above) at room temperature. The primary antibody was used at a dilution of 1:1000 and applied for 72 hours at 4 C, and visualization of staining was with goat antiguinea pig secondary antibody (dilution 1:500; Molecular Probes) for 1 hour. c-fos immunohistochemistry was as above. For glutamate, 266 782 cells per animal were counted in the arcuate nucleus, and 104 266 cells were counted per animal in the POA. The percentage of glutamate/c-fos cells in each region was averaged per ewe. The results of this experiment showed increased c-fos labeling in both kisspeptin and glutamate cells in the arcuate nucleus, so we undertook further studies to determine the extent to which kisspeptin cells in this region receive glutamatergic input, using the VGluT2 and kisspeptin antisera described above. Three sections representing the rostral, middle, and caudal regions of the arcuate nucleus were chosen from each ewe. Antigen retrieval was performed as above and the 2 primary antisera were applied for 72 hours at 4 C. Slides were then washed in Tris-buffered saline and sections incubated with goat antirabbit conjugated to Alexa Fluor 594 at 1:500 and goat antiguinea pig conjugated to Alexa Fluor 488 at 1:500 (Molecular Probes) for 1 hour. Sections were counterstained with 0.3% Sudan Black B. After further washes, coverslips were applied with fluorescence mounting medium (Dako). All kisspeptin cells in the arcuate nucleus in the 3 sections of each animal were counted and close appositions determined by Apotome microscopy with optical rotation (Zeiss AxioVision). A mean percentage ( SEM) was then calculated for close appositions of glutamatergic terminals on arcuate kisspeptin cells. LH assays Plasma LH levels were measured as previously described (20), with a sensitivity of 0.1 ng/ml. For 10 assays, the interassay CV ranged from 6.6% to 10.1% for 4 pools that ranged from 3.6 to 20.5 ng/ml. Maximum precision was between 2.7 and 4.8 ng/ ml, and the intra-assay CV was less than 10% between 0.51 and 14.7 ng/ml. Statistical analysis Data are presented as mean SEM and analyzed by Student s paired or unpaired t test as appropriate. When the hypothesis indicated a one-way analysis, this option was chosen in the analysis. Results Experiment 1 Either thiopentone or ketamine reduced pulsatile GnRH/LH secretion as shown in representative animals

1832 Ezzat et al The GnRH Pulse Generator Endocrinology, May 2015, 156(5):1828 1837 Figure 2. Effect of kisspeptin (A) or saline (B) on plasma LH concentrations in estradiol and progesterone-treated, OVX ewes that were anesthetized with thiopentone and ketamine combined. (Figure 1, A and B), although occasional pulses were seen during anesthesia with single agents (Figure 1A). Combined ketamine/thiopentone anesthesia blocked GnRH/LH secretion in OVX ewes (Figure 1C). Experiment 2 Taking account of the results of experiment 1, this experiment used combined anesthesia and estradiol and progesterone treatment of OVX ewes. This generally held LH levels below 1 ng/ml, although occasional small pulses were seen (Figure 2). To analyze response to treatment or vehicle, basal levels were averaged over 40 minutes before injection and compared with levels over 40 minutes after injection (response) by Student s paired t test. Kisspeptin caused a significant rise in LH levels conscious (P.0001) and anesthetized (P.05) animals, which was not see in the vehicle-treated animals (Figure 3, A and B, respectively). Comparison by unpaired Student s t test showed that the response was greater (P.05) in conscious animals than in anesthetized animals (Figure 3C). Kisspeptin treatment did not increase c-fos labeling in GnRH cells in either conscious or anesthetized animals (Figure 3D). Experiment 3 Microinjections were made into the body of the median eminence, and the localization of this small volume was verified by injection of dye as shown in Supplemental Figure 1. The dye was restricted to the scar caused by entry of the injection needle placed in the median eminence. Kisspeptin microinjection increased the secretion of LH, indicative of a pulse of GnRH (Figure 4). The average percentage increase in LH levels was 314 46 (range 179 480; P.036 vs acsf, by one-way Student s t test). Experiment 4 Examples of LH levels in animals that were killed at the time of a natural pulse or when no pulse was apparent are shown in Figure 