Ion channel networks in the control of cerebral blood flow

The neurovascular unit

The neurovascular unit (NVU) consists of neurons, astrocytes and the cells of parenchymal arterioles, which consist of a single layer of SM cells (SMCs) surrounding the endothelium (Figure 1a). Parenchymal arterioles originate from pial (surface) vessels and penetrate into the brain, where they become almost completely encased by astrocytic endfeet.^30,31 This positions endfeet to act as intermediates between neurons and the vasculature. Similarly, endothelial membrane extensions project through fenestrations in the internal elastic lamina and basement membrane of arterioles to directly contact SMCs (Figure 1b). These structures—termed myoendothelial projections (MEPs)—not only provide direct contact between these two cell types through gap junctions but also offer a unique intracellular and extracellular microdomain signaling environment for controlling vascular tone. In contrast, direct cell–cell contact has not been observed between astrocytic endfeet and SMCs, although the membranes of these cells are closely opposed. Each parenchymal arteriole supplies a large territory of downstream capillaries, an anatomical organization that positions parenchymal arterioles as bottlenecks to the entry of blood into the brain. ³² Therefore, control of parenchymal arteriole diameter by NVC mechanisms is of vital importance to the regulation of downstream blood flow. OPEN IN VIEWER

The ionic composition in the NVU establishes the basal conditions for controlling cerebral blood flow

The choroid plexuses produce cerebrospinal fluid (CSF), which fills the ventricles and the subarachnoid space and circulates from the latter into the brain parenchyma via the Virchow-Robin space, taking a paravascular route through the ‘glymphatic’ system. ³⁴ The composition of CSF is distinct from that of plasma (see Table 2, reproduced from Brown et al. ³⁵ ). Thus, with the luminal surface of ECs exposed to plasma and the parenchymal surface of SMCs bathed in CSF, SMCs and ECs experience extracellular milieus with different ionic compositions. The concentrations of ions in these extracellular compartments dictate their equilibrium potentials and, by extension, both ion channel activity and V_m. This, in turn, influences the level of resting SM tone and tissue perfusion. Variations in local ion concentrations during neuronal activity modulate the SM V_m, thereby exerting a powerful effect on parenchymal arteriolar diameter; this is the central relationship underlying the dynamic regulation of cerebral blood flow.

OPEN IN VIEWER

Table 2.

The composition of plasma and cerebrospinal fluid.

	Plasma	CSF
Na+ (mM)	155	151
K+ (mM)	4.6	3.0
Mg2+ (mM)	0.7	1.0
Ca2+ (mM)	2.9	1.4
Cl− (mM)	121	133
HCO3− (mM)	26.2	25.8
Glucose (mM)	6.3	4.2
Amino acids (mM)	2.3	0.8
pH	7.4	7.4
Osmolality (mosmol.Kg H2O−1)	300	305
Protein a (mg 100 g−1)	6500	25

The key role of the voltage-dependent Ca²⁺ channel as a membrane potential biosensor

The V_m of SM exerts profound control over cerebral artery diameter.^9,36 In isolated parenchymal arterioles, a graded increase in intravascular pressure, for example from 5 mm Hg to 40 mm Hg, causes graded V_m depolarization from approximately –60 mV to –40 mV, an elevation in intracellular calcium ([Ca²⁺]_i) from ∼120 nM to ∼250 nM, and approximately 50% constriction ⁹ (Figure 2). In contrast, V_m hyperpolarization to –60 mV causes near-maximal vasodilation (i.e. almost equivalent to the passive diameter of the vessel).^9,36 The depolarization/vasoconstriction to increasing pressure is termed the ‘myogenic response’ and is an intrinsic property of small resistance arterioles from a range of vascular beds. ³⁷ In brain arterioles, the myogenic response is particularly important for cerebral autoregulation, in which vessels constrict or relax to maintain a constant level of brain perfusion despite varying peripheral blood pressure. ³⁰ Therefore, myogenic regulation sets the level of resting perfusion, while also providing a baseline tone from which arterioles can dilate or constrict in response to factors released during neuronal activity. As such, myogenic mechanisms can be viewed as the foundation upon which NVC mechanisms are constructed. OPEN IN VIEWER

Pial arteries on the surface of the brain develop less myogenic tone in response to pressure compared with parenchymal arterioles, despite the fact that the relationship between SM [Ca²⁺]_i and constriction is identical in both of these types of vessel (see Figure 2d). This is because parenchymal arteriole SM is more depolarized than pial SM at lower pressures (Figure 2b) and thus exhibits a greater influx of Ca²⁺ (Figure 2c) and greater constriction (Figure 2a). The molecular arrangement that leads to the higher sensitivity of parenchymal arterioles to pressure has not yet been elucidated but likely involves a loss of negative feedback control of the SM V_m by Ca²⁺ spark-BK channel interactions (see ‘Ryanodine Receptors in Smooth Muscle’, below).

