Corresponding author J. C. Geerling: Washington University School of Medicine Department of Anatomy & Neurobiology, Box 8108, 660 South Euclid Avenue, St Louis, MO 63110, USA. Email: email@example.com
Sodium appetite, the behavioural drive to ingest salt, is stimulated by prolonged physiological sodium deficiency in many animal species. The same neural mechanisms that are responsible for sodium appetite in laboratory animals may influence human behaviour as well, with particular relevance to the dietary salt intake of patients with diseases such as heart failure, renal failure, liver failure and salt-sensitive hypertension. Since the original experimental work of Curt Richter in the 1930s, much has been learned about the regulation of salt-ingestive behaviour. Here, we review data from physiology, pharmacology, neuroanatomy and neurobehavioural investigations into the stimulatory and inhibitory signals that regulate sodium appetite. A rudimentary framework is proposed for the brain circuits that integrate peripheral information representing the need for sodium with neural signals for the gustatory detection of salt in order to drive a motivated ingestive response. Based on this model, areas of remaining uncertainty are highlighted where future information would allow a more detailed understanding of the neural circuitry responsible for sodium appetite.
‘Some seek not gold, but there lives not a man who does not need salt.’ Cassiodorus
Sodium chloride, more commonly known as salt, is a necessary part of our diet, yet sodium is more than just an essential micronutrient. Wars have been fought over the control of environmental sources of salt, which was a more valuable commodity than gold at various points in history (Kare et al. 1980). The English word ‘salary’ derives from the Latin word for salt (salarium), which was once used to pay Roman soldiers. Today, a productive person is said to be ‘worth his salt’.
Other prominent references to salt are sprinkled throughout history. The poet Homer referred to it as the ‘divine substance’, and Plato noted that salt is ‘especially dear to the gods’ (Kurlansky, 2002). The word ‘salt’ appears more than 50 times in the Bible, in well-known verses such as ‘You are the salt of the earth’ (Matthew 5: 13). Shakespeare's King Lear was based upon a popular European folk story in which a king banishes his daughter for saying that she loves him ‘like salt’, but then forgives her after attempting to live without it.
Salt remains a major ingredient (some would argue that it is too major an ingredient) in our modern diet. We do not crave other minerals, such as magnesium, iodine or potassium, in the way that we desire the taste of sodium. The controversial link between dietary sodium and hypertension (elevated blood pressure) has led a number of groups, including the American Medical Association, to call for sweeping reductions in dietary salt content, particularly in processed foods. While epidemiologists clash over the harms and benefits of different amounts of dietary sodium in healthy versus hypertensive individuals (Taubes, 1998; Mitka, 2004), this review will consider a separate question: why is salt so appealing? Here, we review many of the physiological conditions and neural mechanisms discovered in laboratory animals to help explain the allure of salt.
Why do we need to eat salt?
Body fluid balance is meticulously regulated by neuroendocrine control systems. After a change in the volume or content of the extracellular fluid (ECF; which includes blood plasma), these control systems enact appropriate compensations to bring it back within narrow limits. In the equations of body fluid balance, the two primary variables are water and sodium. Balance is maintained through complementary adjustments in their ingestion and excretion.
The control systems that regulate the intake and excretion of water are the primary means for adjusting the concentration of solutes in the extracellular fluid, whereas control of both water and sodium is necessary for maintaining a volume of blood sufficient for optimal tissue perfusion by the heart (Verbalis, 2003). The basic regulatory mechanisms for controlling water intake (thirst), water excretion and sodium excretion are fairly well characterized and widely appreciated (Andersson, 1978; Stricker & Sved, 2002; McKinley et al. 2004), but less is understood about the regulation of sodium intake.
Sodium is, by far, the most abundant extracellular solute. The osmolarity of the extracellular fluid (including blood plasma) is dictated primarily by the concentration of sodium and its attendant anions (Verbalis, 2003). Extracellular osmolarity is meticulously maintained near a set point of roughly 290 mosmol l−1 (the vast majority of which is composed of ∼140 mm Na+, plus its attendant anions, primarily chloride).
In order to appreciate the role of sodium in body fluid homeostasis, it is critical to understand that extracellular osmolarity is regulated primarily by the ingestion and excretion of water, whereas the volume of extracellular fluid is directly proportional to the total body content of sodium (Verbalis, 2003). That is, the total volume of extracellular fluid in the body depends largely upon the amount of sodium present in the extracellular space, around which water input and output are tailored to tightly control osmotic pressure. This regulatory arrangement is the reason that sodium must be excreted to reduce plasma volume, and it must be ingested and retained to increase plasma volume.
Expansion of the ECF volume (a requirement for growth and for replacing fluid losses) is therefore absolutely limited by dietary sodium intake. Water intake alone is adequate to replace volume losses only when the sodium concentration is elevated owing to a loss of water greater than the loss of sodium. In contrast, as shown in Fig. 1, fluid losses involving large amounts of sodium (prolonged sweating or bleeding, for example) cannot be adequately replaced by water intake alone (Stricker & Jalowiec, 1970; Nose et al. 1988).
In such situations, consuming too much fluid without salt will produce hyponatraemia, which typically inhibits further drinking (Stricker, 1969). Overdrinking water, known as ‘water intoxication’, can be fatal, as in the case of a woman who died after ingesting two gallons of water as part of a radio station contest (Associated Press, 2007). Occasionally, this problem is encountered by endurance athletes, such as marathon runners, who can become dangerously hyponatraemic after drinking large volumes of fluid in excess of their sweat losses (Almond et al. 2005). If salt is consumed in addition to water, however, the body can retain an isotonic mixture of ingested sodium and water to more effectively restore blood volume (Stricker & Jalowiec, 1970; Nose et al. 1988; Twerenbold et al. 2003).
Since sodium represents less than 1% of the extracellular fluid by weight, the mass of sodium required for volume restoration is quite small relative to the amount of water that must be consumed. Accordingly, antidiuresis and thirst are the primary homeostatic drives stimulated in response to fluid loss and, initially, the most critical aspect of sodium regulation is retention by the kidney.
When the ECF volume expands, the blood pressure rises, and the increased perfusion pressure causes the kidneys to excrete more sodium (Guyton, 1991). Conversely, when the ECF volume is reduced, sodium is retained. This retention is primarily mediated by the steroid hormone aldosterone. Elevated levels of aldosterone, which is produced in the adrenal glands, can stimulate near-total conservation of sodium from the urine. This remarkable regulatory mechanism is critical for survival. Removal or gross dysfunction of the adrenal glands is invariably lethal without either exogenous replacement of aldosterone or continuous dietary supplementation of sodium (Richter, 1936; Wilkins & Richter, 1940).
Eventually, however, sodium conservation is only half the battle; the kidneys can retain only what is already present in the body. The other important aspect of this control system is the regulation of salt intake. Under normal conditions, obligatory sodium losses are small and, when necessary, the kidneys can maintain near-total urinary sodium conservation for extended periods of time. This allows animals to survive for many weeks on a sodium-free diet (Orent-Keiles et al. 1937; Fine et al. 1987a). Ultimately, however, ECF volume can be neither increased nor restored without the consumption of sodium. The chronic volume deficit and the secondary increase in plasma potassium that result from sodium deprivation lead insidiously to severe health consequences. Chronic sodium deprivation causes growth retardation, reproductive deficits, reduced muscle mass, alterations in bone composition and various other pathologies, which are eventually lethal (Orent-Keiles et al. 1937; Bursey & Watson, 1983; Fine et al. 1987a).
Put simply, normal growth requires the ingestion and retention of sodium. Without dietary salt, growth slows, reproduction fails, and animals die prematurely (Orent-Keiles et al. 1937; Fine et al. 1987a,b). In humans sustained on sodium-free nutrient infusions, bone mineralization ceases and growth stops in all tissues except fat (Rudman et al. 1975). Even short-term sodium deficiency in humans causes severe muscle cramps, loss of appetite, nausea, fatigue and considerable weight loss (McCance, 1936).
Given these severe health consequences, particularly the deficits in growth and reproduction, it should come as no surprise that a hard-wired behavioural mechanism has evolved to promote salt intake in response to a prolonged sodium deficiency.
What is sodium appetite and why is it important?
Sodium appetite (also known as salt appetite) is a motivated behavioural state that arises in a number of species specifically as a response to sodium deficiency. As the name indicates, it drives an animal to seek and ingest foods and fluids that contain sodium. Sodium appetite is a hard-wired regulatory mechanism and, like thirst, it is vital for restoring extracellular fluid.
It is important to note that when salty foods or fluids are freely available, animals (including humans) spontaneously exhibit a baseline or ‘need-free’ level of intake in excess of any immediate need or growth requirement. This baseline ingestion of salt (and water) is more than adequate for maintaining fluid balance in the absence of significant fluid loss, and any excess sodium or water is simply excreted in the urine. The magnitude of ‘need-free’ salt intake can be influenced by prior episodes of sodium deficiency (Sakai et al. 1989), especially prenatal experience with maternal illness during pregnancy (Nicolaidis et al. 1990; Crystal & Bernstein, 1995, 1998). It remains unclear, however, to what extent spontaneous salt-ingestive behaviours engage the same brain circuits that are responsible for sodium appetite, which is operationally defined as a specific response to sodium deficiency.
