PHYSIOLOGICAL REVIEWS   Vol. 78 No. 3 July 1998, pp. 583-686
Copyright ©1998 by the American Physiological Society

Angiotensin, Thirst, and Sodium Appetite

J. T. FITZSIMONS

The Physiological Laboratory, Cambridge, United Kingdom

I. INTRODUCTION
    A. Angiotensin-Induced Drinking Behavior
    B. Renin-Angiotensin Systems
II. HYPOVOLEMIC THIRST AND SODIUM APPETITE
    A. Causes of Hypovolemic Thirst
    B. Detection of Hypovolemia and the Threshold of Response
    C. Arousal of Sodium Appetite
    D. Initial Evidence Suggesting Involvement of the Renal Renin-Angiotensin System in Drinking Caused by Hypovolemia
    E. The Question of Whether Angiotensins Originating in Brain and Other Tissues Are Involved in Drinking Behavior
    F. Comment
III. ANGIOTENSIN-INDUCED THIRST
    A. Angiotensin-Induced Water Intake in Rat
    B. Angiotensin-Induced Water Intake in Other Mammals
    C. Angiotensin-Induced Water Intake in Birds, Reptiles, and Fish
    D. Amphibians
    E. Importance of Sodium
    F. Comment
IV. ANGIOTENSIN-INDUCED SODIUM APPETITE
    A. Angiotensin-Induced Sodium Intake
    B. Angiotensin-Induced Sodium Appetite, a More Complex Response Than Angiotensin-Induced Thirst
    C. Comment
V. EFFECTS OF ANGIOTENSIN ANALOGS AND ANTAGONISTS ON DRINKING
    A. Angiotensin Structure-Activity and Drinking
    B. Angiotensin Antagonists
    C. Renin and Angiotensin II Precursors
    D. Angiotensin-(1--7) Heptapeptide
    E. Angiotensin-(2--8) Heptapeptide or Angiotensin III
    F. Angiotensin-(3--8) Hexapeptide or Angiotensin IV and Shorter Chain Angiotensin Peptides
    G. Comment
VI. CENTRAL NERVOUS SYSTEM AND ANGIOTENSIN-INDUCED DRINKING
    A. C-fos Expression After Angiotensin
    B. Circumventricular Organs
    C. Anteroventral Third Ventricular Region and Lamina Terminalis
    D. Organum Vasculosum of the Lamina Terminalis
    E. Subfornical Organ
    F. Amygdala
    G. Brain Stem and Area Postrema
    H. Comment
VII. ROLE OF ANGIOTENSIN PEPTIDES FORMED FROM PRECURSORS IN BRAIN IN DRINKING
    A. Components and Distribution
    B. Possible Functions
    C. Role in Drinking
    D. Comment
VIII. NEUROTRANSMITTERS AND ANGIOTENSIN-INDUCED DRINKING
    A. Acetylcholine
    B. Catecholamines
    C. Serotonin
    D. Excitatory and Inhibitory Amino Acids
    E. Tachykinins and Bombesin-Like Peptides
    F. Opioids
    G. Endothelins
    H. Natriuretic Peptides
    I. Vasopressin
    J. Oxytocin
    K. Prostaglandins
    L. Nitric Oxide
    M. Other Neuroactive Substances
    N. Comment
IX. DRUG-INDUCED HYPERRENINEMIA AND DRINKING
    A. -Adrenergics
    B. Furosemide
    C. Angiotensin-Converting Enzyme Inhibitors
    D. Histamine
    E. Serotonin
    F. 3,4-Methylenedioxymethamphetamine
    G. Insulin
    H. Comment
X. EXPERIMENTAL AND CLINICAL CAUSES OF RENIN-DEPENDENT DRINKING
    A. Obstruction of the Inferior Vena Cava
    B. Heart Failure
    C. Hyperoncotic Dialysis
    D. Sodium Appetite of Adrenal Insufficiency
    E. Circulatory Shock
    F. Experimental Renal Hypertension
    G. Diseases of the Kidney and Hyperreninemia
    H. Ureteric Ligation
    I. Diabetes Insipidus and the Brattleboro Rat
    J. Diabetes Mellitus
    K. Comment
XI. CONCLUSIONS
REFERENCES

	ABSTRACT

Top References

Fitzsimons, J. T. Angiotensin, Thirst, and Sodium Appetite. Physiol. Rev. 78: 583-686, 1998.  Angiotensin (ANG) II is a powerful and phylogenetically widespread stimulus to thirst and sodium appetite. When it is injected directly into sensitive areas of the brain, it causes an immediate increase in water intake followed by a slower increase in NaCl intake. Drinking is vigorous, highly motivated, and rapidly completed. The amounts of water taken within 15 min or so of injection can exceed what the animal would spontaneously drink in the course of its normal activities over 24 h. The increase in NaCl intake is slower in onset, more persistent, and affected by experience. Increases in circulating ANG II have similar effects on drinking, although these may be partly obscured by accompanying rises in blood pressure. The circumventricular organs, median preoptic nucleus, and tissue surrounding the anteroventral third ventricle in the lamina terminalis (AV3V region) provide the neuroanatomic focus for thirst, sodium appetite, and cardiovascular control, making extensive connections with the hypothalamus, limbic system, and brain stem. The AV3V region is well provided with angiotensinergic nerve endings and angiotensin AT₁ receptors, the receptor type responsible for acute responses to ANG II, and it responds vigorously to the dipsogenic action of ANG II. The nucleus tractus solitarius and other structures in the brain stem form part of a negative-feedback system for blood volume control, responding to baroreceptor and volume receptor information from the circulation and sending ascending noradrenergic and other projections to the AV3V region. The subfornical organ, organum vasculosum of the lamina terminalis and area postrema contain ANG II-sensitive receptors that allow circulating ANG II to interact with central nervous structures involved in hypovolemic thirst and sodium appetite and blood pressure control. Angiotensin peptides generated inside the blood-brain barrier may act as conventional neurotransmitters or, in view of the many instances of anatomic separation between sites of production and receptors, they may act as paracrine agents at a distance from their point of release. An attractive speculation is that some are responsible for long-term changes in neuronal organization, especially of sodium appetite. Anatomic mismatches between sites of production and receptors are less evident in limbic and brain stem structures responsible for body fluid homeostasis and blood pressure control. Limbic structures are rich in other neuroactive peptides, some of which have powerful effects on drinking, and they and many of the classical nonpeptide neurotransmitters may interact with ANG II to augment or inhibit drinking behavior. Because ANG II immunoreactivity and binding are so widely distributed in the central nervous system, brain ANG II is unlikely to have a role as circumscribed as that of circulating ANG II. Angiotensin peptides generated from brain precursors may also be involved in functions that have little immediate effect on body fluid homeostasis and blood pressure control, such as cell differentiation, regeneration and remodeling, or learning and memory. Analysis of the mechanisms of increased drinking caused by drugs and experimental procedures that activate the renal renin-angiotensin system, and clinical conditions in which renal renin secretion is increased, have provided evidence that endogenously released renal renin can generate enough circulating ANG II to stimulate drinking. But it is also certain that other mechanisms of thirst and sodium appetite still operate when the effects of circulating ANG II are blocked or absent, although it is not known whether this is also true for angiotensin peptides formed in the brain. Whether ANG II should be regarded primarily as a hormone released in hypovolemia helping to defend the blood volume, a neurotransmitter or paracrine agent with a privileged role in the neural pathways for thirst and sodium appetite of all kinds, a neural organizer especially in sodium appetite, or all of these, remains uncertain. ANG II-induced drinking behavior serves as a model of how other complex behaviors involving neural and peptide inputs might be organized.

	I. INTRODUCTION

Top Next References

Angiotensin-induced drinking behavior is a remarkable phenomenon. I discuss it in detail and analyze the different circumstances in which angiotensin peptides may be involved in thirst and sodium appetite. There has been an enormous amount of work on these topics in the past 30 years, and it would be impracticable to survey all that has been published. What follows is therefore selective and very much what has interested me. I have tried to allow for personal bias by citing key review articles or chapters in books for certain aspects of the work described here. I have also given an extended list of monographs, multiauthor works, and conference proceedings devoted to thirst, sodium appetite, and renin-angiotensin systems.

Thirst is a sensation aroused by a need for water, and relief from it is sought by drinking water. We infer from our own experience in similar circumstances that the dehydrated animal eagerly seeking water is also experiencing thirst. But lecturing, singing, eating a spicy meal, seeking pleasure from a refreshing drink of water on a hot day, or just habit may lead to increased consumption of water, although there may be no immediate need for it and no thirst in the strict sense of the word. Perhaps the appropriate term to use here is appetite for water. There are also circumstances in which animals continue to drink water or can be made to do so even though their needs for water have been met many times over. For example, rats infused with more water than they needed in the 24 h, by routes that bypassed the mouth and pharynx, continued to drink spontaneously about one-third or more of their preinfusion intake of water (184). It is important to be aware of circumstances such as these when assessing the results of experiments on drinking behavior. We reasonably assume that animals experience the same sensations as ourselves when they suffer the same deficits, but we should also recognize that animals, like ourselves, drink water for reasons other than to repair a water deficit; they have urges and preferences as well as needs that they seek to gratify. And although it is desirable that we restrict the word thirst to the sensation aroused by a lack of water, in general usage it incorporates both an idea of appetite for water as well as a drive toward relief of a need. We are often tempted by water as well as driven to assuage our thirst.

Increased sodium appetite indicates a need for sodium, and relief will be sought by consuming salt or salty foods. But much more so than with thirst, experience of a previous occasion of sodium deficit and the satisfaction afforded by the taste of salt and consequent intake may so condition behavior that an appetite for repetition of this experience is quickly established. In other words, sodium intake is commonly driven by preference as well as need, and the subject gains satisfaction from it as well as relief. Cannon (89) wrote, "It is not to be supposed that the two motivating agencies  the pang and the pleasure  are as separate as we have been regarding them for the purposes of analysis in the present discussion. They may be closely mingled; when relief from hunger or thirst is found, the appetite may simultaneously be satiated." Cannon was discussing hunger (the pang) and appetite (the pleasure) for food, but his remarks apply with equal force to hunger or appetite for sodium. In most circumstances of sodium need, sodium intake is undoubtedly driven by need, but there is also an element of preference: both the pang and the pleasure operate. These are the reasons why, like most workers in the field from Richter (479) onward, I have preferred to use the word sodium or salt appetite rather than sodium hunger, since appetite implies preference as well as need and it takes account of the powerful effects of learning on the response to a need. I have also preferred to use sodium instead of salt because salt can mean a nonsodium salt.

View larger version (23K): [in this window] [in a new window]   FIG. 1.   Formation of angiotensin peptides. Nonrenin angiotensinogenases include tonin and cathepsins D and G.

	II. HYPOVOLEMIC THIRST AND SODIUM APPETITE

Top Previous Next References

Cellular dehydration and hypovolemia are the two principal causes of deficit-induced drinking. Loss of cell water is detected by osmoreceptors (and possibly sodium-sensitive receptors) located in the hypothalamus and elsewhere that share in the cellular dehydration and give rise to thirst. There is long-standing evidence on monitoring of extracellular fluid volume by stretch receptors in the walls of the heart and vasculature (232, 547). Hypovolemia is detected by these receptors which cause a delayed increase in sodium appetite as well as thirst. It is in drinking behavior induced by hypovolemia that renin-angiotensin systems seem to be most involved. Circulating ANG II derived from renal renin contributes to hypovolemic thirst. It also plays a role in increased sodium appetite, acting with the mineralocorticoids and other hormones. The part played by ANG II formed locally in the brain in these behaviors is uncertain, but in the rat, it may be more important in sodium appetite than in thirst. Other angiotensin peptides formed in the periphery or brain may contribute, but information is lacking. The circumstance in which hypovolemia leads to increased intakes of water and NaCl, some of the methods used to elicit these behaviors, and early evidence implicating renal renin are considered here.

Experimental procedures that have been used to arouse hypovolemic thirst include bleeding, inducing sodium deficiency (371), causing sequestration of extracellular fluid by intraperitoneal (175) or subcutaneous injection (560, 567) of hyperoncotic colloid, and interfering with venous return to the heart by obstructing the inferior vena cava (176, 178, 197, 614). Among clinical causes are severe diarrhea and vomiting, some forms of renal disease, congestive heart failure, and circulatory shock. Many of the experimental and clinical causes are considered in section X. Hypovolemia leads to an increase in circulating ANG II, and this contributes to drinking, although it is not exclusively responsible for the overall drinking response. On the other hand, even in the absence of any other stimulus, thirst can be aroused by injecting ANG II or causing it to be formed by drugs or surgical procedures. Because there is no additional renal renin release in cellular dehydration, circulating ANG II does not contribute to the thirst aroused, although angiotensin of cerebral origin may (see sect. VII). In mixed cellular and extracellular dehydration, ANG II like any other thirst stimulus adds its effect to the overall drinking response. As for spontaneous day-to-day drinking, here water intake is usually in excess of need, and much of the evidence suggests that drinking is determined by habit, showing the characteristics of a nyctohemeral rhythm entrained with feeding, with no fluid deficit or hormonal signal involved (184). However, it has also been suggested that release of histamine and transient hypovolemia associated with feeding may cause release of enough renal renin to stimulate drinking.

Hypovolemia is detected by vascular stretch receptors in various parts of the circulation and possibly by other types of receptor as well (197, 232, 252, 292, 522). Unloading of cardiopulmonary and arterial stretch receptors as venous return and cardiac output drop, pulse pressure narrows, and mean arterial pressure falls, causes a number of compensatory responses, including increased sympathetic activation; decreased vagal tone; increased secretion of renin, aldosterone, ACTH, and vasopressin; and increased drinking. Reflex circulatory adjustments to loss of volume are rapid; renal conservation of water and electrolytes and replacement of the deficits of water and sodium by intake mechanisms are slower responses that depend on both neural and hormonal mechanisms. A deficit in plasma volume of ~8-10% (or this degree of underfilling in critical receptor regions in the heart and blood vessels) causes significant drinking, a threshold that is considerably higher than the deficit of 1 or 2% of cellular water needed to arouse osmotic thirst.

It seems possible that the hypovolemic threshold is higher than the osmotic because the range of blood flow rates required in different activities is so extended and the scope for changes in the distribution of cardiac output and the partitioning of fluid across the capillary wall so great that it is necessary to avoid the controls of extracellular volume being constantly triggered by smaller changes in plasma volume. The ample reserve of fluid in the interstitial fluid compartment acts as a buffer for plasma volume and can be mobilized through operation of the Starling capillary filtration/reabsorption system should the need to correct hypovolemia arise. It would be inappropriate if the circulatory adjustments accompanying different levels of bodily activity were to arouse thirst because of what turned out to be temporary understimulation of vascular stretch receptors, which ended as soon as the activity diminished and circulatory function returned to the resting state. It would make little sense were hypovolemic thirst mechanisms to keep switching on and off in this way because of transient shifts of blood between vascular beds. But once the loss of circulating volume becomes significant and persistent, drinking under the control of hypovolemic thirst is as vigorous as that caused by equivalent cell dehydration, and the amounts of water drunk may be greater than after cellular dehydration of similar magnitude. The requirements of the cellular fluid compartment are different; in contrast to the variable volume demands of the circulation, stability of cellular water content is a necessary condition for normal cell function.

We can only speculate on the mechanisms that are responsible for the long-term precision of blood volume control. This control, which allows for large short-term variations in the distribution of blood and the partitioning of fluid between the plasma and interstitial compartments depending on circulatory needs, implies the accurate monitoring of volume in certain strategic regions of the circulation, at least for part of the time, perhaps integrated over the long-term during periods of rest.

Neural and hormonal effector mechanisms vary the intake, distribution, and excretion of fluid and electrolytes in accordance with long-term needs. The immediate increase in water intake is caused by altered neural inputs from the vasculature reinforced by increases in circulating ANG II. Circulating ANG II contributes directly to the overall drinking response by stimulating structures in the central nervous system and, if the hypovolemia is severe, it is also important in maintaining the blood pressure and therefore the behavioral competence of the animal. However, it is not essential in hypovolemic thirst that can occur in the absence of circulating ANG II.

An increase in sodium appetite is the second behavioral response to hypovolemia. Many mammals in sodium deficit seek and ingest salt, driven to do so by increased sodium appetite (122, 498, 523), an innate behavior that serves sodium homeostasis (148). Sodium appetite is also expressed in many birds, including the pigeon. Animals seek and ingest what they like as well as what they need, and much sodium intake is preference driven; for ourselves, table salt is a common condiment, and sodium-replete rats readily accept and drink large quantities of isotonic NaCl, the saline concentration they prefer. Sodium is normally a major urinary constituent and represents the excess of intake over the amounts needed by the body. It is important in the investigation of need-driven sodium appetite to offer concentrations of sodium solutions that the sodium-replete animals normally avoid. Under these conditions, we can be reasonably sure that any NaCl intake is need driven. Sodium appetite is well developed in inland-dwelling herbivores including some primates, particularly during pregnancy and lactation when calls on fluid and electrolyte reserves of the body are greatest, or when sodium deficiency develops as a result of, for example, intestinal infection. Herbivores in the wild may travel long distances to places where salt is to be found, and this is presumably why their predators also tend to congregate at these places. But omnivores, including humans, and carnivores when they are unable to satisfy their sodium needs from their prey, show increased sodium appetite when sodium deficient.

Inducing sodium deficiency by placing the subject on a low-sodium regime is the simplest way to generate an appetite, and of course, here ANG II levels increase. The combination of dietary restriction and enforced sweating was used by McCance (371) in his classical experiments on himself and colleagues, published in 1936, on the effects of inducing sodium chloride deficiency and, although the study of sodium appetite was not the primary purpose of this experiment, the subjects made some interesting observations on alterations in their sense of flavor and taste. In a more recent study, experimental sodium deficiency in 10 human subjects caused increased preference for salty foods (32). More active procedures to make animals sodium deficient, which may be combined with dietary sodium restriction, are treatment with furosemide or other natriuretic drugs (see sect. IXB); peritoneal dialysis with isosmotic glucose; unilateral fistulation of the parotid salivary gland in sheep, calf, and goat (122); and adrenalectomy (see sect. XD). The methods used to arouse hypovolemic thirst also cause increased sodium appetite, and to these may be added subcutaneous injection of buffered Formalin, which is an effective natriorexigenic stimulus (669).

Sodium appetite can also be induced in the absence of need for sodium by administration of the hormones of sodium deficiency, the mineralocorticoids and ANG II, which are normally secreted in extra amounts in the circumstances just mentioned. But to these should be added the hormones of pregnancy and lactation and the stress hormones of the hypothalamo-pituitary-adrenocortical axis, corticotrophin releasing hormone (CRH), ACTH, and glucocorticoids (122, 597), although these are of varying effectiveness and there are species differences in the effects produced. Angiotensin II-induced sodium appetite is considered in detail in section IV, and the reproductive hormones are considered in sections IIIA4 and IVA3.

Increased sodium appetite caused by mineralocorticoids has been abundantly studied and verified since Richter's discovery of the phenomenon more than 50 years ago (122). A U-shaped function describes the effects of systemic administration of mineralocorticoid on sodium appetite; replacement doses of deoxycorticosterone acetate (DOCA) reverse the increased NaCl intake that follows adrenalectomy, whereas larger than replacement doses of DOCA or aldosterone stimulate NaCl intake in intact and adrenalectomized rats despite the accompanying sodium retention and suppression of renal renin secretion (56, 221, 475, 478, 479, 668). Arousal of sodium appetite depends on central actions of mineralocorticoids. Receptor binding studies have shown that there are two receptor subtypes for corticosteroids, mineralocorticoid (type I or MR), and glucocorticoid (type II or GR). Both are present in rat brain, and their regional distribution is similar. The highest uptakes of aldosterone and corticosterone are in the hippocampus, septum, and amygdala (402), and there is a brief report that bilateral adrenal steroid implants in the amygdala caused an increase in intake of NaCl but not water (472). Central nervous structures involved in increased sodium appetite are discussed in section VI.

