Vasopressin and Oxytocin in the Central Nervous System

Linda Rinaman, Thomas G. Sherman, and Edward M. Stricker

The closely related peptides arginine vasopressin (AVP) and oxytocin (OT), found exclusively in mammals, were originally identified as hormones secreted from the neurohypophysis into the systemic circulation. The neurohypophyseal hormones are synthesized by separate populations of magnocellular neurons whose perikarya primarily occupy the supraoptic (SON) and paraventricular (PVN) nuclei of the hypothalamus, and whose axons terminate within the posterior lobe of the pituitary gland. A smaller number of neurons that synthesize AVP or OT and project to the neurohypophysis are located within the anterior commissural nucleus of the hypothalamus, and others are clustered within magnocellular accessory nuclei scattered between the PVN and SON.

AVP, the antidiuretic hormone, is secreted primarily under circumstances of dehydration. A rise in extracellular solute concentration is one of the most effective stimuli for AVP release, with as little as 2% elevation in plasma osmolality causing a two- to threefold increase in peripheral AVP levels. The kidneys are exquisitely sensitive to AVP, thus enabling adaptive renal water conservation during dehydration. AVP also is a vasoconstrictor agent (hence its name) and is secreted in relatively large amounts during hypovolemia and hypotension, in the absence of changes in plasma osmolality. Only such large amounts of AVP are sufficient to produce vasoconstriction; approximately 40 times more AVP is needed for a pressor response than for antidiuresis.

OT is structurally similar to AVP, presumably reflecting their common evolutionary derivation from the amphibian pituitary hormone, vasotocin (1) (Fig. 1). During labor, sensory signals from the uterus and birth canal stimulate neurohypophyseal OT neurons, and the OT released into the bloodstream binds to uterine OT receptors to facilitate parturition by contracting the smooth muscle. Pituitary secretion of OT also is enhanced by signals related to suckling in lactating mammals. OT receptors are located on myoepithelial cells that are concentrated around the milk ducts and alveoli in the mammary glands; when OT binds to these receptors, the myoepithelial cells contract and stored milk is ejected.

These two familiar functions of systemic OT obviously apply only to sexually mature females. However, a third potent stimulus for pituitary OT secretion, plasma hyperosmolality, has recently been identified in both male and female rats. The OT secretory response to hyperosmolality is adaptive because OT promotes natriuresis in rats. Indeed, OT receptors in renal tissue are just as sensitive to OT as a sodium-excreting hormone as renal AVP receptors are to AVP as a water-retaining hormone. However, there is no evidence that OT has a comparable natriuretic function in other species, and there is in fact evidence that it does not in humans.

This chapter does not further address issues related specifically to neurohypophyseal AVP and OT. Instead, we focus on the roles of these peptides as neurotransmitter or neuromodulatory agents within the central nervous system. The fact that AVP and OT are released from axon terminals within distinct areas of the brain and spinal cord did not gain widespread appreciation until the early 1980s, but during the last decade an impressive amount of research has been performed on the neuroanatomy, neurobiology, and the possible functions of centrally projecting AVP and OT neurons (that is, neurons containing immunocytochemically detectable AVP or OT). This chapter will summarize some of the most firmly established findings that have emerged from these investigations (see also General Overview of Neuropeptides and The Psychopharmacology of Sexual Behavior).

In recent years there has been a very rapid increase in the number of studies examining the structure and function of the AVP and OT genes, as well as studies on the transcriptional regulation of these genes as a function of neuronal secretory activity and physiological manipulation (8, 78). From this work it is clear that AVP and OT are synthesized as portions of large precursor proteins from which AVP and OT are subsequently cleaved. Other portions of the precursor proteins, the neurophysin polypeptides, are also present in the dendritic, somatic, and axonal cytoplasm of AVP and OT neurons and are secreted from their axon terminals. There are separate precursors for AVP and its associated neurophysin and for OT and its associated neurophysin, but these precursors are encoded by a pair of genes that share many similarities in structure and sequence. The high degree of DNA sequence identity and exon–intron distribution between the AVP and OT genes strongly suggests that these genes arose via a duplication event from an ancestral gene. It was of great interest, therefore, when it was determined that the human AVP and OT genes are linked on chromosome 20, separated by less than 12 kb of intervening DNA sequence (65) (see Fig. 2). This chromosomal domain undoubtedly represents the remnants of a gene duplication and inversion event that preceded the establishment of two genes with independent transcriptional regulation and cell-specific expression patterns, regulating distinct physiological systems. Despite occasional claims that neuropeptide hormones related to AVP or OT, such as vasotocin, may be expressed in mammals, careful examination of the human genome have failed to identify other genes with significant similarity to AVP and OT (41).

