Parasympathetic innervation of vertebrobasilar arteries: is this a potential clinical target?

Abstract

This review aims to summarise the contemporary evidence for the presence and function of the parasympathetic innervation of the cerebral circulation with emphasis on the vertebral and basilar arteries (the posterior cerebral circulation). We consider whether the parasympathetic innervation of blood vessels could be used as a means to increase cerebral blood flow. This may have clinical implications for pathologies associated with cerebral hypoperfusion such as stroke, dementia and hypertension. Relative to the anterior cerebral circulation little is known of the origins and neurochemical phenotypes of the parasympathetic innervation of the vertebrobasilar arteries. These vessels normally provide blood flow to the brainstem and cerebellum but can, via the Circle of Willis upon stenosis of the internal carotid arteries, supply blood to the anterior cerebral circulation too. We review the multiple types of parasympathetic fibres and their distinct transmitter mechanisms and how these vary with age, disease and species. We highlight the importance of parasympathetic fibres for mediating the vasodilatory response to sympathetic activation. Current trials are investigating the possibility of electrically stimulating the postganglionic parasympathetic ganglia to improve cerebal blood flow to reduce the penumbra following stroke. We conclude that although there are substantial gaps in our understanding of the origins of parasympathetic innervation of the vertebrobasilar arteries, activation of this system under some conditions might bring therapeutic benefits.

Abbreviations

AChE

acetylcholinesterase

ADMA

asymmetric dimethylarginine

BBB

blood–brain barrier

BP

blood pressure

CBF

cerebral blood flow

CGRP

calcitonin gene related peptide

ChAT

choline acetyltransferase

CMG

carotid mini‐ganglia

ECA

external carotid artery

eNOS

endothelial nitric oxide synthase

ICA

internal carotid artery

NACh receptor

nicotinic acetylcholine receptor

NADPH

nicotinamide adenine dinucleotide phosphate

nNOS

neuronal nitric oxide synthase

NO

nitric oxide

NOS

nitric oxide synthase

NPR‐C

natriuretic peptide clearance receptor

NPY

neuropeptide Y

PACAP

pituitary adenylate cyclase‐activating peptide

PHI

peptide histidine isoleucine

PPG

pterygopalatine ganglia

SHR

spontaneously hypertensive rat

SMC

smooth muscle cell

SSN

superior salivatory nucleus

VAChT

vesicular acetylcholine transporter

VIP

vasoactive intestinal peptide

VPAC

VIP receptor

Introduction

Innervation of the cerebral vasculature was described by Willis as early as 1664, but not until advances in histochemical and immunofluorescent techniques and electrophysiological and pharmacological investigations in the 20th century did we begin to learn about its origins and function. Innervation of the anterior cerebral circulation, which supplies the cerebral hemispheres and delivers the majority of blood to the brain, has been described in detail in several reviews (Purves, 1972; Dahl, 1973; Bevan et al. 1981; Edvinsson, 1987; Bleys & Cowen, 2001; Gulbenkian et al. 2001; Lee et al. 2001; Hamel, 2006; Owman, 2011; Goadsby, 2013). Our aim is to provide an up to date account of the parasympathetic innervation of the posterior vessels (e.g. vertebral and basilar), which have received far less attention and whose normal physiological function is still largely unknown.

Clinically, there is a strong interest in cerebral blood flow regulation because of the devastating personal and socio‐economic consequences of long term hypoperfusion and stroke. As the brainstem controls the most essential brain functions for life, the consequences of reduced blood flow in this part of the brain are devastating. Thus, understanding the control of blood flow to the posterior circulation of the brain is critically important and any vasodilating mechanism highly relevant clinically as it may provide a novel therapeutic target. In the context of ischaemic stroke, the parasympathetic innervation to the vertebral and basilar arteries that supply blood flow to the brainstem (and cerebellum) are of great interest and have not been reviewed systematically.

Gross anatomy of cerebral arteries

The brain is one of the most metabolically demanding tissues in the body and it is not surprising therefore that the main cerebral arteries possess several specialisations that make them different from other arteries in the body enabling them to meet the high demand on blood flow. Their organisation is unique such that multiple feed arteries are interconnected with anastomoses forming a ring‐like structure, the Circle of Willis (Fig. 1). The unique blood vessel organisation of the Circle of Willis acts as a distribution centre permitting blood to flow in any direction to meet increased demand and overcome stenosis. However, under normal circumstances blood flow distribution is normally well demarcated with little mixing from the different feeder arteries (McDonald & Potter, 1951). Thus, in the human and rat cerebral blood flow is mainly provided by the internal carotid arteries (ICAs) whereas the vertebral and spinal arteries supply the brainstem (see Fig. 2).

Figure 1. Human angiogram and schematic diagram of cerebral circulation .

A, CT scan showing the anatomy of the cerebral arterial supply in humans. The image was kindly provided by Dr Nathan Manghat from University Hospital Bristol Trust. B, labelled schematic representation of A including origins of the circulation from the aortic arch. Figure modified from Schuenke et al. (2010). Abbreviations: A, anterior; R, right; P, posterior; L, left.

Figure 2. Approximate spatial contribution to cerebral blood supply by carotid and vertebral arteries in different species .

Schematic representation of the cerebral blood supply contribution by the carotid and vertebral arteries in man (A), rat (B), goat/sheep (C), rabbit (D), dog (E), cat (F) and calf (G). Image based on and adapted from Baldwin & Bell (1963) and Bralet et al. (1977).

The Circle of Willis is formed from several interconnecting arteries, including the posterior cerebral artery, posterior communicating artery, anterior cerebral artery (ACA) and anterior communicating artery; the latter does not exist in rats as the ACAs fuse to form the azygos anterior cerebral artery (Scremin, 2004). There is considerable variation to this organisation both inter‐ (Daniel et al. 1953; McFarland et al. 1979) and intra‐species (Alpers et al. 1959; Brown, 1966; Frąckowiak & Jakubowski, 2008), as seen in Fig. 3. In humans, monkeys and rodents the ICAs supply blood to the Circle of Willis with minimal contribution from the external carotid artery (ECA). In most other mammals including cats, dogs, pigs and sheep, a large proportion of brain blood flow originates from the external carotid artery via a rete mirabili before joining the Circle of Willis with either no or a small contribution from the internal carotids (Daniel et al. 1953). In humans, primates, rabbits and rats there is no rete. A rete mirabili is a complex of arteries and veins in close proximity and built on a counter‐current principle, which creates a gradient that enables exchange of heat, ions or gas between the vessels. The function of the rete mirabili varies between species and is controversial in others (Barnett & Marsden, 1961). It has been suggested that it provides a means of regulating the temperature of blood flowing to the brain in dogs, cats, sheep and goats (Baker, 1982). In the dolphin an extensive rete provides a means of dampening blood pressure in the brain (Nagel et al. 1968). A similar function has been attributed to the rete mirabile in the giraffe (Lawrence & Rewell, 1948).

Figure 3. Species variation in cerebral artery architecture .

