Neuronal Networks in Hypertension: Recent Advances

Abstract

Neurogenic hypertension is associated with excessive sympathetic nerve activity (SNA) to the kidneys and portions of the cardiovascular system. Here we examine the brain regions that cause heightened SNA in animal models of neurogenic hypertension and we discuss the triggers responsible for the changes in neuronal activity within these regions. We highlight the limitations of the evidence and, whenever possible, we briefly address the pertinence of the findings to human hypertension.

The arterial baroreflex reduces BP variability and contributes to the BP set-point. This set-point can also be elevated by a newly described cerebral blood flow-dependent and astrocyte-mediated sympathetic reflex. Both reflexes converge on the presympathetic neurons of the rostral medulla oblongata (RVLM PSNs) and both are plausible causes of neurogenic hypertension. Sensory afferent dysfunction (reduced baroreceptor activity, increased renal or carotid body afferent) contributes to many forms of neurogenic hypertension. Neurogenic hypertension can also result from activation of brain nuclei or sensory afferents by excess circulating hormones (leptin, insulin, AngII) or sodium. Leptin raises blood vessel SNA by activating the carotid bodies and subsets of arcuate neurons. AngII works in the lamina terminalis and probably throughout the brainstem and hypothalamus. Sodium is sensed primarily in the lamina terminalis. Regardless of its cause, the excess SNA is mediated to some extent by activation of presympathetic neurons (PSNs) located in the rostral ventrolateral medulla (RVLM) or the paraventricular nucleus of the hypothalamus. Increased activity of the orexinergic neurons also contributes to hypertension in selected models.

Neurogenic hypertension refers to a chronic increase in arterial blood pressure (BP) that is associated with and presumably driven by excessive sympathetic nerve activity (SNA) to the kidneys and various parts of the cardiovascular system^1–4. We will not discuss here how a global or regional increase in SNA causes or maintains the hypertensive state^5–7. Instead, we focus on two specific aspects of neurogenic hypertension research: the brain regions responsible for heightened SNA in animal models and the triggers responsible for the changes in neuronal activity within these regions (e.g. sensory afferent dysfunction, brain hypoperfusion, gliotransmission, excess circulating hormones, hypernatremia, etc.). The topic is vast and the space limited. Despite their importance, oxidative stress and inflammation are not discussed.

The ventrolateral medulla and hypertension

The rostral ventrolateral medulla (RVLM) is a major node of the BP regulating neural network^{8, 9}. Silencing RVLM neurons indiscriminately produces large reductions of BP and SNA in anesthetized or conscious rodents^9–11. These effects are attributed to the inhibition of excitatory neurons with monosynaptic projections to sympathetic preganglionic neurons, a.k.a. RVLM presympathetic neurons (PSNs). Many RVLM PSNs (C1 cells) synthesize catecholamines inclusive of epinephrine, the rest do not (non-C1)^12–14. Estimates of the ratio between non-C1 and C1 PSNs varies widely (33 to 300%)^{12, 13, 15, 16}. C1 and non-C1 PSNs are glutamatergic and express varying levels of neuropeptides (PACAP, NPY, substance P, enkephalins)⁹.

The degree to which the C1 and other RVLM neurons contribute to resting BP in an intact unstressed and unanesthetized mammal is unsettled. Extensive lesions of the C1 neurons (>85%) compromise BP stability during hypoxic or hypotensive stresses but cause very little hypotension at rest (<10 mmHg in rats)^{17, 18} suggesting that the contribution of these neurons to BP is small under resting conditions. Consistent with this view, in freely behaving rats at rest, bilateral optogenetic C1 cell inhibition (with the proton pump archaerhodopsin) also reduced BP very little (~5mmHg)¹⁹. Yet, a more substantial BP drop (−27 mmHg) was observed using a pharmacogenetic approach (allatostatin receptor)²⁰. The same type of vector (lentivirus with a Phox2b-enhanced promoter, PRSx8) was used in both studies. This vector does not transduce catecholaminergic neurons with total selectivity. The allatostatin receptor is a G-protein coupled receptor and, as such, presumably needs a much lower level of membrane expression than the proton pump to be an effective actuator, thus increasing the probability of off-target effects²¹. Conversely, archaerhodopsin requires a high level of expression for efficacy; its use could have led to the underestimation of the contribution of the C1 neurons to resting BP. Finally, the autonomic effects caused by a purely psychological stress may not be mediated via the RVLM²².

SNA to various organs or tissues is differentially regulated (e.g.^{23, 24}). This point is notable because hypertension may be caused by preferential activation of SNA to the heart, resistance vessels or splanchnic capacitance²⁵. The differential control of regional SNA may occur via differential recruitment of subsets of RVLM premotor neurons that control sympathetic efferents to particular vascular beds (e.g. splanchnic vs. muscle or capacitance vs. resistance) and different regions of the kidneys or the myocardium( Figure 1)^{9, 26–28}. However, the RVLM also contains highly collateralized PSNs, e.g. C1 cells, that can activate multiple sympathetic efferent pathways simultaneously^{29, 30}.

