Journal of Neuroscience Research 25:375-385

( 1990)

A Small Subpopulation of Progesterone Receptor-Containing Neurons in the Guinea Pig Arcuate Nucleus Projects to the Median Eminence P. Poulain, M. Warembourg, and A. Jolivet INSERM U 156. 59045 Lille. (P.P., M.W.); and Groupe de Recherches: hormones et reproduction, INSERM U 135. FacultC de Medecine Paris Sud, 94270 Le Kremlin-Bicktre, France ( A . J . )

In female guinea pigs, a combination of retrograde INTRODUCTION tracing and immunofluorescence for progesterone reRecent studies by immunocytochemistry (Waremceptors (PR) was applied to determine if PR-immu- bourg et al., 1986; Blaustein et al., 1988) have demonnoreactive (PR-IR) neurons in the arcuate nucleus strated the presence of progesterone receptor-immuno(AR) send their axons directly to the median emi- reactive (PR-IR) neurons in the guinea pig hypothalamus. nence (ME). Axonal projections to the ME were stud- confirming previous results obtained by radioautography ied by different techniques using fluorescent dyes. in the same species (Warembourg 1978a) and in other From 31 adult animals, ovariectomized and primed rodents (Sar and Stumpf, 1973; Warembourg, 1978b; by estradiol, small deposits of Lucifer Yellow (LY) Rogers Munn et al., 1983). PR-IR neurons are distributed were made on the cut surface of the ME, either by in different areas of the hypothalamus, a dense group direct application of LY crystals or by iontophoresis. being concentrated in the arcuate nucleus (AR) and adThese techniques were carried out on excised me- jacent lateral regions. It has been shown that all PR-IR diobasal hypothalamus maintained in vitro and al- neurons in the AR also contain estradiol receptors lowed visualization of AR perikarya projecting to the (Warembourg et al., 1989). It is well known that AR ME after dye diffusion in the severed axons. In an- neurons project to a great variety of brain territories (reother group of ten immature animals primed by es- views in Rethelyi, 1985; Chronwall, 1985). Some of the tradiol, Granular Blue (GB) was injected in the jug- AR neurons, classified as tuberoinfundibular neurons ular vein. Blood-borne GB was taken up in the ME by (Szenthagothai, 1964), send short projections to the exintact nerve endings and retrogradely transported to ternal zone of the median eminence (ME) and thus are the perikarya of origin. PR-IR neurons and directly involved in the control of the anterior pituitary. perikarya filled with LY or retrogradely labeled by With the aim of demonstrating the efferent projections of GB were intermingled with each other throughout the PR-IR neurons, the present study was devised to deterrostrocaudal extent of the AR. Double-labeled cells, mine whether or not PR-IR neurons in the AR project to displaying PR immunoreactivity and dye labeling, the ME, by using a Combination of axonal tracing with were observed consistently, but their number was immunocytochemical detection of progesterone receptors small. This result demonstrates that some AR neu(PR). rons sending axonal projections to the ME are target To circumvent the difficulty of reaching the ME in cells for progesterone. As the majority of PR-IR neu- vivo, we developed simple in vitro tracing techniques rons in the AR do not project to the ME, it is sug- that allow backfilling (Iles and Mulloney, 1971; Greggested that most PR-IR neurons present in this nu- ory, 1973) of AR neurons after deposition of Lucifer cleus form local circuit projections or project to Ycllow (LY) fluorescent dye onto transected axons of the distant areas of the central nervous system. ME. In another set of experiments, axonal projections of

Received May 12. 1989: revised September I I . 1989: accepted September 12. 1989.

Key words: immunocytochemistry, tuberoinfundibular system, axonal tracing 0 1990 Wiley-Liss, Inc.

Address reprint requests to P Poulain. INSERM U 156. Place tle Verdun. 59045 Lille Cedex. France.

376

Poulain et al.

AR neurons to the ME were assessed by intravenous injections of Granular Blue (GB) fluorescent dye. Tissue sections through the AR which contained LY-backfilled cell bodies or cell bodies retrogradely labeled by GB were reacted with PR monoclonal antibodies to detect the presence of PR-IR neurons pro-jccting directly to the ME.