Schematic representation of the cerebral arterial tree and difference in contribution to anterior and posterior circulation by the external carotid artery (ECA) and internal carotid artery (ICA) in rats (A), goat/sheep (B) and rabbit (C). Notice the presence of the carotid rete in goat/sheep (B, red arrow), the lack of ICA and the existence of a V‐Oa in the same species. CoW and VA labelled for orientation purposes. Abbreviations: APAr, ascending pharangeal artery residual; BA, basilar artery; CCA, common carotid artery; CoW, Circle of Willis; SA, spinal artery; VA, vertebral artery; V‐Oa, vertebral‐occipital anastomoses. Figure adapted from Daniel et al. (1953), Andersson & Jewell (1956) and Baldwin & Bell (1963).

A unique feature of cerebral arteries is that they are both conduit and resistance vessels and unlike other arteries constrict and dilate (Faraci & Heistad, 1990). Constriction can increase arterial pressure by ∼40% in human (Dickinson & Paton, 2012) and rat (Baumbach & Heistad, 1983). This ability protects the brain from sudden rises in blood pressure (BP) to prevent stroke (Bill & Linder, 1976; Edvinsson et al. 1976 a; Beausang‐Linder & Bill, 1981; Sadoshima et al. 1981), whereas vasodilatation can protect against ischaemia (Santizo et al. 2000; Levi et al. 2012). During constriction maintenance of blood flow to the brain then becomes dependent on compensatory vasodilatation in parenchymal arterioles (Baumbach & Heistad, 1983).

Neural innervation of major cerebral vessels

There are two sources of innervation controlling blood flow to cerebral vessels: extrinsic and intrinsic (Hamel, 2006). ‘Intrinsic’ innervation originates from local neurons within the CNS targeting arterioles to control blood flow in the brain parenchyma. Our focus is on the ‘extrinsic’ innervation from peripheral ganglia that innervate all major cerebral arteries up to the Virchow–Robin space, that is, before they enter the brain parenchyma. Virchow–Robin spaces are perivascular spaces or canals surrounding arteries and veins at the point where they leave the subarachnoid on the surface of brain and turn into the parenchyma. See figure 1 in Hamel (2006).

Extrinsic innervation consists of the sensory system and efferent motor fibres from the autonomic (sympathetic and parasympathetic) nervous system. Numerous proteins have been associated with functionally different subsets of nerve fibres that can be detected using immunocytochemistry. Examples relevant to the cerebral circulation include: acetylcholinesterase (AChE), vesicular acetylcholine transporter (VAChT) and vasoactive intestinal peptide (VIP) within parasympathetic fibres; tyrosine hydroxylase, dopamine β‐hydroxylase, and neuropeptide Y (NPY) in sympathetic nerves; and calcitonin gene related peptide (CGRP) as a marker of sensory afferents. In basilar arteries from human cadaveric fetuses both noradrenergic and AChE containing fibres first appeared around the 12th gestational week whereas NPY and VIP synchronously appeared a month later at the 16th gestational week (Kawamura & Takebayashi, 1994). Interestingly, with ageing, there was a decrease in the expression of vasoconstrictor neurotransmitters (e.g. tyrosine hydroxylase) in cerebrovascular nerves, whereas the expression of vasodilator neurotransmitters (VIP and CGRP) in perivascular nerve fibres supplying the rat cerebral arteries increased with age (Mione et al. 1988).

Stimulation of sympathetic nerves is generally considered to result in vasoconstriction (D'Alecy & Feigl, 1972; Lee et al. 1976; Baumbach & Heistad, 1983), whereas stimulation of parasympathetic nerves is vasodilatory (D'Alecy & Rose, 1977; Talman et al. 2007). As we shall see later, there are exceptions to this rule. Sensory neurons are a heterogeneous population of either fast transmitting Aδ fibres (being purely afferent) or slowly conducting capsaicin sensitive fibres expressing a range of co‐transmitters. Exceptionally, these fibres can also affect cerebrovascular tone due to their additional efferent role; these fibres are referred to as ‘sensory‐motor nerves’ (Rubino & Burnstock, 1996) and contain, for example, substance P and CGRP, a very potent vasodilator.

Extrinsic nerve plexuses and bundles are located on the periphery of the adventitia, mostly run in the longitudinal direction along the vessel and have a smooth (‘string‐like’) appearance, whereas single fibres and terminals tend to be located in the deeper layers of the adventitia, running circumferential/perpendicular to the vessel and close to the smooth muscle cells (SMCs) (Bleys et al. 1996 a; Taktakishvili et al. 2010). The fine terminal nerve fibres are characterised by varicosities and have a ‘pearl‐on‐a string’ appearance (Yoshida et al. 1993). These structures contain the highest vesicle numbers and neurotransmitter contents. Autonomic and sensory nerve fibres can be in close proximity to smooth muscle cells (80–100 nm) and to each other (Iwayama et al. 1970; Matsuyama et al. 1985). However, on basilar arteries close contacts between varicosities and smooth muscle cells are sparse and distances of 1–2 μm are typical (Luff & McLachlan, 1989; Lee et al. 1990). The functional significance of this is unclear but it may reduce efficacy of transmission in these structures.

The parasympathetic innervation of posterior cerebral vessels

Cerebral arteries are richly innervated with neuronal fibres of parasympathetic origin providing a most powerful vasodilatatory mechanism (Goadsby, 2013). Parasympathetic fibres follow the sympathetic nerves along cerebral vessels. In comparison to peripheral arteries (femoral and common carotid), where parasympathetic innervation is negligible, it is generally 10–40 times higher in cerebral arteries (Duckles, 1981). The density of innervation is both species and cerebral artery location dependent. Using VIP immunopositive fibres as a marker for parasympathetic innervation, relatively low densities have been observed in mouse, rabbit and monkey, with intermediate levels in dog and guinea pig, whereas pig, rat and bat exhibited high levels (Edvinsson et al. 1980; Ando, 1988). In general, fibre densities are greater in the rostral part of the cerebral circulation than in the caudal (Edvinsson et al. 1980; Kobayashi et al. 1983; Hara & Weir, 1986; Hara & Kobayashi, 1987) (see figure 4 b). The exception is the bat, demonstrating dense VIP positive innervation of the vertebrobasilar arteries (Ando, 1988). In some animals (dogs, rabbits), either no or low parasympathetic innervation of the vertebrobasilar arteries has been reported (Florence & Bevan, 1979; Edvinsson et al. 1980; Saito et al. 1985). However, dogs have nitric oxide synthase (NOS) positive fibre staining on vertebrobasilar arteries (Yoshida et al. 1993) that was shown not to be sensory, suggesting the presence of some ‘parasympathetic’ fibres.

Figure 4. Parasympathetic innervation to cerebral vasculature .

A, schematic overview of the anatomic position of various parasympathetic ganglia that are sources of parasympathetic input to the cerebral arteries. B, the level of parasympathetic innervation to the cerebral arteries varies as shown by VIP innervations in the posterior communicating artery (dense innervation) and basilar artery (moderate innervation); image from Hara et al. (1985); ×172 in original article, reproduced with permission. C, schematic representation of parasympathetic innervations to the rostral and posterior cerebral vasculature. The gradient represents reported density of innervation, cross‐hatched filling indicates unreported innervation. Note that species variation regarding innervation is not depicted in the figure. CmG, carotid mini‐ganglia; CS, cavernous sinus; OG, otic ganglia; PTG; pterygopalatine ganglia; VIP, vasoactive intestinal peptide. Original figure collated from data from Gibbins et al. (1984 a), Hara et al. (1985), Keller et al. (1985), Shimizu (1994), Kadota et al. (1996), Bleys et al. (2001) and Ayajiki et al. (2012).