Abundant anecdotal evidence for sodium appetite existed for centuries (Kare et al. 1980; Denton, 1982), but a direct experimental demonstration awaited the seminal work of Curt Richter (1936). At this time, it was recognized that removal of the adrenal glands rendered animals unable to conserve urinary sodium owing to the loss of a vital ‘mineralocorticoid’ hormone (aldosterone) produced in the adrenal cortex. Unless their diet was continually supplemented with sodium, adrenalectomized animals deteriorated rapidly and died after roughly 1 week. Richter wanted to know whether animals possess an innate behavioural mechanism that would compel them to seek and ingest extra salt if it suddenly became necessary for their survival. When he gave adrenalectomized rats access to saline, they drank greatly increased amounts, even at a high concentration (3% NaCl; roughly the concentration of seawater), which they had only sampled in small amounts prior to surgery. This dramatic behavioural change is evident in one of his original charts, reproduced in Fig. 2. The voluntary increase in salt intake by these rats was more than sufficient to compensate for their urinary sodium losses, allowing their continued survival.
That this change in ingestive behaviour occurred specifically in response to sodium deficiency was confirmed by the demonstration that saline intake returned to normal when functional adrenal tissue was transplanted back into adrenalectomized rats (Richter & Eckert, 1938). Likewise, their increased saline intake vanished when sodium conservation was re-instated using replacement-dose mineralocorticoid injections (Wolf, 1965; Fregly & Waters, 1966), but promptly reappeared when hormone replacement was withdrawn (McEwen et al. 1986; Tordoff et al. 1993).
Appropriately, the appetite stimulated by sodium deficiency is highly specific for the taste of sodium salts (Richter & Eckert, 1938). Sodium-deficient rats consistently choose sodium over non-sodium salts (potassium, calcium, etc.), and the paired anion (chloride, acetate, etc.) has little or no effect on this preference (Nachman, 1962).
Locating salt remains a necessity for many animals living in the wild, particularly herbivores in sodium-impoverished environments (Blair-West et al. 1968), but not for humans. Sodium is now widely available and present in copious amounts in our diet, such that most people never experience a deficit that is prolonged or severe enough to stimulate sodium appetite. Older clinical trials that failed to produce hypovolaemia (a sustained sodium deficiency) have been cited as evidence against the existence of sodium appetite in humans (Bertino et al. 1982; Beauchamp et al. 1983, 1987), but clinical observations and subsequent experiments involving true sodium deficiency suggest otherwise.
In 1940, Wilkins and Richter described the salt cravings of a child suffering from undiagnosed adrenal disease (probably the severe salt-wasting form of congenital adrenal hyperplasia; see Kochli et al. 2005). As described in this fascinating case report, the boy expressed an extreme and persistent desire for both salt and water from a very early age (Wilkins & Richter, 1940). Even before he could speak, he would demand salt on everything he ate, and ‘salt’ was one of the first words he learned. He did not care for sugar or sweets, preferring instead to eat pure salt or salty foods. Like Richter's adrenalectomized rats, this boy needed to ingest sodium regularly just to survive; shortly after he was forced to eat a standard hospital diet and denied access to salt, he died (his adrenal disease was not diagnosed until autopsy).
More recent investigators have provided evidence for increased voluntary salt intake by children with salt-wasting congenital adrenal hyperplasia, in which adrenal steroid production is lacking as a result of mutations in the gene encoding the enzyme 21β-hydroxylase (Kochli et al. 2005). Similarly, children with Gitelman's syndrome, a genetic salt-wasting disorder caused by a defective renal sodium transporter, eat increased amounts of salt (relative to their unaffected family members) in order to maintain an ECF volume adequate to maintain a low-normal blood pressure (Cruz et al. 2001).
Increased sodium appetite has also been reported in adults: in renal patients after dialysis (Leshem & Rudoy, 1997) and in normal subjects tested after sustained sodium deficiency (Takamata et al. 1994). In the latter study, healthy adult subjects were sodium-depleted by exercise in a heated room, which caused large sweat losses of sodium. After remaining sodium deficient over the following day (with only water and salt-free food available for consumption), subjects' palatability ratings for the taste of concentrated saline peaked with a time course resembling the delayed appearance of sodium appetite in laboratory animals after peritoneal dialysis or drug-induced diuresis (Ferreyra & Chiaraviglio, 1977; Rowland & Morian, 1999). More rigorous laboratory studies of sodium appetite, in humans experiencing a more prolonged sodium deficit, would be useful for determining the influence of sodium appetite on human salt consumption in various physiological and pathophysiological conditions.
Understanding the mechanisms that influence salt intake in humans is important because excess sodium has damaging effects in select groups of people. In most healthy individuals, blood pressure only varies slightly (if at all) even with large changes in dietary salt intake (Graudal et al. 1998; Taubes, 1998; Sacks et al. 2001). Nonetheless, excess dietary sodium is an important risk factor for increasing numbers of patients who suffer from a variety of chronic medical conditions. Salt restriction is a cornerstone of successful therapy for many diseases, including congestive heart failure, liver failure, kidney failure and salt-sensitive hypertension. Unfortunately, patients with these diseases are notoriously non-compliant when they are told to eat a low-sodium diet (Korhonen et al. 1999; Ohta et al. 2004). The normal palatability of sodium certainly contributes to their non-compliance, but it is also likely that many patients experience paradoxical increases in sodium appetite as an attendant symptom of their disease (Langford et al. 1977; DiNicolantonio et al. 1982; Hurley et al. 1987; Leshem & Rudoy, 1997; Francis et al. 2001).
When the mechanisms responsible for sodium appetite are better understood, it may be possible to design evidence-based therapies that aid patients in reducing their hunger for salt. Some currently available drugs are already known to decrease sodium appetite in animal models. For example, the mineralocorticoid receptor antagonist spironolactone, whose beneficial effects in the treatment of human heart failure are firmly established (Pitt et al. 1999), also prevents the increased salt intake normally exhibited by rats in heart failure (Francis et al. 2001). Patients may live longer, healthier lives if this or other inhibitors of sodium appetite can assist them in reducing their salt intake, particularly with drugs already shown to decrease morbidity and mortality (Pitt et al. 1999, 2003).
Beyond any therapeutic potential, however, understanding the neural foundations of sodium appetite will expand our understanding of appetitive brain circuits in general. This most basic of appetites is an important model for other motivated behaviours. The brain circuits that detect the need for sodium and drive salt intake may share components in common (or in parallel) with circuits that control other appetitive drives, such as hunger and thirst. A full understanding of sodium appetite will require an integrated appreciation of the physiological, endocrinological and neural changes that stimulate it and, ultimately, identification of the specific brain circuits that integrate these inputs and drive the consumption of salt.
What causes sodium appetite?
Sodium appetite characteristically increases after a prolonged period of sodium deficiency, but the exact mechanisms responsible for this behavioural change remain incompletely understood. As explained above, a reduction in body sodium content is equivalent to a loss of extracellular fluid volume. Within 1–2 h after the onset of hypovolaemia, an increase in thirst becomes evident as water is consumed in direct proportion to the volume of fluid lost from the extracellular space (Fitzsimons, 1961; Stricker, 1968). In contrast to thirst, the appetite for sodium does not increase until well after the onset of hypovolaemia; many hours or days later, depending upon the experimental model (Fig. 3). Initially, this distinction led to the ‘reservoir hypothesis’ suggesting that, analogous to the central osmoreceptors cells that shrink as a result of increased ECF osmolarity and stimulate thirst, there exist specialized cells or tissues that produce a delayed stimulus for sodium appetite as they gradually release extra sodium into the ECF down an increasing concentration gradient (Wolf & Stricker, 1967).
Sodium appetite, however, is not caused by a decrease in the plasma concentration of sodium (hyponatraemia), despite the intuitive appeal of such an arrangement. Most stimuli for sodium appetite, including dietary sodium deprivation, do not reduce the plasma sodium concentration (McCance, 1936; Bojensen, 1966; Contreras & Hatton, 1975; Stricker et al. 1991). Conversely, hyponatraemia alone (without hypovolaemia) does not increase saline intake (Stricker & Wolf, 1966). Hyponatraemia and sodium appetite can occur together under more extreme circumstances, but in most of these models, sodium appetite generally does not increase until hours after the plasma sodium concentration returns to normal (Ferreyra & Chiaraviglio, 1977). Also, the amount of salt ingested after such experimental manipulations remains unaffected even when additional measures are taken to prevent hyponatraemia. For example, adrenalectomized rats can become severely hyponatraemic (Wolf & Stricker, 1967; Jalowiec & Stricker, 1973), yet they exhibit the same increase in saline intake if their hyponatraemia is prevented by water restriction (Wolf & Stricker, 1967); a similar observation was made in sodium-deficient sheep (Beilharz et al. 1962). Multiple investigators have demonstrated that hypernatraemia inhibits salt intake (Weisinger et al. 1983; Chiaraviglio & Perez Guaita, 1986; Blackburn et al. 1995; Watanabe et al. 2000), but these examples clearly demonstrate that hyponatraemia is neither necessary nor sufficient for stimulating it.