5. Table 1 shows the c-fos labeling in the cell types examined, with examples in Figures 6 and 7. There was no difference between pulse / no pulse animals in c-fos labeling of either GnRH or kisspeptin cells in the POA. There was a highly significant (P.002) increase in c-fos labeling of kisspeptin cells in the arcuate nucleus in the pulse animals and a significant (P.012) increase in c-fos labeling in glutamate cells in the arcuate nucleus (but not in the POA) of pulse animals. Close apposition of vglut2-labeled varicose fibers and kisspeptin cells in the arcuate nucleus was observed in 72 2.5% of cases (n 4 ewes), and examples of such appositions are seen in Figure 8. Figure 3. Mean ( SEM) plasma LH responses (Resp) over basal levels to iv injection of kisspeptin or saline in conscious (A) or anesthetized animals (B) and Fos labeling of GnRH cells in the same animals (n 4/group). Average plasma LH concentrations were calculated for each animal for 40 minutes before and 40 minutes after the administration of either kisspeptin or saline, and a paired Student s t test was applied for either kisspeptin- or saline-treated groups. C, Comparison of the mean responses in conscious and anesthetized animals by unpaired Student s t test. D, Number of GnRH neurons that displayed c-fos labeling in each group of animals. *, P.05; ***, P.0001 compared with levels before injection; #, P.05 compared with conscious. Discussion The present results substantiate the notion that kisspeptin acts on the neurosecretory terminals of GnRH neurons in the median eminence to elicit pulsatile secretion of GnRH.

doi: 10.1210/en.2014-1756 endo.endojournals.org 1833 Figure 4. Example of plasma LH response to microinjection (100 nl) of either acsf or 100 nmol kisspeptin. Group data are found in text. Figure 5. Plasma LH profiles of luteal phase animals that were killed either within 10 30 minutes of a naturally occurring secretory episode (pulse) or not (no pulse)., a significant pulse. Variations in the time of the end of sampling is due to staggered kill times. The so-called kisspeptin/neurokinin B/dynorphin (KNDy) cells of the arcuate nucleus project into the median eminence (8) and are activated at the time of a natural secretory episode of GnRH/LH (experiment 4). The present data provide compelling further evidence that the KNDy cells are fundamental elements of the GnRH pulse generator. Nevertheless, glutamate cells in the arcuate nucleus of the hypothalamus are also activated at the time of pulse secretion, and 72% of KNDy cells receive glutamatergic input. Accordingly, the KNDy cells constitute a component of the pulse generator for GnRH secretion, but glutamatergic elements in the arcuate nucleus may also be an integral component of the same. The role of kisspeptin in the production of pulses of secretion of GnRH seen here is consistent with earlier results showing the effect of the kisspeptin receptor antagonist peptide 234 into the arcuate nucleus, which reduced pulse frequency in OVX rats (10). This prompted the conclusion that kisspeptin/kisspeptin receptor 54 signaling is a critical neural component of the pulse generator and the present work extends this concept. As indicated in the introductory section, work in mice and nonhuman primates shows that kisspeptin application to the median eminence elicits a pulsatile secretory episode of GnRH and LH. The application of kisspeptin to fragments of mediobasal hypothalamus from mice (11) is not unequivocal proof that the agent is acting on the median eminence GnRH terminals, because some neuronal cell bodies may have been included in the fragments. On the other hand, the push-pull methodology used in conscious nonhuman primates provide convincing data of kisspeptin action within the median eminence (13), showing that approximately 75% of pulses of kisspeptin release measured in median eminence perfusates were associated with GnRH pulses. We have substantiated the notion that kisspeptin acts within the median eminence, because iv injection elicited an LH response that reflects a pulse of GnRH secretion (1), when administered to conscious or anesthetized animals or when kisspeptin was applied directly into the median eminence. We suppressed GnRH/LH secretion by a combination of anesthesia and gonadal steroids to provide further evidence of the action of kisspeptin on the GnRH terminals. We did not observe any increase in c-fos labeling in GnRH cells with iv kisspeptin injection in either conscious or anesthetized animals. Nevertheless, the response in anesthetized, estradiol/progesterone-treated animals was lower than that in intact animals, which suggests that some component GnRH pulse generation relies on neural function other than kisspeptin cell activation. This could be nitric oxide or some other moiety that is released from the KNDy cells that project to the external zone of the median eminence. Certainly, kisspeptin cells act on nitric oxide producing neurons in the POA, which express the kisspeptin recep-

