CHF (chronic heart failure) is a multifactorial disease process that is characterized by overactivation of the RAAS (renin–angiotensin–aldosterone system) and the sympathetic nervous system. Both of these systems are chronically activated in CHF. The RAAS consists of an excitatory arm involving AngII (angiotensin II), ACE (angiotensin-converting enzyme) and the AT1R (AngII type 1 receptor). The RAAS also consists of a protective arm consisting of Ang-(1–7) [angiotensin-(1–7)], the AT2R (AngII type 2 receptor), ACE2 and the Mas receptor. Sympatho-excitation in CHF is driven, in large part, by an imbalance of these two arms, with an increase in the AngII/AT1R/ACE arm and a decrease in the AT2R/ACE2 arm. This imbalance is manifested in cardiovascular-control regions of the brain such as the rostral ventrolateral medulla and paraventricular nucleus in the hypothalamus. The present review focuses on the current literature that describes the components of these two arms of the RAAS and their imbalance in the CHF state. Moreover, the present review provides additional evidence for the relevance of ACE2 and Ang-(1–7) as key players in the regulation of central sympathetic outflow in CHF. Finally, we also examine the effects of exercise training as a therapeutic strategy and the molecular mechanisms at play in CHF, in part, because of the ability of exercise training to restore the balance of the RAAS axis and sympathetic outflow.
CHF (chronic heart failure) is the end result of various insults to the myocardium, the most common of which is ischaemic heart disease . Whenever cardiac output falls, more than momentarily, compensatory mechanisms are called into play in an attempt to maintain blood pressure and organ perfusion. Primary to this compensation is activation of the sympathetic nervous system and the RAAS (renin–angiotensin–aldosterone system). Increases in circulating NE [noradrenaline (norepinephrine)] and AngII (angiotensin II) evoke peripheral vasoconstriction and activate aldosterone secretion largely through α-adrenergic receptors and AT1Rs (AngII type 1 receptors) respectively. An increase in renal renin release is undoubtedly caused by renal sympatho-excitation as well as a decrease in renal perfusion pressure in the CHF state [2–4]. Although these compensatory changes are initially beneficial they become counterproductive if sustained for prolonged periods of time. Chronic increases in both plasma and tissue NE not only result in a down-regulation of β-adrenergic receptors in the heart [5,6], but also contribute to further myocyte death . Activation of the RAAS contributes in a major way to salt and water retention and to further sympatho-excitation [8,9].
CENTRAL AT1Rs IN HEART FAILURE
A large body of evidence points to the central RAAS in mediating sympatho-excitation in the setting of CHF [3,10–14]. Areas in the RVLM (rostral ventrolateral medulla) and in the hypothalamus [e.g. the PVN (paraventricular nucleus)] have been shown to mediate an increase in sympathetic outflow in response to microinjection of AngII [15–25]. In studies carried out in rabbits with pacing-induced CHF, Liu and Zucker  showed that central infusion of the AT1R blocker losartan reduced RSNA (renal sympathetic nerve activity). Furthermore, Zucker et al.  showed an increase in AngII in the cerebral spinal fluid of dogs with pacing-induced CHF compared with sham animals. In an attempt to understand further the role of central AT1Rs in the sympatho-excitatory process, Gao et al.  measured AT1R protein and mRNA in the RVLM of rabbits with pacing-induced CHF, both of which were increased (Figure 1). Interestingly, in that study and other studies , it was shown that this increase was associated with increased oxidative stress and RSNA in these animals. Similar observations have been made in rats with coronary artery ligation-induced CHF [30,31]. In the rat model, Zhu et al.  carried out a novel study in which rats were infused with the antisense oligonucleotide to the AT1R while recording RSNA and blood pressure (Figure 2). Compared with animals infused with scrambled antisense, inhibition of AT1Rs resulted in a decrease in RSNA. Up-regulation of the central AT1R is not limited to the RVLM and PVN in CHF; the SFO (subfornical organ), NTS (nucleus of the solitary tract) and area postrema are other areas of the brain that have also shown an up-regulation of AT1Rs in the setting of CHF [33,34].
