In experimental animals and humans blood loss or fluid retention leads to activation of a number of physiological mechanisms which promote restoration of vascular volume and pressure to some set point (Schadt & Ludbrook, 1991). Haemorrhage and venous volume expansion in heart failure are two extreme examples of situations where these mechanisms are clearly brought into play. However, an optimum blood volume is so essential for normal performance of all types of body functions that it is regulated on a moment-by-moment basis via detection of several plasma volume-related signals. Some of these signals are more significant for long-term regulation, and these include circulating levels of angiotensin, aldosterone and osmolality. Other signals more directly reflect changes in vascular volume, and the most important of these are the mechanoreceptors at the venous–atrial juctions of the heart. These volume receptors either fire phasically in time with the cardiac cycle or more tonically, depending on their location, and they provide information to the brain about the central venous volume and force of atrial contraction (Gupta et al. 1966). They are sensitive enough to give a precise rapid indication of the fullness of the thoracic circulation, so that in anaesthetized dogs with closed chests a fall in blood pressure of 10% reduces the number of action potentials per beat to 50%, and a change of 20% in blood volume produces a drop of 80% in receptor discharge, and vice versa (Gupta et al. 1966). This implies that fluctuations in venous volume of less than 1% can be signalled to the brain. The venous volume receptors are therefore ideally suited to act as the afferent arm of a relatively rapid response system using autonomic nerves, whilst at the same time contributing to the slower adjustments via the neuroendocrine system.
It is now well accepted that the sympathetic nervous system responds to specific afferent stimuli in a unique non-uniform fashion. The means by which the brain transforms the signals from a single type of receptor into an appropriate differential sympathetic output is discussed in this brief review. The detection of and response to venous filling are used for illustration. An expansion of blood volume has been shown in a number of species to increase heart rate reflexly via sympathetic nerves and this effect is primarily an action of volume receptors at the venous–atrial junctions of the heart. Stimulation of these volume receptors also leads to an inhibition of renal sympathetic nerve activity. Thus the reflex response to an increase in plasma volume consists of a distinctive unique pattern of sympathetic activity to maintain fluid balance. This reflex is dependent on neurones in the paraventricular nucleus (PVN). Neurones in the PVN show early gene activation on stimulation of atrial receptors, and a similar differential pattern of cardiac sympathetic excitation and renal inhibition can be evoked by activating PVN neurones. Cardiac atrial afferents selectively cause a PVN GABA neurone-induced inhibition within the PVN of PVN spinally projecting vasopressin-containing neurones that project to renal sympathetic neurones. A lesion of these spinally projecting neurones abolishes the reflex. With regard to the cardiac sympathetics, there is a population of PVN spinally projecting neurones that selectively increase heart rate by the release of oxytocin, a peptide pathway that has no action on renal sympathetic outflow. In heart failure the atrial reflex becomes blunted, and evidence is emerging that there is a downregulation of nitric oxide synthesis and reduced GABA activity in the PVN. How this might give rise to increased sympathetic activity associated with heart failure is briefly discussed.
Atrial receptor-mediated adjustments in the autonomic nervous system
Plasma volume expansion or selective stimulation of venous–atrial junctions with balloon distension has been shown to cause a sympathetically mediated increase in heart rate (Bainbridge, 1915; Kappagoda et al. 1975; Kaufman et al. 1981) in anaesthetized dogs and rats. Simultaneously this stimulus results in a decrease in sympathetic activity to the kidneys and renal vasodilatation (Karim et al. 1972; Kopp et al. 1987; Pyner et al. 2002), as well as increasing urine flow and sodium loss (Kaufman & Deng, 1993). Thus the atrial receptors elicit a unique differential pattern of sympathetic activity to the heart and kidney to defend against plasma volume expansion. How then does the brain achieve this remarkable transformation of an increase in activity of a single type of receptor into a differential activation or inhibition of neurones supplying the heart and kidney?
Identification of the command centre for the volume reflex
The afferent pathway to the brain for atrial volume receptors is the vagus nerve and the first synapse is within the nucleus tractus solitarii (NTS) of the medulla oblongata. This was demonstrated by Kappogoda and coworkers (reported by Kidd, 1979) in the only study so far in which increases in unit activity were recorded following inflation of balloons placed at the pulmonary venous–atrial junctions in anaesthetized dogs. From the NTS the volume signals travel to the hypothalamus to influence neurones in the paraventricular nucleus (PVN; Lovick & Coote, 1988b; Pyner et al. 2002).