Sodium appetite can be aroused in rats by inducing the syndrome of apparent mineralocorticoid excess (AME) using the active component of licorice, glycyrrhizic acid, or its hydrolytic product, 18-glycyrrhetinic acid (105, 106). The pattern of increased drinking produced resembles that which occurs after administration of excessive amounts of mineralocorticoid, but plasma corticosteroid levels are not increased. Apparent mineralocorticoid excess has been identified as a cause of human hypertension (529). The clinical picture is similar to that produced by excess mineralocorticoids, with sodium retention, hypertension, potassium loss, and inhibition of the renin-angiotensin system. In type I AME, inactivation of glucocorticoids is impaired because the enzyme 11-hydroxysteroid dehydrogenase (11-OHSD), which oxidises cortisol to cortisone, is congenitally deficient or has been blocked by licorice (557). Type II AME is secondary to a deficiency in the A-ring reduction metabolic pathway. In rats, licorice prevents oxidation of corticosterone, the glucocorticoid in rodents, to 11-dehydrocorticosterone. Failure to inactivate glucocorticoids allows the mineralocorticoid receptors, whose affinities for glucocorticoids and mineralocorticoids in vitro are the same (402), to be powerfully stimulated by the large amounts, relative to aldosterone, of glucocorticoid present. Mineralocorticoid receptors, including those in brain (323), are normally protected from stimulation by glucocorticoids by this enzyme. Licorice inhibits 11-OHSD activity in kidney and other tissues, exposing the mineralocorticoid receptors to glucocorticoid stimulation, but the 11-oxoreductase activity of liver is mainly unaffected, ensuring reconversion of inactive 11-ketosteroid to active glucocorticoid (401). Glycyrrhizic acid-induced increases in sodium appetite were abolished by adrenalectomy but could be restored by administration of cortisol or corticosterone but not by replacement dosage of DOCA (107). Increased sodium appetite in AME suggests that 11-OHSD inactivation of glucocorticoid also operates in those parts of the brain responsible for mineralocorticoid-induced appetite.

The stimulating effect of DOCA is much smaller in rabbit and sheep, and aldosterone seems ineffective in these species (122). Dogs, hamsters, and gerbils also fail to show significant natriorexigenic responses to mineralocorticoids (498). On the other hand, it has been shown that several adrenal steroids in slow-release pellet form, of which deoxycorticosterone (DOC) was the most potent, evoked sodium appetite in BALB/c mice (45), although it was stated earlier that mice do not show a natriorexigenic response to mineralocorticoids (498, 499). The pigeon develops an increase in sodium appetite in response to subcutaneous DOCA (151). Glucocorticoids can generate sodium appetite in rabbit and sheep but in rat only when accompanied by mineralocorticoid (668). Adrenocorticotrophic hormone (Synacthen) stimulates sodium appetite in rabbit, rat (659), and sheep. In rabbit, the effect is mediated partly by the adrenal cortex, but there is also an extra-adrenal effect, since ACTH has some action in adrenalectomized animals, presumably by acting directly on the brain (122). Mineralocorticoid-induced sodium appetite is a primary natriorexigenic effect on brain; it is independent of and unaffected by the developing "escape" from mineralocorticoid action on the kidney. Corticosteroid therapy, or excessive mineralocorticoid secretion whether accompanied by increases or decreases in circulating ANG II, could contribute to inappropriate or excessive sodium intake in human. Continuing sodium intake in the face of sodium retention indicates that the intake is not homeostatically determined.

Mineralocorticoid treatment, or induction of AME, also causes increased water intake, which is usually regarded as being exclusively secondary to NaCl intake, but some increased intake also occurs when water only is available to drink in dog (164, 461) and rat on a high salt (475, 479) or normal (unpublished data) diet. Furthermore, the relation between water intake and NaCl intake is not always close; in mouse, DOC stimulated NaCl intake but had no effect on water intake, whereas the combination of DOC with other corticosteroids, and especially with ACTH, caused increased water intake as well as NaCl (45). Apart from the osmotic effect of the NaCl intake, increased water intake may be the result of mineralocorticoid-induced potassium depletion giving rise to cellular dehydration and failure of renal concentrating ability. Potassium deficiency has long been recognized as a cause of a diabetes insipidus-like syndrome with polydipsia and vasopressin-resistant polyuria both in animals and in humans. The stimulating effect of potassium depletion on water intake may be primary; the polyuria of hypokalemic subjects is often in excess of the urine volume required by their concentrating defect (39).

Increased sodium appetite is an essential and appropriate defense mechanism against sodium deficiency, but other mineral deficiencies may also cause increased NaCl intake. These are circumstances where the relation between sodium appetite and sodium need and/or its hormonal accompaniments appear to be completely absent. In addition to the polydipsia/polyuria syndrome already mentioned, potassium depletion in rat caused increased intake of NaCl as well as KCl (122). Rats also show increased sodium appetite in response to calcium depletion even in the absence of sodium loss or increases in the hormones of sodium homeostasis. The appetite is not related to plasma calcium levels, the hormones controlling calcium, or the renin-angiotensin-aldosterone system, and at present it is unexplained (627, 628).

Because ANG II and mineralocorticoids are central in the body's defenses against sodium loss and consequent hypovolemia, much of the experimental effort to try and elucidate the physiological basis of sodium appetite has been devoted to these hormones. One approach to investigate possible participation of renal renin in sodium appetite has been to lower plasma levels of renin and ANG II by bilateral nephrectomy, or to prevent the effects of renin secretion with appropriate renin-angiotensin system antagonists (see sect. V). Another approach has been to try to stimulate or restore sodium appetite by elevating plasma ANG II levels directly or by administering renin, or by stimulating renal renin secretion pharmacologically or surgically. These approaches complement studies on the possible role of cerebral renin in sodium appetite and its relation, if any, to renal renin; they are essentially those used to assess the renin contribution to thirst.

I proposed that the renal renin-angiotensin system has a direct and physiological role in certain types of drinking behavior as a result of certain findings on caval obstruction as a stimulus to drinking in the rat (176, 178). The procedure of caval obstruction was devised as a method of mimicking the circulatory effects of severe hypovolemia. Within 30 min to 1 h after constriction of the abdominal inferior vena cava just above the renal veins, there was an increase in water intake followed much later by an increase in sodium intake; there was also a marked fall in urine flow and electrolyte excretion so that for a time the animal gained weight as it retained fluid. Because this stimulus to thirst was found to be less effective in the nephrectomized rat than in the intact rat, it seemed that renin or some other renal factor released as a result of the reduced venous return to the heart could have been partly responsible for the increased drinking. Making the rat anuric by bilateral ureteric ligation, a procedure which preserves the endocrine function of the kidney, did not prevent caval obstruction-induced drinking. Experimental support for the theory of renal renin involvement in this behavior came with the finding that saline extracts of the renal cortex of rat caused increased drinking especially in nephrectomized rats. The properties of the rat renal extract were found to be similar to those of renin and purified commercial pig renin caused a dose-dependent increase in drinking (178).

The idea that the kidney might contribute to thirst, although not through renin secretion, was proposed by Linazasoro et al. (341). They suggested that when the supply of water to the body is lacking, the kidney liberates a thirst factor that helps the body to resist dehydration. They found that in bilaterally nephrectomized rats that were allowed free access to water, the loss of body weight and fall in water intake that usually occur could be prevented by injecting an extract of pig kidney. Because renin apparently did not have these effects, Jiménez-Díaz et al. (288) concluded that the renal thirst factor is not renin. However, there were already clues that the renal thirst factor might indeed be renin. In a series of experiments on accelerated hypertensive disease in nephrectomized dogs (368) and rats (367), Masson and co-workers found that the syndrome was intensified by renin, which stimulated water intake. But they considered that increased drinking was secondary to hypovolemia consequent on increased capillary transudation and they stated that ". . . thirst induced by renin in nephrectomized rats is not the result of primary stimulation of thirst centers . . ." (367). Asscher and Anson (17) came to a similar conclusion, finding that renal extracts stimulated water intake in nephrectomized rats, but attributing this to hypovolemia caused by leakage of fluid into the tissues.

Because the caval obstruction experiments described above suggested that there are occasions when endogenously released renin as opposed to injected renin or renal extracts contributes to thirst, it seemed likely that renin acted by more physiological mechanisms than by causing or exacerbating hypovolemia. The question now arose whether, as for other effects of renin, increased drinking in response to caval obstruction was mediated by ANG II (178). This proved to be the case. It was found that ANG II caused increased drinking when infused intravenously into water-replete rats at doses below those which produced marked changes in hematocrit value and that it restored the nephrectomized rat's reduced water intake in response to caval obstruction near to that of a normal rat's response (200). In an investigation of norepinephrine-induced eating in rat, it had been observed that intrahypothalamic injection of ANG II caused drinking (51). It was now shown that injection of small amounts of ANG II into the anterior hypothalamus and preoptic region caused dose-dependent increases in water intake in the water-replete rat (149), suggesting that increased circulating ANG II caused drinking by acting on accessible structures in the central nervous system. Shortly after these early experiments on ANG II-induced water intake, intracranial ANG II was also shown to cause increased NaCl intake (72, 96).

	III. ANGIOTENSIN-INDUCED THIRST

Top Previous Next References

One of the most striking stimulatory effects of any substance on any motivated behavior is the vigorous, short-latency burst of drinking that follows injection of ANG II into sensitive structures in the brain. The initial increase in water intake is followed by a developing increase in NaCl intake (see sect. IV). Systemic injection of ANG II also causes water-replete animals to start drinking, although the response is less predictable than after intracranial administration. Responsiveness to the dipsogenic effect of ANG II is present in all vertebrate groups examined (Table 1), including certain euryhaline bony fish, but excluding amphibians and elasmobranch fish. Renin and some other components of the renin-angiotensin cascade given by intracranial or systemic injection also cause increased drinking.

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Water intake in 15 min after injection of angiotensin (ANG) II in subfornical organ (SFO) or in adjacent lateral ventricle (LV) or third ventricle (3V). Numerator in each fraction is number of rats responding, and denominator is number of animals injected at particular dose and injection site. [From Simpson et al. (537).]

2. Systemic administration

An increase in drinking can also be produced by systemic administration of ANG II or its precursors, or by causing release of renin from the kidney by pharmacological or surgical means (see sects. IX and X). Although it is difficult to make direct comparisons because the techniques of administration are so different, the conditions required for eliciting drinking in response to systemic ANG II are more exacting than those for intracranial administration. With the use of appropriate doses to obtain maximal responses by the two routes, drinking responses caused by intravenous infusion are generally smaller than those following intracranial administration. In contrast to the robust drinking response that is almost invariably obtained with intracranial ANG II, there are well-documented cases in the literature where systemic ANG II has failed to cause drinking.

Angiotension II is usually given by continuous intravenous infusion to unrestrained animals using an external infusion system (0.05-3.0 µg·kg¹·min¹; Ref.

1

1

1

1

View larger version (35K): [in this window] [in a new window]   FIG. 3.   Regression of plasma ANG II concentrations after intravenous infusion of ANG II at 1, 25, 50, or 100 pmol·kg1·min1 for 60 min in nephrectomized rats. Plasma ANG II levels produced by various thirst stimuli in other rats are indicated by arrows on regression line. Top shaded area represents plasma ANG II level at dipsogenic threshold for intravenous ANG II, and bottom shaded area is range of controls. Intact 25 and 100 pmol·kg1· min1 indicates level of circulating ANG II after a 60-min infusion into unanesthetized rats with intact kidneys. DI, diabetes insipidus; RH, renal hypertension; PEG, polyethylene glycol; Isop, isoproterenol. [From Mann et al. (355).]

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

Renin-angiotensin systems are present in birds, reptiles, amphibians, and bony fish (260, 261), and possibly in elasmobranch fish (593). Sensitivity to the dipsogenic action of ANG II occurs in all vertebrate groups except amphibians. Many birds have been found to drink in response to intracranial injection of ANG II (Table 2). The pigeon, Columba livia, is a vigorous drinker, showing a copious, short-latency response to intracranial ANG II (see Fig. 7) that is largely completed within ~10 min of injection (152). It responds to femtomole doses of ANG II, and it is at least as sensitive as the rat but drinks about three times as much water to a given dose. Other birds that drink in response to intracranial ANG II are Japanese quail, Coturnix coturnix japonica (591), white-crowned sparrow, Zonotrichia leucophrys gambelii (646), Pekin duck, Anas platyrhynchos (115), turkey (121), and domestic chicken (525, 548), or chick (645), Gallus gallus and Gallus domesticus, in which the doses (10 µg) required to cause drinking were very large.

View larger version (23K): [in this window] [in a new window]   FIG. 7.   Mean water intakes (in ml; ±SE) after ICV injections of ANG II, precursors of ANG II, and shorter chain ANG fragments into pigeon Columba livia. Birds did not drink between 30 and 60 min after tetradecapeptide (TDP) renin substrate, ANG I, and ANG II. Number of birds tested is given in parentheses. [Two figures combined from Evered and Fitzsimons (153).]

Systemic ANG II also causes drinking in birds. Intravenous (5-100 pmol/min for 15 min) or intraperitoneal ANG II is a highly effective stimulus to drinking in the pigeon, the lowest rate of intravenous infusion (16 pmol· kg¹·min¹) which caused drinking being similar to the lowest effective rate in rats and dogs (152). The Japanese quail drinks in response to intravenous ANG II (5-125 µg), and it also responds to subcutaneous injection (591). Another good responder to intravenous (646) and intraperitoneal (312) ANG II is the white-crowned sparrow. After an intravenous infusion of hypertonic NaCl, the duck drank in response to intravenous ANG II (200 ng/min), but without hypertonic priming, ANG II did not always stimulate drinking (535). The domestic chicken appears to be less sensitive than pigeon to systemic ANG II, but large amounts given by intravenous (300 µg), intramuscular (100-400 µg), or subcutaneous (100 µg) injection caused adult birds (525, 548) and chicks (645) to drink water. Two species of parrot, Barnardius zonarius semitorquatus and Barnardius zonarius zonarius, responded to intraperitoneal ANG II (1-50 µg/100 g), but two other species were relatively insensitive (310). In intraperitoneal injection experiments on 18 species of bird, including Japanese quail and white-crowned sparrow, most granivorous and omnivorous birds drank in response to ANG II (1-10 µg/100 g), but carnivorous species on the whole did not (312).

There have been few studies of dipsogenic responsiveness to angiotensin peptides in reptiles. Intraperitoneal injection of ANG II (5-40 nmol) caused dose-dependent drinking in the common iguana, Iguana iguana, with shorter latencies than after osmotic stimulation (193). The Indian gecko, Hemidactylus flaviviridis, the Japanese lizard, Takydromus tachydromoides, and Japanese skink, Eumeces latiscutatus, two species of Japanese snake, Elaphe quadrivarigata and Elaphe climacophora, and the mud turtle, Kinosternon subrubrum, all responded to intraperitoneal injection of ANG II (5-100 µg/100 g), but the European tortoise, Testudo graeca, needed very high doses (1,000 µg/100 g), and the Indian garden lizard, Calotes versicolor, and Indian water snake, Natrix piscator, failed to drink (312). Based on their experiments on the dipsogenic responsiveness of mammals, birds, and reptiles to intraperitoneal ANG II, Kobayashi et al. (312) concluded that animals that drink little water in their natural environment are relatively insensitive to the dipsogenic action of ANG II.

In fish, cyclostomes and teleosts in seawater drink continuously, presumably because of the dehydrating effect of the hypertonic environment in which they live, but elasmobranchs do not drink, perhaps because they are protected from dehydration by their own hypertonicity. Bony fish respond to similar dipsogenic stimuli as terrestrial vertebrates, although the neural organization is less encephalized and seems to be located in the hindbrain (594), presumably because the task of getting water is much simpler than for a terrestrial animal. Freshwater fish have to cope with osmotic inflow of water, and the problem is getting rid of water, not acquiring it, so that generally they do not drink, although some do. Euryhaline fish that migrate between seawater and fresh water drink while in seawater and stop drinking when in fresh water. In experiments on 20 species of freshwater fish and 17 species of seawater fish, it was found that the freshwater fish which enter and are able to survive in estuarine brackish water drank after intraperitoneal injection of ANG II (1-100 µg/100 g), whereas fish that live exclusively in fresh water or seawater did not respond (311). Among euryhaline fish, the Japanese eel, Anguilla japonica, drank in response to injection of ANG II (100 ng-100 µg) into the pneumogastric artery (594), the flounder, Platichthys flesus, to intravenous infusion (6 and 150 ng/min after captopril blockade) through the caudal vein (26, 93), and the killifish, Fundulus heteroclitus, to intraperitoneal injection (100 µg) (352). Because ANG II-induced drinking characterizes fish that from time to time enter water more hypertonic than that in which they typically live, it may be an emergency response to dehydration; true stenohaline freshwater or seawater species do not respond to ANG II. The mechanisms of ANG II-induced drinking in fish are unknown, nor have the sites of action of ANG II been established. It has been suggested that drinking may be related to ANG II-induced changes in blood pressure and that baroreceptors or volume receptors are involved, or that the chemoreceptors monitoring external chloride are important (261). Vagotomy prevents drinking completely, but the forebrain and midbrain can be destroyed without affecting the response.

Renin-angiotensin systems are present in amphibians in granular epithelioid cells arranged in what is recognizably the beginnings of a juxtaglomerular apparatus (260, 261). In general, amphibians do not drink when challenged by stimuli that cause members of other vertebrate groups to drink, but they maintain their water content by regulating percutaneous water absorption. Little is known about water-seeking behavior by dehydrated amphibians and possible effects of angiotensin peptides on such behavior. Common frogs in water deficit do not appear to find water any more readily than the hydrated frogs, and dehydrated frogs have been observed dying of water lack within a few centimetres of a pool of water (6). The esophagus-cannulated Japanese frog, Rana brevipoda, failed to drink in response to injection or infusion of ANG II through the abdominal vein (264). It also failed to drink after intravenous hypertonic NaCl or subcutaneous hyperoncotic dialysis (see sect. XC) but did drink immediately after exposure to seawater. Water-seeking behavior in newts, salamanders, frogs, and toads was also unaffected by angiotensin (264). A curious observation is that captopril appeared to promote drinking in frogs adapted to a hyperosmotic environment, a response prevented by nephrectomy (261). The ecology of the amphibian is so diverse that it seems likely that there would be differences in behavioral drives toward water, especially among the species that inhabit arid regions, for example, certain Australian frogs, which show burrowing behavior that enables them to reach the damp subsoil. Whether ANG II would cause them to do this more vigorously is an interesting question.

According to Andersson (9), the effectiveness of ANG II as an intracranial dipsogen can be explained by its effect on sodium transport. Drinking, antidiuretic, natriuretic, and pressor responses in the goat were markedly enhanced when ANG II and hypertonic NaCl were infused together into the CSF, whereas the responses with either substance alone were much less (10-12). The angiotensin responses were much attenuated when ANG II was dissolved in isotonic glucose. Intracerebroventricular infusions of hypertonic NaCl are more dipsogenic than hypertonic saccharides in sheep and pig but not in dog (373). Thornton (604) found that pigeons drank normally in response to peripheral osmotic stimuli, but only hypertonic NaCl was effective centrally. Drinking appeared to be inititiated by osmoreceptors, but a certain CSF sodium concentration was required for the osmoreceptor mechanism to operate effectively. The pigeon's response to intracerebroventricular ANG II also depends on CSF sodium, drinking being attenuated when the CSF sodium was lowered (604). The sodium ion undoubtedly occupies a privileged position in the fluid and electrolyte economy of the body, and it would make physiological sense if there were special sodium receptors to monitor sodium concentration at certain critical sites. Angiotensin II could enhance the sodium effect by stimulating sodium transport from the CSF to putative sodium receptors, making the receptors more responsive to the existing sodium concentration, or increasing the intracellular sodium in the receptors (9). Most results are consistent with an osmoreceptor hypothesis, but sodium-sensitive receptors responding to CSF sodium concentration may play a "permissive role," and the responses of these could well be enhanced by increases in ANG II. In physiological circumstances where angiotensin-induced homeostatic mechanisms might be expected to operate, it is not obvious that variations in CSF sodium would be sufficient to affect responses, but sodium sensitivity may become important in hyponatremic or hypernatremic states.