AT1R mRNA and protein expression in rats with CHF
Effects of icv administration of antisense oligodeoxynucleotides and scrambled oligodeoxynucleotides on baseline RSNA, mean arterial pressure and heart rate in sham-operated and CHF rats
The ability of the central RAAS to be a key player in the reverberating circuit of heart failure is not limited to the CNS (central nervous system). Indeed, icv (intracerebroventricular) blockade of the angiotensin receptor improved baroreflex sensitivity and decreased efferent RSNA in rats with CHF [35,36]. Central blockade of ACE (angiotensin-converting enzyme) similarly decreased RSNA, improved the blunted baroreflex sensitivity, and normalized sodium consumption, urine sodium and urine volume in rats with CHF . Furthermore, in an MI (myocardial infarction) rat model with a transgenic deletion of angiotensinogen, left ventricular end-diastolic pressure did not increase to the same extent as control rats with CHF and left ventricular dP/dt max did not decrease to the same extent as control rats with CHF . Taken together, the hyperactive central RAAS is a contributor to the global physiological changes as well as the cardiovascular dysfunction observed in CHF.
This apparent increase in AT1R signalling in the CNS appears to be mediated by a positive feedback of AT1Rs on the transcriptional regulation of the protein. AT1R is up-regulated in response to AngII and in the CHF state [17,29,38,39]. In a neuronal cell line (CATH.a) that express AT1Rs, Mitra and Zucker  showed that, in response to AngII (100 nM), an NF-κB (nuclear factor κB)-dependent increase in AT1R transcription ensued. This increase was also dependent on the downstream activation of two additional transcription factors, namely Elk-1 and AP-1 (activator protein-1). This pathway was confirmed in intact rabbits with CHF in which c-Jun (one of the two transcription factors that dimerize to form AP-1) and JNK (Jun N-terminal kinase) were increased in the RVLM . The NF-κB pathway has been shown to mediate an increase in sympathetic nerve activity in CHF since its blockade reduces sympathetic outflow, AT1R expression and oxidative stress in rats with CHF and hypertension [30,41,42]. This sympatho-excitation in response to central AngII is most likely to be mediated by a decrease in the outward potassium current [43–45]. In a previous study, Gao et al.  showed that potassium channel protein Kv4.3 was down-regulated in the brainstem of rats with CHF, suggesting that this may contribute to enhanced membrane depolarization and action potential generation. Kv4.3 contributes to the transient outward current and is most prominently reduced in cardiac myocytes in CHF [47,48]. The mechanism by which AngII decreased potassium current is not clear, but may result from interaction of a Kv4.3–AT1R complex .
A ROLE FOR ACE2 IN THE SYMPATHO-EXCITATORY PROCESS
The discovery of ACE2 and generation of Ang-(1–7) [angiotensin-(1–7)] [50,51] as important components of the RAAS has resulted in an explosion of studies on the biological and therapeutic effects of this arm of the RAAS. ACE2 activation has been shown to have beneficial effects in a variety of disease states . Furthermore, Ang-(1–7) has been shown to be beneficial in the setting of systemic hypertension [53,54], pulmonary hypertension , renal disease  and cancer [57–59]. Since ACE2 has been found in the brain [60,61], it seems reasonable that this enzyme and its product Ang-(1–7) would modulate the generation of central sympathetic outflow in CHF. Using a unique transgenic mouse model that overexpresses human ACE2 selectively in neurons , Xiao et al.  examined the sympathetic nervous effects in transgenic and wild-type mice who were subjected to a chronic MI and the subsequent development of CHF. Although there were no major differences in cardiac function in both groups of mice, transgenic mice exhibited a lower RSNA and an improvement in arterial baroreflex function (Figure 3). Mice that overexpressed central ACE2 were able to suppress RSNA to zero during increases in blood pressure in contrast with wild-type mice with CHF who could not lower RSNA in response to an increase in blood pressure. Examination of the spontaneous baroreflex control of heart rate also indicated an enhanced sympatho-inhibitory process in these mice. In a study by Zheng et al. , it was shown that viral overexpression of ACE2 reduced RSNA in a rat CHF model. This effect was apparently mediated by an increase in nitric oxide.