1834 Ezzat et al The GnRH Pulse Generator Endocrinology, May 2015, 156(5):1828 1837 Table 1. c-fos Labeling of GnRH, Kisspeptin, and Glutamate Cells Cell type/location No pulse (n 3) Pulse (n 3) Significance GnRH, POA 29.7 8.0 34.5 4.3 NS Kisspeptin, arcuate 34.4 6.1 74.9 2.5 0.002 nucleus Kisspeptin, POA 5.0 0.6 4.3 1.7 NS Glutamate, arcuate 5.8 0.5 27.6 3.6 0.014 nucleus Glutamate, POA 5.9 0.6 5.9 0.7 NS c-fos labeling of GnRH, kisspeptin, and glutamate cells in the POA and/or arcuate nucleus in luteal phase ewes (experiment 4) humanely killed when undergoing a pulse secretion of LH or not doing so (no pulse). The values are means SEM percentage of cells of each type that were immunolabeled for c-fos. Significance between the 2 groups was tested by Student s unpaired t test. Figure 6. Examples of c-fos labeling in kisspeptin cells in luteal phase animals that were sampled within 10 30 minutes of a natural LH secretory episode (pulse) or not (no pulse). Lefthand side panels show low power images of the arcuate nucleus (magnification bar, 100 m), and boxed areas are shown at higher power on the right-hand side (magnification bar, 20 m). Arrowheads indicate kisspeptin cells not colabeled for c-fos, and arrows indicate kisspeptin cells colabeled for c-fos. tor (32). Strangely, however, although KNDy cells also contact nitric oxide containing elements, the latter do not express the kisspeptin receptor. Neither is there any evidence of kisspeptin receptors in the elements of the neurosecretory zone of the median eminence. Thus, the means by which kisspeptin and/or nitric oxide may cause GnRH secretion is not clear. Regarding nitric oxide, the intriguing possibility is that it may act to suppress secretion rather than stimulate the same (33). There are indications that pulsatile GnRH secretion can occur attendant with activation of the GnRH cell bodies, as in the ram effect. Thus, introduction of rams to previously isolated anestrous ewes causes an immediate pulsatile LH secretory response (34). When we measured c- Fos labeling in hypothalamic neurons 3 hours later, activation of both KNDy cells in the arcuate nucleus and GnRH cells in the POA was observed (35). This suggests that the response involved activation of GnRH cell bodies, which are not directly contacted by KNDy cells (36, 37). Relevant to this, it is noted that subpopulations of GnRH neurons exist, some responsive to kisspeptin and others to metabotropic glutamate receptor agonists (38). Other data also indicate that pulse generation in GnRH neurons can involve kisspeptin action at the level of the perikaryon. In this respect, use of a murine model in which a destabilized luciferase reporter was introduced into GnRH cells allowed demonstration of activation of cyclic activity at the level of the cell bodies, as observed in brain slices (15), which was synchronized by kisspeptin. It cannot be discounted, therefore, that GnRH secretion may, in some circumstances and in some species, be controlled at the level of the GnRH cell bodies. Kisspeptin injection (iv) did not increase c-fos labeling in GnRH cells, but it is possible that signaling could occur via the cell bodies without c-fos activation. KNDy cells were activated at the time of a natural GnRH/LH pulse in luteal phase ewes. This is consistent with an increase in activity of the KNDy cells 24 hours after ovariectomy (17) and increased kisspeptin gene expression in OVX ewes (16). Another line of investigation has indicated that MUA in the basal hypothalamus of the goat exhibits the same temporal phasic pattern as pulsatile LH secretion (7). Although it was suggested that this reflects function of KNDy neurons, other neural elements could also contribute to the MUA. Although this indicated correlation of KNDy cell function and GnRH/LH secretion, direct cause-and-effect was not shown. Kisspeptin also modulates the GnRH pulse generator in female rats (39). The present data strongly suggest that the activity of KNDy cells is directly related to the pulsatile secretion of GnRH, as indicated for the nonhuman primate (5, 13). Although these data are correlative in the sense that c-fos activation in KNDy and glutamate cells is contemporaneous with the pulsatile secretion of GnRH/LH, the highly significant statistical result indicates a strong link between cell activation and secretion. As mentioned

