Synaptic and neuronal-glial plasticity in the adult oxytocinergic system in response to physiological stimuli

Brain Research Bulletin, Vol. 20, pp. 681-692. 0 Pergamon Press plc, 1988.Printed in the U.S.A.

0361-9230188 $3.00 t XI0

Synaptic and Neuronal-Glial Plasticity in the Adult Oxytocine~gic System in Response to Physiological Stimuli CATHERINE ~ONTAGNESE, DOMINIQUE A. POULAIN, JEAN-DIDIER VINCENT AND DIONYSIA T. THEODOSIS

Unit4 de Neurobiologie des Compartments, INSERM U. 176 and University of Bordeaux II, I Rue Camille Saint-SaEns, F33077 Bordeaux,

France

MONTAGNESE, C., D. A. IWULAIN, J-D. VINCENT AND D. T. THEODOSIS. Synaptic and neurunal-glialplusticiry oxytocinergic system in response to physiological stimuli. BRAIN RES BULL 20(6) 681-692, 1988.Magnocellular oxytocinergic neurons in the hypothalamus offer a striking example of a mammalian neuronal system whose

in the adult

basic architecture and synaptic circuitry can be reversibly modified in adulthood. During parturition, lactation and prolonged osmotic stimulation, glial coverage of oxytociner~c neurons markedly d~inishes and their surfaces are left in extensive juxtaposition; concurrently, there is formation of new synapses, which are predominantly GABAergic and which couple two or more oxytocinergic neurons simultaneously. These structural changes do not permanently modify the anatomy of the system since upon cessation of stimulation, neuronal juxtapositions and shared synapses disappear, to reappear upon new stimulation. At present, we can only speculate about the cellular mechanisms and factors responsible for these reversible neuroanatomical changes. However, oxytocin itself appears to be of primary importance since it can induce similar anatomical changes when chronically infused into the third ventricle. Mo~holo~c~ Dehydration

Oxytocinergic plasticity Central oxytocin

neurons

Gha

Synapses

Pregnancy

Lactation

PARTURITION AND LACTATION: A NEURONAL-GLIAL DIALOGUE IN THE HYPOTHALAMUS

AS attested by the numerous chapters in this volume, the magnocellular neurons that secrete oxytocin and vasopressin undergo significant changes in their electrical, biosynthetic and secretory activities under conditions that stimulate release of their hormones. Recent observations have demonstrated, moreover, that oxytocin-secreting neurons undergo additional, morphological modifications upon activation (by suckling, vaginal distension or prolonged osmotic stimulation, for example), which alter their relationship to adjacent glial cells and their synaptic input, and therefore, their overall anatomy. These latter changes thus render the oxytocinergic system a striking example of a mammalian neuronal system that can become structurally modified in adulthood. In addition, they illustrate how a dynamic regulation of neuronal structure can be intimately related to normal neuronal function. Nevertheless, as for other adult neuronal systems in which similar plastic changes take place, we can still only speculate about the mechanisms responsible, just as we can only postulate that they have true functional consequences. The relatively easy anatomical accessibility of oxytocinergic neurons and the fact that we can manipulate their activity quite readily with a variety of experimental and physiological stimuli renders this particular neuronal system a very attractive model in which we can now address these questions.

Magnocellular oxytocinergic neurons, together with vasopressinergic cells, are grouped predomin~tly in two areas of the hypothalamus, the supraoptic (SON) and paraventricular (PVN) nuclei. Their cell bodies and dendrites are quite closely packed together in the magnocellular nuclei, but they usually remain separated one from the other by neuropile elements, and in particular, by fine astrocytic processes (Fig. 1A). As in other areas of the mammalian CNS, neuronal surfaces are seen only occasionally to be directly juxtaposed, with no intervening glial elements. In fact, the results of now numerous quantitative ultrastructural analyses have revealed that under basal conditions of neurohypophysial hormone release, neuronal surface juxtapositions in the magnocellular nuclei are not only rare but also limited in extent [18, 39, 40, 421 and occur equally between oxytocinergic and vasopressinergic elements [ 10, 40, 431. Thus, in the SON of normally hydrated male and virgin female rats, about 20% of all oxytocinergic or vasopressinergic cell bodies show direct juxtapositions to an adjacent neurosecretory cell body or dendrite but these neuronal juxtapositions are very small and involve no more than 2% of all the neuronal surface membrane in the nucleus (see Fig. 5). In contrast, glial coverage of magnocellular neurons

*Requests for reprints should be addressed to Dionysia T. Theodosis.

681

682

MONTAGNESE

ET AL.