MATERIALS AND METHODS For in vitro studies, 31 adult female guinea-pigs weighing 450-500 g were ovariectomized to reduce the level of endogenous progesterone. To increase the amount of PR, animals received daily a subcutaneous injection of 10 pg estradiol benzoate (Sigma) dissolved in 0.3 ml sesame oil for 5 days. They were killed 24 hr after the last injection, 10-14 days after ovariectomy. From 14 animals, the brain was quickly dissected after decapitation and placed ventral side up under a dissecting microscope. ME was cut rostrally to the level of its separation from the pituitary stalk with microscissors, and the hypophysis was discarded. The ME stump was surrounded with small pieces of filter paper and a crystal of LY (CH. Fluka), held at the tip of a glass rod, was placed exactly on the cut surface of the ME. In additional control experiments (5 animals), a crystal of LY was placed on the superficial part of the ME left intact. After 1-2 min, superfluous LY was rinsed away with cold incubation medium. A block containing the mediobasal hypothalamus then was excised and immersed in a 5 ml plastic vial containing 1 ml incubation medium. The medium consisted of oxygenated DulbecCO'S modified Eagle's medium containing 20 mM Hepes buffer (Flow Lab.) with 100 IU penicillin and 0.1 mg streptomycin per milliliter. After filling the vial with O,, the preparation was kept at 4" C for 6 hr. The preparation was removed and stored in fixative overnight. Fixative consisted of 2% picric acid and 4% paraformaldehyde in 0.1M phosphate buffer, pH 7.4. From 17 animals, an explant of the mediobasal hypothalamus with attached hypophysis was isolated just after dissection of the brain. It was placed in an interfacetype slice chamber, with the cut surface laid on a piece of lens paper. The explant was maintained at 34" C in an oxygenated atmosphere and perfused through its lower surface with oxygenated Yamamoto medium (Yamamoto, 1973) at a rate of 2 ml/min. Under the dissecting microscope, ME was cut as described above and the hypophysis discarded. LY was applied to the cut surface of the ME by iontophoresis. For that, negative currents of 500 nA were applied for 15 min through glass micropipettes of 30 pm tip diameter containing 5 % LY dissolved in 0.25 M LiC1. The micropipette was positioned with a micromanipulator in the medial or lateral aspects of the ventral surface of the ME. Two symmet-

rical ejections were performed on each explant. After a 45 min period of incubation in the slice chamber, the preparation was fixed as described above. For in vivo studies, ten female immature guinea pigs weighing 160-200 g were primed by estradiol benzoate as described above. Young animals were chosen in order to use small quantities of tracer. Five niilligrams of GB (Dr. Illing, Gross Umstadt, West Germany) were dissolved in 0.1 ml sterile water, sonicated for about I min, and immediately injected in the jugular vein under ether anaesthesia. Survival times after the injection varied from 70-120 hr. Animals were anaesthetized with Nembutal and their brain perfused through the aorta with 50 ml of heparinized 0.970 NaCl followed by 500 ml of fixative. After removal, the brain was blocked and postfixed in the fixative overnight at 4" C. After fixation, the preparations were immersed in 10% sucrose in 0.1 M phosphate buffer 24 hr at 4" C and frozen in isopentane chilled by liquid nitrogen. Fifteen micrometers frontal sections were cut in a cryostat at -20" C and mounted on gelatin-coated slides. The sections were washed in 0.01M phosphate buffer saline (PES), pH 7.4, for 30 min and processed for immunocytochemistry . Monoclonal mouse IgG against PR from rabbit uterus (Let 64) was used to localize PR-IR (Lorenzo et al., 1988). It was diluted (2.5-3.3 pg/ml) in 0.01M PBS containing 5% nonimmune sheep serum. Sections were treated with anti-PR antiserum for 24 hr at 4" C in humid atmosphere. PR immunoreactivity was revealed by incubating the sections in biotinylated sheep antimouse (Amersham) diluted to 1 :30 for 1 hr 30 at 4" C then in Texasred streptavidin (Amersham) diluted to 1:60 for 1 hr 30 at 4" C. The sections were coverslipped with glycerine: PBS (3: I , v/v) and examined with a Leitz Orthoplan fluorescent microscope equipped with Ploemopak filter blocks A (340-380 nm wavelengths) for identification of GB. H2 (390-490 nm) for identification of LY and N, (530-560 nm) for identification of Texas red. Identification of double-labeled cells was made by switching from one filter block to the other during observation.

RESULTS The distribution of PR-IR neurons from explants maintained in vitro and from in vivo experiments using blood-borne GB was in agreement with our previous results (Warembourg et al., 1986, 1989). The quality of immunocytochemical staining was not influenced by the in vitro procedure nor by the presence of the fluorochromes used as tracers. With the filter combinations used, there was no interference between the fluorescence of immunoreactive cells when illuminated through filter block Nz and the fluorescence of LY or GB when illu-