doi: 10.1210/en.2014-1756 endo.endojournals.org 1835 Figure 7. Examples of c-fos labeling in glutamate cells in luteal phase animals that were sampled within 10 30 minutes of a natural LH secretory episode (pulse) or not (no pulse). Lefthand side panels show low power images of the arcuate nucleus (magnification bar, 100 m), and boxed areas are shown at higher power on the right-hand side (magnification bar, 20 m). Glutamate cells were identified with an antibody against vglut2. Arrowheads indicate glutamate cells not colabeled for c-fos, and arrows indicate glutamate cells colabeled for c-fos. Figure 8. Examples of glutamate (vglut2 labeled) (green) input to kisspeptin cells (red) in the arcuate nucleus. A and C, Low power images and the boxed areas are shown in higher power in B and D, respectively. Arrows indicate close appositions between glutamatergic varicosities and the kisspeptin cell. Optical rotation of the cell allowed confirmation that visualization of the close contacts between glutamatergic varicosities and the kisspeptin cells were consistent in all planes of observation. Scale bars, 20 m. above, however, the results obtained in anesthetized ewes suggest that factors other than kisspeptin, emanating from neuronal elements, may also be involved in pulse generation. The data of experiment 3 indicated that kisspeptin can cause release of GnRH within the median eminence in vivo. This relies upon the injection being focal and discrete, and we confirmed that this postmortem (Supplemental Figure 1). Based on the discrete nature of the lesion, into scar tissue, it is likely that not all of the GnRH neurosecretory terminals were reached by the injection. This most likely explains why a higher dose was required for response to microinjections. Although relatively high doses are required for this procedure, the focal injection and the specific nature of the response indicate that the model is valid. A major unresolved question is how kisspeptin may act within the median eminence to cause pulsatile GnRH release from neurosecretory terminals. A number of studies have examined close relationships between various neuronal elements in the median eminence and GnRH secretory terminals. Varicose kisspeptin fibers come into close contact with GnRH fibers in the median eminence in mice (40), nonhuman primates (41), and sheep (8). In the goat median eminence, close contacts between kisspeptin and GnRH terminals have been observed with electron microscopy (42), although there was no evidence of synaptic structures between the 2 elements. In addition, there has been no demonstration of the kisspeptin receptor within the GnRH terminal fibers, due to lack of a suitable antibody. Nevertheless, support for a receptor-mediated mechanism is found in the data of Roseweir et al (43), showing that infusion of peptide 234 blocks pulsatile LH secretion in OVX nonhuman primates, when infused directly into the median eminence. It is possible that the action of kisspeptin within the median eminence is on GnRH fibers entering the neurosecretory zone. The recent description of dendrons projecting from GnRH neuronal cell bodies to the median eminence (44) allows for this possibility. These projections are not typical axons and display spines throughout their entire length that are opposed by boutons that contain synaptic markers. There is, however, no evidence that KNDy cells make synaptic contact with GnRH dendrons. Excitatory amino acid receptors have been identified in the terminal fibers of GnRH neurons in the median eminence of the rat (45), but no synaptic contacts are made between glutamatergic elements and the GnRH fibers at this level. Glutamatergic signaling occurs within the neurosecretory terminals (46), perhaps via volume control.