The results of molecular analyses of human and rat OT promoters have been described in detail (2, 41). Similarly detailed analyses of the AVP promoter have not yet been performed, however. In contrast, studies using transgenic animals to examine the AVP gene are ongoing in several laboratories (3, 28), whereas no similar studies on the OT gene have yet been reported.

Two novel mechanisms involving AVP and OT mRNAs have been observed recently in the hypothalamic–neurohypophyseal system. First, the mRNAs for AVP and OT are actively transported towards the posterior pituitary in the axons of magnocellular neurons (35). The functional significance of AVP and OT mRNA axonal transport is unclear, and whether mRNA transport also occurs in AVP and OT neurons with central axonal projections remains an important question. Second, AVP mRNA injected into the hypothalamus of homozygous Brattleboro rats (which cannot produce AVP due to a genetic defect) is specifically taken up and expressed by magnocellular neurons, resulting in partial amelioration of their diabetic insipidus phenotype (64). Other studies found that AVP mRNAs with short poly(A) tails are taken up more effectively and are transported both retrogradely and anterogradely (43). It is possible that these capabilities are unique to neurohypophyseally projecting AVP and OT neurons, but they also may apply to neurons with central axonal projections.

The large majority of AVP and OT neurons with central axonal projections occupy the caudal portion of the PVN in rodents and primates, including humans (68, 72). These neurons are called "parvocellular" because they are usually smaller than the magnocellular neurons that innervate the posterior lobe of the pituitary gland. Although parvocellular AVP and OT neurons that occupy the PVN do form intrahypothalamic projections, their axons terminate primarily outside of the hypothalamus (see Fig. 3). The density of their fibers within different termination areas varies considerably, from single isolated fibers in the frontal cortex to very dense innervation in the dorsal vagal complex in rats. The ratio of AVP to OT fibers also varies widely in different areas. Quantitative differences in the numbers of fibers present in certain brain areas have been noted in different species, including humans, but qualitatively the distribution of fibers appears to be quite similar. Limbic structures such as the lateral septum and the amygdala are particularly heavily innervated by AVP fibers, whereas OT fibers are predominant in the brainstem and spinal cord.

Not all centrally projecting AVP and OT neurons are located within the PVN. In rodents and in primates, including humans, immunoreactive neurons also occupy the medial posterior region of the bed nucleus of the stria terminalis (AVP and OT), the medial preoptic area (OT), the dorsomedial suprachiasmatic nucleus (AVP), the septal region (AVP), the medial amygdala (AVP), and the locus coeruleus (AVP; not observed in humans) (34, 61, 72). The central projections of many of these neurons have been identified. In rodents, AVP neurons in the suprachiasmatic nuclei innervate the subparaventricular zone and dorsomedial nucleus of the hypothalamus, the medial and lateral preoptic areas, and midline thalamic nuclei (82). AVP neurons in the bed nucleus of the stria terminalis innervate the lateral habenular nucleus, the lateral septum, the medial amygdala, and the periaqueductal central gray (15). AVP neurons in the medial amygdala project to the ventral hippocampus (11). OT immunoreactivity in the locus coeruleus, raphe nuclei, and periaqueductal gray remains unaffected after extensive ablation of the PVN in rats, suggesting that OT fibers in these areas may originate outside of the PVN (29).

Because different populations of AVP and OT neurons project to the neurohypophysis and to central brain areas, stimuli that elicit pituitary secretion of AVP and/or OT need not provoke the release of these peptides within the central nervous system, although they often do. Just as neurohypophyseal secretion of AVP and OT can occur independently of each other, central release of AVP and OT can occur independently of each other and also can differ from region to region. For example, suckling is well known to stimulate magnocellular OT neurons. PVN neurons that contain OT and project to the dorsal vagal complex apparently do not respond to suckling, but they do respond to dehydration, hemorrhage, and exogenous cholecystokinin (30, 39). OT neurons projecting to the amygdala do not respond to suckling or hemorrhage, but do respond to dehydration (39), which also stimulates both AVP and OT magnocellular neurons. These and other data illustrate the selective and differential afferent control of AVP and OT neurons that have specific central projections.