YR-Containing Neurons Projecting to ME

minated through filter blocks H, and A, respectively. Immunoreactive cells displayed a red labeling, which was only present in the cell nucleus. PR-IR neurons were found throughout the entire extent of the AR, but their number varied greatly depending on the rostrocaudal level, Rostrally, very few PR-IR neurons were seen between the third ventricle and the base of the brain. More caudally, at the level of the forniation of the ME, numerous PR-IR neurons were located in the dorsal part of the AR, extending dorsally along the third ventricle (Figs. l A , 2B). At the same level (Figs. IA, 2D), a large collection of PR-IR neurons were present basally and extended laterally, notably in the ventrolateral nucleus, according to the nomenclature of Bleier (1983). Posteriorly, the number of PR-IR neurons gradually decreased. The neurons appeared there closely grouped around the ventricular recess, especially along its lateral and ventral edges (Fig. IB). Following the application of crystalline LY to the cut end of the ME, backfilled neurons were found bilaterally throughout the rostrocaudal extent of the A R . LYbackfilled neurons revealed good cell morphology and, under block filter H,, displayed a brilliant yellow fluorescence of the cell bodies and proximal processes, with nuclear fluorescence prevailing (Figs. 2A. 3A. 4). LYstained neurons were found in the basomedial region of the most rostral portion of the AR, where they were densely packed together in a triangular cluster lying between the tip of the third ventricle and the base of the brain (Fig. 4). In the middle portion of the AR, at the level of the ME, LY-labeled cells were concentrated in the dorsal part of the nucleus, contiguous with numerous labeled cells occupying the periventricular area (Figs. IA, 2A). The labeling extended basolaterally within the cell-poor zone separating the AR and the ventromedial nucleus (Fig. 1A). A few labeled cells were found in the ventrolateral nucleus (Fig. IA). In the basomedial part of the AR, labeled cells were less numerous and more randomly distributed (Fig. IA). A small number were discovered in the internal, subependymal zone of the ME. Posteriorly, the majority of LY-stained neurons was located in the dorsal part of the AR, which merges with the dorsomedial nucleus (Fig. IB). Labeled cells were rarely found more caudally, in the prenianimillary area. It should be emphasized that, as a result of control experiments. no LY-stained neurons were found when the crystal was applied on the intact surface of the ME. As described above, the populations of PR-IR neurons and LY-stained neurons generally shared the same territories within the area of the AR. Groups of PR-IR neurons and LY-stained neurons were frequently found in close proximity to one another and overlapped in several regions (Fig. 1A-B). In these regions were observed double-labeled neurons, i.e, PR-IR neurons exhibiting

377

LY staining. These neurons were red fluorescent and yellow fluorescent when illuminated through filter blocks N, and H,, respectively (Fig. 2A-B, 3A-B). Double-labeled neurons were scattered within the AR without being organized in distinct clusters. Quantification of double-labeled neurons with regard to the populations of single-labeled cells was not systematically done in this study. However, it could be emphasized that the number of double-labeled cells was very low, ranging from 1-2 to 15-16 cells per 15 pm section. which represents a percentage smaller than 4% in the populations of single-labeled cells. As an example, in sections depicted in Figure I , there are 16 doublelabeled neurons among 408 LY-stained neurons and 4 17 PR-IR neurons for the sections of the middle portion of the AR (Fig. IA), and five double-labeled neurons among 535 LY-stained neurons and 315 PR-IR neurons for a more caudal section of the AR (Fig. IB). Populations of PR-IR neurons and LY-stained neurons were intermingled in the most rostral portion of the AR, but double-labeling in this region was extremely rare. The largest number of double-labeled cells was found in the middle portion of the AR (Figs. IA, 2A-B. 3A-B). The majority was observed dorsally and basomedially. in a periventricular position (Figs. IA, 2A-B). Occasionally, double-labeled cells were found in the basolateral part of the AR and in the ventrolateral nucleus. A few cells were also double-labeled in the internal zone of the ME. The proportion of double-labeled cells decreased posteriorly, to reach a number of 0-5 per section. These caudally located neurons were observed laterally to the ventricle recess, where a clear overlap between PR-IR and LY-stained cells was observed (Fig. IB). A similar distribution of LY-backfilled neurons was observed in the cases in which the dye was iontophoretically applied to the ME. LY-stained neurons were distributed in the same territories, but the number of neurons was much less numerous than after crystal application. Moreover, iontophoretic injections resulted in more restricted labeled areas in the AR which occurred at different loci in the various experiments. Double-labeled neurons were distributed in the same way as after crystal application. but their number was fewer, as territories of overlap were smaller (Fig. 3C-D). Following intravenous injections of GB. retrogradely labeled neuronal perikarya and proximal processes showed blue cytoplasm with distinct silvery granules under filter block A . Cell nuclei were only lightly fluorescent, and nucleoli appeared as bright spots (Fig. 3E). GB-labeled neurons were arranged in a triangular formation that corresponded to the classical anatomic landmarks of the AR. As a rule, GB labeling was less extended than after application of crystalline LY. Unlike

378

Poulain et al.

Fig. 1

PR-Containing Neurons Projecting to ME

LY-backfilled neurons, GB-labeled neurons were infrequently observed in the rostra1 portion of the AR. Some neurons were scattered in the ventral periventricular area. This sparse population was clearly delimitated by a zone devoid of labeled cells from the neighbouring paraventricular nucleus, which showed intense labeling. GB-labeled neurons were particularly numerous in the middle portion of the AR, where they appeared as a tightly clustered collection of neurons lying in the area around the tip of the third ventricle (Figs. 2C, 3E). Labeled neurons were randomly distributed in the caudal portion of the AR and totally absent at the level of the premainmillary nuclei. PR-IR neurons partly overlapped with GB-labeled neurons in the middle portion of the AR. In this area, neurons containing both GB granules in the cytoplasm and PR immunoreactivity in the nucleus were found when switching from filter block A to filter block N, (Figs. 2C-D, 3E-F). Their number was low in each section and rather similar to that obtained after in vitro experiments with crystalline LY.

