NEUROENDOCRINOLOGY
Gap Junctions between Neuronal Inputs But Not Gonadotropin-Releasing Hormone Neurons Control Estrous Cycles in the Mouse Rebecca E. Campbell,* Eric Ducret,* Robert Porteous,* Xinhuai Liu, Michel K. Herde, Kerstin Wellerhaus, Stephan Sonntag, Klaus Willecke, and Allan E. Herbison Centre for Neuroendocrinology and Department of Physiology (R.E.C., E.D., R.P., X.L., M.K.H., A.E.H.), University of Otago School of Medical Sciences, Dunedin 9054, New Zealand; and Institute of Genetics (K.We., S.S., K.Wi.), Division of Molecular Genetics, University of Bonn, D-53117 Bonn, Germany
The role of gap junctions in the neural control of fertility remains poorly understood. Using acute brain slices from adult GnRH-green fluorescent protein transgenic mice, individual GnRH neurons were filled with a mixture of fluorescent dextran and neurobiotin. No dye transfer was found between any GnRH neurons, although approximately 30% of GnRH neurons exchanged neurobiotin with closely apposed cells. Dual electrophysiological recordings from pairs of GnRH neurons revealed an absence of electrical coupling. Using adult connexin 36 (Cx36)-cyan fluorescent protein transgenic mice, Cx36 was identified in cells within several hypothalamic brain regions, including 64% of preoptic area kisspeptin neurons but not in GnRH neurons. To assess the potential role of Cx36 in non-GnRH neurons within the GnRH neuronal network (i.e. neurons providing afferent inputs to GnRH neurons), a calmodulin kinase II␣-Cre (CKC)-LoxP strategy was used to generate mice with a neuron-specific deletion of Cx36 beginning in the first postnatal week. Mutant female mice exhibited normal puberty onset but disordered estrous cyclicity, although their fecundity was normal as was their estrogen-negative and -positive feedback mechanisms. The effects of adult deletion of Cx36 from neurons were assessed using a tamoxifen-dependent inducible CKC-Cx36 transgenic strategy. Mutant mice exhibited the same reproductive phenotype as the CKC-Cx36 animals. Together these observations demonstrate that there is no gap junctional coupling between GnRH neurons. However, it is apparent that other neurons within the GnRH neuronal network, potentially the preoptic kisspeptin neurons, are dependent on Cx36 gap junctions and that this is critical for normal estrous cyclicity. (Endocrinology 152: 2290 –2301, 2011)
A
complex network of neurons and glia comprise the GnRH neuronal network that controls fertility in mammals. Much progress has been made over recent years in deciphering the components of this network (1–3). However, one important issue that has remained unclear is the role of gap junctions in this network. In vitro studies using immortalized GnRH-secreting cell lines have found evidence for gap junctions comprised of multiple different
connexins to be critically involved in the episodic release of GnRH from these cultures (4 – 8). The documentation of key roles for gap junctions in the generation of synchronous electrical activity elsewhere in the nervous system (9, 10) has further supported a potential role for gap junctions in the synchronization of GnRH neurons. Whether the gap junction findings from in vitro preparations are relevant to GnRH neurons in vivo are un-
ISSN Print 0013-7227 ISSN Online 1945-7170 Printed in U.S.A. Copyright © 2011 by The Endocrine Society doi: 10.1210/en.2010-1311 Received November 11, 2010. Accepted March 3, 2011. First Published Online March 29, 2011
* R.E.C., E.D., and R.P. contributed equally to this study. Abbreviations: aCSF, Artificial cerebrospinal fluid; AVPV, anteroventral periventricular nucleus; BNST, bed nucleus of the stria terminalis; CaMK, calmodulin kinase II␣; CaMKCreERT2, Cre recombinase linked to a ligand binding domain of the human estrogen receptor activated only by tamoxifen; CFP, cyan fluorescent protein; CKC, CaMK-Cre; cPeN, caudal periventricular nucleus; Cx36, connexin 36; Cx43, connexin 43; GFP, green fluorescent protein; OVX, ovariectomy; rPeN, rostral periventricular nucleus; RP3V, rostral periventricular region of the third ventricle; RT, room temperature; SCN, suprachiasmatic nucleus; TBS, Tris-buffered saline.
2290
endo.endojournals.org
Endocrinology, June 2011, 152(6):2290 –2301
Endocrinology, June 2011, 152(6):2290 –2301
known. The scattered distribution of GnRH neuron cell bodies throughout the basal forebrain of mammals is quite different to the situation in vitro and does not provide an easy anatomical substrate for direct gap junctional communication between GnRH neurons. However, the recent demonstration that the great majority of GnRH neuron dendrites intertwine with one another (11) has provided new impetus to explore the possibility that dendro-dendritic gap junctions may have a physiological role. Again, insights from other neuronal networks (12) suggest that gap junction-coupled dendrites can provide a powerful mechanism for synchronization. At present, the only data indicating that GnRH neurons may form gap junctions in vivo come from two studies in the rat showing connexin 32 and connexin 43 immunoreactivity in GnRH neuron cell bodies and terminals, respectively (13, 14). To address the physiological roles of gap junctional communication within the GnRH neuronal network, we used cell filling, electrophysiology, and fertility analyses in a range of transgenic mouse models. Our first aim was to define whether gap junctions exist between GnRH neurons using conventional dye transfer and electrophysiological approaches in GnRH-green fluorescent protein (GFP) transgenic mice. Having found that GnRH neurons do not form gap junctions with other GnRH neurons, we then examined whether functionally significant gap junctions may exist between other neurons within the GnRH neuronal network. At least 10 different connexins are responsible for forming gap junctions within the brain, but the connexin responsible for the vast majority of neuron-neuron gap junctions is connexin 36 (Cx36) (10, 15). Although initially thought to be exclusive to neuron-neuron gap junctions, Cx36 has also been found outside the brain in the adrenal medulla and pancreatic -cells (16, 17). Accordingly, we examined the role of neuronal Cx36-gap junctions in the control of fertility using Cre-LoxP technology in which calmodulin kinase II␣ (CaMK) directed the deletion of Cx36 from forebrain neurons either from the first postnatal week or in adulthood.
endo.endojournals.org
2291
Generation of neuron-specific Cx36 mutants Two different crosses were used to obtain mice in which Cx36 was deleted from forebrain neurons either at the time of endogenous CaMK expression or after induction in adulthood. The first cross was generated from a CaMK-Cre (CKC) line in which Cre recombinase is driven by CaMK expressed in most forebrain neurons beginning in the first postnatal week (18). This CKC line has been used previously to delete all estrogen receptor-␣ (19) and glucocorticoid receptor (20) expression in the forebrain. In the present experiments, CKC mice were crossed onto a new floxed Cx36 line in which Cre recombinase deletes Cx36 and also activates expression of cyan fluorescent protein (CFP) to mark Cx36-deleted cells (17). Mice were bred to generate mutant [Cx36flox(CFP)/flox(CFP); CKC⫹/⫺] and control [Cx36flox(CFP)/flox(CFP)] mice. The second cross involved the use of a tamoxifen-activated CKC mouse line. In this case, CaMK regulatory elements direct expression of a Cre recombinase linked to a ligand binding domain of the human estrogen receptor activated only by tamoxifen (CaMKCreERT2) (20). Upon treatment with tamoxifen (1 mg twice a day for 5 d; ip), Cre recombinase is directed into the nucleus of the CaMK-expressing cells to recombine any floxed gene (20). Using a CaMKCreERT2 cross with ROSA26-LacZ indicator mice, we also confirmed (not shown) that there is no Cre-mediated recombination in these mice in the absence of tamoxifen treatment (20). In these experiments, the CaMKCreERT2 mouse was crossed onto the Cx36 flox mouse to generate mutants [Cx36flox(CFP)/flox(CFP);CaMKCreERT2] and controls [Cx36flox(CFP)/flox(CFP)]. Initial experiments in wild-type mice demonstrated that the 5-d tamoxifen treatment disrupted estrous cycles for approximately 3– 4 wk after which they returned to normal. Hence, in these studies, mutant and control mice were treated with tamoxifen at approximately 3 months of age and investigated 5 wk later.
