Effects of gonadal steroids on the ultrastructure of GnRH neurons in the rhesus monkey: synaptic input and glial apposition.

0013-7227/91/1292-1083$03.00/0 Endocrinology Copyright vtf 1991 by The Endocrine Society

Vol. 129, No. 2 Printed in U.S.A.

Effects of Gonadal Steroids on the Ultrastructure of GnRH Neurons in the Rhesus Monkey: Synaptic Input and Glial Apposition* JOAN W. WITKIN, MICHEL FERIN, SULLI J. POPILSKIS, AND ANNJUDITH SILVERMAN Departments of Anatomy and Cell Biology (J. W. W., A.-J. S.), Obstetrics and Gynecology (M. F.), and the Institute of Comparative Medicine (S. J. P.), Columbia University College of Physicians and Surgeons, New York, New York 10032

There was also a higher incidence of GnRH neurons with immunostaining confined to secretory granules and a decrease in the volume of nucleoli, both of which could be interpreted as indications that GnRH peptide synthesis was reduced in ovariectomized animals. After an ovarian steroid replacement regimen which mimicked two menstrual cycles, the innervation of GnRH neurons was increased and the glial ensheathment was partially reduced. This was true for both the LtOVX+ and StOVX+ steroid-replacement groups. GnRH neurons in the medial basal hypothalamus received more synaptic input than did those in the preoptic area, regardless of the steroid condition of the animal. The degree of glial ensheathment of GnRH neurons in the preoptic area became significantly greater than that in the medial basal hypothalamus after ovariectomy. These observations suggest there may be differences in the role of GnRH neurons in these two brain regions. These immunocytochemical ultrastructural studies provide strong evidence that alterations in the gonadal steroid milieu can produce morphological changes in the GnRH neuron and its immediate environment in the primate. (Endocrinology 129: 1083-1092, 1991)

ABSTRACT. The secretion of the gonadotropins is modulated by the gonadal steroids, but the means by which these effects are mediated are not well understood. The present anatomical study was undertaken to investigate the possibility that the GnRH system responds to alterations in the gonadal steroid environment with reversible changes in synaptic input and glial wrapping such as have been observed in other neuroendocrine systems. The ultrastructure of GnRH neurons was studied in the preoptic area and medial basal hypothalamus of rhesus monkeys in various steroid conditions including five intact cycling, four long-term ovariectomized animals, two long-term ovariectomized animals with steriod replacement (LtOVX+), and two animals replaced with steroid at the time of ovariectomy (StOVX+). Electron micrographic montages of GnRH neuronal profiles were analyzed using computerized morphometrics, and the percentages of the length of perikaryal membrane immediately apposed by glial processes and that with postsynaptic modification were calculated. Ovariectomy resulted in a significant increase in the apposition of glial processes to GnRH perikaryal membranes and a significant decrease in their innervation in both brain regions.

T

HE HYPOTHALAMIC peptide, GnRH, is secreted in an episodic fashion, imparting a pulsatile pattern upon the secretion of the gonadotropins (1, 2; for reviews see Refs. 3,4). The release of gonadotropins is modulated by gonadal steroid hormones; however, the means by which this modulation is effected are as yet unknown. In primates (5, 6) as well as other species (7), there is strong evidence that at least part of this regulation by steroids is upon the GnRH neuron. It is unlikely to be at the genomic level, as GnRH neurons apparently do not con-

Received March 29,1991. Address all correspondence and requests for reprints to: Dr. Joan W. Witkin, Department of Anatomy and Cell Biology, Columbia University College of Physicians and Surgeons, 630 W. 168th Street, New York, New York 10032. * A preliminary report of a portion of this work has been presented (Soc Neurosci 13:18, 1987). Supported by NIH Grants DK-42323, and HD-05077.

tain estrogen receptors (8-11). It is more likely that GnRH activity is mediated by other means such as through synaptic input from populations of neurons that are steroid-concentrating. In the rodent alterations in synapse formation (12-14) and in neuronal morphology (15-17) have been shown to occur under the influence of gonadal steroids. Glia constitute an important component of the neuronal milieu, and there is a strong indication that the extension of glial processes is highly plastic (for reviews, see Refs. 18, 19). The proportion of neuronal membrane covered by glia diminishes in the oxytocinergic system during lactation (18) and in the vasopressinergic system during dehydration (20), states of activation of the respective neurons. In the neurohypophysis, pituicyte processes recede from between adjacent neurosecretory terminals and from the perivascular space under conditions

1083

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 16 November 2015. at 15:58 For personal use only. No other uses without permission. . All rights reserved.

1084

STEROID EFFECT OF GnRH NEURON ULTRASTRUCTURE

of stimulation (18, 20). Alterations in glial morphology are also seen with changes in the steroid environment (14, 21-24). The present immunocytochemical investigation was undertaken to examine the ultrastructure of GnRH neurons and their surrounding milieu as well as their synaptic input under various steroid conditions. For this purpose, comparisons were made of intact cycling, ovariectomized, and steroid-replaced female rhesus macaques. The aim of the study was to provide insights into the anatomical substrate for the effects of gonadal steroids upon GnRH release in the primate.