Another difference between sodium appetite and thirst is that the amount of salt ingested generally overestimates the sodium deficit. This latter difference between water and salt intake is best illustrated by the natural stimulus for sodium appetite: dietary sodium deprivation. In rats, only 1–2 mmol of sodium are lost during the first 1–2 days of sodium deprivation, followed by near-total urinary retention as aldosterone levels increase (Contreras & Hatton, 1975; Stricker et al. 1991; Lane et al. 1997). Nonetheless, sodium deprivation for an additional 6–7 days stimulates as much salt intake as other manipulations that produce much larger volume deficits (see Fig. 8 of Scheidler et al. 1994). Thus, while sodium appetite is an appropriate response to hypovolaemia in a teleological sense, its magnitude is not always proportional to the original decrease in extracellular fluid volume.
What, then, are the physiological signal(s) that instruct the brain to seek and ingest sodium? This important issue remains unsettled, and many proposed answers remain controversial. Figure 4 highlights a number of the stimulatory and inhibitory factors discussed below.
Aldosterone Aldosterone production in the adrenal cortex is markedly elevated by prolonged sodium deprivation, independent of any change in adrenal glucocorticoid production or angiotensin II (AII; Bojensen, 1966; Okubo et al. 1997; Makhanova et al. 2006). In fact, dietary sodium deprivation is the most potent physiological stimulus for aldosterone production. This steroid hormone promotes two complementary processes to maintain or increase extracellular fluid volume: (1) renal sodium conservation (discussed above) and (2) sodium appetite.
Although Richter's original experiments with adrenalectomy-induced sodium appetite showed that this hormone is not necessary for sodium appetite, he subsequently discovered that treatment with high-dose adrenal mineralocorticoids stimulated the ingestion of large volumes of saline, even in control animals (Rice & Richter, 1943). In low doses that merely restore baseline renal sodium conservation, the administration of an adrenal mineralocorticosteroid (aldosterone or others, such as deoxycorticosterone) reduces sodium appetite in an adrenalectomized animal (Wolf, 1965; Fregly & Waters, 1966; McEwen et al. 1986; Tordoff et al. 1993). Higher doses, however, robustly stimulate the ingestion of large amounts of saline in both adrenalectomized and adrenal-intact rats (Wolf, 1965; Fregly & Waters, 1966).
The ingestive behaviour stimulated by mineralocorticoids is uniquely specific for sodium. In sharp contrast to other stimuli, such as angiotensin II, mineralocorticoids have little or no effect on water intake (Braun-Menendez, 1950; Wolf et al. 1974; Vallon et al. 2005; Geerling & Loewy, 2006c). Unlike the less-than-straightforward pharmacological manipulations required to increase salt intake using systemic infusions of AII (Fitts & Thunhorst, 1996; reviewed by Weisinger et al. 1996), there is no ambiguity regarding the robust ability of systemic mineralocorticoid administration to selectively increase sodium appetite in rats. Unfortunately, this stimulation requires repeated administration of large and generally supraphysiological doses over several days. Acute administration of physiological doses of aldosterone only produces a small and variable increase in salt ingestion (Wolf, 1964b; Wolf & Handal, 1966).
Therefore, although the specificity of aldosterone for stimulating salt intake is unique, its physiological role probably lies more in gradually boosting sodium appetite, and not as an acute stimulus for this behavioural state. An important role has been demonstrated for the activation of mineralocorticoid receptors (MRs) in the brain (Sakai et al. 1986; Francis et al. 2001; Sullivan et al. 2004), although saline intake after diuresis was not reduced by complete inhibition of adrenal steroid synthesis (Rowland & Morian, 1999). The effects of aldosterone are magnified greatly by a concurrent elevation in adrenal glucocorticoids [especially corticosterone, a combined MR and glucocorticoid receptor (GR) agonist that readily enters the brain; Wolf, 1965; Ma et al. 1993; Zhang et al. 1993] or by an intracerebral injection of the peptide hormone angiotensin II (Fluharty & Epstein, 1983). Also, the unique specificity of mineralocorticoids for salt ingestion led to the identification of a unique group of aldosterone-sensitive neurones in the brainstem, which are activated specifically in association with sodium appetite (Geerling et al. 2006a; discussed below, under the subheading Where is sodium appetite regulated in the brain?).
Angiotensin II During hypovolaemia, an elevation in circulating AII is an important stimulus for thirst (Fitzsimons, 1998). Like thirst, circulating AII production increases rapidly in response to hypovolaemia. Following the release of renin from cells in the juxtaglomerular apparatus of the kidney, AII is generated by the enzymatic cleavage of circulating angiotensinogen (produced in the liver) to angiotensin I (AI), which is further processed to AII by angiotensin-converting enzyme (ACE). AII plays an important role in the stimulation of sodium appetite, but does not, on its own, selectively stimulate the ingestion of salt relative to water. Many details regarding the contribution(s) of AII to the stimulation of sodium appetite remain a matter of controversy.
Direct infusion of AII into the brain rapidly stimulates a large increase in water intake, and increases the ingestion of saline (Buggy & Fisher, 1974; Avrith & Fitzsimons, 1980). In sharp contrast to the selective increase in salt intake that is stimulated by dietary sodium deprivation (or high-dose mineralocorticoid administration), however, AII stimulates a high ratio of water-to-saline intake (see, for example, data of Prakash & Norgren, 1991). To reconcile this discrepancy with a possible role for AII in sodium appetite, it has been suggested that AII separately influences thirst versus sodium appetite via the differential activation of separate intracellular signalling cascades under various physiological conditions (Daniels et al. 2005). Also, as discussed in a subsequent subsection (The disinhibition hypothesis), larger increases in AII-stimulated saline intake have been obtained after the inhibition of central oxytocin release (Blackburn et al. 1992a; but see Fitts et al. 2003), or after the injection of various neurotransmitter antagonists into a viscerosensory region in the brainstem that mediates the viscerosensory feedback inhibition of ingestive behaviour (Menani et al. 1996; Menani & Johnson, 1998).
In an apparent contradiction to the hypothesis that peripherally generated AII is a key stimulus for sodium appetite during hypovolaemia, intravenous administration of AII rapidly increases water intake, but a substantial increase in saline intake does not occur until many hours later, and may be secondary to the systemic sodium depletion caused by AII-induced pressure natriuresis (Sakai et al. 1990; Yang & Epstein, 1991; reviewed by Weisinger et al. 1996). Also, boosting AII levels immediately after sodium depletion neither accelerates the onset of sodium appetite nor increases the volume of subsequent saline ingestion (Rowland & Morian, 1999; but see Fitts et al. 1985b). Intravenous administration of AII can, however, increase saline intake acutely if rats are first sodium depleted by diuretic administration on the previous day, then repeatedly administered large doses of captopril (to block endogenous AII production), and then infused with AII on the following day (Fitts & Thunhorst, 1996).
In another contradiction to the hypothesis that circulating AII is a critical stimulus for sodium appetite, the inhibition of AII production by an ACE inhibitor such as captopril somehow increases saline ingestion in rats (Evered & Robinson, 1983; Moe et al. 1984). This captopril-stimulated increase in salt intake can be prevented by the concurrent administration of an ACE inhibitor (or AII receptor antagonist) directly into the brain, or by the peripheral administration of extremely large doses of captopril (e.g. 100 mg kg−1), both of which reduce saline intake after physiological sodium deficiency as well (Elfont et al. 1984; Moe et al. 1984; Weiss et al. 1986). To reconcile these paradoxical findings with the hypothesis that AII is a critical stimulus for sodium appetite, it has been suggested that captopril, by increasing the circulating concentration of unconverted angiotensin I, allows a preferential increase in its conversion to AII in some brain site, presumably a region that is more accessible to blood-borne AI than to captopril, or that produces enough ACE locally to negate inhibition by captopril unless it is given at extremely high doses or infused directly into the brain (Lehr et al. 1973; Evered & Robinson, 1983; reviewed by Thunhorst, 1996).
It is important to note that the increase in sodium appetite (and thirst) stimulated by AII is much greater when it is administered directly into the brain ventricular system after chronic pretreatment with a mineralocorticosteroid (Fluharty & Epstein, 1983; Zhang et al. 1984; Massi & Epstein, 1990; Shade et al. 2002). The doses of AII used to stimulate salt intake generally produce supraphysiological concentrations of this peptide in the cerebrospinal fluid (CSF) and stimulate non-physiological patterns of neuronal activation relative to circulating AII (McKinley et al. 1995).