DISCUSSION In the present study, we took advantage of the properties of diffusion of LY into transected axons for the examination of short projections from AR neurons toward the ME in the guinea-pig. Injection of axonal tracers in the ME in vivo presents technical difficulties. In the rat, some authors have used a dorsal approach to reach the ME (Lechan et al., 1980; Niimi et al., 1988), but some of the clearest results have been obtained with tracers directly injected or applied on the ventral surface of the ME, following a surgical transpharyngeal approach (Wiegand and Price, 1980; Lechan et al., 1982; Kawano and Daikoku, 1987; Silverman eta].. 1987). In the guinea pig, a transpharyngeal approach to the ME poses a severe surgical problem because of the conformation of the palate. The in vitro methods we have used offer a simple way to reach the ME and offer an alternative when surgical approach to the base of the brain proves difficult.

Fig. I . Drawings illustrating the distribution of LY-backfilled cells following application of a LY crystal on the cut surface of the ME (o),PR-IR cells (A),and double-labeled cells (Ajin 15 pm-thick frontal sections of the AR. A: A section through the middle portion of the AR; B: A more caudal section. Alllabeled cells are represented; one symbol corresponds to one labeled cell. AR, arcuate nucleus; DM. dorsomedial nucleus: V. third ventricle; VL, ventrolateral nucleus; VM, ventromedial nucleus.

379

In various brain preparations maintained in vitro, horseradish peroxidase has previously been used for backfilling axons (Mason and Gregory, 1984; Sretavan and Shatz, 1984; Bernard0 et al., 1986) and cell bodies (Ekstrom, 198.5; Bourrat and Sotelo, 1986). To avoid the stage of histochemical processing for the visualization of horseradish peroxidase, thus providing good preservation of PR immunoreactivity, we chose to fill cell bodies with LY. This dye has been proved suitable as horseradish peroxidase for backfilling of live neurons in invertebrates (Stewart, 1981) and has been used successfully to fill neurons through their cut processes in isolated vertebrate brain preparation (Simpson et al., 1986). Incubation procedure following application of crystalline LY on the cut surface of the ME was similar to that described by Ekstrom (198.5), except the incubation time was reduced from 18 hr to 6 hr. In our experiments, incubation for 6 hr appeared suitable for maximal neuronal labeling without loss of cell morphology. The use of explants maintained in the slice chamber should improve cell viability because of a better oxygenation. However, as the explant was placed at the gas-liquid interface, we reduced the time following iontophoretic application of LY to 4.5 min to avoid drying of the superficial part of the explant. Unlike the crystal method, the use of explants maintained in the slice chamber made it possible to perform focal iontophoretic applications. No attempts were made in our present study to detail the precise pointto-point relationships between labeled areas in the AR and sites of iontophoretic applications. It is likely that the differences observed in the distribution and number of LY-filled neurons after crystal or iontophoretic applications are related to the different methods of application rather to the different conditions of in vitro procedures. The in vitro methods used in this study present some disadvantages due, first, to the way of applying the tracer and, second, to misleading positive results obtained as a result of the staining of neurons by local uptake. The first drawback comes from the fact that LY, especially with the crystal method, was applied to the entire section of the ME. This necessarily resulted in the filling of axons projecting not only to the external zone of the ME, where tuberoinfundibular neurons terminate around capillaries of the pituitary portal system, but also of axons that terminate in the internal layer of the ME (Zaborszky and Makara, 1979) or follow their course through the internal layer of the ME toward the posterior lobe of the hypophysis (Page, 1988). A second disadvantage of the in vitro methods comes from the possibility that AR intact neurons might have taken up LY from local extracellular space. A labeling of neurons in brain slices after the addition of low concentrations of LY in the incubation medium has been previously reported (Zimmerman, 1986) and ascertained in our mate-

Fig. 2 .

PR-Containing Neurons Projecting to ME

rial (results not shown). During application of LY, unpredictable amounts of tracer may spread through extracellular spaces, either from the surface exposed to the tracer or from the cerebrospinal fluid, and may be taken up by neurons other than those intended. Yet LYlabeled neurons selected in this study always displayed a strong fluorescence, which in all likelihood indicated that they were filled from their cut processes and not stained after local uptake. In spite of the difference in species, there is a close correlation between our results obtained in vitro and those previously reported in the rat after application of tracers to ME by a transpharyngeal approach (Wiegand and Price, 1980; Lechan et al., 1982). As in these studies, the topography of LY-backfilled neurons reported here shows that AR neurons projecting to the ME are located rostrally and dorsally in the AR, merging with periventricular neurons. They are numerous basolaterally in the more caudal part of the AR, underlying the border of the ventromedial nuclei. They are also present posteriorly, at the level where AR merges with the dorsomedial nuclei. Basomedial portions of the AR, at the level of the ME, contain fewer fluorescent neurons. We have, however, observed LY-backfilled neurons in two areas not previously described to contribute to the tuberoinfundibular system after retrograde tracing in the rat. The first area is the subventricular portion of the rostral pole of the AR. where a dense cluster of labeled neurons was observcd. The second area is the internal zone of the ME, where fluorescent cell bodies, probably belonging to the AR (Zambrano, 1968; Rkthelyi, 1975). were detected. Transport of GB via the bloodstream has also been used in the present study to label neurons with axons entering the ME. It has been reported that vascular administration of horseradish peroxidase (Broadwell and Brightman, 1976) and other fluorescent tracers (Van der Krans and Hoogland, 1983) resulted in retrograde labeling of neurons projecting to areas devoid of blood-brain barrier (Balin et al., 1986). including the neural-hemal zone of the ME, which possesses fenestrated capillaries (Page, 1988). This method obviously does not lead to a