L-type VDCCs are the key voltage sensors in vascular SM. ³⁸ These channels translate changes in the V_m into alterations in [Ca²⁺]_i and thereby adjust the contractile state of the cell. Their central importance is evidenced by the fact that V_m changes induced by pressure or other agents have no effect on vascular tone in the presence of VDCC inhibitors, ³⁶ highlighting the vital role of Ca²⁺ entry through L-type VDCCs in vasoconstriction. Parenchymal arteriolar myocytes express L-type VDCCs composed of pore-forming Ca_V1.2 α_1C-subunits and associated α₂δ-, β- and γ-subunits.^9,39 Recent studies have also presented evidence for the expression and function of T-type VDCCs in SMCs of pial arteries.^40,41 However, no contribution of T-type channels to either [Ca²⁺]_i or arteriolar tone has been shown at physiological pressures in isolated parenchymal arterioles. ⁹

Neurovascular signaling mediators that converge on the SM utilize a variety of mechanisms to drive changes in V_m. Critically, this leads to either (1) closure of VDCCs (in the case of membrane hyperpolarization), a fall in global [Ca²⁺]_i and vasodilation or (2) depolarization and an increase in VDCC open probability (Po), causing an increase in [Ca²⁺]_i and vasoconstriction.^38,42 Thus, tuning the SM V_m to modulate Ca²⁺ entry through VDCCs is the key mechanism for controlling the diameter of parenchymal arterioles and rapidly influencing cerebral blood flow.

A role for smooth muscle transient receptor potential channels in myogenic depolarization: evidence from pial and parenchymal arterioles

Transient receptor potential (TRP) channels are a diverse family of cation-permeable channels activated by a wide range of stimuli. ⁴³ Members of the TRP family share an overall structural similarity, forming tetrameric homomeric or heteromeric channels from subunits containing six transmembrane domains. TRP channels are broadly classified on the basis of sequence homology into six subfamilies in mammals: TRPA (ankyrin), TRPC (classic or canonical), TRPM (melastatin), TRPML (mucoliptin), TRPP (polycystin) and TRPV (vanilloid). ⁴⁴ Most of these channels act as routes for the entry of Ca²⁺, although some, such as a subset of TRPM channels, primarily mediate Na⁺ entry. For a comprehensive overview of TRP channels in the vasculature, see Earley and Brayden. ⁴⁵

Mounting evidence suggests that intravascular pressure or vasoconstrictor agonists can cause V_m depolarization through the activation of TRP channels in pial artery myocytes; this, in turn, increases the activity of VDCCs and promotes vasoconstriction. TRPV4, TRPC3, TRPC6 and TRPM4 channels have all been identified in pial SM,^46–49 and recent studies indicate that pressure causes depolarization through activation of TRPM4 and TRPC6 channels. Monovalent cation-permeable TRPM4 channels, though impermeable to Ca²⁺ ions, ⁴⁴ are activated by sarcoplasmic reticulum (SR) 1,4,5-trisphosphate receptor (IP₃R)-mediated Ca²⁺ release events in pial artery myocytes. ⁵⁰ This results in an influx of Na⁺ through TRPM4, which can be measured as a transient inward cation current (TICC). ⁵⁰ Pressure-induced constriction of ex vivo cerebral arteries requires functional TRPM4 channels, ⁴⁶ and blocking TRPM4 causes almost maximal vasodilation of pressurized arteries, ⁵¹ suggesting that these channels are vitally important for the myogenic response. Similarly, activation of TRPC6 channels, which have a Ca²⁺:Na⁺ permeability ratio of ∼6:1, ⁵² is another contributor to pressure-induced constriction of isolated pial arteries, as evidenced by the fact that knockdown of TRPC6 using antisense oligonucleotides attenuates the myogenic response. ⁴⁷ A recent study, ²¹ conducted in isolated myocytes and pressurized pial arteries, unifies the contributions of TRPC6 and TRPM4 into a single mechanism for myogenic constriction. Here, pressure-induced activation of phospholipase C (PLC) γ1 liberates IP₃ and diacyl glycerol (DAG). Subsequently, DAG activates TRPC6 channels directly and the resulting Ca²⁺ influx across the sarcolemma acts synergistically with IP₃ to activate IP₃Rs, releasing Ca²⁺ from the SR that then activates TRPM4 channels to depolarize the membrane. ²¹