These findings led to the ‘synergy hypothesis’, which proposes that sodium appetite is the result of simultaneous elevations in peripheral aldosterone and AII produced by the brain's own renin–angiotensin system (Epstein, 1982). In support of this hypothesis, subthreshold doses of aldosterone (given peripherally) and AII (given centrally) synergize to stimulate more saline intake than either hormone alone (Fluharty & Epstein, 1983). Also, simultaneous pharmacological blockade of systemic MRs and AII receptors inside the brain prevented sodium appetite after furosemide diuresis (Sakai et al. 1986). Based upon various findings in diverse experimental models for sodium appetite, however, it remains unclear whether the actions of AII and aldosterone are sufficient for the stimulation of sodium appetite in response to physiological sodium deficiency (Coghlan et al. 1981; Weisinger et al. 1996, 1997a,b,c; Blair-West et al. 1997; Rowland & Morian, 1999), and additional signalling mechanisms are likely to play an important role.
Baroreceptor input Since sodium appetite characteristically arises during a sustained reduction in blood volume and is inhibited by hypervolaemia, the peripheral baroreceptor nerves that detect changes in central venous pressure seem well positioned to provide key information. Consistent with this possibility, the salt ingestion stimulated by prolonged hypovolaemia or mineralocorticoid treatment is greatly reduced if the right atrium is distended with a balloon cannula (mimicking the increase in venous return that occurs when blood volume increases), and then rebounds after the cannula is deflated (Toth et al. 1987). Part of this effect could be mediated by atrial or B-type natriuretic peptides, which inhibit salt intake when administered directly into the brain (Fitts et al. 1985a; Antunes-Rodrigues et al. 1986; Blackburn et al. 1995), but this finding also suggests a role for neural input from central venous baroreceptors in the inhibition of sodium appetite.
Whether venous or arterial baroreceptor signalling is relevant for the stimulation of sodium appetite remains unclear. Arterial blood pressure remains well compensated after dietary sodium deprivation or experimental reductions in ECF volume sufficient for the stimulation of sodium appetite (McCance, 1936; Stokes et al. 1986; Webb et al. 1987; Stricker et al. 1994). In fact, some investigators have even found an increase in arterial pressure during chronic dietary sodium deprivation (Webb et al. 1987). Nonetheless, saline ingestion may be potentiated by reduced arterial pressure (Thunhorst & Johnson, 1994). Conversely, sodium appetite was reduced in rats after transection of the nerves that transmit sensory information from arterial baroreceptors (Thunhorst et al. 1994), although no effect was found after destruction of the brainstem region innervated by these nerves (Schreihofer et al. 1999).
Intracerebral sodium concentration The sodium concentration of the CSF is directly related to that of the blood plasma (Doi et al. 1992). Available evidence in multiple species indicates that an increased sodium concentration in this fluid compartment not only stimulates thirst (Andersson, 1978), but also inhibits sodium appetite. For example, infusion of hypertonic saline directly into the brain ventricular system consistently reduced the salt intake of sodium-deficient sheep (Weisinger et al. 1979, 1982), and a similar finding was reported in hypovolaemic rats (Chiaraviglio & Perez Guaita, 1986).
Whether a reduction in intracerebral sodium increases sodium appetite remains unclear. Intraventricular sodium-free infusions of certain osmotically active molecules, such as mannitol or sucrose, increased salt intake in sodium-deprived and non-deprived sheep (Weisinger et al. 1979, 1982). Since one of the effects of these infusions was a reduction in CSF [Na+], these experimenters suggested that sodium appetite is increased by the activation of low-sodium sensors somewhere in the brain (separate from the high-sodium sensors that increase thirst and inhibit sodium appetite). However, when equivalent reductions in CSF [Na+] were produced by infusions of water or a cell-permeable solute (glucose), salt intake did not increase. This discrepancy was interpreted as evidence that cerebral low-[Na+] sensors are located somewhere behind the ventricular epithelial barrier (unlike the high-[Na+] sensors located in forebrain circumventricular organs, which lack a blood–brain barrier) and therefore only detect shifts in [Na+] across this boundary, within the neuropil (see discussion by Weisinger et al. 1982, 1985; Denton et al. 1996).
Both the significance and the location(s) of these hypothetical low-[Na+] sensors remain unknown. These findings could not be reproduced in other laboratory species, including rabbits, rats and mice (Denton et al. 1984; Frankmann et al. 1987; Osborne et al. 1990), and no evidence for a low-[Na+] sensing mechanism has been detected at the cellular or molecular level. Subsequent findings in transgenic mice, however, did confirm the existence of sensors for high-[Na+] in the brain as a unique concentration-activated sodium channel (discussed below, under the subheading Where is sodium appetite regulated in the brain?; Watanabe et al. 2000).
The disinhibition hypothesis The lack of evidence for a critical stimulus that increases sodium appetite (in a physiologically relevant model) led to a new school of thought, which suggested that the appetite for salt is stimulated essentially by the same molecular signal(s) as hypovolaemic thirst (primarily angiotensin II), but that it is usually held in check by a dominant inhibitory signal. To explain the extensive delay between the ingestion of water versus salt after the onset of hypovolaemia, this hypothesis invokes one or more central mechanisms that block sodium appetite (but not thirst) until a sufficient amount of water has been ingested. In this model, sodium appetite is gradually released from inhibition as water ingestion produces osmotic dilution of the extracellular fluid (Stricker & Jalowiec, 1970; Stricker & Verbalis, 1987). Although systemic osmotic dilution is not necessary in various conditions known to stimulate sodium appetite (including dietary sodium deprivation), key inhibitory mechanisms have been identified that are consistent with this model.
Thus, the brain possesses multiple mechanisms for limiting salt intake, but it remains uncertain whether, under physiological conditions, the unique dynamics of sodium appetite are completely explained by disinhibitory control alone. For example, experiments in knockout mice have confirmed a role for oxytocin as an inhibitory modulator of salt intake (Amico et al. 2001; Puryear et al. 2001; Rigatto et al. 2003), but the results did not support the hypothesis that this peptide is individually responsible for gating salt intake in response to stimuli such as AII (see also Polidori et al. 1994; Fitts et al. 2003). Likewise, destruction of the lateral parabrachial nucleus (a key brainstem relay site for ascending post-ingestive signals, in which injections of various neurotransmitter antagonists have been found to increase salt intake) does not appear to potentiate sodium appetite (Johnson & Thunhorst, 1997), despite robustly enhancing thirst (Ohman & Johnson, 1986; Edwards & Johnson, 1991). In fact, sodium appetite is reduced or eliminated, not enhanced, after lesions involving a neighbouring region in the dorsal pons (Flynn et al. 1991; Scalera et al. 1995). These and other findings suggest that sodium appetite is governed by multiple input signals, both stimulatory and inhibitory, that are integrated by the neuronal networks that regulate salt intake.
Where is sodium appetite regulated in the brain?
When a sodium-deficient animal tastes salt, it will excitedly ingest an unusually large quantity (see Supplemental Video). To explain this pronounced behavioural switch, we have proposed a model with three central components (Geerling & Loewy, 2006a), shown in Fig. 5. First, during chronic sodium deficiency, specific groups of neurones provide tonic, increasing signals for sodium need, which motivate salt-seeking behaviour. Second, once salt is tasted, the gustatory apparatus transmits a phasic signal representing sodium detection. Third, these two signals are integrated (along with various inhibitory signals) in one or more forebrain sites that ultimately drive motivated ingestive behaviour.
Determining the general locations of neurones that carry out each of these operations is a fundamental requirement for any model of the central circuitry that regulates sodium appetite. First, the critical involvement of neurones located somewhere in the brainstem is inherently obvious owing to the necessity of its motor neurones and premotor pattern generators for control of the ingestive musculature (Travers & Norgren, 1983), and due to the direct innervation of medullary sites by gustatory nerves from the oral cavity. Two questions arise, then. (1) Does the brainstem also contain cells that detect peripheral sodium deficiency-associated stimuli and generate a tonic signal for sodium need? (2) If so, does it also contain neurones that integrate this signal for sodium need with signals for salt taste detection and, in turn, promote ingestive motor activity? That is, can sodium appetite be reduced to a local, reflexive response within the brainstem?
While the first question remains a matter of debate, the answer to the second question is straightforward: the sensory–integrative–motor network responsible for sodium appetite is not a self-contained reflex pathway in the brainstem. We know that rostral connections must be involved because disconnecting the brainstem from the forebrain eliminates all behavioural evidence for sodium appetite (Grill et al. 1986; Flynn & Stricker, 2003), despite leaving various other hormonal and gustatory–motor reflexes intact (Flynn & Grill, 1983, 1988; Flynn et al. 1995). A critical role for neurones in the forebrain is also supported by evidence that sodium appetite can be eliminated by electrolytic lesions within a dorsolateral region of the hypothalamus (Wolf & Quartermain, 1967), placed in roughly the location of the medial forebrain bundle (a large collection of axonal interconnections between the brainstem and forebrain).