Acute brain slice preparation Studies involving cell filling and electrophysiology used a well-characterized GnRH-GFP transgenic mouse line that enables individual GnRH neurons to be identified in the brain slice (21). Adult mice (60 –75 d old) were killed by cervical dislocation, the brain quickly removed, and 200 m-thick sagittal brain slices prepared with a vibratome (Leica VT1000S; Wetzlar, Germany) in a low-calcium and high-magnesium artificial cerebrospinal fluid (aCSF; NaCl: 118 mM; KCl: 3 mM; D-glucose: 11 mM; HEPES: 10 mM; NaHCO3: 25 mM; CaCl2: 0.5 mM; MgCl2: 6 mM) kept on ice and oxygenated with carbogen (O2: 95%; CO2: 5%). The slices were stored at 35 C for at least 1 h in normal aCSF (NaCl: 118 mM; KCl: 3 mM; D-glucose: 11 mM; HEPES: 10 mM; NaHCO3: 25 mM; CaCl2: 2.5 mM; MgCl2: 1.2 mM) before beginning experiments.
Cell filling
Materials and Methods Animals Experiments were undertaken in five different C57BL/6J transgenic mouse lines and crosses thereof, as detailed below. All mice were housed with 12-h light, 12-h dark cycle (lights on 0700 h) and ad libitum access to food and water. All experiments were approved by the University of Otago Welfare and Ethics Committee.
Brain slices were transferred to a recording chamber, held submerged, and continuously superfused with aCSF at a rate of 2–3 ml/min at room temperature (RT). Whole-cell filling of GnRH neurons was undertaken using a fixed-stage upright fluorescence microscope (BX51WI; Olympus, Tokyo, Japan) with GFP-tagged GnRH neurons identified briefly using fluorescence and then patched under Nomarski differential interference contrast optics (a ⫻40 water-immersion objective). Patch pipettes were pulled from glass capillaries (inner diameter 1.2 mm; outer diameter 1.5 mm) with a microelectrode puller (Sutter Instru-
2292
Campbell et al.
Role of Gap Junctions in GnRH Neuronal Network
ments, Novato, CA) and had 3–5-M⍀ resistances when filled with the pipette solution composed of 130 mM CH3KSO4 (potassium methyl sulfate), 5 mM NaCl, 0.5 mM CaCl2, 20 mM HEPES, 5 mM MgATP, 0.2 mM Na2GTP, 4 mM phosphocreatine-Tris, 6 mM neurobiotin (molecular weight 322), and 0.02 mM dextran-AlexaFluor680 (molecular weight 10,000; Invitrogen, Carlsbad, CA; pH 7.35 adjusted by KOH, ⬃290 mOsmol). The patch pipette solution was passed through a disposable 0.22-m filter before use. After achieving whole-cell mode, the pipette was kept attached to the GnRH neuron for variable times ranging from 30 –240 min to allow diffusion of neurobiotin and 680-dextran. After detaching the pipette from the cell, slices were incubated in aCSF with 0 mM MgCl2 and 0.1 mM CaCl2 at 35 C for 60 min and then placed in ice-cold 4% paraformaldehyde for 24 h at 4 C. Only one cell was filled in each brain slice. Slices were then transferred to Tris-buffered saline (TBS; pH 7.6) and kept at 4 C until processing for immunocytochemistry. Neurobiotinfilled slices were washed in TBS at RT to remove any residual paraformaldehyde. Slices were then incubated with streptavidinAlexaFlour 543 at 1:200 in TBS containing 0.5% Triton-X-100 and 0.3% BSA for 90min in darkness. Slices were then washed in TBS over 1 h, mounted onto glass slides, and coverslipped with Vectashield (Vector Laboratories, Inc., Burlingame, CA).
Electrophysiology Microelectrodes were pulled from borosilicate capillaries (Warner Instruments, Hamden, CT; outer diameter 1.50 mm, inner diameter 1.17 mm) with a microelectrode puller (model p-97; Sutter Instruments). Electrodes (4 –5.2 M⍀) were filled with pipette solution (CH3KSO4 135 mM; NaCl: 4.5 mM; HEPES: 10 mM; Na2GTP: 0.2 mM; Na2ATP: 0.2 mM; MgATP: 2 mM; EGTA: 1.1 mM; CaCl2: 0.1 mM; phosphocreatin-Tris: 7 mM, pH 7.3). Slices were transferred to an upright microscope (BX51WI; Olympus) and perfused throughout the experiment with oxygenated aCSF at RT. Recorded signals were amplified (Multiclamp 700B; Axon Instruments, Inverune, Scotland) and digitized (Digidata 1440A; Axon Instruments). GnRH neurons were identified by brief exposure to fluorescence and the recording electrode was applied to the cell membrane under infrared differential interference contrast. Two GnRH neurons were selected whose dendrites appeared to be in the vicinity of each other or close to one of the cell bodies, and each was patched with a seal resistance greater than 1.2 G⍀. The cell bodies of the GnRH neuron pairs were located within the medial septum and rostral preoptic area. After 5 min in cell-attached mode, the membrane was broken by brief suction and, after holding the neuron potential at ⫺60 mV for more than 10 min, whole-cell recordings were performed. To assess the presence of electrical coupling, 10 current pulses (⫺ or ⫹ 40 pA for 1 sec, 0.5 Hz) were applied to one GnRH neuron (presynaptic cell) and the membrane potential response recorded simultaneously in both the presynaptic and second (postsynaptic) GnRH neuron. The electrical coupling was measured by injecting steps of current (from ⫺40 to 30 pA every 10 pA for 1 sec, 0.3 Hz) and recording the membrane potential in both the injected GnRH neuron (presynaptic neuron) and the second GnRH neuron (postsynaptic neuron). The slope of the curve representing the presynaptic voltage variation as a function of the postsynaptic voltage variation was used to determine the electrical coupling coefficient. To control for coupling between the electrodes, the same protocol was undertaken but with only
Endocrinology, June 2011, 152(6):2290 –2301
one GnRH neuron patched in whole-cell configuration and a second electrode left nearby in the aCSF. The current was injected into the GnRH neuron and the voltage variation recorded for both electrodes.