Fig. 2 . Fluorescent photomicrographs of frontal sections through the AR, showing cells labeled after axonal transport of fluorescent dyes (A,C). and the same sections showing PR-IR cells (B,D). in the periventricular area (A,B) and the middle portion of the AR (C,D). Asterisks indicate third ventricle. Cells were backfilled after application of a LY crystal on the cut surface of the ME (A) or retrogradely labeled after intravenous administration of GB (C). Comparison of A-C respectively with B-D reveals a few dye-labeled cells which are also PR-IR (arrows). Calibration bar = 20 pn.

381

selective labeling of the neurons projecting to the ME, as neurons projecting to the posterior pituitary and to other circumventricular organs are also labeled. Nevertheless, parvocellular neurons retrogradely labeled in the hypothalamus after intravenous injections of horseradish peroxidase (Broadwell and Brightman. 1976, 1983; Armstrong and Hatton, 1980; Poulain et al., 1984; Jennes and Stumpf, 1986; Youngstrom and Numez, 1987) and Fast Blue (H6kfelt et al., 1985; Meister et a]., 1988) were classified as neurons with direct projections to the ME. The major disadvantage of the method when it is intended for labeling the AR is that blood-borne horseradish peroxidase (Broadwell and Brightman, 1976; Armstrong and Hatton, 1980; Van den Pol and Cassidy, 1982; Leonhardt and Eberhardt, 1982; Balin et al., 1986) and fluorescein (Martinez and Koda, 1988) notably invaded the AR up to its lateral edges. Thus, diffusion of tracers within the AR itself may result in labeling of neurons directly by local uptake. However, a diffuse extracellular staining with thc concomitant occurrence of diffusely labeled perikarya following spread of tracers into the AR is only detectable with survival times considerably lower than those used in the present study. According to Van der Krans and Hoogland (1983), we used long survival times, which are needed to obtain the most intense labeling with GB. In these conditions, GBlabeled neurons exhibited silver-blue granules in the cytoplasm, which is a typical feature of labeling due to retrograde axonal transport (Bentivoglio et al., 1979). When comparing distribution of LY-backfilled neurons obtained with applications of LY to the cut end of the ME with that of neurons that have taken up bloodborne GB, we observed a slight difference in the extent of labeling. With the former method, additional LYfilled neurons were found in the rostral and the caudal portions of the AR. Compared with LY-filled neurons, GB-labeled neurons were less numerous in the periventricular area, but they appeared more concentrated basomedially. in the middle portion of the nucleus. This difference is not surprising when the different methods of tracer application are considered. As discussed above, the in vitro methods are particularly suited for characterizing thc total population of neurons whose axons are sectioned within the different layers of the ME, whereas the in vivo method is expected to lead to the labeling of intact neurons terminating around capillaries of the external zone of the ME. On the other hand, the filling of axons in the case of LY applications is restricted to the area where the ME was sectioned, and then intact neurons whose axons terminate rostrally to the level of section probably remain unlabeled. These explanations may account for the differences observed in the extent of labeled neurons. In spite of the inherent problems in the interpreta-

382

Poulain et al.

Fig. 3. Fluorescent photomicrographs of frontal sections through the AR demonstrating localization of dye-labeled cells (A,C,E) and PR-IR cells (B,D,F). Third ventricle is to the right (asterisks), dorsal is at top. Green fluorescent cells were backfilled after application of a LY crystal (A) or iontophoresic of LY (C) on the cut surface of the ME. Blue fluorescent cells

(E) were labeled after intravenous injection of GB. A few cells exhibit both dye-labeling of their perikaryon and PR irnrnunoreactivity (red fluorescence from Texas-red) in their nucleus (arrows). Calibration bar = 20 p n for A-D, 9 p m for E and F.

PR-Containing Neurons Projecting to ME

Fig. 4. Fluorescent photomicrograph of a frontal section through the most rostra1 portion of the AR. Fluorescent cells were backfilled by a LY crystal application on the cut end of the ME. Note the cluster of labeled cells between the third ventricle (asterisk) and the base of the brain. Calibration bar = 100 p m .

tion of each method, experiments incorporating the three approaches have provided consistent information concerning the existence in the AR of a subset of PR-IR neurons that project to the ME. The functional identity of these neurons remains hypothetical. Among the cell groups Concentrated in the rat AR, the majority of the dopamine-containing neurons (A, cell group, Dahlstroni and Fuxe, 1964) project to the ME and represent the major efferent path from the AR to the external zone of the ME (Kawano and Daikoku, 1987). Perikarya of tuberoinfundibular dopamine-containing neurons are located mainly in the dorsomedial and basolateral parts of the AR and are also found in the periventricular area adjacent to the AR and around the lateral border of the ventromedial nucleus (Kawano and Daikoku. 1987). In the guinea pig, dopamine-containing neurons are also