Materials and Methods Thirteen adult female rhesus macaques were used in this study. They had previously been in a variety of noninvasive experiments entailing blood sampling. Five were intact and killed at different phases of the menstrual cycle (two follicular, one midcycle, two luteal), four were ovariectomized (one for 6 weeks and three for more than 1 yr) (OVX), two were ovariectomized (for more than 1 yr) and replaced with ovarian steroids (LtOVX+), and two were given ovarian steroids immediately after ovariectomy (StOVX+). For steroid replacement, estradiol and progesterone-containing capsules were implanted sc for 2 months before the experiment in a design to simulate two menstrual cycles and to reproduce sequential follicular and luteal phase hormone levels. The capsules were prepared from silastic tubing (Silastic brand tubing, 3.3 mm id, 4.6 mm od; Dow Corning Co, Midland, MI) and filled to 3 cm with estradiol or to 4 cm with progesterone. At killing, levels of estradiol ranged between 82-548 pg/ml and progesterone between 4.6-6 ng/ml, values within normal ranges for cycling monkeys. Before brain perfusion, monkeys were sedated with ketamine hydrochloride (Ketaset, Aveco Co., Fort Dodge, 1 A: 5-7 mg/kg), then anesthetized with pentobarbital (25-30 mg/kg). Catheters were inserted into both carotid arteries for fixative infusion, and the aorta was clamped. The brain was perfused with 2 liters of fixative (4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.3 PB). Brains were removed and dissected into blocks encompassing the preoptic area (POA) and the medial basal hypothalamus (MBH). These blocks were postfixed for 2 h in the same fixative at room temperature. Fiftymicrometer sections were cut on the vibratome, collected in PB, and prepared for the immunocytochemical demonstration of sites of GnRH (25). In brief, sections were washed in PB and treated for 20 min with 0.5% H2O2 to reduce endogenous peroxidase activity, again washed and treated for 30 min with 1% normal goat serum to eliminate nonspecific background staining. Sections were then

Endo • 1991 Voll29-No2

incubated in LRl, the antibody to GnRH, (prepared in rabbit using a D-Lys6 LRF glutaraldehyde ovalbumine conjugate, gift of R. Benoit, Montreal General Hospital, Montreal, Quebec, Canada) at 1:20,000 in PB with 0.05% saponin for 48-72 h. (The antigenic determinant of this antibody consists of amino acids 3, 4, 7, 8, 9, and 10 of the decapeptide, and it recognizes the decapeptide both as a prohormone and in its mature, cleaved, and amidated form. Benoit, R., personal communication; 26.) Sites of antigen-antibody complexes were demonstrated using the ABC Vectastain kit (Vector Labs, Inc., Burlingame, CA) and 3,3'-diaminobenzidine (Sigma, St. Louis, MO) (DAB) as the chromogen. In some cases the DAB was intensified using a silver/gold method (27). For controls, the primary antibody was omitted or preabsorbed overnight with synthetic GnRH (Peninsula, Belmont, CA; 1 ng/m\ LRl, 1:20,000; 26). Tissue sections from the POA and the MBH were examined, and regions containing GnRH neurons were removed and osmicated (2% osmium tetroxide in normal saline with 1.5% potassium ferricyanide) for 1 h, dehydrated and embedded in Epon. Cells were located in semithin, 1-ixm sections. Thin (70 nm) sections were then cut on an AO Ultracut microtome (Reichert-Jung, Vienna, Austria) and collected on formvar-coated slot grids. In some cases sections were counterstained with ethanolic uranyl and lead citrate. Each neuron was photographed at three levels of section (with about 300 nm intervening between sections) using a Jeol 1200 EX electron microscope (Jeol USA Inc., Peabody, MA). Electron micrographic montages at a magnification of 8,000x were made of the entire perikaryal region in sections through the plane of the nucleus, averaging about six micrographs per neuronal profile. The perikaryon of all cells was defined as the area within 10 (im of the nuclear membrane. A minimum of five GnRH neurons were photographed for each brain region of each animal. Photographic montages were also made of five random nonidentified neurons per brain region from the same sections from which the GnRH neurons were photographed. These were selected by choosing the first neuron with a nucleus and nucleolus in the plane of section. Micrographs were printed with final magnifications of 20,000x, and measurements were made using the R & M Biometrics (Nashville, TN) morphometry program (Biometric System IV) linked to an IBM AT computer and a digitizing pad. The measurements included the lengths of total GnRH perikaryal membrane and that with glial wrapping or with synaptic modification. Synaptic length was defined for the purposes of this study as membrane along which there was any recognizable synaptic structure: a synaptic cleft, a postsynaptic density or, in the case of en face sections, a density across both pre- and postsynaptic elements. The percentages of


STEROID EFFECT OF GnRH NEURON ULTRASTRUCTURE neuronal membrane ensheathed with glial processes and that with synaptic modification were calculated. Frequency distributions revealed that the data did not conform to normal distributions so that nonparametric statistical tests were required. The Mann-Whitney U statistic (significance, P < 0.05) was used to test for differences among animals within experimental groups and between experimental groups. This statistical test compares the rank order of observations in two samples. It does not require that the two samples contain the same number of observations. However, as a further assurance that our samples were not biased, we performed this test both on a subset of the entire data using only the first five neurons observed for each sample as well as on the entire set of observations which included up to 17 neurons in one brain region for one animal. The results of these tests were identical. Data from the intact animals in various phases of the menstrual cycle were not different, hence they were grouped. Likewise, data from the animal which had been ovariectomized for six weeks were not different from the three which had been ovariectomized for more than 1 yr, and so these animals were considered a single group.

1085

(L

Results Immunocytochemical staining characteristics and morphology of GnRH neurons Heaviest concentrations of GnRH neurons were found in the POA and MBH as had been described earlier in the rhesus monkey (28, 29). The distribution of GnRH neurons and fibers within the brain region studied was not different among the experimental groups. The majority of these neurons was fusiform, though in all groups there were many examples of neurons with more complex dendritic arbors. At the ultrastructural level, the distribution of the DAB reaction product varied among cells (Figs. 1-3). In some it was confined to secretory vesicles (Fig. 3), whereas in others it was dispersed throughout the cytoplasm (Figs. 1 and 2) sometimes so densely that subcellular structures could not be differentiated. The proportion of lightly stained neurons (secretory vesicles only) in both brain regions was greater in ovariectomized animals: 18% (POA) and 17% (MBH) in OVX; 0% (POA) and 6% (MBH) in intact. Reaction product was not seen (in any group) in association with the Golgi apparatus (Fig. 3), except occasionally in some peripheral vesicles which might be interpreted as transfer vesicles, nor was it seen in the cisternae of the rough endoplasmic reticulum (Fig. 3). Changes in the morphology of nuclear constituents have been associated with differences in protein synthetic activity. Under conditions promoting protein synthesis, the nucleolar complex is enlarged, and there is