Figure 1:

Differential control of regional SNA with or without neurogenic hypertension, a hypothesis. Organ-specific SNA regulation operates via selective recruitment of RVLM presympathetic neurons and/or preganglionic neurons by inputs from a variety of CNS nodes (PVN, midline medulla, DMH, ARC, etc.; only one is represented). A) differential recruitment of regional SNA produces very little BP change if the baroreflex (NTS + CVLM; box) is unaffected. B) differential recruitment of regional SNA is associated with a rise in BP (potentially causing hypertension) if the baroreflex (NTS + CVLM; box) is simultaneously down-regulated.

RVLM PSNs, C1 included, innervate many brain regions implicated in BP control besides the spinal cord including the solitary tract nucleus (NTS), parabrachial region, locus coeruleus, periaqueductal gray matter and raphe pallidus (Figure 2)^{9, 31}. Non-bulbospinal RVLM neurons, including C1 neurons, also target hypothalamic nuclei of prime importance to cardiovascular control, notably orexin neurons and the PVN^{31, 32}. In brief, the notion that RVLM controls SNA exclusively via its spinal projections to preganglionic neurons has not been tested rigorously and seems a priori inconsistent with the available anatomical data.

Figure 2:

The C1 neurons innervate many nuclei implicated in circulatory control besides the sympathetic preganglionic neurons (see list of abbreviations). The projections of the rostral C1 cells are shown on the left; the projections of the caudal C1 neurons are shown on the right. Note that that both populations target many of the same nuclei.

Many factors (synaptic, intrinsic or paracrine) determine the discharge rate of the PSNs. Conventional excitatory and inhibitory synaptic inputs clearly play a role¹⁸. The neurons antecedent to RVLM PSNs reside within the lower brainstem reticular formation, with fewer neurons located in the dorsal vagal complex, dorsolateral pons, midbrain and hypothalamus^{33, 34}. The inhibition of RVLM PSNs elicited by arterial baroreceptor activation is mediated by GABAergic interneurons located in the caudal VLM (CVLM; Figure 1); these interneurons are also strongly respiratory modulated and therefore must contribute to the respiratory modulation of RVLM PSNs and, ultimately, SNA^{8, 13, 35–38}. The CVLM region, in turn, receives extensive input from the dorsal vagal complex and the hypothalamus³⁹.

The activity of C1 and other RVLM PSNs is also partly independent of conventional ionotropic synaptic transmission^{40, 41}. This activity relies on intrinsic properties (auto-depolarization) and local factors such as hypoxia, oxidative stress, circulating hormones and paracrine signals released by the surrounding vasculature or glia (ATP, lactate, PGE2, NO)^42–47.

The RVLM contributes to several forms of neurogenic hypertension^48–51. Somewhat higher c-Fos expression has been reported in the RVLM of SH rats⁵². Inhibiting RVLM neurons indiscriminately produces a greater BP drop in hypertensive than normotensive control strains (e.g. SH vs. Wistar-Kyoto) or in a given strain after it has been subjected to a treatment that produces hypertension (salt, overfeeding)^48–51. The favored interpretation of such results is that the discharge rate of the PSNs is higher in hypertensive animals but there other possibilities exist. RVLM unit activity could be unchanged or only marginally increased but downstream efferent connections, e.g. SPGNs, could be hyper-responsive to their input; RVLM unit activity and SNA could both be virtually unchanged but vascular reactivity could be enhanced by sympathetic hyper-innervation and arteriolar hypertrophy^{53, 54} or via sensitized neuroeffector coupling²⁵. Finally, the RVLM could raise BP via projections elsewhere than the spinal cord (e.g. the hypothalamus). These alternative possibilities have been minimally tested⁵⁰ and never in conscious animals. What is generally underappreciated is that the discharge rate of RVLM PSNs depends to an extreme degree on the resting BP and the degree of impairment of the baroreflex, both of which are highly affected by the preparation and the type and depth of anesthesia. In halothane-anesthetized rats, no difference was found between SH and Wistar-Kyoto except a resetting of the baroreflex⁵⁵. In neonatal brainstem spinal preparations, RVLM bulbospinal (putative PSNs) neurons discharged at higher rates (e.g.⁵⁶) and in the arterially perfused SH rat the only difference was an increased respiratory modulation of the PSNs^{57, 58} which, as discussed later, could be related to excess accumulation of metabolically produced CO₂ in preparations with higher vascular resistance. Finally, preferential C1 cell inhibition produced the same BP reduction in control rats as in rats rendered hypertensive by chronic intermittent hypoxia²⁰. This result could mean that the C1 neurons are not hyperactive after CIH and that the postulated heightened SNA might have another cause⁵⁹.