Types of parasympathetic fibres innervating posterior cerebral vessels

Classically, peripheral parasympathetic nerves are considered to be cholinergic, i.e. using acetylcholine (ACh) as a transmitter. However, it is clear that parasympathetic cells innervating cerebral arteries utilise a host of co‐transmitters in addition or in preference to ACh. Some parasympathetic neurones in the cerebral arteries have lost the ability to produce ACh, thus giving rise to the term ‘non‐cholinergic parasympathetic innervation’ (Minami et al. 1994).

Nitric oxide (NO), VIP and/or pituitary adenylate cyclase‐activating peptide (PACAP) have all been identified as co‐transmitters in parasympathetic neurones and all are potent vasodilators (Bevan et al. 1984; McCulloch et al. 1986; Warren et al. 1991; Nozaki et al. 1993; Uddman et al. 1999) producing greater vasodilatation than ACh itself (Lee, 2000). Reviewing the evidence of vasodilatation in large cerebral arteries, Lee (2002) concluded that the main effector for vasodilatation is NO, with the other neurotransmitters, including ACh, acting as presynaptic modulators of NO release. These findings explained the initial confusion that existed after cholinergic nerves had been identified on cerebral arteries, but were found to be resistant to atropine/physostigmine vasodilatation, leading to the misconception that these nerves were non‐functional (Lee et al. 1978; Lee, 1980, 1982). Thus, parasympathetic neurons innervating the posterior cerebral arteries consist of a heterogeneous population of ACh–VIP–NO‐, ACh–NO‐ and VIP–NO‐positive neurons, but this list is not exhaustive (Minami et al. 1994; Kimura et al. 1997; Yu et al. 1998). It is important to note that NO and PACAP are also co‐expressed in sensory neurones on cerebral arteries (Sundler et al. 1996). Therefore, immunocytochemical labelling of parasympathetic neurones is challenging and requires co‐labelling.

Tract tracing studies of parasympathetic input to the cerebral arteries have employed a variety of antibodies, including cholinergic markers choline acetyltransferase (ChAT) or the vesicular acetylcholine transporter (VAChT), or VIP and the NO producing enzymes NO synthase (NOS) or nicotinamide adenine dinucleotide phosphate (NADPH)‐diaphorase. Earlier immunohistochemical studies also used acetylcholinesterase (AChE), though this enzyme is less specific. AChE degrades ACh in the synaptic cleft. It was one of the first available cholinergic markers; however as it stains may adrenergic and sensory nerves in addition to cholinergic cells it is not exclusive to cholinergic neurones and is now more considered as a pan neuronal marker (Bleys et al. 1996 a, 1996 b).

The relative distribution of the different markers (ChAT, VAChT, NOS and VIP) and proportion of neurones expressing them has not been fully elucidated for the vertebrobasilar arteries and reports vary greatly partly due to species differences. ChAT has proven capricious in the periphery and in fine fibres due to low enzyme levels and hence its use underestimates the total cholinergic innervation (Schäfer et al. 1998; Bleys et al. 2001).

Origin of parasympathetic fibres to the posterior cerebral arteries

The origin of the parasympathetic innervation is characterised by a diffuse collection of small extra‐cranial ganglia and small cell groups in both the arachnoid and at the base of the skull (see figure 4).

The anterior circulation is innervated by fibres from the pterygopalatine ganglia (PPG), also referred to as the sphenopalatine ganglia (SPG), the cavernous sinus, the otic and the internal carotid mini‐ganglia (CMG) (Chorobski & Penfield, 1932; Kobayashi et al. 1983; Carvalho, 1985; Keller et al. 1985; Hara & Weir, 1986; Walters et al. 1986; Hara & Kobayashi, 1987; Suzuki et al. 1988; Edvinsson et al. 1989; Hara et al. 1989; Hardebo et al. 1991; Suzuki & Hardebo, 1991, 1993; Nakai et al. 1993; Bleys et al. 1996 b, 2001; Toda et al. 2000 a,b). The origin of parasympathetic innervation to the vertebrobasilar arteries is less well understood. In monkeys, PPG fibres appeared to be the source of NOS‐positive fibres to the anterior but not posterior cerebral arteries (Ayajiki et al. 2012) and in rats parasympathetic VIP‐ (Hara et al. 1989) and NOS‐positive fibres (Nozaki et al. 1993) were found on caudal basilar and vertebral arteries following PPG extirpation, suggesting additional sources. Injecting the anterograde tracer wheat germ agglutinin‐horseradish peroxidase (WGA‐HRP) into the PPG only stained parasympathetic fibres at the anterior end of the basilar artery (Hara et al. 1993). Similar injections to the otic ganglia did not result in labelling on the vertebrobasilar arteries in rats (Shimizu, 1994). However, using retrograde tract tracing with fluoro‐gold applied to the middle portion of the basilar artery resulted in neurone labelling in both the otic ganglia and PPG (Kadota et al. 1996). Labelling in the PPG was also seen following WGA‐HRP application to the middle and caudal part of the basilar arteries in cats (Keller et al. 1985). Bleys & Cowen (2001) suggested that PPG, otic and CMG innervation is restricted to the anterior end of the basilar artery with no observable innervation to the caudal basilar and vertebral arteries. Gibbins et al. (1984 a) lists otic, PPG, submandibular and sublingual ganglia as possible contributions to rostral basilar artery VIP containing fibres. The inconsistencies may be explained by the differences in techniques and species used. Studies where NOS immunocytochemistry has been performed will not be able to delineate between parasympathetic and sensory nerves. Hence, many of the suggested sources of parasympathetic innervation to the posterior cerebral arteries based on NOS immunochemistry, which included the glossopharygeal and vagal nerves, and upper cervical root ganglia (Suzuki et al. 1988; Hardebo et al. 1991), are all likely to be of sensory origin (Arbab et al. 1986).

Despite evidence for parasympathetic innervation of vertebrobasilar arteries, and the existence of VAChT‐ (E. v. L. Roloff & J. F. R. Paton, personal observation), ChAT‐ and VIP‐positive fibres in rat (Suzuki et al. 1988, 1990 c), cat (Gibbins et al. 1984 b) and dog (Seki et al. 1995), we have found sparse information regarding the source(s) of these specific parasympathetic inputs. This clearly needs further clarification.