II FIG. 1. GnRH neuron from MBH of an intact cycling monkey. There are two synapses on this profile (arrows) and little glial apposition (arrowheads). Reaction product is scattered in cytoplasm, exclusive of regions containing Golgi apparatus (G). Nucleus (N) is immunoreactive, except for nucleolus (n), and is highly indented. L, Lysosomes. No contrast staining. Magnification of original photomicrograph, 10,000x.

less infolding of the nuclear membrane (30). As a means of identifying such differences, the nuclei of the GnRH neurons were compared among experimental groups. In order to insure that sampling was random, the photographic montage of the first profile of each neuron in which there was an entire nuclear outline was used. The incidence of nucleoli (which is an estimate of nucleolar volume) was tabulated and found to be greater in the intact and steroid-replaced animals than in OVX (Table 1). Interestingly, the GnRH cells in the MBH of the two StOVX-f animals were radically different from each other. In one, the incidence of nucleoli was zero, whereas in the other it was 80%. (These results are consistent with the divergence in other measurements in these two animals, see other results below and Discussion.) On the same micrographs, the degree of indentation of the nuclei was rated on a scale of zero to 3. Nuclear indentation



1086

'

1 " .

•»

* •

Endo • 1991 Vol 129 • No 2

'

FIG. 3. Portion of GnRH neuron from medial basal hypothalamus of a monkey ovariectomized for 6 weeks. Reaction product is seen only in secretory granules (arrows). Golgi apparatus (G) and rough endoplasmic reticulum are not stained, nor is nucleus (N). No contrast staining. Magnification of original photomicrograph 25.000X. FIG. 2. GnRH neuron from medial basal hypothalamus of a LtOVX+ monkey. The massive glial ensheathment (g) is as hypertrophied as was ever observed in any ovariectomized animal. There are no synapses on this profile. Reaction product is scattered in the cytoplasm, and secretory granules (sg) are prominently stained. Indented (arrowheads) nucleus (N) does not contain a nucleolus in this plane of section. No contrast staining. Magnification of original photomicrograph 10,000x.

was not found to differ between brain regions or among experimental groups, nor did it vary consistently with presence of nucleoli. In about half of the GnRH neurons in all experimental groups, there was DAB reaction product scattered in the nucleus (Figs. 1 and 2). In silverintensified preparations, it appeared to be associated with chromatin, whereas the nucleolus was entirely free of reaction product. The basis for this apparent immunoreactivity is unknown. The nuclei of nonidentified neurons never contained DAB reaction product. Glial ensheathment of GnRH neurons GnRH neurons were in contact with both neuronal and glial elements in the neuropil. There was a wide range in the amount of GnRH neuronal membrane that

TABLE 1. Percent of GnRH neurons with nucleoli in the plane of section (as an indication of the volume of the nucleolus) in two brain regions (POA and MBH) in intact cycling (Intact); ovariectomized (OVX); long term ovariectomized, ovarian steroid-replaced (LtOVX+); ovariectomized and immediately ovarian-steroid replaced (StOVX+) rhesus monkeys

POA MBH

Intact

OVX

U0VX+

StOVX+

41% (36) 32% (32)

17% (26) 21% (23)

50% (12) 50% (12)

10% (10) 0/80% (14)'

One profile was used for each neuron (the first profile photographed in which there was an entire nuclear outline). Number of neurons in each sample are in parentheses. 0 The incidence of nucleoli in the MBH of the two animals in this group was very different. One animal had 0% and the other had 80%.

was apposed by glial processes; a few neurons had no apparent glial wrapping (Fig. 1), whereas some were almost entirely ensheathed (Fig. 2). Comparisons of the percent of GnRH neuronal membrane in apposition to glial processes by brain region among the groups were made using frequency histograms and nonparametric statistics (Mann-Whitney U test) (Fig. 4). In cycling


STEROID EFFECT OF GnRH NEURON ULTRASTRUCTURE MBH

POA

50

1087

40Neuro

s

FIG. 4. Frequency distribution of GnRH neurons with various percentages of GnRH perikaryal membrane ensheathed with glia (percent glial wrap) in POA and MBH in three groups of monkeys: intact, OVX, and LtOVX+. Differences between groups were tested using the Mann-Whitney U nonparametric test on ungrouped data. N is number of neurons observed. Numbers in parentheses are number of animals per group: POA N neurons Intact(5) OVX (4) LtOVX+(2) StOVX+(2) 0

37 35° 12*c 13*

X K

3020-

o 10-

0

10 20 30 40 50 60 70 80 90

0

10 20 30 40 50 60 70 80 90

MBH Nneurons 31 25* 14* 14"

Greater than intact ( P < 0.001).6 Greater than intact ( P < 0.01).c Less than OVX ( P < 0.05).d The glial ensheathment of GnRH neurons in the MBH of the two StOVX+ animals was different. In one it was similar to that in intact and the other, to ovariectomized animals.

