Differential organization of synapses immunoreactive to phenylethanolamine-N-methyltransferase or neuropeptide Y in the parvicellular compartments of the hypothalamic paraventricular nucleus of the rat

Journal of Chemical Neuroanatomy, Vol. 6:55-67 (1993)

Differential Organization of Synapses Immunoreactive to Phenylethanolamine-N-Methyltransferase or Neuropeptide Y in the ParviceHular Compartments of the Hypothalamic Paraventricular Nucleus of the Rat G. Alonso INSERM U 336, Drveloppement, Plasticit6et Vieiilissementdu Systrme Nerveux, Universit6 de Montpellier II, Place E. Bataillon, F-34095 Montpellier Cedex 5, France ABSTRACT The parvicellular compartments of the paraventricular nucleus of the hypothalamus (pPVN) contains particularly high concentrations of neuropeptide (NPY)-containing fibres of two main cellular origins including (i) neurons of the medulla oblongata, most of which co-store phenylethanolamineN-methyltransferase (PNMT), the enzyme characterizing adrenergic neurons, and (ii) noncatecholaminergic neurons of the mediobasal hypothalamus. The aim of the present study is to compare the fine organization of the two types of axons terminating in the pPVN. Immunocytochemistryat light and electron miscroscope levels was used to study both the density and the ultrastructural organization of NPY- and PNMT-immunoreactive fibres in the pPVN of animals bearing surgical lesions disrupting axonal pathways from the hindbrain or from the sublying mediobasal hypothalamus. The brainstem knife-cut induced a strong decrease (65%) in the numerical density of PNMT fibres innervating the pPVN, but was without significant effects on the density of NPY fibres. On the other hand, the hypothalamic knife-cut induced an 80% decrease in the density of NPY fibres within the PVN without affecting the number of PNMT fibres. The electron microscope study showed that in the control pPVN contralateral to the lesions, the majority (64%) of PNMT synapses were asymmetric axo-dendritic synapses, whereas the majority (67%) of NPY synapses form symmetric contacts with both dendrites and perikarya of the hypothalamic nucleus. By contrast, after a hypothalamic knife-cut, the majority (66%) of NPY synapses identified in the pPVN exhibited features of asymmetric synapses. These data indicate that the large majority of NPY-immunoreactive fibres detected within the pPVN arise from non-catecholaminergic neurons located in the mediobasal hypothalamus and mainly form symmetric synapses on neurons of the pPVN, whereas only a minority of them arise from hindbrain regions, and like PNMT fibres innervating this nucleus preferentially form asymmetric axo-dendritic synapses. KEYWORDS: Brain Hypothalamus Immunocytochemistry INTRODUCTION It is generally accepted that the hypothalamic paraventricular nucleus (PVN) plays a major role in the central control of various endocrine and autonomic functions (Sawchenko and Swanson, 1981,1982a; Sofroniew and Schrell, 1981; Luiten et al., 1985). Among the large variety of neuronal systems that innervate this nucleus, special attention has been paid during recent years to neuropeptide Y (NPY)-containing neurons. Indeed, the PVN has been shown to receive an innervation by NPY-nerve terminals that probably represents the most prominent input to the PVN yet identified (Chronwall et al., 1985; Gray and Morley, 1986). Furthermore, numerous physiological studies based on the injection of NPY within or in close vicinity to the PVN 0891-0618/93/020055-13 $11.50 © 1993 by John Wiley and Sons Ltd

provided strong indications that this peptide is involved in the regulation of a number of endocrine and autonomic functions (Moltz and McDonald, 1985; H~trfstrand, 1986; Wahlestedt et al., 1987; Harland et al., 1988; Humphreys et al., 1988; Abe et al., 1989), as well as of mechanisms of eating behaviour (Stanley and Leibovitz, 1985; Leibovitz et al., 1988). Although neurons synthesizing NPY have been detected in various brain regions (Chronwall et aL, 1985), a series of neuroanatomical studies have demonstrated that NPY axons afferent to the PVN mainly arise from two NPY-neuron groups, including (i) neurons of the medulla oblongata (Sawchenko et al., 1985) and (ii) neurons of the arcuate nucleus (Bai et aL, 1985). However, the proportion within the PVN of NPY fibres arising from medullary neurons as opposed to neurons of

56

G. Alonso

A1

rTlm

B1

'-,I ; '.,I.,.!

Fig. 1. Schematic respresentation and low magnification micrographs of the two types of knife-cuts. A~ and B~ respectively represent frontal planes at the levels of a unilateral brainstem knife-cut using a rectangular glass knife, and a unilateral hypothalamic knife-cut using an L-shaped Halasz-type knife. Low magnification micrographs show the extent of the corresponding lesions observed in sagittal (A2) and frontal (B2) sections. AR: arcuate nucleus; CH: optic tract; DR: dorsal raphe nucleus; LC: locus coeruleus; mPVN: magnocellular paraventricular nucleus; pPVN: parvocellular paraventricular nucleus; PY: pyramidal tract; RP: raphe pontis nucleus; $5: senory root of the trigeminal nerve. Bar = 500 p.m.

the mediobasal hypothalamus is still a matter of discussion. Indeed, the PVN is known to receive a rich catecholaminergic (CA) innervation arising from the medullary CA cell groups (Sawchenko and Swanson, 1982b, 1983). This hypothalamic nucleus contains one of the densest concentrations within the brain of axon terminals containing phenylethanolamine-N-methyltransferase (PNMT: the enzyme characterizing the adrenergic neurons; H6kfelt et al., 1974; Zoli et al., 1988; Cunningham et al., 1990). Moreover, it has been shown that virtually all PNMT-containing neurons of the medulla oblongata co-store NPY (H6kfelt et al., 1983a,b; Everitt et al., 1984) and that the vast majority of these medullary PNMT-containing neurons that project to the PVN actually contain NPY (Sawchenko et al., 1985). Surprisingly, however, lesions of the axonal pathways connecting the medullary CA cell groups to the hypothalamus have

been reported to induce non-significant modifications of the NPY innervation of the PVN (Bai et al., 1985; Gustafson and Moore, 1987). The aim of the present study was thus to compare thoroughly the organization of NPY- and PNMT-containing axon terminals within the PVN of animals treated with knife-cuts that respectively disrupt afferent axons originating from hindbrain regions or from sublying hypothalamic regions.

