State-Dependent Plasticity in Vasopressin Neurones: Dehydration-Induced Changes in Activity Patterning


Correspondence to: Colin H. Brown, Department of Physiology, University of Otago, Dunedin 9054, New Zealand (e-mail:


Moderate dehydration impairs concentration and co-ordination, whereas severe dehydration can cause seizures, brain damage or death. To slow the progression of dehydration until body fluids can be replenished by drinking, the increased body fluid osmolality associated with dehydration increases vasopressin (antidiuretic hormone) secretion from the posterior pituitary gland. Increased vasopressin secretion reduces water loss in the urine by promoting water reabsorption in the collecting ducts of the kidney. Vasopressin secretion is largely determined by action potential discharge in vasopressin neurones, and depends on both the rate and pattern of discharge. Vasopressin neurone activity depends on intrinsic and extrinsic mechanisms. We review recent advances in our understanding of the physiological regulation of vasopressin neurone activity patterning and the mechanisms by which this is altered to cope with the increased secretory demands of dehydration.

It is widely accepted that many individuals suffer chronic mild dehydration with little noticeable effect. However, moderate dehydration impairs concentration and co-ordination, and severe dehydration can cause seizures, permanent brain damage or death (1). Hence, it is important to defend the individual, and particularly the brain, from increasingly severe dehydration. One of the key players in the physiological mechanisms that protect the organism from progressive dehydration is the antidiuretic hormone, vasopressin.

Vasopressin maintains body fluid homeostasis principally by activation of V2-receptors to stimulate aquaporin 2 insertion into the apical membrane of kidney collecting duct cells and hence promote water reabsorption from the urine (2). Furthermore, it has recently been reported that vasopressin might act centrally to decrease renal sympathetic nerve activity during osmotic stimulation and thus promote sodium excretion (3). Without vasopressin actions on the kidney, humans excrete up to 30 l of urine each day, as is evident in diabetes insipidus (4). Plasma vasopressin concentrations vary with plasma osmolality around an osmotic set-point of approximately 300 mosmol/kg (5) and, when plasma osmolality increases during dehydration, vasopressin secretion is increased proportionately to reduce water loss in the urine until body water can be replenished by drinking.

Vasopressin has a short half-life in plasma of approximately 2 min in the rat (6) and so antidiuresis requires sustained secretion of vasopressin from nerve terminals in the posterior pituitary gland to maintain plasma concentrations of the hormone. Vasopressin secretion is largely determined by action potential (spike) discharge initiated at the cell bodies of magnocellular neurosecretory vasopressin neurones in the hypothalamic supraoptic and paraventricular nuclei (7, 8). Nevertheless, the release of vasopressin can be modulated at the level of the nerve terminals in the posterior pituitary gland in a state-dependent manner; it has recently been demonstrated that dehydration up-regulates salusin-β expression in vasopressin neurones, which stimulates vasopressin release from nerve endings (9), and that L-type Ca2+channel expression in posterior pituitary gland pituicytes is increased in dehydration (10) both of which might be involved in increasing the efficiency of vasopressin release from the nerve terminals under stimulated conditions.

Both the firing rate and firing pattern of vasopressin neurones are important determinants of the overall secretion of vasopressin and these are regulated by several extrinsic and intrinsic mechanisms, each of which are modulated by plasma osmolality. We review the mechanisms by which vasopressin neurones respond to dehydration to limit avoidable water loss in the urine.

Firing patterns of vasopressin neurones

Under basal conditions, vasopressin neurones display a range of activity patterns (Fig. 1a) that maintain plasma vasopressin concentrations at approximately 1 pg/ml, which is sufficient to cause the reabsorption of approximately 30 l of water from the urine each day: some do not fire spikes (silent), some display irregular activity, some are continuously active and some display rhythmic ‘phasic’ activity, comprised of alternating bursts of activity, during which there is clear spike frequency adaptation (i.e. a reduction in firing rate with time), and silent periods that each last tens of seconds (7, 8).

Figure 1.

 Dehydration increases the firing rate of continuously-active and phasic, but not irregular, vasopressin cells. (a) Ratemeter records (in 1-s bins) of the firing rates of supraoptic nucleus vasopressin neurones in nondehydrated (euhydrated) rats (left) and 24-h dehydrated rats (right), displaying spontaneous continuous activity (top), phasic activity (middle) and irregular activity (bottom). (b) The mean  ± SEM firing rates of continuous (top), phasic (middle) and irregular (bottom) vasopressin neurones in euhydrated rats and dehydrated rats (*P < 0.05, unpaired t-test). Reproduced with permission from Scott et al. (17).

Why vasopressin neurones should express such widely varying activity patterns under the same physiological conditions is not immediately obvious, until it is considered that the output of the vasopressin neurone population determines the plasma concentration of vasopressin, and not the activity of any particular individual within the population (11).