..llli.il. 11. .. 0

10 20 30 40 50 60 70 80 90

0

10 20 30 40 50 60 70 80 90

0

10 20 30 40 50 60 70 80 90 % Glial Wrap

0

10 20 30 40 50 60 70 80 90 % Glial Wrap

animals, there were no completely ensheathed GnRH neurons in either the POA or the MBH. After ovariectomy, ensheathment increased significantly in both brain regions, particularly in the POA. In the POA of intact animals, more than 40% of the GnRH neurons had less than 10% of their membrane with glial apposition. After ovariectomy, 90% of the GnRH neurons in the. POA had more than 10% glial wrapping. Although there was an increase in glial ensheathment of GnRH neurons in the MBH after ovariectomy, it resulted in a level significantly below that in the POA. In LtOVX+ animals, the ensheathment was reduced in the POA, but was still significantly above the intact level, whereas in the MBH the ensheathment was not reduced. Levels of ensheathment were the same in the MBH as in the POA in these animals. StOVX+ animals had glial ensheathment in the POA which was not significantly different from that in the ovariectomized group, i n the MBH of these two

StOVX+ animals, the degree of glial ensheathment differed. In one the ensheathment was similar to OVX, whereas in the other it was similar to intact animals (and was consistent with the incidence of nucleoli). The degree of ensheathment of random nonidentified neurons was compared in both brain regions using the same nonparametric test. It ranged between zero and 54% and was found not to vary systematically across the treatment groups. Density of synaptic input to GnRH neurons The synaptic input to the GnRH neurons in the MBH was significantly greater than to those in the POA in all experimental groups (Fig. 5). There was a range in percent of GnRH neuronal membrane with synaptic modification between 0-3% in the POA and between 0-5% in the MBH. After ovariectomy, the input was significantly decreased in both brain regions such that in the POA



1088

Bndo«1991 Vol 129 »No 2

MBH

POA 80

40

INTACT

INTACT F I G . 5. Frequency distribution of G n R H neurons with various percentages of G n R H perikaryal membrane with postsynaptic modification (percent synaptic) in POA and M B H in three groups of monkeys: intact; OVX; and L t O V X + . Differences between groups were tested using the Mann-Whitney U nonparametric test on ungrouped data. N is number of neurons observed. Numbers in parentheses are number of animals per group: POA N neurons Intact(5) OVX (4) LtOVX+(2) StOVX+(2)

37 35°

60-

30-

40-

20-

20

10-

it 0

.5 1 1.5 2 2.5 3 3.5 4 4.5 5

0

0

.5

0

till.. . .5

1 1.5 2 2.5 3 3.5 4 4.5 5

.5

1 1.5 2 2.5 3 3.5 4 4.5 5

MBH N neurons 31 25" 14C 14"

0 Less than intact (POA, P < 0.01; MBH, P < 0.001).* Greater than intact (P < 0.001).c Greater than OVX ( P < 0.001). d T h e density of synaptic input to the M B H neurons of the two S t O V X + animals differed. In one it was similar to that in intact and the other, to ovariectomized animals. G n R H neurons in the POA were less highly innervated than those in the M B H in all experimental groups: intact and OVX, P < 0.001; U 0 V X + , P < 0.05.

1 1.5 2

3 3.5 4 4.5 5

100

UOVX+

LtOVX*

80-

30-

60-

1

204010-

20-

0- •

0

.5

1 1.5 2 2.5 3 3.5 4 4.5 5 % Synaptic

more than 80% of the GnRH neurons had less than 1% of their membrane innervated. In the MBH this decrease was even more pronounced. Steroid replacement after long term ovariectomy resulted in an increase in innervation to GnRH neurons in both brain regions. In animals that were replaced immediately, the synaptic input to GnRH neurons in the POA was also increased and was greater than that in intact animals. As for the other parameters, the synaptic input in the MBH of these two animals was different. In the animal with more glial apposition and a low incidence of nucleoli, there was little synaptic input, whereas the animal with less glial apposition and a higher incidence of nucleoli had more synaptic input. We do not know how to interpret these differences in the various parameters measured in the MBH of the two animals. This will require further experiments in the future. Random nonidentified neurons had between 0-28% of

0

1.1. 1, 1 .5

1 1.5 2 2.5 3 3.5 4 % Synaptic

7.5

their perikaryal membrane with synaptic modification. The density of synaptic input in these nonidentified neurons did not vary systematically across treatment groups. The density of innervation and the degree of glial ensheathment were not inversely related on a neuron by neuron basis. This was tested by performing regression analysis (data not shown).

Discussion This immunocytochemical ultrastructural study was undertaken to examine the effects of ovariectomy and gonadal steroid replacement on the GnRH neuron in the rhesus macaque. In intact cycling monkeys, GnRH neurons were often partially ensheathed with glial processes, and they received a modest synaptic input. They had both more glial apposition and less innervation than


STEROID EFFECT OF GnRH NEURON ULTRASTRUCTURE random nonidentified neurons in the same brain regions. After ovariectomy, GnRH neurons became more fully ensheathed and concomitantly less well innervated. Both of these effects were specific to GnRH neurons and were more pronounced in the preoptic area than in the medial basal hypothalamus. A 2-month ovarian steroid replacement protocol which restored physiological levels of estradiol and progesterone resulted in increases in synaptic density to GnRH neurons but only a slight reduction in their glial ensheathment. That is, some aspects of the milieu of these neurons resembled those of ovariectomized monkeys with no steroid treatment. It has been demonstrated in the rhesus monkey that the alterations in gonadotropin release resulting from ovariectomy (31) can be reversed by estrogen treatment, the effect of which is exerted at least in part at the level of the hypothalamus (32). The response of the GnRH system may include changes in release and/or of synthesis (and processing) of the hormone (33). Although the present data do not speak directly to this problem, some observations may be relevant to the discussion. First, there was a higher incidence of nucleoli in random photomicrographs of GnRH neurons (an estimate of relative nucleolar volume) of intact animals than in the castrates, and steroid replacement was accompanied by an incidence of nucleoli even higher than that of intacts. Increases in ribonuclear-protein constituents of the nucleolus have been reported to accompany genetic transcription (30). This suggests that higher levels of protein synthesis occur in GnRH neurons of animals exposed to gonadal steroids. This observation is consistent with data indicating increased GnRH messenger RNA in the presence of estrogen (34, 35), although there is conflicting evidence (36). Other studies in rats suggest that there are similar steroid effects in other populations of neurons. Neurons in the ventromedial nucleus were apparently stimulated by estrogen treatment (15, 13, 37), whereas those in the arcuate nucleus appeared to be stimulated by castration (38). Second, in ovariectomized animals more than 15% of the GnRH neurons had DAB reaction product confined to a few secretory granules, as compared to an average of 3% of the GnRH neurons in intact animals. This may be another indication that there is a lower rate of synthesis in this group. This latter observation, however, could also be interpreted as an indication of an increased rate of movement out of the cell body and/or increased release at the terminals, depleting cellular stores of GnRH or its prohormone. The POA and MBH contain many cells that bind and retain estradiol (39), but there is no evidence that GnRH neurons themselves contain estrogen receptors (8-11). There is however considerable evidence from studies in rat that sex steroids affect synaptic connectivity. Castration results in decreases in various indices of synaptic

