D.F. Swaab, M.A. Hofman, M. Mirmiran, R. Ravid and F.W. van Leeuwen (Eds.) Progress in Brain Research, Vol. 93 0 1992 Elsevier Science Publishers B.V. All rights reserved.

167

CHAPTER 12

Ontogeny of peptides in human hypothalamus in relation to sudden infant death syndrome (SIDS) N. Kopp', M. Najimi', J . Champierl, F. Chigr', Y. Charnay3, J . Epelbaum2 and D. Jordan' I

Laboratoire d'Anatomie Pathologique, Facultt! de Mddecine Alexis Carrel, 69372 Lyon Cedex 08, France; INSERM U 159, Centre Paul Broca, 75014 Paris, France; and Laboratoire de Neuropathologie, Clinique Be1 Air, Geneva, Switzerland

Introduction The mammalian brain is not mature at birth. This is particularly true for humans. During the end of gestation and in the post-natal period, growth and development of the brain are influenced by the hormonal state. This is well illustrated in the hypothalamus. For instance, it has been shown that the hypothalamo-pituitary adrenal axis is transformed from a state of hyperactivity during early stages of gestation to a state of partial control during late stages (Donovan, 1980). The maturation of the regulation of the hypothalamo-pituitary system by the brain is not'completed at birth and goes on during a long post-natal period (Donovan, 1980). In addition, the maturation of the hormonal pattern, as illustrated physiologically by the chronobiological variations of thyrotropin-releasing hormone (TRH) and thyro-stimulating hormone (TSH) of the hypothalamo-pituitary axis, has been well documented in the neonatal period up to adulthood in rat (Jordan et al., 1989). The authors report'that a hypothalamic TRH circadian rhythm was present at all ages studied (between 15 and 70 days old) whereas the low amplitude of pituitary TSH rhythms detected in the young rat disappeared in the adult. In contrast, the serum TSH rhythm becomes consistent to reach the well-characterized circadian mid-day peak in the adult rat. In fact, during the

post-natal period, circadian variations of the concentration of some pituitary hormones, such as thyroid, growth and gonadotrophic hormones, have been observed in both rats and humans. TSH pituitary concentrations are very low at birth and reach a maximum at the end of the second post-natal week (Winter, 1983). The control of growth hormone (GH) secretion does not reach a complete functional state before the post-natal period. Serum concentrations of GH are more elevated in neonates than in 3-month-old humans (Winter, 1983). This suggests a post-natal set-up of the regulatory system and of the control of GH secretion. Gonadotropic hormones play a very important role during the neonatal period in the development of the hypothalamo-pituitary axis. Correlations have been established between temporary changes of the concentration of sexual hormones and the alteration of the hypothalamo-hypophyso-gonadal system during the first post-natal year of cerebral differentiation (Darner, 1985). Serum concentrations of gonadotropins are elevated between birth and the third post-natal month (Forest, 1983). Furthermore, interrelations were established between hormonal changes and their hypothalamic control factors (TRH, somatostatin (SRIF), luteinizing hormone releasing-hormone (LHRH)). Biochemical and immunohistochemical studies performed in rat and man have revealed important

168

differences in the distribution of these hypothalamic factors between neonates and adults. Compared to animal research, human data in the field of the anatomical distribution of hypothalamic factors and their binding sites are very limited. To our knowledge, the only reports on SRIF and LHRH distribution based on immunocytochemistry are those by Bugnon et al. (1977a,b), Bloch (1978) and Fellman (1978) in the fetus and by Barry (1978) and Bouras et al. (1987) in adults. No precise binding site study of SRIF or TRH in the human hypothalamus is available. We therefore performed immunohistochemical mapping of SRIF and LHRH, and binding site studies by quantitative autoradiography of SRIF and TRH in the hypothalamus of normal infants and adults. Some of these markers have been applied to four cases of sudden infant death syndrome (SIDS). In addition, we report here the first preliminary data on the immunohistochemical distribution of delta sleep-inducing peptide (DSIP) in the control infant hypothalamus. Disorders of thermoregulation are hypothesized in SIDS and the peptides and areas studied here might be implicated in this regulatory mechanism. Generally speaking, detailed and precise anatomical normative data will serve as a basis for neuropathological studies and might be useful for neuro-imaging of neurologic, psychiatric and neuroendocrinologic patients. Methodological questions raised by studies of peptides and binding sites in the human hypothalamus We shall review briefly the general pitfalls that arise from human neuropathological studies as well as specific inset of hypothalamus and/or peptides. We shall consider ante- and post-mortem factors. Ante-mortem factors Sex. Many studies have failed to take the factor sex into consideration. However, several macro-

scopic studies indicated the possibility of differences associated with sex in brain morphology (Kopp et al., 1977; De Lacoste-Utamsing and Holloway, 1982). Some differences are described at a microscopic level: number of cells in Onuf‘s nucleus of the spinal cord (Forger and Breedlove, 1986), size of the hypothalamic sexually dimorphic nucleus as well as the number of its cells (Swaab and Fliers, 1985). Laterality. Lateralisations for dopamine, cholineacetyltransferase, noradrenaline and GABA have been reported (Oke et al., 1978; Amaducci et al., 1981; Click et al., 1982). Radioimmunological measurement of LHRH and TRH in twelve microdissected areas of the adult human hypothalamus recently revealed a higher concentration of TRH but not LHRH in three areas corresponding to the paraventricular, dorsal and ventromedial nuclei on the left side (Borson-Chazot et al., 1986). Otherfactors. The influence of nutritional factors or drug treatment on peptidergic systems in humans is not, to our knowledge, well taken into account. Such factors should not be underestimated, especially in the hypothalamus (this part of the brain is highly implicated in the control of satiety; some side effects of psychotropic treatments are of the neuroendocrine type). Quite obviously individual variations raise the problem of “normality”, which is as much a philosophical problem as a scientific one. Age is, of course, crucial in an ontogenic study; it should be stressed here that “normal” autopsy cases of late infancy, adolescence and young adulthood are very rare. This often brings a gap, especially around puberty, in the series aiming at covering all periods of life. Post-mortem factors Post-mortem delay is usually understood as the period of time between death and fixation of tissue or freezing of tissue. Both immunohistochemistry and binding site studies of peptides are influenced by post-mortem delay.