In sum, the RVLM contributes somehow to the elevated BP present in most hypertensive animal models. PSN hyperactivity is a plausible explanation in need of further evidence.

Cardiorespiratory coupling and hypertension

This topic has been recently reviewed⁹ and will only be briefly updated. SNA exhibits respiratory-synchronous fluctuations for two reasons. First, cardiopulmonary receptors (including baroreceptors) discharge in phase with chest movements and these periodic fluctuations are transmitted to the network that generates SNA without obligatory relay through the respiratory pattern generator. Secondly the network that generates the SNA, for example CVLM neurons and RVLM PSNs receive inputs from the respiratory pattern generator^{8, 13, 57}. In this section, we discuss whether these inputs are abnormally strong in two animal models of neurogenic hypertension, the SH rat and rats subjected to CIH.

Arterially perfused preparations of SH rats have a higher central respiratory pattern activity than control Wistar rats; this is manifested by narrow, non-ramping, phrenic nerve discharges and by the presence of late expiratory activity in abdominal nerves⁵⁸. In this model, enhanced respiratory modulation of SNA and RVLM PSNs has been observed in all three phases of the respiratory cycle (inspiration, post-inspiration and late expiration) with qualitative differences between individual RVLM PSNs, between sympathetic nerves and according to the animals’ age^{57, 58, 60}. Similar observations have been made in rats subjected to CIH except that SNA appears enhanced predominantly during the inspiratory or late-expiratory phases, depending on the animals’ sex^{13, 61}. The exaggerated respiratory phasic components of SNA have been attributed to enhanced central respiratory-sympathetic coupling^{13, 57}. However, the enhanced respiratory fluctuations of SNA may be the normal consequence of an increase in central chemoreceptor activity because the abnormal pattern of the hypertensive animals is normalized by lowering perfusate PCO₂^{58, 62}. This explanation has been rejected by some authors because they could not find any difference in the respiratory chemoreflex of the SH rat⁶⁰. Yet, hypertensive strains have an elevated cerebrovascular resistance^{60, 63}. If flow is limited, metabolically produced CO₂ will inevitably accumulate to a higher level and overstimulate the brain chemoreceptors, ultimately enhancing fictive breathing and the respiratory modulation of SNA^64–66.

Unlike in a perfused rodent preparation, breathing at rest is the same in intact unanesthetized SH and WKY rats⁶⁷ and neither breathing nor the respiratory fluctuations of SNA differ noticeably between normotensive humans and individuals with essential hypertension⁶⁸. One study did report the presence of active expiration in a few awake rats subjected to CIH but these rats were equally hypertensive during sleep despite the absence of active expiration⁶⁹. Finally, although daytime SNA is indeed elevated in patients with OSA, resting breathing and SNA respiratory fluctuations seem normal⁷⁰. That breathing is unaffected in the awake resting state might appear counterintuitive given the well-documented hyperactivity of the carotid bodies (see next section); the powerful countervailing effect of the central chemoreceptors on breathing is one explanation⁷¹.

Sensory afferent dysfunction and hypertension

Renal afferents, most of which are unmyelinated, contribute to hypertension and cause a widespread increase in SNA in the DOCA-salt and the 2K-1C (2-kidney-1 clip) models^{72, 73}. Renal inflammation is the suspected trigger. Unmyelinated afferents typically terminate in the dorsal-most laminae of the spinal cord. This region projects in turn to the intermediolateral cell column, the NTS, the parabrachial nucleus and the rostral medulla, all of which could contribute to raise SNA⁷⁴.

The arterial baroreflex is critically important to BP. Sinoaortic denervation (SAD) causes BP lability. Baroreflex modulation elicited by CNS inputs to the NTS, the CVLM and elsewhere, is likely required for BP to rise or fall appropriately during various behaviors (Figure 1)⁷⁵. The arterial baroreflex is down-regulated by carotid body afferents, which contributes to various forms of neurogenic hypertension⁷⁶. Incomplete resetting of the baroreceptor afferents may also cause hypertension even though SAD produces very small BP increases (<10 mmHg)^{6, 19, 77, 78}. Silencing baroreceptors in mice by deleting both Piezo channels from vagal afferents (Phox2bCre⁺;Piezo1^f/fPiezo2^f/f) produces hypertension (~15 mmHg) and a modest but significant increase of BP lability⁷⁹. This genetic approach has limitations (gene knock-out from early developmental stage; gene excision not limited to baroreceptor afferents, etc.) but it suggests that a chronic reduction of the activity of baroreceptor afferents and, by extension incomplete baroreceptor resetting, may indeed cause hypertension.

