Endocrinology. First published April 24, 2003 as doi:10.1210/en.2002-0188 EN-02-0188-REV3b
Seasonal Plasticity within the GnRH System of the Ewe: Changes in Identified GnRH Inputs and in Glial Association. 1
2
3
2
Heiko T. Jansen , Christopher Cutter , Steven Hardy , Michael N. Lehman , and Robert L. 3 Goodman 1
Department of Veterinary and Comparative Anatomy, Pharmacology and Physiology (VCAPP) and Center for Reproductive Biology, Washington State University College of Veterinary Medicine, Pullman, WA 99164-6520 2 Department of Cell Biology, Neurobiology and Anatomy, University of Cincinnati College of Medicine, Cincinnati, OH 45267-0521 3 Department of Physiology, West Virginia University Health Sciences Center, Morgantown, WV 26506-9229 Short Title: Neural inputs to GnRH neurons vary seasonally Number of text pages: 35 Number of Figures: 8 Number of Tables: 2
Correspondence should be addressed to: Heiko T. Jansen, Dept. VCAPP, Programs in Neuroscience and Center for Reproductive Biology, Washington State University, 205 Wegner Hall, Pullman, WA 99164-6520. VOICE: 509-335-7056 FAX: 509-335-4650 E-mail:
[email protected]. Acknowledgements: This work was supported by USDA 98-35203-6321 to HTJ. The authors are grateful to Mr. Jeremy Kornoely for his technical assistance. A preliminary report of these findings appeared at the 31st Annual Meeting of the Society for Neuroscience, San Diego, CA., 2001. Key Words: GnRH, Season, Plasticity, Glial fibrillary acidic protein, Glutamate decarboxylase, Neuropeptide-Y, Luteinizing hormone, Synapsin-I, estradiol
-1Copyright (C) 2003 by The Endocrine Society
EN-02-0188-REV3b
ABSTRACT The annual reproductive cycle in sheep may reflect a functional remodeling within the GnRH system. Specifically, changes in total synaptic input and association with the polysialylated form of neural cell adhesion molecule (PSA-NCAM) have been observed. 5
Whether seasonal changes in a specific subset(s) of GnRH inputs occur, or if glial cells specifically play a role in this remodeling is not clear. We therefore examined GnRH neurons of breeding season (BS) and non-breeding season (anestrus, AN) ewes and tested the hypotheses that, a) specific (i.e., GABA, catecholamine, NPY, or ß-Endorphin (ß-EN)) inputs to GnRH neurons change seasonally, and b) concomitant with any changes in neural inputs is a change in
10
glial apposition. Using triple-label immunofluorescent visualization of GnRH, glial acidic fibrillary protein (GFAP), and neuromodulator/neural terminal markers combined with confocal microscopy and optical sectioning techniques, we confirm that total numbers of neural inputs to GnRH neurons vary with season and demonstrate that specific inputs contribute to these overall
15
changes. Specifically, NPY and GABA inputs to GnRH neurons increased during BS while ßEN inputs were either greater during AN (GnRH somas) or greater during BS (GnRH dendrites). Associated with the changes in GnRH inputs were seasonal changes in glial apposition, in GFAP density, and in the thickness of glial fibrils. These findings are interpreted to suggest an increase in net stimulatory inputs to GnRH neurons during the BS contributes to the seasonal changes in
20
GnRH neurosecretion and that this increased innervation is perhaps stabilized by glial processes.
-2-
Jansen, HT, et al.