169

tions, following a plane parallel to the plane defined by the anterior limit of the optic chiasm and of the anterior white commissure. Finally, the sections were mounted on chromalun-gelatin-coated slides Posr-mortem delay and ~mm~no~istochemistry.and stored at - 20°C. We found that storage up to In our experience of immunohistochemistry of 6 months gave no different results. SRIF, LHRH and DSIP, delays of up to 12 h did not For immunohistochemistry, the peroxidaseantiperoxidase (PAP) technique (Sternberger et al., interfere noticeably with the quality of the results. 1970) was used with minor modifications: (1) preWe tried, as others before, TRH immunohistotreatment of the sections with a 1Yo hydrogen peroxchemistry with several antibodies in the human brain without any success. This may have been due ide for 20 min at room temperature to inhibit the ento the rapid post-mortem diffusion, explainable by dogenous peroxidases; and (2) addition of nickelthe very small size of this peptide, or to degradation. ammoniumsulfate to the final colour developing solution (Kitahama et al., 1985) to increase the senBiochemical studies have shown a good stability of LHRH and SRIF, even in cases with a delay as long sitivity. In the immunohistochemical controls, we as 36 h (Parker and Porter, 1982; Sorensen, 1984). first omitted the specific antiserum or substituted it by non-immune serum (which gave no positive reaction); second, each antiserum used was absorbed Post-mortem delay and binding sites. Most autoonly by its homologous peptide (peptide concentraradiographic studies performed on human brain tion of 20 pM) and produced the same results, irtissue tend to show a good stability up to 24 h or even 36 h for peptide binding sites (Cash et al., respective of its being preabsorbed or not by other 1987). Our own results do corroborate previous data peptides tested (such as neurotensin, vasoactive intestinal peptide, beta-endorphin, cholecystokinin (Manaker et al., 1986; Reubi et al., 1986) for SRIF and substance P). and TRH binding sites. This stability may be a The primary antisera used were: antisomatostatin general property of binding sites since we have also 1-12 (Chayvialle et al., 1978), antisomatostatin 15found it in non-peptidergic ligands such as serotonin 28 (SRIF) (Tohyama, Osaka), antisomatostatin 28 and benzodiazepine, confirming data from the liter(Peninsula Laboratories), anti-LHRH (Bugnon et ature (Palacios et al., 1983; Zezula et al., 1988). al., 1977a) and anti-DSIP (Charnay et al., 1989). Irnmunohistochemistry Quantitative autoradiography Fixation was achieved after autopsy by perfusion After the brain had been removed from the skull, a through the carotis interna. After 5 min of perfusion block of fresh tissue including the hypothalamus with 200 ml saline buffer, 1.5 1 of paraformand adjacent structures was excised. This block had aldehyde 4% (0.1 M phosphate buffer, pH 7.4) was approximately the shape of a parallelepiped with perfused in approximately 30 min at room anterior and posterior aspects parallel to the plane temperature. The leptomeninges were carefully defined by the anterior limit of the chiasm and of the peeled away and the entire brain was immersed in anterior white commissure. The caudal plane was the same fixation medium overnight at 4°C. A block just behind the mammilary complex. Inorder to preof brain including the hypothalamus and adjacent vent anatomical distortions as much as possible, this structures was excised and immersed in a renewed fresh block was supported and surrounded by a sort identical fixation medium for 6 - 10 days at 4°C. of extemporaneously prepared and adapted boxFollowing 1 week of immersion in 20070 sucrose at shaped mould made of strong aluminium foil. The 4"C, the tissue block was frozen in liquid nitrogen mould, with the tissue inside it, was laid on a copper and cut with a cryostat into 20 pm thick coronal secPost-mortem delay and morphology. The number and size of dendritic spines are very sensitive to postmortem delay (De Ruiter, 1983).

170

block previously cooled at -80°C. By using this freezing procedure, minimal crystal artefacts were obtained. Quantitative autoradiography (Young and Kuhar, 1979) was preferred to a membrane binding method because of (1) the high degree of anatomical resolution and (2) the possibility of quantification with image analysis. The apparatus used was either an Imstar-Starwise (Paris) station or a Rag 200 Biocom image analysis system (Les Ulis, France). For anatomical localization of the hypothalamic nuclei and areas, adjacent sections to those used for autoradiography were stained with cresyl violet. Nuclei and areas were identified according to an atlas of the human hypothalamus developed in our laboratory and according to the cytoarchitectural reports on the human hypothalamus of Diepen (1962) and Braakand Braak (1987). SRIF binding sites