383

concentrated in the AR (Barry and Leonardelli, 1967). It is quite conceivable that a great number of the neurons identified in the present study as projecting to the ME correspond to dopamine-containing neurons. especially those found in the dorsomedial and basolateral parts of the AR and in the periventricular area. Recently, it has been demonstrated in the rat that tyrosine-hydroxylase immunoreactive neurons accumulated tritiated estradiol (Sar, 1984) and progestin (Sar, 1988). It can be hypothesized that some of the double-labeled cells observed in the present study correspond to tuberoinfundibular dopamine-containing neurons possessing steroid receptors. PR-IR neurons located in the AR and projecting to the ME may also belong to different classes of neuropeptide-containing neurons. It has been demonstrated in the rat that substance P-containing neurons originating in the AR and in the ventral portion of the ventromedial nuclei and terminating in the external zone of the ME (Tsuruo et al., 1983) accumulate estradiol (Akesson and Micevych, 1988). It is conceivable that some substance P-containing neurons may be included in the population of double-labeled cells identified in the present study. The presence of estradiol target sites in P-endorphincontaining neurons (Morrell et al., 1985; Jirikowski et al., 1986) and in dynorphin-containing neurons (Morrell et a]., 1985) has been described in the AR and adjacent areas in different rodent species. As P-endorphin-containing neurons preferentially terminate in the internal zone of the ME. at least in the rat (Hisano et al., 1982; Mezey et a]., 1985), and dynorphin-containing neurons have only few projections to the ME (Everitt et al., 1986), they do not belong to the tuberoinfundibular system. Nevertheless, in our in vitro studies, as discussed above, axons passing through or terminating in the internal zone of the ME may have taken up the tracer. Thus, it is likely that P-endorphin-containing neurons in the AR are labeled after application of LY and that certain double-labeled cells may be P-endorphin-containing neurons endowed with PR. In the present investigation, results of in vitro studies, as well as results obtained after intravenous injections of GB, clearly indicated that only a small population of PR-IR neurons in the AR gave rise to projections to the ME. It seems evident from these results that the majority of PR-IR neurons present in the AR establish local interconnections or project to parts of the brain other than the ME. It has already been demonstrated, by using retrograde axonal transport of fluorescent dyes and steroid radioautography, that estradiol-concentrating cells in the rat AR projected to the dorsal midbrain (Morre11 and Pfaff, 1982; Morrell et al., 1984) and to the medial preoptic nucleus (Akesson et al., 1988). In the future, tract-tracing techniques combined with immunocytochemistry should shed light on the other territories of

384

Poulain et al.

the brain that receive afferent projections from neurons that possess receptors for steroid hormones.

REFERENCES Akesson TR. Micevych PE (1988): Emxipen concentration by wbstance P-imniunoreactive neurons in the medial basal hypothalamus of the female rat. J Neurosci Res l9:412-419. Akesson TR. Siinerly RB. Micevych PE ( 1988): Estrogen-concentrating hypothalamic and limbic neurons project to the medial preoptic nucleus. Brain Res 451:381-38S. Armstrong WE. Hatton GI (1980): The localization of projection neurons in the rat hypothalamic paraventricular nucleus following vascular and neurohypophysial injections of HRP. Brain Res Bull 51473-477. Balin BJ. Broadwell RD, Salcnian M , El-Kalliny M (19x6): Avenues for entry of peripherally administrated protein to the central nervous system in mouse. rat, and squirrel monkey. J Comp Neurol 2S 1260-280. Barry J , LConardelli J (1967): Etude de la topographie des fibres et des neurones rnonoaminergiques au niveau de I’hypothalamus chez le cobaye normal ou stirkotaxe. CR Acad Sci (Paris) 265: 557-560. Bentivoglio M, Kuypers HGJM, Catsman-Berrevoets CE, Dann 0 (l97Y): Fluorescent retrograde neuronal labeling in rat by means of substances binding specifically to adenine-thymine rich DNA. Neurosci Lett 12:235-240. Bernardo KL. Ma PM. Woolsey TA ( 1986):In vitro labeling of axonal projections in the mammalian central nervous system. J Neurosci Methods 16:8Y-101. Blaustein JD, King JC. Toft DO. Turcotte J (198X): Immunocytochemical localization of estrogen-induced progestin receptors in guinea-pig brain. Brain Res 474:l-IS. Bleier R ( 19x3): The hypothalamus of the guinea-pig: A cytoarchitectonic atlas. The University of Wisconsin Press, 134 pp. Bourrat F, Sotelo C ( 1986): Neuronal migration and dendritic niaturation o f the medial cerehellar nucleus in rat embryos: An HRP study using cerebellar slabs. Brain Res 378:6Y -85. Broadwell RD. Brightman MW ( 1976): Entry of peroxidase into neuron\ of the central and peripheral nervous systeins from extracerebral and cerebral blood. J Conip Neurol 166:257-284. Broadwell RD. Brightman MW (1983): Horseradish peroxidase: a tool to study the neuroendocrine cell and other peptide secreting cells. Methods Enzymol 103:187-218. Chronwall BM (1985): Anatomy and physiology of the neuroendocrine arcuate nucleus. Peptides 6 [suppl 21: 1-1 I . Dahlstriim A . Fuxe K ( 1964): Evidence for the existence of monoamine containing neurons in the central nervous system. 1. Demonstration of nionoamines in the cell bodies of brain stem neurons. Acta Physiol Scand 62 [suppl 232l:I-SS. Ekstriim P (1985): Anteropade and retrograde filling of central neuronal systems with horseradish peroxidase under in vitro conditions. J Neurosci Methods 15:21-35. Everitt BJ. Meister B. Hokfelt T , Melander T. Terenius L, Riikaeus A, Theodorsson-Norheini E, Dockray G , Edwardson J. Cuello C , Elde R. Goldstein M. Hemmings H . Ouiniet C. Walaas 1, Greengard P, Vale W, Weber E. Wu J Y , Chang KJ ( 1986): The hypothalamic arcuate nucleus-median eminence complex: Inimunohistocheniistry of transmitters. peptides and DARPP-32 with special reference to coexistence in dopamine neurons. Brain Res 11:Y7-155. Gregory GE (1973): Simple fluorescence staining of insect central nerve fibres with Procion Yellow. Stain Techno1 48:85-87.