INTRODUCTION Yearly cycles of fertility and infertility in sheep and other seasonal breeders are thought to reflect the outcome of several endogenous neural and endocrine processes. Together, these interactions impinge upon the GnRH neuron, the final common pathway in the control of 5
reproduction, to produce a neuroendocrine rhythm of GnRH secretion and subsequent gametogenesis of approximately one year in length (1). An important component of the annual GnRH rhythm in the female is that it reflects a profound change in the brain's sensitivity to estradiol negative feedback (2, 3). Although the neural mechanism of steroid feedback is unknown, it is likely due in part, to changes in the phenotype and activation of specific GnRH
10
afferents (4). Intriguingly, evidence from a number of different species suggests that the brain is capable of undergoing annual synaptic remodeling; thus, seasonal neural plasticity in the GnRH system may provide an important contribution to seasonal breeding. Perhaps the best-studied example of seasonal plasticity is that expressed in the songbird brain. For example, annual song learning is associated with recurring changes in both the size of
15
specific nuclear groups and in neural connectivity (5-8). Both types of changes are dependent upon gonadal steroids and can be driven by changing the steroid milieu (9, 10). Within the GnRH system of some bird species, changes in neuronal morphology and gene expression have also been reported (11-15). However, whether these changes in the avian GnRH system also reflect an underlying change in neuronal connectivity, glial associations, or other mechanisms, is
20
not clear. In mammals, examples of short-term (hours to days) neural plasticity are abundant. Dehydration and lactation both result in profound increases in synaptic inputs onto hypothalamic magnocellular neurons and changes in glial association (16, 17). During the estrous cycle,
-3-
Jansen, HT, et al.
changes in neural connectivity (18, 19) and glial associations (20) have been documented. Examples of long-term (months) seasonal neural plasticity in mammals are less abundant, despite the fact that many physiological parameters (e.g. behavior, endocrine status, reproduction) in temperate zone species exhibit robust seasonal variation. At best, our 5
knowledge of the neural processes underlying these events is incomplete; however, a recent report demonstrated that the number of synaptic inputs onto GnRH neurons increase significantly during the BS when compared to AN in sheep (21). A variety of different types of neural inputs to GnRH neurons have been identified and together with a large body of experimental evidence suggest they may serve important
10
physiological functions (for reviews: see (22-26). For seasonal breeders, such as sheep, these inputs may be necessary to convey steroid feedback information to the GnRH neuron. Indeed, the sheep has become a model species in which to examine the neural mechanisms underlying seasonal changes in steroid feedback within the reproductive neuroendocrine axis. For example, sheep exhibit seasonal changes in GnRH neurosecretion (3, 27), seasonal changes in total
15
numbers of GnRH inputs (21) and seasonal changes in activation of potential GnRH afferents (28, 29) associated with gonadal steroid feedback. Anatomical and physiological studies in this species have implicated catecholaminergic, NPYergic, GABAergic and opioidergic systems, among others, as potential transducers of steroid feedback onto the GnRH system. Specifically, dopamine is suggested to mediate estrogen negative feedback via an inhibitory influence on
20
GnRH system during anestrus (29-32) whereas the role of norepinephrine may vary depending on the stage of the annual reproductive cycle (30, 33, 34). An inhibitory influence of GABA on the GnRH system is suggested and may be expressed during negative feedback in anestrus and during negative feedback throughout the estrous cycle (35, 36). Opioid peptides, such as ß-
-4-
Jansen, HT, et al.
endorphin, also appear to exert an inhibitory influence on GnRH during progesterone-mediated and estradiol-mediated negative feedback (37-39). For NPY, both stimulatory and inhibitory influences on the GnRH system have been reported (40-43). Despite the evidence in support of a role for various neurochemical systems in regulating GnRH at different times of the year and 5
under different reproductive conditions, it has not been established, in any seasonal-breeding mammal, whether this regulation occurs because of changes in inputs to GnRH neurons directly. While GnRH neurons receive direct neural innervation (albeit, less than their non-GnRH expressing neighbors within the POA in sheep (21, 44-47)) they are also extensively associated with non-neuronal elements, specifically glial cells (45-47). Therefore, glial cells may provide
10
an important method for modulating the number of inputs onto the GnRH neuron, as has been suggested in the monkey (45, 46) and sheep (25). Additional evidence supporting a role for glia in modulating GnRH function comes from our own studies demonstrating that the association between GnRH neurons and the cell adhesion molecule, PSA-NCAM, varies seasonally (48). PSA-NCAM is expressed in both glia and neurons (49-51) and studies in the rat demonstrate that
15
removal of PSA from the NCAM molecule in vivo blocks both the glial retraction and increase in synaptic inputs onto magnocelluar neuroendocrine neurons during lactation and dehydration or after estradiol administration (49, 52). Given the observation that a seasonal change in total innervation of GnRH neurons is associated with seasonal changes in reproductive activity in the ewe, we asked whether this
20
reflected changes in specific neural afferents, changes in glial associations, or both.