Preliminary desaturation by guanosine triphosphate (GTP) appeared to be a necessary condition to obtain a good visualization of binding sites in some areas. In vitro GTP treatment has a dramatic desaturating effect on sites occupied by SRIF (Lerouxet al., 1988). Indeed, it was shown that binding sites to SRIF in the brain are coupled to G proteins (Enjalbert et al., 1983). It has also been reported that GTP increases the dissociation of the radioactive ligand from its receptor (Moyse et al., 1989). Leroux et al. (1988) have used this property of GTP by adding it to the preincubation medium; the result was the detection of somatostatinergic binding sites in several more hypothalamic structures (preoptic area, suprachiasmatic periventricular nuclei and mediolateral region of posterior hypothalamus), whereas adjacent sections not treated with GTP were not labeled. In the present study, M of GTP in the preincubation medium dramatically increased SRIF binding in the preoptic area, diagonal band of Broca, ventromedial and dorsomedial nuclei. Conversely, it had no effect on some structures, such as the infun-

dibular nucleus and tuber nuclei. The effect of GTP was not due to a change in affinity but to a difference in accessibility. This would tend to prove, in an indirect manner, the coupling of SRIF binding sites to G proteins. However, this coupling could be differential. After the addition of GTP in different concentrations (10- loM) during incubation, a decrease of autoradiographic labeling (70% at l o p 7 M) was noticed in both the preoptic area and the infundibular nucleus. This supports the hypothesis that SRIF binding sites couple to G proteins in GTP-sensitive regions (i.e., preoptic area) and GTP-insensitive areas (i.e., infundibular nucleus). Saturation studies made in anterior and mediobasal levels of the hypothalamus show that binding sites to SRIF have a high affinity (Kd in the nanomolar range). It is not different in infants and adults. Furthermore, such affinity is comparable to those usually reported for SRIF binding. It should also be mentioned that displacement experiments with SRIF-14 - as well as withTyr-0-Trp 8 SRIF-14 - show that the specific binding is competitively inhibited. TRH binding sites [3H]3MeTRH was used as a ligand for TRH binding sites in different animal species. This analogue binds to TRH sites with a high affinity, as has been shown both in human brain homogenates (Parker and Capdevila, 1984)and in sections by quantitative autoradiography (Manaker et al., 1986; Jordan et al., 1989). In the present study, the specific binding represented more than 90% of total binding in the hypothalamic structures. Saturation experiments were performed in two different hypothalamic areas (preoptic area and infundibular nucleus). They demonstrated that [3H]3MeTRH binding is saturable. Furthermore, the ligand bounds with a high affinity (dissociation constant of the nanomolar range) to TRH sites in both the preoptic area and the infundibular nucleus. Inhibition experiments showed that among ana-

171

logues tested only 3MeTRH and TRH are efficient in their inhibiting action. However, 3MeTRH is 10 times as potent as TRH. The characteristics are the same in infants and adults. In contrast to the effect on SRIF, GTP preincubation had no significant effect on the density of TRH binding sites. Distribution of SRIF immunoreactive neurons Our study is based on four male infant brains (26, 30, 40 and 60 days old, respectively). The infants died of circular of umbilical cord, pulmonary hypoplasia associated with operated diaphragmatic hernia, acute staphylococcus sepsis, therapeutic perfusion error. The presence of several molecular varieties leads to cautiousness in the interpretation of our results. Indeed, besides the SRIF-14 variety, other forms of SRIF, also called prosomatostatins, are present in both the brain and the peripheral nervous system (Benoit et al., 1985). Other studies have shown wide differences in the distribution of SRIF-28 immunoreactive neurons and of SRIF-28 1 - 12 immunoreactive neurons in the monkey (Bakst et al., 1985). A predominance of the SRIF-14 has been reported in human and rat brain whereas the SRIF28 proportion varies widely (Emson et al., 1981;Biggins et al., 1984). Immunohistochemical studies show an exhaustive distribution of SRIF-28 in cell bodies as well as in fibres (Bakst et al., 1985). Binding sites of SRIF-14 and of SRIF-28 have the same general distribution (Srikant and Patel, 1981). Comparison of the labeling obtained with our three different antibodies did not reveal any noticeable difference. Immunohistochemical controls consisted of first, the omission of the specific antiserum or its substitution by non-immune serum; second, each antiserum was blocked only with its homologous peptides (1 pm); and third, the same results were obtained whether antiserum was preabsorbed or not with other peptides. Our results (see Fig. 1) are in general agreement with other data in fetuses and adult humans

(Bugnon et al., 1977b; Bouras et al., 1987). In fetuses, immunoreactive cell bodies are found in paraventricular and suprachiasmatic nuclei as well as in the periventricular area. Immunoreactive fibres are localized in the anterior hypothalamus and in mammillary bodies. However, we find a higher density in the paraventricular nucleus, especially in the parvocellular part. We found a higher density of cell bodies in infundibular and posterior nuclei than has been found in adults (Bouras et al., 1987). In the fetus, no fibres have been reported in the dorsomedial, ventromedial, posterior and mammillary nuclei (Bugnon et al., 1977b). In adults, only a low density of fibres has been observed in the ventromedial nucleus and the median eminence (Bouras et al., 1987). In mammillary bodies, a decrease of fibres has been described in adults, but without specifying the nucleus. These differences, observed mainly between infants and adults, could be due to differences in methods, differences in antibodies or differences between the anatomical atlases used by different authors. However, they could also be due to the fact that this peptidergic system is implicated in maturation and development of neurons. In rats, Inagaki et al. (1982) found SRIF immunoreactive neurons in the granular layer of the cerebellar cortex of neonates but not of adults. In our laboratory, a decrease of another peptide, neurotensin, was found in diencephalic structures, especially in mammilary bodies (Sakamoto et al., 1986). Long and large diameter fibres (“woolly” fibres, described in the rat pallidum by Haber and Nauta, 1983, and in the human septum by Gaspar et al., 1987) were found in the median preoptic area, anterior hypothalamic area and, to a lesser extent, in the lateral preoptic area. In our laboratory, in the same cases and in three additional ones, substance P immunoreactive “woolly” fibres were stable from 26 to 60 days of age. Compared with the distribution in rats, our results show a higher density of SRIF cell bodies in the anterior hypothalamus (including the preoptic

172

H

J

. . . . , . . . . . . ...