Immunocytochemistry and microscopy For all immunocytochemistry, the mice were killed with an overdose of pentobarbital followed by perfusion fixation through the heart with 4% paraformaldehyde in PBS (pH 7.4). Brains were removed, postfixed for 1 h at RT, and then saturated in 30% sucrose, TBS solution overnight at 4 C until processing for immunocytochemistry. Brains were cut on a freezing microtome into three sets of 30-m-thick coronal sections throughout the brain. To evaluate Cx36 expression in the brain, Cx36-CFP (22) and Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ (see above) mice were used. Immunofluorescent detection of CFP was performed on free-floating sections obtained from adult Cx36-CFP mice (n ⫽ 8 females) and Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ mice (n ⫽ 5 females). One set of tissue (every third section through the brain) was placed in an incubation solution (TBS solution with 0.3% Triton-X-100 and 0.25% BSA) containing rabbit anti-GFP antibodies (1:5,000; Invitrogen) for 48 h at 4 C. This antibody also detects CFP. After washes in TBS, tissue was incubated in AlexaFluor 488 goat antirabbit immunoglobulins (1:200; Invitrogen) in incubation solution for 90 min. After a final wash in TBS, tissue was mounted onto gelatin-coated glass slides, coverslipped with Vectasheild mounting media (Vector Laboratories), and kept in the dark at 4 C until analysis. Omission of primary antibody resulted in an absence of staining as did GFP immunolabeling of wild-type brain sections. Dual-label immunofluorescence labeling for GnRH and GFP, or kisspeptin and GFP, was undertaken by incubating coronal brain sections in rabbit anti-GFP antiserum (1:2500) and either monoclonal mouse anti-GnRH (1:1000 HU11B, a kind gift of H. Urbanski, Portland, OR) antiserum or monoclonal mouse antimetastin antibodies (23) (0.5 g/ml, kind gift of Takeda Pharmaceuticals, Osaka, Japan). After 48 h in the cocktail of primary antibodies, sections were washed and incubated in antirabbit fluorescein isothiocyanate (1:200; Jackson ImmunoResearch Laboratories, West Grove, PA) and biotinylated antimouse immunoglobulins (1:200; Jackson ImmunoResearch Laboratories) for 90 min at RT followed by washes and a final incubation in streptavidin-AlexaFluor 568 (Invitrogen) for 90 min at RT. Tissue was then washed and mounted on slides as above. For examination of colocalization with GnRH neurons, analysis was undertaken by examining all GnRH neurons in two sections at each of three levels of the GnRH neuron continuum (medial septum, rostral preoptic area, anterior hypothalamic area) in each mouse as detailed previously (24). Single- and double-labeled kisspeptin neurons were counted in two sections at each of three levels of the rostral periventricular region of the third ventricle (RP3V) in each mouse, as detailed previously (25). Dual-label immunofluorescence for GnRH and connexin 43 (Cx43) was carried out in coronal sections from four adult (two males, two females) GnRH-GFP mice (21) after antigen retrieval to enhance Cx43 labeling. Coronal brain sections were incubated in Tris HCl (0.1 M, pH 10) for 10 min in a 100 C water bath. Tissue was subsequently incubated in a cocktail of primary antibodies against Cx43 (mouse anti-Cx43, 1:500; Chemicon International Inc., Temecula, CA) and GFP (1:5000, expressed in
Endocrinology, June 2011, 152(6):2290 –2301
GnRH neurons). Secondary antibodies included biotinylated antimouse immunoglobulins (1:200; Jackson ImmunoResearch Laboratories) followed by streptavidin-AlexaFluor 568 (Invitrogen) to visualize Cx43 and antirabbit fluorescein isothiocyanate (1:200; Jackson ImmunoResearch Laboratories) to visualize GFP. Every third section through the median eminence was examined by confocal microscopy for colocalization. This antibody has been shown to specifically colocalize with gap junctions at the ultrastructural level (26). All fluorescent-labeled tissue was imaged and analyzed on a Zeiss 510 LSM upright confocal laser-scanning microscope system using LSM 510 control software (version 3.2; Carl Zeiss, Jena, Germany). Stacks of confocal images were captured using the following objectives: ⫻40 Plan Neofluar (numerical aperture 1.3) and ⫻63 PlanApochromat (numerical aperture 1.4) with ⫻2 zoom function. A red helium neon laser exciting at 633 nm was used to image the 568-nm fluorophore. In neurobiotin/680-dextran-filling experiments, the red helium neon laser was used to excite the 543- and 680-nm fluorophores at 543 and 633 nm, respectively, using 560 – 615-nm band pass and 450-nm long pass filters. An argon laser exciting at 488 nm was used to image the 488 fluorophore. A series of images at 0.3- to 0.5-m intervals were collected for analysis. Images are presented as projections of optical images stacks. The brightness and contrast of the images were adjusted in Photoshop (Adobe Systems, San Jose, CA) to match microscope visualization. Chromagen detection of CFP was performed on one 1:3 set of free-floating sections obtained from mutant [Cx36flox(CFP)/flox(CFP); CaMKCreERT2] and control [Cx36flox(CFP)/flox(CFP)] mice used in fertility analyses (n ⫽ 4 per group). After 48 h of incubation in rabbit anti-GFP (1:5000) antisera at 4 C, sections were incubated in biotinylated antirabbit immunoglobulins (1:200; Vector) for 90 min, washed in TBS, and then incubated in Vectastain Elite ABC reagents (Vector Laboratories) for 90 min. GFP immunoreactivity was visualized using a glucose oxidase, nickel-enhanced diaminobenzidine hydrochloride method. Sections were mounted on gelatin-coated glass slides and after 24 h, dehydrated in ethanol followed by xylene and then coverslipped with DPX. Brightfield images were imaged on an Olympus BX51 microscope (Olympus, Hamburg, Germany).
endo.endojournals.org
2293
1 g per 20 g body weight estradiol benzoate. The following day mice were anesthetized and killed at 1900 –1930 h with blood taken for LH RIA and the brain perfused to enable cFos-GnRH dual immunocytochemistry to be undertaken as reported previously (29). Estrogen-negative feedback was assessed by ovariectomy and estradiol replacement. Mice were anesthetized with Avertin (200 l, sc), a tail blood sample taken, and bilaterally ovariectomized. Two weeks later, mice were anesthetized with Avertin and a further tail blood sample taken. One or two days later, this was followed by a sc injection of 1 g per 20 g body weight 17-estradiol and, 3 h later, cervical dislocation and decapitation, with blood collected for LH assay. Statistical analysis was undertaken by nonparametric Mann-Whitney tests. Plasma LH concentrations were determined by RIA using the antirabbit LHS-11 antiserum and mouse LH-RP reference provided by Dr. A. F. Parlow (National Hormone and Peptide Program, Torrance, CA). The intra- and interassay coefficients of variation were 8.3 and 11.2%, respectively.
Results Absence of dye filling between GnRH neurons No evidence of dye coupling was found between any GnRH neurons after filling 34 GnRH neurons from 18 GnRH-GFP mice (15 male and three female, two estrus, and one diestrus) with neurobiotin/680-dextran (Fig. 1A).
Fertility testing and estrogen feedback experiments The testing of puberty onset, estrous cyclicity, fecundity, and estrogen feedback was undertaken exactly as described previously (27, 28). In brief, female mice were inspected from the time of weaning for vaginal opening and subsequent to this vaginal smears were taken each morning to detect the first estrus. Adult female mice were then examined for an 18-d period with daily vaginal smears to assess estrous cyclicity. Estrous cycle length was determined by counting the numbers of days between consecutive proestrus for each mouse. The percentage of time spent in diestrus was determined by counting the number of diestrus and metestrus days within the 18-d test period. Fecundity was assessed by putting mutant and control adult female mice with wild-type adult male mice for a period of 3 months and the numbers of litters and pups recorded. After this, some mice were perfused for immunocytochemistry as detailed above. Estrogen-positive feedback was assessed by ovariectomizing female mice, implanting a 1-g 17-estradiol-containing SILASTIC capsule (Corning, Midland, MI) and 6 d later treating the mice with
FIG. 1. Absence of dye coupling between GnRH neurons after cell filling with neurobiotin. A, Neurobiotin-filled GnRH neuron (red) surrounded by other GnRH neurons (green), showing a complete absence of neurobiotin spread. B, Representative neurobiotin filled GnRH neuron (yellow, arrow) demonstrating coupling with another non-GnRH cell in the slice (red, arrowhead). C–E, GnRH neuron (green), filled with dextran (blue), and neurobiotin (red), showing coupling to adjacent cells (red outside cell boundary indicated by dotted line). Scale bars, 10 m.
2294
Campbell et al.