Baroreflex response to elevation in blood pressure induced by phenylephrine
In an attempt to understand the cellular mechanisms by which overexpression of ACE2 in neurons regulates AT1R expression, we carried out in vitro studies where the neuronal cell line CATH.a was transfected with a lentivirus that resulted in overexpression of human ACE2. As demonstrated in Figure 4, AngII stimulated an increase in AT1R expression, which was blocked by the AT1R antagonist losartan. Overexpression of ACE2 completely prevented the increase in AT1Rs in response to AngII. Interestingly, this suppression was not reversed by the Mas receptor antagonist A-779. These results suggest that the efficacy of ACE2 to prevent the up-regulation of AT1Rs may be related to its ability to degrade AngII rather than to the generation of Ang-(1–7). Because overexpression of human ACE2 may be quite non-physiological and expression may be increased severalfold over endogenous mouse ACE2, the effects of exogenous Ang-(1–7) was tested in order to determine whether this peptide could alter AT1R expression in vitro. Figure 5 shows the results from this experiment. Again losartan prevented the up-regulation of AT1Rs in response to AngII. Ang-(1–7) also increased AT1R expression. Both the response to AngII and the response to Ang-(1–7) were blocked by losartan and neither was blocked by A-779. Taken together, these data strongly suggest, at least in vitro, that Ang-(1–7) does not prevent the up-regulation of AT1Rs and that the efficacy of ACE2 in this response is mediated by an AT1R-dependent reduction in AngII. Therefore increases in Ang-(1–7) may be secondary to a reduction in AT1R activation as a therapy in the setting of CHF. However, therapies that both stimulate the Ang-(1–7) pathway and decrease AT1Rs would be of additive benefit. It is also possible that Ang-(1–7) signalling through the Mas receptor augments nitric oxide production, which has been shown to decrease AT1R expression . Although manipulations of ACE2 may provide insight into the autonomic effects of this enzyme, they do not directly address the question of whether Ang-(1–7) is capable of modulating sympathetic nerve activity in a beneficial direction, especially in cardiovascular disease states. There has been controversy over the effects of Ang-(1–7) on sympathetic outflow and baroreflex function. In early experiments, Potts et al.  and da Silva et al.  provided evidence for a sympatho-excitatory effect of Ang-(1–7). On the other hand, Xia et al.  and Diz et al.  have provided evidence that Ang-(1–7) exerts sympatho-inhibitory effects. In a study by Kar et al. , conscious rabbits with and without CHF were infused by the icv route with Ang-(1–7) for several days. Autonomic balance to the heart was assessed by evaluating the heart rate response to an acute bolus injection of either atropine to assess vagal tone or metoprolol to evaluate cardiac sympathetic outflow. Ang-(1–7) increased vagal tone in rabbits with CHF (i.e. a greater increase in heart rate in response to atropine) and decreased sympathetic tone (a smaller decrease in heart rate in response to metoprolol). There was no effect of Ang-(1–7) in sham rabbits. In addition, chronic icv infusion of Ang-(1–7) profoundly reduced RSNA in conscious rabbits with CHF while having no effect in normal rabbits. Importantly, this effect was mediated by the Mas receptor since it was reversed when Ang-(1–7) was co-infused with A-779 (Figure 6). Consistent with these results, baroreflex gain for both heart rate and RSNA was increased in rabbits with CHF.
AT1R protein expression in Cath.a neurons which were incubated with AngII alone or in combination with either a Mas receptor antagonist A-779 or AT1R antagonist losartan
AT1R expression is up-regulated by both AngII and Ang-(1–7) in CATH.a cultured neurons
GRK5, AT1R and NF-κB expression in PVN and RVLM of CHF and CHF-exercise-trained rats
MODULATION OF CENTRAL RAS (RENIN–ANGIOTENSIN SYSTEM) COMPONENTS BY EXERCISE TRAINING IN CHF
There is a growing trend to consider the use of non-pharmacological therapy in the treatment of CHF. One such intervention is ExT (exercise training). In 2003, a position statement by the American Heart Association concluded that ExT in patients with CHF is safe and likely to be an effective treatment paradigm . There is now clear evidence that ExT in patients with CHF increases quality of life and improves survival from all cardiac events [71–74], even if cardiac function is not improved. However, in elderly patients the benefits may not be as great [75,76]. In experiments carried out in rabbits with pacing-induced CHF, Mousa et al.  and Liu et al.  showed that ExT reduced AT1R expression in the RVLM and reduced plasma AngII. Similar results have been reported in rats with MI-induced CHF . In humans, studies by Roveda et al.  and Negrão et al. [81–83] have clearly shown a decrease in MSNA (muscle sympathetic nerve activity) in CHF patients following ExT. These studies raise important questions concerning the role of the central RAAS in mediating the sympatho-inhibitory effects of ExT in the setting of CHF. What does ExT do to central AngII signalling and the components of the RAAS system in the CNS? ExT in experimental CHF has been shown to reduce central AT1Rs [77,84] and oxidative stress [85–87] while at the same time increasing the sympatho-inhibitory effects of nitric oxide [84,88] and improving baroreflex function [77,78,86,89]. There is much less known about ACE and ACE2 in the CNS following ExT in the CHF state. In a study carried out in rabbits with pacing-induced CHF, Kar et al.  examined the relationship between ExT and the expression of ACE and ACE2 in the RVLM and PVN. In both regions, ACE was elevated in the CHF state and ACE2 was decreased. Following ExT the expression of these two proteins were reversed and looked very similar to sham animals. Although there are several studies showing that ExT reduces ACE in the myocardium [91,92], the study by Kar et al.  is the first to show a reversal of the ACE/ACE2 ratio in the brain of animals with CHF. Because Ang-(1–7) is thought to be sympatho-inhibitory and AngII sympatho-excitatory, a decrease in the ratio of ACE to ACE2 should be beneficial in reducing sympathetic outflow in CHF and hypertension. This in turn would also reduce oxidative stress and increase nitric oxide at the local level. Theoretically, the ACE/ACE2 balance would also mediate the relative concentrations of AngII and Ang-(1–7) in the brain. In vitro evidence has shown that Ang-(1–7) can mediate an increase in neuronal potassium current by a nitric oxide- and Mas receptor-dependent mechanism .