MATERIALS AND METHODS

Animals Male adult (200-250 g) Sprague-Dawley rats were used. They were kept in light (12 h light, 12 h dark) and temperature (24+ I°C) controlled rooms and had free access to standard dry food and tap water.

PNMT- and NPY synapses in the paraventricular nucleus

57

Preparation of tissues

I00 "~ 80 t.

60 40 20 0

m l

Brainstem knife-cut

Hypothalamic knife-cut

11 PNMT.IR [] NPY-IR Fig. 2. Image analysis quantification of PNMT-IR and NPY-IR fibres in the pPVN of animals treated with brainstem or hypothalamic knife-cuts. Data represent the surface of immunostained profiles measured in the pPVN homolateral to the lesion expressed as % of control (i.e. of the surface of immunostained profiles measured on the same section, in the pPVN contralateral to the lesion). Means+SEM of 10 to 15 analysed sections are given for each type of axonal lesioning. Statistical analysis according to Mann-Whitney U test. **= significantly different from control at P < 0.01.

Axonal lesioning After deep pentobarbital anaesthesia (60 mg/kg), the animals were fixed in a sterotaxic device. The axonal lesions were placed unilaterally and were positioned according to the stereotaxic atlas of Paxinos and Watson (1982). Brainstern knife-cut (n = 10)

Glass knives, 2.5 mm wide and at least 15 mm long, were made from commercial histological cover glasses as described by Palkovits et al. (1982). Transection of the mesencephalic CA bundle was performed by moving the knife downwards to the base of the skull, in the vertical plane in a position located 0.2 mm before the interaural plane and 0.3 mm lateral to the midline (Fig. 1A).

Hypothalamic knife-cut (n = 10) This surgical lesioning was performed according to the method originally described by Haldsz and Pupp (1965), by means of an L-shaped knife (radius 1.8 mm) with an obtuse angle of 110° made from a 25 gauge 3 1/2 spinal needle (Fig. 1B). The tip of the knife was oriented backwards and placed 7 mm anterior to the interaural line. It was then lowered through the midline to 10 mm below the surface of the skull and turned 180° to one side. In a series of preliminary experiments, a maximum decrease in the catecholaminergic innervation of the PVN was observed as soon as 7 days after a brainstem lesion. In the present study, for each type of knife-cut, animals were thus allowed to survive for 7-10 days before being treated for immunocytochemistry.

The animals were anaesthetized with sodium pentobarbital (60 mg/kg) and perfused through the ascending aorta with phosphate-buffered saline (PBS) followed by 300 ml of fixative composed of 4% paraformaldehyde, 0.5% glutaraldehyde and 0.2% picric acid in 0.1 M-phosphate buffer at pH 7.4. The hypothalamic region was dissected and fixed by immersion in the same fixative without glutaraldehyde for 2-4 days. Brain pieces including the hypothalamus or the brainstem were then cut frontally and sagitally respectively, into 40-50 lam thick sections with a vibratome. The sections were carefully rinsed with PBS before subsequent treatment.

Histofluorescence method The 'Faglu' method (Furness et al., 1972) was used to evaluate the effectiveness of the lesion of the ascending CA bundle. Some vibratome sections cut from the paraformaldehyde-glutaraldehyde-fixed brains were put on histological glass slides, mounted in PBS under a cover slide, and rapidly observed in a Zeiss epifluorescence microscope under violet illumination. They were then mingled with other vibratome sections and subsequently treated for immunocytochemical labelling of NPY or PNMT.

lmmunocytochemical procedures Vibratome sections were successively incubated (i) for 48 h at 4°C with the antibody to NPY (UCB, Belgium) or to PNMT (kindly provided by L. Denoroy, Lyon, France) diluted 1:20 000 in PBS + 0.1% saponin, (ii) for 18 h at 4°C with a peroxidaselabelled Fab immunoglobulin fragment of sheep anti-rabbit globulin (Biosys, Compirgne, France) diluted 1:1000 in PBS + 0.1% saponin, and (iii) with 0.1% 3,3'-diaminobenzidine diluted in 0.05% Tris buffer at pH 7.3, in the presence of 0.02% H202. The histochemical specificity of antibodies is guaranteed by the furnisher for anti-NPY, and has been assessed in previous studies for anti-PNMT (see Denoroy, 1979 and Kithahama et al., 1985). Present controls included incubation of the sections without primary antiserum, or with primary antisera previously adsorbed with their specific antigens (50 ~tg antigen/ ml diluted antiserum). About half of the labelled sections were then mounted on gelatin-coated slides and observed under a light microscope. A Samba 2005 image analysis system (Alcatel) was used for quantifying the differences in the densities of PNMT- or NPY-immunostained fibres throughout the PVN, under the different conditions ofaxonal lesioning used here. The microscope magnification was chosen to obtain a camera video field that can be included in the pPVN. The area analysed was thus represented by a square of 400 ~tm each side, which was centred on the pPVN so that its dorsal boundary was located at the level of the roof

0

~.~ ~ • "~i ~

~-~ ~,,~,~ ~ ~ ~,~

~ ~ S ~- ,~~ ~

~'~

-i ~, ~ ~ii~,~.ii~, •

~"