-5-
Jansen, HT, et al.
MATERIALS AND METHODS Animals Mature blackface ewes (n=12) were ovariectomized using sterile techniques under pentobarbital anesthesia at least 6 weeks prior to sacrifice and immediately received 5
subcutaneous estradiol implants to maintain constant low (luteal phase) physiological estrogen concentrations (2, 53). Sheep were maintained outdoors in open pens under natural photoperiod at the West Virginia University Research Farm (39º18’N latitude), were fed silage daily and had free access to water and mineral blocks1. Prior to euthanasia, jugular blood samples were collected by venipuncture for four hours
10
from all ewes at either 10 min (BS) or 20 min (AN) intervals. Serum was subsequently assayed for luteinizing hormone (LH) concentrations as previously described (54). Following blood collection, animals were heparinized and then killed with an overdose of sodium pentobarbital during AN (August, N=6) or BS (December, N=6). The heads were immediately perfused with 6l 4% paraformaldehyde in 0.1M phosphate buffer (PB; pH 7.4) containing 0.1% sodium nitrite
15
as a vasodilator (26). The forebrain and hypothalamus was removed, blocked and then placed in fixative for an additional 24 h followed by cryoprotectant (20% sucrose in PB) until infiltrated (2-3 days). Frozen sections (55µm) extending from the diagonal band of Broca rostrally to the mammillary nuclei, caudally, were then collected in six series (distance between sections within a series = 330µm) and subsequently stored in cryopreservative (55). Each series of sections was
20
then processed for immunocytochemical identification of GnRH, glial processes, and either total neural inputs or specific neuromodulatory afferents as described below. Postmortem
1
All procedures were approved by the West Virginia Institutional Animal Care and Use Committee and were performed in accordance with the ‘Guide for the Care and Use of Laboratory Animals’. -6-
Jansen, HT, et al.
examination revealed that two BS ewes had lost their estrogen implants at some time before being euthanized.
Western Blotting 5
Sheep brains (N=3) were obtained from the Meat Science Laboratory at Washington State University. After slaughter, the preoptic area/hypothalamus was dissected out and immediately frozen at –80ºC. Total protein was extracted from each of the three brains as described previously (56) and then subjected to SDS-PAGE (7.5% resolving + 4% stacking gel) using a Bio-Rad Criterion apparatus. Proteins were resolved at 50mA for 60 min and then were
10
transferred to PVDF membranes (12V for 2h; Idea Scientific, Minneapolis, MN). Proteins on gels were visualized with Coomassie Blue staining. After transfer, the membranes were incubated overnight in blocking buffer (Bio-Rad, Hercules, CA) containing 3% BSA and on the following day were transferred to primary antiserum (anti-Synapsin I, Molecular Probes, Eugene OR; diluted 1:200) containing blocking buffer, 3% BSA and 5% normal donkey serum.
15
Membranes were incubated overnight for 16 hours at room temperature with shaking. On the following day, the membranes were washed in PBS and transferred to horseradish peroxidaseconjugated donkey anti-rabbit IgG (diluted 1:10,000 in blocking buffer) and incubated for 1 hour at room temperature with shaking. Immunoreactive bands were visualized using chemiluminescence methods (ECL, Amersham) and subsequently recorded on X-ray film.