STRUCTURAL

PLASTICITY

OF THE OXYTOCINERGIC

markedly diminishes in the nuclei of parturient and lactating animals (Fig. 1B) [18, 24, 39, 421, and almost every oxytocin-secreting neurone is seen directly juxtaposed, at one point or another of its surface, to an adjacent cell body and/or dendrite, which usually is also oxytocinergic (Figs. 2,9) [24,40,43]. These changes are quite specific to oxytocin neurons since plasmalemma juxtapositions between the neighboring vasopressin neurones remain few and no different in size and frequency to those seen in the nuclei of unstimulated animals (Fig. 5). To date, most quantitative analyses have been carried out on the SON, because of its cellular homogeneity, and in this nucleus, the majority of directly juxtaposed oxytocinergic elements were found in the dorsal confines of the nucleus, corroborating the well-known dorsal preponderance of oxytocin cells in this nucleus (see [35,43]). However, ultrastructural analyses on identified neurosecretory elements (by morphological and/or immunocytochemical means) clearly show that oxytocinergic profiles exist everywhere in the SON, including the ventral glial lamina which contains many oxytocinergic dendritic profiles [ 10,431. What must be kept in mind, especially in terms of functional consequences, is that one often finds clusters of 3-5 directly juxtaposed oxytocinergic neuron profiles and this may be found anywhere within the nucleus. Recently, we have also studied the occurrence of such changes in the PVN, using ultrastructural immunocytochemical methods to identify the neurons. Our morphometrical analysis has indicated that the more heterogeneous PVN also becomes structurally reorganised at lactation and, as in the SON, the changes affect exclusively its magnocellular oxytocinergic system. Individual plasmalemma juxtapositions between oxytocinergic cell bodies and/or dendrites in stimulated animals can be quite extensive and, altogether, can involve more than 10% of all the oxytocinergic surface membrane in the SON (Fig. 5) [43]. Nevertheless, it is important to note that, at the level of each juxtaposition, the extracellular space between the directly juxtaposed neuronal surfaces does not diminish but remains similar in size to that seen between the other cellular elements in the nucleus (11-13 nm). The only visible plasmalemma specializations are punctae adhaerentes or attachment plates, a type of adhesive cell junction present in all nervous tissue. The appearance of attachment plates highlights further the plasticity of the system since their incidence increases and decreases significantly as the amount of juxtaposed neuronal surface increases and decreases with changing secretory conditions [IO, 39, 42, 431. On the other hand, there are now several reports of dye coupling [2,1 I] and even of gap junctions [2], the morphological correlates of electronic synapses in the CNS (see [6,36]), between magnocellular neurons. However, we have never detected any gap junctions between any neurosecretory element within the magnocellular nuclei, by conventional or freeze fracture electron microscopy [ 10,24,39, 42-441. We can indeed visualize gap junctions in these areas of the hypothalamus, but they are seen to couple glial processes. This is not to say that gap junctions do not exist within the SON or PVN but if they do, they are extremely rare.

SYSTEM

683

Another possibility is that they may occur outside the magnocellular nuclei themselves, coupling neurosecretory processes that course in the hypothalamic regions outside the nuclei proper. A striking characteristic of the neuronal-glial changes that lead to the formation of the neuronal plasmalemma juxtapositions is that they do not permanently modify the anatomy of the magnocellular nuclei: once stimulation ceases, glial processes reappear between oxytocinergic neurons (Fig. 3). The disappearance of glial processes appears to take place relatively quickly. For example, within a few hours prior to parturition, there are as many directly juxtaposed oxytocinergic neurons in the SON as after 10 days of lactation [24]. On the other hand, once stimulation ceases, the reappearance of glial processes between the neurons appears to depend on the length of time the system has remained stimulated. In rats that have undergone two consecutive pregnancies and lactations, the incidence of plasmalemma juxtapositions returns to ‘virgin’ values only two months after weaning, whereas one month is sufftcient after a first lactation ([39] and Fig. 3). On the other hand, if the young are taken away immediately after birth, glial reinsertion takes place as early as two days post pat-turn [24]. Conversely, one can maintain the lack of glial coverage by maintaining sttmulation. Rats are usually weaned after 2 l-22 days of lactation but a perfectly normal lactation can be maintained for much longer periods of time, by providing the mothers with new litters. Such a prolonged lactation (for over a month, for example) is accompanied by the same anatomical reorganisation of the oxytocinergic system as that seen during a normal lactation [24]. Interestingly, the proportion of directly juxtaposed elements is the same after one or 50 days of lactation. This is not too surprising when one considers that close to 90% of all oxytocin-secreting profiles (in random stereological analyses) are juxtaposed to another neuronal element within a few hours of parturition [43]. It remains to be seen whether the amount of neuronal surface membrane involved in the juxtapositions is the same after a short or prolonged period of stimulation. PARTURITION

AND LACTATION: SYNAPTIC THE HYPOTHALAMUS

REMODELLING

OF

As described in many electrophysiological and anatomical studies (see, for example, [20,28,41]), magnocellular cell bodies and dendrites receive a rich synaptic input. In animals in which hormone demands are minimal, most terminals impinge onto one single postsynaptic element in any plane of section examined. In the magnocellular nuclei of parturient and lactating animals, however, we often see a different synaptic configuration, namely, axon terminals that synapse onto 2 or even 3 postsynaptic elements, in the same plane of section [18, 39, 421. In the main, the postsynaptic elements are oxytocinergic (Figs. 2,9) [24, 40, 41, 431. In the SON of lactating rats, for example, such shared synapses couple over 20% of all oxytocinergic cell bodies (relative to less than 3% of all vasopressinergic cells) (Fig. 6). This amplification of synaptic input coincides with the neuronalglial changes and it is not uncommon to see a cluster of

FACING PAGE FIG. 1. Electron micrographs of magnocellular neurons in the supraoptic nucleus (SON) of virgin (A) and lactating (B) rats. In A, 2 neurosecretory soma profiles are seen to occur quite close together but they remain separated by an intervening astrocyte and its process. Note glial filaments (arrowheads) in the astrocyte cytoplasm. B also shows 2 neurosecretory soma profiles close together but now their plasmalemmae are directly juxtaposed (between arrows), without a glial or neuropile element between them. sg, neurosecretory granules.