Little is known about the chemical coding of afferent pathways that might differentially stimulate AVP and OT neurons in the PVN. The afferent innervation of the PVN is exceedingly complex, with most inputs appearing to innervate more than one PVN region (67, 72). In many cases these connections have been confirmed by both anterograde and retrograde neural tract tracing and/or by electrophysiological recording. However, it seldom is known whether the neurons of some extrahypothalamic cell group synapse upon AVP or OT neurons, or both, or whether the postsynaptic targets are magnocellular or parvocellular. One exception is a recently identified projection arising from neurons in the caudal medulla. A small number of neurons in the nucleus of the solitary tract and in the region ventrolateral to it coexpress b-inhibin and somatostatin, and these neurons appear to preferentially innervate OT neurons in the PVN and SON (66). The SON targets obviously are magnocellular, but it is not known whether the PVN targets are magnocellular, parvocellular, or both. The b-inhibin/somatostatin projection is distinct from the well-documented, dense noradrenergic projection from the caudal medulla that innervates both magnocellular and parvocellular populations of AVP and OT neurons (67).

Under normal physiological conditions, AVP and OT are present within the cerebrospinal fluid (CSF) in concentrations slightly higher than those in plasma. The separate regulation of AVP and OT in the blood and CSF is sustained by the blood–brain barrier, which prevents appreciable amounts of these peptides from entering the CSF after their secretion into the systemic circulation; that which does enter is presumed to do so through circumventricular structures, where the blood–brain barrier is absent (12). This arrangement does not exclude a possible central contribution of AVP and OT from neurons with neurohypophyseal projections, because peptide exocytosis occurs not only from their terminal boutons in the posterior pituitary but also along their perikarya, axons, and dendrites within the hypothalamus (51).

It is generally presumed that AVP and OT released into brain extracellular fluid will readily diffuse into the CSF, and vice versa, although the extent to which this is true is uncertain. It has been suggested that AVP and OT also may be released directly into the CSF, because AVP- and OT-immunoreactive processes have been observed in the ependymal lining of the third ventricle (12). The purpose of OT in the CSF remains unclear, but AVP is believed to play an important role in the regulation of blood flow to the choroid plexus and the rate of CSF production (12, 44). Just as circulating catecholamines secreted from the adrenal glands supplement norepinephrine released from postganglionic sympathetic neurons, one could speculate that AVP and OT released into the CSF might supplement AVP and OT released into central synaptic clefts, thereby supporting the various central functions of these neuropeptides that are discussed below.

The available evidence suggests that the affinity and ligand selectivity of OT receptors in the brain do not differ from those in the mammary gland or uterus. Thus, central OT receptors bind both OT and AVP with high affinity (16, 73, 75). The reported maximal binding capacities for central OT receptors are relatively small (30–44 fmol/mg protein). The transduction mechanism triggered when OT binds to its receptors in the brain has not yet been ascertained, although central OT receptors appear not to be linked to either adenylate cyclase or to phosphoinositide breakdown (70).

AVP acts on three receptor subtypes: V1a, V1b, and V2. Only the V1a (vascular) subtype has been found in the central nervous system (6, 75), although some evidence for central V2-type (renal) receptors has been reported (16). The maximal binding capacities for central V1a receptors also are small (20–80 fmol/mg protein) (6). AVP receptors in the anterior pituitary are classified as V1b on the basis of different antagonist displacement profiles of [3H]AVP binding compared with classical V1a or V2 receptors (33). Activation of both V1a and V1b receptors results in phosphoinositide breakdown (70).

cDNAs encoding the human OT receptor (37) and the rat V1a receptor (50) were recently cloned. They are prototypic G-protein-coupled receptors, with seven putative transmembrane domains. They are more closely related to each other than to any other known G-protein-coupled receptors, perhaps accounting for the high affinity of AVP for OT receptors and (to a much lesser extent) vice versa. However, recent studies indicate that the central distribution of OT receptors is quite distinct from that of AVP receptors (42, 75). Whenever present in the same general area, AVP and OT binding sites appear to be located in different parts of the structure. For example, AVP binding sites in the rat hippocampus are located mainly in the inner and outer blade of the dentate gyrus and in a subregion of the CA1 field, whereas OT binding sites are located in the subiculum and in a different part of the CA1 field. AVP binding sites in the rat dorsal vagal complex are concentrated in the nucleus of the solitary tract and the area postrema, whereas OT binding sites are concentrated in the dorsal motor nucleus of the vagus. Collectively, such data provide a structural basis for observations that AVP and OT differentially influence centrally regulated brain functions (below).