	IV. ANGIOTENSIN-INDUCED SODIUM APPETITE

Top Previous Next References

Angiotension II causes a delayed and persistent increase in sodium appetite following the immediate thirst. This sequence reproduces the pattern of drinking in hypovolemia. Increased NaCl intake is an appropriate behavior in hypovolemia and sodium deficiency because extracellular solute as well as water is required for the proper restoration of blood volume. Increased ANG II-induced sodium appetite contributes to this, complementing the renal sodium-retaining actions of circulating ANG II. A major uncertainty is the relative importance of angiotensin peptides formed in the brain and those in the periphery in arousing the appetite. Most experiments on sodium appetite have been carried out on laboratory rats and mice, and on herbivores such as sheep, cattle, and wild rabbits. Some experiments have been performed on pigeons.

View larger version (15K): [in this window] [in a new window]   FIG. 4.   Top, left (ANG II infusion), mean 24-h 3% NaCl intakes of rats offered continuous access to water and NaCl during intracerebroventricular (ICV) infusion of 0.9% NaCl and last 2 days of a 4-day ICV infusion of ANG II at 6, 60, 600, or 6,000 ng/h. Right (post-ANG II infusion), mean 3% NaCl intakes for 4 days after termination of ANG II infusion in same animals arranged according to rates that they had previously been infused (rates in parentheses). Mean 24-h water intakes of these rats are given in corresponding bottom panels. Results are in ml and are means ± SE, and there were 5 rats in 6 ng/h group and 3 in each of other groups; no rat received more than one rate of infusion of ANG II. [Redrawn from Bryant et al. (69).]

1

1

1

1

1

1

Like the responses to other natriorexigenic challenges, ANG II-induced increases in sodium appetite are generally slow to develop, tend to be persistent, and are affected by previous experience of the conditions accompanying the natriorexigenic stimulus. The sequence of response to ANG II resembles that to any hypovolemic stimulus. The effect of hypovolemia on drinking behavior is an immediate increase in water intake followed by a slowly developing and persistent increase in NaCl intake. This is the desirable sequence of drinking response because in hypovolemia ". . . there is an immediate need for restoration of volume, primarily to maintain the circulation. Water is best suited for this because it is a substance of uniform composition which presents no problem of variable palatability and because the mechanisms controlling its intake are quantitative" (178).

How is the appropriate sequence of initial thirst followed by increased sodium appetite achieved? Part of the delay in onset of sodium appetite may be attributable to the fact that rats maintained on standard laboratory diet have an excess body sodium that acts as a buffer when natriorexigenic stimuli are applied. For example, the onset of NaCl intake was more rapid in rats made hypovolemic by hyperoncotic dialysis (see sect. XC) and which had been maintained on a sodium-deficient diet than it was in rats maintained on a normal sodium-containing diet (563, 571). However, this is unlikely to be the entire explanation for the difference in latencies of onset between ANG II-induced thirst and sodium appetite. Even in the sodium-depleted adrenalectomized rat where the need to take salt was more urgent than in a normal animal, intracranial injection of components of the renin-angiotensin system stimulated water intake before NaCl intake (206). The delay in onset of NaCl intake suggests that for full expression of sodium appetite, synthesis of a neurotransmitter or paracrine agent, modification of receptors, alteration in morphology of the neural connections, or any of these may be necessary. The stimulating effects of mineralocorticoids and ANG II on sodium appetite would accord well with these possible effects, since the hormones are known to stimulate protein synthesis. Denton (122) has suggested that a change in intracellular sodium concentration in the neurons subserving sodium appetite might initiate transcription and synthesis of a specific protein, which would lead to alterations in the ionic or membrane characteristics of the relevant neurons. Although the length of time needed for transcription and protein synthesis may account for some of the features of sodium appetite, especially the delay in onset, a more physiological mechanism of active inhibition of sodium appetite by central nervous pathways involving various neuroactive substances may operate to prevent any diversion of interest away from water until the water intake necessary for restoration of volume is well on the way to being completed. There is some evidence that inhibitory neuroactive substances released in hypovolemia, e.g., oxytocin (see sect. VIII), inhibit sodium appetite until the falling plasma osmolality or sodium concentration as water is drunk brings to an end any further release, leading to increasing acceptability of sodium solutions.

	V. EFFECTS OF ANGIOTENSIN ANALOGS AND ANTAGONISTS ON DRINKING

Top Previous Next References

Different angiotensin peptides are formed by the renin-angiotensin system (see Fig. 1), and for some of these there are specific receptors. Renin and the precursors and the shorter chain analogs of ANG II possess varying degrees of dipsogenic and natriorexigenic activity, and some angiotensin-related peptides act as ANG II antagonists. The effects of renin, angiotensin analog agonists, and antagonists, including the newer receptor subtype selective blockers and antisense oligonucleotides, on thirst and sodium appetite are discussed here. Angiotensin-converting enzyme inhibitors and drinking are dealt with in section IXC.

Of the 10 amino acid residues in the angiotensin decapeptides isolated from different vertebrate groups (Table 3), varying substitutions have occurred in the course of evolution in positions 1, 5, and 9 (260, 261, 405). The angiotensinogen gene is therefore remarkably well conserved. Two angiotensin octapeptides have been identified in mammals, differing by a single amino acid. Human, horse, pig, rat, and guinea pig and probably dog and rabbit have [Ile⁵]ANG II; cow, probably sheep, and nonmammalian vertebrates have [Val⁵]ANG II. In rat (189) and pigeon (152), there was no difference in dipsogenic potency between the two ANG II, and this seems to be true for dog and rhesus monkey, although direct comparisons have not been made. The angiotensin decapeptides of all nonmammalian vertebrate groups examined so far have Val⁵ but show differences from the mammal in position 9 (260, 261). The structural requirements for the dipsogenic activity of ANG II (181, 190) are similar to those for its myotropic and pressor actions (76, 405, 468). The aromatic side groups of Tyr⁴ and His⁶, the guanidino group of Arg², and the hydrophobic phenyl ring and carboxyl of Phe⁸ are involved in binding to the receptor site, whereas Tyr⁴ and Phe⁸ are needed for most biological responses (275). Conversion of ANG I to ANG II by ACE exposes Phe⁸, providing the aromatic ring necessary for myotropic, pressor, and dipsogenic actions. Angiotensin analogs with a COOH-terminal aliphatic amino acid instead of Phe⁸ have very little agonistic activity and act as antagonists of ANG II; the best known of these is [Sar¹,Ala⁸]ANG II or saralasin which has found widespread use in thirst and sodium appetite research. But for intestinal water and sodium absorption, Phe⁸ can be replaced by aliphatic amino acids with retention of full agonistic activity (339).

Other angiotensin analogs of varying dipsogenic effectiveness are also produced. However, their involvement in drinking behavior remains to be established. Cleavage of ANG II by the aminopeptidases present in many tissues including angiotensin receptors yields [des-Asp¹]ANG II or ANG-(2--8), which is also called ANG III. In some circumstances, ANG III is a dipsogenically active peptide. An alternative pathway for ANG III formation is by aminopeptidase action on ANG I to form [des-Asp¹]ANG I, which is then acted upon by ACE. Further processing of ANG II or III from the NH₂ terminus by aminopeptidases, endopeptidases, and carboxypeptidases leads to smaller fragments with progressive loss of dipsogenic activity. The [des-Phe⁸]ANG II, or ANG-(1--7), can also be formed from either ANG I or ANG II by various enzymes that include prolyl endopeptidase, a serine protease present in brain, other peptidases, e.g., enkephalinase, and angiotensinase C, although it is uncertain which of these peptidases is responsible for formation of ANG-(1--7) in vivo (86, 162). Formation of ANG-(1--7) from ANG I by-passes ANG II formation so that it is not blocked by ACE inhibitors. On the contrary, preventing ANG II formation in this way results in greatly increased production of ANG-(1--7). Angiotensin-(1--7) is without any immediate stimulatory effect on drinking, but whether it has longer term or more subtle effects on drinking behavior remains to be investigated.

View larger version (24K): [in this window] [in a new window]   FIG. 5.   Mean water intakes (+SE) of 5 spontaneously hypertensive rats in response to 50 ng ANG II ICV preceded by isotonic NaCl, scrambled oligodeoxynucleotide (SC-ODN), or antisense oligodeoxynucleotide (AS-ODN) for AT1 receptor mRNA. ** P < 0.01. [From Meng et al. (384).]

Renin, TDP renin substrate, and ANG I are effective intracranial dipsogens in rat (79, 181, 190), dog (194, 471), pig (411), and pigeon (153), although there were differences in sensitivities to the different dipsogens in the different species. These substances also stimulate drinking when given systemically in rat, dog, and pigeon. Increased drinking caused by intracranial injection of renin, or the peptide precursors of ANG II, is mediated by ANG II generated in situ by components of a brain renin-angiotensin system (see sect. VIIC). Renin produces a pattern of drinking different from that produced by ANG II, ANG I, or TDP renin substrate. In the rat, it is an exceptionally powerful stimulus to both thirst and sodium appetite (Fig. 6). The onset of drinking after intracranial injection of lyophilized preparations of pig renal renin (10-50 mU) or purified mouse submaxillary gland renin (26.5-265 ng) tends to be delayed by a minute or so compared with ANG II-induced drinking, but the main difference is that renin-induced drinking may last for many hours or indeed days (20, 21). There are species differences in responsiveness to different renins. In rat, intrahypothalamic injections of partially purified rat renin were ~10 times as potent as pig renin in causing drinking (79). Pig renin was an effective intracranial dipsogen in the pigeon, but it was less potent than in the rat (153). After causing an immediate thirst, injection of purified renin (1 ng) into the third ventricle of the pigeon caused a delayed increase in sodium appetite (151).

View larger version (18K): [in this window] [in a new window]   FIG. 6.   Mean daily intakes (in ml; ±SE) of 5 rats offered water only (left) and 5 rats offered water and 2.7% NaCl (right) in response to a single injection of 26.5 ng purified renin into preoptic region. All injections were first injections. * P < 0.05, *** P < 0.001 compared with day preceding injection by paired t-test. Purified renin was prepared from mouse submaxillary gland by Drs. T. Inagami and K. Misonoby and was generously donated by them. [Modified from Avrith and Fitzsimons (21).]

Drinking in response to peptide precursors of ANG II is indistinguishable from that caused by ANG II itself. In the dog, the relative dipsogenic responsiveness to TDP renin substrate, ANG I, or ANG II, injected intracranially (1-1,000 pmol) or infused through a vein (250 pmol/min) or through one carotid artery (40 pmol/min), were similar (194, 195). Renin (0.05-0.5 U/min) infused intravenously was an effective dipsogen, but when renin was infused at the higher rate through one carotid artery, the amounts drunk were much less (195). Regardless of whether it was given by the intracranial or systemic route, ANG II caused the most drinking, especially in the dog. With intracranial administration of doses of 100 pmol or more, the differences in dipsogenic effectiveness between TDP renin substrate and ANG I and ANG II tended to disappear. After equipotent doses, the temporal pattern of drinking was similar for the different peptides, and drinking was mostly completed within 15 min of injection. The relative dipsogenic effectiveness (Fig. 7) and temporal pattern in response to intracranial injection were similar in the pigeon (153). A tridecapeptide renin substrate was less effective than TDP renin substrate as an intracerebroventricular dipsogen in the rat (625). In normal (21) or adrenalectomized (206) rats, intracranial injection of renin or TDP renin substrate caused a delayed increase in NaCl intake as well as water within an hour of injection when both fluids were available to drink. The stimulating effects were large and not secondary to renin-induced urinary losses.

The two angiotensin heptapeptides, ANG-(1--7) and ANG-(2--8), are quite different in their effects on drinking. The thirst-stimulating effect of angiotensin peptides depends critically on the presence of Phe⁸; very large doses (up to 100 nmol) of ANG-(1--7) given by intracranial injection were almost devoid of dipsogenic activity in the rat (104, 181, 190, 350, 584) (see Fig. 9) and pigeon (see Fig. 7) (153). Apart from it having no stimulating effect on drinking, ANG-(1--7) does not cause aldosterone release, and it has no myotropic or direct pressor actions (468). However, it has been shown to possess excitatory actions on some nervous and other tissues. Angiotensin-(1--7) caused increased PG synthesis in astrocytes and C6 glioma cells with a potency equal to or greater than ANG II (162). In the case of ANG-(1--7)-induced PG synthesis, there was no associated rise in intracellular calcium as there was after ANG II, and the effect could be blocked by the AT₂ antagonist CGP-42112A but not by the AT₁ antagonist losartan, whereas the ANG II effect on PG synthesis was blocked by both receptor antagonists. Even though a few results with angiotensin antagonists suggest that AT₂ receptors may mediate some ANG-(1--7) effects, the grounds for this conclusion are weak. Angiotensin-(1--7) applied microiontophoretically to the PVN excited most neurons, but more weakly than ANG III or ANG II, and the excitatory effects were inhibited by lower concentrations of AT₁ antagonists than of AT₂ antagonists (8). In the brain, ANG-(1--7) competes less effectively than ANG II for AT₁ receptors and even less effectively for AT₂ receptors (496). If ANG-(1--7) acts at AT₂ (or AT₁) receptors, it would have to possess very high efficacy to compensate for the low affinity. It is unlikely, therefore, that ANG-(1--7)-induced responses are mediated by these receptors. Some other as yet uncharacterized angiotensin receptor type may be involved.

View larger version (30K): [in this window] [in a new window]   FIG. 9.   Top: amounts of water and 1.8% NaCl drunk by Lister hooded rats in 1 h after ICV injections indicated. Bottom: number of c-fos positive nuclei in defined areas of 104 µm2 in OVLT, ventral MnPO nucleus, dorsal MnPO nucleus, and SFO after similar ICV injections in separate groups of rats that were not allowed to drink. Results are means + SE, with number of rats given in parentheses. * P < 0.05, ** P < 0.01 compared with saline control (top) or corresponding areas of ANG II-treated rats (bottom) by analysis of variance with post hoc Sheffé test. [Redrawn from Mahon et al. (350).]

Angiotensin-(1--7) caused increased release of substance P immunoreactivity from a superfused preparation of hypothalamic tissue (132). It was equipotent with ANG II for vasopressin release from rat hypothalamo-neurohypophysial explants (520), although ANG-(1--7) failed to release vasopressin into the bloodstream of the conscious rat when injected into the lateral cerebral ventricle at 1,000 times the effective molar dose of ANG II (350). It elicited depressor and cardioinhibitory responses similar to those caused by ANG II when microinjected into the dorsal vagal complex of the medulla oblongata of anesthetized rats (85). Angiotensin-(1--7) could lower the blood pressure by competing for ANG II at a pressor site; in the periphery, ANG-(1--7) is an antagonist of AT₁ receptors (351), or it could act by releasing PGs in the brain or periphery (236).

Although intracranial ANG-(1--7) has no immediate stimulatory action on thirst or sodium appetite, its actions on PG synthesis are of potential interest because PGs of the E series inhibit thirst, especially ANG II-induced drinking, and it has been suggested that these substances act as natural satiety factors responsible for terminating bouts of drinking (see sect. VIIIK). The same remarks could apply to the effect of ANG-(1--7) on substance P, which in rat is inhibitory to ANG II-induced drinking (see sect. VIIIE). An attractive speculation is that the intensity of thirst (or sodium appetite) depends on a balance between production of stimulatory factors such as ANG II (or ANG III) and inhibitory factors such as ANG-(1--7) (among other potential candidates) and that the relative amounts of these produced change as drinking progresses. Another possibility is that ANG-(1--7) has both stimulatory and inhibitory effects that cancel out. However, there is no evidence at present of an inhibitory or other role in drinking behavior for ANG-(1--7), and these ideas remain speculative.

In contrast to the lack of any short-term stimulatory effect of ANG-(1--7) on drinking, ANG III given by intracranial injection retains considerable dipsogenic activity in the rat (181, 190), pig (411), and pigeon (see Fig. 7) (153), but molar doses must be increased more than 10 times to give water intakes similar to those produced by ANG II in the dose range of 1-100 pmol, and ANG III was ineffective when injected intravenously. It was also a much weaker intracranial, intravenous, and intracarotid dipsogen than ANG II in the dog (194, 195).

It has been argued that ANG III is the biologically significant angiotensin ligand, formed from ANG II before it reacts with angiotensin recognition sites, and that it is responsible for the pressor and dipsogenic activity of ANG II much as it is for ANG II stimulation of aldosterone secretion (673, 678). Angiotensin III binding is tighter and more widely distributed in various brain regions in all species so far examined than is the case for ANG II. In the rat, the dipsogenic and pressor potencies of ANG III and ANG II were similar at intracranial doses below ~10 pmol, and in the gerbil, water intakes were larger after ANG III than after ANG II (679). The dipsogenic responsiveness to ANG II in different species is better correlated with the distribution of binding sites for ANG III than for ANG II. After microiontophoresis of ANG III, neuronal firing rates in the SFO (see sect. VI) of the cat (160) and PVN of the rat (254) were higher than after ANG II. Neuronal firing in the PVN of rat in response to iontophoretically applied ANG II was inhibited by coapplication of amastatin, an aminopeptidase A inhibitor that prevents the conversion of ANG II to ANG III, but the response to ANG III itself was little affected (255). Neuronal firing in response to iontophoretically applied ANG II and ANG III was enhanced by coapplication of bestatin, an aminopeptidase B inhibitor. Bestatin prevents degradation of ANG III to ANG-(3--8) so that this result is also consistent with the idea that ANG III is the important ligand. Neither bestatin nor amastatin affected neuronal firing by itself, although intracerebroventricular bestatin was stated to be dipsogenic (673).

It has been suggested that the greater effectiveness of higher doses of ANG II on drinking, compared with ANG III, depends on the more stable ANG II providing a continuous supply of the more rapidly degraded ANG III; the half-life of ANG III in the lateral ventricle is 6.5 s contrasted with 22.5 s for ANG II (673, 678). At lower doses, the greater stability of ANG II is offset by the greater affinity of angiotensin binding sites for ANG III. The ANG II receptor antagonist saralasin was equally effective at reducing pressor and drinking responses induced by injections of ANG II and ANG III into the lateral ventricle, which could mean that either the two peptides act at a common site or ANG II must be converted to ANG III to serve as the ligand for angiotensin-induced drinking. However, when the peptides were continuously infused into the lateral ventricle instead of being given by bolus injection, water intakes were larger after ANG II than after ANG III in Sprague-Dawley, SH, and WK rats (680). It has also been found that ANG II is a more powerful dipsogen than ANG III when injected into the PVN of the rat (287). It remains to be seen whether ANG III or other angiotensin peptides act as natural ligands instead of or in addition to ANG II.

In intracranial injection experiments in the rat (181, 190, 626) and pigeon (see Fig. 7) (153), it was found that loss of Arg² from ANG III to give ANG-(3--8), also known as ANG IV, resulted in a substantial reduction in dipsogenic activity; the importance of Arg² is also shown by the fact that the D-Arg²-substituted octapeptide was found to be inactive (181). Angiotensin IV binds to AT₄ sites that are pharmacologically distinct from AT₁ and AT₂ sites, having low affinity for ANG II and III and for saralasin. The AT₄ sites (Table 4) are expressed in the cerebral cortex, hippocampus, thalamus, basal ganglia, habenula, lateral geniculate, colliculi, central gray, nucleus accumbens, and cerebellum, overlapping with the AT₁ and AT₂ distribution but with differences. The distribution of AT₄ sites in the brain suggests that ANG IV might be involved in learning, memory, and exploratory behavior, possibly being responsible for effects on these functions normally ascribed to ANG II. Whether this has a bearing on some aspects of drinking behavior remains to be investigated.

The pentapeptide ANG-(4--8) and the tetrapeptides ANG-(5--8) and ANG-(1--4) showed little dipsogenic activity in the rat, and then only at much higher dosage than for ANG II (181, 626), and the pigeon (153). In structure-activity studies with angiotensin fragments, the dipsogenic responsiveness to the different fragments (181, 626) was generally similar to their iontophoretic responsiveness, the COOH-terminal sequence being necessary for receptor recognition (160). It is noteworthy that in pigeon, ANG-(1--4) which has tyrosine, an aromatic amino acid similar in structure to phenylalanine, at the COOH-terminal end of the molecule, caused more drinking than ANG-(1--7), which has the aliphatic proline.