684

MONTAGNESE

ET AL.

FIG. 2. Low power electron micrograph of magnocellular neurons in the SON of a parturient rat. In this example, 2 neurosecretory soma profiles (underlined in black) are identified as oxytocin-secreting by colloidal gold particles over their neurosecretory granules (g, insets on the left), after postembedding immunocytochemical staining with antioxytocin serum and immunoglobulin G-colloidal gold (IgG-gold). Note that substantial portions of the surface membranes of the contiguous neurons are directly juxtaposed (between arrows). The 2 somata are also contacted synaptically by the same presynaptic terminal (asterisk and inset at lower right).

oxytocinergic profiles whose surface membranes are in extensive juxtaposition also sharing one or more synaptic terminal. The shared synapses are very plastic, since they disappear once the system is no longer stimulated (Fig. 4) and reappear upon new stimulation. Like the plasmalemma juxtapositions, the rapidity with which they disappear depends on how long the system had remained stimulated: after a normal lactation, at least one month is necessary for their incidence to return to virgin values [39]; if the young are

removed immediately at birth, they are no longer visible within 10 days [24]. On the other hand, prolonging stimulation does not increase their numbers further: the proportion of oxytocinergic profiles sharing a common presynaptic terminal is the same after 50 days of lactation as it is after parturition [24]. Finally, preliminary observations have recently indicated that the synaptic changes noted within the magnocellular nuclei may be even more pronounced than first suspected. It would appear that there is not only a change in the incidence

STRUCTURAL

PLASTICITY

OF THE OXYTOCINERGIC

SYSTEM

50%

Virgin

Ed gestation 21d gemion lad lacm.ion

FIG. 3. Evolution of neuronal plasm~emma juxtapositions in the SON of rats at different reproductive stages. The proportion of neurosecretory soma profiles directly juxtaposed to another soma or dendritic profile (ordinate) increases suddenly at the end of gestation (21-day gestation) and remains elevated throughout lactation. After weaning, this proportion reverts back to its original ‘virgin’ value. Note that the same neuronal-glial changes take place during a second gestation-lactation period (diparous). The q~ntitative analyses represented here, as those depicted in Fig. 4, were performed on tissue in which the peptidergic content of the neurosecretory profiles was not determined. Thus, close to 40% of any neurosecretory soma profile in this nucleus is directly juxtaposed to another neuronal element at parturition and lactation (Adapted from [40]).

proportion of soma profiles contacted by a shared synapse

virgin gestation

18d

gestation

19d

parturition lactation post-weaning

FIG. 4. Evolution of shared synapses in the SON at various stages of reproduction. The graph illustrates the sudden increase in the proportion of neurosecretory soma profiles contacted synaptically by the same axonal terminal at p~u~tion and during lactation; this proportion decreases after weaning (Adapted from 1391).

685

686

MONTAGNESEET

AL.

LACTATION

OT cells I t

iz .‘5 20. z

B

6

g fj

4

OT+VP ceils

.+ 0

control t.E

10 v1

4

Lmating

111 C0ntvJl

w%

OT-Cells

LE

LaGwng w

S

0

dJ Virgin LE

Lactating Br 01

Br

VP-Cells

OT OT

LE

w

6r

LE

w

Br

LE

ceils **I

untreated Br 01

FIG. 5. Specificity of the neurons-glial changes to the oxytocinergic system. The magnocellular neurons were identified with the electron microscope by morphological means [in Brattleboro homozygote (Br DI) rats] or immunocytochemical criteria [in normal Long-Evans (LE). Wistar (W), and Brattleboro homozygote (Br) rats, with antisera against oxytocin and vasopressin] (for further details, see 110,431). Similar changes are seen during lactation and prolonged dehydration. Dehydration was induced experimentally in the W&tar rats by giving them hypertonic saline to drink; Brattleboro homozygotes suffer genetically from diabetes insipidus and are in a constant state of dehydration, even with water provided ad lib, unless they are treated with vasopressin. The histograms on the left represent the proportion of neurosecretory soma profiles directly juxtaposed to an adjacent soma or dendritic profile while those on the right show the proportion of neurosecretory plasmalemma involved in these juxtapositions. Note that only an increased number of oxytocinergic neurons (OT-cells) become juxtaposed during both stimuli and that this is accompanied by a large increase in the extent of juxtaposed oxytocinergic surface membrane. Vasopressinergic neurons (VP-cells) do not become more juxtaposed upon stimulation and the extent of vasopressinerg~c surface in dire& juxtaposition with another neuronal element is low and no different from that measured under basal conditions of hormone release. For Br Dl rats, VP refers to the abnormal magnocellular cells unable to synthesize vasopressin (Adapted from [ 10,431).

of shared synapses with changing stimulation, but also a significant increase in the density of all terminals synapsing onto neurosecretory profiles, at least within the SON [9]. THE STRUCTURAL CHANGES ARE SPEClPIC TO OXYTO~t~ERGlC NEURONS BUT NOT TO PARTURlTION AND LACTATION