3v

L

173

area in infants). In rat, the densest concentration of cell bodies is present in the periventricular nucleus,

whereas in human infants, it is in the paraventricular nucleus. As in previous studies in humans, we did not detect SRIF fibres in the fornix, as has been described in rats by Roberts et al. (1982). Distribution of SRIF binding sites

Fig. 2. Autoradiography of somatostatin binding sites in the tuber nuclei (TN) of an adult human. Thesenucleiareprominently rich in binding sites and can be easily located between optic tract (OT) and fornix (F).

This study was performed on 16 brains: 7 infants and 9 adults. The general distribution pattern (Fig. 2) shows a rostro-caudal decrease in the hypothalamus and inside some of the hypothalamic nuclei and structures. However, in the mediobasal hypothalamus, the SRIF binding sites increase in the infundibular nucleus in a rostro-caudal direction. Similar observations were made in human fetal spinal cord (Charnay et al., 1988). The same distribution was found in infants and adults. However, some differences should be emphasized: there was a higher density of SRIF binding sites in the posterior part of the infundibular nucleus and in the anterior hypothalamic area in infants. In contrast, the developmental pattern of SRIF binding sites was rather delayed in the tuberal nuclei as compared with other regions. It is interesting to note that these nuclei are well differentiated only in humans and not in lower animals (Grunthal, 1929). Thus, a later ontogenic appearance could reflect the recent phylogenic evolution of these nuclei. These specific differences in the regional distribuOf a peptide between infant and suggest that this peptide could be involved in neuronal maturation. Other studies support the presence of a

Fig. 1. Topographical distribution of somatostatin immunoreactive perikarya fibres in the human infant hypothalamus. Asterisks represent cell bodies; large dots, large fibres; small dots, small fibres. Abbreviations: AA, anterior hypothalamic area; AC, anterior commissure; CP, cerebral peduncle; DA, dorsal area; DBBh or DBH, diagonal band horizontalis; DBBV or DBV, diagonal band verticalis; DM, dorsomedial nucleus; DPC, decussatio pedunculorum cerebellorum superiorum; F, fornix; FF, field of Forel; FLM, fasciculus longitudinalis medialis; I, infundibular nucleus; IC, inferior colliculus; IP, interpeduncular nucleus; LH, lateral hypothalamic area; LM, lateral mammillary nucleus; LP, lateral preoptic area; LT, lamina terminalis; LTu, lateral tuberal nucleus; ME, median eminence; MES, mesencephalon; MM, median mammillary nucleus; MP, medial preoptic area; MT, medial tuberal nucleus; OC, optic chiasma; ON, optic nerve; OT, optic tract; PA, posterior hypothalamic area; PE, periventricular nucleus; PV, paraventricular nucleus; s c , suprachiasmatic nucleus; SO, supraoptic nucleus; SPT, supratrochlearis nucleus; TM, tubero mammillary nucleus; VM, ventromedial nucleus; 21, zona incerta; 3V, third ventricle. (Reproduced from Najimi et al., 1989a, with permission.)

174

...

d

b

.......

f SRIF-LR fibers densities 1251-SRIF binding sites densities

3

+

I +

,6000

moderate

verylow

0 0

dpm/mm2

6000-5000 dpm/mm2 5000-2000 dpm/mm2

~ 2 0 0 0dprn/mrn2

Fig. 3. Comparison of somatostatin immunoreactive densities (left side) and binding site densities (right side). Density scales are in dicated on graph. Abbreviations: see legend of Fig. 1. (Reproduced from Najimi et al., 1991a, with permission.)

175

similar action of SRIF on molluscs in vitro (Bulloch, 1987; Grimm-Jorgensen, 1987). The difference (higher number of binding sites in medial preoptic area, diagonal band of Broca, ventromedial and dorsomedial nuclei) in labeling of SRIF binding sites in tuber were found in both sexes: this suggests that gonadotrophins and sex steroids do not influence the expression of these SRIF binding sites. The differences (higher number of binding sites in medial preoptic area, diagonal band of Broca, ventromedial nucleus) between our results and those of Reubi et al. (1986) could be explained, at least partly, by the use of GTP. Indeed, GTP preincubation enhanced the binding in both infants and adults, especially in the diagonal band of Broca. Comparison of the distribution of SRIF and its binding sites (Fig. 3)

sites is relatively lower and increases in the posterior parts. In the posterior hypothalamus, a moderate SRIF binding site density was also demonstrated in the lateral mammillary nucleus, where, at least in the newborn infant, moderate to high densities of SRIF immunoreactive fibres had been reported. However, the posterior hypothalamic area (containing a moderate to high density of SRIF immunoreactive fibres) as well as the medial mammillary nucleus (with no SRIF immunoreactive fibres) display a low SRIF binding site density. It was originally suggested (Reubi et al., 1986) that the mismatch between the peptide and its binding sites could be explained by the fact that several members of the somatostatin family, not detectable by immunocytochemistry, could bind to the binding site. Following desaturation with GTP, the SRIF binding sites were generally in good correlation with immunohistochemical data.