Surprisingly, however, extensive lesions of sensory and motor cortex do not reduce salt ingestion in response to sodium deficiency (Wolf et al. 1970; Wirsig & Grill, 1982). Subregions of medial prefrontal cortex and probably other cortical areas do appear to exert behaviourally relevant influences over salt ingestion (Chiaraviglio, 1984), but these findings indicate that the cerebral cortex is not necessary for the basic regulation of sodium appetite, and that the forebrain neurones primarily responsible for the integrative control of sodium appetite are located within subcortical nuclei.
A wide range of subcortical brain sites have been tested for potential influences on salt intake (see reviews in Wolf et al. 1974; Denton, 1982; Schulkin, 1991; Johnson & Thunhorst, 1997; Daniels & Fluharty, 2004). Rather than providing an exhaustive catalogue of these brain sites, the following summary offers a tentative framework of brain sites specifically linked to sodium appetite in the context of the model described above (their locations within the rat brain are shown in Fig. 6).
Sodium need: humoral and visceral sensory pathways Dehydration-induced thirst (water intake) is critically dependant upon neurones in the preoptic hypothalamus, lying along the anterior wall of the third ventricle (Andersson et al. 1975; Johnson & Buggy, 1978). This region of the brain, the lamina terminalis, contains specialized neurones and glial cells uniquely sensitive to elevations in extracellular osmolarity and sodium concentration (Oldfield et al. 1991; Watanabe et al. 2000, 2006; Hiyama et al. 2002, 2004). Many cells within this region are also directly sensitive to circulating AII, and are activated by increased concentrations of this peptide during hypovolaemia. Angiotensin II-sensitive neurones here are concentrated in two sensory circumventricular organs: the subfornical organ (SFO) and the organum vasculosum of the lamina terminalis (OVLT; Mendelsohn et al. 1984), both of which are critical for the stimulation of fluid-ingestive behaviour by AII (Simpson & Routtenberg, 1973; Buggy et al. 1975; Morris et al. 2002).
Interestingly, however, lesions of the SFO, OVLT and surrounding hypothalamic tissue only cause a partial and variable decrease in the stimulation of salt intake by prolonged sodium deficiency. For example, destroying the thirst-critical ‘AV3V’ (anteroventral third ventricular) region around the preoptic recess of the third ventricle (including the OVLT), which consistently abolishes the homeostatic control of water intake, does not eliminate sodium appetite in rats (Johnson & Buggy, 1978; De Luca et al. 1992) and has no detectable effect on sodium appetite in sheep (Weisinger et al. 1993). In some studies, lesions of the SFO did attenuate saline intake (Thunhorst et al. 1990; Weisinger et al. 1990), whereas other investigators found no effect (Andersson et al. 1975; Ruhf et al. 2001; Wilson et al. 2002). Destruction of the both the SFO and the OVLT produces a more consistent reduction in saline ingestion, but does not consistently eliminate it (Fitts et al. 2004).
Conversely, some of the sites interposed within the AV3V region appear to inhibit salt intake. Evidence for this inhibition initially derived from observations in goats and rats that both spontaneous and mineralocorticoid-stimulated saline intake were increased after lesions involving the OVLT and the adjacent median preoptic nucleus (Andersson et al. 1975; Gardiner et al. 1986; Fitts et al. 1990; Fitts, 1991).
Subsequently, Noda and colleagues uncovered a molecular mechanism for the inhibition of salt intake by specialized sodium detectors in the lamina terminalis (Noda, 2006). Many glial cells in the SFO and other parts of the lamina terminalis express a sodium-selective channel, NaX, which makes them uniquely sensitive to large increases in the extracellular sodium concentration (Watanabe et al. 2000, 2006; Hiyama et al. 2002, 2004; Grob et al. 2004). In mice, NaX in the SFO is necessary for the rapid stimulation of thirst by hypertonic saline infusion, as well as the inhibition of salt intake after 24 h water deprivation (Watanabe et al. 2000; Hiyama et al. 2004). Also, relative to wild-type mice, NaX knockout mice over-ingest saline when sodium appetite is stimulated by diuresis and overnight sodium deprivation (Watanabe et al. 2000).
Despite claims that the SFO is ‘essential and sufficient for the control of salt-intake behaviour’ (Noda, 2006), this channel is probably not directly involved in the stimulation of sodium appetite in response to sodium deficiency. The NaX channel is activated by large increases in the extracellular sodium concentration (greater than ∼150 mm, as occurs after prolonged water deprivation or after the ingestion of hypertonic saline), but cannot detect a deficit in body sodium content (or concentration). Accordingly, mice lacking the NaX channel still exhibit an increase in salt intake in response to prolonged sodium deficiency (Watanabe et al. 2000). Interestingly, although rats show a blunting of the rapid thirst response to central infusion of hypertonic saline after lesion of the SFO (similar to NaX knockout mice), their depletion-induced saline intake is decreased or unchanged, not increased (Thunhorst et al. 1990; Weisinger et al. 1990; Wilson et al. 2002; Fitts et al. 2004), suggesting the existence of both inhibitory and excitatory sensory elements within the SFO. Thus, NaX-mediated inhibition of salt intake in the SFO may operate indirectly, by inhibiting excitatory output neurones that increase the ingestion of water and salt. This implication is supported by the twofold increased neuronal activation in the SFO of NaX knockout mice versus wild-type mice after 24 h water deprivation (Watanabe et al. 2000), as well as the NaX-mediated stimulation of GABAergic interneurones within the SFO in response to hypertonic saline (Shimizu et al. 2007).
Thus, a variety of AII-sensitive circumventricular organs exert stimulatory and inhibitory effects on salt intake, none of which are individually necessary for the stimulation of sodium appetite by prolonged sodium deficiency. Collectively, these findings suggest that an important signal for sodium need is generated by cells in one or more brain sites not targeted in previous lesion studies.
One finding from these prior studies hinted that such neurones may be found immediately ventral to the AP, in the nucleus of the solitary tract (NTS). Specifically, salt intake increased very little (or not at all) in rats with AP lesions if collateral damage beneath this structure extended through any more than a limited region of the underlying subpostremal NTS (Contreras & Stetson, 1981; Wang & Edwards, 1997; Curtis et al. 1999). This finding suggests that the inhibitory effect of neurones in the AP may counterbalance a stimulatory signal arising from neurones in the NTS.
Aldosterone-sensitive neurones. Within this region of the NTS, we identified a group of neurones that generate an output signal specifically associated with sodium need. These neurones were originally identified based upon their unusual sensitivity to aldosterone (Geerling et al. 2006a,c). As shown in Figs 7 and 8, they express the mineralocorticoid receptor (MR) and the enzyme 11β-hydroxysteroid dehydrogenase type 2 (HSD2). This enzyme, which inactivates glucocorticosteroids (primarily cortisol in humans; corticosterone in rodents), is necessary for aldosterone sensitivity. This is because glucocorticoids, which bind to the MR with comparable affinity and circulate at 100- to 1000-fold higher concentrations, out-compete aldosterone in most MR-expressing cells. Only cells that express both MR and HSD2 are sensitized to the relatively low levels of circulating aldosterone (Funder et al. 1988; Naray-Fejes-Toth et al. 1998).
The aldosterone-sensitive HSD2 neurones are also unique in that, unlike most other neurones in the NTS (and most other neurones in the brain), they are selectively activated by prolonged sodium deficiency (Geerling et al. 2006a; Geerling & Loewy, 2007b). In response to a variety of stimuli for sodium appetite (including chronic dietary sodium deprivation, colloid-induced hypovolaemia, diuresis and high-dose mineralocorticoid administration), they exhibit a marked increase in nuclear c-Fos (an established marker for neuronal activity; see Sagar et al. 1988; Hoffman et al. 1993). Dietary sodium deprivation progressively increases the activation of the HSD2 neurones in parallel with the development of sodium appetite, as shown in Fig. 9. Then, after sodium-deprived rats drink saline, the HSD2 neurones are rapidly inactivated, in contrast to surrounding neurones in the NTS and AP, most of which exhibit marked activation after salt intake (Geerling et al. 2006a; Geerling & Loewy, 2006c, 2007a).
The HSD2 neurones, like sodium appetite, are still activated by dietary sodium deprivation after adrenalectomy, showing that they are more than just aldosterone sensors (Geerling et al. 2006a). They receive input from multiple brain sites implicated in the regulation of sodium appetite, including direct axonal projections from neighbouring neurones in the AP and dorsomedial NTS (Sequeira et al. 2006), a prominent descending projection from the central nucleus of the amygdala (Geerling & Loewy, 2006b) and a descending projection from the paraventricular hypothalamic nucleus (J. C. Geerling & A. D. Loewy, unpublished observations). Unlike many other neurones in the NTS, the HSD2 neurones do not innervate nuclei in the caudal brainstem that directly modulate the autonomic nervous system. Instead, their axons project rostrally, targeting nuclei in the pons and basal forebrain previously implicated in sodium appetite (for details, see below and Geerling & Loewy, 2006a,b).