The neuronal-giial

and synaptic changes, which are par-

ticu~arIy s~king during parturition and lactation, occur, in fact, every time the oxytocinergic system is under strong and/or prolonged stimulation, as for example, during prolonged osmotic challenge, Dehydration is usually considered a stimulus for vasopressin secretion, but both oxytocinergic and vasopressinergic systems are in fact activated, especially after long periods

STRUCTURAL

PLASTICITY

OF THE OXYTOCINERGIC

SYSTEM

proportion of soma profiles contacted by a shared synapse

virgin

lactating

Br W

dehydrated Br icv oxytocin

FIG. 6. Specificity of the synaptic changes to the oxytocinergic system. In virgin rats, there are very few shared synapses in the SON and they contact about the same proportion of oxytocin (OT) and vasopressin (VP) neurons. Under stimulated conditions (lactation or prolonged dehydration, as

in Fig. 5), the increase in the incidence of shared synapses affects exclusively those contacting oxytocinergic neurons. Note that chronic (8-day) intracerebroventricular infusion of oxytocin (ICV oxytocin) reproduces the same anatomical changes as those seen during lactation and dehydration (see also Fig. 7). As in the analyses depicted in Fig. 5, the neurons were identified ultrastructurally by morphological homozygote rats.

or immunocytochemical

of such stimulation (see [28]). Although 24 hr of water deprivation does not significantly alter the anatomy of the magnocellular nuclei [42], prolonged dehydration, induced experimentally by chronic drinking of hypertonic saline [ 10,451 or genetically, in Brattleboro homozygote rats suffering from diabetes insipidus [9,10], induces the same structural reorganisation as that seen during lactation. Surprisingly, as during lactation and parturition, the morphological changes affect exclusively the oxytocinergic system (Figs. 5,6) [lo]. Upon cessation of stimulation, that is, upon rehydration of the animals, neuronal juxtapositions and shared synapses again diminish in frequency [ 10,451, as they do after weaning [24,39]. Similar changes have also been noted in the oxytocinergic system of nonpregnant, primiparous rats that were continuously exposed to suckling litters. In these animals, suckling quickly leads to pseudogestation and, within 18-21 days, to a quite normal lactation. Neuronal-glial and synaptic changes were evident already after 12 days of pseudogestation, and by the first day of such suckling-induced lactation, the oxytocinergic system was as reorganised as on the first day of a normal, post partum lactation [24]. THE ROLE OF HORMONAL FACTORS IN THE STRUCTURAL CHANGES

Since the morphological

changes

in the magnocellular

criteria.

W, normal

Wistar rats; Br, Brattleboro

nuclei involve exclusively oxytocinergic neurons and since they consistently occur whenever the oxytocinergic system is activated, an obvious hormonal factor that may be involved in their genesis is oxytocin itself. In order to test this hypothesis, we recently infused oxytocin, or its analogues, into the third ventricle of normally hydrated, nongestating, nonlactating female rats for a period of one week and found that oxytocin did indeed engender the structural reorganisation characteristic of the oxytocin system when it is physiologically activated (Figs. 6,7) [44]. The morphological consequences were quite specific to oxytocin since infusions of the closely similar neuropeptide, vasopressin, left the magnocellular nuclei basically unmodified. These observations thus offered a striking demonstration that a centrally circulating neuropeptide can induce anatomical changes in an adult neuronal center. They suggested, moreover, that oxytocin can regulate its own peripheral release by contributing to the dramatic restructuring of the nuclei containing the neurons responsible for its secretion. If our experimental intracerebroventricular infusions of oxytocin indeed mimic the presence of oxytocin normally present in the hypothalamus, then the question arises as to where this neuropeptide comes from under physiological conditions. A peripheral origin is highly unlikely since oxytocin appears incapable of passing through the bloodbrain barrier. On the other hand, oxytocin is present in the

MONTAGNESE

688

*

CSF

AVP

OT

4-T-OT

FIG. 7. Anatomical changes in the SON in response to central administration of oxytocin. Wnstimulated female rats (normally hydrated, nonpregnant, nonlactating) received a continuous infusion of either oxytocin (OT), ethreonine oxytocin (4-T-OT), vasopressin (AVP) or artificial cerebrospinal fluid (CSF) within the third ventricle during 8 days, but means of miniosmotic pumps delivering fluid at a rate of 0.5 plihr. The pumps contained 2 @g/ml of peptide dissolved in CSF. Oxytocin or its analogue, 4-T-OT, induced a significant increase in the proportion of neurosecretory profiles in juxtaposition. The numbers represent the total number of neurosecretory profiles examined. In the group which received oxytocin, a further analysis (inset), carried out on immunocytochemically identified profiles (postembedding procedures with IgG-gold, see [43,44]), revealed that only oxytocin cells were involved in this reorganisation. See Fig. 6 for the effect of central oxytocin on shared synapses in the SON.