There is an overall good correlation between the distribution of SRIF immunoreactive fibres and SRIF binding sites. The highest densities of SRIF binding sites are located in the preoptic area, corresponding to a high density of fibres in the same area. Furthermore, it should be mentioned that both fibres and binding sites display the same rostrocaudal distribution. There is also a good correlation between the distribution of SRIF fibres and SRIF binding sites in the anterior hypothalamic area, where both exhibit moderate densities. In the mediobasal hypothalamus, the dorsomedial and ventromedial nuclei are homogeneous with respect to densities of SRIF binding sites and SRIF fibres. In contrast, the infant ventromedial nucleus, displaying highly SRIF immunoreactive fibres, contains only a moderate density of SRIF binding sites. This could be due to the fact that a large number of these fibres are fibres “en passage”, with their corresponding binding sites located outside the ventromedial nucleus. In the infundibular nucleus, there is a good correlation between immunoreactive fibres and binding sites. Thus, the density of SRIF immunoreactive fibres is higher in the anterior parts of this nucleus, whereas the density of SRIF binding

Functional implications The presence of SRIF fibres terminating on the third ventricle in infants would favour a possible release of the neuropeptide to the cerebrospinal fluid. It has been suggested that this release would be followed by a transport by cerebrospinal fluid to extra-hypothalamic regions in order to evoke some of its behavioural effects. The very dense innervation of the median eminence, where fibres are concentrated around blood vessels, probably originates from the paraventricular and infundibular nuclei. Neuropeptide release in the portal blood stream is in good agreement with the generally proposed neuroendocrine role of hypothalamic SRIF. The differences found between infants and adults support the hypothesis that SRIF concentrations decrease with age. A decrease of the concentrations of some hypothalamic factors has also been described in rat (Steger et al., 1979; Wise and Ratner, 1980; Rice et al., 1983; McDonald, 1987; Pekary et al., 1987;Morimoto et al., 1988). Thedecreasein somatostatin innervation of the median eminence with age in man and rat strongly suggests a decrease of secretory activity of these hypothalamic neuronal

176

systems implicated in the regulation of pituitary functions. The results of immunohistochemical and autoradiographic studies reveal the presence of SRIF cell bodies, SRIF fibres and SRIF binding sites in some hypothalamic nuclei and areas implicated in the control of distinct physiological functions. For instance, the preoptic area contains a dense group of SRIF cell bodies and fibres as well as a high density of SRIF binding sites, which is of interest, since this area seems to be implicated in the control of thermoregulation in rat (Briese, 1989). Moreover, other studies indicate that SRIF may indeed play a role in thermoregulation (Lipton and Glyn, 1980; Brown et al., 1981; Chandara et al., 1981;Wakabayashi et al., 1983). All these data confirm a role of SRIF in the control, from birth onwards, of thermoregulation in man. Distribution of TRH binding sites

Anatomical distribution The distribution of TRH binding sites in the

human hypothalamus was studied in 14 cases (8 adults and 6 infants of both sexes) (see Table I). This distribution is widespread, heterogeneous and predominant in the anterior and mediobasal hypothalamus (Figs. 4 and 5 ) . As for SRIF, the study of TRH binding sites reveals regional differences in the distribution pattern between infants and adults. In infants, the density is higher in the diagonal band of Broca, whereas in adults the density is greater in the tuber nuclei. The decrease with aging in the density of TRH binding sites in the diagonal band of Broca suggests that these sites could be implicated in functions other than synaptic transmission such as cellular differentiation and migration (Zagon and McLaughlin, 1986; Hauser et al., 1987). As in the case of SRIF binding, TRH binding sites appear much later in the tuber nuclei than in other hypothalamic areas. This latency would suggest that even though the systems of binding sites of these two hypophysiotropic factors are already in their definitive location in the post-natal period, they have not yet reached their

TABLE 1 Cases in which TRH autoradiographic binding site studies were performed -

Cases

Sex

Age

Post-mortem delay

Cause of death

Adults A B C D E F G H

M F M F F M M F

22 years 27 years 34 years 42 years 45 years 41 years 67 years 82 years

30 h 25 h 28 h 30 12 h 20 h 12 h l h 2 h 30

Cardiac failure Cardiac failure Sudden death Sudden death Coronary thrombosis Cardiac failure Myocardic infarct Cardiac failure

Neonates I J

F M

2h 35 h

14 h 34 h

K

M

1 day

10 h

L M N

M M F

1 month 1 month 1 year

20 h 5h 7h

Pulmonary hypoplasia Amniotic inhalation with gastric regurgitation Oedematic alveolitis with refractory hypoplasia Liver lesions Enterocolitis Hepatic necrosis

177

Comparison between TRH binding sites and the radioimmunological distribution of TRH

Fig. 4. Autoradiography of TRH binding sites in an adult human: diagonai band of Broca.

Fig. 5. Autoradiography of TRH binding sites in an adult human: mammillary nuclei (MN).

complete maturation and adult functional level at that time. It should be noted that, as for SRIF binding sites, the distribution of TRH sites is unaffected by sex in humans. Some differences with the rat should be mentioned: no TRH binding site was found in the infundibular, ventromedial and tuber nuclei in this species (Rostkne et al., 1984; Sharif, 1989). The discrepancy may be explained, for tuber nuclei, by their very small size in the rat.