Preganglionic parasympathetic neurones destined for the cerebral vasculature

The location of preganglionic neurones projecting to the PPG of rats has been determined by retrograde transneuronal pseudorabies virus labelling (Spencer et al. 1990). Neuronal labelling was obtained in the ipsilateral superior salivatory nucleus (SSN). Pharmacological stimulation of the greater petrosal nerve cell group within the SSN reduces ipsilateral cerebral vascular resistance. The effect can be countered by PPG section but not with systemic muscarinic antagonism; another confirmation that the vasodilatory mechanism (via the PPG) is mediated by a non‐cholinergic mechanism in the cerebral arteries (Nakai et al. 1993). The following nuclei were also labelled and most likely upstream of the SSN: the nucleus tractus solitarii and the dorsomedial part of the spinal trigeminal nucleus and gigantocellular reticular nucleus as well as pontine structures of the parabrachial nucleus, A5 catecholaminergic cells and subcoeruleus region of the pons. Midbrain areas included the lateral, dorsomedial and paraventricular hypothalamic nuclei and lateral preoptic area. The bed nucleus of the stria terminalis, substantia innominata and the amygdalopiriform were also labelled in the forebrain. Given this extensive convergent connectivity it underpins the importance of the vasodilatory role of the SSN and those regions that may influence it. There is scant information regarding sources of input to the other parasympathetic ganglia innervating cerebral arteries. Gibbins et al. (1984) and Gibbins & Morris (1988) reported that the SSN was also a source of input to the otic ganglia and CMG.

Functional role of the parasympathetic transmitters and receptors

The transmitters and receptors are presented in order of importance as vasodilatory mechanisms.

Nitrergic

NO is the main vasodilator in the cerebral arteries (Lee, 2002). It has an estimated diffusion range of up to 300 μm and an estimated half‐life of 5–15 s in the cerebral circulation (Ignarro, 1991; Garthwaite & Boulton, 1995). Nerve fibres with NO production capabilities on cerebral arteries include pure nitrergic, VIPergic and/or cholinergic parasympathetic neurones, as well as sensory neurones (Minami et al. 1994; Kimura et al. 1997; Edvinsson et al. 2001). Note that there are multiple additional sources of NO that could produce vasodilatation in the vertebrobasilar arteries, such as endothelial cells, astrocytes, neurones (Bredt et al. 1990; Faraci & Brian, 1994) and immune cells (Hibbs et al. 1988; Bogdan, 2001). The distribution of NO containing fibres in the cerebral arteries is highly variable in terms of exact location on the cerebral arteries and species. In rats, although present on the basilar artery, the densest NOS/NADPH‐diaphorase‐positive fibre staining occurred on the anterior part of the cerebral circulation, whereas in humans NOS immunoreactivity was predominantly on posterior cerebral arteries, especially the basilar artery (Iadecola et al. 1993; Nozaki et al. 1993; Minami et al. 1994; Edvinsson et al. 2001; Taktakishvili et al. 2010).

Cholinergic

Cholinergic receptors have been located on endothelial cells, smooth muscle cells and perivascular nerves on cerebral vessels (see Table 1). Electrical or pharmacological stimulation of cholinergic fibres on cerebral arteries can, depending on the species, the site of action, receptor subtype and concentration of ACh released, induce either vasodilatation or vasoconstriction (Edvinsson et al. 1977; Duckles, 1981; Miao & Lee, 1991; Dauphin & Mackenzie, 1995; Toda et al. 1997). In cats and dogs, low concentrations of ACh (10⁻⁶ to 10⁻⁵ m) applied to cerebral arteries, including the basilar, caused vasodilatation, whereas higher concentrations result in vasoconstriction (Tsukahara et al. 1986; Dauphin & Hamel, 1992). In humans, however, only vasodilatation was observed (Edvinsson et al. 1976 b; Tsukahara et al. 1986). ACh can exert its effects via either metabotropic muscarinic (M‐) or ionotropic nicotinic (N‐type) receptors.

Table 1.

Cholinergic receptors identified on cerebral arteries

Receptor subtype	Cellular location	Vessel location	Effect	Proportion of total receptors	Species	Reference
M1‐AChR	SMC	MCA	Vasoconstriction	15–27%	Cat	(Dauphin et al. 1991)
				26–54%	Human	(Dauphin & Hamel, 1992)
M2‐AChR	Perivascular nerves		Auto regulatory	35%	Cat and	(Dauphin & Hamel, 1992)
		Absent!! BA		0%	human
	Nitrergic nerves		NO‐modulatory inhibition of NO‐vasoconstriction	—	Pig	(van Charldorp & van Zwieten, 1989; Liu & Lee, 1999)
	Prejunctionally NO‐nerves	MCAa	Attenuate vasodilatation by inhibiting NO release from nitrergic nerves		Monkey	(Toda et al. 1997)
M3‐AChR	Endothelial	MCA	Vasodilatation	10–18%	Cat and human	(Dauphin & Hamel, 1990; Dauphin et al. 1991)
M5‐AChR	unknown, whole vessel analysed for mRNA	BA	not determined	not determined	Rat	(Phillips et al. 1997; Tayebati et al. 2003)
	Endothelium and SMC (less) media	CoWs	Vasodilatation		GM‐mouse	(Yamada et al. 2001 a)
α7‐NAChR				None	Cat and human	(Dauphin & Hamel, 1992)
		BA	Modulation of sympathetic		Pig	(Zhang et al. 1998 a; Si & Lee, 2001)
		BA	perivascular neurones		Rat	(Chang et al. 2012)
α3β2‐NAChR			Modulation of sympathetic perivascular neurones		Rat	(Lee et al. 2011)

^aEndothelial denuded vessel preparation. AChR, acetylcholine receptor; GM, genetically modified; MCA, middle cerebral artery; NAChR, nicotinic acetylcholine receptor.

Muscarinic receptors

Four subtypes of muscarinic and one nicotinic receptor have been found in on cerebral arteries. Using isolated vessel preparations, Furchgott & Zawadzki (1980) demonstrated that the vasodilatory effect of ACh is endothelium dependent. ACh acts on muscarinic receptors expressed on endothelial cells that stimulate endothelial NO synthase (eNOS) activity, releasing NO, which then triggers vasodilatation (Fig. 6 in Wahl & Schilling, 1993).

Figure 6. Mechanisms of endothelial dependent cross‐talk leading to NO induced vasodilatation .

There are multiple sources of nitric oxide production, including the endothelium and nitregic neurones. A, ACh exerts endothelium dependent vasodilatation acting on muscarinic (M) type 3 and/or 5 receptors on endothelial cells inducing an increase in intracellular calcium which leads to NO production via eNOS activity. ACh still evokes a dilatation in denuded vessels suggesting a non‐endothelial source of NO. B, noradrenaline (NA) can stimulate the production of NO from cholinergic/nitrergic neurons via β₂ receptors leading to increases in Ca²⁺ and calcium channel stimulation. The NO diffuses to nearby smooth muscle cells causing relaxation. Modulation of NO release in this system is achieved by ACh‐mediated NO inhibition via prejunctional muscarinic (M) type 2 receptors on nitrergic neurons. Abbreviations: ACh, acetylcholine; cGMP, cyclic guanosine monophosphate; GC, guanylate cyclase; GTP, guanosine triphosphate; NO, nitric oxide; NOS, nitric oxide synthase. Figure based on and adapted from Wahl & Schilling (1993), Zhang et al. (1998 b), Liu et al. (2000) and Si & Lee (2002).

The endogenous origin of ACh for this mechanism is unclear, as it is rapidly broken down by AChE in the blood. Release from perivascular nerves is a possibility, as tritiated ACh applied to the outside of intact vessels readily crossed the media and appeared in the luminal perfusion fluid (Gonzalez et al. 1997). An endothelial origin has also been suggested, as endothelial cells contain ChAT and can produce and release ACh (Parnavelas et al. 1985; Kawashima et al. 1990), but what controls this is unclear.