cerebrospinai fluid, even under normal conditions [34] and may be released from oxytocin-containing neurones located in many areas of the central nervous system [35]. In addition, oxytocin appears to be released within the magnoceliular nuclei themselves [23,25]. By using push-pull cannulas inserted into the SON of suckled rats, we ourselves have recently shown that the neuropeptide is specifically released inside the nucleus during milk ejection (see also chapter in this volume). Moreover, by means of ult~structural immunocytochemistry, we have been able to visualize oxytocinergic synapses within the SON (Fig. 8) [38]. Most significant is that such synapses contact neurons which themselves are oxytocinergic. In addition to oxytocin, parturition and suckling-induced or post partum lactation are characterized by increased plasma levels of ovarian and placental steroids and these hormones may also take part in the restructuring of the nuclei. Magnocellular neurons possess steroid receptors 227, 30, 32, 331, and increased levels of estrogens and progesterone can modulate their electrical activity [ 1,261. Observations from our own laboratory indicate that steroid hormones probably do take part in the restructuring of the magnocellular nuclei: in recent experiments where we examined the effects of oxytocin chronically infused into the third

ET AL.

ventricle, we noted that the centrally administered neuropeptide was capable of altering the anatomy of its own system only in animals whose cycles were arrested in diestrus. Nevertheless, it can be argued that sexual steroids may only play a minor role since during dehydration, the nucleus is as reorganised as during lactation; moreover, prepartum ovariectomy does not prevent the structural changes from taking place. Dehydration, however, is associated with an increased secretion of corticosteroids [ 13,161, as is lactation in ovariectomized rats [22], which suggests that corticosteroids may take over the function of the sexual steroids. That steroid hormones may participate in restructuring of nervous tissue is not too far-fetched when one considers that steroid-accumulating neurones exist in a number of areas of the CNS, and that the hormones have been implicated in sexual dimo~hism of certain nervous structures (see [3]), in dendritic growth as well as formation of new synapses (see [3,7]) and in compensatory collateral sprouting after lesion (see [4]). By analogy, then, it can be postulated that steroids may affect the anatomy of the oxytocinergic system by acting directly on the morphology of oxytocinergic neurones, their surrounding glia and/or their impinging synapses. Alternatively, steroids may act secondarily, by changing other parameters, such as the electrical activity of ma~ocellular neurones [ 1,261, which in turn would take part in the anatomical restructuring. Although it is tempting to speculate that elevated levels of steroid hormones may play a role in the structural plasticity of the SON, it is obvious that, in themselves, they cannot be sufficient, since the nuclei remain unmodi~ed for quite long periods of time when their plasma levels are significantly enhanced, as for example, during pseudogestation and gestation [24,39]. A more likely hypothesis is that centrally circulating steroids, or their functional consequences, act in synergy with central oxytocin to promote the plastic changes. CELLULAR MECHANISMS Just as we are only beginning to have some clues as to the overall factors that may take part in the structural reorganisation of the magnocellular nuclei, we are also just starting to understand the intimate cellular mechanisms involved. From our ultrast~ctural observations it appears very likely that both neurones and glial cells are actively involved in the neuronal-glial changes. The active participation of the neurones is obvious. The neurones hypertrophy under stimulation, and it is not unlikely that the general morphology and geometry of their cell bodies and dendrites are also signiflcantly altered, as has been noted in other adult neuronal systems under stimulation (see [29]). Such mo~holo~cal changes may have important consequences, not only in modifying the relation of the neurones to their adjacent glial cells, in changing their synaptic connectivity, but even in modifying their basic electrical properties. That the neurones take part in the restructuring of their own system is also evident from the fact that they produce and destroy attachment plates in concert with the mount of their surface in juxtaposition; they also take part in the concurrent synaptic remodelling. The participation of astrocytes in these changes is less obvious. It could have been imagined that glial processes are squeezed away when the neurones hypertrophy. This is certainly not the case since both types of neurone hypertrophy when chronically stimulated, as for example during de-

STRUCTURAL

PLASTICITY

OF THE OXYTOCINERGIC

689

SYSTEM

FIG. 8. Oxytocin synapses on oxytocinergic neurons in the SON. In (A), after preembedding immunocytochemistry using antioxytocin serum and peroxidase-anti-peroxidase (PAP), the electron dense peroxidase immunoprecipitate has filled an axonal terminal and secretory vesicles in the soma it is contacting, thus identifying both pre- and postsynaptic elements as oxytocinergic (OXY). In (B), after postembedding staining with antioxytocin and IgG-gold, electron dense colloidal gold particles have covered dense-cored vesicles in the presynaptic terminal and in its postsynaptic soma, thus identifying both as oxytocinergic. Note that in (B) the procedure has permitted to show that such OXY-immunoreactive synapses also contain numerous small clear synaptic vesicles. Similar synaptic profiles are also immunoreactive for oxytocin-related neurophysin (for further details, see [38]).