Intuitively, silent and (slow) irregular vasopressin neurones will release little hormone into the circulation, whereas (fast) continuous vasopressin neurones would be expected to make the greatest contribution to the circulating hormone concentrations. However, the evoked secretion of vasopressin rapidly reduces (fatigues) over a few tens of seconds at high firing rates. Hence, neurones that are continuously active over long periods might not contribute a large proportion of the circulating hormone. Remarkably, only a few tens of seconds of silence are needed for full recovery from this fatigue (12) and so it is the neurones displaying a phasic firing pattern that likely contribute most to circulating vasopressin concentration by making optimal use of their terminal properties.

Another function of phasic activity might be to allow the actively-secreting population to act as a low-pass filter and thus prevent transient perturbations from causing large fluctuations in circulating vasopressin (11); brief stimulation of afferent inputs to phasic vasopressin neurones can trigger bursts when applied during a silent period, or terminate bursts when applied during an active period (13). Hence, the population response of vasopressin neurones to transient stimuli might comprise no overall change in secretion; only sustained changes, such as occur in chronic dehydration, will result in a shift in the average activity of the population as a whole.

Firing patterns of vasopressin neurones during dehydration

It has long been accepted that acute osmotic stimulation increases the spike discharge rate of all vasopressin neurones (14). However, such stimuli are extreme and can raise plasma osmolality to over 330 mosmol/kg in only 30–60 min. It has recently been demonstrated that more modest (but nevertheless strong) acute osmotic stimulation induces variable changes in the activity of phasic vasopressin neurones, with both increases and decreases in overall spike discharge rates of individual neurones being observed (15). Hence, the pervading opinion that osmotic stimulation is always excitatory to every vasopressin neurone might not accurately represent the response to more ‘physiological’ situations, such as a slow progressive dehydration.

It has been reported that all vasopressin neurones display phasic activity after 24 h of water deprivation (16), suggesting that chronic dehydration drives the entire population of vasopressin neurones to adopt phasic activity. However, these experiments were completed in lactating rats, and the plasma osmolality reached approximately 340 mosmol/kg over 24 h (16), which is an extreme osmotic stimulus. We recently reported no change in the proportions of active vasopressin neurones displaying irregular, phasic or continuous spike discharge between nondehydrated virgin female rats and virgin female rats deprived of access to water for 24 h (with a plasma osmolality of approximately 310 mosmol/kg) (17). Plasma osmolality after only 6 h of water deprivation in lactating rats more closely matched the levels reached after 24 h of water deprivation in our study (16, 17). However, at 6 h of dehydration, again, almost all vasopressin neurones in lactating rats exhibited phasic activity (16). A confounding factor in comparing these two studies is the surgical approach used; for lactating rats, vasopressin neurones were recorded using an electrode placed through the dorsal aspect of the skull, whereas, in our experiments, a more invasive technique was used (transpharyngreal surgery) to expose the ventral surface of the brain for placement of a microdialysis probe to pharmacologically manipulate the local environment of the vasopressin neurones. Hence, the additional stress of more extensive surgery in our experiments could account for the differences in basal activity patterning between the two studies. Nevertheless, our observations are similar to previous reports of the effects of more modest chronic osmotic stimulation by water deprivation in male rats (18) or 2% saline drinking in male rats (19) in which the dorsal approach was also used. Hence, long-term changes in physiological state (e.g. pregnancy and lactation) might alter the ‘basal’ activity of vasopressin neurones. Basal vasopressin neurone activity maintains a plasma concentration of approximately 1 pg/ml in conscious rats and this increases to approximately 10 pg/ml during chronic osmotic stimulation (20), whereas, in response to extreme acute osmotic stimuli, plasma concentrations can approach 100 pg/ml (21). Urethane anaesthesia itself (as used in the electrophysiological experiments described above) increases plasma vasopressin concentrations to approximately 5 pg/ml, or higher when combined with surgery (22), and so it is likely not appropriate for all vasopressin neurones to adopt an activity pattern that results in excessive secretion of vasopressin to concentrations that might also cause vasoconstriction.

In our experiments, the response to modest chronic osmotic stimulation was a doubling of the overall spike discharge rate of phasic vasopressin neurones (Fig. 1b) (17). The dehydration-induced increase in spike discharge of phasic neurones was achieved by an increase in intraburst firing rate, without a change in burst duration or interburst interval. In combination with frequency-facilitation of secretion from the posterior pituitary gland, such increases in intraburst firing rate would be expected to increase circulating vasopressin to around the concentrations evident during modest chronic osmotic stimulation without the need to recruit a larger proportion of vasopressin neurones into phasic activity.