20 Immunocytochemistry Tissue sections were removed from cryopreservative and washed 4 times, 5min each, in PB containing 0.1% Triton-X 100 (PBTX) with shaking followed by a 1hr incubation in a
-7-
Jansen, HT, et al.
cocktail of blocking sera (1% each) specific to the species in which the secondary antibodies were raised. Next, the sections were incubated in a cocktail of primary antibodies against GnRH, glial fibrillary acidic protein (GFAP) and one of the following: Synapsin-I (SynI), tyrosine hydroxylase (TH), neuropeptide Y (NPY), ß-endorphin (ß-EN), or glutamate decarboxylase 5
(GAD). The antibody cocktail was made in PBTX containing 3% blocking sera plus 0.01% sodium azide; sections were incubated for 48hr at 4ºC with shaking. On the third day, sections were rinsed in PB, 4 times, 5 min each and then incubated in fluorescent secondary antibodies (see Table 1; diluted 1:100) appropriate for GnRH (anti-rabbit or mouse Cy3-conjugated), GFAP (anti-guinea pig Cy5-conjugated) for 1 hr at room temperature with shaking. Sections were
10
washed 4 times, 5 min each. To enhance the visualization of Syn-I, GAD, ß-EN, and TH immunoreactivity (ir), the fluorescent signal (Alexa 488) was intensified using a multiplebridging amplification method as follows. Sections were transferred to fluorescent secondary antibody (diluted 1:100) for 30 min and then washed. This was followed by incubation in unlabeled bridging IgG (10µg/ml) specific to the species in which the primary antibody was
15
raised (e.g. rabbit, mouse) for 30 min. After additional washes, the entire process (primarysecondary) was repeated twice more. Because of the intensity of NPY fluorescence, amplification was not required. All sections were processed through a final series of washes and then mounted on SuperFrost-plus slides (Fisher), dried in the dark overnight and coverslipped with Gelvatol (57).
20 Antibodies Antibodies used in the present study are listed in Table 1. The characteristics of GnRH antibodies have been previously described (58, 59). For the other antibodies, controls consisted
-8-
Jansen, HT, et al.
of either preincubating the antibody cocktail with 1-10µg purified antigen (anti-NPY, -Syn-I, -ßEN) or omission of the primary antibody (anti-GAD, -GFAP, -TH); in each case, specific staining was lost. Additional controls consisted of omission of one of the three secondary antibodies; again, this resulted in the complete loss of signal for the particular antigen (species) 5
against which the antibody was raised (not shown).
Imaging and Analysis Slides were viewed and the confocal images obtained using a Zeiss LSM-510 laserscanning microscope equipped with Ar, and He/Ne lasers capable of producing monochromatic 10
excitation at 458/488/514, and 543/643nm wavelengths, respectively. The pinhole diameter was optimized to 1.0 Airy disk. Image size was set at 512 x 512 pixels. Scans at each wavelength were performed independently (multitracking) to eliminate bleed-through between individual channels. Individual GnRH neurons were visualized at x20 magnification under epifluorescent
15
(mercury vapor) illumination. Then, the illumination was switched to confocal mode (laser) and optical sectioning (Z-thickness, 0.45µm) of a selected GnRH neuron was performed using an x63 C-Apochromat water immersion objective (numerical aperture 1.2). For each GnRH neuron, twenty optical sections, each composed of scans through three separate channels (488, 543, and 643nm excitation) were captured and saved (Fig. 1). Composite images, as well as individual
20
images from each channel, were then converted to TIF format, saved, and subsequently analyzed using NIH Image software (Object IMAGE, v. 2.07) as described below. GnRH neurons were grouped, post hoc, into either rostral or caudal populations based on their location rostral to, or caudal to, the midline crossing of the anterior commissure, respectively. The plane of sectioning
-9-
Jansen, HT, et al.
was adjusted for each brain such that similar landmarks were present within a section between animals. For each ewe, approximately 50 GnRH neurons were analyzed (30 rostral, 20 caudal). The following parameters were then recorded from each optical section for each GnRH neuron: 5
1) total length of GnRH soma membrane, 2) total length of membrane from each proximal dendrite (defined as beginning 4µm from the nuclear envelope), and 3) the area of the cell soma. Then, for each GnRH neuron the number of inputs (either total, i.e, Syn-I-ir, or specific) onto the soma and dendrite was determined as described below. For each Syn-I-ir input, the length of the immunoreactive portion of the terminal in contact with the GnRH neuron was determined.