hydration, and glial coverage of both oxytocinergic and vasopressinergic cells actually increases [lo]. What is peculiar to oxytocinergic cells is that the proportion of their plasmalemma covered by glial elements diminishes after stimulation [10,43]. Neuronal juxtapositions result not so much from active glial retraction, therefore, but, rather, from an insufficient glial coverage. Another question raised by such observations is why should the glial environment become modified only around oxytocinergic neurones? One possibility is that glial cells react to varying concentrations of extracellular K+ consequent to changing neuronal activity (see [12]). Again, the specificity of the changes argues against such a possibility: although the electrical activity of both oxytocin and vasopressin cells is enhanced under chronic dehydration (see [28]), it is only oxytocinergic cells that are left with less glial coverage [lo]. A more specific signal must therefore be responsible. We can also only be highly speculative as to the mechanisms responsible for the accompanying synaptic remodelling. There are now numerous examples of synaptic plasticity in the adult mammalian brain but little is known of the cellular processes involved (see [14]). From many of these studies, it is clear that synaptogenesis involves the active

participation of both pre- and postsynaptic partners. Moreover, it appears that the rate of synapse formation may depend on the intensity of stimulation (see [14, 17, 311). Within the activated oxytocinergic system, then, an obvious possibility is that shared synapses are formed as extensions of preexistent terminals, which are continuously hyperactive, and which have thus induced an adjacent postsynaptic element (cell body and/or dendrite) to create a postsynaptic density (see also [31]). CONCLUSIONS

The morphological reorganization of the oxytocinergic system described in this review may be considered important not only in itself, as an example of how the adult CNS can be modified structurally in response to varying physiological conditions, but also because the plastic changes may have significant consequences to the subsequent functioning of the system. During parturition and lactation, oxytocinergic neurons display a quite spectacular electrical activity, consisting in periodic high frequency discharges which are synchronised in the whole oxytocinergic population in the hypothalamus [5,2 1,461. The nature of the anatomical changes and the fact

690

MONTAGNESE

ET AL.

FIG. 9. GABAergic synapses on oxytocinergic neurons in the SON. Preembedding immunostaining with anti-GABA and PAP, followed by postembedding staining with antioxytocin and IgG-gold were performed on the same section from the SON of a lactating rat. Such multiple immunostaining permits showing, simultaneously, an oxytocinergic soma (OXY) (note colloidal gold particles in its secretory granules, sg, also shown at higher magnification in inset, top left) contacted synaptically by a GABAergic terminal (asterisk) (shown at higher magnification in inset, top right). Note that the GABAergic terminal also synapses onto an adjacent dendrite (d), and that the surface membranes of the two postsynaptic structures are in direct juxtaposition. (From ref. [41], with permission).

that they are always associated with these physiological conditions make it very likely that they do play a role in the neuronal activation. The possible impact of shared synapses on coordination of cell tiring appears evident since they allow several postsynaptic elements to receive the same information simultaneously. GABA is a major inhibitory transmitter in the magnocellular nuclei (see [41]) and we recently showed that the majority of shared synapses in the rat SON are GABAergic (Fig. 9) [41]. It may be, then, that an increased inhibitory input to oxytocin neurones serves to coordinate their accelerated fiing by permitting activity to occur only at particular intervals, as is the case during the milk ejection reflex induced by suckling [S, 21,461 or the expulsion of each fetus at parturition [37]. On the other hand, some shared synapses are dopaminergic [8] and this would imply a shared excitatory input (see [28]), which could further facilitate neuronal coordination.

Extensive neuronal surface juxtapositions may also contribute to synchronisation of firing by leading to electrical field interactions and/or changes in the concentration of extracellular ions, particularly K+. Electrical field effects imply increased flow of electrical current through the extracellular space without the participation of any specialized connections (see [6,15]). The electrical activity of a group of neurons firing simultaneously, as is the case of oxytocin neurons during parturition and lactation, could thus modify the excitability of their neighbours and even recruit them into activity if the latter are close to activation threshold. Neuronal fling is also accompanied by important movements of water and ions across neuronal membranes, which can modify not only the composition of the extracellular fluid but also the size and resistance of the extracellular space (see [ 121). Because of the membrane properties of glial cells, and because of their position between neurons, they are usually considered essential to the maintenance of the homeostasis

STRUCTURAL

PLASTICITY

OF THE OXYTOCINERGIC

of the extracellular space. Lack of glial elements would then result in accumulation of ions, particularly K+, which could considerably modify the excitability of the extensively juxtaposed oxytocinergic somata and dendrites in the hypothalamus. At present, we have no evidence that such a phenomenon occurs in the magnocellular nuclei but it may not be that unlikely in view of recent findings that high frequency discharges during suckling do lead to a transient but significant increase in extracellular K+ at the level of the neurohypophysial terminals [ 191. Finally, what we must stress is that in no case can a shared synaptic input or ephaptic and/or effects be held completely responsible for the synchronisation of neuronal

691

SYSTEM

firing in the oxytocinergic system. The neurons that exhibit synchronous firing are located in four separate nuclei in the hypothalamus [5]. What we suggest is that neuronal juxtapositions and shared synapses may constitute additional facilitatory mechanisms, within each magnocellular nucleus which would act together with other mechanisms, more directly responsible for synchronisation, such as the afferent drive involved in each particular neuroendocrine reflex. ACKNOWLEDGEMENTS

We thank Mme. R. Bonhomme, Mr. F. Rodriguez and Mr. S. Senon for their constant support and excellent technical assistance.