PNMT- and NPY synapses in the paraventricular nucleus of the third ventricle and the medial boundary immediately lateral to the wall of the third ventricle (Fig. 3A). For each section, the camera video field was first centred on the pPVN on the side opposite to the lesion. The image was digitalized and a level of filtration was adjusted to eliminate background staining. The numerical density of immunoreactive fibres was estimated as the surface of immunostained structures measured in this field. It was then compared to that measured, in the same section and with the same level of filtration, in an identical field centred on the pPVN homolateral to the axonal lesion. For each rat, such an analysis was performed on at least three sections containing the hypothalamic nucleus, and four or five rats displaying optimal conditions for both the location of stereotaxic lesions and the immunostaining reactions were used for each type of unilateral axonal lesion. A M a n n Whitney U-test was used for the statistical comparison of the mean labelled surface measured in the PVN homo- and contralateral to the lesion. The remaining labelled vibratome sections were further treated for electron microscopic observations, after being postfixed with 1% OsO 4 solution in 0.1 M-cacodylate buffer (pH 7.4) containing 1.25% potassium ferricyanide. After dehydration in graded concentrations of ethanol, the sections were embedded in epon. For each labelled section studied, one punch of 1.5 mm diameter was cut through the pPVN in both the lesioned and intact sides, once these regions had been microscopically localized. Punches were then mounted on epon blocks and cut horizontally into ultrathin sections, which were observed in an EM 900 Zeiss electron microscope, either without subsequent staining, or after counterstaining with 5% uranyl acetate. Quantitative assessments for the organization of labelled synapses were performed on counterstained ultrathin sections (see Milner et al., 1988). In each group of animals bearing brainstem or hypothalamic knife-cuts, tissues from the three animals with the best immunocytochemical labelling and preservation of ultrastructural morphology were used for quantitative analysis. From each of these animals, eight to ten grids supporting two to five ultrathin sections were collected through both PVN from three or more plastic-embedded vibratome sections. All labelled profiles within one thin section from each grid were photographed for analysis. Differences in the frequencies in the types of PNMT- or NPY-immunoreactive (IR) synapses and the types of neuronal structures connected were statistically analysed by means of a Mann-Whitney U-test. Such

59

an analysis of PNMT- and NPY-IR synapses was additionally performed in the pPVN of two intact rats similarly treated for immunocytochemistry. RESULTS The patterns of immunostaining observed for both the anti-PNMT and anti-NPY immunosera in the PVN of the unlesioned side fully conformed to previous descriptions (see Zoli et al., 1988; Cunningham et al., 1990). After a series of attempts to quantify immunocytochemical labelling of intracellular antigens, it is generally accepted that a number of variables bound to the immunocytochemical treatment can greatly influence the intensity of the immunostaining reaction (Berod et al., 1981). Thus, in order to eliminate possible immunostaining differences between different sections of individual or different rats, the estimations of the variations of density in immunostained fibres were based on the comparison of the immunostaining pattern observed on the same frontal brain section, in the sides homo- and contralateral to a unilateral lesion. Moreover, since both PNMT- and NPY-IR fibres were essentially concentrated within the medial parvicellular part of the PVN (pPVN), the quantitative estimations at light and electron microscopic levels of the modifications induced by both types of axonal lesioning were restricted to this subdivision of the nucleus.

Light microscopy The data concerning the image analysis quantification of the modifications of PNMT- or NPYimmunostaining within the pPVN are shown in Fig. 2. Brainstem knife-cut Since the hypothalamic PVN is known to receive a bulk of noradrenergic fibres from the medullary regions, the effectiveness of the lesions was controlled on each lesioned animal by observing three to four vibratome sections cut through the PVN in the fluorescence microscope under violet illumination. The animals retained for this study were those which exhibited a strong decrease of fluorescent fibres within the PVN homolateral to the lesion. In the brainstem of these animals, dense accumulations of both PNMT- and NPY-IR fibres were observed just proximal to the lesion (Fig. 3C,D). In the hypothalamus, the density of PNMT-

Fig. 3. Light micrographs of vibratome sectionsimmunostained to PNMT (A and C) or NPY (B and D) in an animal treated with a unilateral brainstem knife-cut. (A) and (B) Frontal sectionsthrough the PVN (the lesioned side is indicated by the asterisk in (A). The number of PNMT-IR fibresis dramaticallydecreasedin the PVN homolateralto the lesionas comparedto the control side (A), whereas the densityin NPY-IR fibresis comparableon both sides(B).The squarein (A) representsthe area analysedto performthe quantification of the densities of immunostained fibres in the pPVN. (C,D) Sagittal sections through the brainstem. Both PNMT (C) and NPY (D) denselyaccumulatewithin the proximalcut ends of fibresabutting on the surgicalcut. V: third ventricle;bar = 50 ~m.

i

I

r

, - J , ¸~' ' ¸

P N M T - and NPY synapses in the paraventricular nucleus IR fibres was markedly decreased within the nucleus homolateral to the lesion as compared to the contralateral control side (Fig. 3A). By contrast, no significant variation of the density of immunostained NPY-fibres could be detected throughout the pPVN homolateral to the cut (Fig. 3B). Hypothalamic knife-cut In the animals retained for this study, the cut was arching from 1 to 1.5 mm above the ventral surface of the medial hypothalamus to 1.5 to 2 mm lateral to the midline up to the optic tract (Fig. 1B). In these animals, only slight if any modifications of the density of P N M T - I R fibres could be detected in the pPVN (Fig. 4A). By contrast, in all animals a marked decrease in the density of NPY-IR fibres was observed in the pPVN on the lesioned side (Fig. 4B). In the periventricular portion of the lesion, P N M T - I R fibres were not modified (Fig. 4C). On the other hand, a strong pile-up of large varicose NPY-IR fibres was observed beneath the cut (Fig. 4D). In some instances NPY-IR perikarya could be detected within the periventricular hypothalamic region located beneath the transection site.