Hence, ‘physiological’ osmotic variations of less than approximately 10 mosmol/kg from the set-point (5) are unlikely to force every vasopressin neurone to adopt phasic activity; rather, such extreme responses of the vasopressin neurone population are probably reserved for the most extreme life-threatening cases of dehydration. It appears likely that modest chronic osmotic stimuli such as water deprivation initially increases circulating vasopressin concentrations by first increasing the spike discharge rate of the neurones that display phasic activity; only as dehydration progresses, and becomes more severe, will a larger proportion of vasopressin neurones be recruited to adopt phasic activity. Although the proportion of vasopressin neurones that display phasic activity is static close to the osmotic set-point, it is probable that the individual neurones that display phasic activity cycle to spread the secretory load across population of vasopressin neurones (11)

Spike discharge in vasopressin neurones results from the interaction between intrinsic and extrinsic mechanisms (23). Intrinsic mechanisms include osmotically-regulated ion channels and post-spike potentials, whereas extrinsic mechanisms include afferent synaptic inputs and glial modulation of the local extracellular environment. The interplay between these intrinsic and extrinsic factors modulates the probability that any individual vasopressin neurone will fire a spike at any given time and so determines the rate and pattern of spike discharge.

Intrinsic osmoreception in vasopressin neurones

As for all cells, vasopressin neurones will shrink when the osmolality of the extracellular fluid increases and water leaves the cells, or swell when extracellular osmolality decreases. Although most cells rapidly correct osmotically-induced volume changes by altering the flux of osmolytes such as taurine, vasopressin neurones maintain their osmotically-induced volume changes for at least 1 h (24). This sustained change in cell volume is important for osmoreception in vasopressin neurones because these neurones express non-inactivating, stretch-inactivated nonselective cation channels (25) that are mechanically-gated via the actin cytoskeleton (26). Hence, cell shrinkage in hyperosmotic conditions will cause a sustained membrane depolarisation, increasing the probability of spike discharge by bringing baseline membrane potential closer to spike threshold; these responses are not evident in trpv1 knockout mice and so it appears likely that the channel responsible is the transient receptor potential vanilloid-1 ion channel (27). Regardless of their molecular identity, these channels are partially activated at the osmotic set-point, and swelling in hypo-osmotic conditions will cause a sustained hyperpolarisation and thus reduce the probability of spike discharge. However, it remains to be determined whether these responses are sustained over a long enough time-course to contribute to the increased spike discharge of vasopressin neurones in response to chronic hyperosmotic challenge such as dehydration.

Nonsynaptic post-spike potentials in vasopressin neurones

As alluded to above, phasic spike discharge is important for vasopressin neurone function. Phasic activity patterning is shaped by the nonsynaptic post-spike potentials expressed by these cells (Table 1), principally afterhyperpolarisations (AHPs) and afterdepolarisations (ADPs) (Fig. 2) that translate random synaptic input into a phasic output pattern via mechanisms that are reviewed at length elsewhere (7, 8).

Table 1.   Nonsynaptic Post-Spike Potentials Expressed by Vasopressin Neurones.
Post-spike potentialChannelProposed functionReferences
Fast after hyperpolarisation (fAHP)Large conductance potassium (BK) channelSpike repolarisation(98)
Medium after hyperpolarisation (mAHP)Small conductance potassium (SK) channelSpike frequency adaptation(35, 99)
Slow after hyperpolarisation (sAHP)Intermediate conductance potassium (IK) channelBurst termination(36, 37)
Fast after depolarisation (fADP)Nonspecific cation (CAN) channelBurst initiation(32)
After depolarisation (ADP)Nonspecific cation (CAN) channelPlateau potential generation and burst termination(31, 100)
Figure 2.

 Post-spike potentials and phasic activity in vasopressin neurones. (a) After each spike in vasopressin neurones, there is a prominent post-spike fast afterhyperpolarisation (fAHP) that is responsible for the markedly unexcitable state of magnocellular neurosecretory cells for approximately 10 ms immediately following each spike. (b) Post-train medium AHP (mAHP) and afterdepolarisation (ADP) that follow a spike train of five spikes (arrowhead, spikes truncated) that contribute to phasic activity. (c) Summated post-train slow component of the AHP (sAHP) that follows a 50-Hz (1 s) spike train (arrowhead, spikes truncated), showing the sAHP more prominently; the sAHP becomes apparent after spike trains of 20 Hz or more and contributes to burst termination. (d, e) The currents underlying the mAHP and ADP, blockade of the mAHP with apamin reveals an ADP of earlier onset (d) and blockade of the ADP with CsCl reveals a deeper and longer lasting mAHP (e). (f) Single spikes (truncated) activate AHPs and ADPs but the amplitudes of the mAHP and ADP are relatively small. Nevertheless, the long time-course of these post-spike potentials allows temporal summation when spikes occur close enough together. Summation of ADPs contributes to burst initiation and the summed mAHP contributes to spike frequency adaptation during phasic bursts. The balance between the currents that underlie the mAHP and ADP (as well as other currents) sets the level of the plateau potential that sustains activity during bursts by bringing the membrane potential closer to the threshold for spike activation.