A projection to the PVN is unsurprising, since it has long been known that the magnocellular peptidergic neurohypophyseal neurones play a key part in neuroendocrine control of fluid balance. Recent data, however, show activation of the early gene c-fos in many smaller neurones lying in the parvocellular subnuclei, in response to plasma volume signals (Deng & Kaufman, 1995; Badoer et al. 1997; Pyner et al. 2002). The parvocellular neurones project to extrahypothalamic sites, including sites in the spinal cord and brainstem that are involved in cardiovascular regulation (Shafton et al. 1998; Pyner & Coote, 2000). The PVN spinally projecting neurones are especially interesting because they display characteristics consistent with a role in the plasma volume expansion reflex. In a series of studies, Lovick & Coote (1988a,b, 1989) antidromically identified these neurones and showed that some were activated and others inhibited by stimulation of cardiac vagal afferents or by circulating atrial natriuretic peptide (ANP). Furthermore, in a follow-up study, Lovick et al. (1993) selectively destroyed more than 80% of this population and found that the renal vascular response to systemic volume load was virtually abolished. This important result was confirmed in similar experiments by Haselton et al. (1994), who showed that a reflex reduction in renal sympathetic nerve traffic in response to plasma volume expansion was markedly attenuated following destruction of PVN parvocellular neurones. Although this evidence does not rule out an involvement of other parvocellular groups of neurones, it does strongly indicate that the PVN is a ‘command centre’ for the atrial receptor reflex. This idea is supported by the demonstration in anaesthetized rabbits that quite low amounts of the neurone excitant DL-homosysteic acid preferentially evoke excitation of cardiac sympathetic activity and simultaneously evoke inhibition of renal sympathetic activity from some parvocellular regions of the PVN (Deering & Coote, 2000).
Location of the volume reflex pattern generator
The observation that microinjection of DLH at specific locations in the PVN differentially affects both cardiac and renal sympathetic outflows is revealing. The patterned response is unlikely to be due to activation of a mixed pool of PVN neurones projecting directly to cardiac and renal sympathetic preganglionic neurones, since retrograde transynaptic labelling studies with pseudorabies virus indicate that those PVN neurones projecting to the stellate ganglion lie more medially in the dorsal division of PVN (Strack et al. 1989) and some distance from those PVN neurones projecting to the kidney (Schramm et al. 1993). This would also appear to make it less probable that the response is due to activation of intermingled dendrites from the two populations. It was therefore concluded by Deering & Coote (2000) that the patterned response is more likely to be a consequence of excitation of an intrinsic PVN network of neurones ‘hard wired’ to respond appropriately to blood volume-related stimuli. This is strongly supported by studies which show the output neurones of the PVN are dominantly sympatho-excitatory. Microinjection into PVN of the GABA agonist muscimol reduces renal sympathetic activity and blood pressure (Zhang & Patel, 1998), while injection of the GABAA antagonist bicuculline does the opposite. Furthermore, the bicuculline-induced, as well as the DLH-induced, sympatho-excitatory effects on renal neurones can be selectively blocked at the spinal level by a vasopressin antagonist applied intrathecally to the lower thoracic spinal cord (Yang et al. 2002). The vasopressin innervation of spinal sympathetic neurones most probably only arises from the PVN (Cechetto & Saper, 1988), and sympathetic preganglionic neurones express vasopressin V1a receptors (Sermasi et al. 1998) and are directly excited by vasopressin (Ma & Dun, 1985; Sermasi & Coote, 1994). Therefore, the evidence indicates that there is a PVN spinally projecting vasopressin neurone excitatory influence on renal sympathetic preganglionic neurones which is tonically inhibited by a PVN network of GABA neurones. The next obvious question is whether GABA-containing interneurones within the PVN mediate the inhibition of renal sympathetic nerve activity associated with the volume expansion reflex. To address this we tested the effect of atrial balloon distension before and during microinjection of bicuculline into the PVN. The volume reflex inhibition of renal sympathetic activity was abolished by the GABA antagonist (Yang & Coote, 2003). Interpretation of this result is not straightforward, however, since the GABA antagonist alone increased sympathetic activity. Therefore Yang & Coote (2004) compared the effect of bicuculline with that of NO synthase inhibitors (l-NAME and l-NMMA) injected into PVN, which produce a similar increase in renal sympathetic activity. The latter effect was prevented by atrial balloon inflation, indicating that the rise in sympathetic activity was being prevented by the inhibitory reflex. Prevention of the action of other putative sympatho-inhibitory systems at the spinal level (Yang et al. 2002) by intrathecal application of GABA, glycine or dopamine antagonists also was without effect on the atrial receptor reflex. Therefore, the evidence strongly favours the view that inhibition of renal sympathetic activity following selective stimulation of venous volume receptors is mediated by a population of GABA interneurones within the PVN (Fig. 1).