REFERENCES 1. Akaishi, T. and Y. Sakuma. Estrogen excites oxytocinergic but not vasopressinergic cells in the paraventricular nucleus of female rat hypothalamus. Brain Res 335: 302-305, 1985. 2. Andrew, R. D., B. A. McVicar, F. E. Dudek and G. I. Hatton. Dye transfer through gap junctions between neuroendocrine cells of rat hypothalamus. Science 211: 1187-1189, 1980. 3. Arnold, A. P. and R. A. Gorski. Gonadal steroid induction of structural sex differences in the central nervous system. Annu Rev Neurosci 7: 413-442, 1984. 4. Azmitia, E. C. and F. C. Zhou. Chemically induced homotypic collateral sprouting of hippocampal serotoninergic afferents. Exp Bruin Rrs 13: Suppl, 129-141, 1986. 5. Belin, V. and F. Moos. Paired recordings from supraoptic and paraventricular oxytocin cells in suckled rats: recruitment and synchronization. J Physiol (Land) 377: 369-390, 1986. 6. Bennet, M. V. L. Electrical transmission: a functional analysis and comparison with chemical transmission. In: Cellular Biology of Neurons: Handbook of Physiology, vol 1, edited by E. R. Kandel. Baltimore: Williams and Wilkins, 1977, pp. 357-416. 7. Bottjer, S. W. and A. P. Arnold. Hormones and structural plasticity in the adult brain. Trends Neurosci 7: 161171, 1984. 8. Buijs, R. M., M. Geffard, C. W. Pool and E. D. M. Hoomeman. The dopaminergic innervation of the supraoptic and paraventricular nucleus. A light and electron microspinal study. Brain Res 323: 65-72, 1984. 9. Chapman, D. B., H. W. Sokol, J. F. Morris and D. T. Theodosis. Control of magnocellular neurosecretory neurones: analysis of synapses and cell-to-cell contacts in normal rats and in rats with hereditary diabetes insipidus. JAnat 139: 734, 1984. 10. Chapman, D. B., D. T. Theodosis, C. Montagnese, D. A. Poulain and J. F. Morris. Osmotic stimulation causes ultrastructural plasticity of the neurone-glia relationships of the oxytocin- but not vasopressin-secreting neurones in the hypothalamic supraoptic nucleus. Neuroscience 17: 67%686, 1986. 11. Cobbett, P. and G. I. Hatton. Dye coupling in hypothalamic slices: dependence on in vivo hydration state and osmolality of incubation medium. J Neurosci 4: 3034-3038, 1984. 12. Coles, J. A. Homeostasis of extracellular fluid in retinas of invertebrates and vertebrates. Prog Sensory Physiol 6: 105-138, 1986. 13. Coover, G. D., .I. P. Heybach, J. Lenz and J. F. Miller. Corticosterone “basal levels” and response to ether anesthesia in rats on a water deprivation regimen. Physiol Behav 22: 653-656, 1979. 14. Cotman, C. W. and M. Nieto-Sampedro. Cell biology of synaptic plasticity. Science 225: 1287-1294, 1984. 15. Dudek, F. E., R. W. Snow and C. P. Taylor. Role of electrical interactions in synchronization of epileptiform bursts. In: Advances in Neurology, Vol44, edited by A. V. Delgado-Escueta, A. A. Ward, Jr., D. M. Woodbury and R. J. Porter. New York: Raven Press, 1986, pp. 593-617.

16. Ezzarani, E.-A., J.-P. Laulin and R. Brudieux. Effets de la privation d’eau sur la production de corticosterone chez le rat m&le Brattleboro gCnCtiquement privC de vasopressine. J Physiol (Paris) 80: 157-162, 1985. 17. Greenbugh, W. T., R. W. West and T. J. De Voogt. Subsynaptic plate perforations: changes with age and experience in the rat. Science 202: 1096-1098, 1979. 18. Hatton, G. I. and C. D. Tweedle. Magnocellular neuropeptidergic neurons in the hypothalamus: increases in membrane apposition and number of specialized synapses from pregnancy to lactation. Brain Res Bull 8: 197-204. 1982. 19. Leng, G. and K. Shibuki. Extracell& potassium changes in the rat neurohypophysis during activation of the magnocellular neurosecretory system. J Physiol (Lond) 392: 97-l 11, 1987. 20. LerBnth, C., L. Zhborsky, J. Marton and M. Palkovits. Quantitative studies on the supraoptic nucleus in the rat. I. Synaptic organization. Exp Brain Res 22: 509-523, 197.5. 21. Lincoln, D. W. and J. B. Wakerley. Factors governing the periodic activation of supraoptic and paraventricular neurosecretory cells during suckling in the rat. J Physiol (Land) 250: 443-461, 1975. 22. MacDonald, G. J. and D. W. Matt. Adrenal and placental steroid secretion during pregnancy in the rat. Endocrinology 114: 2068-2073, 1984. 23. Mason, W. T., G. I. Hatton, Y. W. Ho, C. Chapman and I. C. A. F. Robinson. Central release of oxytocin, vasopressin and neurophysin by magnocellular neurones depolarization: evidence in slices of guinea pig and rat hypothalamus. Neuroendocrinology 42: 3 1l-322, 1986. 24. Montagnese, C., D. A. Poulain, J. D. Vincent and D. T. Theodosis. Structural plasticity in the rat supraoptic nucleus during gestation, post-partum lactation and suckling-induced pseudogestation and lactation. J Endocrinol 115: 97-105, 1987. 25. Moos, F., M. J. Freund-Mercier, Y. Guem& J. M. GuemC, M. E. Stoeckel and P. Richard. Release of oxytocin and vasopressin by magnocellular nuclei in vitro: specific facilitatory effect of oxytocin on its own release. J Endocrinol 102: 63-72, 1984. 26. Negoro, H. S. Vissussewan and R. C. Holland. Unit activity in the paraventricular nucleus of female rats at different stages of the reproductive cycle and after ovariectomy with or without estrogen or progesterone treatment. J Endocrinol 59: 545-558, 1973. 27. Parsons,