Electron microscopy In the ultrathin sections examined, the peroxidase reaction product contained in labelled profiles was mainly associated with the core of large (80120 nm) vesicles and with the membrane o f the small (25-55 nm) microvesicles. Within axonal profiles immunostained with anti-PNMT or anti-NPY, the number and size of labelled organelles appeared very similar. For both antibodies, immunostaining was restricted to unmyelinated structures that were identified as (i) axons when the structures were smaller than 0.5 lam and did not form synapticjunctions, and (ii) axon terminals when they measured 0.5-1.4 ~tm in diameter and contained a few mitochondria, numerous clear microvesicles and one to five large dense-cored vesicles. The present study essentially focused on labeled synapses which could be clearly identified according to the morphological criterion defined by Gray (1959): (1) asymmetric synapses were characterized by prominent postsynaptic thickening and by a large synaptic cleft (Gray's type 1); and (2) symmetric synapses were characterized by narrow postsynaptic density and a synaptic cleft that was not clearly distinguishable from normal membrane appositions (Gray's type 2). Besides these two types of identified synapses, some individual synapses were difficult to characterize as symmetric or asymmetric because o f a heavy

61

accumulation of peroxidase reaction product on the pre- or post-synaptic side of the membrane. Electron micrographs were analysed to determine (i) the types of junction involved in the well-characterized synapses, and (ii) the types of neuronal structure connected to the synapses. The data obtained for both intact and operated animals are summarized in Table 1. They indicate that in operated animals, the organization of both types of labelled synapses in the pPVN contralateral to the lesion was similar to that observed in the nuclei of intact animals. The effects of the different types of lesions were thus evaluated in operated animals, by comparing the organization of labelled synapses in the pPVN of both lesioned and unlesioned sides. Unlesioned side For both antibodies used, labelled synapses were found to contact unlabelled perikarya or dendritic processes. In both cases axo-somatic synapses represented less than 20% of the total labelled synapses and were essentially of the symmetric type (Fig. 5B,E). On the other hand marked differences were detected between the NPY- or P N M T - I R synapses involved in axo-dendriticjunctions. A large majority of P N M T - I R synapses was made o f asymmetric axo-dendritic synapses (64% of P N M T - I R total synapses) (Fig. 5A), whereas this type of synapse only represents a minority (20%) of the NPY-IR detected in the pPVN. On the other hand 67% of NPY-IR total synapses were found to form symmetric synapses on either dendrites or perikarya of this nucleus (Fig. 5C,D and E). Lesioned side With both types of axonal lesions, numerous unlabelled profiles exhibited ultrastructural features of degenerating axons, i.e. they were engulfed by glial processes and their cytoplasm frequently appeared electron dense. In animals which had received a brainstem lesion, P N M T - I R profiles were composed of a large number of degenerating axons and of intact synapses that, as in the unlesioned side, preferentially formed asymmetric axo-dendritic contacts (69% of intact P N M T - I R synapses). In these animals, the NPY-IR synapses identified in the pPVN showed a slight but significant decrease in the proportion of asymmetric synapses connected to dendrites (14% vs 20% in the unlesioned side, P < 0.05). In animals bearing a hypothalamic horizontal lesion, the organization of the P N M T - I R synapses was not modified as compared to the control side (Fig. 6A). In these animals, numerous

Fig. 4. Lightmicrographsof frontal vibratomesectionsimmunostainedto PNMT (A and C) and NPY (B and D) in an animal treated with a unilateral hypothalamicknife-cut(the lesioned side is indicated by an asterisk in (A). (A,B) Frontal sections through the PVN. The densityin PNMT-IR fibresis similarthroughout the PVN of control and lesionedsides (A). In contrast, a marked decreasein the density in NPY-IR fibresis observedthroughout the PVN on the lesionedside(B). (C,D) Frontal sectionsthrough the pcriventricularhypothalamic region underlyingthe PVN. No accumulation of PNMT is observedaround the cut (C), whereas strong accumulationof NPY is observedin numerousfibreslocated beneath the cut (D). V: third ventricle;bar = 50 p.m.

62

G. Alonso Table 1. Frequency of the different types of P N M T and NPY-IR synapses in the pPVN contralateral (control) and homolateral to brainstem or hypothalamic knife-cuts Type of synapse

as-som

s-som

as-den

s-den

s?

% of P N M T - I R synapses % of N P Y - I R synapses

1 (n=l) I (n=l)

12 (n=10) 20 (n=20)

63 (n=52) 20 (n=20)

12 (n=10) 48 (n=48)

12 ( n = 10) 20 (n=10)

pPVN in intact rats

% of PN M T - I R synapses % of N P Y - | R synapses

2 (n = 2) 1 (n = 2)

15 (n = 16) 15 (n = 30)

64 (n = 70) 20 (n = 41)

10 (n = 11) 52 (n = 104)

9 (n = 10) 12 (n = 25)

Control pPVN (contralateral to knife-cuts)

% of P N M T - I R synapses % of NPY-IR synapses

2 (n = 1) ! (n=l)

l0 (n = 6) 13 (n=22)

69 (n = 42) 14 (n=24)

11 (n = 7) 59 (n=100)

8 (n = 5) 13 (n=22)

pPVN homolateral to brainstem knife-cut

% of P N M T - I R synapses % of N P Y - I R synapses

2 (n = 2) 2 (n=5)

6 (n = 6) 8 (n=17)

69 (n = 66) 59 (n=127)

17 (n = 16) 19 (n=42)

6 (n = 6) 12 (n=26)

pPVN homolateral to hypothalamic knife-cut

as-som: asymmetric axo-somatic synapses; s-som: symmetric axo-somatic synapses; as-den: asymmetric axodendritic synapses; s-den: symmetric axo-dendritic synapses; s?: unidentified synapses.