It is likely that muscarinic M₃ receptors located on endothelial cells mediate the NO induced vasodilatation, at least in rabbits (Garcia‐Villalon et al. 1991), cats (Dauphin & Hamel, 1990) and human (Dauphin & Hamel, 1992). In the absence of endothelium, ACh (exogenously applied or released by transmural stimulation) induces vasoconstriction via M₁ receptors located on the vascular smooth muscle cells (VSM). M₁ receptors have been identified in VSM of cerebral arteries including basilar arteries of rats (Phillips et al. 1997) and in numerous other species including cat (Dauphin et al. 1991), pig (Garcia‐Villalon et al. 1991) and human (Dauphin & Hamel, 1992). Since M₁ receptors have a lower affinity for ACh than M₃ receptors, higher concentrations of ACh are required for their activation and this may provide a mechanism for the differential vasomotor actions of ACh. From a study of pial vessels in mice (Shimizu et al. 1993), there is evidence that the M₁ receptor mechanism is uncoupled/non‐functional in the presence of a functional endothelial M₃ receptor, explaining the apparent dominance of M₃ receptor induced vasodilatation.

Prejunctional M₂ receptors are located on nitrergic perivascular neurones and attenuate the vasodilatation (or increase vasoconstriction) by inhibiting NO production and release. This has been found in several species (pig, cat, monkey) and in various locations along the posterior (and anterior) cerebral arteries (van Charldorp & van Zwieten, 1989; Dauphin & Hamel, 1992; Phillips et al. 1997; Toda et al. 1997; Liu & Lee, 1999). In contrast, there is a total absence of M₂ receptors in human cerebral arteries (Dauphin & Hamel, 1992), explaining the absence of ACh induced contraction in man mentioned above.

The M₅ receptor was the last muscarinic receptor to be identified in cerebral arteries and, as with M₁ and M₃ receptors, it is postsynaptically located. The receptor was found on both smooth muscle cells and endothelium (Phillips et al. 1997; Tayebati et al. 2003). The mRNA content for M₅ in rat basilar artery was found to be equal to that of M₂ and M₃ receptors, and greater than M₁ receptor mRNA (Phillips et al. 1997). Due to the lack of selective M₅ muscarinic receptor ligands, little is known of the functional significance of this receptor on cerebral arteries. To circumvent this problem, Yamada et al. (2001 b) created a M₅ receptor knockout mouse, in which large cerebral arteries and arterioles did not dilate following topical application of acetylcholine (both in in vivo and in organ bath preparation). The effect was specific to the cerebral blood vessels, as peripheral vessels retained the ability to dilate. It has been suggested that some of the effects previously ascribed to the M₃ receptor may be mediated via the M₅ receptor (Elhusseiny & Hamel, 2000).

Nicotinic ACh receptors

There is a dearth of evidence supporting a major role for nicotinic receptors on vertebrobasilar arteries. Although Dauphin & Hamel (1992) did not find nicotinic receptors in cat or human cerebral arteries, Edvinsson et al. (1977) found evidence of the α7‐nicotinic acetylcholine (NACh) receptor on adrenergic nerves of cat middle cerebral arteries. As its function is relevant to modulation of adrenergic signalling, it will be described in the ‘cross‐talk’ section (see below).

Vasoactive intestinal peptide

VIP has been found to co‐localise with ChAT in some parasympathetic nerves and ganglia innervating the cerebral vasculature (Gibbins & Morris, 1988; Suzuki et al. 1990 a; Hardebo et al. 1992; Uddman et al. 1999), including the basilar artery (Saito et al. 1985; Miao & Lee, 1990). In basilar arteries from cats, only 5% co‐localisation between VIP‐ and ChAT‐positive fibres was found despite these running close to each other (Miao & Lee, 1990). The density of VIP innervation of cerebral arteries is species dependent with highest densities in the cat, rat and rabbit (Duckles & Said, 1982; Gibbins et al. 1984 b; Zhang et al. 1991) and relatively sparse in human (Edvinsson et al. 1987, 1994 b). This concurs with a study by White (1987) in which 87% of human basilar arteries pre‐contracted with prostaglandin F_2α (PGF_2α) displayed a sustained concentration dependent vasodilatation when exposed to VIP. Despite the lack of VIP positive fibre staining on vertebral and basilar arteries in dogs (Edvinsson et al. 1980), the vascular conductance of vertebral arteries was increased by 100% on exposure to VIP (Blitz & Charbon, 1983). These data emphasise the importance of this peptide in mediating vasodilatation of vertebrobasilar arteries but the neural source remains to be ratified.

There are two VIP receptors with similar affinity for VIP: VIP₁ (VPAC₁) and VIP₂ (VPAC₂). In addition to these, VIP can also act via the natriuretic peptide clearance receptor (NPR‐C) (Akiho et al. 1995; Murthy et al. 2000; Grant et al. 2005). Immunohistochemical labelling for VPAC₁ in rat cerebral arteries found it on the surface of smooth muscle cells in the medial layer only (Fahrenkrug et al. 2000). This was confirmed using in situ hybridisation mRNA, where VPAC₁ and VPAC₂ located into smooth muscles cells in posterior (and anterior) cerebral vessels (Baun et al. 2011). Similar results have been obtained in humans, using real time polymerase chain reaction (RT‐PCR) against the two VPAC receptors (Knutsson & Edvinsson, 2002). In pig basilar arteries, mRNA is present for VPAC₁, VPAC₂ and NPR‐C. Using immunocytochemistry, VPAC₁ was found in the endothelium, VPAC₂ on the outer layers of SMCs and endothelium and NPR‐C throughout the artery, including the perivascular nerves themselves (Grant et al. 2005).

The vasodilatory effect of VIP may or may not involve a NO dependent mechanism. Denuded cerebral arteries from sheep and cow showed blunted VIP induced vasodilatation, indicating the presence of a mechanism involving NO of endothelial origin (Gaw et al. 1991; Gonzalez et al. 1997). VIP stimulation of isolated rat basilar arteries results in vasodilatation, which is inhibited by the neuronal (n)NOS selective antagonist 7‐nitroindazole (7‐NI), indicating the involvement of NO from perivascular nitrergic nerves (Seebeck et al. 2002). VIP receptors present on SMCs, however, induce vasodilatation directly, via an NO independent mechanism, probably through the cAMP pathway (Grant et al. 2005).

Pituitary adenylate cyclase‐activating peptide

PACAP belongs to the same family of glucagon/secretin neuropeptides as VIP and shares 68% amino acid sequence homology. The two co‐transmitters are expressed in different subpopulations of parasympathetic neurones. In addition, PACAP is expressed by both trigeminal sensory neurones and sympathetic neurons making its association with the parasympathetic nervous system equivocal (Sundler et al. 1996). PACAP mRNA is expressed in most otic and PPG cells in the rat (Sundler et al. 1996). PACAP acts through three different G protein coupled receptors: the VIP receptors VPAC₁ and VPAC₂, and PAC₁ – all increase adenylate cyclase activity. There is evidence of vasodilatory effects of PACAP in rat basilar arteries mediated by nitrergic nerves (Seebeck et al. 2002). Although PACAP is up to 100× more potent as a vasodilator than VIP (Kashimoto et al. 1996), PACAP fibre innervation of cerebral arteries is less dense than that of VIP and NO (Warren et al. 1991; Edvinsson et al. 2001). In pre‐constricted rat basilar arteries a stronger vasodilatory capacity of VIP than PACAP was found, whereas VIP and PACAP acted similarly in the anterior vessel segments, which may indicate a differential receptor distribution between these arteries (Baun et al. 2011).