Why SAD does not cause hypertension remains unexplained. SNA and the activity of the C1 PSNs are greatly elevated immediately after SAD in rats but return to control within a few days¹⁹. Accordingly, whatever mechanism restores SNA and BP to control after SAD likely resides within or upstream from the RVLM (NTS, CVLM). This adaptation could result from some form of intrinsic neuron homeostasis⁸⁰, enhanced synaptic inhibition driven by sensory afferents other than the arterial baroreceptors or an enhancement of the cerebral blood flow-mediated regulation of RVLM neuronal activity⁸¹. Interestingly, chronic baroreceptor stimulation produces a reduction in SNA and BP that persists for weeks with very little adaptation⁶. This remarkable property justifies the use of arterial baroreceptor stimulation to chronically reduce BP in individuals with drug-resistant hypertension⁸².

Carotid body hyperactivity is another trigger of hypertension. These sensory organs are activated by arterial hypoxia in a pH-dependent manner; they also respond to hypoglycemia, temperature, hormones (angiotensin, leptin) and low blood flow^83–86. Carotid body sensory afferents innervate the NTS via the glossopharyngeal nerve and respond primarily to ATP and acetylcholine, transmitters released by the oxygen sensing Type-I glomus cells. Acutely, carotid body stimulation can activate RVLM PSNs via at least four mechanisms whose relative importance to neurogenic hypertension is unsettled: direct excitatory inputs from the NTS to the RVLM, enhanced cardiorespiratory coupling, arterial baroreflex down-regulation and, if the stimulus is intense enough, general arousal^87–89. The carotid bodies contribute to the elevated BP of SH rats⁷⁶, leptin-treated mice⁹⁰, 2K-1C hypertensive rat⁹¹, rats subjected to a hypercaloric diet⁹² or rodents exposed to chronic intermittent hypoxia (CIH)^{93, 94}. In rodents, the arterial baroreflex is down-regulated when the carotid bodies are activated⁷⁶. Carotid body denervation in juvenile SH rats also delays the development of hypertension^{76, 95}. In every model, the principal evidence is that bilateral carotid body excision or denervation attenuates the hypertension.

In several of these models (SH rats, CIH) carotid body afferents develop a heightened responsiveness to stimuli (cyanide, hypoxia) and “tonicity”, defined as an increased discharge at rest^96–98. In the SH rat, this hyperexcitability is attributed to the upregulation of channels expressed by carotid body afferents such as ASIC3, TASK1 (Kcnk3) or P2X3^{76, 96, 99}. In CIH, it is attributed to excessive ROS production by type-I glomus cells¹⁰⁰. In mice, hyperleptinemia activates the carotid bodies by activating leptin receptors that are coupled to Trpm7 channels⁹⁰. Carotid body hyperplasia may also contribute to the outsized influence of these organs on the sympathetic outflow in hypertension^{85, 101, 102}. Finally, the carotid bodies receive a dense sympathetic innervation from the superior cervical ganglion. CIH hypertension is greatly attenuated by both sympathectomy and angiotensin II (AngII) receptor blockade^{85, 93} suggesting that carotid body hyperactivity could be driven by SNA in a feed-forward loop involving AngII, blood flow restriction via catecholamine-mediated vasoconstriction or a proinflammatory action⁸⁵.

Unilateral carotid body ablation produced transient BP reductions in a few individuals with essential hypertension¹⁰³. The potentially deleterious effects of bilateral glomectomy (loss of hypoxic ventilatory reflex, hypoventilation during sleep, breathing difficulties at altitude including in commercial air planes, effects during and after general anesthesia)⁸⁵ and the possibility of surgical mishaps reduce the translational potential of this intervention but a pharmacological approach based on the use of P2X3 receptor antagonists is promising⁹⁶.

The hypothalamic paraventricular nucleus (PVN)

Besides the circulation, the parvocellular portion of the PVN controls food intake, appetitive responses to sodium deficiency, gastric, pancreatic and esophageal function, glucose counter-regulation, ventilation, the protection of the cornea and various mucosal tissues and the regulation of cerebral blood flow and possibly body temperature^{39, 104, 105}. Based on their projections, PVN neurons have the potential to elicit cardiovascular stimulation (e.g. via their direct projections to the SPGNs, the caudal pressor area, the C1 neurons), or depression (via sympathoinhibition and parasympathetic bradycardia)^{39, 106} or the differential regulation of regional SNA¹⁰⁷. The PVN undoubtedly regulates SNA and BP but we still do not know whether a subset of PVN neurons is dedicated to regulating the cardiovascular system or whether every PVN neuron influences the cardiovascular system in a unique manner that is best suited to the particular behavior to which this subset of neurons contribute (feeding, drinking, arousal, etc.).

The PVN has been implicated in countless hypertension models and in other conditions associated with increased SNA such as heart failure¹⁰⁸. The following section focuses on the SH rat. The contribution of PVN to obesity or salt-dependent hypertension is examined later.