10
Inputs to GnRH neurons were identified based on the proximity of terminal immunoreactivity to the presumptive GnRH cell membrane. Although the limit of resolution with the light microscope does not allow identification of synaptic specializations, and thus bona fide synapses, additional criteria were developed to maximize the identification of axon terminals (i.e., GnRH inputs). Previous studies using confocal microscopy to examine close appositions
15
between axon terminals and GnRH neurons (60) were limited in their ability resolve close associations and thus we incorporated GFAP immunoreactivity as an additional means to eliminate false positives. Thus, only those terminals (green pixels) apposed to the GnRH cell membrane (red pixels) and not having any GFAP-ir elements (blue pixels) interposed between them were considered (Fig. 1). The sum of all Syn-I-ir GnRH inputs (soma and dendrite) was
20
then used to provide an estimate of the total number of afferents per GnRH neuron. Based on these results, the percentage of the total number of somatic and dendritic GnRH inputs represented by each type of neurotransmitter / neuropeptide input was derived.
- 10 -
Jansen, HT, et al.
Those GnRH neurons processed for determination of total numbers of inputs (i.e., were Syn-I-ir), were also used to estimate the amount of glial association with the soma and proximal dendrites. Data from the individual channels in three optical sections were used to calculate the percentage of GnRH-ir membrane (red pixels; both soma and proximal dendrite) in direct contact 5
with GFAP–ir elements (blue pixels) according to the formula: (LGFAP/LGnRH ) X 100, where LGFAP = length of GFAP-ir pixels in contact with GnRH-ir membrane and LGnRH = length of GnRH-ir pixels associated with either the soma or dendrite(s). The overall density of GFAP-ir was also estimated in 100 randomly selected fields of GnRH neurons using the method described previously (48) with a minor modification to increase the size of the area analyzed from 150
10
pixels to 260 pixels in order to incorporate more of the GnRH proximal dendrite(s). First, for each 512 x 512 pixel field, GFAP density was estimated within an area containing the GnRH neuron and proximal dendrite(s) and then in an area of the same field but devoid of GnRH-ir (usually the lower right quadrant). Finally, within the area lacking GnRH-ir, the diameter of individual glial filaments was quantified by drawing a line (15µm in length) bisecting the region
15
of interest and oriented perpendicular to the majority of glial filaments in the field. The distribution of gray scale values along the line was then plotted and for each plot, a threshold value 2 SD above the minimum was computed. The diameter of each glial fiber was estimated from the points at which the peak crossed the threshold value. This procedure was repeated in three adjacent optical sections centered about the middle of each z-series.
20 Statistical Analysis For each parameter, the mean ± SEM was determined for all rostral or caudal neurons in each animal. Statistical comparisons were made by three-way ANOVA (main effects: season,
- 11 -
Jansen, HT, et al.
cell location, dendrite vs. soma; and interactions). LH Pulse parameters were identified using the Pulsar algorithm as described previously (61, 62) and compared between seasons by MannWhitney-U test. Post-hoc comparisons of group means were performed by Fisher’s LSD analysis. Differences were considered statistically significant if p < 0.05. 5 RESULTS A. Neuroendocrine Status Examination of LH profiles during BS and AN confirmed the expected seasonal neuroendocrine status of ewes. Those ewes from which tissue was collected during BS 10
(December) exhibited nearly three times as many LH pulses when compared to AN ewes (August; p