Despite these major advances in our understanding, the mechanism by which pressure activates TRP channels has not been firmly established. In addition to the possibility that TRP channels directly sense membrane deformation, ⁵³ they may be attached to other cellular structures that respond to pressure, such as the cytoskeleton. ⁵⁴ Alternatively, upstream mechanosensitive signaling elements may transduce physical forces into second messenger signaling (such as PLC activation), ²¹ which is then detected by TRP channels. ⁵⁵ These possibilities, which are not mutually exclusive, have recently been reviewed in detail. ⁵⁶

In contrast to the vasoconstriction associated with TRPC6 and TRPM4 activation, SM TRPV4 channels, which are non-selective but possess a high permeability to Ca²⁺, have been implicated in vasodilatory mechanisms in pial arteries. ⁴⁹ According to this mechanism, reported by Earley et al., epoxyeicosatrienoic acids (EETs) activate TRPV4 channels, which conduct Ca²⁺ into the cell and enhance ryanodine receptor (RyR)-mediated Ca²⁺ spark activity (see below), ⁴⁹ resulting in vasodilation. ⁵⁷

The studies described above were conducted using pial arteries. Recently, Li et al. ²⁰ pioneered the exploration of SM TRP channel contributions to myogenic tone in parenchymal arterioles. In keeping with their myogenic role in pial arteries, TRPM4 channels were found to be vitally important for the development of tone in isolated arterioles from this vascular bed. Here, TRPM4 activation and membrane depolarization lies downstream of direct mechanoactivation of purinergic P2Y₄ and P2Y₆ G_q-type G-protein coupled receptors (G_qPCRs) and consequent stimulation of PLC activity by intravascular pressure,^20,58,59 suggesting that mechanisms similar to those reported for pial arteries may also be at play in parenchymal arterioles (Figure 3). OPEN IN VIEWER

Ryanodine receptors in smooth muscle: Ca²⁺-release channels that oppose myogenic constriction

Although extracellular Ca²⁺ entry through VDCCs has a central role in the delivery of Ca²⁺ for SM contraction, Ca²⁺ release from the SR through RyRs has the opposite effect. ⁶¹ Thus, Ca²⁺ in SM has a dichotomous role, with distinct Ca²⁺ signaling modalities either promoting or opposing vasoconstriction.

RyRs are homotetrameric assemblies located on the sarco-/endoplasmic reticulum (SR/ER) in many cell types. They are gated by Ca²⁺ and modulated by Mg²⁺, ATP, calmodulin and FK506-binding proteins as well as by phosphorylation by protein kinases such as protein kinase A (PKA), Ca²⁺/calmodulin-dependent kinase II and cGMP-dependent protein kinase (PKG). ⁶² Ca²⁺ release through RyRs can lead to two types of signal: waves and sparks. Ca²⁺ waves are propagating events with considerable spatial spread that result from Ca²⁺-induced Ca²⁺ release (CICR), a process whereby the presence of a sufficiently high concentration of local Ca²⁺ evokes Ca²⁺ release from nearby RyRs and/or IP₃Rs in the SR membrane. This released Ca²⁺ then evokes further release from adjacent channels; the process repeats, leading to a regenerative Ca²⁺ wave that propagates along the SR. The role of Ca²⁺ waves in vascular SM is not fully understood, although waves in pial arteries may contribute to myogenic tone development, particularly at lower pressures. ⁶³ In contrast, Ca²⁺ sparks are highly localized, brief events mediated by the simultaneous opening of several clustered RyRs that cause local Ca²⁺ to reach micromolar concentrations.^61,64,65 In pial artery SM, sparks occur at physiological pressures (60 mm Hg). ⁶⁶ In these vessels, and a range of other vascular beds, large-conductance Ca²⁺-activated K⁺ (BK) channels are localized in the sarcolemma adjacent to SR RyR Ca²⁺ spark sites. These channels are activated by sparks, releasing K⁺ and resulting in a spontaneous transient outward current (STOC) that causes a brief hyperpolarization. ⁶¹ In the pial circulation, this process acts as a negative feedback on membrane depolarization and Ca²⁺ entry through VDCCs^61,67 and thus is an important determinant of arteriolar tone.