Peptide histidine isoleucine

VIP‐like neuropeptides such as peptide histidine isoleucine (PHI) in rat and the human equivalent peptide histidine methionine (Linder et al. 1987) are also co‐expressed in the VIP population of parasympathetic fibres on basilar arteries and have weaker vasodilatory properties relative to VIP (Suzuki et al. 1984; Uddman et al. 1993).

Cross‐talk between parasympathetic and sympathetic nerves

Parasympathetic and sympathetic nerves are often intertwined on cerebral arteries (Owman et al. 1974) or run in parallel within the same perineural sheath (Iwayama et al. 1970; Edvinsson et al. 1972 a) (see figure 5). Axo‐axonal contact distances between sympathetic and parasympathetic fibres can be as little as 25 nm, compared to the 100 nm seen for neuromuscular contacts in the same vessels (Edvinsson et al. 1972 a). This close apposition suggests cross‐talk between parasympathetic and sympathetic nerve fibres.

Figure 5. Electron microscopy image showing the close proximity of sympathetic and parasympathetic terminals on a cerebral artery .

Image of the anterior cerebral artery of cat showing the close apposition of a parasympathetic (Ch, cholinergic) and a sympathetic (A, adrenergic) neuron. Schwann cell is visible at the bottom. Image from Edvinsson et al. (1972 b). ×60,000 in original reference, with permission.

Support for interactions between the parasympathetic and sympathetic innervations to cerebral arteries is that they develop simultaneously in both mice and rats (Kobayashi et al. 1981; Tsai et al. 1992) and human (Kawamura & Takebayashi, 1994). In pre‐ and postnatal rats, parasympathetic and sympathetic fibres (visualised with VIP and NPY markers, respectively) show similar patterns of development in their distribution, density and pattern (Tsai et al. 1992). This close association in development was not observed for sensory fibres (visualised using CGRP), their development occurs after sympathetic innervation (Tsai et al. 1989). There is not the same degree of close apposition between sensory fibres with either sympathetic or parasympathetic nerves (Tsai et al. 1989; Edvinsson et al. 1994 a). Functional studies in basilar arteries from rat and pig have found that noradrenaline and ACh have limited vasoreactivity, yet adrenergic nerves express nicotinic α‐7 NACh receptors, which upon stimulation cause increases in intracellular calcium leading to noradrenaline release from their varicosities (Si & Lee, 2001, 2002). This triggers vasodilatation via activation of β₂‐adrenergic receptors; vasoconstriction does not occur as there is paucity of α‐adrenoceptors. β₂‐Adrenoceptors are located on the intertwined acetylcholinergic–nitrergic fibres leading to NOS activation and NO production, and stimulation of guanylyl cyclase (Fig. 6). Removal of the endothelium does not affect the arteries’ ability to dilate, indicating that the source of NO must be neuronal (Lee, 1980; Toda, 1982). Indeed, NO is stored and released with ACh. ACh acts on nitrergic neurones stimulating prejunctional M₂ receptors and inhibits NO synthesis, thus modulating NO release (Liu et al. 2002). Following unilateral excision of the superior cervical ganglion in rats, NO‐mediated vasodilatation in cerebral arteries was reduced, indicating a dependence on an intact sympathetic innervation (Smith et al. 2002).

Cross‐talk may have a role in the development and maintenance of autonomic fibres and their phenotype. For example, after sympathectomy: (i) parasympathetic fibres proliferate and make more direct contacts with vascular smooth muscle cells, whereas prior to sympathectomy they mainly contact sympathetic fibres (Smith et al. 2002); (ii) parasympathetic cells can display increased dopamine‐β‐hydroxylase immunoreactivity (though this is functionally irrelevant), and this is matched by a similar decrease in the number of PPG neurones expressing parasympathetic markers (Mione et al. 1991; Fan & Smith, 1993); (iii) both NOS and VIP activity in the PPG is decreased, resulting in a reduction in NOS dependent vasodilatation in various target areas (Fan & Smith, 1993; Smith et al. 2002).

In the face of increased sympathetic drive as occurring during ‘fight or flight’ situations vertebrobasilar arteries may thus vasodilate (instead of constrict) via the mechanisms described above to ensure the brainstem and or cortex remains well perfused at times when peripheral organs experience vasoconstriction and reduced vascular conductance (Lee et al. 2011).

Other proposed functions of parasympathetic innervation to cerebral arteries in vivo

With such a dense parasympathetic innervation of the cerebral arteries it is a natural question to ask what are the physiological circumstances in which these nerves are activated? Besides the suggested function in ensuring adequate cerebral blood flow (CBF) during periods of high sympathetic activity, what else does it do? The question has not been answered equivocally, though various clues to its function are appearing: as mentioned earlier, activation of the parasympathetic innervation causes vasodilatation. Thus it is likely to be involved in any process requiring centrally regulated increases in CBF that are independent of the endothelial NO system. Examples include increased brain metabolism, thermoregulation (it is essential to keep the brain thermoregulated) and hypercapnia.

Electrical stimulation of the PPG in rats increases cerebral blood flow by 40–50% on the ipsilateral side as measured by Doppler flow over the parietal cortex and tissue $P_{{\mathrm{O}}_2}$ (Seylaz et al. 1988; Talman et al. 2007). In dogs, stimulation of the PPG and the superficial petrosal evoked vasodilatation in the middle cerebral artery, posterior communicating arteries and basilar artery (D'Alecy & Rose, 1977; Toda et al. 2009). Anti‐cholinergic drugs failed to attenuate the response, emphasising that the vasodilatory response is mediated via a non‐cholinergic mechanism (Nakai et al. 1993).

What evidence is there for vasodilatory tone? Following removal of the PPG in the awake rat Suzuki et al. (1990 b) reported no change in blood flow, suggesting an absence of tone. However, these data were not confirmed by Boysen et al. (2009), who saw ablation of the PPG evoking a 30% decrease in blood flow increasing cerebrovascular resistance in rats. Given that the muscarinic antagonist, atropine, does not affect cerebral blood flow, non‐cholinergic transmission must underlie this tone (Toda et al. 1993, 2000 a; Yoshida et al. 1993). Further evidence of tonic parasympathetic tone in the form of reduced blood flow upon NO inhibition has also been observed in monkeys (Toda et al. 2000 b) and dogs (Toda et al. 1993, 2000 a; Yoshida et al. 1993), where the vasodilatory response to nerve stimulation was found to be attenuated by the NO inhibitor N ^G‐nitro‐l‐arginine (l‐NA). Indeed, NO of neuronal origin has been associated with vasodilatation and decreased cerebral vascular resistance (Toda & Okamura, 2003). In nNOS deficient mice a reduced vasodilator response following stimulation of both efferent and afferent nerves was observed, an effect not seen in eNOS or inducible NOS (iNOS) deficient mice (Toda et al. 2009). Thus, there appears to be an apparent loss of ACh dependency as primary transmitter in the parasympathetic innervation and transfer to NO mediated vasodilatory processes, which is distinct to the innervation of peripheral arteries.