Silencing PVN neurons produces a larger BP drop in the SH rat than in Wistar normotensive controls, anesthetized or conscious^{109, 110} suggesting that PVN neurons could be “hyperactive”. This interpretation is subject to similar caveats as those evoked for the RVLM. PVN neuron discharge rate was not directly monitored, PVN is a very small nucleus that cannot be selectively manipulated except in mice lines in which Cre recombinase is restricted to this nucleus¹¹¹ or to subsets of its neurons¹¹². The BP reduction evoked by inhibiting PVN in a hypertensive animal could be larger, not because these neurons are more active, but because of increased vascular reactivity to SNA, increased excitability of the downstream circuits engaged by PVN stimulation, increased or decreased expression of secondary transmitters by PVN neurons, etc. Finally, assuming that PVN activity is indeed elevated, the cause is still speculative e.g. enhanced input from orexin neurons, the VLM, the carotid bodies, direct effects of local inflammation, neuroplasticity and perhaps a difference in gut microbiota^{32, 108, 113, 114}.

The arcuate nucleus (ARC) and obesity-related hypertension

Obesity and a high-fat diet cause a mild hypertension typically associated with elevated muscle SNA in humans and animals^{115, 116}. Hyperleptinemia, hyperinsulinemia, AngII, or even hypoxia caused by hypoventilation are suspected triggers.

The ARC is the primary CNS target of leptin and insulin, blood-borne hormones that control appetite and weight gain (Figure 3)¹¹⁷. ARC neurons elicit behaviors and sensations required for energy homeostasis (appetitive, hedonic, motor), and regulate metabolism and BP via the autonomic nervous system¹¹⁷. Global activation of the ARC can raise or lower BP via changes in SNA and activates brown adipose tissue (BAT) SNA^{118, 119}. ARC lightly innervates the SPGNs and the RVLM^120–122 but it probably exerts its effects on SNA predominantly via hypothalamic relays (paraventricular, dorsomedial, and ventromedial nuclei)^{121, 123, 124}.

Figure 3:

leptin and insulin raise BP by activating POMC neurons and inhibiting AgRP neurons in the arcuate nucleus. POMC neurons activate PVN neurons by releasing α-MSH. AgRP neurons inhibit PVN by releasing NPY. Leptin also raises BP by stimulating the carotid bodies. Dashed lines, minor projections of POMC neurons plausibly involved in SNS control.

Microinjection of leptin into the ARC increases BP and SNA to the kidneys, the lumbar chain, and thermogenic BAT^{117, 125}. Deletion of the leptin receptor from the ARC eliminates the effect of systemic leptin administration on BP but leptin may also act elsewhere in the hypothalamus¹²⁶. Insulin, on the other hand, probably acts solely in the ARC to increase SNA^{127, 128}. Leptin⁹⁰, contrary to insulin¹²⁹, can also increase SNA by activating the carotid bodies (Figure 3). After binding to receptors in the ARC, leptin and insulin regulate SNA via projections from both POMC and AgRP neurons to the PVN (Figure 3). The POMC neurons release α–melanocyte-stimulating hormone (α-MSH) that binds to MC4R and AgRP neurons release neuropeptide Y (NPY) that binds to Y1R. In short, leptin and insulin cause sympathoexcitation by enhancing the excitatory effect of POMC neurons and by reducing the inhibitory effect of NPY neurons in the PVN¹³⁰.

Hypothalamic sites shown to support basal SNA in diet-induced obese animals include the ARC via leptin¹²⁶ and insulin¹³¹ stimulation, the PVN via α-MSH and glutamatergic stimulation¹³¹ and via loss of tonic NPY inhibition¹³², and the VMH, driven by leptin and α-MSH¹³³. Actions of leptin in the DMH may also support increased BP in mice¹³⁴ but this point is controversial^{131, 133}.

Leptin has received much attention as the link between increased adiposity and sympathoexcitation and hypertension, yet obese Zucker rats, lacking a functional leptin receptor, also exhibit elevated SNA¹³⁵. Excessive but leptin-independent activation of MC3/4R may underlie the hypertension¹³⁶ and CNS mediators other than leptin also contribute to hypertension in this obesity model. Candidates include AngII in the RVLM⁵¹ and orexin in the PVN¹³⁷.

ARC neurons can generate complex patterns of autonomic responses, including a rise in BP, depending on which cell type is recruited. Activation of ARC POMC neurons by leptin and insulin increases BP and SNA to resistance vessels¹¹⁷. This effect is mediated by the release of α-MSH in the PVN and perhaps elsewhere (VMH). One may surmise that in lean and healthy animals, the BP changes elicited by ARC neurons are transient and adapted to periods of food seeking and consumption. Overfeeding may exacerbate and prolong the sympathetic stimulation elicited by ARC under the combined effect of hyperleptinemia, hyperinsulinemia and local tissue inflammation¹³⁸.