Most of the neural target areas that contain AVP- and/or OT-immunoreactive axon terminals have been shown to also contain AVP and/or OT binding sites (putative receptors) in rodents, monkeys, and humans (42, 75), and the presence of AVP or OT binding sites is consistently associated with AVP or OT neuronal responsiveness (18, 19, 75, 77). Comparison of electrophysiological recordings obtained from guinea-pig brain and rat brain have shown that species differences in OT receptor binding correlate positively with differences in OT neuronal sensitivity, suggesting that binding sites detected in the brains of other species also represent functional receptors (18, 75). However, when immunocytochemical data are compared with receptor autoradiographic data, some discrepancies appear. For example, binding sites for OT are abundant in the suprachiasmatic nuclei in rats, where very few or no OT-immunoreactive axons appear to be present. It remains to be established whether these observations reflect the technical limits of receptor autoradiography or immunocytochemistry, or whether they reveal authentic peptide–receptor mismatches suggestive of central paracrine-like effects of AVP and OT following their release in remote locations.

AVP and OT agonists are categorized primarily on the basis of the type of receptors they bind to. As stated above, most studies indicate that central AVP receptors are of the V1a type, whereas central OT receptors appear pharmacologically identical to OT receptors in the uterus and mammary gland. The central binding characteristics and receptor-mediated effects of various analogues of AVP and OT are similar to their performance in peripheral tissues. Thus, agents that act as AVP agonists in vascular smooth muscle also mimic many of the central effects attributed to AVP (see below), and agents that act as agonists of OT in the uterine myometrium appear similarly effective on OT receptors within the central nervous system. However, because marked differences in rat, monkey, and human V1a peripheral receptors have been demonstrated (55), and species differences may also exist in central AVP (and OT) receptors, the published pharmacological data for AVP and OT analogues serve as a useful guide only in that species for which data are reported.

The ring structure of the AVP molecule is an absolute requirement to elicit agonist responses (73), and no known compound displays higher V1 agonist activity than AVP itself. Many centrally acting AVP agonists display selective binding to V1 rather than V2 receptors, but AVP analogues with high V1 agonist activity and reduced affinity for OT receptors have not been reported. Central OT receptors bind both AVP and OT with high affinity; they also bind the agonist HO[Thr4, Gly7]OT (TGOT) with high affinity (16). In fact, TGOT is more selective for OT receptors than is OT itself, and is effective in smaller doses than OT in certain behavioral tests in rats (52), although it has a lower efficacy than OT in causing neuronal excitation in vitro (74).

The known antagonists of AVP and OT fall into four classes: (a) related cyclic peptides, (b) linear peptides, (c) cyclic peptides of bacterial origin, and (d) nonpeptides. A potent cyclic antagonist of central V1a receptors is d(CH2)5Tyr(Me)AVP. It is the most widely used V1a antagonist, and it is 100-fold more potent as a V1a antagonist than as an OT antagonist in rats (45). Many linear V1a antagonists are similarly potent, but most have not yet been evaluated for their specificity with regard to OT receptors (45). Substantial gains in both anti-OT potency and selectivity have been made; one recently reported OT antagonist has an anti-OT/anti-V1a potency ratio of 17:1 (55).

The electrophysiological actions of AVP and OT on target neurons have been well studied in several brain areas, including hippocampus, septum, and dorsal vagal complex (75, 77). There is strong evidence supporting physiological roles for both peptides as neurotransmitter substances in the descending pathway from the PVN to the dorsal vagal complex in rats, which contains high-affinity receptors for AVP and OT (75). Electron microscopy has revealed that AVP- and OT-immunoreactive axon terminals synapse on dendrites within motor and sensory subnuclei of the dorsal vagal complex (81). Both peptides are released in the dorsal vagal complex in a Ca2+-dependent and K+-stimulated manner (9), and their concentrations increase there following electrical stimulation of the PVN (38). Microinjection of low concentrations of AVP (5 nM) or OT (100–500 fmol) into the rat dorsal vagal complex both in vitro and in vivo potently enhances the spontaneous activity of responsive neurons in a receptor-mediated, dose-related, reversible, postsynaptic, TTX-resistant, and voltage-dependent manner (18, 19, 47).

The onset of neuronal responses to AVP or OT in the dorsal vagal complex occurs between 15 sec and 2 min after injection, and lasts from 30 sec to 20 min (18, 19, 47). The relatively slow onset and long duration of neuronal responses to AVP and OT are consistent with the possibility that these neuropeptides produce their effects by activating a postsynaptic second messenger cascade. This possibility also is consistent with the G-protein-coupled nature of the central AVP and OT receptors (37, 50) (for detailed discussion of receptor structure–function relationships, see Cholinergic Transduction, Signal Transduction Pathways for Catecholamine Receptors, Serotonin Receptors: Signal Transduction Pathways, and Neuropharmacology of Endogenous Opioid Peptides). Because the currents generated by AVP or OT do not show pronounced time-dependent inactivation, they might contribute to persistent depolarizing membrane potentials that could modify the repetitive firing properties of their target neurons. By virtue of their voltage dependence, these currents could participate in regulating other synaptic inputs by causing preferential amplification of excitatory postsynaptic potentials (18, 47). Some of the functions attributed to central AVP and OT have been postulated to involve modifications of synaptic transmission, and they are consistent with the electrophysiological effects of these neuropeptides within the dorsal vagal complex. These functional correlates are considered in the next section.