Hisano S, Kawano H, Nishiyama T. Daikoku S ( 1982): Immunoreactive ACTHiP-endorphin neurons in the tubero-infundibular hypothalamus of rats. Cell Tissue Rcs 224303-3 14. Hokfelt T , Dalsgaard CJ. Meister B, Everitt B (1985): Tracing of afferents to the median eminence using combined immunohistochemistry and intravenous administration of fluorescent compounds. Neurosci Lett [Suppl] 22:S565. lles JF. Mulloney B (1971): Procion Yellow staining of cockroach motor neurones without the use of microelectrodes. Brain Res 30:397- 400. Jennes la, Stumpf WE (1986): Gonadotropin-releasing hormone immunoreactive neurons with access to fenestrated capillaries in mouse brain. Neuroscience 18:403-416. Jirikowski GF. Merchenthaler I, Rieger GE. Stumpf WE (1986): Estradiol target sites immunoreactive for P-endorphin in the arcuate nucleus of rat and mouse hypothalamus. Neurosci Lett 65: I 2 1-1 26. Kawano H, Daikoku S (1987): Functional topography of the rat hypothalamic dopamine neuron systems: Retrograde tracing and iiiiniunohistocheniical study. J Conip Neurol 265:242-253. Lechan RM, Nestler JL. Jacobson S ( 1982): The tuberoinfundibular system of the rat as demonstrated hy immunohistochernical localization of retrogradely transported wheat g e m agglutinin (WGA) from the median eminence. Brain Res 245:1-15. Lechan RM. Nestler JL. Jacobson S. Reichlin S (1980): The hypothalamic *‘tuberoinfundibular” system o f the rat as demonstrated by horseradish peroxidase (HRP) microiontophoresis. Brain Res 195:13-27. Leonhardt H, Eberhardt HG (1982): Dye transport from the median eminence to the hypothalamic wall. In Knigge KM, Scott DE. Weindl A (eds): “Brain Endocrine Interaction-Median Eminence: Structure and Function.” New York: Karger, pp 335341. Lorenzo F, Jolivct A, Loosfelt H, Thu Vu Hai M , Brailly S. PerrotApplanat M. Milgroni E (1988): A rapid method of epitope mapping. Application to the study of immunogenic domains and to the characterization of various forms o f rabbit progesterone receptor. Eur J Biochem 176:53-60. Martinez JL Jr. Koda L (1988): Penetration of fluorescein into the brain: A sex difference. Brain Res 450:81-85. Mason CA, Gregory E (1984): Postnatal maturation of cerebellar mossy and climbing fibers: Transient expression of dual features on single axons. J Neurosci 4:1715-1735. Meister B, HBkfelt T. Geffard M. Oertel W (1988): Glutamic acid decarboxylase- and y-aniinobutync acid-like immunoreactivities in corticotropin-releasing factor-containing parvocellular neurons of the hypothalamic paraventricular nucleus. Neuroendocrinology 48:s 16-526. Mezey F.. Kiss JZ, Mueller GP, Eskay R. O’Donohue TL, Palkovits M ( 1985): Distribution of the pro-opiomelanocortin derived peptides. adrenocorticotrope hormone, u-melanocyte-stimulating hormone and P-endorphin (ACTH, a-MSH, P-end) in the rat hypothalamus. Brain Res 328:341-347. Morrell J I , McGinty JF. Pfaff DW (1985): A subset of P-endorphinor dynorphin-containing neurons in the medial basal hypothalamus accumulates estradiol. Neuroendocrinology 41 :417-426. Morrell J I . Pfaff DW ( 1982): Characterization of estrogen-concentrating hypothalamic neurons by their axonal projections. Science 2 17: 1273-1 276. Morrell JI. Schwanzel-Fukuda M , Fahrbach SE, Pfaff DW (19x4): Axonal projections and peptide content of steroid hormone concentrating neurons. Peptides 5 [Suppl 1]:227-239. Niimi M, Takahara J , Hashimoto K. Kawanishi K (1988): Immunohistochemical identification of corticotropin releasing factor-