Orexin and hypertension

The orexin neurons reside in the perifornical and lateral hypothalamic region; they are implicated in brain functions as diverse as the control of food intake and energy expenditure, breathing and blood pressure, motivation, circadian rhythm of sleep and wake, memory, cognitive functions, and the cardiovascular, thermoregulatory and respiratory effects of various stresses (including transient hypercapnia)^139–143. The orexin neurons are active during waking, when postural muscle tone is high, less active during quiet waking in the absence of movement, and virtually cease firing during sleep¹⁴⁴. These neurons are probably further activated by any form of stress including hypercapnia or hypoxia, and they clearly activate breathing and BP^{143, 145, 146}.

The orexin system contributes to hypertension in the SH rat, the Schlager BPH/2J mouse and the obese Zucker rat^{114, 145, 147}. The principal evidence is that systemic administration of a broad-spectrum orexin receptor antagonist, almorexant, reduces BP and plasma catecholamines^147–149. This evidence is predicated on two assumptions: the selectivity of almorexant and the belief that orexin does not have pertinent peripheral effects. Also, the brain of the SH rat contains greater numbers of detectably orexin-immunoreactive neurons, increased orexin-A mRNA and denser brainstem orexinergic projections than normotensive controls^149–151. In addition, the RVLM seems hyper-responsive to orexin in this strain¹⁵⁰. Finally, the orexin system may also contribute to hypertension in the Dahl salt-sensitive (Dahl-S) rat by enhancing vasopressin synthesis in the PVN^{114, 152}.

The SH rat and the BPH/2J mice are also models of behavioral hyperactivity and orexin neuron firing is highly correlated with locomotor activity¹⁴⁴. Orexin neurons regulate spontaneous physical activity, non-exercise thermogenesis, the effects of psychological stress and their autonomic correlates (hyperpnea and elevated BP)¹⁵³. The BP of the SH rat and the BPH/2J mice is partially normalized by amygdala lesions^{148, 154, 155} implying that a limbic forebrain dysfunction could be driving the orexin system in these rodents. The autonomic effects of increased orexinergic neuron activity are probably relayed through several hypothalamic and brainstem nuclei^{142, 156}. Almorexant reduces CSF noradrenaline dramatically in rats¹⁵⁷, possibly because this drug nullifies the excitatory effect of orexin on the locus coeruleus, a major waking-promoting structure^{158, 159}. Orexin also excites the dorsal raphe¹⁶⁰. By reducing noradrenaline and serotonin release, orexin receptor antagonists such as almorexant reduce the ability of rodents to wake up from sleep and likely produce sedation^{161, 162}. Accordingly, the sympathoinhibition and hypotension elicited in animal models by orexin receptor antagonists could largely result from a reduction in locomotor activity and, during rest, from interference with wake-promoting systems and circuits (noradrenaline, serotonin)¹⁴⁸.

There is no clear indication that the orexin system contributes to hypertension in humans¹⁶³. Orexin-receptor antagonists could conceivably be effective antihypertensive agents in humans but their current therapeutic use is as sedative-hypnotic drugs¹⁶⁴. From past experience, sympatholytic drugs with notable sedative effects (e.g. alpha-2 adrenergic receptor agonists) have fared poorly as anti-hypertensive agents.

The OVLT-PVN connection and hypertension

The lamina terminalis (anterior wall of the 3^rd ventricle) includes the median preoptic nucleus (MnPO) and two circumventricular organs, the subfornical organ (SFO) and the organum vasculosum lamina terminalis (OVLT)^{24, 165, 166}. SFO and OVLT detect circulating and brain AngII, [Na⁺], osmolality and cytokines via receptors expressed by principal neurons, astrocytes and ependymal cells (Figure 4)^167–169. The SFO and OVLT have significant reciprocal connections with the MnPO which serves as an integrative hub for signaling from multiple subcortical structures that regulate neuroendocrine and autonomic function and behavior¹⁶⁵. The interoceptive function of the SFO and OVLT, like that of the arcuate nucleus¹⁷⁰, is regulated by synaptic inputs from regions implicated in drinking and food consumption, thermoregulation, circadian and diurnal patterns, and cardiovascular reflexes^{166, 171}. Sodium sensing by the OVLT and SFO is attributed to several mechanisms. Na(x), a sodium channel encoded by the Scn7a gene and expressed by astrocytes and ependymal cells may be the primary sensing mechanism (Figure 4)¹⁶⁷. Other candidates include an N-terminal variant of the transient receptor potential cation channel subfamily V member 1 (TRPV1) and the epithelial sodium channel (ENaC)⁷.

Figure 4:

Selected nuclei and pathways that contribute to salt or angiotensin-induced hypertension. Inset: mechanism of action of Na in the OVLT (after¹⁶⁷). Sodium present in the CSF and ECF binds to and opens Na(x) channels, elicits the release of protons from astrocytes and ependymal cells. [H⁺] activates OVLT neurons via acid-sensing ion channel-1a (ASIC). Other types of sodium sensing mechanisms have also been proposed⁷ . Other abbrs: see list.