Role of Gap Junctions in GnRH Neuronal Network
Endocrinology, June 2011, 152(6):2290 –2301
The majority of cells were filled with dyes for a 30- to 60-min period, but increasing this time to up to 4 h (n ⫽ 4) had no effect on the absence of dye transfer to other GnRH neurons. Evidence of dye coupling between GnRH neurons and other cells was found in 13 of the 34 (38%) filled GnRH neurons (Fig. 1, B–E). Studies using glialfibrillary acid protein antibodies to try and define the nature of these coupled cells were inconclusive (data not shown). The dye coupling observed in these experiments was not an artifact of inadvertent dye spread from the electrode tip before patching as all coupled cells exhibited neurobiotin without 680-dextran, the latter remaining only in the injected cell (Fig 1, C–E). Absence of electrical coupling between GnRH neurons Although the absence of dye coupling between GnRH neurons is suggestive of an absence of gap junctions it is not definitive proof. To examine for the presence of gap junctions in an alternative manner, we undertook dual recordings from GnRH neurons to assess electrical coupling. Twelve pairs of GnRH neurons, obtained from eight adult diestrous females and four adult males, that exhibited GFP-expressing dendrites in close apposition were targeted for investigation. The initial observation of firing patterns in cell-attached mode revealed no obvious synchronization in firing between any of the pairs of recorded neurons. In whole-cell mode, injection of current generating either depolarization or hyperpolarization in the first (presynaptic) GnRH neuron failed to generate any response in the second (postsynaptic) GnRH neuron (Fig. 2). Similarly, injection of current into the pairs in the other direction (current injection into the second GnRH neuron) failed to initiate any response in the other recorded GnRH neuron. Of the 12 pairs of neurons tested, none of the postsynaptic GnRH neurons showed any significant change in membrane potential. Overall, the mean coupling coefficient of ⫺0.019 ⫾ 0.019 (n ⫽ 24; Fig. 2C) was not different from the coupling of the electrodes alone (0.0002 ⫾ 0.0003). Absence of Cx36 and Cx43 in GnRH neurons The above observations suggest that GnRH neurons do not form gap junctions between themselves. To support this finding further and also to determine the locations of Cx36-expressing neurons in the hypothalamus, we examined the distribution of the neuronal connexin, Cx36, using a Cx36-CFP transgenic mouse model (22). CFP expressing cells were distributed in an heterogeneous manner throughout the forebrain as reported previously (16) and included the hypothalamic anteroventral periventricular nucleus (AVPV) and suprachiasmatic nu-
FIG. 2. GnRH neurons are not electrically coupled. A, Representative simultaneous recordings from two GnRH neurons in which 10 hyperpolarizing or depolarizing current pulses (⫾40 pA, top trace) were injected in the presynaptic neuron (pre), causing a depolarization of its membrane potential and firing of action potentials. The membrane potential of the paired GnRH neuron (post) was not affected by the injected current. Holding potential was ⫺60 mV for all cells. B, Representative simultaneous recordings from two GnRH neurons. Coupling coefficient was measured by applying current steps of increasing amplitude in the presynaptic neuron (pre, from ⫺40 to 30 pA, with 10 pA steps) and measuring the voltage variation in the pre- and postsynaptic (post) GnRH neurons. C, No coupling correlation was found with the postsynaptic cell failing to respond to the range of membrane polarization changes evoked in the presynaptic cell.
cleus (SCN) (Fig. 3A-C). The GnRH neurons were found located in the typical inverted Y distribution within the medial septum and hypothalamus (2). Although numerous CFP-expressing cells were detected nearby GnRH neuron cell bodies, there was a complete absence of CFP immunoreactivity in all GnRH neurons irrespective of their location within the GnRH neuron continuum, in each of the eight mice examined (Fig. 3, D–F). A previous paper has reported that GnRH nerve terminals exhibit Cx43 immunoreactivity (13). To reexamine this issue, we undertook a dual-label immunofluorescence study using a well-characterized Cx43 antibody (26). Although Cx43 immunoreactivity was found within the brain in expected areas, including the hippocampus (Fig. 4A), we found no evidence for Cx43 in GnRH neurons including the GnRH nerve terminals in the median eminence in any of the four mice examined (Fig. 4B). Cx36 is expressed by a subpopulation of preoptic area kisspeptin neurons The studies above using Cx36-CFP and Cx36flox(CFP)/flox(CFP); CKC⫹/⫺ mice indicated that one brain area with Cx36-expressing neurons was the AVPV
Endocrinology, June 2011, 152(6):2290 –2301
FIG. 3. Cx36 expression in Cx36-CFP and Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ mice. A–C, CFP expression driven by Cx36 regulatory elements in Cx36-CFP mice (green) is identified in the striatum (A), SCN (B), and AVPV (C). D–F, Low-level CFP expression was detected in cells of the rostral preoptic area in Cx36-CFP mice, but there was a complete absence of colocalization with GnRH neurons. G–L, The same distribution of CFP-expressing cells was found in Cx36flox(CFP)/flox(CFP); CKC⫹/⫺ mice, indicating areas in which Cx36 expression has been deleted. J–L, Dual-immunofluorescence labeling for CFP and GnRH in Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ mice also showed no evidence of colocalization. Scale bars (A–D and G–J), 50 m; (E, F, K, and L), 10 m.
and periventricular preoptic nucleus, together referred to as the RP3V (30). Given that this is the location of the preoptic kisspeptin neuron population (31) and their likely importance in control of GnRH neuron excitability (32), we examined whether Cx36 is expressed specifically by kisspeptin neurons in this region. Counting kisspeptin neurons in two sections from the level of the AVPV, rostral periventricular nucleus (rPeN), and caudal periventricular nucleus (cPeN), an average of 165 ⫾ 22 neurons were examined for colocalization. Kisspeptin cell numbers were not different across transgenic mouse lines. In Cx36-CFP mice, the majority of kisspeptin neurons throughout the RP3V (64%) were found to coexpress CFP (Fig. 5A). A greater proportion of colabeled neurons were identified in rostral and caudal regions of the periventricular nucleus (67.8⫾5.5and75.9⫾4.7%)thanintheAVPV(47.8⫾5.6%). Asimilarproportionofcolabeledneuronswereidentifiedinour other model, the Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ mice (Fig. 5B).
endo.endojournals.org
2295
Effects of neuron-specific deletion of Cx36 on fertility The above evidence indicated that GnRH neurons do not form gap junctions with one another. However, it was readily apparent that Cx36 was expressed by hypothalamic brain regions thought to contain neurons innervating GnRH neurons (3). To assess the possibility that Cx36 gap junctions in these non-GnRH neurons of the GnRH neuronal network may be important for fertility, we used a Cre-LoxP strategy to delete Cx36 from only forebrain neurons. This was achieved by crossing the CKC and floxed Cx36 mouse lines. Recombination in mutant Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ mice was demonstrated by PCR of brain tissue (not shown) and the appearance of CFP within the brain of mutant mice (Fig. 3, G–I). Control mice [Cx36flox(CFP)/flox(CFP)] never exhibited CFP immunoreactivity in any brain region. The distribution of CFPexpressing cells in Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ mice (n ⫽ 5) was the same as that observed in Cx36-CFP mice; for example, cells expressing CFP (and deleted Cx36) were detected throughout the hypothalamus including the SCN and RP3V (Fig. 3, G–I). Similarly, the distribution and number of GnRH neurons was normal (e.g. rostral preoptic area: control, 22.0 ⫾ 1.2 and mutant 24.5 ⫾ 2.9 GnRH neurons/section). As in the Cx36-CFP mouse, dual-label immunofluorescence showed a complete absence of CFP labeling in all GnRH neurons in each of the five mice examined (Fig. 3, J–L). Puberty onset in Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ female mice (n ⫽ 10) was found to be normal with the day of vaginal opening the same between genotypes (postnatal d 29.2 ⫾ 3.4 vs. 28.9 ⫾ 1.2) and a nonsignificant delay to first estrus (postnatal d 34.0 ⫾ 1.0 vs. 37.5 ⫾ 0.5) compared with controls [Cx36flox(CFP)/flox(CFP) and Cx36flox(CFP); n ⫽ 11]. Estrous cycles of Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ mutant mice were abnormal with prolonged periods in diestrus interspersed with occasional normal cycles (Fig. 6, C and D). Quantitative evaluation showed that Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ mice (n ⫽ 12) had double the mean cycle length (P ⬍ 0.001; Fig. 6A) with more time spent in diestrus (P ⬍ 0.01; Fig. 4B) compared with controls [Cx36flox(CFP)/flox(CFP) and Cx36flox(CFP); n ⫽ 12)]. The control (Cx36 flox only) mice exhibited mean estrous cycle lengths of 6.4 ⫾ 0.7 d that are not different from those found in C57BL/6J wild-type mice in our colony (5.5 ⫾ 0.3) (27). In terms of fecundity, no differences were detected between mutant and control mice in terms of the number of litters delivered or pups per litter (Fig. 6, E and F). In addition, male mice exhibited normal fecundity (data not shown).