1836 Ezzat et al The GnRH Pulse Generator Endocrinology, May 2015, 156(5):1828 1837 The involvement of glutamate neurons in the generation of GnRH pulses may occur at the level of the median eminence, but the present data implicate glutamate cells that provide input to KNDy cells. Other data obtained from mice indicate that ip N-methyl-D-aspartate injection activates KNDy cells, and there is a lack of LH response to NMDA in Kiss1-null and kisspeptin receptor-null mice (12). Our present data indicate that glutamatergic cells provide input to at least 75% of KNDy cells, and glutamate cells of the arcuate nucleus are activated at the time of a natural secretory pulse of LH, reflecting GnRH secretion. It is likely that at least some of the glutamatergic input to KNDy cells emanates from beyond the arcuate nucleus. In summary, we present evidence that the KNDy cells of the arcuate nucleus of the ovine brain are 1 component of the pulse generator acting to stimulate GnRH secretion in the basal mode. Thus, it was observed that these KNDy cells are activated at the exact time of a natural GnRH/LH pulse in the ewe. An upstream element of this pulse generator appears to include glutamatergic cells within the arcuate nucleus. Evidence is also presented to indicate that the pulse generator involves neural elements in addition to kisspeptin neurons, because the LH response to iv administered kisspeptin is greater in animals that are conscious than in those that are anesthetized. The pulse generator does not appear to involve the kisspeptin neurons of the POA. Kisspeptin acts within the median eminence to cause pulsatile secretion of GnRH, but the mechanism for this remains to be elucidated. Acknowledgments We thank Bruce Doughton, Lynda Morrish, Alexandra Rao, and Alda Pereira for their technical assistance. Address all correspondence and requests for reprints to: Professor Iain J. Clarke, Department of Physiology, Monash University, Clayton, Victoria 3800, Australia. E-mail: iain.clarke@monash.edu. Present address for A.E.: Department of Theriogenology, Faculty of Veterinary Medicine, South Valley University, Qena 83523, Egypt. This work was supported by the National Health and Medical Research Council of Australia. A.E. was supported by the South Valley University, Qena, Egypt. Disclosure Summary: The authors have nothing to disclose. References 1. Clarke IJ, Cummins JT. The temporal relationship between gonadotropin releasing hormone (GnRH) and luteinizing hormone (LH) secretion in ovariectomized ewes. Endocrinology. 1982;111(5): 1737 1739. 2. Clarke I, Moore L, Veldhuis J. Intensive direct cavernous sinus sampling identifies high-frequency, nearly random patterns of FSH secretion in ovariectomized ewes: combined appraisal by RIA and bioassay. Endocrinology. 2002;143(1):117 129. 3. Nunemaker CS, DeFazio RA, Geusz ME, Herzog ED, Pitts GR, Moenter SM. Long-term recordings of networks of immortalized GnRH neurons reveal episodic patterns of electrical activity. J Neurophysiol. 2001;86(1):86 93. 4. Pitts GR, Nunemaker CS, Moenter SM. Cycles of transcription and translation do not comprise the gonadotropin-releasing hormone pulse generator in GT1 cells. Endocrinology. 2001;142(5):1858 1864. 5. Terasawa E, Keen KL, Mogi K, Claude P. Pulsatile release of luteinizing hormone-releasing hormone (LHRH) in cultured LHRH neurons derived from the embryonic olfactory placode of the rhesus monkey. Endocrinology. 1999;140(3):1432 1441. 6. Wilson RC, Kesner JS, Kaufman JM, Uemura T, Akema T, Knobil E. Central electrophysiologic correlates of pulsatile luteinizing hormone secretion in the rhesus monkey. Neuroendocrinology. 1984; 39(3):256 260. 7. Ohkura S, Takase K, Matsuyama S, et al. Gonadotrophin-releasing hormone pulse generator activity in the hypothalamus of the goat. J Neuroendocrinol. 2009;21(10):813 821. 8. Smith JT, Li Q, Yap KS, et al. Kisspeptin is essential for the full preovulatory LH surge and stimulates GnRH release from the isolated ovine median eminence. Endocrinology. 2011;152(3):1001 1012. 9. Pinilla L, Aguilar E, Dieguez C, Millar RP, Tena-Sempere M. Kisspeptins and reproduction: physiological roles and regulatory mechanisms. Physiol Rev. 2012;92(3):1235 1316. 