Based upon their sodium appetite-specific activation, the HSD2 neurones clearly integrate stimuli associated with prolonged sodium deficiency. A confirmed role for the HSD2 neurones in stimulating sodium appetite would explain the previously paradoxical finding that systemic administration of an HSD2 antagonist (allowing unrestricted MR activation by glucocorticosteroids) increases saline intake, despite the simultaneous increase in sodium retention by the kidneys, which should otherwise inhibit sodium appetite (Cooney & Fitzsimons, 1996). Interestingly, a pronounced increase in salt appetite was also reported in a human with impaired HSD2 function (Ingram et al. 1996).
The aldosterone sensitivity, sodium appetite-associated activation and input/output connections of the HSD2 neurones suggest that they play a role in driving sodium appetite. If so, this group of cells could represent an attractive target for pharmacotherapy, especially given the increased blood–brain barrier permeability within this subregion of the NTS (Gross et al. 1990; Broadwell & Sofroniew, 1993). However, these provocative associations do not establish functional involvement, and a causal role for HSD2 neurones in the stimulation of sodium appetite remains to be tested.
While the precise molecular mechanisms for sodium detection and transduction within the taste buds remain unclear (Chandrashekar et al. 2006), the ascending axonal pathways that convey gustatory input through the brainstem and into the forebrain have been mapped in many neuroanatomy, electrophysiology and lesion studies (Travers et al. 1987). First, information from taste buds located in the tongue and oropharynx is collected by peripheral gustatory nerves (branches of cranial nerves VII, IX and X). While this information is carried by multiple nerves, the main contributor is the chorda tympani branch of cranial nerve VII, which innervates the anterior two-thirds of the tongue and contains the largest complement of sodium-sensitive fibres (Frankmann et al. 1996; Roitman et al. 1999; Blonde et al. 2006). As indicated in Fig. 10, these nerves directly innervate neurones in the rostral one-third of the lateral NTS (distant from the caudal-medial visceroceptive region of the NTS, containing HSD2 neurones below the AP).
Next, in rodents, gustatory neurones in the rostral NTS project to the ‘pontine taste area’ of Norgren, a caudal-medial subregion of the parabrachial complex (PB) in the dorsolateral pons (Norgren & Leonard, 1971; Norgren & Pfaffmann, 1975; Herbert et al. 1990). Lesions in this region virtually eliminate the behavioural expression of sodium appetite (Flynn et al. 1991; Scalera et al. 1995). This effect is presumably due to the interruption of gustatory signalling to the forebrain, although these lesions also involved the neighbouring pre-locus coeruleus (pre-LC), a site that is heavily innervated by HSD2 neurones in the NTS (Geerling & Loewy, 2006a) and selectively activated by sodium deprivation (Geerling & Loewy, 2007a). Therefore, although critical signals are relayed to the forebrain by neurones within the PB complex to regulate sodium appetite, the specific contributions of individual functional groups within the PB remain unclear because adjacent subnuclei within this complex region transmit qualitatively different sensory information associated with sodium need, gustatory salt detection and visceral sodium detection (Geerling & Loewy, 2007a). It should also be noted that in monkeys (and probably, by extension, in humans) most axons from gustatory neurones in the rostral NTS bypass this pontine relay site and instead project directly to the forebrain (Beckstead et al. 1980).
Before continuing to a discussion of the forebrain sites that integrate signals for sodium need and salt detection, gustatory-centric models for explaining the consummatory phase of sodium appetite should be considered. Unlike the present model, such models emphasize functional changes at one or more points along these gustatory sensory pathways as the reason for increased salt intake after prolonged sodium deficiency.
Originally, Richter noted that adrenalectomized rats ingest larger volumes of extremely dilute saline, which normal animals did not seem able to distinguish from water, based upon altered ‘preference’ ratios. He interpreted this difference as an increase in the sensory threshold for sodium (Richter, 1939). Subsequently, it was found that neither the sensory threshold for gustatory nerve activation nor the behavioural threshold for discriminating dilute saline solutions (as measured by a more sensitive operant protocol) were increased by sodium depletion (Pfaffmann & Bare, 1950; Carr, 1952; Nachman & Pfaffmann, 1963). Thus, during sodium deficiency, it appears that the motivation to ingest salt increases, not the ability to detect it.
The behavioural significance of the above changes in gustatory sensory processing remains uncertain. Reduced responsiveness of taste neurones in the rostral NTS (and the reduction in sharply defined taste categories; Jacobs et al. 1988) could explain the dulled taste sensations reported by humans during sodium deprivation (McCance, 1936). However, changes in gustatory processing alone cannot explain the motivated salt-seeking behaviour exhibited during sodium deficiency. For example, sodium-deficient animals will trek extended distances to locate and ingest salt (Blair-West et al. 1968; Denton, 1982) and will perform increased amounts of work to obtain it (Wagman, 1963; Quartermain et al. 1967; McCutcheon & Levy, 1972). Such motivational changes are more likely to be driven by sites in the forebrain that integrate multiple signals associated with sodium homeostasis.
Forebrain integration of the signals for sodium need and salt detection A logical first step towards identifying the forebrain sites that regulate sodium appetite is to consider sites that receive input signals specifically associated with sodium need (originating in the lamina terminalis or from the HSD2 neurones of the NTS, for example) as well as signals for salt detection (via ascending gustatory pathways). Based upon neuroanatomical tracing data, major sites of convergence can be found within two sites in the central extended amygdala complex: the ventrolateral bed nucleus of the stria terminalis and the central nucleus of the amygdala. For each of these sites, connectional data are complemented by experimental demonstrations of lesion-induced reductions in sodium appetite. Below, existing evidence for each of these sites is discussed, along with additional forebrain regions implicated in the integrative regulation of sodium appetite.
The BSTvl sends massive output projections to the substantia innominata, lateral hypothalamic area (LHA), paraventricular hypothalamic nucleus and a number of other brain regions (Dong et al. 2001b), placing it in a key position to co-ordinate changes in arousal and ingestive behaviour with neuroendocrine and autonomic activity. The LHA, it should be noted, is the only forebrain site in which discrete lesions reliably eliminated sodium appetite (Wolf & Quartermain, 1967). The BSTvl projection to the LHA provides substantial input to the orexin neurones (Yoshida et al. 2005), which influence both arousal and ingestive behaviour (Yamanaka et al. 2003).
One notable mechanism by which BSTvl neurones may influence sodium appetite is by disinhibiting it from the action of central oxytocinergic pathways, which originate exclusively in the hypothalamus (discussed above, in the subsection The disinhibition hypothesis). Neurones in the fusiform subnucleus of the BSTvl, which are predominantly GABAergic, project directly to the paraventricular nucleus of the hypothalamus (PVH), where they primarily target subregions containing excitatory neurones that project directly to visceroceptive and autonomic sites in the brainstem (Cullinan et al. 1993; Dong et al. 2001b). As shown in Fig. 11, the activation of HSD2 neurones during sodium deficiency probably stimulates BSTvl GABAergic neurones, which would then inhibit neurones in the PVH, many of which release oxytocin as a cotransmitter. Since these PVH neurones represent the only known source of oxytocin in the brain (De Vries & Buijs, 1983), they are presumably responsible for inhibition of sodium appetite by central oxytocinergic release (Stricker & Verbalis, 1987, 1996). It remains unknown how or where exactly in the brain sodium appetite is inhibited by oxytocin, although excitatory projections from oxytocinergic neurones in the PVH back to viscerosensory feedback centres in the NTS and parabrachial nucleus are likely to play a role (Sawchenko & Swanson, 1982; Luiten et al. 1985). This model offers one testable hypothesis for future exploration of the brain circuitry responsible for the (dis)inhibitory control of sodium appetite.
Regardless of the mechanism, a functional role for the BST in regulating sodium appetite is supported by demonstrations that large lesions in this region virtually eliminate the spontaneous ingestion of saline and substantially reduce salt intake in response to sodium depletion or mineralocorticoid administration in rats (Reilly et al. 1994; Zardetto-Smith et al. 1994). Lesions of the BST also reduce the magnitude of salt intake stimulated by yohimbine, an α2-adrenergic receptor antagonist (Zardetto-Smith et al. 1994), suggesting a modulatory role for the dense noradrenergic input to the BSTvl from A1 and A2 neurones in the medulla (Terenzi & Ingram, 1995). Owing to the large lesions produced in these initial studies (Reilly et al. 1994; Zardetto-Smith et al. 1994), specific conclusions cannot be drawn regarding the relative contributions of individual subnuclei, such as the fusiform BST. Nonetheless, converging data from neuroanatomical and physiological experiments indicate that the BST plays a key integrative role in the regulation of sodium appetite.
Central nucleus of the amygdala. The central nucleus of the amygdala (CeA) is heavily interconnected with the BST (Dong et al. 2001a) and plays an established role in modulating salt intake (Johnson et al. 1999). Salt intake can be increased or decreased by electrical stimulation or by large lesions involving multiple subregions of the amygdala (Gentil et al. 1968, 1971). Destruction of the entire amygdala reduces spontaneous fluid intake (both water and saline) and attenuates the stimulation of salt intake by sodium depletion and mineralocorticoid administration (Cox et al. 1978).