2296
Campbell et al.
Role of Gap Junctions in GnRH Neuronal Network
Endocrinology, June 2011, 152(6):2290 –2301
the deleted gene. To examine the effects on fertility of a more acute deletion of Cx36 from adult neurons, we undertook experiments in which a tamoxifen-inducible CKC mouse line (CaMKCreERT2) (20) was used to delete Cx36. These mice also have a more restricted distribution of Cre expression compared with CKC mice (20). For these experiments, mutant [Cx36flox(CFP)/flox(CFP);CaMKCreERT2] and control [Cx36flox(CFP)/flox(CFP)] female mice were bred and all treated with tamoxifen at FIG. 4. Cx43 is not coexpressed in GnRH nerve terminals. A, Punctate labeling 3– 4 months of age (n ⫽ 8 –9 each). Mutant of Cx43 is present within the CA1 region of the hippocampus. B, Cx43 mice treated with tamoxifen (examined after immunoreactivity was present around the median eminence, but no evidence of completion of the breeding experiments), colocalization between GnRH nerve terminals (green) and Cx43 (red) could be exhibited a less widespread distribution of identified. Scale bars, 10 m. 3V, Third ventricle. GFP immunoreactivity compared with Cx36CFP and Cx36 flox(CFP)/flox(CFP) ;CKC ⫹/⫺ Effects of neuron-specific deletion of Cx36 on mice. Cells expressing CFP, indicating sites of Cx36 deestrogen feedback mechanisms letion, were found in the striatum (Fig. 8A), neocortex, The above results indicated that Cx36flox(CFP)/flox(CFP); hippocampus, accumbens, ventral diagonal band of CKC⫹/⫺ female mice were not able to exhibit normal esBroca, bed nucleus of the stria terminalis (BNST), medial trous cycles and one possible explanation for this would be and lateral preoptic area (including the AVPV; Fig. 8B), a partial failure of the GnRH surge mechanism. Two brain sparsely in the SCN (Fig. 8C), lateral hypothalamus, and areas strongly implicated in this mechanism in the mouse, medial amygdala. Mutant mice not given tamoxifen, or the AVPV and SCN (30), were found to have numerous control mice treated with tamoxifen, had no CFP expresCx36-expressing cells (Fig. 3, H and I). Accordingly we sion in any brain region. examined the estrogen-positive feedback mechanism in Mice treated with tamoxifen exhibited a similar repromutant [Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ (n ⫽ 5)] and conductive phenotype to Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ mice flox(CFP)/flox(CFP) trol [Cx36 (n ⫽ 5)] female mice. In renormal fecundity (Fig. 8). sponse to the estrogen treatment, all control mice and mu- with disordered estrous cycles but flox(CFP)/flox(CFP) ;CaMKCretant mice exhibited an LH surge with LH levels greater The estrous cycle length of Cx36 T2 ER mice was significantly prolonged compared with conthan 1 ng/ml (Fig. 7C) with the exception of one mouse in trols (Fig. 8D), although the amount of time spent in diestrus each group (LH levels of 0.2 and 0.54 ng/ml, respectively). All control and mutant mice showed robust cFos expres- was not significantly different between the genotypes (Fig. flox(CFP)/flox(CFP) ;CKC⫹/⫺ sion in 43 ⫾ 2 and 43 ⫾ 10% of rostral preoptic area 8E). In comparison with Cx36 flox(CFP)/flox(CFP) females (Fig. 6D), this was due to Cx36 ; GnRH neurons, respectively (Fig. 7B). T2 Another possible reason for the disordered estrous CaMKCreER mice showing a mixture of prolonged pericycles was that basal levels of LH and estrogen-negative ods in estrus as well as diestrus (Fig. 8D, F, and G). In terms feedback were abnormal. This was investigated using an of fecundity, no differences were detected between mutant ovariectomy (OVX) ⫹ estradiol replacement paradigm and control mice in terms of the number of litters delivered with three blood samples taken from each mouse. Both or pups per litter (Fig. 8, H and I). control [n ⫽ 5; Cx36flox(CFP)/flox(CFP)] and mutant [n ⫽ 4; Cx36flox(CFP)/flox(CFP);CKC⫹/⫺] mice exhibited normal basal levels of LH (0.2 ⫾ 0.01 vs. 0.28 ⫾ 0.08 ng/ml, Discussion respectively) and responded in an identical manner to OVX with an increase in LH (2.9 ⫾ 1.0 vs. 3.4 ⫾ 0.9 We find here that GnRH neurons do not form gap juncng/ml) that was then suppressed (1.0 ⫾ 0.3 vs. 0.9 ⫾ 0.5 tions between themselves at the level of their cell bodies and dendrites. Although a subpopulation (⬃35%) of ng/ml) 3 h after treatment with estradiol (Fig. 7A). GnRH neurons was able to transfer dye to immediately adjacent non-GnRH cells, this never occurred between Effects of adult induction of neuron-specific Cx36 GnRH neurons. It is noteworthy that dye transfer has simdeletion on fertility One potential confounding factor with the use of ilarly not been observed between embryonic GnRH neuknockout models is that of compensation over time for rons in the explant preparation (33). Using paired elec-
Endocrinology, June 2011, 152(6):2290 –2301
endo.endojournals.org
2297
FIG. 5. Cx36 is expressed in RP3V kisspeptin neurons in Cx36-CFP and Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ mice. A, Histogram shows the mean ⫾ SEM percentage of kisspeptin neurons in the AVPV, rPeN, and cPeN coexpressing CFP in the Cx36-CFP mouse (n ⫽ 6). Representative images of CFP (Cx36, green) and kisspeptin (red) labeling indicate colocalization (yellow, arrowheads). B, Mean ⫾ SEM percentage of kisspeptin neurons in the AVPV, rPeN, and cPeN coexpressing CFP in the Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ mouse (n ⫽ 4). Representative images of CFP (excised Cx36 gene, green) and kisspeptin (red) labeling indicate colocalization (yellow, arrowheads).