10. Li XF, Kinsey-Jones JS, Cheng Y, et al. Kisspeptin signalling in the hypothalamic arcuate nucleus regulates GnRH pulse generator frequency in the rat. PLoS One. 2009;4(12):e8334. 11. d Anglemont de Tassigny X, Fagg LA, Carlton MB, Colledge WH. Kisspeptin can stimulate gonadotropin-releasing hormone (GnRH) release by a direct action at GnRH nerve terminals. Endocrinology. 2008;149(8):3926 3932. 12. d Anglemont de Tassigny X, Ackroyd KJ, Chatzidaki EE, Colledge WH. Kisspeptin signaling is required for peripheral but not central stimulation of gonadotropin-releasing hormone neurons by NMDA. J Neurosci. 2010;30(25):8581 8590. 13. Keen KL, Wegner FH, Bloom SR, Ghatei MA, Terasawa E. An increase in kisspeptin-54 release occurs with the pubertal increase in luteinizing hormone-releasing hormone-1 release in the stalk-median eminence of female rhesus monkeys in vivo. Endocrinology. 2008;149(8):4151 4157. 14. Caraty A, Lomet D, Sébert ME, Guillaume D, Beltramo M, Evans NP. Gonadotrophin-releasing hormone release into the hypophyseal portal blood of the ewe mirrors both pulsatile and continuous intravenous infusion of kisspeptin: an insight into kisspeptin s mechanism of action. J Neuroendocrinol. 2013;25(6):537 546. 15. Choe HK, Kim HD, Park SH, et al. Synchronous activation of gonadotropin-releasing hormone gene transcription and secretion by pulsatile kisspeptin stimulation. Proc Natl Acad Sci USA. 2013; 110(14):5677 5682. 16. Smith JT, Clay CM, Caraty A, Clarke IJ. KiSS-1 messenger ribonucleic acid expression in the hypothalamus of the ewe is regulated by sex steroids and season. Endocrinology. 2007;148(3):1150 1157. 17. Merkley CM, Porter KL, Coolen LM, et al. KNDy (kisspeptin/neurokinin B/dynorphin) neurons are activated during both pulsatile and surge secretion of LH in the ewe. Endocrinology. 2012;153(11): 5406 5414. 18. Barker-Gibb ML, Scott CJ, Boublik JH, Clarke IJ. The role of neuropeptide Y (NPY) in the control of LH secretion in the ewe with

doi: 10.1210/en.2014-1756 endo.endojournals.org 1837 respect to season, NPY receptor subtype and the site of action in the hypothalamus. J Endocrinol. 1995;147(3):565 579. 19. Grasshoff C, Drexler B, Rudolph U, Antkowiak B. Anaesthetic drugs: linking molecular actions to clinical effects. Curr Pharm Des. 2006;12(28):3665 3679. 20. Lee VW, Cumming IA, de Kretser DM, Findlay JK, Hudson B, Keogh EJ. Regulation of gonadotrophin secretion in rams from birth to sexual maturity. I. Plasma LH, FSH and testosterone levels. J Reprod Fertil. 1976;46(1):1 6. 21. Clarke IJ. Variable patterns of gonadotropin-releasing hormone secretion during the estrogen-induced luteinizing hormone surge in ovariectomized ewes. Endocrinology. 1993;133(4):1624 1632. 22. Goodman RL, Legan SJ, Ryan KD, Foster DL, Karsch FJ. Two effects of estradiol that normally contribute to the control of tonic LH secretion in the ewe. Biol Reprod. 1980;23(2):415 422. 23. Rice GE, Jenkin G, Thorburn GD. Comparison of particle-associated progesterone and oxytocin in the ovine corpus luteum. J Endocrinol. 1986;108(1):109 116. 24. Scott CJ, Clarke IJ, Tilbrook AJ. Neuronal inputs from the hypothalamus and brain stem to the medial preoptic area of the ram: neurochemical correlates and comparison to the ewe. Biol Reprod. 2003;68(4):1119 1133. 25. Scott CJ, Cumminst JT, Clarke IJ. Effects on plasma luteinizing hormone levels of microinjection of noradrenaline and adrenaline into the septo-preoptic area of the brain of the ovariectomized ewe: changes with season and chronic oestrogen treatment. J Neuroendocrinol. 1992;4(1):131 141. 26. Clarke IJ. Evidence that the switch from negative to positive feedback at the level of the pituitary gland is an important timing event for the onset of the preovulatory surge in LH in the ewe. J Endocrinol. 1995;145(2):271 282. 27. Smith JT, Li Q, Pereira A, Clarke IJ. Kisspeptin neurons in the ovine arcuate nucleus and preoptic area are involved in the preovulatory luteinizing hormone surge. Endocrinology. 2009;150(12):5530 5538. 28. Franceschini I, Lomet D, Cateau M, Delsol G, Tillet Y, Caraty A. Kisspeptin immunoreactive cells of the ovine preoptic area and arcuate nucleus co-express estrogen receptor. Neurosci Lett. 2006; 401(3):225 230. 