Identification of the PVN cardiac arm of the volume reflex
At a number of sites within the PVN parvocellular subnuclei, microinjection of DLH was shown to evoke sympathetic-medated increases in heart rate (Yang et al. 2004). These increases were dependent on activation of a PVN spinally projecting oxytocin pathway, since they were blocked by a highly selective non-peptide oxytocin antagonist (Yang et al. 2002) given intrathecally to the upper thoracic spinal cord. They were not affected by the glutamate antagonist kynurenic acid or a selective vasopressin V1a antagonist (Yang et al. 2004). There is also evidence provided by an earlier study (Yashpal et al. 1987) that is suggestive of a cardiac-selective supraspinal oxytocin pathway in which intrathecal application of oxytocin to the upper thoracic spinal cord caused a sympathetic-mediated increase in heart rate without an effect on blood pressure. There is also electrophysiological evidence that sympathetic preganglionic neurones in the upper thoracic cord are directly excited by oxytocin (Desaules et al. 1995), whereas sympathetic neurones in the lower thoracic cord are not affected (Sermasi & Coote, 1994). In line with this, neuroanatomical studies show that sympathetic preganglionic neurones in the upper thoracic segments are innervated by oxytocin-containing fibres most probably originating from the PVN (Hosoya et al. 1995), whereas such fibres appear to avoid many sympathetic preganglionic neurones in the lower thoracic cord (Appel & Elde, 1988).
Therefore, although the effect of intrathecal application of an oxytocin antagonist on a volume reflex increase in heart rate has not so far been tested, the idea that it is mediated by activation of PVN oxytocin neurones is attractive.
Central organization of the volume receptor reflex
In summary, we can envisage the basic organization of the reflex response to activation of the stretch receptors at the venous–atrial junctions as comprising an afferent arm in the vagus to the NTS. From there it seems to connect to the PVN, where there is a discrete targetting of spinally projecting oxytocin neurones that innervate cardiac sympathetic preganglionic neurones in the upper thoracic spinal cord and of a specific pool of GABA interneurones that innervate a population of spinally projecting vasopressin neurones that synapse with renal sympathetic preganglionic neurones (Fig. 2). There is also a vasopressin/glutamate projection from the PVN to rostral ventrolateral medulla spinally projecting vasomotor neurones (Yang & Coote, 1998; Pyner & Coote, 1999; Yang et al. 2001), although its involvement in the volume expansion reflex is unclear. In addition, a recent study shows that the excitability of the spinal renal sympathetic neurone is modulated by a NO–glycine negative feedback loop (Yang & Coote, 2004).
The PVN and chronic heart failure
A characteristic feature in patients with chronic heart failure is fluid retention and this is accompanied by increased sympathetic nerve activity (Leimbach et al. 1986). Why does this occur? Although there may be some impairment of the sensitivity of peripheral receptors (Zucker, 1991) which regulate the level of sympathetic activity, there is a growing body of evidence pointing towards a central nervous abnormality. In a rat model of heart failure produced by coronary artery ligation, it has been shown that the renal sympatho-inhibition in response to acute volume expansion is blunted. In this model, PVN neuronal activity is increased and the endogenous GABA-mediated inhibition within the PVN is less. The latter was tested by comparing the ability of different concentrations of either bicuculline or muscimol to increase or decrease renal sympathetic nerve activity in control rats and in rats with heart failure. Neither the GABA antagonist nor the agonist was as effective in rats with heart failure as they were in control rats (Zhang et al. 2002; Li & Patel, 2003).
Now that we have a relatively good understanding of the volume expansion reflex pathway, the work is at an exciting stage. We are now in a position to address the question at the cellular level of what central nervous mechanisms lead to alterations in the GABA neurone influence in pathological states such as heart failure. It is likely that NO signalling is involved, since this molecule has a regulatory role in the release of GABA in the PVN (Li & Patel, 2003).
Many of the experiments reported in this brief review were supported by a project grant from The Wellcome Trust.