Subsequent to a period of approximately 15 s after the end of the previous burst, phasic burst onset essentially comprises a random process (28) of the summation of synaptic potentials to reach the threshold for spike initiation. Each spike is followed by a post-spike ADP that can persist for up to 3 s in vasopressin neurones (29–31) and, when consecutive spikes occur close enough together, a persistent plateau potential is generated by summation of their ADPs. Recently, a fast ADP (fADP) has been characterised that has been proposed to ‘bootstrap’ the classical ADP and thus contribute to burst initiation (32). Once established, the plateau potential promotes further spike firing by bringing the membrane potential closer to threshold for spike initiation (28). The amplitude of the plateau potential is moderated by activity-dependent activation of other post-spike potentials: a fast AHP (fAHP) paradoxically increases firing rate during bursts via activation of a hyperpolarisation-activated inward current (IH) (33), whereas a medium AHP (mAHP) generates the spike frequency adaptation evident over the course of bursts (34, 35). After approximately 20 s of firing, burst termination is (similar to burst onset) essentially a random process. Burst termination involves a progressive decline in plateau potential amplitude via activity-dependent inhibition of ADPs (31) and activation of a slow AHP (36, 37). The decline in plateau potential amplitude does not itself terminate bursts but reduces the probability that on-going (random) synaptic inputs will reach the threshold for spike initiation (28). After each burst, there is a marked and prolonged hyperpolarisation that prevents the onset of the next burst for approximately 15 s; only after sufficient relaxation of this post-burst hyperpolarisation can summation of synaptic inputs (and their associated ADPs) establish the next cycle of burst firing (28).

Although the involvement of post-spike potentials in the generation of phasic firing has been well characterised, the potential contribution of changes in post-spike potentials to the increased activity evident during dehydration has not been addressed, perhaps as a result of the inherent difficulty in controlling for likely acute changes in osmolality when setting up the in vitro preparations from which these potentials can be measured.

However, the influence of post-spike potentials, particularly the mAHP and ADPs, on spontaneous firing patterns can be revealed in vivo by calculation of the probability of spike firing (hazard) from the inter-spike interval histogram of individual vasopressin neurones; the hazard function gives the inferred probability of a neurone firing a subsequent spike in any interval after a spike, given that another spike has not occurred earlier (38–40) and is a measure of the post-spike excitability of the recorded neurone. A constant hazard is produced by a random firing pattern that is deduced to reflect the baseline membrane potential and the on-going synaptic input activity, whereas divergence below and then above a constant hazard after each spike reveals periods of prolonged decreased and then increased post-spike excitability that are deduced to reflect the hyperpolarising influence of the mAHP, followed by the depolarising influence of the ADP (40) (Fig. 3). However, it should be noted that the time-course of the changes in post-spike excitability (40) is more similar to that of the recently-identified fADP (32) than the classical ADP and so might reflect changes in the fADP.

Figure 3.

 Contribution of intrinsic membrane properties to phasic activity in vasopressin neurones. The schematic on the left shows an individual spike from a vasopressin neurone, with the associated changes in membrane potential caused by afterhyperpolarisations (AHPs) (principally the medium AHP) and afterdepolarisations (ADPs) (the classical ADP and the fast AHP) after each spike (not to scale). Normally, vasopressin neurones exhibit a prominent post-spike AHPs and post-spike ADPs (solid line). After each spike, AHPs initially hyperpolarise the neurone, making it less likely to reach spike threshold (less excitable) and then the ADPs increase excitability by bringing the membrane potential closer to spike threshold, making it more likely that on-going (random) synaptic input will reach spike threshold to trigger a further spike. The influence of these changes in post-spike membrane potential on activity patterning can be revealed by calculating the probability of spike firing from the inter-spike interval histogram in vivo (a hazard function), shown in the graph on the right for 13 phasic neurones. Immediately after each spike (at inter-spike interval = 0), there is a reduced probability of a further spike firing (post-spike refractoriness) that is inferred to result from the AHPs. The post-spike refractoriness is followed by an increased probability of a further spike firing (post-spike hyperexcitability) that is inferred to reflect the influence of the ADPs. If a spike does not fire, the probability of spike firing returns a steady-state hazard that is inferred to reflect the baseline membrane potential and the on-going synaptic input activity. Modified with permission from Brown et al. (51).

Analyses of hazard functions from phasic and continuous vasopressin neurones suggest that both the mAHP and ADP decay more rapidly in continuously active neurones than in phasic neurones (40). Our recent (unpublished) work shows that the shape of the hazard function derived from irregular vasopressin neurones is similar to that from continuous vasopressin neurones, opening the possibility that changes in mAHPs and ADPs have the potential to contribute to plasticity in activity patterning, with longer-lasting mAHPs and ADPs being associated with the expression of phasic activity by vasopressin neurones. The mechanisms by which the mAHP and ADP might be modulated in such a way are currently unknown. However, somato-dendritic release of the κ-opioid peptide, dynorphin, inhibits ADPs (31) and endogenous adenosine (probably generated from breakdown of ATP exocytosed with vasopressin from the soma and dendrites) enhances the mAHP (41). Hence, autocrine dynorphin and adenosine feedback inhibition are mechanisms that might contribute to plasticity in vasopressin neurone activity (see below).