The study of the distribution of TRH in 10 adult human hypothalami performed in our laboratory (Borson-Chazot et al., 1986), based on “punch” microdissection technique and radioimmunological assays showed a significant predominance of the TRH concentration in three areas (ventromedial, dorsal and paraventricular) on the left side. In our TRH binding site study, we studied the distribution on both sides but found no difference in laterality. To our knowledge, no precise and detailed TRH distribution study is available in man since no immunohistochemical data have been reported. This can probably be explained by rapid diffusion of the peptide TRH in post-mortem tissues. Attempts have been made in our laboratory with different antibodies, different fixation media and different procedures, on relatively short post-mortem cases (as short as 5 h), without any success. However, the comparison of our autoradiographic study with available microbiochemical studies in man (Parker and Porter, 1983; BorsonChazot et al., 1986) tends to reveal a generally good correlation between TRH and its binding sites, especially in the preoptic area, the infundibular and ventromedial nuclei and mammillary bodies.

Functional implications Besides its role in the control of the release of prolactin and TSH, the implication of TRH in thermoregulation is well established. In several species, intracerebral injection of TRH induces an alteration of thermoregulation. In rats that are awake, the direct intra-hypothalamic injection of TRH is followed by an evaluation of rectal temperature (Lin and Yang, 1989). The site of action of TRH seems to be the anterior hypothalamus and, more precisely, the preoptic area, since lesions of the preoptic area induce alterations of thermoregulation. In the preoptic area, TRH would be responsible for the excitation of coolness-sensitiveneurons and of inhibition of neurons sensitive to warmth (Lin and Yang, 1989).

178

J

Fig. 6. LHRH immunoreactive neurons in human brain. Large dots symbolize cell bodies; small dots represent fibres. Abbreviation! see legend of Fig. 1 . (Reproduced from Najimi et al., 1990, with permission.)

179

Our findings of a high density of TRH binding sites in the human preoptic area could be considered an anatomical support of such a control, in adults as well as infants. No autoradiographic labeling was observed in the median eminence, so that the occurrence of a local action in this structure could not be confirmed. The terminal TRH fibres seen around vessels in that structure in animals would therefore support the hypothesis of a hypophysiotropic action of this peptide. Immunohistochemical distribution of LHRH neurons

Anatomical distribution This study was carried out on five infant hypothalami (ages of subjects: 26 days, 30 days, 40 days, 60 days and 9 months; three males and two females). The highest density of cell bodies is found in the mediobasal hypothalamus, principally in the infundibular nucleus. The preoptic area has the second highest population. In these two structures, the majority of immunoreactive cell bodies are densely marked, large and bipolar. Concerning fibres, three main structures can be distinguished by their density and the distribution: the lamina terminalis, the periventricular and paraventricular nuclei. The median eminence has the highest density of fibres, with a preferential distribution around vessels. The posterior hypothalamus has the lowest density of fibres (see Fig. 6).

Comparison with human data published Our findings are in general agreement with those reported in the literature (Barry, 1976, 1977, 1978; Bugnon et al., 1977a; King and Anthony, 1984). However, these studies showed high densities in the supraoptic nucleus, in contrast with the low density of LHRH immunoreactive neurons which we found. This discrepancy could be explained by the difference in antibodies and by differences in anatomical definitions. The overall distribution of LHRH immunoreac-

tivity is in good agreement with that described in adults and fetuses. However, King and Anthony (1984) do not report a high density of LHRH innervation in the preoptic region and only a low density of fibres in ,adults. In the median preoptic area, we observed a moderate labeling of cell bodies whereas high concentrations were found in the fetus (Bugnon et al., 1977a). In the anterior part of the hypothalamus, LHRH innervates the following structures: lateral preoptic area, anterior hypothalamic nucleus, dorsal hypothalamic nucleus, supraoptic nucleus and suprachiasmatic nucleus. We have also for the first time observed dense aggregations of LHRH fibres and rare cell bodies in periventricular and paraventricular nuclei.

Functional implications The presence of LHRH immunoreactive fibres around thickenings of third ventricle walls may be of great interest. One possibility is that LHRH is released into the cerebroventricular fluid and transported, probably by specialized ependymocytes, towards the median eminence and portal circulation, thus leading to an elevation of plasma LH levels (Knigge et al., 1978). There are several indications for this: the presence of LHRH in the cerebrospinal fluid (Joseph et al., 1975), the absorption by tanycytes of LHRH injected in ventricles inducing an elevation of plasma LH (Uemura et al., 1975). The discrepancies between our results in infants and those published by others in adults (Barry, 1976; King and Anthony, 1984) could be explained by cell death, which is known as one of the major characteristics of normal development (Parnavelas and Cavanagh, 1988). Immunohistochemical and autoradiographic studies have shown that, in the occipital cortex of rats, there is an overproduction of SRIF neurons with a maximum at 2 weeks after birth; immediately after that, a decrease is observed, suggesting neuronal death (Parnavelas and Cavanagh, 1988). Thus, this peptide could be produced “in surplus” and have a role in the establishment of a given con-