The physiological and behavioral effects of central AVP and OT usually have been studied by direct injection of these neuropeptides into the cerebroventricular system. Although it cannot be determined what concentrations are achieved subsequently at critical synapses, many of the reported effects are produced only when the administered doses are in the high (microgram) range. The acute effects of such large central doses of AVP and OT often resemble each other (e.g., barrel rotation, seizures, or ataxia), and they may reflect the cross-reactivity of AVP and OT with each other's receptors and/or the toxic effects of large doses. Bearing this in mind, we have focused our current discussion on results that have been corroborated by studies using more refined techniques such as stereotaxically localized injections and specific receptor antagonists.

The various putative functions of central AVP and OT that we review briefly do not represent an exhaustive survey, but are intended to reflect those areas in which the most significant recent progress has been made. Most of the experiments demonstrating central effects of AVP or OT have been performed in rats and cannot be extended with assurance to humans. Nevertheless, the distribution of AVP and OT neurons, fibers, and binding sites in experimental animals closely resembles their distribution in humans, thus encouraging speculation that functional similarities exist as well.

AVP released from axon terminals in the median eminence enters the portal blood system and is carried to the anterior lobe of the pituitary gland, where it acts as a secretagogue for adrenocorticotropic hormone (ACTH) by binding to V1b receptors (4). The parvocellular system containing corticotropin-releasing hormone (CRH) is a probable source of portal AVP. AVP and CRH occupy the same secretory vesicles in terminals that contain both peptides, and AVP potentiates CRH-induced ACTH secretion. Two approximately equal subsets of CRH perikarya occupy the medial parvocellular subdivision of the PVN, those that express pro-AVP peptides and those that do not (69). Because the two populations of CRH neurons have different topographical distributions, they may be regulated differentially.

The precise role of OT on the hypothalamic–hypophyseal–adrenal system remains controversial, but its effect seems to be limited to controlling ACTH release only in some instances (26, 34, 61). OT immunoreactivity is colocalized in many of the AVP-negative, CRH-positive neurons in the PVN (69). Although OT-positive fibers are rare in the median eminence, large amounts of OT are present in the hypophyseal portal blood of the rat (26) and rhesus monkey (58), and high-affinity receptors for OT have been reported in the anterior pituitary of the rat (4). OT is less potent than AVP in enhancing ACTH release, but passive immunoneutralization of OT blunts the ACTH response to some types of stress in rats (26). In contrast, OT appears to inhibit rather than to stimulate ACTH release in primates, including humans, under certain conditions such as physical exercise (34).

The extensive research examining the effects of central AVP and OT on memory has been reviewed elsewhere (7, 16). Intracerebroventricular (i.c.v.) injections of very small (picogram) doses of AVP or the AVP derivative [pGlu4,Cyt6]AVP-(4, 5, 6, 7, 8) (a natural proteolytic fragment of AVP) induce long-lasting facilitation of learned passive avoidance behavior in rats. In contrast, OT and its natural derivative [pGlu4,Cyt6]OT-(4, 5, 6, 7, 8) hasten the extinction of learned avoidance responses, and central antagonism of AVP or OT can attenuate or facilitate, respectively, these conditioned avoidance behaviors (16). AVP appears also to have facilitating effects on "social memory" (the tendency of rodents to investigate unfamiliar conspecifics more intensely than familiar ones), a perhaps more ethologically relevant model with which to investigate learning and memory in rats (13).

Studies using stereotaxic microinjections of AVP and OT to alter passive avoidance behaviors have supported distinct sites of action for the two peptides. The dorsal septal nucleus and the ventral hippocampus are most sensitive to AVP, whereas the hippocampal dentate gyrus and the dorsal raphe nucleus are most sensitive to OT (5). In addition to directly stimulating hippocampal neurons, very low doses of AVP or OT also modify catecholamine turnover in the midbrain–limbic areas where they are believed to induce their effects on memory. Specifically, AVP increases and OT decreases alpha-methyl-p-tyrosine-induced disappearance of catecholamines in these areas, suggesting that changes in catecholamine activity might be involved in the behavioral effects of AVP and OT (5). Other recent data support a modulatory role of AVP on the activity of central nicotinic mechanisms that are critical for memory retrieval (21).