B., T. C. Rainbow, N. J. MacLusky and B. S. MacEwen. Progestin receptor levels in rat bvpothalamic and limbic nuclei. JNeurosci i: 14461452, 1982. __ 28. Poulain, D. A. and J. B. Wakerlev. Electronhvsioloev of hvoothalamic magnocellular neurones-secreting-oiytocinand viiopressin. Neuroscience 7: 773-808, 1982. 29. Purves, D., and J. T. Voyvodic. Imaging mammalian nerve cells and their connections over time in living animals. Trends Neurosci 10: 398-404, 1987.

692

30. Rainbow, T. C., B. Parsons, N. J. MacLusky and B. S. MacEwen. Estradioi receptor levels in rat hypothalamic and limbic nuclei. 3 Neurosci 2: 1439-1445, 1982. 3 1. Raisman, G. and P. M. Field. A quantitative investigation of the development of collateral reinnervation after partial deafferentation of the septal nuclei. Brain Res 50: 241-264, 1973. 32. Rhodes, C. H., J. I. Morrell and D. W. Pfaff. Distribution of estrogen-concentrating, neurophysin-containing magnocellular neurons in the rat hypothalamus as demonstrated by a technique combining steroid autoradiography and immunohistoIogy in the same tissue. ~=~~~)end~)~~~n~~~~y 33: 18-23, 1981. 33. Rhodes. C. H.. J. I. Morrell and D. W. Pfaff. Estrogenconcentrating neurophysin-containing hypothalamic magnocellular neurons in the vasopressin-deficient (Brattleboro) rat: a and imcombining steroid autoradiography study munocytochemistry. J Nrurnsci 2: 1718-1724, 1982. 34. Robinson, I. C. A. F. and P. M. Jones. Plasma and CSF oxytotin during suckling in the guinea-pig. Nr,uroend~~~rinolo~~ 34: 59-63, 1982. 35. Sofroniew, M. V. Vasopressin, oxytocin and their related neurophysin. In: Handbook ofChemical Neuroanatomy, Vol4. CABA and Neuropeptidrs in the CNS, Part I. Chapter 111. edited by A. Bjorklund and T. Hokfeld. New York: Elsevier Science Publishers, 1985, pp. 93-165. 36. Sotelo, C. and H. Kom. Morphological correlates of electrical and other interactions through low-resistance pathways between neurones of the vertebrate central nervous system. Int Rev Cytol 55: 67-107, 1978. 37. Summerlee, A. J. S. Extracellular recordings from oxytocin neurones during the expulsive phase of birth in unanaesthetized rats. J Physiol (Land) 321: l-9, 1981.

MONTAGNESE

ET AL.

38. Theodosis, D. T. Oxytocin-immunoreactive terminals synapse on oxytocin neurones in the supraoptic nucleus. Nature 313: 682-684, 1985. 39. Theodosis, D. T. and D. A. Poulain. Evidence for structural plasticity in the supraoptic nucleus of the rat hypothalamus in relation to gestation and lactation. Neuroscience 11: 183-193, 1984. 40. Theodosis, D. T. and D. A. Poulain. Oxytocin-secreting neurones: a physiological model for structural plasticity in the adult m~malian brain. Trends ~~ur~~s~i 10: 4X-430, 1987. 41. Theodosis, D. T., L. Paut and M. L. Tappaz. Immunocytochemical analysis of the GABAergic innervation of the oxytocin and vasopressin-secreting neurones in the rat supraoptic nucleus, Nruroscirnw 19: 207-222, 1986. 42. Theodosis, D. T., D. A. Poulain and J. D. Vincent. Possible morphological bases for synchronization of neuronal tiring in the rat supraoptic nucleus during Iactation. Ne~~~~~.s~~~fz[,e 6: 919-929, 1981. 43. Theodosis, D. T., D. B. Chapman, C. Montagnese, D. A. Poulain and J. F. Morris. Structural plasticity in the hypothalamic supraoptic nucleus at lactation affects oxytocin-, but not vasopressin-secreting neurones. Ner(roscient,e 17: 661-678, 1986. 44. Theodosis, D. T., C. Montagnese, F. Rodriguez, J. D. Vincent and D. A. Poulain. Oxytocin induces morphological plasticity in the adult hypoth~amo-neurohy~physi~ system. Nature 322: 738-740, 1986. 45. Tweedle, C. D. and G. I. Hatton. Synapse formation and disappearance in adult rat supraoptic during different hydration states. Bruin Rcs 30% 373-376, 1984. 46. Wakerley, J. B. and D. W. Lincoln. The milk-ejection reflex of the rat: a 20- to 40-fold acceleration in the tiring of paraventricular neurones during oxytocin release. I ~~dl~~~rj~~~i 57: 477-493. 1973.