degenerating NPY-IR axonal profiles could be observed throughout the PVN (Fig. 6B). However, the majority of the NPY-axonal profiles exhibiting the normal features of synapses formed asymmetric junctions with dendritic profiles (59% vs 20% in the unlesioned side P < 0.01; Fig. 6B). DISCUSSION The observations that the brainstem knife-cut used here induced a dramatic decrease of both the catecholamine-fluorescent and PNMT-IR fibres throughout the PVN homolateral to the lesion clearly show that this lesion disrupted the ascending CA pathway connecting the CA medullary neurons to the hypothalamus. Such a lesion, however, was not found to induce any detectable modification in the numerical density of NPY fibres within the pPVN. Such data, which are in full agreement with a series of previous immunocytochemical studies (Bai et al., 1985; Gustaffson and Moore, 1987) strongly suggest that only a very minor part of the NPY-IR fibres detected in the PVN arise in medullary CA neurons. This, however, appears contradictory with data indicating that (i) the large majority of the PNMT-IR perikarya of the medulla oblongata that innervate the PVN also contain NPY (Sawchenko et al., 1985) and (ii) similar surgical lesioning of the brainstem induces a 50-60% decrease in NPY content of the PVN as measured by radioimmunoassay of the peptide in microdissected hypothalamic nuclei (Sahu et al., 1988). A possible explanation for such discrepant results could be that the postlesional delays used are not long enough for detecting a decrease of immunoreactivity within transected NPY-IR fibres. Indeed, in a recent study

using similar surgical lesions through the brainstem, it was reported that the decrease of PNMT-IR within the PVN was much more spectacular at 14 than at 10 days after the lesion (Palkovits et al., 1992). It is also possible that the medullary PNMTIR fibres projecting to the PVN contain low amounts of NPY that are hardly detected by the immunocytochemical approach used here. This is supported by the present observation that the number of NPY-IR fibres remaining in the PVN after a horizontal hypothalamic knife-cut always appeared far less than that of PNMT-IR fibres detected in the same nucleus (Fig. 4A,B). A series of neuroanatomical studies have shown that axonal fibres of medullary origin penetrate the PVN either laterally by running between the zona incerta and the fornix, or, to a lesser extent, ventrally by running along the periventricular area (Ter Horst et al., 1989; Cunningham et al., 1990). It is thus likely that the horizontal hypothalamic knifecuts performed here transected these medullary fibres ascending vertically through the periventricular zone. However, the present finding that such lesions did not induce any significant decrease in the density of PNMT-IR fibres innervating the PVN strongly suggests that the ventral ascending pathway provides a minor contribution to the medullary innervation of the PVN. On the other hand, that such horizontal knife-cuts placed between the PVN and the mediobasal hypothalamus induced a dramatic decrease in the density of NPY-IR fibres throughout the PVN, strongly reinforces the theory that a large majority of NPY-IR fibres that innervate the PVN arise from non-catecholaminergic neuronal perikarya located in the mediobasal hypothalamus (Bai et al., 1985). That the majorit,y of NPY-IR axons terminating in the PVN do not

PNMT- and NPY synapses in the paraventricular nucleus

63

Fig. 5. Electron micrographs of synapses immunostained to PNMT (A and B) or NPY (C,D and E) in the parvocellular paraventricular nucleus of the control side. Two axon terminals immunostained to PNMT form typical asymmetric and symmetric synapses respectively on unlabelled dendrites (A) or perikarya (B). Axon terminals immunostained to NPY form symmetric synapses both on small (C) or large (D) dendritic profiles and on a neuronal perikarya (E). G: Golgi apparatus; m: mitochondria; N: nucleus; bar = 0.5 jam,

64

G. A l o n s o

Fig. 6. Electron micrographs of synapses immunostained to PNMT (A) and NPY (B) in the parvocellular paraventricular nucleus homolateral to a hypothalamic knife-cut. Axon terminals immunostained to PNMT (A) or NPY (B) form typical asymmetric synapses on dendritic profiles. Note that in (B), a degenerating axonal profile surrounded by numerous glial processes exhibits slight immunostaining (asterisk). Bar = 0.5 ~tm.

arise f r o m c o m b i n e d P N M T - N P Y - I R n e u r o n s o f the m e d u l l a o b l o n g a t a is also s u p p o r t e d b y the present evidence o f a m a r k e d u l t r a s t r u c t u r a l differe n t i a t i o n between P N M T - a n d N P Y - I R synapses

detected within the nucleus. Since i m m u n o r e a c t i v e synapses were n o t e x a m i n e d on serial sections, it is clear that the d a t a r e p o r t e d here do n o t represent the exact q u a n t i t a t i v e d i s t r i b u t i o n o f s y m m e t r i c a n d