The relevance of these findings to human memory is uncertain. Inconsistent effects of AVP on human memory in various experimental paradigms have been reported (22). Nevertheless, clinical researchers have sought to determine whether the severe memory disturbance characteristic of Alzheimer's disease might be attributed, at least in part, to alterations in central endogenous AVP or OT. Patients with Alzheimer's disease were found to have abnormal levels of AVP and OT in the CSF and in certain brain areas (42, 46, 61)). The control of pituitary AVP release is impaired in patients with Alzheimer's disease, and AVP levels in the brain and CSF also are reduced (83). Conversely, a 35% increase in the concentration of hippocampal OT has been measured in men with Alzheimer's disease (46). Although these data appear to be consistent with the effects of AVP and OT in memory tests, clinical trials have not yet shown any beneficial effect of AVP analogues given chronically to patients with Alzheimer's disease (83); see also Experimental Therapeutics).

Central injections of AVP decrease body temperature in rabbits, guinea pigs, cats, rats, and sheep. Most studies examining the role of AVP in thermoregulation focus on its antipyretic effect in animals with experimentally induced fever, but a few studies have investigated the effect of AVP on basal body temperature. These latter studies have shown a marked decrease in body temperature, lasting up to 30 min after i.c.v. injection of only 30 ng of AVP (17). Conversely, low doses of a V1 receptor antagonist (10 ng, i.c.v.) produce hyperthermia in euthermic, conscious, unrestrained rats, and they block the effect of exogenous AVP (up to 100 ng, i.c.v.) to attenuate induced fever. The antipyretic effects of AVP appear similar to those of indomethacin or aspirin; moreover, central blockade of V1 receptors eliminates the antipyretic effects of peripherally administered indomethacin (36).

The ventral septal area, which is thought to mediate the antipyretic effects of AVP, contains both AVP binding sites and AVP-immunoreactive terminals derived from the bed nucleus of the stria terminalis (36, 75). Electrical stimulation of the bed nucleus produces antipyresis, and this effect is blocked by V1 receptor antagonists in the ventral septal area (36). The ventral septal area contains neurons that receive synaptic input transmitted from temperature sensors in the body, and these neurons alter their glutamate sensitivity in response to local application of AVP. Push–pull perfusion of the ventral septal area in sheep and rabbits reveals a direct relation between AVP levels and rising or falling body temperatures. These data have been interpreted to suggest that endogenously released AVP acts as a "brake" on febrile increases in body temperature (36). Because physiological stimuli that release AVP peripherally (such as plasma hyperosmolality) are antipyretic in rats and in guinea pigs, it was proposed that concurrent central release of AVP by these stimuli might mediate their effects on body temperature. Recent data have supported this hypothesis (57, 63).

In conscious, unrestrained rats, low doses of AVP given i.c.v. induce tachycardia, whereas higher doses cause tachycardia preceded by significant bradycardia (17). Injection of a V1 receptor antagonist i.c.v. increases heart rate, but this effect could be secondary to the behavioral activation that also results. Changes in both sympathetic and parasympathetic nervous outflow have been reported to underlie the central cardiovascular effects produced by i.c.v. injections of AVP, but conflicting data have been obtained along with marked interspecies variability (10, 17).

Smaller doses of AVP and AVP antagonists injected into the caudal medulla have provided data that can be interpreted more easily. AVP-immunoreactive terminals derived from neurons in the PVN form synaptic contacts with dendrites in both the dorsal vagal complex (81) and the lateral reticular nucleus (27) in rats. Injection of AVP antagonists into the dorsal vagal complex reduces the pressor and tachycardic responses elicited by electrical stimulation of the PVN (59). Microinjections of as little as 1 pmol of AVP into the lateral reticular nucleus produces dose-related increases in arterial pressure and heart rate, and injections of AVP into the dorsal vagal complex have a similar but somewhat smaller effect (27). Bilateral microinjections of a specific AVP antagonist into the lateral reticular nucleus following acute, small hemorrhage in rats produces a decrease in arterial pressure, whereas neither the AVP antagonist nor the hemorrhage alone alters arterial pressure. These data have been interpreted to suggest that under conditions (such as hemorrhage) demanding increased sympathetic drive to maintain arterial pressure, a functional AVP receptor mechanism in the caudal medulla may be activated to help restore normal blood pressure (27)). One conjecture is that AVP may enhance the sympathoinhibitory influence over the vagally mediated arterial baroreflex (10, 59).