PR-Containing Neurons Projecting to ME containing neurons projecting to the stalk-median eminence of the n t . Peptides 9:.589-593. Page RB (1988): The anatomy of the hypothalamo-hypophyseal c o n plex. In Knobil E. Neil1 J.D (eds): “The Physiology of Reproduction.” New Yorh: Raven Preas. pp 1161-1233. Poulain P. Martin-Bouyer L. Beauvillain JC. Tramu G (1984): Study of the efferent connections of the enkephalinergic magnocellular dorsal nucleus in the guinea pig hypothalamus using lesions. retrograde tracing and immunohistochemistry: Evidence for a projection to the lateral septum. Neuroscience I I :33 1-343. Rkthelyi M (1975): Neurons in the subependymal layer of the rat median eminence. Neuroendocrinology I7:330-339. Rkthelyi M ( 1985): Dendritic arborization and axon trajectory of neurons in the hypothalamic arcuate nucleus of the rat-updated. Neuroscience 16323-33 1 Rogers Munn A I l l . Sar M, Sturnpf WE (1983): Topographic distribution of progestin target cells in hamster brain and pituitary after injection of (‘H) R.5020. Brain Res 274:l-10. Sar M (1984): Estradiol is concentrated in tyrosine hydroxylase-containing neurons of the hypothalamus. Science 223:938-940. Sar M (1988): Distribution of progestin-concentrating cells in rat brain: colocalization of (‘11) ORG.20.58, a synthetic progestin, and antibodies to tyrosine hydroxylase in hypothalamus by combined autoradiography and iiiiiiiuiiocytochemistry. Endocrinology 123:1110-1118. Sar M , Stumpf WE ( 1973): Neurons of the hypothalamus concentrate (’H) progesterone or its inetabolites. Science 183:1266-1268. Silverman AJ. Jhamandas J, Renaud LP ( 1987): Localization of luteinizing hormone-releasing hormone (LHRH) neurons that project to the median eminence. J Neurosci 7 2 3 12-2319. Simpson HB. Tohias ML. Kelley IIB ( 1986): Origin and identification of fibers in the cranial nerve IX-X complex of X e t i o p s lortis: Lucifer Yellow backfills in vitro. J Cornp Neurol 244:430444. Sretavan D , Shatz CJ (1984): Prenatal development of individual retinogeniculate axonn during the period of segregation. Nature 308:84S -848. Stewart WW (19811: Lucifer dyes-highly fluorescent dyes for biological tracing. Nature 292: 17-2 I .

385

Szentipothai J ( 1 9 6 4 ~The : parvicellular neurosecretory 5ystem. Prog Brain Res 5:13.5-146. Tsuruo Y. Kawano H, Nishiyama T, Hisano S. Daikoku S (1983): Substance P-like iiiiiiiunoreactive neurons i n the tuberoinfundibular area of rat hypothalamus. Light and electron niicroscopy. Brain Res 289:1-9. Van der Krans A, Hoogland PV ( 1983): Labeling of neurons following intravenous injections o f fluorescent tracers in mice. J Neurosci Methods 9:9.5-103. Van den Pol AN, Cassidy JR ( 1982): The hypothalamic arcuate nucleus of rat: A quantitative Golgi analysis. J Comp Neurol 204:65 -98. Warembourg M (1978a): Radioautographic study of the brain and pituitary after (‘H) progesterone injection into estrogen-primed ovariectomized guinea-pigs. Neurosci Lett 7: 1-5. Warembourg M (l978b): Radioautographic study of the rat brain, uterus and vagina after (’H) R.5020 injection. Mol Cell Endrocrinol I2:67-79. Warembourg M, Jolivet A, Milgrom E ( 1989): Immunohistochernical evidence of the presence of estrogen and progesterone receptors in the same neurons of the guinea pig hypothalamus and preoptic area. Brain Res 48O:l-IS. Warembourg M. Logeat F. Milgroni E ( 1986): Imniunocytocheinical localization of progesterone receptor in the guinea pig central nervous system. Brain Res 384: 12 I- 131 . Wiegand SJ, Price JL (1980): Cells of origin of the afferent fibers to the median eminence of the rat. J Comp Neurol 192:1-19. Yamamoto C ( 1973): Propagation of aAer-discharges elicited in thin brain sections i n artificial media. Exp Neurol 40: 183-188. Youngstrom TG, Nuinez AA (1987): Neurons in the suprachiasniatic area are labelled after intravenous injections of horseradish peroxidase. Exp Brain Res 67:127-130. Ziborszky D. Makara GB ( 1979): Intrahypothalaniic connections: An electron microscopic study in the rat. Exp Brain Res 34:201215. Zarnbrano D (1968): On the presence of neurons with granulated vesicles in the median eminence of rat and dog. Neuroendocrinology 3:141-1.5.5. Zimmerman RP (1986): Specific neuronal staining by in vitro uptake of Lucifer Yellow. Brain Res 383:287-298.