Components of the lamina terminalis (SFO or OVLT or MnPO) contribute to BP elevation in several hypertension models: DOCA-salt¹⁷², AngII infusion¹⁶⁸, chronic intermittent hypoxia⁵⁹ and high-salt diet^{7, 167}. The sympathetic component of the hypertension probably occurs via an MnPO-PVN connection and PVN projections to SPGNs (direct and indirect via the RVLM; Figure 4)^{7, 39, 59, 167, 173}.

Other brain regions potentially involved in hypertension

SNA is regulated at some level by the entire brain¹⁷⁴. Important regions that are little explored from a hypertension perspective include the periaqueductal gray matter, the dorsal colliculi and the cerebellum, dorsomedial hypothalamic nucleus (DMH) and the midline medulla^175–178. The DMH relays the cardiovascular response to stress¹⁷⁹ and could therefore play a role in stress-induced hypertension. Stimulation of the ventral periaqueductal gray matter may relieve hypertension associated with chronic pain in humans¹⁸⁰. The midline medulla oblongata contains the majority of the PSNs identified using the retrograde transport of pseudorabies virus^{30, 181} and regulates cutaneous flow, the cardiac output and the sympathetic innervation of the brown adipose fat in the context of thermoregulation and body weight homeostasis^{182, 183}.

Hypertension: role of brainstem hypoperfusion and hypoxia

Recent work has reinvigorated the theory that hypertension could be an adaptive mechanism to maintain cerebral blood flow when cerebrovascular resistance increases^{81, 184–186}. A relatively moderate (7–20 mmHg) rise in cerebral pressure increases BP and SNA in unanesthetized sheep and humans^{186, 187}. In anesthetized rats, the BP rise requires the exocytotic release of ATP by RVLM astrocytes⁸¹ and ATP may directly activate the PSNs (Figure 5)¹⁸⁸. This enticing theory is predicated on the assumption that there is no off-target expression of dnSNARE (in neurons, vessels, etc.) when adeno-associated vectors engineered with an artificial GFAP promoter are used to transduce astrocytes⁸¹. Also, whether this glial mechanism can cause hypertension or merely mediates the transient effects of severe brainstem hypoxia or brain hypoperfusion on BP merits additional scrutiny. Finally, BP also rises when blood flow is reduced through the NTS only¹⁸⁹, therefore astrocytes may contribute to SNA regulation there too.

Figure 5:

BP regulation via arterial baroreceptors and brainstem perfusion: a hypothesis. A reduction in cerebral blood flow within the RVLM (box) is detected by astrocytes via hypoxia, acidification or end-feet mechanosensitivity elicited by changes in vascular diameter, the latter symbolized by double arrows. The flow-mediated reflex (slow pathway) and the conventional baroreflex (fast pathway, at right of diagram) converge on RVLM PSNs.

The physiological variable detected by astrocytes when cerebral blood flow is limited could be hypoxia, acidification, mechanical stretch of astrocytic end-feet or chemical signals from the vasculature^{43, 81}. In the absence of baroreceptors, this CBF-dependent pathway could conceivably mediate BP homeostasis¹⁹⁰.

This astrocyte-mediated homeostatic mechanism evidently does not prevent the substantial hypotension elicited by chronic baroreceptor activation (>25 mmHg; 3 weeks)⁶. Maybe, baroreceptor activation elicits a countervailing cerebrovascular response that maintains brainstem blood flow despite the hypotension. Alternately, the capacity of astrocytes to depolarize the RVLM PSNs may be limited and may be overridden by the powerful GABA-mediated hyperpolarization elicited by baroreceptor activation (Figure 5). Also unknown is whether excessive flow through the medulla produces the opposite effect, namely sympathoinhibition, and whether this could explain the return to normotension following baroreceptor denervation¹⁹.

Conclusions

Myriad brain regions and sensory afferents have now been implicated in various forms of neurogenic hypertension. Progress is constrained by the difficulty to record SNA or brain neurons in conscious animals for long periods and to compare the data between animals. Progress is also limited by our imperfect knowledge of the network that controls SNA and BP. Powerful methods to interrogate the CNS connectome have been recently developed¹⁹¹. They should be more intensively applied to the neural control of BP.

Neurogenic hypertension is typically, and plausibly, attributed to an increase in the activity of RVLM PSNs. However, the supportive evidence has largely consisted of showing that silencing the RVLM (with drugs or vector-delivered actuators), modifying the RVLM redox state or manipulating the RVLM glia produces a larger BP drop in hypertensive rodents. What is actually being measured by these manipulations is the “neurogenic pressor potential” of this brain region, which depends not only on the discharge rate of RVLM neurons but also on a long series of integrative processes that occur between these neurons and vascular smooth muscle and cardiac contraction¹⁹². The primacy of the RVLM in BP control is a notion that derives primarily from experiments conducted under anesthesia when the PSNs are disinhibited and sympathetic tone is extremely elevated. SPGNs receive direct input from many sources besides the RVLM e.g. spinal cord, hypothalamus and several brainstem regions¹⁹³. Most likely, every type of PSN contributes to the differential regulation of regional SNA, albeit in distinct physiological contexts. Brainstem regions such as the NTS and CVLM are clearly as important as the RVLM to BP control and deserve far more attention in the context of hypertension.