29. Goodman RL, Lehman MN, Smith JT, et al. Kisspeptin neurons in the arcuate nucleus of the ewe express both dynorphin A and neurokinin B. Endocrinology. 2007;148(12):5752 5760. 30. Fujiyama F, Furuta T, Kaneko T. Immunocytochemical localization of candidates for vesicular glutamate transporters in the rat cerebral cortex. J Comp Neurol. 2001;435(3):379 387. 31. Pompolo S, Pereira A, Scott CJ, Fujiyma F, Clarke IJ. Evidence for estrogenic regulation of gonadotropin-releasing hormone neurons by glutamatergic neurons in the ewe brain: an immunohistochemical study using an antibody against vesicular glutamate transporter-2. J Comp Neurol. 2003;465(1):136 144. 32. Hanchate NK, Parkash J, Bellefontaine N, et al. Kisspeptin-GPR54 signaling in mouse NO-synthesizing neurons participates in the hypothalamic control of ovulation. J Neurosci. 2012;32(3):932 945. 33. Clasadonte J, Poulain P, Beauvillain JC, Prevot V. Activation of neuronal nitric oxide release inhibits spontaneous firing in adult gonadotropin-releasing hormone neurons: a possible local synchronizing signal. Endocrinology. 2008;149(2):587 596. 34. Schnickel PG. The effect of the ram on the incidence and occurrence of oestrus in ewes. Aust Vet J. 1954;30:189 195. 35. De Bond JA, Li Q, Millar RP, Clarke IJ, Smith JT. Kisspeptin signaling is required for the luteinizing hormone response in anestrous ewes following the introduction of males. PLoS One. 2013;8(2): e57972. 36. Pompolo S, Rawson JA, Clarke IJ. Projections from the arcuate/ ventromedial region of the hypothalamus to the preoptic area and bed nucleus of stria terminalis in the brain of the ewe; lack of direct input to gonadotropin-releasing hormone neurons. Brain Res. 2001; 904(1):1 12. 37. Backholer K, Smith J, Clarke IJ. Melanocortins may stimulate reproduction by activating orexin neurons in the dorsomedial hypothalamus and kisspeptin neurons in the preoptic area of the ewe. Endocrinology. 2009;150(12):5488 5497. 38. Dumalska I, Wu M, Morozova E, Liu R, van den Pol A, Alreja M. Excitatory effects of the puberty-initiating peptide kisspeptin and group I metabotropic glutamate receptor agonists differentiate two distinct subpopulations of gonadotropin-releasing hormone neurons. J Neurosci. 2008;28(32):8003 8013. 39. Kinsey-Jones JS, Li XF, Luckman SM, O Byrne KT. Effects of kisspeptin-10 on the electrophysiological manifestation of gonadotropin-releasing hormone pulse generator activity in the female rat. Endocrinology. 2008;149(3):1004 1008. 40. Clarkson J, Herbison AE. Postnatal development of kisspeptin neurons in mouse hypothalamus; sexual dimorphism and projections to gonadotropin-releasing hormone neurons. Endocrinology. 2006; 147(12):5817 5825. 41. Ramaswamy S, Guerriero KA, Gibbs RB, Plant TM. Structural interactions between kisspeptin and GnRH neurons in the mediobasal hypothalamus of the male rhesus monkey (Macaca mulatta) as revealed by double immunofluorescence and confocal microscopy. Endocrinology. 2008;149(9):4387 4395. 42. Matsuyama S, Ohkura S, Mogi K, et al. Morphological evidence for direct interaction between kisspeptin and gonadotropin-releasing hormone neurons at the median eminence of the male goat: an immunoelectron microscopic study. Neuroendocrinology. 2011; 94(4):323 332. 43. Roseweir AK, Kauffman AS, Smith JT, et al. Discovery of potent kisspeptin antagonists delineate physiological mechanisms of gonadotropin regulation. J Neurosci. 2009;29(12):3920 3929. 44. Herde MK, Iremonger KJ, Constantin S, Herbison AE. GnRH neurons elaborate a long-range projection with shared axonal and dendritic functions. J Neurosci. 2013;33(31):12689 12697. 45. Kawakami SI, Hirunagi K, Ichikawa M, Tsukamura H, Maeda KI. Evidence for terminal regulation of GnRH release by excitatory amino acids in the median eminence in female rats: a dual immunoelectron microscopic study. Endocrinology. 1998;139(3):1458 1461. 46. Kermath BA, Riha PD, Sajjad A, Gore AC. Effects of chronic NMDA-NR2b inhibition in the median eminence of the reproductive senescent female rat. J Neuroendocrinol. 2013;25(10):887 897.