Injecting OT into the dorsal vagal complex in rats increases gastric acid secretion and reduces gastric motility; these effects are eliminated by vagotomy (62). The OT innervation of the dorsal vagal complex arises from the PVN (68), and electrical stimulation of the PVN increases gastric secretion and has a biphasic effect on gastric motility, with a strong increase followed by a weaker decrease. OT antagonists injected into the dorsal vagal complex prior to PVN stimulation block the increase in gastric secretion and suppress the inhibitory phase of gastric motility, whereas central OT antagonism alone or acute PVN lesions increase baseline gastric motility (24, 62). These data suggest that the PVN exerts a tonic, OT-mediated influence on the activity of vagal gastric motor neurons in rats.

Stimulation of gastric vagal sensory fibers affects gastric secretion and motility in a manner similar to that of central OT. Vagal sensory information is transmitted directly to the PVN by ascending neural pathways (67). Gastric vagal stimulation excites OT neurons in the PVN (60), some of which project to the dorsal vagal complex in a reciprocal manner (53). Because microinjections of OT in the rat dorsal vagal complex specifically increase the excitability of neurons that receive mechanoreceptive gastric vagal sensory input (47), long-loop reciprocal pathways between the dorsal vagal complex and OT neurons in the PVN may enhance the efficiency of certain gastric vagal reflexes in rats. Whether OT plays a similar role in other species remains to be examined. In particular, OT binding sites are not discernible in the dorsal vagal complex of guinea pigs, and vagal neurons in guinea pigs are unresponsive to OT in vitro (77). These data caution against generalizing the gastric (and other) effects of central OT demonstrated in rats to other species.

Pituitary secretion of OT in rats is elicited by a variety of treatments that also inhibit ingestion of food and NaCl, including plasma hyperosmolality or hypotension, gastric distension, and systemic administration of cholecystokinin or lithium chloride (79, 80). Each of these treatments also stimulates parvocellular OT neurons with central axonal projections, resulting in increased central release of OT. Peripheral administration of OT or OT antagonists does not affect food or NaCl ingestion, mitigating against a role of systemic OT in these behavioral effects. In contrast, central administration of OT or a specific OT agonist reduces or completely eliminates ingestion of food or NaCl, and central injection of specific OT receptor antagonists blunts the ability of many treatments to inhibit food or NaCl intake (80). The inhibition of food and NaCl intake produced by treatments that stimulate central and systemic release of OT may reflect a coordinated effort to reduce solute concentrations in the body, because circulating OT derived from neurohypophyseal secretion increases urinary Na+ excretion in rats (80).

Physiological or pharmacological treatments that inhibit feeding also inhibit gastric motility and emptying in rats (25, 48). This association may be mediated by central OT systems, because OT acts in the dorsal vagal complex to inhibit gastric motility (discussed above). The inhibition of gastric motility may prolong or enhance gastric distension produced by food in the stomach, and thereby inhibit further food intake. It is likely that many of the chemical agents that potently stimulate central and pituitary OT secretion and inhibit gastric motility and food intake in rats also generate an aversive sensation such as nausea, because those agents produce strong learned taste aversions (79). However, as discussed above, activation of central OT transmission seems to hasten the extinction of other learned avoidance behaviors in rats. It remains to be determined whether anorexigenic/nauseogenic stimuli in rats enhance release of OT in the limbic forebrain regions that have been associated with memory. Similarly, although nauseogenic stimuli induce pituitary secretion of AVP rather than OT in several other species examined, including humans (49) (which could fit the "AVPmemory" hypothesis more neatly), it is not known whether central release of AVP rather than OT is stimulated in these species.

The triggering of maternal behavior in mammals is linked to endocrine changes that accompany the end of gestation and parturition. Considerable evidence now implicates central OT release in the initiation of maternal behavior during the immediate postpartum period. Parturition has been proposed to activate central OT release in addition to OT secretion from the pituitary, to provide coordinated maternal responses that could contribute to the survival of the young as well as to their birth. Many studies have provided support for this attractive hypothesis. Central injections of OT induce the rapid onset of maternal behaviors (i.e., retrieval, grooming, and nursing of the young) in steroid-primed virgin female rats in a dose-related manner (54). Central antagonism of OT transmission or PVN lesions each inhibit the onset of maternal behavior in postparturient rats (20). The site(s) of central action at which OT might act to stimulate maternal behavior has been localized variously to the ventral tegmentum, olfactory bulb, amygdala, midbrain, bed nucleus of the stria terminalis, and preoptic area; the reports are controversial, with some investigators seeing no effect of central OT at all (61). It seems likely that no one particular brain site is solely responsible for maternal behavior, but rather several sites that each facilitate different components of this complex behavior.