In contrast to pressurized pial arteries, where Ca²⁺ spark activity is a prominent feature, spark activity is absent in the SM of ex vivo pressurized (40 mm Hg) parenchymal arterioles; instead, asynchronous ryanodine-sensitive Ca²⁺ waves are the predominant form of Ca²⁺ signal here. ¹⁴ However, the RyR Ca²⁺ spark-BK channel STOC signaling architecture is still present and can be engaged under certain circumstances, ¹⁴ making this a potential target for neurovascular signaling mediators. Dabertrand and co-workers ¹⁴ demonstrated this principle by acidifying artificial CSF from pH 7.4 to pH 7.0, which converted SM RyR-sensitive Ca²⁺ waves to sparks in parenchymal arterioles. The resultant increase in spark activity opposed pressure-induced vasoconstriction and led to a rapid and sustained vasodilation that was sensitive to both ryanodine and the BK channel blocker paxilline in a non-additive manner. ¹⁴ It is presently unclear why Ca²⁺ sparks do not occur under basal (experimental) conditions in isolated parenchymal arterioles, but this observation may help to explain why these arterioles develop higher tone at lower pressures compared with pial arteries. Indeed, one possibility is that RyRs in parenchymal arterioles have a higher basal open probability than their counterparts in pial arteries. ⁶⁸ Such higher basal sensitivity to Ca²⁺ could lead to the preferential generation of waves rather than sparks in parenchymal arterioles, leaving an absence of the hyperpolarizing influence of spark-STOC signaling and higher tone. This contrasts with pial arteries, where basal spark-STOC signaling activity acts as a negative feedback that resists SM membrane depolarization, resulting in lower tone at the same pressure.

IP₃ receptors

IP₃Rs are ubiquitous ligand-gated Ca²⁺-permeable channels located in the membrane of the SR/ER. ⁶⁹ To date, IP₃R signaling in parenchymal arteriole SM has not been examined, but in pial arteries, UTP-induced PLC activation has been linked to IP₃R-dependent Ca²⁺ waves that lead to vasoconstriction. ¹³ Moreover, Xi and co-workers identified a possible physical interaction of IP₃Rs with TRPC3 in pial SM that was independent of SR Ca²⁺ release. The resultant cation current was demonstrated to promote vasoconstriction in response to endothelin-1. ⁴⁸ Since TRPC3 channels do not participate in pressure-induced constriction, ⁷⁰ this interaction might be important for receptor activation-induced constriction of cerebral arteries. As previously noted, pial SM IP₃R activation leads to depolarizing TRPM4 currents. ²¹ Since TRPM4 channels have a similar role in parenchymal arterioles, ²⁰ IP₃R-mediated Ca²⁺ signaling is a likely candidate for mediating activation of TRPM4 channels here as well. IP₃Rs are also involved in activation of the Ca²⁺-dependent transcription factor NFATc3 and subsequent regulation of pial SM gene expression. ⁷¹

K⁺ channels in smooth muscle

The resting V_m of SM (−35 to −45 mV) indicates that the ratio of K⁺ permeability (pK) to the permeability of other ionic species (assuming a collective reversal potential of 0 mV) is around 0.5–0.8:1. ²⁶ Thus, the SM V_m is quite positive to the K⁺ equilibrium potential (E_K), which is approximately −103 mV in CSF (assuming 3 mM extracellular K⁺ and 140 mM intracellular K⁺). These basal conditions mean that the activation of K⁺ channels can impart a major hyperpolarizing influence on cerebral SM, and thereby exert rapid and robust control over vascular diameter.