Whether post-spike potentials are modulated by dehydration is not known but 24 h of water deprivation up-regulates the L-type Ca2+ current in supraoptic nucleus neurones (42), which would be expected to enhance ADPs and mAHPs, both of which are dependent on extracellular Ca2+ (35, 43). In vivo, an enhanced ADP would be expected to increase the spike discharge rate and an enhanced mAHP would be expected to strengthen spike frequency adaptation, both of which are evident in phasic neurones from dehydrated rats (17).

However, our hazard function analyses of vasopressin neurone activity in nondehydrated and dehydrated rats indicate the dehydration-induced increase in spike discharge was not associated with gross alterations in the post-spike excitability of the neurones that could be attributed to altered mAHP or ADP dynamics. Although this does not preclude a contribution of changes in the mAHP and/or ADP to the vasopressin neurone response to dehydration, the likely primary drivers of increased firing rate of vasopressin neurones during dehydration are increased synaptic drive (44) and/or increased activation of stretch-inactivated cation channels (24).

Autocrine modulation of vasopressin neurone activity during dehydration

Vasopressin neurones also contain large amounts of neuropeptide within their soma and dendrites, from which release occurs by exocytosis (45, 46). This somato-dendritic neuropeptide release is activity-dependent (39, 47), although it is unlikely that there is tight coupling between individual spikes and individual exocytotic events in the somata/dendrites (48). Nevertheless, somato-dendritic exocytosis appears to be critical for phasic spike discharge in vasopressin neurones (7, 8). The neurosecretory vesicles of vasopressin neurones express V1a and V1b receptors (49), which will presumably be inserted into the cell membrane during exocytosis. Hence, activity-dependent vasopressin release will be coupled with activity-dependent insertion of surface receptors, providing a mechanism for autocrine activity-dependent feedback by somato-dendritically released vasopressin.

However, vasopressin is present in the extracellular space of the supraoptic nucleus in measurable quantities under basal conditions (50) and so it is likely that somato-dendritic vasopressin release affects not only the cell of origin, but also neighbouring neurones (51). Indeed, V1a- receptor antagonists increase the firing rate of phasic vasopressin neurones throughout bursts (39), indicating that endogenous vasopressin is present before burst onset and could be acting as a paracrine modulator within the supraoptic nucleus.

Although vasopressin inhibits vasopressin neurones displaying robust phasic spike discharge via V1a- receptors, exogenous vasopressin administration can excite irregular or weakly phasic vasopressin neurones (52, 53). Hence, it has been proposed that somato-dendritic vasopressin serves as a ‘population feedback signal’ that distributes the secretory load across the population of vasopressin neurones (54).

Vasopressin reduces the amplitude of the excitatory post-synaptic current (55) and increases the frequency of the inhibitory post-synaptic current (56) in supraoptic nucleus neurones and these actions probably underpin autocrine/paracrine inhibition by vasopressin. Although the mechanism of vasopressin excitation of vasopressin neurones is not known, it might occur via activation of phospholipase C (PLC) (57) because vasopressin stimulation of adrenocorticotrophic hormone secretion from anterior pituitary corticotrophs is mediated by PLC activation (58).

Dehydration increases vasopressin expression in the supraoptic nucleus (59) as well as the concentration of vasopressin within the supraoptic nucleus (60). This increased intra-supraoptic nucleus release might prevent over-excitation of vasopressin neurones during dehydration and hence allow the neurones to respond to changes in plasma osmolality in a graded manner over a wide dynamic range.

Autocrine κ-opioid receptor modulation of phasic activity in dehydration

Vasopressin neurones also synthesise and secrete several other factors that modulate spike discharge in vasopressin neurones, including the neuropeptides, apelin, dynorphin and galanin, as well as other factors such as adenosine and nitric oxide. The mechanisms by which each of these factors modulates vasopressin neurone spike discharge under basal conditions has recently been reviewed at length elsewhere (46).

During chronic hyperosmotic stimulation, prodynorphin mRNA expression is up-regulated in vasopressin neurones (17, 61), suggesting that autocrine modulation by dynorphin might be important in shaping the spike discharge patterns of vasopressin neurones during dehydration.

Dynorphin activates μ-and κ-opioid receptors. However, although κ-opioid receptor agonists inhibit vasopressin neurone activity in hypothalamic slices (62, 63), the potent and selective μ-opioid agonist, endomorphin-1, does not affect the spike discharge of vasopressin neurones in hypothalamic slices at doses that potently inhibit neighbouring oxytocin neurones (63). Hence, dynorphin inhibition of vasopressin neurones is likely mediated exclusively via κ-opioid receptors, which are expressed by vasopressin neurones (64).