Parasympathetic innervation from the PPG acting directly or indirectly on nociceptive sensory fibres has also been suggested to participate in regulation of blood–brain barrier (BBB) permeability (Delépine & Aubineau, 1997; Yarnitsky et al. 2004 a,2004 b) and development of cluster headaches/migraine (Yarnitsky et al. 2003).

Additionally, it is involved in the dilatation of cerebral arteries following ‘breakthrough of autoregulation’ in acute hypertension (Agassandian et al. 2003).

Parasympathetic targets with clinical relevance for the cerebral vasculature

The critical importance of the brainstem for life means that its perfusion needs to be tightly regulated. At rest, autonomic tone of the cerebral vessels is small (Heistad & Marcus, 1978; Heistad et al. 1978). However, it is reasonable to assume that any imbalance in the activity of autonomic nerves innervating the posterior cerebral circulation could have pathological implications for blood flow control and be a potential clinical target for various conditions in which cerebral blood flow is impeded, such as stroke or hypertension. The importance of the brainstem is highlighted by the finding that during extreme acute hypotension, cerebral blood flow is prioritised to the brainstem (Mueller et al. 1977). In this regard, we consider how disease is affected by the parasympathetic regulation of the cerebral vasculature and ways in which modulation of parasympathetic input to the cerebral arteries may be used to alleviate pathology.

During pathological increases of blood pressure above the upper limit of autoregulation, cerebral vessels dilate. Denervation of parasympathetic fibres originating from the PPG dampens this effect suggesting a baroreceptor mediated cerebral vasodilatation in rats (Talman & Dragon, 2000). It was later found that nNOS drove the observed dilatory response; using a specific nNOS inhibitor the breakthrough dilatation that occurs above the autoregulatory threshold was attenuated (Talman et al. 2007). One possibility is that the shift in the autoregulatory curve to the right in hypertension (e.g. as in the spontaneously hypertensive rat (SHR)) may reflect reduced parasympathetic nitrergic input.

The importance of NO in cerebral vascular regulation is highlighted by the association of the endogenous NOS inhibitor asymmetric dimethylarginine (ADMA), which is elevated in conditions of hypoperfusion such as stroke and hypertension (Yoo & Lee, 2001; Napoli et al. 2004; Kielstein & Zoccali, 2005). By targeting ADMA availability (Fan et al. 2013) and reducing plasma levels (Fan et al. 2013; Tsai et al. 2014) hypertension is prevented in young SHRs. Furthermore, ADMA infusion at levels that do not affect systemic blood pressure decreased CBF and induced vascular stiffness in healthy volunteers (Kielstein et al. 2006). In rats topical application of ADMA results in basilar vasoconstriction and an inhibition of ACh mediated vasodilatation and in rabbits it has the same effect in cortical vessels (Faraci et al. 1995). Thus, NO, in conjunction with parasympathetic innervation, plays a major role in maintaining blood flow to the brain in healthy and disease states. The role of ADMA in ischaemic stroke and its potential as a predictive disease marker has been discussed and reviewed elsewhere (Chen et al. 2012).

The endogenous NOS inhibitor monomethylarginine (l‐NMMA) has been reported to be present at lower concentrations than ADMA in the circulation (Vallance et al. 1992). Like ADMA, l‐NMMA was shown to reduce cortical blood flow when injected into the internal carotid artery without affecting systemic blood pressure (Thompson et al. 1996). When challenged with hypercapnia, the chemoregulatory response normally resulting in increased cerebral blood flow in monkeys was prevented by prior intracarotid injection of l‐NMMA. The same was not true for cerebral autoregulation due to changes in blood pressure (Thompson et al. 1996). This result provides further evidence for the role of NO in the control of cerebral blood flow.

Nitric oxide is synthesised by NOS from l‐arginine. The active site of NOS is subject to inhibition by the asymmetric methylarginines, ADMA and l‐NMMA. These inhibitors could be a good clinical target in conditions where the synthesis or bioavailability of NO is compromised (Leiper & Nandi, 2011). However, it is not specific to nNOS but also work on eNOS.

The finding of markers associated with cardiovascular risk, and possibly autonomic imbalance, such as ADMA (Faraci et al. 1995), could provide other therapeutic pathways with targeted prophylactic medicine. In humans, using sulfhydryl angiotensin converting enzyme (ACE) inhibition to reduce blood pressure also restored the NO/ADMA balance in hypertension (Napoli et al. 2004). Similarly, in SHRs, 10 weeks of treatment with an ACE inhibitor not only decreased blood pressure but improved the vasodilatatory response of the basilar artery and blood flow during hypotension as well as the NO‐mediated dilatation during topical application of ACh; these data all suggest an improvement in endothelial function of the basilar artery (Toyoda et al. 1998). Furthermore, NO or NO donor administration in stroke models has been shown to reduce lesion size and increase cerebral perfusion (Willmot et al. 2005).

The direct effects of parasympathetic innervation can be seen during a hypotensive challenge in SHRs in which forebrain blood flow was reduced after chronic parasympathetic denervation compared to sham controls (Koketsu et al. 1992). These findings were not corroborated in normotensive Sprague–Dawley rats (Branston et al. 1995). The parasympathetic system is involved in vasodilatation at the lower end of autoregulation (Branston et al. 1995). To what extent this is true for the posterior circulation of the brainstem remains to be established. In stroke, parasympathetic denervation exacerbates infarct size following middle cerebral artery occlusion (Kano et al. 1991; Koketsu et al. 1992). This evidence supports a role for parasympathetic nerves in protecting against brain ischaemia. In this context, stimulation of the PPG provided protection by increasing CBF and reducing the infarct size following cerebral artery occlusion (Henninger & Fisher, 2007). A reduction in ischaemic area as well as protection of blood–brain barrier integrity were found in a model of photothrombotic stroke with PPG stimulation (Levi et al. 2012).

It is worth considering the potential clinical importance of cross‐talk between sympathetic and parasympathetic branches mentioned above, as it can result in alterations in autonomic regulation in pathological circumstances. For instance stimulation of sympathetic nerves in normotensive rats results in basilar artery vasodilatation and increased blood flow; the same is not true for hypertensive animals or animals with chronic superior cervical ganglion denervation (Chang et al. 2012). The attenuated response is likely to be, at least in part, due to a reduction in the sympathetic to parasympathetic terminal density at the level of the basilar artery in the hypertensive animals (Lee & Saito, 1986) or a lack of sympathetic input (Chang et al. 2012). Conversely, an overactive sympathetic and underactive parasympathetic nervous system limits basilar artery dilatation in the face of a hypotensive challenge (Barry et al. 1982). This may cause ischaemic damage especially in conditions where cerebrovascular resistance is high (e.g. SHRs) (Paton et al. 2009; Cates et al. 2011, 2012). Furthermore the limited vasodilatation of the basilar artery (Toyoda et al. 1998) contributes to the shift of the autoregulatory curve to higher BP levels in hypertension (Barry et al. 1982), further increasing the susceptibility of hypoperfusion.