PNMT- and NPY synapses in the paraventricular nucleus asymmetric labelled synapses that were actually present in the sections examined. For instance, the number of asymmetric synapses may be underevaluated since postsynaptic thickening, which characterizes this type of synapse, can be observed in only about one third of the ultrathin sections cut through the synaptic axon terminal (Beaudet and Sotelo, 1981). On the other hand, symmetric synapses may be overevaluated, since a number of axon terminals exhibiting the morphological criteria attributed to this type of synapse may represent sections of asymmetric synapses that do not include the postsynaptic density. Moreover, a number of immunostained synapses could not be identified as symmetric or asymmetric. It is clear, however, that such mismatches of quantitative assessment affect in the same fashion the proportion of synapses within the population of PNMT- or NPY-IR synapses detected in the PVN. Thus, the present finding that, within the control pPVN, the large majority of PNMT synapses form asymmetric contacts, whereas the majority of NPY synapses form symmetric contacts, certainly attests for markedly differentiated patterns ofultrastructural organization between PNMT- and NPY-IR synapses. These data appear contradictory with those of a previous study reporting that besides symmetric NPY-IR synapses, a number of asymmetric NPYIR synapses were observed throughout the PVN (Sawchenko and Pfeiffer, 1988). However, such discrepant data very likely result from the fact that these authors also observed the magnocellular compartments of the PVN where asymmetric NPY synapses were predominant. Interestingly, however, the organization of NPY-IR synapses remaining in the pPVN after a hypothalamic horizontal knife-cut was considerably modified as compared to the unlesioned control side. In these animals, NPYIR synapses were predominantly axo-dendritic synapses of the asymmetric type, which appeared similar to the organization of PNMT synapses. This supports the hypothesis that NPY axons of medullary origin co-store PNMT and mainly form asymmetric axo-dendritic synapses, whereas non-catecholaminergic NPY axons originating in the mediobasal hypothalamus predominantly form symmetric synapses. It is known from a series of anatomical and physiological data that (i) NPY axons terminating in the pPVN are connected to corticotropinreleasing hormone (CRH-41)-producing neurons, on which they mainly form asymmetric synapses (Liposits et al., 1988) and (ii) the local injection of NPY produces a powerful activation of the pituitary-adrenocortical axis (H~rfstrand et al., 1987; Wahlestedt et aL, 1987; Leibovitz et al., 1988). The CRH neurons of the pPVN have also been shown to be connected by asymmetric PNMT-IR synapses (Liposits et al., 1988), and as for NPY, the local injection of adrenaline stimulates the pituitary-adrenocortical axis (Szafarczyk et al.,

65

1987; Leibovitz et al., 1988). This provides strong support for the idea that NPY and adrenaline that are contained in the same synaptic axons of medullary origin act synergistically to stimulate CRHproducing neurons. Moreover, this suggests that, as proposed for other neuron types (Uchizone, 1965; Cohen et al., 1982) asymmetric NPY and PNMT synapses mediate excitation. Similar to the PVN, NPY synapses have been reported to predominantly form symmetric synapses in several brain regions exhibiting a dense NPY innervation such as the striatum (Vuillet et al., 1989), the suprachiasmatic nucleus (Ibata et al., 1988) and the cortex (Hendry et al., 1984). So far, the nature of the neurons of the pPVN that receive the symmetric NPY-IR synapses is mostly unknown. It is possible that a number of them represent neurons that project to the medulla oblongata and/or to the spinal cord and are involved in the central control of autonomic functions (Sofroniew and Schrell, 1981; Sawchenko and Swanson, 1982a; Luiten et al., 1985). In this connection, the injection of NPY in the PVN has been shown to inhibit gastric acid secretion (Humphreys et al., 1988) and to induce both bradycardia and fall in systolic blood pressure (Harland et al., 1988). It could thus be assumed that, similarly to symmetric synapses described in other brain regions (Milner et al., 1987; Pickel et al., 1988), NPY-IR synapses of the pPVN exert an inhibitory control of the neurons to which they are connected. During recent years, special interest has been paid to the role of NPY in the control of endocrine and physiological functions that modulate energy metabolism (see Leibovitz, 1991). Due to the strong effects produced by the injection of NPY in the vicinity or in the PVN on these specific functions, it is likely that they are directly or indirectly controlled by the NPY axon terminals that innervate the pPVN. However, a better understanding of the possible roles played by NPY synapses in the regulation of these functions obviously will need a better anatomical and biochemical characterization of the postsynaptic neurons of the pPVN connected by these synapses. In conclusion, light and electron microscope examinations of NPY innervation of the pPVN of rats treated with surgical lesions disrupting axonal pathways from hindbrain or mediobasal hypothalamic regions indicate that (i) the large majority of NPY-IR fibres detected in the pPVN with current methods form symmetric synapses and arise from neurons of the mediobasal hypothalamus, and (ii) a minority of these fibres form asymmetric synapses and originate in adrenergic medullary neurons that co-store PNMT. •

ACKNOWLEDGEMENTS The author wishesto thank A. Legrandfor her excellenttechnical assistance. This studywas supported by grant no. 886001 from INSERM.