Like the BSTvl, the input and output connections of the CeA suggest a role in the regulation of sodium appetite. For example, the CeA receives substantial gustatory and viscerosensory input from the NTS and AP, primarily via projections relayed through the pontine PB complex (Norgren, 1976; Bernard et al. 1993; Karimnamazi & Travers, 1998). These relayed inputs probably include information from the aldosterone-sensitive HSD2 neurones because tracer injections into the lateral PB that produced the most retrograde labelling in the HSD2 neurones also produced extensive anterograde labelling of axons terminating in the lateral CeA (Geerling & Loewy, 2006b). In turn, neurones in the medial CeA provide a major descending projection back to the caudal NTS, heavily targeting the HSD2 neurones (Geerling & Loewy, 2006b). Together, these findings indicate that the involvement of the CeA in mineralocorticoid-stimulated salt intake (Sakai et al. 1996, 2000) may depend upon its synaptic connections with the intrinsically mineralocorticoid-sensitive HSD2 neurones (see discussion by Geerling & Loewy, 2006b).
Neurones in the central extended amygdala complex (a term that encompasses the CeA, subnuclei of the lateral BST and additional neurones in between) may powerfully influence ingestive behaviour via output connections to forebrain sites that regulate motor activity, in parallel with their well-known output to autonomic and neuroendocrine nuclei. Both the CeA and the BSTvl provide heavy output connections to pallidal-derived neurones in the substantia innominata (Petrovich & Swanson, 1997; Bourgeais et al. 2001; Dong et al. 2001b), which in turn heavily innervate premotor regions of the midbrain (Swanson et al. 1984). In addition to these indirect pathways, neurones in both the BST and the CeA provide direct axonal input to oral premotor regions of the brainstem reticular formation (Hopkins & Holstege, 1978; Price & Amaral, 1981; Holstege et al. 1985; Shammah-Lagnado et al. 1992; Dong & Swanson, 2003; Yasui et al. 2004). Neurones in these regions, particularly within the dorsomedial medullary reticular formation, are the gatekeepers for controlling ingestive behaviour. They represent the final common pathway for oral motor behaviours, providing the vast majority of direct axonal input to cranial motor neurones in the trigeminal, facial and hypoglossal motor nuclei, which innervate muscles necessary for behaviours such as licking and drinking (Travers & Norgren, 1983; Cunningham & Sawchenko, 2000). The relative importance of these direct versus indirect pathways from forebrain integrative sites to motor regions in the brainstem and spinal cord remains an important subject for future investigation.
Lateral hypothalamic area. Neurones in the LHA are critically involved in homeostatic ingestive behaviours, such as drinking in response to acute dehydration (Clark et al. 1991). This region may also play a necessary role in driving salt intake. The dorsolateral LHA receives axonal input from many sites implicated in sodium appetite, including the BSTvl (Dong et al. 2001b), the pre-LC (the main recipient of efferent projections from the HSD2 neurones in the pons), and even a small direct projection from the HSD2 neurones (Geerling & Loewy, 2006a). As noted above, electrolytic lesions in the dorsolateral LHA eliminate the normal salt-ingestive responses to sodium deficiency and mineralocorticoid administration (Wolf, 1964a; Wolf & Quartermain, 1967). Effective lesions were placed lateral to the fornix and just medial to the cerebral peduncle, a region that envelops many axonal pathways connecting sensory and motor regions of the brainstem to integrative sites in the forebrain, including axons from the HSD2 neurones in the NTS ascending to the BSTvl (Geerling & Loewy, 2006a) and descending axons from the substantia innominata to the midbrain locomotor region (Swanson et al. 1984).
It remains unclear whether the critical elements destroyed in these experiments were neurones located in the dorsolateral LHA, or simply axons that pass through this region in the medial forebrain bundle. To our knowledge, no data are available regarding the effects of axon-sparing excitotoxic lesions in the LHA in relation to sodium appetite. However, an important integrative role for intrinsic LHA neurones was supported by the finding that flanking lesions, which targeted the axon bundles entering and exiting the LHA, did not reproduce the sodium appetite deficit caused by lesions centred directly in the dorsolateral LHA (Wolf et al. 1974). Also, sodium appetite was relatively unaffected by the destruction of nearly all the catecholaminergic axons projecting through this region (by intraventricular injection of 6-hydroxydopamine), which are responsible for many of the other motivational and locomotor deficits caused by electrolytic lesions in the LHA (Stricker & Zigmond, 1974).
Besides the orexin neurones (mentioned above, under the subheading Bed nucleus of the stria terminalis), two other neuronal subpopulations in the LHA may be of particular interest in the regulation of sodium appetite. First, the parasubthalamic nucleus, which borders the cerebral peduncle and subthalamic nucleus at caudal levels of the LHA, provides a unique pattern of axonal projections to a variety of sites implicated in ingestive behaviour and sodium appetite. Its output targets include the BSTvl, CeA, gustatory regions of the rostral NTS and caudal PB, and the HSD2 and A2 viscerosensory subregions of the caudal NTS (Goto & Swanson, 2004). Second, a distinct population of neurones in the perifornical LHA increase their expression of corticotrophin-releasing hormone (CRH) in response to extreme dehydration (Watts, 1992), a response which may be driven by direct input from circumventricular organs in the lamina terminalis and/or neurones within the BSTvl (Kelly & Watts, 1996). When injected into the brain ventricular system, CRH increases saline intake (Tarjan et al. 1992). Interestingly, neurones in this same region of the LHA project extensively to some of the brainstem sites discussed above, including subregions of the parabrachial complex and medial NTS (Kelly & Watts, 1998).
Nucleus accumbens. Although sodium appetite is usually characterized experimentally by a salt-ingestive response (the rapid and voluminous ingestion of concentrated saline) it probably represents a more complex motivational-affective state, more than just the reflexive activation of motor behaviour. The motivational aspects of sodium appetite discussed above (see What is sodium appetite and why is it important?) suggest the involvement of the nucleus accumbens (NAc). This structure occupies the rostral-ventral extent of the striatum in the basal forebrain and plays an important role in behavioural motivation (Mogenson et al. 1980). Neurones in the NAc receive input from limbic cortical regions and from select subcortical sites that include the amygdala, BST and dopaminergic neurones in the midbrain (Brog et al. 1993). Via output projections from the NAc to the substantia nigra and ventral pallidum/substantia innominata, among other sites (Groenewegen & Russchen, 1984), this structure exerts a powerful behavioural influence. The substantia nigra provides dopaminergic input to somatomotor regions of the dorsal striatum (caudate-putamen), and the ventral pallidum/substantia innominata provides feedback to limbic cortical areas via the thalamus, in addition to a major route of descending control to a locomotor region of the midbrain reticular formation (Swanson et al. 1984; Mogenson et al. 1985).
Consistent with the importance of the NAc in other motivated ingestive behaviours, various anatomical and functional findings have implicated this structure in the motivational aspects of sodium appetite. The NAc core receives multisynaptic input from aldosterone-sensitive neurones in the NTS (Shekhtman et al. 2007), and both sodium deprivation and exogenous aldosterone cause a decrease in dopamine reuptake in this region (Roitman et al. 1999). Sodium deprivation also causes subtle changes in NAc dendritic morphology (Roitman et al. 2002).
Expression of the dopamine reuptake transporter and various neuropeptides in the NAc is altered after sodium depletion or high-dose mineralocorticoid treatment (Lucas et al. 2000, 2003a,b). Also, sodium-depleted rats drank less saline after the administration of the δ-opioid receptor antagonist naltrindole into the midbrain ventral tegmental area (VTA), which provides dopaminergic input to the NAc and other limbic structures (Lucas et al. 2007).
Although the basic stimulation of sodium appetite remains unaffected after large lesions in the VTA or destruction of all dopaminergic neurones in the brain (Wolf, 1967; Stricker & Zigmond, 1974), a role for dopamine in the motivational aspects of salt-ingestive behaviour was demonstrated by Roitman et al. (1997). They used a sham-drinking model to eliminate post-ingestive feedback (all ingested fluids were drained through a gastric fistula), allowing the ingestion of much larger volumes of saline over a longer period of time. After sodium deprivation, sham-drinking rats (gastric fistula open) ingested much more saline than similarly deprived control animals (gastric fistula closed). After systemic dopamine D2 receptor blockade, however, sham-drinking rats no longer drank any more than control animals. Importantly, dopaminergic blockade only diminished the increased salt intake of sham-drinking animals, not the amount normally ingested by sodium-deprived control rats (Roitman et al. 1997). In a complementary finding, increased c-Fos activation in the NAc was found only if the amount of saline available to sodium-deprived sham-drinking rats was limited to that ingested by control rats (Roitman et al. 1999).