180

nection, and later disappear. It should be noticed, however, that arguments in favour of atrophic role of this peptide become more and more obvious (Bulloch, 1987; Palacios et al., 1988; Parnavelas and Cavanagh, 1988). Cell death after the neonatal period might indeed also occur in the human hypothalamus. As already mentioned, we have shown the presence of immunoreactive neurons in the suprachiasmatic nucleus. It is well established that this nucleus plays a major role in the control of circadian rhythms (Brown-Grant and Raisman, 1977; Raisman and Brown-Grant, 1977; Moore, 1979; Hery et al., 1982). Our observations may have important functional implications. Thus, in rat, the electrical stimulation of the suprachiasmatic nucleus increases the release of LHRH and LH (Maxwell and Fink, 1988) in the portal circulation (Chiappa et al., 1977). Immunohistochemical studies have revealed the presence of LHRH neurons in the suprachiasmatic nucleus which project towards the median eminence by pathways running dorsally and ventrally to the optic chiasm (Rethelyi et al., 1981). These observations would be in favour of the implication of the suprachiasmatic nucleus in the “generation” of the circadian rhythm responsible for the release of luteinizing hormone. It would be of interest to follow the fate of LHRH neurons in the suprachiasmatic nucleus in diseases such as Alzheimer’s disease and Prader-Willi syndrome (Swaab et al., 1987). The cell bodies present in the preoptic area may project towards the median eminence via the infundibular nucleus. Thus, tracing studies have revealed direct connections between the preoptic area and the mediobasal hypothalamus (Conrad and Pfaff, 1976; Fink and Jamieson, 1976)on the one hand and the increase of metabolic activity of the infundibular nucleus after electrical stimulation of the preoptic area on the other hand (Maxwell and Fink, 1988). Our observations of the presence of LHRH immunoreactive cell bodies in the preoptic region and of LHRH immunoreactive fibres in the infundibular nucleus and median eminence confirm the existence of such a projection in infants.

Comparison of the distribution of LHRH immunoreactive elements in the left and right hypothalamus did not reveal differences in their nature for their density. These results are in good agreement with radioimmunological microchemical studies of Borson-Chazot et al. (1986), which did not show lateralization on the concentration of LHRH in the human adult hypothalamus. Distribution of delta sleep inducing peptide (DSIP) This unpublished study revealed a great similarity in the patterns of distribution of DSIP immunoreactive neurons and that of LHRH (Fig. 7). It has been shown that, in the rabbit, DSIP and LHRH are colocalized in the same neuronal population (Charnay et al., 1989). Electron microscopic studies in the rat median eminence demonstrated that both DSIP and LHRH immunoreactivities are present within single granules of secretion (Vallet et al., 1991). Studies on sudden infant death syndrome Several convergent data presently support the hypothesis that immaturity of brain areas is responsible for the control of vegetative functions (respiration, heart and blood pressure, swallowing,

Fig. 7. Distribution of delta sleep inducing peptide immunoreac. tive neurons in the human hypothalamus. Abbreviations: set legend of Fig. 1.

181

temperature, sleep . . .). The areas regulating these processes are mainly confined to the spinal cord, brain-stem and hypothalamus. Possible thermoregulation and respiratory control disorders in SIDS prompted us to look at the hypothalamus and lower brain-stem; we studied SRIF immunoreactive neurons since SRIF is known to stimulate or inhibit respiration. We therefore investigated the topographical distribution of SRIF immunoreactive neurons as well as the distribution of LHRH immunoreactive neurons in the hypothalamus of four cases of SIDS of 1 month, 1 month, 1.5 months and 3 months of age (with a post-mortem delay of 6, 7, 8 and 10 h, respectively). The controls were 1 month, 1.5 months, 2 months and 5 months (post-mortem delay: 4,6,8 and 10 h, respectively). The causes of death were, respectively: pulmonary hypoplasia in a case of operated diaphragmatic hernia, staphylococcus acute sepsis, perfusion (therapeutic) error, congenital cardiomyopathy. In both series, there are one female and three males. The general autopsy includes gross examination and a protocol of 24 tissue samples from lung, myocardium, heart septum, thymus, oesophagus, stomach, liver, adrenal gland, kidney, pancreas, rib. No significant lesion was found in SIDS cases, which were therefore diagnosed as real unexplained SIDS. We found no difference for the SRIF distribution study between controls and SIDS cases. With LHRH immunohistochemistry, the same general distribution was found in the hypothalamus of controls and SIDS cases. However, in the periventricular and paraventricular nuclei, a dramatic difference appeared in the number of LHRH immunoreactive fibres, which was greatly decreased in the four SIDS cases (Fig. 8). This diminution of the number of fibres of LHRH immunoreactive neurons, in four cases of SIDS as compared with four matched controls, has to be considered with cautiousness until artefacts or other causes unrelated to the syndrome can be definitely discarded. However, this deficiency in LHRH fibres cannot be clearly explained by a post-mortem artefact since

Fig. 8. LHRH immunoreactive fibres in human infant hypothalamic paraventricular nucleus. Upper part: control infant; lower part: case of sudden infant death syndrome. (Reproduced from Najimi et al., 1989b, with permission.)

SIDS cases and controls were well matched for postmortem delay. Sex is probably not responsible for our results either since there were males and females in the two series. This deficiency could thus reflect a real diminution of the number of fibres in periventricular and paraventricular nuclei. However, it could also reflect a depletion of LHRH by which fibres could not be visualized any longer. A nycthemeral factor should also be taken into consideration since in SIDS cases death occurred at night, whereas it occurred during the day for the matched controls. Clearly the present data need to be confirmed by further investigations. It should be mentioned here that previous work from our laboratory has shown changes in SIDS in other neuroregulator systems:

182

absence of catecholaminergic neurons in the subnucleus gelatinosus of the nucleus tractus solitarius and increase of the number of neurotensin binding sites in the nucleus tractus solitarius (Denoroy et al., 1987; Chigr et al., 1989b, 1991). If the present finding of a decrease of LHRH immunoreactive fibres in the mediobasal part of the hypothalamus in SIDS is confirmed, the functional significance will justify further investigation and reflection. Presently we can only speculate on a possible implication of a deficiency in LHRH fibres in the anterior part of hypothalamus. The LHRH fibres of the periventricular and paraventricular nuclei are known in animals to originate from cell bodies localized in the anterior hypothalamus, an area known to be implicated in thermoregulation and thermoregulation disorders have been mentioned in infants at risk of SIDS. However, we are fully aware of the fragility of this hypothesis. Since DSIP and LHRH are colocalized in the same neurons, one could take also less DSIP in SIDS as an argument in favour of the reality of our finding since DSIP has been implicated in sleep mechanisms and thermoregulation (as well as in other mechanisms) and sleep anomalies are well documented in infants at risk of SIDS (Challamel et al., 1981; Harper et al., 1983). However, this is pure speculation since DSIP was not studied in our SIDS cases. Furthermore, one must notice here that the anterior hypothalamus is strongly implicated in the control sleep-wakefulness states (Sakai et al., 1990). The present findings are to be considered as preliminary data and should encourage investigators in neuroscience and neuroendocrinology to become involved in SIDS research. Summary and conclusions The brains of mammals are not mature at birth, in particular in humans. Growth and brain development are influenced by the hormonal state in which the hypothalamus plays the major regulatory role. The maturation of the hormonal patterns leads to the physiological establishment of chronological variations as revealed by the circadian variations of

both hypothalamic peptides and pituitary hormones (as illustrated for hypothalamic-pituitary-thyroid axis by the determination of thyro-stimulating hormone (TSH) and thyrotropin-releasing hormone (TRH) circadian rhythms in the rat (Jordan et al., 1989)). It has been established that hypothalamic peptide variations are regulated by hormonal feed-back and amine systems, with the maturation of the latter also being dependent upon the whole functional maturation of the brain. Though these systems have been studied in the rat, very little information is currently available with regard to the human brain. The only biochemical or immunohistochemical information published to date concerns either the fetus or the adult. We have studied four main peptidergic systems (somatostatin-releasing inhibiting factor (SRIF), thyrotropin-releasing hormone (TRH), luteinizing hormone-releasing hormone (LHRH) and delta sleep inducing peptide (DSIP) in post-mortem adults and infants and in sudden infant death syndrome (SIDS) brains either by autoradiography and/or immunochemistry of radioimmunology. From a technical point of view, human brain studies display certain pitfalls not present in animal studies. These may be divided into two subclasses: ante- and post-mortem. Ante-mortem problems concern mainly sex, laterality, nutritional and treatment patterns while post-mortem problems concern post-mortem delay and conditions before autopsy and hypothalamic dissection. This might induce dramatic changes in morphological, immunochemical and autoradiographic evaluations. The matching of pathological subjects with controls is particularly difficult in the case of SIDS because of the rapid changes which take place in physiological regulatory processes during the first year of life. Thus, the treatment of hypothalamic tissue samples both for immunochemistry, radioimmunology and autoradiographic studies required techniques which must be rigorously controlled. For example, SRIF studies were carried out with three different antibodies, which gave similar results. The use of different technical procedures as well as dif-

183

ferent antibodies is discussed. These types of differences might explain, at least in part, the discrepancy observed until now. As previously described in the fetus (Bugnon et al., 1977b; Bouras et al., 1987), we confirmed that in the infant hypothalamic SRIF immunoreactive cell bodies are present in the paraventricular and suprachiasmatic nuclei and in the periventricular area. We also showed the presence of a higher density of cell bodies in the infundibular and posterior nuclei as compared with adults. In contrast to results obtained in rat (Roberts et al., 1982), we did not detect SRIF fibres in the fornix of either infants or adults. The distribution of SRIF binding sites was similar in infants and adults. However, two major specific regional differences between infants and adults were evident. A higher density of labeling was found in the posterior infundibular nucleus and the anterior hypothalamus of infant brain whereas the tuber nuclei were much more labeled in adults. It is noteworthy that 2 - 3 times as many SRIF binding sites were revealed by adding guanosine triphosphate (GTP) to the preincubation medium. A high density of immunoreactive cell bodies and binding sites for SRIF was found in the preoptic area. This area is known to be involved in thermoregulation and therefore this result is particularly important when considering SIDS pathology. TRH Previous radioimmunological determinations of the concentration of TRH in hypothalamic structures in our laboratory revealed a left prominence for the ventromedial, dorsal and paraventricular nuclei (Borson-Chazot et al., 1986) in adults. Our present results show no evidence of laterality in hypothalamic TRH binding sites in adults or infants. These comparative measurements in infants and adults give evidence of specific structural differences. First of all, the highest density of binding sites was found in the preoptic area of the infant. This density was similar to that found for SRIF; TRH is also known to be involved in thermoregula-

tion. A high density was also found in the diagonal band of Broca in infants while in contrast in adults, the density was higher in the tuber nuclei. The results that levels of both TRH and SRIF binding sites in the tuber nuclei are significantly lower in infants than in adults suggest that this structure has not achieved functional maturation at this stage in development. LHRH and DSIP Here, we report for the first time the presence of a dense aggregation of LHRH fibres in periventricular and paraventricular nuclei. These observations are of particular interest since similar studies performed in four SIDS pathology cases revealed a dramatic decrease in the number of LHRH immunoreactive fibres in these two hypothalamic nuclei. Confirmation of these results would support the hypothesis of a link between the control of thermoregulation and SIDS pathology since, at least in animals, these fibres originate in the anterior hypothalamus. Furthermore, DSIP and LHRH are colocalized in the same neurons (Charnay et al., 1989) and indeed we found, in the human brain, a similar immunohistochemical distribution for the two peptides. DSIP is known to be implicated in sleep mechanisms. Since sleep abnormalities are well documented in infants at risk of SIDS and since in almost all SIDS cases death occurred during the night, it might be of interest to further investigate the poFsible role of DSIP in this pathology. /

/