Endogenous dynorphin reduces phasic burst duration in vivo (65) and in vitro (31). This inhibition likely involves two complementary mechanisms: activity-dependent inhibition of ADPs (31) to progressively decrease plateau potential amplitude during phasic bursts (28), as well as retrograde inhibition of excitatory synaptic transmission (66). By contrast to the endogenous V1a-receptor inhibition of phasic vasopressin neurones, dynorphin inhibition of phasic neurones is absent at the onset of each burst but emerges as bursts progress (39), suggesting that dynorphin is cleared from the immediate vicinity of each phasic neurone between bursts. As the bursts develop, activity-dependent dynorphin release from the soma and dendrites causes progressive κ-opioid feedback inhibition of plateau potentials to reduce the probability of (dynorphin inhibited) excitatory post-synaptic synaptic potentials reaching spike threshold, eventually terminating bursts.

Because the activity of adjacent phasic vasopressin neurones is asynchronous (7, 67), it is likely that dynorphin acts as an autocrine, rather than paracrine, inhibitor of vasopressin neurones, at least under basal conditions. Extracellular peptidases limit the effects of endogenous dynorphin (31) and so might guard against dynorphin-induced synchronisation of phasic bursts between vasopressin neurones to avoid co-ordinated pulsatile secretion of vasopressin into the circulation.

During dehydration, phasic neurones display an increased intra-burst firing rate (16–19). Intuitively, this increased firing rate would be expected to trigger increased somato-dendritic dynorphin release and so result in shorter bursts. However, in response to intense dehydration, phasic bursts are prolonged (16) and, in our hands, the duration of phasic bursts during moderate dehydration was the same as that recorded form nondehydrated rats (17). During moderate dehydration, autocrine dynorphin feedback inhibition of phasic firing is increased in proportion to firing rate because the κ-opioid receptor antagonist, nor-BNI, induces a marked increase in burst duration in vasopressin neurones in dehydrated rats similar to that in nondehydrated rats (17).

Autocrine κ-opioid receptor modulation of vasopressin neurones activity patterning in dehydration

Although the mechanisms that generate phasic activity are well-characterised (8), the mechanisms that induce continuous and irregular firing patterns have not been extensively investigated; our recent work suggests that dynorphin autoregulation might be involved in determining the activity pattern displayed by individual vasopressin neurones.

Continuously active vasopressin neurones are strongly inhibited by κ-opioid agonist administration but are not affected by κ-opioid receptor antagonists (17, 65) (Fig. 4), indicating that continuously active vasopressin neurones express functional κ-opioid receptors that are not exposed to endogenous dynorphin under basal conditions. Hence, continuously active vasopressin neurones might have ‘escaped’ autocrine feedback inhibition to enter into a nonterminating burst (i.e. continuous activity). Indeed, as alluded to above, continuously active vasopressin neurones express a shorter post-spike inhibition in vivo (40), which might result from a more prominent ADP that is free from endogenous dynorphin inhibition.

Figure 4.

 Phasic and irregular, but not continuously-active, vasopressin neurones are excited by κ-opioid receptor antagonism in nondehydrated rats and dehydrated rats. (a) Ratemeter records (in 10-s bins) of the activity of continuously-active vasopressin neurones in a nondehydrated rat (left) and a dehydrated rat (right). Injection of the κ-opioid receptor agonist, U50,488H (U50; 1 mg/kg, i.v.), profoundly inhibited the firing rate of both neurones and this inhibition was partially blocked (left) by intra-supraoptic nucleus microdialysis administration of the κ-opioid receptor antagonist nor-BNI over 60 min (BNI; 200 μg/ml). However, BNI alone did not alter the firing rate of these neurones, indicating that continuously active neurones express functional κ-opioid receptors that are not activated by endogenous opioid peptides. (b) Ratemeter recordings (in 1-s bins) of the firing rates of phasic vasopressin neurones in a nondehydrated rat (left) and a dehydrated rat (right) before and during microdialysis administration of BNI (200 μg/ml). The inserts show a 2-min period of firing before and during nor-BNI administration in each example; BNI increased burst duration and intraburst firing rate to a similar degree in nondehydrated and dehydrated rats. (c) Ratemeter recordings (in 1-s bins) of the firing rates of irregular vasopressin neurones in a nondehydrated rat (left) and a dehydrated rat (right) before and during microdialysis administration of BNI (200 μg/ml). Again, the inserts show a 2-min period of firing before and during nor-BNI administration; BNI increased firing rate to a similar degree in irregular neurones from nondehydrated and dehydrated rats. (a) and (b) are modified with permission Scott et al. (17).

We have recently demonstrated that, similar to phasic neurones, irregular vasopressin neurones are inhibited by endogenous dynorphin because the administration of a κ-opioid receptor antagonist into the supraoptic nucleus increases the activity of vasopressin neurones displaying irregular activity (Fig. 4) (17). Hence, neurones displaying different activity patterns are differentially regulated by endogenous dynorphin and so autocrine dynorphin feedback inhibition might partition vasopressin neurones into various activity patterns. These mechanisms appear to persist during dehydration because, similar to its effects in nondehydrated rats, κ-opioid receptor antagonist administration induces a similar increase in the firing rate of irregular vasopressin neurones in dehydrated rats but does not affect the firing rate of continuously active neurones (17).