66

G. Alonso

REFERENCES Abe, M., Saito, M. and Shimazu, T. (1989). Neuropeptide Y and norepinephrine injected into the paraventricular nucleus of the hypothalamus activate endocrine pancreas. Biomed. Res. 10, 431-436. Bai, F. L., Yamano, M., Shiotani, Y., Emson, P. C,, Smith, A. D., Powell, J. F. and Tohyama, M. (1985). An arcuato-paraventricular and -dorsomedial hypothalamic neuropeptide Y-containing system which lacks noradrenaline in the rat. Brain Res. 331, 172-175. Beaudet, A. and Sotelo, C. (1981). Synaptic remodeling of serotonin axon terminals in rat agranular cerebellum. Brain Res. 206, 305-329. Berod, A., Hartman, B. K. and Pujol, J. F. (1981). Importance of fixation in immunocytochemistry: use of formaldehyde solutions at variable pH for the localization of tyrosine hydroxylase. J. Histochem. Cytochem. 29, 844-850. Chronwall, B. M., DiMaggio, D. A., Massari, V. J., Pickel, V. M., Ruggiero, D. A. and O'Donohue, T. L. (1985). The anatomy of neuropeptide Y-containing neurons in rat brain. Neuroscience 26, 461-478. Cohen, R. S., Carlin, R. K., Grab, D. J. and Siekevitz, P. (1982). Phosphoproteins in postsynaptic dendrites. Prog Brain Res. 56, 49-76. Cunningham, E. T., Bohn, M. C. and Sawchenko, P. E. (1990). Organization of adrenergic inputs to the paraventricular and supraoptic nuclei of the hypothalamus in the rat. J. Comp. Neurol. 292, 651-667. Denoroy, L. (1979). Utilisation de la phenylethanolamineN-methyltransferase comme marqueur des structures adr6nergiques. Application ~, l'6tude de l'hypertension art6rielle. Th+se de 3° cycle, Lyon, France. Everitt, B. J., H6kfelt, T., Terenius, L., Tatemoto, K., Mutt, V. and Goldstein, M. (1984). Differential coexistence of neuropeptide Y (NPY)-Iike immunoreactivity with catecholamines in the central nervous system of the rat. Neuroscience 11,443-462. Furness, J. B., Heath, J. W. and Costa, M. (1978). Aqueous aldehyde (Faglu) Methods for the fluorescence histochemical localization of catecholamines and for ultrastructural studies of central nervous tissue. Histochemistry 57, 285-295. Gray, E. G. (1959). Axo-somatic and axo-dendritic synapses of the cerebral cortex: an electron microscopic study. J. Anat. 93, 420-431. Gray, T. S. and Morley, J. E. (1986). Neuropeptide Y: anatomical distribution and possible function in mammalian nervous system. Life Sci. 38, 389-401. Gustafson, E. L. and Moore, R. Y. (1987). Noradrenaline and neuropeptide Y innervation of the rat hypothalamus are differentially affected by 6-hydroxydopamine. Neurosci. Lett. 83, 53-58. Hal~sz, B. and Pupp, L. (1965). Hormone secretion of the anterior pituitary gland after physical interruption of all nervous pathways to the hypophysiotropic area. Endocrinology 77, 553-562. H~rfstrand, A. (1986) Intraventricular administration of neuropeptide Y (NPY) induces hypotension, bradycardia and bradypnoea in the awake unrestrained male rat. Co unteraction by NPY-induced feeding behaviour. Acta Physiol. Scand. 128, 121-123. H~rstrand, A., Eneroth, P., Agnati, L, and Fuxe, K. (1987). Further studies on the effects of central administration of neuropeptide Y on neuroendocrine

function in the male rat: relationship to hypothalamic catecholamines. Regul. Pept. 17, 167-179. Harland, D., Bennett, T. and Gardiner, S. M. (1988) Cardiovascular actions of neuropeptide Y in the hypothalamic paraventricular nucleus of conscious Long Evans and Brattleboro rats. Neurosci. Lett. 85, 239-243. Hendry, S. H. C., Jones, E. G. and Emson, P. C. (1984). Morphology, distribution, and synaptic relations of somatostatin- and neuropeptide Y-immunoreactive neurons in rat and monkey neocortex. J. Neurosci. 4, 2497-2517. H6kfelt, T., Fuxe, K., Goldstein, M. and Johansson, O. (1974). Immunohistochemical evidence for the existence of adrenaline neurons in the rat brain. Brain Res. 66, 235 251. H6kfelt, T., Lundberg, J. M., Tatemoto, K., Mutt, V., Terenuis, L., Polak, J., Bloom, S., Sasek, C., Elde, R. and Goldstein, M. (1983a). NeuropeptideY (NPY)- and FMRFamide neuropeptide-like immunoreactivities in catecholamine neurons of the rat medulla oblongata. A cta Physiol. Scand. 117, 315-318. H6kfelt, T., Lundberg, J. M., Lagercrantz, H., Tatemoto, K., Mutt, V., Lindberg, J., Terenius, L., Everitt, B. J., Fuxe, K., Agnati, I. and Goldstein, M. (1983b). Occurrence of neuropeptide Y (NPY)-like immunoreactitivity in catecholamine neurons in the human medulla oblongata. Neurosci. Lett. 36, 217-222. Humphreys, G. A., Davison, J. S. and Veale, W. L. (1988). Injection of neuropeptide Y into the paraventricular nucleus of the hypothalamus inhibits gastric acid secretion in the rat. Brain Res. 456, 241-248. Ibata, Y., Takahashi, Y., Okamura, H., Kubo, T. and Kawakami, F. (1988). Fine structure of NPYcontaining neurons in the lateral geniculate nucleus and their terminals in the suprachiasmatic nucleus of the rat. Brain Res. 439, 230--235. Kitahama, K. J., Pearson, J., Denoroy, L., Kopp, N., Ulrich, J., Maeda, T. and Jouvet, L. (1985). Adrenergic neurons in human brain demonstrated by immunocytochemistry with antibodies to phenylethanolamine-Nmethyltransferase (PNMT): discovery of a new group in the tractus solitarius. Neurosci. Lett. 53, 303-308. Leibovitz, S. F. (1991). Brain neuropeptide Y: an integrator of endocrine, metabolic and behavioral processes. Brain Res. Bull. 27, 333-337. Leibovitz, S. F., Sladek, C., Spencer, L. and Tempel, D. (1988). Neuropeptide Y, epinephrine and norepinephrine in the paraventricular nucleus: stimulation of feeding and the release of corticosterone, vasopressin and glucose. Brain Res. Bull. 21,905-912. Liposits, Zs., Phelix, C. and Paull, W. K. (t986). Electron microscopic analysis of tyrosine hydroxylase, dopamine-B-hydroxylase and phenylethanolamine-Nmethyltransferase immunoreactive innervation of the hypothalamic paraventricular nucleus in the rat. Histochemistry, 84, 105-120. Liposits, Zs., Sievers, L. and Paull, W. K. (1988). Neuropeptide-Y and ACTH-immunoreactive innervation of corticotropin releasing factor (CRF)-synthesizing neurons in the hypothalamus of the rat. Histochemistry 88, 227-234. Luiten, P. G. M., Ter Horst, G. J., Karst, H. and Steffens, A. B. (1985). The course of paraventricular hypothalamic efferent to autonomic structures in the medulla and the spinal cord. Brain Res. 329, 374-378.