1089

input and neuronal activity, whereas estradiol replacement reverses these (12-14,16,17, 40-42). Plasticity and synaptic remodeling have not been studied in the hypothalamus of primates, and such studies in the rat have largely been confined to the ventromedial and arcuate nuclei, areas which although important for the modulation of GnRH release do not, in the rat, contain GnRH neurons. We found that ovariectomy resulted in a significant decrease in the synaptic input to GnRH neurons in the monkey and that gonadal steroid replacement reversed this effect. We did not find these effects in random nonidentified neurons from the same brain regions. These results in GnRH neurons in monkey are consistent with the earlier general observations in rat but speak more directly to the issue of GnRH secretion, as we used immunocytochemically identified neurons. Many neurotransmitter systems known to affect GnRH release are themselves affected by estrogen, including galanin (43), neuropeptide Y (44), 7-aminobutyric acid (GABA) (45), the catecholamines (46, 47), and serotonin (48). Several of these populations have been found to be estradiol concentrating, including catecholaminergic neurons (49, 50) and neurons immunoreactive for ,8-endorphin (51, 52) and GABA (53). In addition, some of these (GABA, /3-endorphin, serotonin, and the catecholamines) have been shown to make direct synaptic contacts onto GnRH neurons (54-59). It remains to be seen whether ovariectomy differentially affects any of these classes of synaptic inputs to GnRH neurons. In the present study, ovariectomy was found to have an effect specifically on the glial ensheathment of GnRH neurons in both the POA area and the MBH. There was a significant increase in the percentage of the perikaryal membrane that was directly apposed by glia, and this was particularly marked in the POA. A role for astrocytic activity in remodeling has been suggested both for the arcuate nucleus and for the vasopressinergic and oxytocinergic magnocellular systems, at the level of the perikarya, as well as at the terminals in the neurohypophysis (20, 60). In these systems under conditions of excitation and neurohormone release, ensheathing glial processes retract, exposing neuronal membranes. Although glial cells have not been reported to contain steroid receptors (39), it has been shown in in vitro preparations that estradiol modifies both their shape and the distribution of glial fibrillary acidic protein (the major constituent of astrocytic filaments) (61, 62). Gila have been shown to serve many functions beyond the maintenance of neuropil structure (for reviews see Refs. 63, 64). They may act as ion sinks during neuronal activity, contain receptors for various neurotransmitters, sequester released neurotransmitters, release neurotransmitters, and influence the maintenance of postsynaptic surfaces (65).


1090


It is important to note that although steroid replacement of ovariectomized animals reduced glial ensheathment of GnRH neurons somewhat, it remained significantly greater than that in intact animals. This suggests that the absence of the ovaries results in effects upon GnRH neurons and their milieu which are more complex than the withdrawal of gonadal steroids. Some studies in a seasonal breeder indicate a similar phenomenon. In the anestrous sheep, GnRH neuron were found to be almost entirely surrounded by massive glial processes (66), and the synaptic input to these neurons was significantly less than that in animals from the breeding season (67). However, this seasonal difference was also seen in ovariectomized ewes bearing implants which maintained a constant level of estradiol (Lehman, M. N., personal communication), again suggesting that there are potent influences beyond changes in the steroid environment. The present studies suggest two caveats. The first relates to the use of the castrate, steroid-replaced animal in making predictions for the intact animal. We found that a steroid replacement regimen which mimics ovulatory cycles did not result in complete restoration of the anatomical milieu of GnRH neurons. This is an indication that caution must be exercised in the interpretation of results generated through the use of this presumably physiological model. It is possible, however, that a simulation more precisely replicating the changes observed during each phase of the menstrual cycle or of a longer duration than that used in our protocol may be required to reverse these effects of ovariectomy. The second caveat relates to cross-specific generalizations. The ensheathment of GnRH neurons with conspicuous glia is not characteristic of all species. There are occasional thin glial lamellae seen in apposition to these neurons in hamsters, rats, and guinea pigs (68, Witkin, J. W., and A.-J. Silverman, unpublished observations). However, interestingly, although ovariectomy results in effects on LH secretion similar to those seen in monkey, female rats that have been ovariectomized for periods up to 20 months have no conspicuous glial ensheathment of GnRH neurons (Witkin, J. W., unpublished observations). In light of the hypothesis that there is a differential role for the MBH in modulation of GnRH pulsatility in the monkey (69), it may be significant that GnRH neurons in the MBH were consistently more densely innervated than those in the POA. It is important to point out however, that a recent tracing study in rhesus revealed that GnRH neurons as far anteriorly as the POA send projections to the hypophysial portal vasculature (70). In summary, the present studies provide compelling evidence that the anatomical milieu of GnRH neurons is highly plastic, and that gonadal steroids are among the

Endo«1991 Vol 129-No 2

important factors in its definition. Although both the density of innervation and the degree of glial ensheathment of the GnRH neuron are affected by ovariectomy, steroid replacement is chiefly effective in restoring synaptic input. The significance of glial ensheathment of GnRH neurons and the role of the ovary in inhibiting it remain to be determined.

Acknowledgments The authors gratefully acknowledge the assistance of Dr. Y. J. Feng in the experiments, Kafui Demasio, Katharine Rosa, and Honor O'Sullivan for technical and photographic work, and the laboratory of Dr. W. Hembree for performing RIAs.