	VI. CENTRAL NERVOUS SYSTEM AND ANGIOTENSIN-INDUCED DRINKING

Top Previous Next References

Because ANG II is a powerful dipsogen when injected directly into the brain, it is generally accepted that circulating ANG II stimulates drinking behavior mainly by acting directly on sensitive structures in the central nervous system. It is also becoming increasingly evident that angiotensin peptides generated in situ in the central nervous system may affect drinking behavior by acting on sensitive neurons, some of which may be inaccessible to blood-borne peptide. Evidence just discussed introduces the possibility that analogs of angiotensin other than ANG II may also be involved. The view that the brain monitors the state of the body fluids goes back a long way. Bernard (40) believed that the central nervous system was responsible for maintaining the constancy of the milieu intérieur and that the sensation of thirst played a major part in achieving this by enabling fluid intakes to be matched to fluid losses, but neither he nor his contemporaries were able to state how this was done. The idea of specific receptors in the central nervous system to detect and measure particular aspects of the body's hydrational state owes much to Verney's (637) work on the factors that determine the release of antidiuretic hormone. The concept of hypothalamic osmoreceptors that register cellular water content and control vasopressin release and drinking (666) in response to cellular dehydration is generally accepted, although the anatomic identity and location of such receptors are elusive, and questions concerning their relation to possible sodium-sensitive receptors remain to be explored (54, 55, 336, 373).

Early evidence for the existence of ANG II-sensitive neurons in the central nervous system accessible to circulating peptide was Bickerton and Buckley's (41) finding, published in 1961, that ANG II injected into the arterial supply of the head of a dog in a cross-circulation experiment caused an increase in arterial blood pressure in the rest of the body even though the only connection between head and body was through nerves. In other pioneering experiments at about this time, it was found that pressor responses to infusion of small amounts of ANG II into the vertebral artery of the conscious rabbit exceeded those for intravenous infusion (131). Since then, a large number of behavioral, electrophysiological (158), binding (543, 544), and mapping studies (342) have shown that there are ANG II-sensitive neurons in various parts of the brain. Angiotensin II is a polar molecule unable to penetrate the blood-brain barrier easily. However, many of these neurons were found to be present in highly vascularized nervous structures, the CVOs, where the blood-brain barrier is deficient and that are therefore accessible to circulating polar molecules such as ANG II, although they themselves are isolated from the rest of the brain by other barriers. The neurons of the SFO and OVLT seem to be mainly responsible for drinking in response to circulating ANG II, although the AP may also be involved. Some angiotensin-sensitive structures involved in thirst and sodium appetite lie inside the blood-brain barrier and cannot be reached directly by circulating ANG II. One of the most important of these is the MnPO nucleus in the lamina terminalis. Others are the PVN (287), POA (149, 194), and the central gray of the midbrain (585) to which the POA projects. It is possible that neurons in these structures are stimulated by angiotensin peptides generated in the brain (see sect. VII). In the case of the MnPO nucleus that is close to and has extensive connections with the OVLT, there may be uptake of blood-borne hormone by specialized cells such as tanycytes. The ventral lamina terminalis and surrounding tissue are a crucial part of the limbic system and are a focus for the integration of body fluid homeostasis and blood pressure control.

Immunohistochemical identification of the protein products of inducible transcription factors or immediate-early genes (IEGs) such as c-fos and c-jun (274, 326) has been used to map the pattern of activation of neurons in the basal forebrain following various dipsogenic and natriorexigenic stimuli including ANG II. The IEGs are expressed by neurons in response to growth factors, neurotransmitters, or other stimuli, and the protein products of their expression may influence transcription of other genes and function in a variety of signal transduction systems. Intracranial injection of ANG II into conscious rats caused intense c-fos expression in a band of neurons extending through the tissue surrounding the anteroventral third ventricle or AV3V region, including the OVLT and MnPO nucleus (263). There were also high levels of expression in the SFO, SON, and PVN, the lateral division of the central nucleus of the amygdala and the bed nucleus of the stria terminalis (BNST). It seems that intracranial ANG II induced highly localized expression of c-fos in areas known to be involved in ANG II-induced drinking. Anatomic differences in c-fos expression were found after injection of ANG II, carbachol, or ANG-(1--7) into the lateral cerebral ventricle (350). The patterns of expression seemed to be related to the differences in drinking behavior induced by the three ligands. Of the two dipsogens, ANG II caused intense c-fos expression in the AV3V region including the OVLT and MnPO nucleus, whereas carbachol caused less expression in the AV3V region but more in the SON and PVN (Figs. 8 and 9). Angiotensin-(1--7), which is nondipsogenic in rats even in large doses (see sect. VD), caused a small amount of staining in the AV3V region. Angiotensin II-induced Fos-like immunoreactivity was prevented by the concurrent administration of losartan, whereas carbachol-induced Fos was mainly unaffected by this except in the MnPO nucleus (503).

View larger version (183K): [in this window] [in a new window]   FIG. 8.   C-fos expression after ICV injection of 100 pmol ANG II (1), 500 ng carbachol (2), 1,000 pmol ANG-(1--7) (3), or cerebrospinal fluid (CSF) (4) in SFO (a), dorsal median preoptic (MnPO) nucleus (b), ventral MnPO nucleus (c), and organum vasculosum of the lamina terminalis (OVLT) (d). a.c., anterior commisure. [From Mahon et al. (350).]

Conscious rats infused intravenously with ANG II (30-55 pmol·kg¹·min¹), giving physiological increases in plasma levels of ANG II, also showed increased c-fos expression in the SFO and OVLT and in many other hypothalamic and medullary sites, including the BNST, SON, magnocellular and parvocellular PVN, central nucleus of the amygdala, AP, and NTS (374, 429). The pattern was similar after subcutaneous injection of ANG II, with staining in the SFO, OVLT, MnPO nucleus, SON, and PVN, but in these experiments, the amygdala was not stained (324). Sodium depletion, mineralocorticoids, and other causes of increased sodium appetite also cause increased c-fos expression especially in the OVLT as described later (see sect. VID). In rabbit, intravenous infusion of ANG II caused significant increases in the numbers of Fos-positive cell nuclei in the SFO and OVLT, but unlike in the rat, intravenous ANG II causes little drinking (23). Fos staining of neurons in the SFO and OVLT after intravenous ANG II is consistent with the role of ANG II in hemorrhage, since these neurons also show increased c-fos expression after hemorrhage, although this could also be a response to synaptic inputs from elsewhere in the brain (24). Hemorrhage also resulted in c-fos expression in the lamina terminalis, SON, PVN, AP, and NTS.

It is unlikely that c-fos expression is an essential part of ANG II-induced water intake, since the drinking response is largely completed within 15 min of injection, whereas Fos protein is hardly detectable by the end of this time, taking ~60-120 min to reach maximal levels. Herbert (262) has pointed out that induction of c-fos, a transcriptional regulator which then must alter expression of target genes, can hardly be a necessary component of the rapid dipsogenic action of ANG II, and indeed, it is possible to dissociate ANG II-induced drinking and c-fos expression (see sect. VIIIL). These remarks may not apply to the much slower process of ANG II-induced NaCl intake, but other evidence also suggests that the relation between Fos and the act of drinking may be quite loose. When the rats drank water immediately after intracerebroventricular ANG II, c-fos expression was markedly reduced in the SON and magnocellular PVN but not when they drank 0.9% NaCl, which suggests that these structures respond to changes in osmolality rather than to the act of drinking. The high level of ANG II-induced c-fos expression in the AV3V region was, however, unaffected by the fluid drunk. Preloading rats with water through an indwelling gastric cannula prevented rats from drinking in response to ANG II, but the effects on c-fos expression were the same as when rats were allowed to drink; c-fos expression was reduced in the SON and PVN, but there was no detectable effect on expression in the MnPO nucleus, SFO, or OVLT (682). Mapping expression of c-fos and other IEGs allows identification of individually active neurons, possibly provides a measure of second messenger levels, and gives an overall view of the pattern of neural activation in response to ANG II. It is, however, an empirical technique.

The CVOs contain neural elements that have uniquely close contacts with blood and CSF and that connect with other brain regions, enabling the brain to monitor the body fluids in the periphery while at the same time it is protected by the blood-brain barrier against major homeostatic imbalances and harmful substances reaching it in the bloodstream (242). Circumventricular organs are richly vascularized with fenestrated capillaries that allow blood-borne substances such as ANG II to reach and stimulate afferent nerve endings in the pericapillary spaces. They are usefully thought of as lying outside the blood-brain barrier because there are further impediments to the free diffusion of substances into brain tissue beyond them (Fig. 10). Three CVOs, the OVLT, SFO, and AP, have been implicated in body fluid homeostasis and are rich in ANG II receptors. There is much evidence that they play important roles in the control of drinking behavior, renal function, and blood pressure. The AP was regarded as the likely target tissue for the neurogenic pressor response to blood-borne ANG II, and by analogy, it was suggested that the SFO might be concerned in systemic ANG II-induced drinking (182). The importance of the SFO in ANG II-induced drinking was quickly established (539), and early evidence had already implicated the OVLT in body fluid homeostasis (416). The CVOs contain ANG II-sensitive receptors that can be reached and stimulated by ANG II in the bloodstream, although there are also ANG II-sensitive neurons inside the blood-brain barrier that may normally be stimulated by ANG II formed locally in the brain. Angiotensin II may stimulate neurons directly, although some of the evidence has been interpreted in terms of an indirect action secondary to local vasoconstriction (418).

View larger version (67K): [in this window] [in a new window]   FIG. 10.   Model of a circumventricular organ (CVO) with a glial barrier. In this case, model is of OVLT that is separated from adjacent MnPO nucleus. Ependyma over OVLT has tight junctions, and substances can only move into OVLT from CSF by uptake at tanycytes. Plasma ANG II has access to OVLT receptors (r), but glial barrier prevents diffusion into MnPO nucleus (x at arrowhead). Cells containing endogenous brain ANG II have been found in OVLT. [From Phillips in Gross (242).]

The CVOs are covered on their cerebral ventricular surface by nonciliated ependymal cells (436). The tight junctions between the cells of the overlying ependyma insulate the CVOs from the CSF; they are absent from the ependyma elsewhere in the ventricles. The blood-brain barrier can be thought of as being shifted to the ependymal layer, creating a blood-CSF barrier to free diffusion (243). Within the ependymal layer are specialized ependymal cells or tanycytes that send processes into the pericapillary spaces of the CVOs. The tanycytic processes sealed by tight junctions isolate the pericapillary spaces from the adjacent brain parenchyma, forming another barrier to free diffusion. The high vascularity and capillary leakiness of CVOs together with the barriers to diffusion that surround them result in the blood-borne messenger molecules being contained within the neuropil, a possible way of improving the efficiency of hormonal transduction. However, ependymal cells show secretory and transport characteristics, and the tanycytes themselves are in contact with both CSF and blood. It is possible to envisage exchange of substances between the CSF and pericapillary space so that receptors could be accessible to both CSF-borne and blood-borne ANG II; for example, ANG II injected or generated inside the blood-brain barrier could be transported by the tanycytes to sensitive structures such as the OVLT outside the barrier and stimulate drinking by acting on these structures. Similarly, circulating ANG II may reach and stimulate the MnPO nucleus and other sensitive structures inside the barrier.

The tissue surrounding the anteroventral third ventricle in the anterior hypothalamus and ventral portion of the lamina terminalis, the AV3V region, plays a key role in body fluid homeostasis. The lamina terminalis contains three structures the OVLT, MnPO nucleus, and, at its upper boundary, the SFO, all of which are extremely sensitive to the dipsogenic action of ANG II. It has extensive connections with the hypothalamus and other limbic structures that allow information about the milieu intérieur to be processed and translated into appropriate behavior. Andersson (9) produced unequivocal evidence that the anterior hypothalamus is important for controlling drinking behavior. Nicolaïdis (416) in an electrophysiological study in the cat was the first to suggest that structures in the median and anterior hypothalamus in the region of the suprachiasmatic nucleus serve as a receptive and integrative center for cardiovascular and body fluid regulation. Using concentric bipolar microelectrodes, he found that there was increased integrated unit activity in this region in response to a fall in blood pressure produced by either distension of the carotid sinus or removal of blood. Neurons responding to hypovolemia have been found at a "carrefour" of structures responding to blood volume and pressure, osmolality, temperature, and gustatory signals in the cat and rat (416, 607). Iontophoretic application of ANG II, vasopressin, ANP, and mineralocorticoids, hormones involved in the regulation of fluid and electrolyte balance, influences the activity of neurons located here (609, 610). Neurons sensitive to the iontophoretic application of ANG II and aldosterone have been identified in the medial septum and in and around the MnPO nucleus, and these neurons show greater and more prolonged responses after pretreatment with systemic DOCA (611). This could be the electrophysiological basis for the synergism between ANG II and mineralocorticoids in arousing sodium appetite (207).

Lesioning evidence supports the view that the AV3V region serves as an integrative center for cardiovascular and body fluid regulation (71, 293, 343-345). Lesions in rat resulted in attenuated pressor responses to centrally acting agents, failure to develop most forms of experimental hypertension, and impaired function of fluid intake and fluid output controls, leading to severe dehydration and hypernatremia. The effects of different thirst challenges were variably attenuated depending on the stimulus, but drinking in response to ANG II by intracranial or subcutaneous injection or to intracranial renin was abolished (73, 120, 292). The AV3V region contains the OVLT and MnPO nucleus just dorsal to it, and it continues upward into the anterodorsal lamina terminalis that contains the SFO. The OVLT and SFO may be important for drinking responses to blood-borne ANG II, but the presence of rich connections between them and the MnPO nucleus makes it difficult to quantify their respective contributions (449). The MnPO nucleus is crucial for ANG II-induced drinking responses, and there are extensive two-way connections between it and the many adjacent structures, including the SFO, OVLT, anterior hypothalamus, periventricular preoptic region, and the medial and lateral POAs (Fig. 11). These are the structures that show intense c-fos expression after ANG II (see sect. VIAA). It is claimed that injection of ANG II into MnPO nucleus near the OVLT causes increased intakes of both water and NaCl, but injection into the SFO causes only water intake (173). However, damage to the SFO may impair sodium depletion-induced sodium intake.

View larger version (40K): [in this window] [in a new window]   FIG. 11.   A: a midsagittal section of rat brain showing most of afferent and efferent projections of lamina terminalis (stippled area). Direction of pathways is indicated by arrows, and reciprocal projections are denoted by double arrows. B: Angiotensinergic (solid lines) and catecholaminergic (dotted lines) pathways projecting to and from lamina terminalis. Note presence of intrinsic angiotensinergic neurons in SFO. Arc, arcuate nucleus; BNST, bed nucleus stria terminalis; Cg, cingulate cortex; CG, central gray; CM, central medial thalamic nucleus; DG, dentate gyrus; DMH, dorsal medial hypothalmus; DR, dorsal raphe; DTg, dorsal tegmentum; IL, infralimbic cortex; LC, locus ceruleus; LH/PFA, lateral hypothalamus perifornical area; LPBN, lateral parabrachial nucleus; LS, lateral septum; MnPO nucleus, median preoptic nucleus; MnR, medial raphe; MPO, medial preoptic nucleus; MS, medial septum; NTS, nucleus solitary tract; OVLT, organum vasculosum lamina terminalis; PV, paraventricular nucleus of thalamus; PVN, paraventricular nucleus hypothalamus; Re, reuniens nucleus; SI, substantial innominata; SON, supraoptic nucleus; VLM, ventrolateral medulla; ZI, zona incerta. [From Oldfield in Ramsay and Booth (462).]

From the anterior third ventricle region and anterior hypothalamus there are projections to the SON and PVN and a pathway courses down to the brain stem. There are numerous projections to other parts of the limbic system. The OVLT, MnPO nucleus, and SFO contain moderate- to high-density ANG II immunoreactive neurons, the distribution of which corresponds well with ANG II binding sites (342). The AV3V region and especially the MnPO nucleus also receive ascending noradrenergic and other inputs from cell groups in the NTS and the various structures in the brain stem that are concerned in monitoring blood volume and pressure (289).

Angiotension II-sensitive neurons have also been identified in the walls of the cavity of the septum pellucidum (417). These neurons line a hollow sagittal structure that in the rat consists of a horizontal horn lying in the septum below the corpus callosum and a vertical horn that extends down from the genu, overlying the anterior aspect of the lamina terminalis as far as the OVLT. The cavities of the two horns communicate freely with each other but not with the subarachnoid or cerebroventricular spaces. The structure, called the organum cavum prelamina terminalis or OCPLT, is extremely responsive to the dipsogenic action of ANG II injected into it. When water and 3.0% NaCl were offered, water was drunk but the NaCl intake was unaffected. The effect was blocked by losartan but not by CGP-42112A so that the angiotensin receptor type is AT₁, the same as for angiotensin-induced drinking responses elsewhere (146). It is possible that blood-borne ANG II may reach and stimulate the OCPLT through a highly vascular dorsoventral network that lines its vertical horn. The significance of the OCPLT in ANG II-induced drinking and its separate identity from other angiotensin-sensitive structures in the vicinity have yet to be established.

The OVLT is a remarkably vascularized structure possessing fenestrated capillaries, numerous glial, tanycytic and neural processes, small neurons, and many peptidergic nerve terminals. It lies in the ventral portion of the lamina terminalis in the tissue surrounding the AV3V region. The OVLT in rat possess high-affinity ANG II binding sites (383), and it has been shown by microiontophoresis to contain angiotensin-responsive neurons (158). High-affinity binding sites are also found in the OVLT of the dog (552) and sheep (376). Although it is clear that the AV3V region is involved in ANG II-induced drinking, the part played by the OVLT itself in this response is less certain. Tissue in the region of the AV3V or suprachiasmatic recess is extremely sensitive to the dipsogenic action of ANG II (418, 448, 449). Early experiments showing reduced drinking and pressor responses to intracerebroventricular ANG II after covering the ventricular surface of the OVLT with cream were interpreted as meaning that the OVLT responds to changes in the levels of ANG II in the CSF (266). However, injections into the AV3V seem as likely to affect structures inside the blood-brain barrier as the OVLT, which is protected on its ependymal surface by tight junctions. The MnPO nucleus is an obvious target because it is rich in angiotensinergic nerve terminals and ANG II-sensitive cells (342). In the rat (149) and dog (194), injection of ANG II restricted to the POA, i.e., adjacent to the MnPO nucleus and inside the blood-brain barrier, elicits drinking. However, the OVLT shows intense c-fos expression after intracranial as well as systemic administration of ANG II so that it is clearly influenced by ANG II in the CSF (see sect. VIA). After the peripheral natriorexigenic challenges of subcutaneous implantation of DOCA pellets, or dietary or furosemide-induced sodium depletion, the increased c-fos expression was more restricted than after ANG II, being confined mainly to the OVLT (324). This suggests that it may play a key role in sodium appetite in the rat even though the relation between NaCl intake and c-fos in the OVLT was not quantitative (94). Other structures in the central nervous system are also involved.