P N M T - and N P Y synapses in the paraventricular nucleus Milner, T. A., Abate, C., Reis, D. J. and Pickel, V. M. (1988). Ultrastructural localization of phenylethanolamine-N-methyltransferase-like immunoreactivity in the rat locus coeruleus. Brain Res. 478, 1-15. Milner, T. A., Pickel, V. M., Chan, J., Massari, V. J., Oertei, W. H., Park, D. H., Joh, T. H. and Reis, D. J. (1987). Phenylethanolamine-N-methyltransferase-containing neurons in the rostral ventrolateral medulla of the rat: II. Synaptic relationships with GABAergic terminals. Brain Res. 411, 46-57. Moltz, J. H. and McDonald, J. K. (1985). Neuropeptide Y: direct and indirect action on insulin secretion in the rat. Peptides 6, 1155-1159. Palkovits, M., Tapia-Arancibia, L., Kordon, C. and Epelbaum, J. (1982). Somatostatin connections between the hypothalamus and the limbic system of the rat brain. Brain Res. 250, 223-228. Palkovits, M., Mezey, E., Skirboll, L. R. and H6kfelt, T. (1992). Adrenergic projections from the lower brainstem to the hypothalamic paraventricular nucleus, the lateral hypothalamic area and the central nucleus of the amygdala in rats. J. Chem. Neuroanat. 5, 407-415. Paxinos, G. and Watson, C. (1982). The Rat Brain in Stereotaxic Coordinates. Academic Press, London. Pickel, V. M., Towle, A. C., Joh, T. H. and Chan, J. (1988). Gamma-aminobutyric acid in the medial nucleus accumbens: ultrastructural localization in neurons receiving monosynaptic input from dopaminergic afferents. J. Comp. Neurol. 272, 1-14. Sahu, A., Kalra, S. P., Crowley, W. R. and Kalra, P. S. (1988). Evidence that NPY-containing neurons in the brainstem project into selected hypothalamic nuclei: implication in feeding behavior. Brain Res. 457, 376-378. Sawchenko, P. E. and Pfeiffer, S. W. (1988). Ultrastructural localization of neuropeptide Y and galanin immnoreactivity in the paraventricular nucleus of the hypothalamus in the rat. Brain Res. 474, 231-245. Sawchenko, P. E., Swanson, L. W., Grzanna, R., Howe, P. R. C., Bloom, S. R. and Polak, J. (1985). Colocalization of neuropeptide Y immunoreactivity in brainstem catecholaminergic neurons that project to the paraventricular nucleus of the hypothalamus. J. Comp. Neurol. 241, 128-153. Sawchenko, P. E. and Swanson, L. W. (1981). Central noradrenergic pathways for the integration of hypothalamic neuroendocrine and autonomic responses. Science 214, 685-687. Sawchenko, P. E. and Swanson, L. W. (1982a). Immunocytochemical identification of neurons in the paraventricular nucleus of the hypothalamus that project to the

67

medulla or to the spinal cord. J. Comp. Neurol. 21}5, 260-272. Sawchenko, P. E. and Swanson, L. W. (1982b). The organization of noradrenergic pathways from the brainstem to the paraventricular and supraoptic nuclei in the rat. Brain Res. Rev. 4, 275-325. Sawchenko, P. E. and Swanson, L. W. (1983). The organization and biochemical specificity of afferent projections to the paraventricular and supraoptic nuclei. Prog. Brain Res. 60, 19-21. Sofroniew, M. V. and Schrell, U. (1981). Evidence for a direct projection from oxytocin and vasopressin neurons in the hypothalamic paraventricular nucleus to the medulla oblongata: immunohistochemical visualization of both the horseradish peroxidase transported and the peptide produced by the same neurons. Neurosci. Lett. 22, 211-217. Stanley, B. G. and Leibovitz, S. F. (1985). Neuropeptide Y injected in the paraventricular hypothalamus: a powerful stimulant of feeding behavior. Proc. Natl. Acad. Sci. USA 82, 3940-3943. Szafarczyk, A., Malaval, F., Laurent, A., Gibaud, R. and Assenmacher, I. (1987). Further evidence for a central stimulatory action of catecholamines on adrenocorticotropin release in the rat. Endocrinology 121,883-894. Ter Horst, G. J., De Boer, P., Luiten, P. G. M. and Van Willigen, J. D. (1989). Ascending projections from the solitary tract nucleus to the hypothalamus. A Phaseolus vulgaris lectin tracing study in the rat. Neuroscience 31, 785-797. Uchizono, K. (1965). Characterization of excitatory and inhibitory synapses in the CNS of the cat. Nature 207, 642-643. Vuillet, J., Kerkerian, L. and Nieoullon, A. (1989). Ultrastructural features of NPY-containing neurons in the rat striatum. Brain Res. 477, 241-251. Wahlestedt, C., Skagerberg, G., Ekman, R., Heilig, M., Sundler, F. and Hakanson, R. (1987). Neuropeptide Y (NPY) in the area of the hypothalamic paraventricular nucleus activates the pituitary-adrenocortical axis in the rat. Brain Res. 417, 33-38. Zoli, M., Agnati, L. F., Fuxe, K., Zini, I., Merlo Pich, E., Grimaldi, R., H~irfstrand, A., Goldstein, M., Wikstr6m, A. C. and Gustafsson, J. A. (1988). Morphometrical and microdensitometrical studies on phenylethanolamine-N-methyltransferase- and neuropeptide Y-immunoreactive nerve terminals and on glucocorticoid receptor-immunoreactive nerve cell nuclei in the paraventricular hypothalamic nucleus in adult and old male rats. Neuroscience 26, 479-492.

Accepted 24 November 1992