Lesions that attenuate hypertension typically enhance the baroreflex (carotid bodies or PVN in SH rats, renal nerves in the 2K-1C model)^{76, 109, 194} suggesting that baroreflex down-regulation could be a major contributing factor to hypertension. This calls for further investigations of the role of the CVLM and the NTS in hypertension.

The SH rats have been the subjects of most studies. Countless factors are described as playing a “critical” role in the elevated BP of the SH rat (orexin: −33mmHg¹⁵⁷, carotid bodies: −17 mmHg⁹⁵, PVN: −26 mmHg¹⁰⁹ or −61 mmHg¹¹⁰; intestinal dysbiosis, −38 mmHg¹¹³, amygdala −15 mmHg¹⁵⁵, RVLM: −40 mmHg¹¹). This is paradoxical. The homeostasis principle would predict that large and, especially, persistent hypotension should require interfering with multiple pathways simultaneously. Perhaps, at rest, the orexin system, PVN, carotid bodies, amygdala etc. contribute to a roughly comparable extent to the excitatory drive of SPGNs and most of the input summation occurs below action potential threshold in these neurons. If so, the removal of any one of the various inputs could dramatically reduce their firing rate, giving the impression that each is “essential”.

Carotid body hyperactivity contributes to several forms of hypertension but is generally not the predominant factor except when the hypertension results from chronic intermittent hypoxia⁹⁴. This type of hypertension is also eliminated by adrenal demedullation, sympathetic blockade, AT1 receptor blockade and lesions of the AV3V region^{59, 93}. It is difficult to conceptualize why each of these manipulations would eliminate rather than merely attenuate the effect of carotid body hyperactivity.

SNA elevation and neurogenic hypertension can be caused by excessive levels of circulating hormones (leptin, insulin, AngII). Leptin raises blood vessel SNA by activating the carotid bodies and subsets of ARC neurons but other sites of action may exist¹⁹⁵. AngII works in the lamina terminalis and probably throughout the brainstem and hypothalamus via its proinflammatory activity and by reducing the blood brain barrier. Its neurogenic action probably also includes the facilitation of noradrenaline release by sympathetic postganglionic neurons. Sodium works in part via astrocytes and ependymal cells located in the circumventricular organs (OVLT, SFO).

One would also like to know which of the many detected abnormalities have a common origin. Perhaps, all organs with a high metabolic demand (brain, kidneys and the carotid bodies), regulate their blood flow at the expense of systemic BP by eliciting sympathoexcitatory reflexes and the root cause of these neural reflexes is a cerebral or peripheral vascular abnormality. Is hypoxia, oxidative stress or inflammation the trigger in both carotid bodies and brainstem? Are the astrocytes located in the RVLM fundamentally different from those located elsewhere in the brainstem or from type II glomus cells? Finally, the SH rat also highlights the importance of environmental factors (gut microbiome) in hypertension even in this most genetic of models.

Abbreviations:

A1, A2, A5

pontomedullary noradrenergic nuclei

AgRP

agouti-related peptide

AngII

angiotensin II

ARC

arcuate nucleus

AV3V region

see lamina terminalis

BP

arterial blood pressure

C1

RVLM adrenergic/glutamatergic neuron

CBF

cerebral blood flow

CIH

chronic intermittent hypoxia

CSF

cerebrospinal fluid

CVLM

caudal ventrolateral medulla

DMH

dorsomedial hypothalamic nucleus

DOCA

di-hydroxy corticosterone acetate

ECF

extracellular fluid

2K-1C

two-kidney one clip

LC

locus coeruleus

MCR

melanocortin receptor

MnPO

median preoptic nucleus

αMSH

melanocyte-stimulating hormone

Na(x)

sodium channel encoded by Scn7a

NPY

neuropeptide Y

NTS

solitary tract nucleus

Orx

orexin

OSA

obstructive sleep apnea

OVLT

organum vasculosum lamina terminalis

PAG

periaqueductal gray matter

PBL

lateral parabrachial nucleus

POMC

proopiomelanocortin

PSN

presympathetic neuron

PVN

paraventricular nucleus of hypothalamus

ROS

radical oxygen species

RPa

raphe pallidus

RVLM

rostral ventrolateral medulla

SAD

sinoaortic denervation

SFO

subfornical organ

SGN

sympathetic (post) ganglionic neuron

SH rat

spontaneously hypertensive rat

SNA

sympathetic nerve activity

SNS

sympathetic nervous system

SPGN

sympathetic pre-ganglionic neuron

VMH

ventromedial hypothalamic nucleus