Lesions of the AV3V region eliminated drinking in response to peripheral ANG II and isoproterenol in the rat (343, 345), but a more restricted lesion confined to the OVLT, or infusion of sarthran into it, did not affect isoproterenol-induced drinking (172). In sheep, near-complete ablation of the anteroventral wall of the third ventricle, including the OVLT and adjacent tissue but usually leaving the SFO intact, abolished drinking in response to intracarotid infusion of ANG II (378, 379). In the dog, lesions confined to the tip of the optic recess, destroying most of the OVLT, resulted in markedly reduced drinking in response to intravenous infusion of ANG II (612). The obvious interpretation, that drinking fails because of loss of a specific role for the OVLT or MnPO nucleus in ANG II-induced drinking, may be too simple. The lesion could have destroyed vital neural pathways from the SFO or other structures involved in drinking behavior, although this is unlikely in view of the widespread distribution of connections between the CVOs and hypothalamic structures involved in water balance. A more likely reason for variable results in lesion experiments is that the OVLT and/or MnPO nucleus serve several functions and respond to more than one stimulus. For example, in dogs, drinking and vasopressin release in response to systemic hyperosmotic NaCl were disrupted by lesions of the OVLT (613), which suggests that the OVLT is an important site for osmoreceptors. Similar results were obtained in sheep after large lesions of the anterior wall of the third ventricle (377), and in rat after AV3V lesions. Loss of osmoreceptor function after lesioning might be expected to have the same effect as a water load because in each case the stimulatory nerve outputs from the osmoreceptors that cause drinking and vasopressin release would be reduced. Failure of intravenous ANG II-induced drinking and vasopressin release after ablation of the OVLT might therefore have been a consequence of loss of osmoreceptors rather than of specific angiotensin receptors.

Although the OVLT is a major site for osmoreceptors, osmosensitivity is more widespread than this (54, 55). There appears to be significant osmosensitivity in the SFO of the rat, since ablation results in some reduction in osmotically induced water intake (271) and marked disruption in vasopressin release (353). Earlier experiments on drinking responses to intracranial injections of hyperosmolal solutions in conscious rats indicated that osmosensitivity was diffusely distributed in the basal forebrain region, POAs, and anterior parts of the hypothalamus (49, 442). Osmosensitive cells have been found in the hypothalamus of conscious monkeys injected with hypertonic NaCl, and the firing rate of these cells decreased when the monkey drank water (642). Magnocellular neurons in the SON are sensitive to osmotic stimulation (336), and this is confirmed by selective c-fos expression in response to drinking solutions of different osmolalities. It is scarcely possible to avoid damaging osmosensitive tissue when lesions are made in an angiotensin-sensitive region containing osmosensitive cells, and it could be argued that it is the loss of tonic discharge from osmoreceptors that interferes with ANG II-induced drinking rather than the loss of ANG II-sensitive cells per se. However, even though a mistakenly perceived hyposmolality after ablation of one or other CVO might account for failure to drink, this does not rule out the possibility that the intact CVO responds both to increases in circulating ANG II and to plasma hyperosmolality.

Of the three CVOs that have been implicated in ANG II-induced drinking behavior, the SFO is thought to be the principal site for the stimulating effect of systemic ANG II on water intake in rat and dog, although the OVLT may also be important. Morphological and metabolic changes occur in the SFO during dehydration (576). Routtenberg and Simpson (494) found that direct application of carbachol to the SFO caused increased water intake in the rat. Because there are negligible amounts of circulating acetylcholine, this result has no direct bearing on SFO responses to blood-borne molecules, but the SFO was also found to be very sensitive to the dipsogenic action of ANG II, responding to femtomole doses injected directly into it (537, 539). It is a matter for speculation whether there would have been the dual response of increased intakes of NaCl as well as water had both fluids been available to drink as has been described for AV3V injections of ANG II (173). However, the SFO may be implicated in sodium appetite in the rat because NaCl intake after furosemide-induced sodium depletion was decreased by damage to the SFO, although other brain areas are clearly involved because low dosage of captopril, which increases the ANG I reaching the brain (see sect. IXC), prevented the decrease (656).

The SFO protrudes into the third ventricle at the level of the interventricular foramina of Monro where the crura of the fornix begin to diverge. It is derived from the ependyma at the boundary of the tela choroidea and lamina terminalis. It has extensive reciprocal connections with the MnPO nucleus and OVLT in the AV3V region and, like the OVLT, it sends projections to the SON and PVN (119, 241, 390). The SFO is composed of a variety of small neurons and their processes, glial cells, and tanycytic ependymal cells. It contains a close network of fenestrated capillaries which in places appear to merge with the nearby choroid plexus. The fenestrations make the neurons in the perivascular spaces highly accessible to blood-borne ANG II and other circulating molecules. The SFO contains a dense plexus of ANG II immunoreactive nerve fibers, including an angiotensinergic input from the lateral hypothalamus, and angiotensinergic efferents projecting to the MnPO nucleus, OVLT, BNST, infralimbic cortex, PVN, and further afield to the midbrain raphé system (342). Angiotensin II-responsive neurons, identified by measuring neuronal discharge after the direct or microiontophoretic application of ANG II, or in response to intravenous ANG II, are present in the SFO (158). Discharge by these cells was blocked by saralasin, whereas other less selective neurons that responded to acetylcholine as well as to ANG II were unaffected. Because spontaneous firing of these neurons fell when saralasin was applied, it has been suggested that their activity depends on tonic release of angiotensin peptides derived from the action of brain renin, but this is uncertain.

Destruction of the SFO results in the loss of drinking in response to intravenous ANG II in rat (1, 540), dog (615), opossum (165), and Japanese quail (592) and to intraperitoneal ANG II in pigeon (359, 360). But in sheep, the SFO does not appear to be involved in drinking in response to intracarotid infusion of ANG II (378, 379). In the rat, destruction of the SFO or infusion of sarthran into it also caused a fall in the amount drunk in response to isoproterenol (172). On the face of it, blood-borne ANG II-induced drinking is mediated mainly by the OVLT and adjacent tissue in sheep and by the SFO in rat, opossum, quail, and pigeon. In the rat, both CVOs show intense c-fos expression after intracranial and systemic administration of ANG II (see sect. VIA), and this was also true of the rabbit, an equivocal responder to the dipsogenic effect of ANG II (see sect. III). In the dog, ablation of the OVLT also attenuated intravenous ANG II-induced drinking (612) so that compatible with the present evidence is a role for both the SFO and OVLT in dog. However, as already discussed, the obvious interpretation, that drinking fails after ablation of a CVO because ANG II-sensitive receptors have been destroyed, is not necessarily the entire explanation. The CVOs subserve other functions, and it could be that the loss of these functions accounts for the reduction in ANG II-induced drinking. In the case of OVLT lesions, we saw (sect. VID) that loss of the drinking response to intravenous ANG II could have been caused partly by a mistaken perception of hyposmolality owing to destruction of osmoreceptors, although this does not rule out the possibility that there are also ANG II-induced responses from the intact OVLT. The SFO of rat is also osmosensitive, but the argument of dilutional inhibition to explain the loss of drinking after destruction of the OVLT can be discounted in the case of SFO lesions. This is because saralasin applied directly to the SFO of rat attenuated drinking in response to intravenous ANG II in a dose-dependent fashion, without affecting drinking in response to peripheral hypertonic NaCl (537). A similar type of selective blocking experiment on the OVLT against circulating ANG II has not been described. The result might be different from the SFO result because infusion of sarthran into the OVLT did not affect isoproterenol-induced drinking (172).

Much evidence, most of it from the rat, suggests that the amygdala plays an important part in sodium appetite, including that induced by ANG II. It is ideally suited to this purpose because it receives olfactory, gustatory, and visceral inputs and has extensive connections with the preoptic, hypothalamic, and hindbrain regions (122, 523). Bilateral lesions of the amygdala result in exaggerated affective responses and a general perceptual deficit so that the rats fail to be alerted by novelty (413). In the case of sodium appetite, absence of neophobia means a deficiency in the learning of taste aversions. Chiaraviglio (95) found that damage to the amygdala, BNST, or ventral amygdalofugal pathway impairs or abolishes the increased NaCl intake that follows sodium depletion induced by peritoneal dialysis. It has been suggested that the amygdala and BNST were once a single structure that has become separated by the internal capsule (292). The idea of an "extended amygdala" makes it reasonable to envisage an integrative function in sodium appetite with, for example, the possibility of both stimulatory and inhibitory roles. Multiple roles could explain some of the conflicting results that have been obtained.

The amygdala is also one of the regions of the brain with the highest uptake of aldosterone, a property it shares with the septum and hippocampus (see sect. IIC) (402). After damage to the medial region of the amygdala, rats failed to increase their NaCl intake in response to mineralocorticoid but remained responsive to the natriorexigenic effects of captopril (see sect. IXC) and sodium depletion (424, 524, 689). Bilateral injections of mineralocorticoid receptor antisense oligonucleotides into the amygdala (500 ng each side) inhibited the increase in NaCl intake caused by subcutaneous DOCA (2.5 mg) but not the increase following adrenalectomy; glucocorticoid receptor antisense or scrambled oligonucleotides had no effect on the increased NaCl intake (512). In an abstract, it was reported that bilateral adrenal steroid implants in the amygdala caused an increase in intake of NaCl but not water (472). On the other hand, in one experiment, there was no significant c-fos expression in the amygdala after subcutaneous implantation of DOCA pellets which, however, resulted in c-fos expression in the OVLT (324). In experiments on the role of the central nervous system in mineralocorticoid hypertension, it was stated, although no results of fluid intake were given, that the "salient difference between the hypertension in the ICV aldosterone model and that of natural or induced mineralocorticoid excess is the absence of saline polydipsia and polyuria in the animals receiving aldosterone ICV" (235). Polydipsia/polyuria occurred only with systemic and not intracerebroventricular administration of aldosterone, perhaps because aldosterone was not placed directly in the amygdala, but it was given by intracerebroventricular infusion, which could mean that mineralocorticoid was not reaching the critical target tissue in sufficient amounts to arouse sodium appetite.

Angiotension II immunoreactivity is found in the amygdala and BNST (342). The central nucleus of the amygdala contains angiotensinergic endings and receptors as well as mineralocorticoid receptors and has been proposed as a possible site of interaction between ANG II and mineralocorticoids in generating an increase in sodium appetite (523). Lesions here result in impairment of the NaCl intake caused both by intracerebroventricular injection of renin and subcutaneous injection of DOCA (225). Lesions of the central nucleus of the amygdala or BNST also caused impairment of NaCl intake in response to furosemide-induced or yohimbine-induced sodium depletion, circumstances in which the renin-angiotensin system is activated (688). As we saw, the central nucleus showed c-fos expression after intracerebroventricular or intravenous injection of ANG II (although not after subcutaneous ANG II). The NK-3 selective tachykinins (see sect. VIIIE) injected into the medial amygdala inhibited intracerebroventricular renin-induced but not subcutaneous DOCA-induced sodium appetite. However, whatever the role of the amygdala in ANG II/mineralocorticoid interaction, the ventral lamina terminalis containing the MnPO nucleus and OVLT is more important than the amygdala for ANG II-induced sodium appetite. Injection of ANG II into the POA, MnPO nucleus, or OVLT, but not the SFO, caused an increase in intake of NaCl as well as water (20, 173), and lesions of the AV3V region that damage the MnPO nucleus and OVLT impaired the natriorexigenic action of intracerebroventricular renin or ANG II more than that of subcutaneous mineralocorticoid treatment (523). This region contains neurons that show increased and prolonged discharges to ANG II after DOCA pretreatment. The AV3V region is also the region that shows the greatest c-fos expression accompanying the increased sodium appetite caused by intracerebroventricular or systemic ANG II, or dietary or furosemide-induced sodium depletion.

The AP is adjacent to and projects strongly into the NTS at the transition of the fourth ventricle and central canal of the spinal cord (391). It is richly supplied with fenestrated capillaries that appear to merge with the vessels of the nearby choroid plexus. The ventricular surface is covered by differentially penetrable ependyma with tanycytes that have slender processes ending on the capillaries. The AP is implicated in the neurogenic pressor response to circulating ANG II and in fluid homeostasis including drinking behavior. The AP and caudal NTS receive vascular stretch receptor inputs from the carotid sinus, aortic arch, pulmonary vasculature, venoatrial junctions, and heart. Chemoreceptors and the receptors originating in the liver provide other sources of afferent inputs. The AP may also play a role in taste, control of feeding behavior, and vomiting, but these are not considered here. Direct application of ANG II to the AP causes a rise in blood pressure and also an increase in NaCl intake, but effects on the adjacent NTS system are more complicated and depend on dose and the structures affected. Application of low doses of ANG II to the NTS system produces a decrease in blood pressure, but higher doses are pressor (77). Angiotensin II is depressor when applied to the dorsal motor nucleus of vagus.

In rats, lesions of the AP and the caudal NTS immediately adjacent to it resulted in larger increases in water intake than in unlesioned rats in response to subcutaneous isoproterenol or hypovolemia induced by hyperoncotic dialysis with polyethylene glycol (see sect. XC) (138). However, drinking responses to systemic hypertonic NaCl or intracranial carbachol (see sect. VIIIA) were unaffected by these lesions. In some cases, excessive drinking after AP lesions may have been caused by primary renal fluid losses, since these may have been excessive (278), but others have not observed excessive losses (101, 650). Bilateral lesions in the LPBN, to which the AP and NTS make substantial secondary projections, produced by electrolytic lesions or injection of ibotenic acid which destroys neurons but spares the fibers passing through the nucleus, caused similar alterations in drinking behavior, with enhancement of the water intake caused by intracerebroventricular or systemic ANG II, or by isoproterenol, but not that caused by systemic hypertonic NaCl or intracerebroventricular carbachol (137, 426, 427). Intake of NaCl is also affected by lesions of the AP/NTS system. When water and NaCl were freely available, rats increased their spontaneous intake of NaCl after destruction of the AP and immediately adjacent NTS (101, 650). Intake was greatest when lesions were restricted to the AP with minimal damage to the NTS. However, in contrast to their effects on water intake, LPBN lesions did not affect NaCl intake. When challenged with systemic ANG II, rats drank increased amounts of NaCl after destruction of the AP, or if NaCl was not available, more water (651). The large spontaneous intake of NaCl was unaffected by saralasin injected subcutaneously or into the lateral ventricle, although these treatments diminished drinking in response to systemic ANG II. On the other hand, prolonged administration of ANG II into the intact AP also caused rats to increase their NaCl intake, presumably by stimulating neurons directly; the effect was greater when the capillary tip lay in the anterolateral zone of the AP than in the anteromedial zone, and it was less when the tip lay in the fourth ventricle.

The neuroanatomic studies by Johnson and Thunhorst (292) have clarified our understanding of the central pathways for extracellular volume control. Much of the functional and anatomic evidence suggests that the AP, immediately adjacent NTS, and LPBN are part of the system that monitors extracellular fluid volume on the basis of vascular stretch receptor information and circulating ANG II levels, sending this information via ascending noradrenergic and other pathways to the AV3V region, and in particular to the MnPO nucleus, and to many other forebrain sites including SFO, PVN, SON, and amygdala (see Fig. 11). Reduced filling in the volume receptors in the heart and large veins and the arterial baroreceptors results in a fall in inhibitory traffic from vascular stretch receptors (197, 614). This activates the ascending noradrenergic pathways and norepinephrine released in the region of the ventral lamina terminalis (see sect. VIII) facilitates ANG II effects and leads to stimulation of the circulation, increased drinking behavior, and vasopressin release. There is also evidence of ascending serotonergic pathways that are inhibitory, and these are less active in hypovolemia and hypotension. There are probably other neurotransmitter pathways involved as well. The integrated activity in these various ascending facilitatory and inhibitory pathways will presumably depend on the level of cardiovascular receptor traffic and will determine drinking behavior and circulatory homeostasis. The selective enhancement of drinking caused by ANG II and other hypovolemic stimuli after lesions of the AP/NTS/LPBN system suggests that the inhibitory pathways are dominant or at least most affected by these lesions. The enhancement could therefore be explained by the increased effectiveness of responses from angiotensin receptors elsewhere in the brain, owing to removal of an inhibitory feedback. A similar explanation could apply to the angiotensin contribution to increased drinking in hypovolemia where the inhibitory feedback is also removed, but in this case by circulatory underfilling. The mechanism for increased NaCl intake produced by direct stimulation of the AP by ANG II is clearly different but could also contribute to the overall ANG II-induced increase in drinking.

The CVOs, MnPO nucleus, and tissue surrounding the anterior third ventricle in the lamina terminalis provide the neuroanatomic focus for thirst, sodium appetite, and cardiovascular control. The SFO, OVLT, and AP contain ANG II-sensitive receptors accessible to blood-borne hormone that are mainly responsible for dipsogenic, natriorexigenic, and other effects of circulating ANG II. Other ways in which circulating ANG II might cause or modify drinking, such as by acting on vascular stretch receptors or peripheral chemoreceptors, or indirectly by changing the excretion or internal distribution of fluid and electrolytes, have not been ruled out. Some of the controversy about the natriorexigenic effect of systemic ANG II revolves around the extent to which ANG II-induced NaCl intake is the consequence of ANG II-induced natriuresis (see sect. IVA2). In hypovolemia, it is possible that increases in circulating ANG II alter the setting of vascular stretch receptors making the vessel wall in which they are located less compliant to stretch. This would result in a reduction in inhibitory sensory discharge and therefore in increased drinking. More speculative still is the possible involvement of peripheral chemoreceptors in ANG II-induced drinking. There is some evidence that chemoreceptors may be involved in altered drinking behavior, although the evidence for this is sparse (269) and difficult to interpret. It is conceivable that ANG IIinduced alterations in blood flow to peripheral chemoreceptors may be involved in angiotensin-induced drinking behavior.

The CVOs have many connections with other regions of the brain. The SFO and OVLT project directly and via the MnPO nucleus and periventricular tissue to the preoptic, anterior hypothalamic, and limbic structures to initiate drinking, and to the SON and PVN to mediate release of vasopressin and oxytocin from the neurohypophysis and from vasopressinergic and oxytocinergic nerve endings. In addition to its connections with the forebrain CVOs, the AV3V region receives inputs from the AP, immediately adjacent NTS, LPBN, and other brain stem structures that form part of the negative-feedback system for extracellular fluid volume control. This serves to monitor blood volume and provides the signal, reinforced by circulating ANG II, to cause increased intakes of water and NaCl in hypovolemia. Direct application of ANG II to the AP also arouses sodium appetite. Both the AV3V region and amygdala are crucial in the control of sodium appetite. Olfactory, gustatory, and visceral inputs project to ANG II- and aldosterone-sensitive tissue in the AV3V region, amygdala, and BNST. It has been suggested that the amygdala is important in mineralocorticoid-induced sodium appetite, and it could also be a site of interaction between ANG II and mineralocorticoids.

There are many ANG II-sensitive receptors inside the blood-brain barrier, notably in the MnPO nucleus. Some of these could have a particular role to play in sodium appetite. Angiotensin II-sensitive structures inside the blood-brain barrier may normally be stimulated by ANG II generated locally by components of a brain renin-angiotensin system. This view is compatible with the finding that intracranial injections of renin, TDP renin substrate, or ANG I in rat, dog, and pigeon cause increased drinking and that this does not occur if formation of ANG II at the injection site is prevented by prior intracranial injection of appropriate blockers in amounts insufficient to affect components of the renin-angiotensin in the bloodstream (see sect. VB). However, the relation between ANG II-sensitive structures on either side of the blood-brain barrier in drinking behavior is uncertain. The idea that circulating ANG II contributes to hypovolemic thirst and sodium appetite is straightforward and supported by the evidence, although lacking quantitative precision. Less well understood is the possible role of angiotensin peptides generated inside the blood-brain barrier in drinking behavior, which is discussed next.

	VII. ROLE OF ANGIOTENSIN PEPTIDES FORMED FROM PRECURSORS IN BRAIN IN DRINKING

Top Previous Next References

Renin and angiotensinogen genes are expressed in many organs other than kidney; these include salivary gland (see sect. IIIA4), brain, pituitary, adrenal cortex, heart, vascular smooth muscle, ovary, and testis (135, 453). The amounts of renin found in brain are unaffected by nephrectomy, which shows that nervous tissue can synthesize renin independently of the kidney (171, 229, 230). Functional and neuroanatomic evidence indicates that angiotensin receptors for increased drinking behavior exist inside the blood-brain barrier as well as outside. Because ANG II probably cannot cross the barrier, it is reasonable to suppose that ANG II-sensitive neurons inside the barrier are normally stimulated by peptide derived from angiotensin precursors made in the central nervous system. Hypovolemia causes an increase in hypothalamic ANG II, and sodium restriction causes some increase in ANG II binding, suggesting that the brain renin-angiotensin system is functionally responsive to altered physiological demands. Brain renin itself may not be responsible for local formation of angiotensin peptides because mRNAs for renin and angiotensinogen are distributed differently in nervous tissue, and the amounts of renin are very small. Other enzymes may be responsible. Much of the evidence favors involvement of brain angiotensin peptides in body fluid homeostasis, thirst, sodium appetite, and blood pressure control, but there are unanswered questions about the extent and nature of the involvement.