References 1. VanVugt DA, Diefenbach WD, Alston E, Ferin M 1985 Gonadotropin-releasing hormone pulses in third ventricular cerebrospinal fluid of ovariectomized rhesus monkeys: correlation with luteinizing hormone pulses. Endocrinology 117:1550-1558 2. Levine JE, Duffy MT 1988 Simultaneous measurement of luteinizing hormone (LH)-releasing hormone, LH, and follicle-stimulating hormone release in intact and short-term castrate rats. Endocrinology 122:2211-2221 3. Kalra SP 1986 Neuronal circuitry involved in the control of LHRH secretion: a model for preovulatory LH release. In: Ganong WF, Martini L (eds) Frontiers in Neuroendocrinology. Raven Press, New York, vol 9:31-75 4. Lincoln DW, Fraser HM, Lincoln GA, Martin GB, McNeilly AS 1985 Hypothalamic pulse generators. Recent Prog Horm Res 41:369-411 5. Ferin M, Carmel PW, Zimmerman EA, Warren M, Perez R, Vandewiele RL 1974 Location of intrahypothalamic estrogen-responsive sites influencing LH secretion in the female rhesus monkey. Endocrinology 95:1059-1068 6. Levine JE, Norman RL, Gliessman PM, Oyama TT, Bangsberg DR, Spies HG 1985 "In vivo" gonadotropin-releasing hormone release and serum luteinizing hormone measurements in ovariectomized, estrogen-treated rhesus monkeys. Endocrinology 117:711— 721 7. Petersen SL, Barraclough CA 1989 Suppression of spontaneous LH surges in estrogen-treated ovariectomized rats by microimplants of antiestrogen into the preoptic brain. Brain Res 484:279289 8. Shivers BD, Harlan RE, Morrell JI, Pfaff DW 1983 Absence of oestradiol concentration in cell nuclei of LHRH-immunoreactive neurones. Nature 304:345-347 9. Karsch F, Lehman MN 1988 Do gonadotropin releasing hormone or dopaminergic neurons in the sheep contain estradiol receptor? Soc Neurosci Abstr 14:1069 (Abstract) 10. Watson RE, Langub MC, Landis JW 1990 Further evidence that LHRH neurons are not directly estrogen responsive: LHRH and estrogen receptor immunoreactivity in the guinea pig brain. Soc Neurosci Abstr 16:1200 (Abstract) 11. Sullivan KA, Silverman AJ, Witkin JW, Ferin M 1990 Distribution of estrogen receptor (ER) containing and GnRH neurons in the rhesus macaque. Soc Neurosci Abstr 16:1201 (Abstract) 12. Arai Y, Matsumoto A 1978 Synapse formation of the hypothalamic arcuate nucleus during post-natal development in the female rat and its modification by neonatal estrogen treatment. Psychoneuroendocrinology 3:31-45 13. Carrer HF, Aoki A 1982 Ultrastructural changes in the hypothalamic ventromedial nucleus of ovariectomized rats after estrogen treatment. Brain Res 240:221-233 14. Garcia-Segura LM, Baetens D, Naftolin F 1986 Synaptic remodelling in arcuate nucleus after injection of estradiol valerate in


STEROID EFFECT OF GnRH NEURON ULTRASTRUCTURE adult female rats. Brain Res 366:131-136 15. Cohen RS, Chung SK, Pfaff DW 1984 Alteration by estrogen of the nucleoli in nerve cells of the rat hypothalamus. Cell Tissue Res 235:485-489 16. Frankfurt M, Gould E, Woolley CS, McEwen BS 1990 Gonadal steroids modify dendritic spine density in ventromedial hypothalamic neurons: a golgi study in the adult rat. Neuroendocrinology 51:530-535 17. Garcia-Segura LM, Olmos G, Tranque P, Aguilera P, Naftolin F 1987 Nuclear pores in rat hypothalamic arcuate neurons: sex differences and changes during the oestrous cycle. J Neurocytol 16:469-475 18. Theodosis DT, Poulain DA 1987 Oxytocin-secreting neurones: a physiological model for structural plasticity in the adult mammalian brain. Trends Neurosci 10:426-430 19. Hatton GI 1988 Cellular reorganization in neuroendocrine secretion. Curr Top Neuroendocrinol 9:1-27 20. Hatton GI, Perlmutter LS, Salm AK, Tweedle CD 1984 Dynamic neuronal-glial interactions in hypothalamus and pituitary: implications for control of hormone synthesis and release. Peptides 5:121-138 21. Brawer J, Schipper H, Robaire B 1983 Effects of long term androgen and estradiol exposure on the hypothalamus. Endocrinology 112:194-199 22. McQueen JK, Wright AK, Arbuthnott GW, Fink G 1990 Glial fibrillary acidic protein (GFAP)-immunoreactive astrocytes are increased in the hypothalamus of androgen-insensitive testicular feminized (Tfm) mice. Neurosci Lett 118:77-81 23. Tranque PA, Suarez I, Olmos G, Fernandez B, Garcia-Segura LM 1987 Estradiol-induced redistribution of glial fibrillary acidic protein immunoreactivity in the rat brain. Brain Res 406:348-351 24. Tweedle CD, Modney BK, Hatton GI 1988 Ultrastructural changes in the rat neurohypophysis following castration and testosterone replacement. Brain Res Bull 20:33-38 25. Witkin JW, Silverman A-J 1985 Synaptology of luteinizing hormone-releasing hormone neurons in rat preoptic area. Peptides 6:263-271 26. Silverman AJ, Witkin JW, Millar RP 1990 Light and electron microscopic immunocytochemical analysis of antibodies directed against GnRH and its precursor in hypothalamic neurons. J Histochem Cytochem 38:803-813 27. Liposits Z, Gores T, Gallyas F, Kosaras B, Setalo G 1982 Improvement of the electron microscopic detection of peroxidase activity by means of the silver intensification of the diaminobenzidine reaction in the rat nervous system. Neurosci Lett 31:7-11 28. Silverman AJ, Antunes JL, Ferin M, Zimmerman EA 1977 The distribution of luteinizing hormone-releasing hormone (LHRH) in the hypothalamus of the rhesus monkey. Light microscopic studies using immunoperoxidase technique. Endocrinology 101:134-142 29. Silverman AJ, Antunes JL, Abrams GM, Nilaver G, Thau R, Robinson JA, Ferin M, Krey LC 1982 The luteinizing hormonereleasing hormone pathways in rhesus (Macaca mulatto) and pigtailed (Macaca nemestrina) monkeys: new observations on thick, unembedded sections. J Comp Neurol 211:309-317 30. Fakan S, Puvion E 1980 The ultrastructural visualization of nucleolar and extranucleolar RNA synthesis and distribution. Int Rev Cytol 65:255-299 31. Dierschke DJ, Bhattacharya AN, Atkinson LE, Knobil E 1970 Circhoral oscillations of plasma LH levels in the ovariectomized rhesus monkey. Endocrinology 87:850-853 32. Ferin M, VanVugt D, Wardlaw S 1984 The hypothalamic control of the menstrual cycle and the role of endogenous opioid peptides. Recent Prog Horm Res 40:441-485 33. Lee W-S, Smith S, Hoffman GE 1990 Progesterone enhances the surge of luteinizing hormone by increasing the activation of luteinizing hormone-releasing hormone neurons. Endocrinology 127:2604-2606 34. Kelly MJ, Garret J, Bosch MA, Roselli CE, Douglass J, Adelman JP, Ronnekleiv OK 1989 Effects of ovariectomy on GnRH mRNA, proGnRH and GnRH levels in the preoptic hypothalamus of the female rat. Neuroendocrinology 49:88-97 35. Roberts JL, Dutlow CM, Jakubowski M, Blum M, Millar RP 1989