Autocrine modulation of spike frequency adaptation

Phasic activity is characterised by prominent spike frequency adaptation during bursts (typically from approximately 15–20 spikes/s at burst onset to approximately 6–8 spikes/s after 5–10 s of firing) (7, 8). Spike frequency adaptation in vasopressin neurones is induced by activation of the mAHP (35) and we have recently shown that activity-dependent adenosine A1 receptor activation increases the mAHP amplitude to enhance spike frequency adaptation and hence reduce burst duration (41, 68) (Fig. 5). Adenosine also reduces voltage-dependent calcium currents (69), inhibits glutamate (and GABA) release via pre-synaptic receptors and reduces spike duration and hyperpolarisation through post-synaptic receptors (70, 71) in vasopressin neurones; each of these effects might also contribute to the adenosine-induced increase in burst duration.

Figure 5.

 A1 receptor antagonism reduces medium afterhyperpolarisation (mAHP) amplitude to increase activity in phasic neurones. (a) Current-clamp recording of the membrane potential of a supraoptic nucleus neurone during spontaneous phasic activity, before (Pre-CPT, left) and during superfusion of the A1 adenosine receptor antagonist, 8-cyclopentyl-1,3-dimethylxanthine (10 μm CPT; right). Note that CPT increased the duration of spontaneous phasic bursts, typical of all neurones tested. (b) mAHPs and afterdepolarisations (ADPs) (averages of five traces) that each follow a five-spike train (arrowheads, spikes truncated) evoked by an 80-ms depolarising pulse (+200 pA) before (Pre-CPT, left) and during superfusion of 10 μm CPT (right). CPT decreases mAHP amplitude to expose a more prominent ADP, indicating that endogenous adenosine enhances the mAHP to decrease excitability and thus contribute to spike frequency adaptation and burst termination.

Although endogenous adenosine enhancement of spike frequency adaptation is activity-dependent, the source of the adenosine is not known. Adenosine is likely generated by rapid breakdown of ATP in the extracellular space (72). This ATP might be released by neighbouring glia because the amplitude of miniature excitatory post-synaptic currents is increased by glial ATP release in the paraventricular nucleus (73). However, vasopressin neurosecretory granules also contain ATP (74) and so another potential source is from somato-dendritic exocytosis from the vasopressin neurones themselves.

Similar to the effects of endogenous dynorphin, endogenous adenosine does not alter the firing rate of continuously active vasopressin neurones (68), which do not display spike frequency adaptation. Hence, it appears that continuously active vasopressin neurones are not exposed to endogenous adenosine or dynorphin. Indeed, the firing rate of continuously active vasopressin neurones is not affected by vasopressin receptor antagonists (53), suggesting that continuously active neurones might not be actively secreting from their soma and dendrites.

The effects of dehydration on endogenous adenosine feedback are not known. Nevertheless, given that somato-dendritic vasopressin and dynorphin release is probably increased (17, 75), it is likely that ATP release is also increased. Indeed, the magnitude of spike frequency adaptation in phasic vasopressin neurones is increased during dehydration and so the contribution of endogenous adenosine to spike frequency adaptation might also be increased.

Glial regulation of vasopressin neurone activity

Vasopressin neurones are ensheathed by astrocytes that control the local environment of the neurones; this is not simply a support role because these glia are actively involved in shaping the activity of vasopressin neurones through several mechanisms, each of which might contribute to the vasopressin neurone response to dehydration. During dehydration, glia withdraw their processes from the extracellular environment that separates vasopressin neurones, leading to increased apposition of neuronal perikarya membrane (76) and dendritic bundling (77).

Dehydration increases the release of both glutamate and GABA within the supraoptic nucleus, as revealed by an increase in the frequency of spontaneous excitatory and inhibitory post-synaptic currents in supraoptic nucleus neurones recorded from hypothalamic slices prepared from dehydrated rats (44). Although increased activity of excitatory glutamate inputs would be expected during dehydration, increased activity of inhibitory GABA inputs appears paradoxical at a time when increased output is required; it appears likely that increased GABA is required to moderate the dehydration-induced excitation because increased GABA inhibition of vasopressin neurones during acute osmotic stimulation reduces the gain of the excitatory response (14).

Glia express transporters for glutamate (78) and GABA (79) that are important for modulating synaptic efficacy. However, the role of these glial transporters is not limited to the modulation of synaptic efficacy; when glia retract during dehydration, spill-over from these synapses is increased to generate persistent changes in membrane potential through the activation of extra-synaptic receptors (79) that might be important for modulating the increased activity of vasopressin neurones during dehydration.

In addition to the modulation of neuronal activity by modifying extracellular neuronal transmitter concentrations, glia also regulate neuronal activity via secretion of substances that directly affect vasopressin neurones; as stated above, glial ATP release increases the amplitude of miniature excitatory post-synaptic currents in vasopressin neurones (73).