Various components of the renin-angiotensin system are present in different parts of the brain, but sites of production, concentrations, and location of components in relation to angiotensin receptors present some puzzles. Not all components are necessarily found in any one place, and important topological mismatches have been identified (77, 78). The distribution of renin mRNA and renin is not well characterized and has been described as "the missing link" in the anatomy of the brain renin-angiotensin system (77). In some areas, all components are present but high-affinity ANG II binding sites are absent. This might mean that other angiotensin analogs are produced and exert physiological effects by acting on their specific binding sites. The presence of angiotensinogen mRNA in astrocytes suggests that extracellular processing of angiotensinogen and diffusion of biologically active angiotensin peptides to specific receptors on neurons nearby may take place. However, topological mismatches do not occur in brain regions responsible for body fluid homeostasis and cardiovascular control and where the functional ligand is presumably ANG II. Hypothalamic and limbic structures concerned in the control of fluid balance and release of hormones, and the brain stem/NTS system for extracellular fluid volume control, have high concentrations of all renin-angiotensin components closely associated with high-affinity ANG II binding sites.

Brain renin has been localized intracellularly in neurons and neuroglia of humans, dogs, rats, mice, rabbits, pigs, sheep, and cattle. The concentrations are low, and there are marked variations in amounts in different brain regions and between species (231, 342). In the pig, brain renin is widely distributed throughout the central nervous system and resembles renal renin in its physical, enzymatic, and immunological properties; it is true renin and not cathepsin D-like acid protease (265). A specific renin mRNA has been identified in rat and mouse brains by hybridization techniques (136), suggesting that renin is synthesized in the brain, although Ganten and colleagues (231, 342) have stated that final proof of this will require cell-free translation of brain mRNA, identification of the newly synthesized protein as renin, and characterization of the mRNA/cDNA or cRNA hybrids. The low levels of mRNA isolated from brain tissue may not be derived from neural elements but from endothelial cells in the cerebral vessels or choroid plexus (162).

Angiotensinogen is present in high concentration in CSF. With the use of recombinant DNA techniques and cell-free translation of mRNA, angiotensinogen identical to that synthesized in liver has been found in rat brain (87). Large amounts of angiotensinogen mRNA are found in the forebrain and medulla of the rat, with astrocytes being the chief or even exclusive source (559). However, small amounts may be expressed in neurons (453). Although angiotensinogen mRNA may largely be confined to neuroglia, angiotensinogen itself is found both in neuroglia and neurons. The scarcity or even absence of mRNA for angiotensinogen in neurons presumably means that neuronal angiotensinogen comes mainly from the astrocytes, although the possible existence of a mechanism of neuronal uptake after formation and release of the peptide by astrocytes remains to be investigated (77, 78). Angiotensinogen mRNA is also expressed by human astrocytes, but astrocytes expressing the gene for angiotensinogen failed under the same conditions to show the presence of mRNAs encoding either renin or ACE (162). Glucocorticoid and mineralocorticoid treatment caused increases in angiotensinogen mRNA in presumed astrocytes in the anterior hypothalamus, POA, PVN, SON, suprachiasmatic nucleus, and periventricular nucleus, whereas adrenalectomy had the reverse effect (505). This could explain the synergism between ANG II and mineralocorticoids in arousing sodium appetite (see sect. II). Where angiotensinogen is processed once it is synthesized and what other peptides are formed are uncertain. Extracellular processing is certainly possible, but angiotensinogen may also be taken up for intraneuronal processing, giving rise to angiotensin peptides that do not necessarily depend on renin for their formation. Which of the angiotensin peptides formed locally will presumably be governed by which enzymes are present in the particular tissue (see Fig. 1).

Angiotensin converting enzyme is widely distributed in the brain, with high concentrations found in choroid plexus, SFO, caudate-putamen, substantia nigra, globus pallidus, and ME; it is found inside neurons as well as on blood vessels (231, 342). There may be more than one isoform of ACE (691). Angiotensin II and ACE immunoreactivities are not codistributed (67), which could mean that angiotensin peptides are formed by enzymatic pathways that are independent of ACE. For example, ANG-(1--7) is produced directly from ANG I through action of prolyl endopeptidase in brain tissue, and some of the actions of angiotensin peptides in the central nervous system may be mediated by it (see sect. VD). In some tissues, the responses to ANG I are not completely blocked by ACE inhibitors, showing that other enzymes capable of forming ANG II from ANG I are present, including the serine proteases, kallikrein, tonin, and chymase (15, 35, 276).

Angiotensin I, ANG II, and ANG III have been extracted from the brains of nephrectomized rats, rabbits, and primates in amounts that again highlight the question of whether brain renin alone is responsible for generating angiotensin peptides, or whether other proteases or even renal renin taken up by glia or tanycytes may not also play a part (162). The amino acid sequences of rat brain angiotensins are the same as those of plasma angiotensins; identical peptides were cleaved from brain and plasma angiotensinogen by renin (228). Immunohistochemical techniques have shown that ANG II is widely distributed throughout the central nervous system (Fig. 12). In contrast to angiotensinogen, ANG II seems to be located exclusively in neurons. Angiotensin II immunoreactivity in varying amounts has been detected in the cerebral cortex, hippocampal formation, basal ganglia, cerebellum, lateral geniculate nucleus, and spinal cord, as well as in the nervous structures more directly concerned in the control of body fluids such as the ventral lamina terminalis, medial and lateral POAs, anterior hypothalamus, amygdala and BNST, lateral septum, other forebrain limbic structures, and the medulla oblongata (77, 78, 342, 630).

View larger version (24K): [in this window] [in a new window]   FIG. 12.   Distribution of ANG II immunoreactivity and ANG II receptor binding in a midline sagittal section of rat brain. Shading represents receptors; solid circles and lines from them represent cells and axons. AP, area postrema; HIPP, hippocampus; LC, locus ceruleus, ME, median eminence; MnPO nucleus, median preoptic nucleus; NTS, nucleus tractus solitarius; OVLT, organum vasculosum laminae terminalis; PVN, paraventricular nucleus; SEPT, septum; SC, superior colliculus; SCN, suprachiasmatic nucleus; SON, supraoptic nucleus; STH, subthalamic nucleus. [From Phillips in Harding et al. (256).]

There are uncertainties about the functions of brain angiotensin peptides, how they act, and their relation to circulating ANG II. In addition to actions on thirst, sodium appetite, and blood pressure control, ANG II affects cell growth, membrane function, protein synthesis, PG release, pituitary hormone synthesis and release, and learning and memory (231, 406, 675). Different angiotensin receptors (see sect. VB3) may mediate opposing effects on cell proliferation and possibly on drinking behavior (631). After substantial blood loss in the rat, ANG II levels were increased in the hypothalamus independently of the increases in plasma ANG II, which could mean a role for brain ANG II in drinking and blood pressure control in hypovolemia (451). Regional increases in ANG II binding also occur in response to dehydration or sodium restriction (382), supporting the view that the brain system is responsive to changing physiological need. Dehydration leads to increased ANG II receptor binding in the SFO. In sodium restriction, the increase was small but significant and confined to the PVN, suggesting that there is a limited subset of receptors responding to sodium deprivation (684).

Whether angiotensin peptides derived from brain precursors are to be regarded simply as neurotransmitters or as also exercising a paracrine role is undecided. In view of the discrepancy in concentrations between renin and angiotensin peptides in brain tissue, renin is probably not an obligatory component of the processes that lead to the formation of angiotensin peptides. The anatomic distribution of components also may not match that of receptors. Because renin activity and angiotensin immunoreactivity and binding in the brain are so widespread and because they are present in regions as unexpected as the cerebellum, it is more than probable that some of the actions of centrally generated angiotensin peptides have little to do with body fluid homeostasis. The physiological consistency of the peripheral actions of ANG II is unlikely to be repeated for centrally generated ANG II, although most experimental evidence supports the view that brain ANG II is particularly important in the regulation of blood pressure, sodium homeostasis, and blood volume (342). These are related physiological functions that have a direct bearing on thirst and sodium appetite. Drinking behavior is affected by changes in blood pressure, decreases causing hypovolemic thirst and sodium appetite in which ANG II is implicated, and acute increases attenuating ANG II-induced thirst (see sect. III, A2 and B2) and probably sodium appetite.

The SH rat provides much of the evidence for involvement of a brain renin-angiotensin system in blood pressure control, and it has been suggested that an overactive brain renin-angiotensin system in this rat strain is responsible for the hypertension (77, 78, 674) and increased sodium appetite, although whether this means overproduction of angiotensin peptides or heightened sensitivity to their action or both of these is not always clear. Certainly, injection of ANG II into the lateral ventricle caused larger pressor responses in SH rats than in WK controls (267), and septal neurons in stroke-prone SH rats were more sensitive to iontophoretically applied ANG II, showing a lower threshold and extended afterdischarge compared with WK rats (158, 159). Furthermore, plasma renin and ANG II are not elevated in SH rats, whereas renin activity is elevated in brain catecholaminergic nuclei and ANG II levels are increased in the hypothalamus, NTS, and dorsal motor nucleus of the vagus in SH rats compared with WK rats (674). The central pressor actions of ANG II and the increased responsiveness of SH rats have been regarded as favoring a role for ANG II generated from brain precursors in the control of blood pressure, especially since some of the sensitive structures lie inside the blood-brain barrier. Pressor sites accessible to blood-borne or locally generated ANG II have been identified in various parts of the brain, including the MnPO nucleus, OVLT, SFO, central gray, superior colliculus, AP, and NTS.

Direct evidence suggesting participation of brain ANG II in the control of blood pressure is provided by experiments with antagonists of the renin-angiotensin system that necessarily can only produce their effects by preventing the endogenous formation or action of ANG II. In SH rats, intracerebroventricular administration of ANG II receptor antagonists (357, 372, 452) or angiotensin-converting enzyme inhibitors (277, 581) lowered the blood pressure independently of an effect on circulating ANG II. In SH rats, there was a dose-dependent reduction in blood pressure after acute blockade of brain angiotensin receptors by lateral ventricular saralasin, whereas intravenous infusion of saralasin caused a rise in blood pressure (357). The hypotensive effect of intracerebroventricular saralasin was still present 15-20 h after nephrectomy. These effects were not seen in WK controls. However, others found no lowering of blood pressure in SH rats after intracerebroventricular administration of saralasin or sarthran (37, 68), and injection of losartan into the lateral ventricle did not lower the blood pressure of SH rats, whereas oral administration was extremely effective (126). Intracerebroventricular losartan also prevented the pressor response to intracerebroventricular injection of ANG II but not to intravenous ANG II (see sect. VB3). According to these results, the antihypertensive effect of losartan seemed to be exerted outside the blood-brain barrier, which if confirmed would leave unresolved the question of the role of brain ANG II in the elevated blood pressure of SH rats.

Despite the conflicting reports, more recent evidence with intracerebroventricular injection of losartan or antisense oligonucleotides supports the view that ANG II formed locally in the brain is responsible for the high blood pressure in SH rats and that circulating ANG II plays little part in this type of hypertension. Single injections of losartan into the lateral ventricle caused a slowly developing reduction in blood pressure in SH but not WK rats (437). Antisense inhibition of either AT₁ receptor mRNA or angiotensinogen mRNA reduced the blood pressure of SH rats to normotensive levels, but neither of these antisense oligonucleotides affected the blood pressure of normotensive WK controls (250). These results suggest that ANG II formed locally in the brain is responsible for the elevated blood pressure of SH rats. However, they also seem to suggest that ANG II plays no part in the control of blood pressure in normotensive animals. The latter conclusion is surprising. The differences between SH and WK rats seem more likely to be quantitative than representing fundamental differences in the central mechanisms of blood pressure control.

The possible role of ANG II formed in the brain has also been examined in other models of hypertension, with findings that are not always consistent. Intravenous saralasin lowered the blood pressure in the rat with two-kidney, one-clip Goldblatt hypertension (see sect. XF), showing the importance of circulating ANG II for maintaining the elevated blood pressure (357) as it is for the accompanying increased drinking behavior. However, intracerebroventricular saralasin or captopril (357, 581) also lowered the blood pressure in doses that were ineffective when given peripherally. This does not necessarily mean that ANG II formed in the brain is responsible for the rise in blood pressure in renal hypertension because these treatments also caused a fall in plasma renin by unknown mechanisms so that the results are compatible with an exclusive role for circulating ANG II. In rats with DOCA-salt hypertension, intravenous and intracerebroventricular saralasin or capto pril were pressor. The agonism and lack of any hypotensive effect of saralasin does not suggest involvement of renal or brain renin in the increased blood pressure of this animal preparation. However, results with captopril in DOCA-salt hypertension make it unwise to eliminate altered activity of brain renin-angiotensin altogether as a factor in DOCA-salt hypertension; a single intracerebroventricular injection of captopril was pressor (581), but long-term infusion over 7 days attenuated the increase in blood pressure (283). On the other hand, there is little doubt that central mineralocorticoid effects are important in this model because intracerebroventricular infusion of RU-28318, a specific mineralocorticoid antagonist, prevented the hypertension caused by mineralocorticoid excess (235). Again, as suggested for SH and WK rats, a fall in blood pressure produced by presumed blockade of a brain renin-angiotensin system in these models of hypertension might simply be the result of eliminating a basal control mechanism. It does not mean that the increased blood pressure in renal or DOCA-salt hypertension is accounted for by overactivity of the brain system.

	VIII. NEUROTRANSMITTERS AND ANGIOTENSIN-INDUCED DRINKING

Top Previous Next References

Many substances that are thought to be released by neurons, and possibly in some cases by neuroglial cells as well, affect drinking behavior when injected into the brain and may interact with and modify ANG II-induced drinking. These substances act as neurotransmitters producing effects localized to the postsynaptic membrane or presynaptic endings close to where they are released, or they may diffuse into the extracellular fluid and exert paracrine or volume effects on nervous structures some distance from their source, or they are secreted into the bloodstream and function as hormones. The time course of action varies, and some substances could have long-term trophic actions. The nature of the interaction between angiotensin peptides derived from whatever source and these other substances is generally unknown. Where it has been established that ANG II is colocalized in particular neurons with other peptides or neurotransmitters, the substances could be coreleased in response to the same event and could act presynaptically or postsynaptically at the same synapses so that interaction takes place within an angiotensinergic system itself. There are other possibilities, and interaction may be more remote, with different neuronal pathways employing different neurotransmitters and subserving different aspects of drinking behavior, operating separately and only converging to produce an integrated response at some final common pathway.

Considered in this section are neuroactive substances (except angiotensin peptides themselves) released in the nervous system that are known to stimulate or inhibit drinking behavior when injected into parts of the brain (Table 5). Most of these have been shown to interact with ANG II-induced thirst or sodium appetite. They include classical neurotransmitters, peptides, and other substances such as PGs and nitric oxide. Thirst and sodium appetite are also affected by neuroactive substances released outside the central nervous system, and many of these also interact with ANG II-induced drinking. These are dealt with elsewhere: corticosteroids in section IIC, reproductive hormones (except oxytocin) in sections IIIA4 and IVA3, and increases in circulating ANG II caused by drugs or experimental procedures in sections IX and X. Complete lists of substances causing increased thirst or sodium appetite appear in Table 6.

Some substances stimulate drinking behavior by participating directly in central mechanisms. Others inhibit it, sometimes quite selectively. Substances may have opposite effects in different animal species (e.g., tachykinins stimulate drinking in birds but inhibit it in mammals), or even in the same species (e.g., opioids may both stimulate and inhibit drinking behavior). The actions of a substance released in the brain may complement the effects of the same substance when it is released peripherally and is acting as a circulating hormone, although in some cases the effects may be quite the opposite (e.g., serotonin acting centrally may sometimes inhibit drinking but acting peripherally it stimulates).

Acetylcholine is distributed throughout the central nervous system, occurring in all parts of the forebrain, midbrain, and brain stem. Cholinergic stimulation of limbic structures by intracranial injection of acetylcholine or carbachol causes vigorous drinking in the rat, but not in the cat, monkey, and pigeon, and very little in the dog and rabbit (244-246). Cholinergic stimulation inhibits sodium appetite in the rat (192). The rat's SFO, which is especially sensitive to the dipsogenic action of circulating ANG II, responds to carbachol (494), but ANG II-induced drinking does not seem to depend on cholinergic pathways. Intracranial or subcutaneous injection of doses of atropine that were fully effective at preventing carbachol-induced drinking did not block ANG II-induced drinking (199). Angiotensin II-induced drinking was only inhibited by doses of atropine large enough to produce signs of general atropinization. Angiotensin II-induced drinking was also unaffected by injection of atropine into the lateral ventricle of the cat which does not drink to carbachol (102). This and other evidence suggest that in the rat and cat ANG II-induced drinking does not depend on cholinergic mechanisms that seem to be more involved in thirst of cellular origin than in hypovolemic thirst (184). However, there are species differences because intracerebroventricular atropine blocked ANG II-induced drinking in the dog which shows a small and variable increase in drinking in response to carbachol (463).

On the other hand, it has been suggested that ANG II may be involved in cholinergic thirst because intracerebroventricular injection of large doses of the AT₂ antagonist PD-123319 inhibited carbachol-induced drinking in the rat (219, 501). Drinking in response to other thirst challenges was also inhibited by PD-123319, but eating behavior was not, giving rise to the suggestion that there is a final common AT₂ pathway for all types of thirst including cholinergic as discussed in section VB. However, intracerebroventricular carbachol-induced drinking was not inhibited and may even have been slightly increased by pretreatment with the nonselective angiotensin antagonist [Sar¹,Gly⁸]ANG II (189), and it was unaffected by the AT₂ antagonists CGP-42112 (38) and PD-123177 (103), the AT₁ antagonist losartan (38, 306), and AT₁ antisense oligonucleotide (see sect. VB4) (510). The significance of the inhibitory effect of PD-123319 on carbachol-induced drinking is not obvious.

The peripheral action of -adrenergics on drinking behavior is dealt with in section IXA. There are many monoaminergic pathways that project widely from the brain stem to the forebrain structures that are responsible for the control of drinking behavior and body fluid homeostasis. Dopamine is the major mammalian catecholamine in the central nervous system. In addition to the nigrostriatal system, dopaminergic neurons form a mesolimbic system with fibers arising in the ventral tegmentum and projecting to the hypothalamus and various parts of the limbic system. Noradrenergic neurons from locus ceruleus, NTS, dorsal motor nucleus of the vagus, and other nuclei in the pons and medulla send ascending fibers running in dorsal and ventral noradrenergic bundles and then the medial forebrain bundle to many parts of the forebrain, including the hypothalamus, amygdala, thalamus, cerebral cortex, and hippocampus. Other noradrenergic fibers supply the cerebellum and spinal cord. Adrenergic fibers are much more restricted in their distribution, projecting from brain stem to hypothalamus and spinal cord. The dipsogenic action of intracranial ANG II may involve interaction with the catecholamines that are released from neurons by ANG II and that are known to elicit ingestive behavior when injected into the brain of the rat (244). The neurogenic pressor response to close arterial infusion of ANG II is an example of central nervous angiotensin/adrenergic interaction. In the rat, ANG II-induced drinking depends on the integrity of catecholaminergic pathways. There is a close parallelism between the amounts of brain angiotensin (see sect. VII) and norepinephrine in various parts of the brain of dog and rats (171, 342), and intracerebroventricular injection of ANG II has been shown by microdialysis to release norepinephrine from the anterior hypothalamus of the conscious rat (460).