The proposed role of central OT in certain aspects of male and female sexual behavior is discussed elsewhere in this volume (see chapter by Pfaus and Everitt).

Central AVP may play a role in the control of male territorial displays and aggression. Studies of the golden hamster implicate an AVP-responsive region in the preoptic area of the anterior hypothalamus that is important in flankmarking (scent marking), a critical element in the behavioral repertoire associated with establishing and maintaining social dominance in males of this species (23). Flankmarking behavior is stimulated by intrahypothalamic administration of AVP and is inhibited by AVP antagonists. These site-specific effects are regulated by testosterone, which increases the presynaptic AVP innervation of the preoptic area (76).

It has been suggested that natural differences in AVP and OT receptor distribution may be functionally associated with species-typical patterns of social organization in mice and in voles (31). For example, voles that are monogamous and highly parental have very different distributions of central AVP and OT receptors than do voles that are polygamous and minimally parental, and these species differences are associated with pronounced differences in the behavioral responses to exogenous AVP and OT. There is some evidence that AVP and OT may be involved in the process of pair bonding in monogamous species. OT infused centrally (10–100 ng/hr) confers striking partner preferences in monogamous female voles, whereas infusion of AVP (50–100 ng) in male monogamous voles markedly increases aggression towards all conspecifics except the mate (31). Because partner preferences and aggression in monogamous voles are normally reinforced by mating, mating may release endogenous central AVP and/or OT in these species. Data showing that central administration of a specific AVP antagonist (50 pg to 500 ng) blocks the aggressive behavior that normally follows mating in male monogamous voles support this idea, but it remains to be seen whether OT antagonism blocks the formation of partner preferences in monogamous female voles (31).

In addition to their release into the systemic circulation, AVP and OT act as neurochemicals within the central nervous system to modulate neuronal activity through receptor-mediated mechanisms. Centrally released AVP and OT derive largely (but not exclusively) from PVN neurons that have central axonal projections. It seems unlikely that neurohypophyseal and central secretion of AVP and OT are generally coordinated under basal conditions. However, certain stimuli (i.e., parturition, dehydration, toxins) induce pronounced secretion of AVP and/or OT from the neurohypophysis as well as from axon terminals in discrete brain regions, and in such situations the peripheral and central actions of AVP and OT may be complementary in function. Identification of the physiological stimuli that induce central release of AVP or OT, and the functional consequences of such release, is an area of increasing investigation.

Given the broad array of complex behaviors that are affected by supplementing or blocking central AVP and OT in experimental animals, there is understandably considerable interest regarding the central functions of these neuropeptides in humans, especially in humans with neurological and psychiatric disorders. There are many speculations but few findings relating changes in central AVP and OT systems to human psychopathology. Alterations in CSF levels of AVP and/or OT have been observed in patients with schizophrenia, anorexia, obesity, clinical depression, alcoholism, Alzheimer's disease, or Parkinson's disease (14, 40, 46), but the functional significance of these observations is unknown.

AVP and OT systems are anatomically and functionally linked with catecholaminergic systems (67). Roles for both peptides as "stress hormones" have been considered (32, 34), consistent with their ability to alter ACTH secretion and thereby influence the adrenomedullary response to stress. In addition to their activation in situations of stress (71), many other functional and structural aspects of the central AVP and OT systems also are reminiscent of the central and peripheral catecholaminergic systems. For example, AVP and OT neurons are localized in discrete brain regions that receive afferent information from literally dozens of different cell groups, and they have axonal projections that extend throughout the central nervous system from the cerebral cortex to the spinal cord. Furthermore, AVP and OT may have effects on brain function that extend beyond classically defined (e.g., synaptic) limits, similar to what has already been proposed for centrally released catecholamines (71). These characteristics would permit the small, localized groups of AVP and OT neurons to influence neuronal events in many parts of the central nervous system, much as peripherally released catecholamines act on sympathetic targets throughout the periphery. A further similarity between the AVP, OT, and catecholamine systems is their demonstrated ability to modulate the responsiveness of their target neurons to other inputs. These influences often are exerted not over milliseconds but rather over minutes or hours, presumably due to receptor-mediated activation of second messenger systems. The diversity of physiological and behavioral effects that have been attributed to central AVP and OT supports the idea that these neuropeptides play a general role in modulating neural activity in hypothalamic, limbic, and autonomic circuits, and that the complexity of their effects arises from the complexity of the circuits in which they operate.

published 2000