In the case of ischaemic stroke, stimulation of the parasympathetic nervous system to increase cerebral blood flow and restrict ischaemia has clinical potential particularly considering the limiting treatments currently available and the restrictions of their use, as is the case with thrombolysis therapy (Cheyuo et al. 2011). Stroke in the vertebrobasilar circulation is a clinical emergency and accounts for about 20% of all ischaemic strokes (Schoen et al. 2011) and has a high morbidity and mortality rate. Evidence shows that parasympathetic input provides some protection from ischaemic damage (Kano et al. 1991; Koketsu et al. 1992; Henninger & Fisher, 2007; Cheyuo et al. 2011; Levi et al. 2012). Importantly this can be effective even after 24 h of stroke onset (Solberg et al. 2008; Levi et al. 2012). The mechanisms of action require further investigation, as does the relevance of this treatment to the vertebrobasilar circulation. A comprehensive review on the potential of harnessing the parasympathetic system as a therapy in stroke has been published recently (Cheyuo et al. 2011). Whether this can also be targeted to assist with vertebrobasilar artery hypoperfusion in neurogenic hypertension (Cates et al. 2012; Marina et al. 2015), for example, remains unknown but holds therapeutic potential. Indeed, the BrainsGate Ischemic Stroke System comprises a stimulator implanted adjacent to the PPG via the greater palatine canal under local anaesthesia (Khurana et al. 2009). This device is currently undergoing phase 3 trials (Clinical Trials.gov: NCT00826059 and NCT01874093). Another device directed to stimulation of the otic ganglion has also been described (Shalev & Gross, 2010). By extrapolation, once the source(s) of input to the posterior cerebral circulation has been clarified, these nerves/ganglia could be targeted in a similar fashion.

Systemically, activation of the parasympathetic efferent vagus nerve has been shown to inhibit pro‐inflammatory cytokine release and decrease the inflammatory response. This function is coined ‘the cholinergic anti‐inflammatory pathway’ (Borovikova et al. 2000; Pavlov et al. 2003; Tracey, 2007). Evidence suggest it works via nicotinic α7‐acetylcholinergic receptor stimulation (Wang et al. 2003) and has been reported to have beneficial effects in a plethora of diseases from irritable bowel syndrome to depression (Wang et al. 2004; Rush et al. 2005; van Westerloo et al. 2006; Ghia et al. 2007; van Maanen et al. 2009) as well as improving kidney transplant outcomes (Hoeger et al. 2010). We propose that activation of the parasympathetic pathways innervating the cerebral arteries could also yield a similar anti‐inflammatory response. In the anterior circulation this may be relevant for Alzheimer's disease (de la Torre, 2002; Reale et al. 2004), and for posterior cerebral arteries, hypertension, which has been linked to inflammatory activity in brainstem walls (Waki et al. 2007, 2010, 2013; Zubcevic et al. 2011).

In this regard, both VIP and PACAP have been reported to exert similar anti‐inflammatory effects through the VPAC₁ receptor, which could be a relevant mechanism for the parasympathetic innervation of cerebral arteries (Leceta et al. 2000; Martinez et al. 2002; Grimm et al. 2003).

Yarnitsky et al. (2005) have shown that electrical stimulation of the PPG can prevent cerebral vascular spasm after subarachnoid haemorrhage and improve blood flow. Acetylcholine (given via intracisternal injection) can counter effects of cerebral vasospasm in basilar arteries after subarachnoid haemorrhage by suppressing the production of cytokines (Song et al. 2014). However, it is worth noting that PPG stimulation has also been reported to disrupt the BBB and induce neurogenic inflammation via a reflex involving nociceptive sensory fibres (Delépine & Aubineau, 1997). A similar mechanism has also been ascribed to pain in migraine: Yarnitsky et al. (2003) reported that PPG activation during migraine is involved in sensitisation of intracranial nociceptors by direct or indirect activation of sensory pain fibres by the release of ACh, VIP and NO. This effect of PPG stimulation has been used clinically to deliver macromolecules through the BBB (Yarnitsky et al. 2004 a). Hence the (over‐) stimulation of the parasympathetic innervation to the cerebral arteries may not only yield positive effects. Clinical application will require careful adaptation depending on the effect sought.

Finally, and hypothetically, peptide targeting of the cerebral vasculature may also be possible. Using phage display, proteins were identified that targeted the endothelial cells of the aorta and heart (Greig et al. 2010). Using such a technique may allow selective targeting of the cerebral arteries and may even differentiate between posterior versus the anterior circulation. Such targeting might allow enhancement of efficacy of vasodilating transmitter substances, for example.

Concluding remarks

Control of blood flow to the brain is highly specialised and regionally specific. Uniquely, resistance is controlled in part by the main supply vessels as well as the microcirculation. We have discussed how the Circle of Willis and the vertebrobasilar circulations have different patterns of nervous innervation from parasympathetic, sympathetic and sensory systems. Because of similar phenotypes, we stress caution as to how to separate parasympathetic from sensory innervations immunocytochemically. Whilst the parasympathetic innervation of the anterior cerebral circulation is well described, this review highlights the need for clarification of the sources of parasympathetic innervation to the posterior cerebral circulation. It is clear that there are multiple sources of parasympathetic innervation and these change with age and in disease states and between species. The distinct sources of origin provide an opportunity for targeted regulation affecting sub‐components of the cerebral circulation, which may be relevant to the treatment of stroke. The review highlights the number of transmitter systems used by the parasympathetic nerves innervating the cerebral vessels (e.g. ACh, NO, VIP, PACAP and PHI); this may provide fail‐safe mechanisms, state‐dependent control, and regulation of the magnitude of the vasoactive response, for example. The evidence is unequivocal that there is a strong interdependence between the sympathetic and parasympathetic innervation both developmentally and functionally as they modulate each other. An example is the basilar artery vasodilatation induced by sympathetic nervous stimulation triggered by NO release from parasympathetic terminals. Malfunction of one of these systems will functionally alter the performance of the other. Thus, therapy targeting the parasympathetic nervous system to increase brain blood flow should not only be assured of a functional parasympathetic but also sensory and sympathetic innervation.

Innovative methods are being used to selectively stimulate specific nerves and ganglia innervating components of the cerebral circulation to increase blood flow. Whether specific endothelial markers on aspects of the cerebral circulation can be identified and targeted to modulate parasympathetically mediated vasodilatation remains a project for future scientific enquiry.

Additional information

Competing interests

None declared.

Funding

This article was written with the support of the British Heart Foundation (RG/12/6/29670 ).

Biographies

Eva Roloff is currently a Research Associate specialising in systems physiology and neuroscience. She holds a Masters in Biology from Copenhagen, Denmark (1995) and a PhD from Aberdeen, UK (2003).

Ana Tomiak‐Baquero completed her PhD in 2013 at the University of Bristol. She is currently a Research Associate working on brain blood flow control in rodents with hypertension.

Sergey Kasparov holds a MD PhD (Moscow State University, Russia) and has a Chair in Molecular Neuroscience. Current interests are in astrocyte signalling.

Julian Paton completed his PhD in 1987 (University of London) and trained in London, USA and Germany. In 2004 he was awarded a British Heart Foundation Lectureship. He leads a translational research group in cardiovascular and respiratory disease.