The expression of AT1Rs in the brain is critically tied to signalling through AT1Rs in animals with CHF. This positive feedback nature was shown in CHF rabbits that underwent an ExT regimen while simultaneously receiving a chronic infusion of AngII so that plasma AngII would not be normalized . Under these conditions, AT1R mRNA and protein in the RVLM remained elevated (compared with non-ExT CHF rabbits). Furthermore, resting RSNA and arterial baroreflex function remained high and low respectively. These data, along with those from chronic central losartan infusion, strongly suggest that both high levels of AngII and increased AT1R signalling are necessary for sympatho-excitation and that the normalization of these parameters by ExT are mediated, at least in part, by a reduction in both.
REGULATION OF CENTRAL RAS COMPONENTS
Because plasma AngII levels are reduced following ExT in CHF animals  (although it is unclear whether this is also true for AngII in the brain), it is of some interest to determine the influence of AngII on ACE and ACE2 expression. In a recent in vitro study, Xiao et al.  clearly showed that AngII treatment of neurons resulted in an increase in ACE and a decrease in ACE2 in a dose-dependent manner. At the mRNA level, the changes for both proteins in response to AngII could be inhibited by blocking p38 MAPK (mitogen-activated protein kinase) or ERK1/2 (extracellular-signal-regulated kinase 1/2). This seems to be selectively mediated by the AT1R as it was also blocked by losartan, but not by the AT2R (AngII type 2 receptor) blocker PD123319. This reciprocal relationship appears to hold in other tissues and in other disease states such as hypertension [95–98]. These data suggest that AngII can set into motion a series of transcriptional events through common cell signalling pathways to regulate the balance between ACE and ACE2. However, the transcription factors that drive the regulation of both ACE and ACE2 have not yet been identified.
Although the transcription of new AT1R protein may be an important contributor to central AngII signalling and another mechanism of potential importance relates to the way the AT1R is turned over. The AT1R is a G-protein-coupled receptor and signals through a Gq/11 and other G-protein mechanisms . As such, its phosphorylation following agonist binding is mediated by GRKs (G-protein receptor kinases) . Following phosphorylation, the protein is targeted for internalization by β-arrestin, after which it is degraded in lysosomes . Recent experiments carried out in rats with CHF have shown that GRK5 is up-regulated in the RVLM and PVN (Figure 6) and binds to the AT1R . On the other hand, GRK2, the more classical β-adrenergic receptor kinase that can also regulate AT1R is not changed. Interestingly, the increase in GRK5 occurs at the same time that the AT1R is up-regulated. Following ExT, both GRK5 and AT1R were decreased. This parallel change in both proteins suggests that the increase in GRK5 is a compensatory response to the increased AT1R expression. The increase in GRK5 may not be effective in decreasing AT1R expression due to intense stimuli that increase the transcription of this receptor (e.g. NF-κB) in CHF.
In vitro experiments carried out in CATH.a neurons confirmed that substantial up-regulation of GRK5 results in a decrease in AT1R protein. Under these conditions, the AngII-mediated increase in AT1R expression was completely blocked (Figure 7). On the other hand, GRK5 knockdown with an siRNA caused an increase in AT1R expression in response to AngII. Taken together, these studies suggest that a balance exists between the transcriptional regulators of AT1Rs and the pathways responsible for degrading the AT1Rs. In the setting of CHF, the former apparently predominate, thus promoting an AngII-dependent neuronal depolarization and increase in sympathetic outflow.