36.

37.

38. 39. 40. 41.

42. 43. 44.

45.

1091

Estradiol stimulates preoptic area-anterior hypothalamic proGnRH-GAP gene expression in ovariectomized rats. Mol Brain Res 6:127-134 Zoeller RT, Seeburg PH, Young III WS 1988 In situ hybridization histochemistry for messenger ribonucleic acid (mRNA) encoding gonadotropin-releasing hormone (GnRH): effect of estrogen on cellular levels of GnRH mRNA in female rat brain. Endocrinology 122:2570-2577 Jones KJ, Pfaff DW, McEwen BS 1985 Early estrogen-induced nuclear changes in rat hypothalamic ventromedial neurons: an ultrastructural and morphometric analysis. J Comp Neurol 239:255-266 Zambrano D, deRobertis E 1968 The effect of castration upon the ultrastructure of the rat hypothalamus. II. Arcuate nucleus and the outer zone of the median eminence. Z Zellforsch 87:409-421 Morrell JI, Krieger MS, Pfaff DW 1986 Quantitative autoradiographic analysis of estradiol retention by cells in the preoptic area, hypothalamus and amygdala. Exp Brain Res 62:343-354 Clough RW, Rodriguez-Sierra JF 1983 Synaptic changes in the hypothalamus of the prepuberal female rat administered estrogen. Am J Anat 167:205-214 Olmos G, Aguilera P, Tranque P, Naftolin F, Garcia-Segura LM 1987 Estrogen-induced remodelling in adult rat brain is accompanied by the reorganization of neuronal membranes. Brain Res 425:57-64 Olmos G, Naftolin F, Perez J, Tranque PA, Garcia-Segura LM 1989 Synaptic remodeling in the rat arcuate nucleus during the estrous cycle. Neuroscience 32:663-667 Gabriel SM, Koenig JI, Kaplan LM 1990 Galanin-like immunoreactivity is influenced by estrogen in peripubertal and adult rats. Neuroendocrinology 51:168-173 Crowley WR, Kalra SP 1987 Neuropeptide Y stimulates the release of luteinizing hormone-releasing hormone from medial basal hypothalamus in vitro: modulation by ovarian hormones. Neuroendocrinology 46:97-103 Lamberts R, Vijayan E, Graf M, Mansky T, Wuttke W 1983 Involvement of preoptic-anterior hypothalamic GABA neurons in the regulation of pituitary LH and prolactin release. Exp Brain Res 52:356-362

46. Barraclough CA, Wise PM 1982 The role of catecholamines in the regulation of pituitary luteinizing hormone and follicle-stimulating hormone secretion. Endocr Rev 3:91-119 47. Kalra SP, Kalra PS 1984 Opioid-adrenergic-steroid connection in regulation of luteinizing hormone secretion in the rat. Neuroendocrinology 38:418-426 48. Kerdelhue B, Bojda F, Lesieur P, Pasqualini C, El Abed A, Lenoir V, Douillet P, Chiueh MC, Palkovits M 1989 Median eminence dopamine and serotonin neuronal activity. Neuroendocrinology 49:176-180 49. Grant LD, Stumpf WE 1974 Hormone uptake sites in relation to CNS biogenic amine systems. In: Stumpf WE, Grant LD (eds) Anatomical Neuroendocrinology. Karger, Basel, pp 445-463 50. Sar M 1984 Estradiol is concentrated in tyrosine hydroxylasecontaining neurons of the hypothalamus. Science 223:938-940 51. Jirikowski GF, Merchenthaler I, Rieger GE, Stumpf WE 1986 Estradiol target sites immunoreactive for /?-endorphin in the arcuate nucleus of rat and mouse hypothalamus. Neurosci Lett 65:121-126 52. Morrell JI, McGinty I, Pfaff DW 1985 A subset of beta-endorphinor dynorphin-containing neurons in the medial basal hypothalamus accumulates estradiol. Neuroendocrinology 41:417-426 53. Leranth C, Sakamoto H, MacLusky NJ, Shanabrough M, Naftolin F 1985 Estrogen responsive cells in the arcuate nucleus of the rat contain glutamic acid decarboxylase (GAD): an electron microscopic immunocytochemical study. Brain Res 331:376-381 54. Leranth C, MacLusky NJ, Sakamoto H, Shanabrough M, Naftolin F 1985 Glutamic acid decarboxylase-containing axons synapse on LHRH neurons in the rat medial preoptic area. Neuroendocrinology 40:536-539 55. Chen W-P, Witkin JW, Silverman A-J 1989 Beta-endorphin and gonadotropin-releasing hormone synaptic input to gonadotropinreleasing hormone neurosecretory cells in the male rat. J Comp