trophysiological recordings from GnRH neurons, we also found no evidence for synchronized firing or electrical coupling between these cells. Although only limited numbers of GnRH neuron dendritic bundles have been examined so far, we have not found any evidence for gap junctions between GnRH dendrites at the level of the electron microscope (11). Using both the Cx36-CFP and Cx36flox(CFP)/flox(CFP);CKC transgenic mouse models, no evidence was found for Cx36, the principal neuronal connexin, in GnRH neurons. Finally, we have been unable to confirm in the mouse the observation that GnRH nerve terminals in the rat express Cx43 (13). Ultrastructural studies demonstrate that Cx43 is not expressed by neurons in the central nervous system and most likely exists between astrocytes (26, 34). Although it is appreciated that it is difficult to prove the absence of something, we believe that the present data, drawing together multiple lines of cell filling, electrophysiological, morphological, and connexin investigations, provide a consistent and compelling case that adult GnRH neurons do not communicate among themselves through gap junctions in vivo. Although evidence suggests a similar scenario in immature GnRH neurons (33), the possibility of functional gap junctions at other developmental time points is not ruled out. If gap junctions do exist between adult GnRH neurons, then they must be quite atypical and, certainly, different in nature to those found in GT1 cells in which each of the lines of investigation used here has provided a positive result in the past (4 – 8). Those prior observations of functionally important gap junc-
tions between immortalized GnRH-secreting neurons reinforce the dangers of extrapolating findings from immortalized cell models to adult GnRH neurons in vivo. Some GnRH neurons transfer dye to immediately adjacent unidentified cells. Although evidence for gap junctions between neurons and glial cells is limited (35), one possibility is that GnRH neurons form gap junctions with ensheathing glial cells (36). This might explain the prior identification of connexin 32 in adult GnRH neurons of the rat (14). It is also noteworthy that the heterozygous Cx43 knockout mouse has disordered reproductive functioning, although the key sites of Cx43 dysfunction in bringing about this phenotype are not known (37). The ultrastructural and functional data indicate that Cx43 is not expressed by neurons (26, 34). Hence, although the present findings discount the existence of gap junctions between GnRH neurons, they leave open the possibility of non-Cx36 gap junctions between GnRH neurons and adjacent non-GnRH cells. The observation of substantial Cx36 expression in hypothalamic brain regions thought to contain afferent inputs to GnRH neurons led us to speculate that neuronneuron gap junctions in other parts of the GnRH neuronal network might be important. For example, neurons in the SCN form Cx36-dependent gap junctions important for their synchronized activity (38), and SCN inputs project to the dispersed GnRH neuron population (39). Similarly, estrogen-receptive neurons within the AVPV express Cx36, and these cells are also believed to form direct synaptic connections with the majority of the GnRH neuron
2298
Campbell et al.
Role of Gap Junctions in GnRH Neuronal Network
FIG. 6. Abnormal estrous cycles in neuron-specific Cx36 mutant mice. A and B, Histograms showing the mean (⫾SEM) estrous cycle length (A) and percentage of time spent in diestrus (B) for Cx36flox(CFP)/flox(CFP) (controls) and Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ (neuron-Cx36) female mice (n ⫽ 5 each). C and D, Charts showing estrous cycle activity for individual mice over an 18-d period. Two representative examples from each of the control and neuron-Cx36 mice are given. E and F, Histograms showing mean (⫾ SEM) number of litters born (E) and pups per litter (F) over a 3-month period. **, P ⬍ 0.01; ***, P ⬍ 0.001.
population (30, 32). Indeed, we have been able to show here that 50 –75% of kisspeptin neurons located in the RP3V express Cx36. Ideally, it would be very useful to delete Cx36 from specific cell populations in the hypothalamus to evaluate their specific roles. However, apart from the kisspeptin-Cre lines that have only just been reported (40, 41), there are few Cre driver lines available to undertake this task. As such, we started by examining the effects of deleting Cx36 from the majority of forebrain neurons on fertility. Because Cx36 is also expressed outside the brain (16), we used a CaMK-directed Cre transgenic line to target only neurons. Intriguingly, Cx36flox(CFP)/flox(CFP);CKC mice exhibited quite disordered estrous cycles but no defects in puberty onset, fecundity, or estrogen feedback mechanisms, including the generation of the preovulatory LH surge. Mutant mice exhibited estrous cycles in which they remained in the diestrous stage for prolonged periods before showing isolated cycles through proestrus and estrus. The normal fecundity in the presence of disorderd estrous cy-
Endocrinology, June 2011, 152(6):2290 –2301
FIG. 7. Normal estrogen feedback in neuron-specific Cx36 mutant mice. A, Mean (⫾ SEM) LH levels in Cx36flox(CFP)/flox(CFP) (controls) and Cx36flox(CFP)/flox(CFP);CKC⫹/⫺ (neuron-Cx36) female mice (n ⫽ 4 –5 each) when intact, 2 wk after OVX, and 3 h after an injection of estradiol (E2). LH levels are significantly increased by OVX and then suppressed by E2 in both genotypes. ***, P ⬍ 0.01. B, Estrogen-positive feedback results in equivalent numbers of GnRH neurons expressing cFos at the time of the OVX-E2-evoked LH surge in the two genotypes (n ⫽ 5 each). C. Similarly, no significant difference was found in the surge levels of LH detected in the two genotypes.
clicity very likely reflects normal mating-induced ovulation in these mice (42). This unusual reproductive phenotype suggested a situation in which the essential core parts of the GnRH neuronal network mediating pulsatility and spontaneous/induced surge generation may be normal, whereas modulatory inputs that help determine the occurrence of each cycle are disordered. For example, adverse environmental conditions inappropriate for reproduction have the ability to suppress the estrous cycle and generation of the GnRH surge mechanism (43– 46). As yet, it has not been possible to ascribe defined afferent inputs to specific stress modalities with respect to GnRH neurons, but the present data suggest that Cx36 gap junctional coupling is important for one or more of these inputs. An acknowledged caveat of global- and cell-specific gene knockout strategies is the potential for the deleted gene to be compensated for over embryonic/postnatal development. Indeed, amino acid transmitters are thought to be key regulators of GnRH neuron excitability (1, 47), but embryonic deletion of GnRH neuron NMDA or GABAA receptor expression has been found to have minimal effects upon fertility (28, 48). In the course of undertaking this work, the inducible CaMKCreERT2 mouse line (20) became available and provided us with the opportunity to
Endocrinology, June 2011, 152(6):2290 –2301
endo.endojournals.org
2299
mice were given time to recover. Given the impact of gonadal hormones on a wide range of neuronal networks in the brain (49), one corollary is that investigators using tamoxifen-inducible transgenes likely need to wait for 4 wk before examining their mice to ensure the phenotypetheyencounterisnotconfoundedbyabnormal gonadal steroid hormone concentrations. We encountered a similar fertility phenotype in the adult-induced neuronal Cx36-deleted mice; animals exhibited abnormal estrous cycles but normal fecundity. In this case, the estrous cycle phenotype was not as robust as that seen in Cx36flox(CFP)/flox(CFP);CKC females in that the percentage of time spent in diestrus was not significantly different from that of controls, although cycle length remained abnormally elongated. The same phenomenon of persistent diestrus with occasional cycles was evident, but the cycles themselves were slightly different in that, having gone through proestrus, mice then stayed in estrus for 2 or 3 d. Although the inducible Cx36 deletion did not reveal any further fertility phenotype, it does help define the possible brain regions in which Cx36 is required for normal estrous cycles. One of the major differences between Cx36flox(CFP)/flox(CFP); CKC and Cx36flox(CFP)/flox(CFP);CaMKCreERT2 mice is that Cx36 deletion is restricted to a smaller number of brain regions in the latter model. This is due to CaMK being developmentally regulated so that at birth it is expressed by most forebrain neurons, but by adulthood it is expressed only by subsets of possibly glutamatergic neurons (20). Of relevance to the GnRH neuronal network, we found evidence for Cx36 deletion in the medial amygdala, BNST, and both FIG. 8. Abnormal estrous cycles in inducible neuron-specific Cx36 mutant mice. A–C, GFP-immunoreactive cells in the striatum (A), AVPV (B), and SCN (C) of the medial and lateral preoptic area, including tamoxifen-treated Cx36flox(CFP)/flox(CFP);CaMKCreERT2 mice. Scale bars, 50 m. D and the RP3V, in Cx36flox(CFP)/flox(CFP);CaMKCreE, Histograms showing the mean (⫾ SEM) estrous cycle length (D) and percentage of flox(CFP)/flox(CFP) flox(CFP)/flox(CFP) ERT2 mice. This narrows considerably the potime spent in diestrus (E) for Cx36 (controls) and Cx36 T2 CaMKCreER (i-neuron-Cx36) female mice. F and G, Charts showing estrous cycle tential locations of GnRH neuronal network activity for individual mice over an 18-d period. Two representative examples from neurons dependent on Cx36 gap junctions. each of the control and neuron-Cx36 mice are given. H and I, Histograms showing Of particular interest, the medial amygdala mean (⫾ SEM) number of litters born (H) and pups per litter (I) over a 3-month period. *, P ⬍ 0.05. and BNST are implicated in olfactory pathways to GnRH neurons (50), whereas the SCN provides an important circadian input to examine the effects of more acute Cx36 deletion in adults GnRH neurons. We note, however, that OVX estrogenon fertility. The estrogen receptor antagonist tamoxifen flox(CFP)/flox(CFP) ;CKC mice, with Cx36 deused to induce CaMK-Cre expression in this mouse line treated Cx36 leted from the SCN, exhibited normal surge activation of unfortunately has profound effects on reproductive functioning. Nevertheless, wild-type mice given the 5-d tamox- GnRH neurons and LH secretion. This is somewhat surifen regimen returned to normal estrous cyclicity within 4 prising, given the reported importance of Cx36 in enabling wk, indicating that this strategy was useful as long as the the correct circadian output pattern from the SCN (38)
2300
Campbell et al.