The brain site(s) that receive this dopamine-dependant motivational signal for salt intake remain unknown. Many regions of the brain receive dopaminergic input, including various limbic cortical areas and the entire striatum. For example, a ventrolateral region of the striatum located just rostral to the CeA (referred to as the fundus of the striatum) is heavily innervated by taste-relay neurones in the gustatory thalamus (VPMpc), and its dopaminergic afferents play a critical role in homeostatic ingestive behaviour (Dunnett & Iversen, 1982; Salamone et al. 1993). The NAc may play a role as well, although microinjections of various D1 and D2 receptor agonists and antagonists specifically into its ventromedial shell region did not affect sodium appetite (Lucas et al. 2007). It is also worth considering the potential involvement of parallel projections that originate from midbrain dopaminergic neurones in the retrorubral fields, which target extensive regions of the extended amygdala complex (Zahm, 2006).
Individually, none of these associational data prove that the NAc itself is necessary for or involved in regulating sodium appetite, but in combination they suggest that it may influence motivational aspects of this behaviour. The specific functional interactions between the NAc and other dopamine-innervated regions of the ventral striatum and extended amygdala with other sodium appetite-associated brain sites implicated in motivation, learning and ingestive behaviour remains an important topic for future investigation.
The findings described here represent initial progress towards defining the brain sites that regulate sodium appetite. This information provides a foundation for the eventual completion of a circuit diagram that will show the neural components responsible for this behavioural state. Despite significant progress, much work remains before we understand the phenotypes, connections and functional interactions of the pertinent subpopulations of neurones within the brain sites listed here, and in as-yet unidentified brain sites that may represent additional components of this circuitry.
Currently, the main problems hindering advances in the study of sodium appetite are uncertainty and controversy regarding the basic signal(s) that stimulate it. Sodium deficiency leads to a variety of autonomic and neuroendocrine changes, including increases in circulating AII and aldosterone. Much of the evidence supporting a critical role for these two hormones in the physiological stimulation of sodium appetite involves contrived pharmacological models (see, for example, Fitts & Thunhorst, 1996; Thunhorst, 1996), or infusion of AII directly into the brain ventricular system, in supraphysiological doses, producing a somewhat different pattern of neuronal activation than physiological increases in circulating AII (McKinley et al. 1995). Many studies support an important role for AII in stimulating sodium appetite (reviewed by Thunhorst, 1996; Weisinger et al. 1996), while other findings suggest that it is not the principal stimulus for salt intake (Coghlan et al. 1981; Weisinger et al. 1997a,b,c; Rowland & Morian, 1999). Similarly, the relative importance of oxytocin as the principal inhibitory control mechanism for sodium appetite remains a point of contention (Fitts et al. 2003; Stricker & Verbalis, 2004). Meanwhile, despite unambiguous evidence to the contrary, some investigators continue to suggest that salt intake is controlled exclusively by changes in the extracellular sodium concentration (Noda, 2006).
It is unfortunate that most existing studies of this appetite do not include more straightforward, non-invasive experiments using sodium deprivation. Chronic dietary sodium deprivation, which mimics the physiological circumstances that stimulate sodium appetite in wild animals in their natural habitats (Blair-West et al. 1968), robustly increases salt ingestion without many of the pronounced autonomic, endocrine and behavioural changes that confound other drug regimens frequently used to produce a more rapid increase in ingestive behaviour (see, for example, Prakash & Norgren, 1991). Also, in our experience, the neural activation produced by this more physiological stimulus (as measured by c-Fos activation) is highly restricted relative to other, more dramatic models used to increase sodium appetite, such as furosemide diuresis, colloid-induced hypovolaemia, or adrenalectomy (differences in c-Fos labelling are particularly apparent in autonomic control regions such as the NTS and ventrolateral medulla; J. C. Geerling & A. D. Loewy, unpublished observations,). Since dietary sodium deprivation is a more selective stimulus for sodium appetite (versus thirst) than other, more invasive manipulations, this experimental model may represent a more useful ‘filter’ for investigators seeking to minimize the number of tangential changes that may obscure subtle mechanisms responsible for sodium appetite versus those primarily responsible for thirst. In any event, additional experiments that resolve the peripheral mechanisms responsible for driving sodium appetite will be important for stimulating future progress in studying the central circuits that control this behaviour.
The finding that phenotypically identifiable neurones are activated specifically during sodium appetite (Geerling et al. 2006a), whether or not they are functionally involved in driving this appetite, may be useful as an adjunct assay for evaluating hypothetical input signals. Clearly, the output signal generated by the HSD2 neurones is progressively and selectively associated with sodium deficiency (Geerling et al. 2006a; Geerling & Loewy, 2007b), suggesting that they receive many of the same input signals that regulate sodium appetite. It remains to be determined, however, whether their output is functionally involved in the stimulation (or disinhibition) of salt-ingestive behaviour. The attenuated salt intake of AP-lesioned animals exhibiting collateral damage to this underlying subregion of the NTS (Contreras & Stetson, 1981; Wang & Edwards, 1997; Curtis et al. 1999) is consistent with the possibility that salt intake is influenced by the activity of HSD2 neurones, while lesions placed in the dorsolateral NTS, just lateral to the HSD2 subregion, did not affect sodium appetite (Schreihofer et al. 1999). Unfortunately, the technical challenges associated with producing a non-lethal lesion of this specific group (without damaging other, immediately adjacent and intermingled neuronal subpopulations in the AP and NTS), or with selectively stimulating the HSD2 neurones (without activating surrounding ingestive-inhibitory regions of the NTS and AP) have thus far prevented a definitive answer to this question.
Much remains to be learned about these unique neurones at the cellular and local-circuit levels, including exactly how aldosterone influences their baseline activity and excitability. Much also remains unknown regarding the extended input–output networks influenced by the HSD2 neurones and those of neurones within sensory circumventricular neurones implicated in sodium appetite. Future functional–neuroanatomical studies should build upon existing data by deciphering the specific contributions of the myriad inputs integrated by subpopulations of neurones in forebrain sites, including the BSTvl and dorsolateral LHA. The ultimate goal of such work is to delineate the entire circuitry responsible for sodium appetite, from primary input sites, through forebrain integrative networks, to motor output channels in the brainstem and spinal cord.
Despite the technical challenges involved, cell-type-specific manipulations targeting the individual components of this circuitry will be critical for achieving this goal. For example, evaluating the potential role of HSD2 neurones in driving sodium appetite may require a transgenic approach to produce neurone-selective lesions (Saito et al. 2001) or the selective expression of light-activated microbial ion channels (Zhang et al. 2007). Alternatively, the identification of a membrane receptor expressed specifically by HSD2 neurones could allow targeted immunotoxin lesions (Wiley & Lappi, 2003). Previous lesion studies targeting the entire BST (Reilly et al. 1994; Zardetto-Smith et al. 1994) must be refined to sort out the specific functional contributions of its myriad subnuclei, particularly the BSTvl, in light of neural tracing data revealing major connectional differences among the various tightly packed subnuclei in this region (for examples, see Dong et al. 2001b; Dong & Swanson, 2003). Similarly, in the dorsolateral pons, lesions targeting specific neuronal subpopulations will be necessary to assign functional roles to the tightly packed, functionally diverse relay nuclei in this critical region of the brainstem. Also, the finding that electrolytic lesions of the dorsolateral LHA can completely eliminate sodium appetite (Wolf, 1964a; Wolf & Quartermain, 1967) should be extended using cell-specific lesion techniques to learn whether this effect involves neurones located in the LHA itself or simply axons traversing this region in the medial forebrain bundle.
In addition to studying these brain circuits in experimental animals, more rigorous data are needed regarding the regulation of sodium appetite in humans, including patients with conditions such as heart failure and salt-sensitive hypertension, especially now that various animal models have indicated that sodium appetite is elevated in such conditions (DiNicolantonio et al. 1982; Francis et al. 2001). Sodium appetite may not be as strong a behavioural drive in humans, but it is well established that we, too, possess an innate behavioural mechanism for increasing salt intake in response to a sustained sodium deficiency (Wilkins & Richter, 1940; Takamata et al. 1994; Leshem & Rudoy, 1997; Cruz et al. 2001; Kochli et al. 2005), even if the relative importance of this appetite in various disease states remains unclear. Finally, the potential for pharmacological intervention in decreasing salt intake in these conditions should be evaluated using drugs that have been shown to reduce salt intake in experimental animals (Sakai et al. 1986; Francis et al. 2001; Sullivan et al. 2004), especially those already known to provide long-term health benefits in human patients, such as mineralocorticoid receptor antagonists (Pitt et al. 1999, 2003).
Since Richter's original work, many advances have increased our understanding of sodium appetite, but much remains unknown. Sodium appetite clearly involves complex brain circuitry, similar to other homeostatic behavioural drives such as hunger and thirst. Beyond any practical application, elucidating the brain circuitry responsible for sodium appetite will both contribute to and benefit from our knowledge of other appetitive and motivated behaviours.
The authors gratefully acknowledge the financial support of the National Institute of Heart, Lung, and Blood of the NIH, grant no. HL-25449 (to A.D.L.) and the American Heart Association, Predoctoral Fellowship, award no. 0510050Z (to J.C.G.).