One of the best characterised mechanisms by which glia regulate the activity of vasopressin neurones is via the osmotically-regulated release of the osmolyte, taurine, as reviewed elsewhere (80). Taurine inhibits vasopressin neurones via extra-synaptic glycine receptors and glial taurine release is reduced in hyperosmotic conditions to disinhibit the neurones. However, dehydration produces similar increases in urine osmolality in mice that lack the Na+-coupled taurine transporter (taut−/− mice) as in wild-type littermates (81) but, upon rehydration, taut−/− mice continue to generate hyperosmotic urine, whereas that of wild-type littermates rapidly returns to normal osmolality. Hence, decreased glial taurine release might not substantively increase the activity of vasopressin neurones during chronic dehydration but increased taurine release could contribute to the reduction of vasopressin neurone activity when plasma osmolality falls.

Another mechanism by which glia appear to modulate activity of vasopressin neurones is via regulation of ambient glutamate concentrations in the extracellular environment. Vasopressin neurones express glutamate and GABA transporters (78, 79) and, upon glial retraction during dehydration, the ambient levels of glutamate and GABA increase (78, 79) to more strongly activate metabotrophic glutamate receptors and GABAA receptors, each of which inhibit vasopressin neurones. Hence, the modulation of ambient transmitter concentrations might serve to limit the excitation of the neurones.

Osmosensory afferent inputs to vasopressin neurones

Although many mechanisms (both excitatory and inhibitory), including those described above, modulate the gain of vasopressin neurone excitation during dehydration, the driving force that underpins the excitation is likely an increase in synaptic input activity from osmosensory afferents. Removal of even a portion of the excitatory synaptic drive to supraoptic nucleus neurones in vivo will cause them to fall silent (82–85), even during intense acute osmotic stimulation (86).

Vasopressin neurones receive afferent inputs from various brain regions that principally convey cardiovascular and osmotic information. The most important osmotic inputs are from the osmoreceptor complex in the forebrain that is comprised of the organum vasculosum of the lamina terminalis, the median preoptic nucleus and the subfornical organ, each of which project to vasopressin neurones (14, 87–92). The organum vasculosum of the lamina terminalis and the subfornical organ lie outside the blood–brain barrier and at least some of these neurones are osmosenstive (93).

Osmoreceptor complex inputs to vasopressin neurones are mediated by a mixed population of excitatory and inhibitory transmitters, including glutamate (88) and GABA (14). Nevertheless, lesion of these inputs reduces osmotic stimulation of vasopressin release (94) indicating that the function of these inputs is to mediate excitation; the inhibitory (GABA) afferents reduce the gain of the excitation to allow the vasopressin neurones to respond dynamically across the physiological range of plasma osmolalities (14). Although osmosensitive inputs have largely been mapped using acute osmotic stimulation, studies of the activation of these inputs during chronic dehydration support their involvement in chronic, as well as acute, osmotic stimulation of vasopressin neurones (92).

Vasopressin neurones also receive osmotic information from peripheral osmoreceptors that is relayed via noradrenergic projections from the brainstem; lesion of noradrenergic neurones also reduces vasopressin release in response to osmotic stimuli (95, 96), indicating that these areas also contribute to osmotic stimulation of vasopressin neurones. Whether noradrenergic projections are direct is unclear and these effects might be mediated via other brain areas, including the parabrachial nucleus (97).


During dehydration, various intrinsic and extrinsic mechanisms modulate the activity of each vasopressin neurone to allow the population of neurones to respond appropriately through secretion of vasopressin to limit further water loss in the urine. Accordingly, vasopressin neurones display a range of activity patterns even during dehydration, including irregular, phasic and continuous patterns. Because of fatigue of secretion at the posterior pituitary gland, it is likely that phasic neurones contribute a large proportion of the vasopressin that can be measured in the blood. Hence, vasopressin concentrations can be increased by two main mechanisms: by increasing the proportion of neurones that display phasic activity and by increasing the firing rate of neurones during phasic bursts. During moderate dehydration, the first response appears to be an increase in the intraburst firing rate (but not the duration of bursts) of neurones displaying phasic activity (17) and an increase in the proportion of neurones that display phasic activity only occurs when dehydration increases in severity (16). Shaping the response to dehydration in this way might prevent over-excitation of individual vasopressin neurones and thus reduce the risk of excitotoxic cell death.

Although the proportion of vasopressin neurones that display phasic activity during moderate dehydration is static, we have previously argued that the sub-population of neurones that display phasic activity at any one time is likely dynamic (11). Various mechanisms (some of which we have highlighted above) contribute to the response of individual vasopressin neurones to dehydration and plasticity in these mechanisms might permit individual neurones to transition between activity patterns and contribute to the appropriate population response. It remains to be determined how the thousands of vasopressin neurones, distributed across two bilateral hypothalamic nuclei, determine which activity pattern is the appropriate contribution of each individual neurone to the population response.


This work was supported by a New Zealand Lottery Health Research Grant (#223744).