Angiotension II could interact with both afferent and efferent central autonomic pathways. A matrix of ANG II-immunoreactive fibers and varicosities enclose catecholamine-containing cells in the ventrolateral medulla. Angiotensin II-immunoreactive fibers also innervate dopamine-containing neurons throughout the hypothalamic periventricular zone. Ascending noradrenergic pathways from cell groups in the brain stem reach the MnPO nucleus, a well-defined focus of angiotensinergic and adrenergic innervation. Angiotensin II and catecholamines are therefore present together in structures associated with the control of blood pressure and body fluid homeostasis so that in hypovolemia, ANG II could influence the responses to the sensory information reaching the ventral lamina terminalis from volume receptors and baroreceptors via ascending noradrenergic and other pathways. Electrophysiological evidence shows that there is increased activity in the ascending noradrenergic pathways projecting to the MnPO nucleus in hypovolemia or hypotension, which could provide the mechanism for integration of neural inputs from volume receptors and baroreceptors and ANG II effects in the ventral lamina terminalis (289).

Injection of 6-hydroxydopamine into the lateral ventricle, which selectively destroys catecholaminergic neurons, attenuated drinking in response to intracerebroventricular ANG II but not carbachol (199). When in the intact rat injection of ANG II or carbachol was preceded by injection of catecholamine antagonists through the same intracranial cannula, the dopamine antagonists haloperidol and spiroperidol inhibited ANG II-induced drinking but not carbachol-induced drinking. High intracranial doses of - and -adrenergic receptor antagonists also inhibited drinking, but the -antagonist phentolamine was no more effective against ANG II than it was against carbachol, and the -antagonists propranolol and sotalol were also without differential inhibitory effects on the two dipsogens. High doses of D- and L-propranolol were equally inhibitory, although only the D-isomer is a receptor blocker. In the cat, - and -adrenergic neurons seem to be involved, since both intracerebroventricular phentolamine and D-propranolol inhibited ANG II-induced drinking (102). Other results in rat using various blockers implicate adrenergic and dopaminergic pathways in ANG II-induced drinking. Some evidence is contradictory, but this may depend on which brain structures are affected and whether intake of water or NaCl is in question. For example, intracranial injection of norepinephrine has been found to inhibit thirst (246) but stimulate sodium appetite (97). However, in another study, injection of norepinephrine, clonidine, the presynaptic ₂-adrenergic agonist, or phenylephrine into the third ventricle inhibited intakes of 3% NaCl and water caused by intracerebroventricular renin or subcutaneous DOCA (127). The results in this experiment suggested that ₂-adrenergic receptors were responsible for the inhibition of hormone (ANG II, aldosterone)-induced NaCl intake, but the presence of norepinephrine in the ventral lamina terminalis seems critical for ANG II-induced drinking. Impairment after 6-hydroxydopamine (34) can be reversed by central norepinephrine infusions or transplants of fetal noradrenergic cells in this region (110, 289). Norepinephrine is a widespread neurotransmitter, and it is not unreasonable to assume that it may have both excitatory and inhibitory effects on fluid intake.

Serotonergic pathways from midline raphé neurons of the brain stem project to parts of the limbic system, hypothalamus, hippocampus, and cortex. There are similarities in distribution between serotonergic and noradrenergic pathways, and the close parallelism between brain angiotensin and norepinephrine in various parts of the brain has already been mentioned. Angiotensin II-immunoreactive nerve terminals are found in serotonin-positive regions in the hypothalamus and brain stem. Angiotensin II stimulates serotonin synthesis and release from nervous tissue (78), introducing the possibility that central serotonergic mechanisms play a role in the ANG II drinking response. Systemically injected serotonin causes increased drinking in rat, an effect that is probably brought about by activation of the renal renin-angiotensin system (see sect. IXE), but evidence of a direct excitatory effect of centrally injected serotonin on drinking behavior is much weaker than it is for systemic injection. In rhesus monkey, serotonin (6-25 µg) caused eating and drinking on 27 of 55 occasions when injected in the fiber systems associated with the fields of Forel (530). The response latency was long (15-30 min), but some animals drank without eating. The central dipsogenic effect is clearly very weak, and there is little evidence of it in the rat; depleting central serotonin stores pharmacologically had no effect on the pressor or dipsogenic effects of intracerebroventricular ANG II (84). Serotonin effects on brain are widespread and complex, affecting wakefulness, mood, aggression, feeding, sexual behavior, motor neuron excitability, body temperature, anterior pituitary secretion, and autonomic functions in the gut, heart, and circulation. Some of these other actions of serotonin could have secondary effects on drinking behavior.

Even though there may be a weak drinking response to intracranial serotonin, activating central serotonergic pathways on drinking can also be inhibitory. Water and 0.3 M NaCl intakes induced by injection of ANG II (50 ng) into the lateral ventricle of the rat or by subcutaneous injections of furosemide and captopril were inhibited by serotonergic agonists and augmented by the serotonergic receptor antagonist methysergide (4 µg/200 nl), injected bilaterally 10 min beforehand into the LPBN (380, 381). Bilateral injections of methysergide into the LPBN also enhanced NaCl intake in response to injections of ANG II (20 ng/200 nl) into the SFO, an effect that was blocked by preinjection of losartan (1 µg/200 nl) into the SFO (100). Systemic dexfenfluramine, a drug enhancing serotonergic transmission, inhibited spontaneous and depletion-induced NaCl intake without affecting water intake, whereas the serotonin antagonist metergoline caused significant increases, suggesting that serotonin may have a tonic inhibitory role on NaCl intake (493). These results indicate that there is a serotonergic inhibitory pathway associated with the LPBN. It is reasonable to suppose that this forms part of the ascending brain stem pathway for volume control referred to in section VIG, but such a mechanism is not incompatible with the presence in the periphery and possibly in other parts of the brain of serotonergic mechanisms that stimulate drinking.

Glutamate, the major excitatory neurotransmitter, and GABA, the inhibitory transmitter synthesized from glutamate, are present in high concentrations in neurons in many brain areas where they function in projections between areas, raising or lowering the level of neural activity. Among excitatory amino acid receptors, the N-methyl-D-aspartate (NMDA) subtype predominates in the hippocampus and hypothalamus. Two excitatory amino acids, NMDA and kainate (a stable analog of L-glutamate), have been found to be dipsogenic in pigeons (31), but glutamate itself does not appear to be (152). The relation of excitatory amino acids to ANG II-induced drinking in pigeon is unknown. The possible role of excitatory amino acids in drinking behavior in the mammal is also not known, but NMDA receptors may mediate drinking and c-fos expression in response to intracerebroventricular ANG II. Intracerebroventricular infusion of dizocilpine, which prevents calcium entry through open NMDA channels, inhibits ANG II-induced drinking and c-fos expression in the MnPO nucleus, SON, PVN, and NTS (683). This suggests that an NMDA receptor-dependent mechanism is involved in both the behavioral and cellular responses to ANG II. -Aminobutyric acid attenuates intracerebroventricular ANG II-induced intakes of water and NaCl (3, 294) and ANG II-induced pressor responses and vasopressin release (58, 59).

Tachykinins and bombesin-like peptides are unrelated biologically active peptides with similar effects on drinking behavior so that it is convenient to consider them together. Both affect ANG II-induced drinking behavior in rats. Tachykinins are involved in body fluid homeostasis, in a variety of motor, sensory, and autonomic functions, and they have potent and specific effects on drinking behavior in mammals and birds (113, 118). Tachykinins share a common COOH-terminal sequence, and they include substance P found in mammalian brain, gut, and skin, physalaemin from the skin of the South American frog, kassinin from the skin of the African frog, and eledoisin from the Mediterranean octopus. They and their receptors are widely distributed in the central nervous system, but there is not always good correspondence between tachykinin nerve fibers and tachykinin receptor density, possibly because of the presence of low-affinity binding sites or undiscovered receptor subtypes (433). Among the tachykinins isolated from mammalian brain and the preferred receptors on which they act are the undecapeptide substance P (NK-1 receptor), the decapeptides neurokinin A (NK-2 receptor) and neurokinin B (NK-3 receptor), and the NH₂-terminal extended peptides neuropeptide K and neuropeptide  (NK-2 receptor) (259). The NK-1 receptor is found in brain and peripheral tissue, NK-2 in highest concentration in the periphery, and NK-3 mainly in brain. Receptors are the seven membrane spanning type coupled to G protein. Mammalian tachykinins are rapidly hydrolyzed by ACE and other peptidases, but the nonmammalian tachykinins are more resistant to degradation, and because they bind to mammalian cerebral receptors, they are more potent and have longer lasting effects when injected into the mammalian brain. Members of a second group of neuroactive peptides from amphibian skin, the bombesin-like peptides, also affect drinking behavior in mammals and birds. They include bombesin, ranatensin, and litorin. The mammalian analog of bombesin is gastrin-releasing peptide. Bombesin receptors have been identified in the pancreas and central nervous system.

Tachykinins and bombesin-like peptides injected into sensitive structures in the brain are generally dipsogenic in pigeons but antidipsogenic in rats (113, 154, 191). In thirsty sodium-depleted wild rabbits and sheep, eledoisin infused into the lateral ventricle caused increased water intake in addition to inhibiting sodium intake (596) so that the difference in the effects of tachykinins on drinking in pigeon and rat is not a distinction between birds and mammals but probably represents differences in the distribution of receptors and receptor types. In the pigeon, injection of eledoisin, physalaemin, or bombesin into the third cerebral ventricle caused vigorous short-latency dose-dependent drinking, similar to the ANG II response, but requiring ~10 times the molar dose for the same water intake, and generally unaffected by prior treatment with competitive angiotensin antagonists (113). Intracerebroventricular injections of eledoisin, physalaemin, and kassinin were of equal potency, neurokinin A and substance P were less potent, and neurokinin B was the least potent (364). In ducks, third ventricular injections of bombesin caused dose-dependent drinking, but ANG II was ~10 times more potent; eledoisin was less effective than bombesin, and substance P was almost completely inactive (115). Tachykinins and bombesin were also effective dipsogens injected intraperitoneally, but doses had to be increased at least 1,000 times. Destruction of the SFO in pigeons did not affect drinking in response to intracerebroventricular tachykinins or bombesins, and although the response to intraperitoneal but not intracerebroventricular injection of ANG II was reduced, the responses to intraperitoneal injection of bombesins and the more modest responses to intramuscular injection of tachykinins were unaffected (359, 360).

In the rat, intracranial injection of eledoisin, physalaemin, substance P, or bombesin caused dose-dependent inhibition of drinking to various dipsogenic stimuli including central ANG II, an effect opposite to that produced by these substances in the pigeon (116, 117, 191). Tachykinins also caused vasopressin release and water conservation (90). Eledoisin-induced vasopressin release may have been mediated by central ANG II, because it was blocked by saralasin (445). Inhibition of drinking by tachykinins was centrally mediated and behaviorally selective in that feeding was unaffected, or less affected in the case of bombesin, than drinking. Angiotensin II-induced drinking tended to be inhibited by lower intracranial doses of eledoisin, physalaemin, and substance P than drinking induced by other thirst stimuli such as intracranial carbachol, water deprivation, or systemic hypertonic NaCl. Because ANG II caused increased substance P immunoreactivity release from superfused preparation of hypothalamic and medullary tissue (132), this could be one of the ways in which ANG II-induced drinking is terminated. Angiotensin II-induced drinking was more powerfully inhibited by the nonmammalian tachykinins, eledoisin and physalaemin, than by substance P (118), presumably because of the greater resistance of nonmammalian tachykinins to degradation in rats, but mammalian neuropeptide  was equipotent with eledoisin and physalaemin and has a similar spectrum of antidipsogenic action. Inhibition of ANG II-induced drinking by neuropeptide  can be reversed by the nonpeptide NK-1 receptor antagonist WIN-62577 (458). On the basis of the rank order of inhibition of drinking by naturally occurring tachykinins and tachykinin receptor-selective synthetic peptides, it has been suggested that the tachykinin receptor subtypes involved in the central inhibitory effects of the tachykinins in rats are NK-1 for ANG II-induced water intake and NK-2 for cell dehydration-induced drinking (366). Some tachykinins have a much narrower spectrum of antidipsogenic activity that seems to exclude their involvement in angiotensinergic thirst.

Tachykinins also inhibit sodium appetite in rats (118) and other mammals. Tachykinin inhibition of NaCl intake may be mediated by NK-3 receptors (366, 456). Intracranial eledoisin, followed by kassinin and neurokinin A, exerted potent and long-lasting inhibition of NaCl intake, whereas physalaemin and neurokinin B were much less effective (118). Intracerebroventricular injection of kassinin attenuated sodium depletion-induced NaCl intake in the rat and the increased NaCl intake caused by intraventricular renin or subcutaneous DOCA (365). Kassinin also inhibited need-free or palatability-induced NaCl intake, but it did not interfere with food intake, indicating that it has a general suppressive effect on salt intake, irrespective of the natriorexigenic treatment or whether sodium was needed or not. The sodium intake of sodium-depleted sheep, rabbits, and cows was inhibited by intracerebroventricular eledoisin (596). Continuous intracerebroventricular infusion of eledoisin, NH₂-senktide and, to a lesser extent, substance P attenuated the enhanced need-free NaCl intake in female rats that had been subjected to repeated sodium depletion (456). When the NK-3-selective tachykinins, eledoisin and NH₂-senktide, were injected directly into the medial amygdala of the rat, the increased sodium appetite after sodium depletion or injection of renin into the lateral ventricle was reduced but not the appetite aroused by DOCA treatment (363). Because there is evidence that the amygdala is the target for mineralocorticoid natriorexigenic action (see sect. VIF), this might mean that inhibition of NaCl intake by the tachykinins is selective for angiotensinergic mechanisms. However, this selectivity may only apply to the amygdala because after intracerebroventricular injection of tachykinins there is also inhibition of DOCA-induced sodium appetite.

Opioid peptide-producing neurons and opiate receptors are present in the hypothalamus, SFO, ventral lamina terminalis, and other brain regions associated with ANG II-induced drinking behavior and the control of body fluid and electrolyte homeostasis. Opioids and their antagonists have marked effects on drinking in response to ANG II and other dipsogenic and natriorexigenic challenges. In the rat, intracerebroventricular injection of the naturally occurring opioids, -endorphin, Leu-enkephalin, or Met-enkephalin, inhibited intracerebroventricular ANG II-induced water intake, vasopressin release, and increases in blood pressure (113, 577-579). These effects were prevented by the opioid antagonist naloxone. After the initial inhibition of drinking and antidiuresis, there was a delayed polydipsia apparently mediated by an increase in plasma renin caused by -adrenergic stimulation of central origin affecting the kidney. The amphibian opioid heptapeptide dermorphin and especially the synthetic derivative [D-Ala²,D-Leu⁵]enkephalin were even more effective against ANG II-induced drinking (113). Dermorphin-like immunoreactivity is found in the SFO, OVLT, and arcuate nucleus of the rat, and it therefore has a more restricted distribution in the central nervous system than the enkephalins and endorphins. When injected into the SFO, dermorphin inhibited intracerebroventricular ANG II-induced drinking more effectively than when it was injected into the lateral ventricle (444). After destruction of the SFO, higher doses of dermorphin were needed to attenuate drinking. Inhibition seemed to be selective; drinking induced by subcutaneous injection of hypertonic NaCl or after a period of water deprivation was unaffected by doses of dermorphin that were effective against ANG II-induced drinking.

Although these results suggest that opioid pathways are mainly inhibitory to drinking behavior, other results show that opioids may cause increased drinking. The opioid antagonists naloxone and naltrexone inhibit spontaneous and water deprivation-induced drinking, acting on sites within the central nervous system (64). Low doses of morphine and other opioids cause dose-dependent increases in water intake after an initial depression in the first 2 h after injection (16, 446), and naloxone inhibits ANG II-induced drinking (63, 577) and may inhibit sodium appetite (108), results which suggest that endogenous opioids could also stimulate, or at least facilitate, thirst and sodium appetite. It is worth noting that opioid peptides also have mixed stimulatory or inhibitory effects on the release of many of the hormones of the hypothalamo-pituitary axis (358). Morphine-induced drinking was suppressed by bilateral nephrectomy (446), which together with the delay in onset of drinking could mean that renal renin was responsible for any stimulatory effects on drinking, although this is unlikely to be the entire explanation. The pattern of drinking of water-deprived animals after naloxone or naltrexone suggests that satiety may be enhanced by the drug treatment (108), but naloxone in low doses may also stimulate drinking (430), giving a different perspective on the possible role of endogenous opioids in drinking behavior. There are two opposing interpretations of the way in which they could act: opioid receptor antagonists inhibit ANG II-induced drinking by blocking the action of the endogenous opioid ligands that normally facilitate drinking, or, alternatively, endogenous opioid ligands that are displaced by naloxone or naltrexone inhibit drinking (577).

It is likely that some of the complex effects of opioid peptides and their antagonists on drinking behavior, whether stimulatory or inhibitory, depend on anatomic location, dose, and which type of receptor is affected (222, 577). The effects of receptor-selective antagonists on drinking behavior are complex and not easy to interpret. Three serpentine membrane-spanning opiate receptors coupled to G protein have been characterized, µ, , and , with different pharmacological properties and functions. For example, µ- and -opiate receptor agonists increase striatal release of dopamine, whereas -receptor agonists inhibit dopamine, and µ-opiate receptor agonists inhibit norepinephrine release from the locus ceruleus (358). Systemic naltrexone inhibited drinking in response to subcutaneous ANG II (500 µg/kg) or hypertonic saline, but naltrexone (1-50 µg) injected into the lateral ventricle only inhibited drinking in response to intracerebroventricular ANG II (20 ng) (504). Intracerebroventricular ANG II- and systemic hypertonic saline-induced drinking were both reduced by intracerebroventricular injection of the selective -receptor antagonist nor-binaltorphamine and by the µ-receptor antagonist -funaltrexamine, whereas the µ₁-antagonist naloxonazine stimulated drinking after hypertonic saline. The ₁-antagonist [D-Ala²,D-Leu⁵,D-Cys⁶]enkephalin caused a significant reduction in ANG II- but not hypertonic saline-induced drinking. The conclusion was that µ₂-, ₁-, and -opioid binding sites modulate ANG II-induced drinking. Although the effects of administered opioids and their antagonists on drinking are powerful, the physiological role of endogenously released opioids in ANG II-induced and other drinking behavior is unresolved. It is likely that they are involved in a number of different mechanisms.

Endothelins (ETs) are antidipsogenic. Endothelin-1, ET-2, and ET-3 are 21-amino acid peptides, originally isolated from endothelial cell cultures. They are powerful vasoconstrictors and have widespread cardiovascular and renal effects. Two serpentine transmembrane receptors have been cloned, ET_A and ET_B , and these are coupled to G proteins. Endothelins are produced in the brain as well as in the periphery. The distribution of receptors for ETs in mouse brain (348) is remarkably similar to that for ANG II in rat and includes the AV3V region and SFO. Drinking in rat after an 18-h period of water deprivation, systemic hypertonic NaCl, or intracerebroventricular ANG II was inhibited by prior injection of ET-3 (the ET synthesized in the hypothalamus) into the third ventricle (515). In other experiments, inhibition of ANG II-induced drinking by ETs was confirmed, and the receptor subtype was identified (514). Injection of ET-2 into the third ventricle just before ANG II caused significant dose-related inhibition of drinking. Endothelin-2 does not discriminate between ET_A and ET_B receptors, but because ANG II-induced drinking was unaffected by the ET_B-selective agonist but was significantly accentuated by the ET_A antagonist BQ-123, the antidipsogenic effects of ETs may be mediated by the ET_A receptor subtype. Angiotensin II is known to release ET from cultured endothelial cells. If it does this also in brain, a possible physiological role for ETs is to terminate a bout of ANG II-induced drinking, adding yet another neuroactive substance to the list of potential inhibitors.