GRK5 overexpression normalizes AT1R and p65 NF-κB protein levels following stimulation with AngII
Despite our increasing knowledge on the regulatory pathways of AT1R and central RAS components, the induction of the central RAS, especially in regions with an intact blood–brain barrier like the RVLM, remains unclear. Although angiotensinogen is found in brain extracellular fluid and cerebrospinal fluid, astrocytes and, more recently in the neurons of many brain regions including the PVN, NTS, RVLM and SFO, the cell types in which renin, ACE, aminopeptidase A and aminopeptidase N are found in the brain are still controversial. To date, only low levels of AngI (angiotensin I), AngII and Ang-(1–7) have been identified in brain tissue . Therefore one possibility is that the AngII from the periphery detected in CVOs (circumventricular organs) would induce a signalling cascade in non-CVO nuclei. Conversely, given that much of the RAS is expressed centrally between neurons, astrocytes and endothelial cells, angiotensins are not only present in the brain, but may function as neurotransmitters .
Data from this laboratory and others strongly suggest that circulating AngII is a primary driver of the imbalance for AT1R/ACE and AT2R/ACE2. Both in vitro and in vivo studies have demonstrated that AngII mediated the increase in AT1Rs and its pathway components are dependent on AT1R signalling; pre-treatment with losartan blunts this imbalance [29,38].
Clearly, the regulation of sympathetic nerve activity is a complicated and multifactorial process. The present review only highlights one mechanism that plays a role in this process. AngII along with other peptide and non-peptide mediators can alter neuronal membrane potential, in part, by reducing outward potassium currents. Most central pre-sympathetic neurons express all the receptors involved in signalling through the RAS.
Therefore the balance between AngII and other peptides such as Ang-(1–7) and the balance between AT1Rs, AT2Rs and Mas receptors may be critical in establishing the level of neuronal activation. Furthermore, the synthesis of AngII and Ang-(1–7) due to ACE and ACE2 also appears to contribute to sympatho-excitation in CHF. Figure 8 outlines the major mechanisms in the RAS system that have thus far been defined to regulate sympatho-excitation in CHF. Increases in central AngII initiate a positive feedback process whereby the AT1R is up-regulated by an AngII/AT1R-dependent mechanism. Intracellular activation of NF-κB and downstream transcription factors, Elk-1 and AP-1, increase the transcription of the AT1R. This process is accompanied by a compensatory increase in GRK5 in an attempt to limit AT1R up-regulation. However, the apparent intensity of the stimuli to up-regulate this protein far outweighs the ability of GRK5 alone to reduce the AT1R protein.
Schematic diagram showing the relationship between RAS metabolites, the AT1R signalling cascade, and sympathetic outflow in CHF and following ExT
The relative paucity of new pharmacological agents in the treatment of CHF has stimulated a search for non-pharmacological therapies. In addition to novel device therapy (e.g. baroreflex stimulation, renal denervation and vagal stimulation), ExT has been promoted as a way of reducing mortality and increasing the quality of life for patients with CHF. The mechanisms by which ExT is efficacious in this regard are not well understood. Although ExT is known to affect virtually every organ system, the focus on central sympathetic remodelling is starting to define some of the pathways affected by this intervention . ExT has an impact on the central RAS in a major way. Importantly, it reduces oxidative stress in the RVLM and causes an up-regulation of both SOD1 (superoxide dismutase 1) and SOD2 (superoxide dismutase 2) . Since AngII signals, in part, through the NADPH oxidase-dependent production of superoxide, ExT is likely to have a major impact on AngII signalling. Indeed ExT lowers plasma AngII and reduces AT1R expression which is dependent on activation of the AT1R (thus positive feedback). Furthermore, ExT reduces ACE and increases ACE2 in the setting of CHF. Therefore the role of ExT in modulating the central RAS would seem a fruitful area of continued investigation. This simple and inexpensive intervention may provide some of the benefits of currently used pharmacological therapies and can also be used as adjunctive therapy.
Many questions still remain in the regulation of central RAS in the setting of CHF. It will be important to determine the precise location and mechanism(s) by which central AngII is generated, and what initiating signal drives the feed-forward activation of the RAS in the setting of CHF. Conversely, the precise central and peripheral signals generated by ExT that trigger the downstream effects outlined in the present review which lead to protection in the setting of CHF also remain unclear. Additionally, the mechanism(s) by which nitric oxide can negatively regulate AT1Rs are still unknown. Finally, because many existing therapies that target the RAS do not improve cardiac parameters, it will be important to develop novel therapies that improve both autonomic imbalance and hemodynamic parameters.
AngII type 1 receptor
AngII type 2 receptor
chronic heart failure
central nervous system
G-protein receptor kinase
nuclear factor κB
nucleus of the solitary tract
renal sympathetic nerve activity
rostral ventrolateral medulla
Some of the data shown in this review was supported by the National Heart, Lung and Blood Institute [grant number HL PO1 62222]. K.K.V.H. was supported by an F3 National Institutes of Health post-doctoral fellowship [grant number HL F32-116172].