Smooth muscle inwardly rectifying K⁺ channels: Targets of endfoot K⁺ release

A modest elevation of external K⁺ is a rapid and powerful vasodilator of pial arteries and parenchymal arterioles.^22,23 For example, raising K⁺ from 3 mM to 8 mM hyperpolarizes the SM in isolated pressurized parenchymal arterioles to near the new E_K of −76 mV, and causes maximum vasodilation. ²² For this to occur, K⁺ permeability has to increase more than 50-fold; this is achieved through activation of SM inwardly rectifying K⁺ (K_IR) channels. ²⁶

K_IR channels, which form tetramers composed of two-transmembrane α-subunits, can be distinguished by their rectification properties. ⁷² The K_IR subtypes expressed in parenchymal arterioles have yet to be definitively established, but the pharmacological sensitivity of the parenchymal SM K_IR current to Ba²⁺ ions in situ^22,28 is consistent with the presence of functional K_IR2-containing channels. In addition, mRNA data indicate the expression of K_IR2.1 and K_IR2.2 isoforms in rat parenchymal arterioles. ²⁸ In pial arteries, knockout of K_IR2.1, but not K_IR2.2, completely ablates the dilatory response of vessels to small increases in extracellular K⁺, ²⁹ suggesting a principal role for the K_IR2.1 subtype. Furthermore, downregulation of SM K_IR2.1 expression leads to impairment of K⁺-mediated dilation of rat parenchymal arterioles, ²⁸ strongly suggesting that K_IR2.1 is of similar central importance in these smaller vessels. K_IR2 family members are strongly rectifying and are activated by hyperpolarization and external K⁺, with their half-activation voltage corresponding to E_K. ⁷³ At potentials positive to E_K, outward K⁺ current through K_IR channels progressively declines with depolarization as a result of intracellular polyamine block. ⁷⁴ As described previously, at basal extracellular K⁺ (3 mM), E_K is approximately −103 mV and the SM V_m is about −35 to −40 mV.^9,10 Under these conditions, K_IR channel activity is very low. ²⁶ K⁺ is released from neurons and from endfoot BK channels during NVC (see below), and raising K⁺ by a small amount (e.g., to 8 mM) leads to rapid unblock of K_IR channels and an enormous increase in K_IR conductance. ²⁶ The ensuing increase in K⁺ permeability drives the SM V_m to the new E_K (−76 mV with 8 mM [K⁺]_o), in the process promoting even greater K_IR channel activation by virtue of their steep voltage dependence. ²⁶ This ∼40 mV hyperpolarization closes VDCCs and leads to maximal vasodilation, making K_IR activation by K⁺ released during neuronal activity a powerful mechanism for increasing local cerebral blood flow. The stark hyperpolarizing effect of K⁺ on parenchymal SM V_m likely also involves K_IR channels located on the underlying endothelium, for which there exists preliminary evidence. ²⁶ Here, K⁺ may diffuse through the SM, or hyperpolarizing signals could be conducted upstream from capillaries (see below), to activate EC K_IR channels. This would promote EC hyperpolarization which can be transmitted via gap junctions to reinforce SM hyperpolarization, further amplifying K_IR channel activation and locking V_m at E_K (Figure 4). For an extended discussion of K⁺-induced activation of parenchymal arteriole K_IR channels in the control of cerebral blood flow, see Longden et al. ²⁶ OPEN IN VIEWER

K_IR subunits are also components of ATP-sensitive K⁺ (K_ATP) channels, which are found in many peripheral vascular beds. K_ATP channels, composed of K_IR6.1 or 6.2 with one of four types of auxiliary sulfonylurea (SUR) receptors, are weakly rectifying and their gating is determined by local concentrations of intracellular nucleotides. ⁷² K_ATP channels composed of K_IR6.1 and K_IR6.2 plus SUR2B subunits are expressed in basilar and middle cerebral pial arteries ⁷⁵ and may play a role in the dilatory responses of large surface arterioles to hypoxia ⁷⁶ and acidosis. ⁷⁷ However, these channels appear to be absent from the SM of parenchymal arterioles, as evidenced by a lack of vasodilatory response to the K_ATP channel opener cromakalim. ¹⁴

Ca²⁺-activated K⁺ channels: the hyperpolarizing influence of smooth muscle BK channel activity can be exploited by NVC mediators