Role of Gap Junctions in GnRH Neuronal Network
and the role of the SCN in generating the circadian-locked GnRH surge mechanism (39, 51). Furthermore, the same estrous cycle disruption was seen in Cx36flox(CFP)/flox(CFP); CaMKCreERT2 mice that had only minor Cx36 deletion from the SCN. Although the exact nature of Cx36-deleted SCN neurons is not known, these findings indicate that the SCN may not be the key site of Cx36 deletion underlying the fertility phenotype. In contrast, Cx36 was deleted in the RP3V of both mouse models exhibiting disordered estrous cycles, and this was shown to involve the majority of kisspeptin neurons. These RP3V neurons are established to provide direct inputs to rostral preoptic area GnRH neurons (32) and have been strongly implicated in the estrogen surge mechanism (30, 52). Although estrogen feedback appears normal in Cx36-deleted mice, the concept of gap junctioncoupled RP3V kisspeptin and/or other neurons is intriguing and should now be amenable to testing in the future using kisspeptin-Cre mice. In summary, we provide evidence here that GnRH neurons in vivo do not communicate directly with each other through gap junctions at the level of their cell bodies and dendrites. These neurons do not exhibit dye or electrical charge transfer and do not express Cx36 or Cx43. However, using inducible, cell-specific transgenics, we show that neurons providing afferent inputs to GnRH neurons are dependent on Cx36 gap junctions and that this is critical for normal estrous cyclicity.
4.
5.
6. 7.
8.
9. 10. 11.
12.
13.
14.
Acknowledgments 15.
We thank Dr. H. Urbanski for the gift of the GnRH antibody. 16.
Address all correspondence and requests for reprints to: Professor Allan Herbison, Centre for Neuroendocrinology, Department of Physiology, University of Otago School of Medical Sciences, P.O. Box 913, Dunedin, New Zealand. E-mail:
[email protected]. This work was supported by the New Zealand Health Research Council. Work and personnel in the Bonn laboratory were supported by a grant from the German Research Foundation Wi 270/31-1 (to K.W.). Disclosure Summary: The authors have nothing to disclose.
17.
18.
19.
References 20. 1. Christian CA, Moenter SM 2010 The neurobiology of preovulatory and estradiol-induced gonadotropin-releasing hormone surges. Endocr Rev 31:544 –577 2. Herbison AE 2006 Physiology of the GnRH neuronal network. In: Neill JD, ed. Knobil and Neill’s physiology of reproduction. 3rd ed. San Diego: Academic Press; 1415–1482 3. Campbell RE 2007 Defining the gonadotrophin-releasing hormone
21.
22.
Endocrinology, June 2011, 152(6):2290 –2301
neuronal network: transgenic approaches to understanding neurocircuitry. J Neuroendocrinol 19:561–573 Bose S, Leclerc GM, Vasquez-Martinez R, Boockfor FR 2010 Administration of connexin43 siRNA abolishes secretory pulse synchronization in GnRH clonal cell populations. Mol Cell Endocrinol 314:75– 83 Vazquez-Martinez R, Shorte SL, Boockfor FR, Frawley LS 2001 Synchronized exocytotic bursts from gonadotropin-releasing hormone-expressing cells: dual control by intrinsic cellular pulsatility and gap junctional communication. Endocrinology 142:2095–2101 Charles AC, Kodali SK, Tyndale RF 1996 Intercellular calcium waves in neurons. Mol Cell Neurosci 7:337–353 Matesic DF, Hayashi T, Trosko JE, Germak JA 1996 Upregulation of gap junctional intercellular communication in immortalized gonadotropin-releasing hormone neurons by stimulation of the cyclic AMP pathway. Neuroendocrinology 64:286 –297 Hu L, Olson AJ, Weiner RI, Goldsmith PC 1999 Connexin 26 expression and extensive gap junctional coupling in cultures of GT1–7 cells secreting gonadotropin-releasing hormone. Neuroendocrinology 70:221–227 Bennett MV, Zukin RS 2004 Electrical coupling and neuronal synchronization in the mammalian brain. Neuron 41:495–511 Connors BW, Long MA 2004 Electrical synapses in the mammalian brain. Annu Rev Neurosci 27:393– 418 Campbell RE, Gaidamaka G, Han SK, Herbison AE 2009 Dendrodendritic bundling and shared synapses between gonadotropin-releasing hormone neurons. Proc Natl Acad Sci USA 106:10835– 10840 Traub RD, Pais I, Bibbig A, LeBeau FE, Buhl EH, Hormuzdi SG, Monyer H, Whittington MA 2003 Contrasting roles of axonal (pyramidal cell) and dendritic (interneuron) electrical coupling in the generation of neuronal network oscillations. Proc Natl Acad Sci USA 100:1370 –1374 Tsukahara S, Maekawa F, Tsukamura H, Hirunagi K, Maeda K 1999 Morphological characterization of relationship between gap junctions and gonadotropin releasing hormone nerve terminals in the rat median eminence. Neurosci Lett 261:105–108 Hosny S, Jennes L 1998 Identification of gap junctional connexin-32 mRNA and protein in gonadotropin-releasing hormone neurons of the female rat. Neuroendocrinology 67:101–108 So¨hl G, Maxeiner S, Willecke K 2005 Expression and functions of neuronal gap junctions. Nat Rev Neurosci 6:191–200 Degen J, Meier C, Van Der Giessen RS, So¨hl G, Petrasch-Parwez E, Urschel S, Dermietzel R, Schilling K, De Zeeuw CI, Willecke K 2004 Expression pattern of lacZ reporter gene representing connexin36 in transgenic mice. J Comp Neurol 473:511–525 Wellershaus K, Degen J, Deuchars J, Theis M, Charollais A, Caille D, Gauthier B, Janssen-Bienhold U, Sonntag S, Herrera P, Meda P, Willecke K 2008 A new conditional mouse mutant reveals specific expression and functions of connexin36 in neurons and pancreatic -cells. Exp Cell Res 314:997–1012 Casanova E, Fehsenfeld S, Mantamadiotis T, Lemberger T, Greiner E, Stewart AF, Schu¨tz G 2001 A CamKII␣ iCre BAC allows brainspecific gene inactivation. Genesis 31:37– 42 Wintermantel TM, Campbell RE, Porteous R, Bock D, Gro¨ne HJ, Todman MG, Korach KS, Greiner E, Pe´rez CA, Schu¨tz G, Herbison AE 2006 Definition of estrogen receptor pathway critical for estrogen positive feedback to gonadotropin-releasing hormone neurons and fertility. Neuron 52:271–280 Erdmann G, Schu¨tz G, Berger S 2007 Inducible gene inactivation in neurons of the adult mouse forebrain. BMC Neurosci 8:63 Spergel DJ, Kru¨th U, Hanley DF, Sprengel R, Seeburg PH 1999 GABA-and glutamate-activated channels in green fluorescent protein-tagged gonadotropin-releasing hormone neurone in transgenic mice. J Neurosci 19:2037–2050 Kreuzberg MM, Deuchars J, Weiss E, Schober A, Sonntag S, Wellershaus K, Draguhn A, Willecke K 2008 Expression of con-