1092


Neurol 286:85-95 56. Chen W-P, Witkin JW, Silverman A-J 1989 Gonadotropin releasing hormone (GnRH) neurons are directly innervated by catecholamine terminals. Synapse 3:288-290 57. Kiss J, Halasz B 1985 Demonstration of serotoninergic axons terminating on luteinizing hormone-releasing hormone neurons in the preoptic area of the rat using a combination of immunocytochemistry and high resolution autoradiography. Neuroscience 14:69-78 58. Watanabe T, Nakai Y 1987 Electron microscopic cytochemistry of catecholaminergic innervation of LHRH neurons in the medial preoptic area of the rat. Arch Histol Jpn 50:103-112 59. Leranth C, MacLusky NJ, Shanabrough M, Naftolin F 1988 Catecholamine innervation of luteinizing hormone-releasing hormone and glutamic acid decarboxylase immunopositive neurons in the rat medial preoptic area. Neuroendocrinology 48:591-602 60. Montagnese C, Poulain DA, Vincent JD, Theodosis DT 1988 Synaptic and neuronal-glial plasticity in the adult oxytocinergic system in response to physiological stimuli. Brain Res Bull 20:681692 61. Garcia-Segura LM, Suarez I, Segovia S, Tranque PA, Cales JM, Aguilera P, Olmos G, Guillamon A 1988 The distribution of glial fibrillary acidic protein in the adult rat brain is influenced by the neonatal levels of sex steroids. Brain Res 456:357-363 62. Garcia-Segura LM, Torres-Aleman I, Naftolin F 1989 Astrocytic

63. 64. 65. 66.

67. 68. 69. 70.

Endo • 1991 Voll29«No2

shape and glial fibrillary acidic protein immunoreactivity are modified by estradiol in primary rat hypothalamic cultures. Dev Brain Res 47:298-302 Hertz L 1979 Functional interactions between neurons and astrocytes. I. Turnover and metabolism of putative amino acid transmitters. Prog Neurobiol 13:277-323 Walz W 1989 Role of glial cells in the regulation of the brain ion microenvironment. Prog Neurobiol 33:309-333 Meshul CK, Seil FJ, Herndon RM 1987 Astrocytes play role in regulation of synaptic density. Brain Res 402:139-145 Lehman MN, Karsch FJ, Robinson JE, Silverman A-J 1988 Ultrastructure and synaptic organization of luteinizing hormone-releasing hormone (LHRH) neurons in the anestrous ewe. J Comp Neurol 273:447-458 Karsch F, Lehman MN, Silverman AJ 1987 Seasonal plasticity in synaptic input to luteinizing hormone-releasing hormone (LHRH) neurons in the sheep. Soc Neurosci Abstr 13:1527 (Abstract) Lehman MN, Silverman AJ 1988 Ultrastructure of luteinizing hormone-releasing (LHRH) neurons and their projections in the golden hamster. Brain Res Bull 20:211-222 Knobil E 1980 The neuroendocrine control of the menstrual cycle. Recent Prog Horm Res 36:53-88 Goldsmith PC, Thind KK, Song T, Kim EJ, Boggan JE 1990 Location of neuroendocrine gonadotropin-releasing hormone neurons in the monkey hypothalamus by retrograde tracing and immunocytochemistry. J Neuroendocrinol 2:158-168


The ultrastructure of Rhesus monkey prostate.

Subicular input from temporal cortex in the rhesus monkey.

Ultrastructure of pre-synaptic input to motor neurons in Onuf's nucleus: controls and motor neuron disease.

Effects of anabolic steroids on liver cell ultrastructure in sheep.

The effect of gonadal steroids on the behavioral and biochemical effects of hippocampal sympathetic ingrowth.

Dual mode of corticothalamic synaptic termination in the mediodorsal nucleus of the rhesus monkey.

Morphology and mosaics of VIP-like immunoreactive neurons in the retina of the rhesus monkey.

The role of gonadal hormones in the sexual behavior of the rhesus monkey and human: A literature survey.

The effects of denervation and stimulation upon synaptic ultrastructure.

Effects of hypoxic preconditioning on synaptic ultrastructure in mice.

The effects of naloxone on behavior maintained by cocaine and heroin injections in the rhesus monkey.

Golgi, histochemical, and immunocytochemical analyses of the neurons of auditory-related cortices of the rhesus monkey.

Effects of gonadal steroids on clearance of growth hormone at steady state in the rat.

The morphology and laminar distribution of cortico-pulvinar neurons in the rhesus monkey.

Effect of input multiplicity on the establishment of simian virus 40 persistent infections in rhesus monkey kidney cells.

Combined effects of androgen anabolic steroids and physical activity on the hypothalamic-pituitary-gonadal axis.

Diversity of glutamatergic synaptic strength in lateral prefrontal versus primary visual cortices in the rhesus monkey.

Neurons of the lateral entorhinal cortex of the rhesus monkey: a Golgi, histochemical, and immunocytochemical characterization.

Mapping Synaptic Input Fields of Neurons with Super-Resolution Imaging.

Effect of potassium on ventilation in the rhesus monkey.

Effects of Cannabis sativa on ultrastructure of the synapse in monkey brain.

Effects of behavioral performance on single-unit firing patterns in inferior colliculus of the rhesus monkey.

Effects of long-term salicylate administration on synaptic ultrastructure and metabolic activity in the rat CNS.

Synaptic input from the pontine reticular nuclei to absucens motoneurons and internuclear neurons in the cat.