Atrial natriuretic peptides inhibit drinking behavior. Three endogenous ligands, atrial natriuretic peptide (ANP), brain natriuretic peptide (BNP), and C-type natriuretic peptide (CNP), and two bioactive receptor subtypes, ANP-A and ANP-B, have been characterized (279); a third receptor, ANP-C, may be a clearance receptor. Originally isolated from porcine brain, BNP is now known to be produced mainly in the heart. C-type natriuretic peptide is a neuropeptide. The peptides and their receptors are widely distributed in brain tissue, especially in the AV3V region, basal medial hypothalamus, NTS, brain stem, and CVOs, the brain regions that are particularly involved in drinking behavior, body fluid homeostasis, and blood pressure regulation (248). In rat, the major brain receptor is ANP-B for which the ligand is CNP. Central nervous and peripheral effects of ANP are mainly if not exclusively mediated through cGMP. Natriuretic peptides may contribute to the tonic control of blood pressure, sympathetic outflow, and baroreceptor sensitivity especially in SH rats and to a lesser extent in WK rats (432). In addition to their inhibitory effects on thirst and sodium appetite, circulating natriuretic peptides inhibit sympathetic nervous activity, vasopressin release, renal renin release and aldosterone secretion, leading to a fall in blood pressure and increased urinary loss of water and sodium. The actions of natriuretic peptides are generally the opposite of those of ANG II and are seen best in the setting of ANG II stimulation of thirst and sodium appetite.

In the rat, ANP given by intracerebroventricular injection or intravenously caused dose-dependent inhibition of water intake in rats made thirsty by mild dehydration or intracerebroventricular ANG II but not by intracerebroventricular carbachol (13, 415). Intracerebroventricular BNP was also effective against intracerebroventricular ANG II-induced drinking (285). One site of the inhibitory effect is the SFO, since ANP injected here attenuated drinking induced by ANG II or water deprivation (139). Atrial natriuretic peptide injected into the third ventricle also reduced NaCl preference in sodium-depleted rats (14). Atrial natriuretic peptide infused into the lateral cerebral ventricle attenuated the exaggerated sodium appetite shown by SH rats but did not affect NaCl intake in WK controls (284). This result is of interest in view of evidence that ANG II may be important in the increased sodium appetite of SH rats as discussed in section VII. Intracerebroventricular ANP also inhibited water and NaCl intake in wild rabbits (600).

The inhibitory effects of natriuretic peptides on drinking behavior complement their effects on the renin-angiotensin-aldosterone system, kidney, and vasculature which led to increased excretion and a fall in blood pressure. These effects are physiologically consistent with the possible role of natriuretic peptides in normalizing an expanded extracellular fluid volume. It is not known whether ANP of cardiac or brain (including CNP) origin is the more important, but because the inhibitory effects could be mediated by intravenous ANP and one of the sites of inhibition was the SFO, ANP of cardiac origin could certainly play a part. Also consistent with the inhibitory effect of ANP on drinking is the action of exogenous ANP on hindbrain sites, especially the NTS, to lower blood pressure, although it is pressor when injected into the anterior hypothalamus. In humans, ANP has also been reported to depress osmotic thirst but not osmotically stimulated vasopressin release; it is not known whether it also affects angiotensin responsiveness (81). In dogs, no evidence was found to support the view that central ANP is a physiological regulator of the renin-angiotensin or vasopressin systems during osmotic and hypovolemic challenges; third ventricular infusions of ANP did not affect the changes in plasma ANG II and vasopressin induced by dehydration and bleeding (588). However, drinking responses were not examined in this experiment, and the possible involvement of other natriuretic peptides in these functions has yet to be ruled out.

Magnocellular and parvocellular vasopressin-containing neurons are present in the SON, PVN, and suprachiasmatic nucleus. The magnocellular neurons terminate in the posterior pituitary, where vasopressin is stored and released into the bloodstream in increased amounts in response to increases in effective osmolality or hypovolemia, but many of the central nervous structures involved in the control of water intake have vasopressin receptors that are innervated by vasopressinergic fibers originating from parvocellular neurons. For both vasopressin and oxytocin, the magnocellular systems are representative of neurosecretory endocrine neurons, whereas it seems that parvocellular neurons constitute peptidergic neurotransmitter pathways. Peripheral administration of ANG II can stimulate vasopressin release by acting on CVOs or directly on the pituitary; intracranial ANG II can also cause vasopressin release (506). Vasopressin is released by cellular dehydration and hypovolemia, and it is the major hormone of water conservation through its antidiuretic effect (54, 55). It may also play a role in control of water intake by increasing the sensitivity of neural thirst systems (186, 586). These peripheral and central actions of vasopressin complement each other and highlight the importance of the hormone in the water economy of the body.

In dogs, intravenous vasopressin lowered the threshold of drinking in response to intravenous hyperosmotic NaCl (587), and injection of vasopressin into the third ventricle stimulated water intake (590). There was a small and inconsistent stimulating effect of intracranial vasopressin in the rat (418). Increased thirst has been observed in the human clinical syndrome of inappropriate secretion of vasopressin or SIADH (586). Reduction in vasopressin secretion in hypothalamic diabetes insipidus may affect the setting of the osmotic thirst mechanism so that thirst may not keep pace with fluid loss, leading to significant plasma hyperosmolality. Infusion of V₁ or V₂ vasopressin antagonists into the third cerebral ventricle suppressed osmotic thirst in the dog (589). In view of the stimulating effect of vasopressin on water intake, and because cellular dehydration releases vasopressin, stimulates water intake, and inhibits NaCl intake (205), an attractive speculation is that vasopressin released by vasopressinergic neurons determines this pattern by inhibiting NaCl intake and favoring water intake. It has to be said that in the rat present evidence is that oxytocin may do this as discussed in the next section.

Vasopressin is colocalized with ANG II in the magnocellular neurosecretory cells of the SON and PVN (342). Angiotensin II and vasopressin are also colocalized with CRH in the parvocellular system of the tuberoinfundibular component of the PVN, where the three peptides may be responsible for ACTH secretion. Whether there is corelease and interaction of peptides on a single target tissue in these examples of colocalization is unknown. In the rat, iontophoretically applied vasopressin excited neurons in the SFO, medial POA, and diagonal band, and these neurons also responded to ANG II (421). When vasopressin and ANG II were applied simultaneously, there was usually additivity or even potentiation of the effects of each peptide applied individually. The common responsiveness of neurons to the two hormones that play homeostatic roles in cellular dehydration and hypovolemia could partly account for the additivity of the drinking responses to cellular dehydration and hypovolemia that is observed when these fluid deficits are simultaneously present (184). However, there is no evidence that lack of vasopressin affects ANG II-induced drinking; the Brattleboro rat responds normally to intracranial ANG II (see sect. XI).

Oxytocin is synthesized in the magnocellular neurons of the PVN and SON and is released from the posterior pituitary into the bloodstream. In addition to its well-known effects in parturition and milk ejection, systemic oxytocin is natriorexigenic in wild rabbits, and it and other hormones associated with the various stages of pregnancy and lactation are thought to be responsible for the spectacular rises in sodium appetite that occur in some species at these times (122). However, evidence in rats suggests that oxytocin from parvocellular neurons acting as a neurotransmitter or paracrine agent may also inhibit sodium appetite, including that induced by ANG II, in circumstances where water is the more urgent and immediate need as in cellular dehydration or severe hypovolemia. There are parvocellular oxytocin-containing neurons in the PVN and other scattered clusters of neurons that project widely in the central nervous system to receptors in limbic sites and as far as the dorsal motor nucleus of the vagus in the brain stem. The function of these neurons is not well understood, but because oxytocin is released by increases in effective osmolality and by hypovolemia (568), it has been suggested that a subset of oxytocinergic neurons in the PVN may operate as an inhibitory system to sodium appetite, switching drinking toward water away from NaCl (565, 571, 572). A similar role could be envisaged for vasopressin, the hormone released by these stimuli, although evidence is lacking.

Water is always the immediate requirement whether the stimulus to increased drinking is cellular dehydration or hypovolemia. The oxytocinergic system could operate to ensure that thirst takes precedence over sodium appetite in either of these states. It is well-established that hyperosmolality resulting from administration of osmotically effective solutes such as hyperosmotic NaCl or sucrose causes increased intake of water and that an existing sodium appetite is inhibited (205). Sodium appetite is absent in hyperosmolality because, according to the oxytocin hypothesis, it is inhibited by increased activity of the oxytocinergic neurons in the PVN. The fact that in rat systemic oxytocin is also natriuretic (636) complements the proposed central action. In severe hypovolemia with high plasma ANG II, drinking behavior is initially similar to that caused by cellular dehydration with increased water intake only, but it changes later to increased NaCl intake, again an appropriate response because extracellular solute is also required. Water needed to repair the plasma volume deficit is drunk immediately. As water is drunk, the body fluids become hypotonic and oxytocin levels that were high before rehydration start to fall. The excitatory effects of hypovolemia on sodium appetite now become manifest, and NaCl intake starts to increase. Such a mechanism could explain the delay in the onset of NaCl intake in hypovolemia. The evidence for this hypothesis is mainly indirect and may apply only to the rat (635). In other experimental conditions in which there is increased sodium appetite, there is an inverse relation between plasma oxytocin and NaCl intake. However, raising circulating oxytocin levels did not inhibit NaCl intake, nor did systemic administration of an oxytocin antagonist enhance intake in polyethylene glycol-treated rats. This may be because plasma oxytocin is not itself the inhibitory signal, but simply a peripheral marker of parallel activity in the central oxytocinergic pathways that inhibit sodium appetite. Direct support for the hypothesis is provided by the finding that intracerebroventricular ANG II caused a larger and earlier intake of NaCl without any change in water intake after injection of the oxytocin antagonist, ornithine vasotocin, or inactivation of oxytocin receptors by lateral cerebroventricular injections of the cytotoxin, ricin-A conjugated to oxytocin, than after ANG II alone (42, 572). Similarly, after oxytocin antagonists or inactivation of oxytocin receptors, rats increased their intakes of NaCl as well as water in response to increased plasma osmolality induced by intraperitoneal hyperosmotic mannitol instead of just taking water (43, 572). Hypertonic mannitol was found to inhibit NaCl intake induced by hypovolemia but not after inactivation of central oxytocin receptors. These results are compatible with the view that the delay in the onset of ANG II-induced NaCl intake is partly because of inhibition by central oxytocinergic neurons. How the hypothesis is to be reconciled with increased sodium appetite during lactation described long ago by Richter (479) and abundantly confirmed since, and the natriorexigenic effect of systemic oxytocin in wild rabbits (122), has yet to be addressed. The seeming contradiction could be accounted for by peripheral and central oxytocin having diametrically opposite effects on NaCl intake. The arguments applied to oxytocin may also apply to peripheral and central release of vasopressin, although this has not yet been examined in species other than the rat. In dogs, intracerebroventricular vasopressin stimulated water intake and intracerebroventricular infusion of V₁ or V₂ vasopressin antagonists suppressed osmotic thirst (see sect. VIIII); it is easy to imagine opposing effects on NaCl intake, although this remains to be tested.

Prostaglandins, prostacyclin, and other eicosanoids act as local hormones or paracrine agents in many physiological processes, among which may be control of renal renin release and inhibition of drinking. In rat, PGE₁ or PGE₂ injected into the brain or the periphery inhibited drinking in response to water deprivation, hypovolemia induced by hyperoncotic dialysis (see sect. XC), and especially drinking to intracerebroventricular or systemic ANG II (304, 305). Drinking in response to pure cellular dehydration was not affected by intracerebroventricular PGE, but it was inhibited by intraperitoneal injection. Food intake was not affected by PGEs. Opposing the inhibitory effect of PGEs by pretreatment with indomethacin enhanced the dipsogenic effect of ANG II (443). Because of these results, it has been suggested that endogenous PG production may act as a natural satiety signal, bringing drinking to an end and in this way preventing the animal from overhydrating itself when it responds to various thirst stimuli but especially to ANG II. In view of the stimulating effect of ANG-(1--7) on PG synthesis (see sect. VD) and the possibility that PGs act as natural satiety factors, the intensity of thirst (or sodium appetite) may depend on a balance between production of ANG II (or III) and ANG-(1--7). However, there are difficulties with this speculation. Some PGs are implicated in fever produced by various pyrogens, linking receptor activation in the hypothalamus to an increase in the temperature set point. This means that the effect of PGs on drinking may be secondary to alterations in core temperature so that reduced water intake could simply be the result of general malaise. There are also apparently marked species differences; for example, intracerebroventricular PGE augmented drinking in the goat in response to intracerebroventricular ANG II, eliciting the comment that the central actions of PGE₁ and ANG II were strikingly similar, although PGE was not dipsogenic on its own (335). There are also many other substances that inhibit ANG II-induced drinking so that whatever the inhibitory role of PGs in drinking behavior, it is unlikely to be unique.

Nitric oxide is considered to exert a neurotransmitter or paracrine role in the central nervous system, and it is therefore potentially important in the regulation of ANG II-induced and other types of drinking behavior. However, there have been contradictory reports on the effects of central nitric oxide on drinking. Constitutive nitric oxide synthase has been identified in varying amounts in all areas of animal and human brain tissue (399). Nitric oxide synthases are expressed in various tissues including neurons and astrocytes in the central nervous system as well as endothelial cells (643). Inducible enzyme activity occurs in astrocyte cultures and may be associated with neurotoxicity. Nitric oxide is an unstable and readily diffusible nonpolar gas that cannot be stored in vesicles in nerve terminals but is synthesized by the constitutive enzyme by a calcium-dependent process when excitation reaches the nerve terminals. It stimulates soluble guanylate cyclase in brain homogenates, acting directly on the enzyme and not via a cell surface receptor. Angiotensin II may affect cGMP through the AT₂ receptor, and neurotransmission by acetylcholine and excitatory amino acids has long been known to be associated with elevated cGMP levels. Therefore, nitric oxide could be involved in short-term effects of these substances. It could be involved in long-term potentiation of synaptic transmission linked to learning and memory formation, in effects on brain development, as well as in the biological effects of other neurotransmitters whose actions involve cGMP (399, 400).

Some work indicates that nitric oxide is inhibitory. In one experiment, activation of the L-arginine-nitric oxide pathway by injection of L-arginine into the lateral ventricle of water-deprived rats or rats given a large intracranial dose (250 ng) of ANG II inhibited the drinking caused by these two stimuli in a dose-dependent fashion (82, 83). The POA was suggested as one of the central sites for the antidipsogenic activity of nitric oxide because a smaller dose of L-arginine that was ineffective when injected into the lateral ventricle inhibited drinking by water deprived-rats when injected into the POA. The D-enantiomer was inactive, and inhibition of water deprivation-induced drinking by L-arginine could be reversed by antagonizing guanylate cyclase with methylene blue and nitric oxide synthase with N^G-nitro-L-arginine methyl ester (L-NAME). The effect of L-NAME on the L-arginine attenuation of ANG II-induced drinking was not examined. Acetylsalicylic acid injected into the POA also antagonized the antidipsogenic effect of POA injection of L-arginine, suggesting that the antidipsogenic action requires prostaglandin synthesis. Nitric oxide may also be inhibitory to sodium appetite; rats maintained on 1.0% NaCl containing L-NAME for 6 wk showed increases in NaCl intake compared with the untreated controls, but there were no differences in intake between L-NAME-treated rats maintained on water and their controls (161).

Other evidence supports a stimulatory role for nitric oxide in drinking behavior, including ANG II-induced drinking. Central inhibition of nitric oxide synthase with N^G-monomethyl-L-arginine has been found to attenuate drinking by a small amount in rats that had been deprived of water for 24 h, and to reduce glucose utilization in the SFO without affecting the enhanced glucose utilization of water deprivation in the hypothalamo-hypophysial system (297). Intracerebroventricular injection of L-arginine or the nitric oxide donor diethylenetriamine nitric oxide adduct augmented drinking by water-deprived rats (409). Recently, it has been found that intracerebroventricular injection of L-NAME (125 and 250 µg) antagonized drinking in response to intracerebroventricular ANG II (25 pmol) and that this effect could be reversed by coinjection of L-arginine (690). However, L-NAME did not alter the pattern of ANG II-induced c-fos expression in the SFO, OVLT, MnPO nucleus, SON, and PVN, suggesting that c-fos expression is not a necessary part of ANG II-induced drinking, although it serves as an excellent marker of neuronal activation (see sect. VIA). There was a high rate of codistribution of NADPH-diaphorase (a nitric oxide synthase) staining and c-fos expression in response to intracerebroventricular injection of ANG II but a low rate of colocalization of the two markers to individual cells, which also suggests that ANG II stimulates nitric oxide production and c-fos expression in different populations of neurons. These experiments suggest that ANG II interacts with nitric oxide in the control of drinking but not in c-fos expression.

Adrenomedullin is a 52-amino acid peptide first isolated from human pheochromocytoma tissue, found in many tissues including the brain, especially in the thalamus and hypothalamus (519). It is vasodilator, diuretic, and natriuretic. It acts on both parts of the adrenal gland to inhibit catecholamine and aldosterone secretion and on the pituitary to inhibit ACTH. When injected into the lateral ventricle, adrenomedullin caused a dose-dependent reduction in the amounts of water drunk in response to intracerebroventricular ANG II (100 pmol), hyperosmolality, and overnight deprivation of water, without affecting blood pressure, heart rate, or motor activity (410).

Glucagon-like peptide (GLP)-1-(7--36) amide is a gastrointestinal hormone found in the L cells of the gut, belonging to the glucagon/secretin group of peptides. Both GLP-1-(7--36) and its pancreatic receptors are important in the control of blood glucose, but GLP-1 receptors are also present in rat brain in the hypothalamus and CVOs. Injection of GLP-1-(7--36) amide into the lateral ventricle inhibited food intake in rats, but intraperitoneal injection was ineffective. The GLP-1-(7--36) amide was also antidipsogenic, with both intracerebroventricular and intraperitoneal administration causing marked inhibition of intracerebroventricular ANG II-induced (50 ng) drinking, whereas GLP-1-(1--36) was inactive (595). The inhibitory effect was accompanied by diuresis and natriuresis and was prevented by coinjection of the GLP-1 antagonist exendin-(9--39) amide.

Histamine-containing neurons in the posterior hypothalamus run via the medial forebrain bundle to limbic and thalamic structures and large areas of the cortex and hippocampus. Many functions have been ascribed to this diffuse histaminergic system, including arousal, autonomic and endocrine function, temperature regulation, and drinking behavior. Histamine injected into the POA, LH, and LHA caused dose-dependent drinking in rat (334), but how this is related to the central dipsogenic action of ANG II is not known. Histamine is also a systemic dipsogen, the effects of which are partly mediated through renal renin. This is discussed in section IXD.

Neuropeptide Y (NPY) is a 36-amino acid peptide belonging to the pancreatic polypeptide family of hormones and is a central regulator of energy homeostasis (624, 661). It is expressed by neurons in brain and the periphery and acts on three distinct receptor types, Y₁ , Y₂ , and Y₃ . It is widely distributed in the central nervous system, where it is colocalized with norepinephrine and epinephrine in catecholamine cell groups in the brain stem that send afferent fibers to the PVN (555). Levels of NPY are particularly high in the hypothalamus. In the rat, injection of NPY into the PVN caused a large dose-dependent increase in feeding and some increase in water intake (555), but in the mouse, intracerebroventricular NPY inhibited drinking associated with feeding or resulting from water deprivation (404). The hypothalamic receptor for feeding behavior is a subtype of Y₁ . Occurrence of NPY with catecholamines in the brain stem of rat suggests a possible connection with ANG II-induced drinking, but there is no evidence that this is the case.

Neurotensin is a tridecapeptide originally isolated from bovine hypothalamus, and it is a putative neurotransmitter. It is localized in neuronal cell bodies, receptors exist, and its release from storage vesicles in nerve terminals is calcium dependent. Iontophoretic application increases neuronal firing. Neurotensin is released from the intestine during feeding, and when injected into the PVN, it caused dose-dependent reductions in food intake but it had no effect on water intake (554). In contrast to the lack of effect of intracranial neurotensin on drinking, intravenous administration caused an increase in water intake in food-deprived rats. It may therefore contribute to food-associated drinking. It is not known whether neurotensin affects sodium appetite. Given intravenously, neurotensin is hypotensive and causes histamine release. Drinking could therefore be mediated by renal renin.

Pituitary adenylate cyclase activating polypeptide 38 (PACAP-38) has been isolated from ovine hypothalamic tissue. It is widely distributed in peripheral tissues and in the central nervous system in rats with the highest concentrations in the hypothalamus. Also, PACAP-38 caused dose-dependent increases in water intake when injected bilaterally into the perifornical lateral hypothalamus (459).