Endocrinology, June 2011, 152(6):2290 –2301
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
nexin30.2 in interneurons of the central nervous system in the mouse. Mol Cell Neurosci 37:119 –134 Yamada S, Uenoyama Y, Kinoshita M, Iwata K, Takase K, Matsui H, Adachi S, Inoue K, Maeda KI, Tsukamura H 2007 Inhibition of metastin (kisspeptin-54)-GPR54 signaling in the arcuate nucleusmedian eminence region during lactation in rats. Endocrinology 148:2226 –2232 Herbison AE, de Tassigny X, Doran J, Colledge WH 2010 Distribution and postnatal development of Gpr54 gene expression in mouse brain and gonadotropin-releasing hormone neurons. Endocrinology 151:312–321 Clarkson J, Herbison AE 2006 Postnatal development of kisspeptin neurons in mouse hypothalamus; sexual dimorphism and projections to gonadotropin-releasing hormone neurons. Endocrinology 147:5817–5825 Rash JE, Staines WA, Yasumura T, Patel D, Furman CS, Stelmack GL, Nagy JI 2000 Immunogold evidence that neuronal gap junctions in adult rat brain and spinal cord contain connexin-36 but not connexin-32 or connexin-43. Proc Natl Acad Sci USA 97:7573–7578 Herbison AE, Porteous R, Pape JR, Mora JM, Hurst PR 2008 Gonadotropin-releasing hormone neuron requirements for puberty, ovulation, and fertility. Endocrinology 149:597– 604 Lee K, Porteous R, Campbell RE, Lu¨scher B, Herbison AE 2010 Knockdown of GABA(A) receptor signaling in GnRH neurons has minimal effects upon fertility. Endocrinology 151:4428 – 4436 Clarkson J, d’Anglemont de Tassigny X, Moreno AS, Colledge WH, Herbison AE 2008 Kisspeptin-GPR54 signaling is essential for preovulatory gonadotropin-releasing hormone neuron activation and the luteinizing hormone surge. J Neurosci 28:8691– 8697 Herbison AE 2008 Estrogen positive feedback to gonadotropinreleasing hormone (GnRH) neurons in the rodent: the case for the rostral periventricular area of the third ventricle (RP3V). Brain Res Rev 57:277–287 Clarkson J, d’Anglemont de Tassigny X, Colledge WH, Caraty A, Herbison AE 2009 Distribution of kisspeptin neurones in the adult female mouse brain. J Neuroendocrinol 21:673– 682 Liu X, Porteous R, d’Anglemont de Tassigny X, Colledge WH, Millar R, Petersen SL, Herbison AE 2011 Frequency-dependent recruitment of fast amino acid and slow neuropeptide neurotransmitter release controls gonadotropin-releasing hormone neuron excitability. J Neurosci 31:2421–2430 Kusano K, Fueshko S, Gainer H, Wray S 1995 Electrical and synaptic properties of embryonic luteinizing hormone-releasing hormone neurons in explant cultures. Proc Nat Acad Sci USA 92:918 – 3922 Theis M, So¨hl G, Speidel D, Ku¨hn R, Willecke K 2003 Connexin43 is not expressed in principal cells of mouse cortex and hippocampus. Eur J Neurosci 18:267–274 Alvarez-Maubecin V, Garcia-Hernandez F, Williams JT, Van Bockstaele EJ 2000 Functional coupling between neurons and glia. J Neurosci 20:4091– 4098
endo.endojournals.org
2301
36. Silverman A, Livne I, Witkin JW 1994 The gonadotrophin-releasing hormone (GnRH), neuronal systems: immunocytochemistry and in situ hybridization. In: Knobil E, Neill JD, eds. The physiology of reproduction. 2nd ed. New York: Raven Press; 1683–1706 37. Gulinello M, Etgen AM 2005 Sexually dimorphic hormonal regulation of the gap junction protein, CX43, in rats and altered female reproductive function in CX43⫹/⫺ mice. Brain Res 1045:107–115 38. Long MA, Jutras MJ, Connors BW, Burwell RD 2005 Electrical synapses coordinate activity in the suprachiasmatic nucleus. Nat Neurosci 8:61– 66 39. van der Beek EM 1996 Circadian control of reproduction in the female rat. Prog Brain Res 111:295–320 40. Cravo RM, Margatho LO, Osborne-Lawrence S, Donato Jr J, Atkin S, Bookout AL, Rovinsky S, Fraza˜o R, Lee CE, Gautron L, Zigman JM, Elias CF 2011 Characterization of Kiss1 neurons using transgenic mouse models. Neuroscience 173:37–56 41. Mayer C, Acosta-Martinez M, Dubois SL, Wolfe A, Radovick S, Boehm U, Levine JE 2010 Timing and completion of puberty in female mice depend on estrogen receptor ␣-signaling in kisspeptin neurons. Proc Natl Acad Sci USA 107:22693–22698 42. Bakker J, Baum MJ 2000 Neuroendocrine regulation of GnRH release in induced ovulators. Front Neuroendocrinol 21:220 –262 43. Rivier C, Vale W 1984 Influence of corticotropin-releasing factor on reproductive functions in the rat. Endocrinology 114:914 –921 44. Tomaszewska-Zaremba D, Herman A 2009 The role of immunological system in the regulation of gonadoliberin and gonadotropin secretion. Reprod Biol 9:11–23 45. Dobson H, Ghuman S, Prabhakar S, Smith R 2003 A conceptual model of the influence of stress on female reproduction. Reproduction 125:151–163 46. Bronson FH, Marsteller FA 1985 Effect of short-term food deprivation on reproduction in female mice. Biol Reprod 33:660 – 667 47. Iremonger KJ, Constantin S, Liu X, Herbison AE 2010 Glutamate regulation of GnRH neuronal excitability. Brain Res 1364:35– 43 48. Shimshek DR, Bus T, Grinevich V, Single FN, Mack V, Sprengel R, Spergel DJ, Seeburg PH 2006 Impaired reproductive behavior by lack of GluR-B containing AMPA receptors but not of NMDA receptors in hypothalamic and septal neurons. Mol Endocrinol 20: 219 –231 49. McEwen BS, Alves SE 1999 Estrogen actions in the central nervous system. Endocr Rev 20:279 –307 50. Wersinger SR, Baum MJ 1996 The temporal pattern of matinginduced immediate-early gene product immunoreactivity in LHRH and non-LHRH neurons of the estrous ferret forebrain. J Neuroendocrinol 8:345–359 51. de la Iglesia HO, Meyer J, Schwartz WJ 2003 Lateralization of circadian pacemaker output: activation of left- and right-sided luteinizing hormone-releasing hormone neurons involves a neural rather than a humoral pathway. J Neurosci 23:7412–7414 52. Petersen SL, Ottem EN, Carpenter CD 2003 Direct and indirect regulation of gonadotropin-releasing hormone neurons by estradiol. Biol Reprod 69:1771–1778