The Ca²⁺-activated potassium (K_Ca) channel family is divided into three subgroups: small-conductance (SK) channels consisting of K_Ca2.1 (SK1), 2.2 (SK2) and 2.3 (SK3) isoforms, the intermediate-conductance (K_Ca3.1; IK) channel and the large conductance (K_Ca1.1; BK) channel. ⁷⁸ BK channel α-subunits possess six transmembrane domains with an extended intracellular C-terminus. These subunits gain their Ca²⁺-sensitivity from C-terminal regions, termed the Ca²⁺-bowl and the RCK (regulator of conductance for K⁺) domain. Four α-subunits associate with four β-subunits, the latter of which modulate the overall gating properties of the channel. α-subunits also possess a functional voltage sensor in their S4 region whose sensitivity is modified by local Ca²⁺ as well as by phosphorylation by a number of intracellular enzymes, making these channels responsive to a diverse range of intracellular stimuli. ⁷⁹

In line with observations in other vascular beds, parenchymal myocytes express functional BK channels, but do not express SK or IK channels. ¹⁰ In contrast to their important role in opposing pressure-induced constriction in pial artery SM, ⁶⁷ BK channels contribute only modestly to parenchymal arteriolar tone. ¹⁰ However, these channels represent a key target for many putative NVC mediators (e.g. EETs, nitric oxide and prostaglandins), which may harness BK channel activity to bring about membrane hyperpolarization.

K_V channels exert a tonic hyperpolarizing influence on the smooth muscle membrane potential

Voltage-dependent potassium (K_V) channels constitute the largest subfamily of the K⁺ channel superfamily. ⁸⁰ Functional channels are formed from four six-transmembrane α-subunits, either homo- or heteromeric, plus additional β-subunits. ⁸¹ Rat parenchymal arterioles express mRNA for K_V1.2 and K_V1.5 α-subunits. ²⁵ Inhibiting K_V channels in parenchymal arterioles causes a significant constriction, indicating that these channels provide a hyperpolarizing K⁺ efflux that opposes vasoconstriction at physiological pressures. ²⁵ Similarly, pial arteries express protein for K_V1.2 and K_V1.5 subtypes, which are thought to assemble into heterotetramers; these channels are tonically active in isolated pressurized pial arteries. ⁸² Figure 5 provides an overview of the ion channel influences on SM V_m, the balance of which determines arteriolar tone at any given point. OPEN IN VIEWER

IP₃Rs and TRPV4 play central roles in Ca²⁺ signaling in the endothelium and endfoot

In contrast to the RyR-mediated Ca²⁺ signaling that predominates in SM, Ca²⁺ release through IP₃Rs has a key role in communication to the SM by both the parenchymal endothelium and astrocytic endfeet. Although the mechanisms by which extracellular Ca²⁺ ions enter the endothelium and endfoot remain unsettled, it is unlikely that VDCCs contribute to Ca²⁺ influx in either cell type. Instead, local Ca²⁺ entry through TRPV4 channels has been shown to be important in both ECs ⁹⁰ and astrocytes. ¹⁶

IP₃Rs

Astrocytes are equipped with a range of neurotransmitter receptors, ⁹¹ enabling them to respond to a diverse array of neuronal signaling molecules. Glutamate is the predominant excitatory neurotransmitter in the brain, and the prevailing view is that astrocytes detect glutamatergic neuronal activity through G_q-coupled metabotropic glutamate receptors (mGluRs) expressed on processes that contact synapses. ⁹² According to this model, activation of these G_qPCRs causes the hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP₂) by PLC, resulting in the production of IP₃ and DAG. ⁹³ IP₃, in turn, acts on IP₃Rs to increase their sensitivity to [Ca²⁺]_i and increase Ca²⁺ efflux from the ER (Figure 6). The IP₃R2 isoform is thought to be primarily responsible for the release of Ca²⁺ from astrocytic intracellular stores in hippocampal and cortical astrocytes, as some studies have observed that ablation of the gene encoding this channel attenuates G_qPCR-stimulated and neuronal activity-evoked astrocytic Ca²⁺ signaling.^94–96 Further studies, using IP₃R2-knockout mice, have raised questions about the role of IP₃R-mediated astrocytic Ca²⁺ signaling in the control of cerebral blood flow, reporting that functional hyperemia is unaffected in these mice.^97,98 While these studies raise the intriguing possibility of IP₃R2-independent NVC, their broader conclusion—that astrocytic Ca²⁺ signaling is dispensable for NVC and functional hyperemia—is premature and directly contradicts a substantial body of evidence to the contrary. For example, several laboratories have firmly established that uncaging Ca²⁺ or IP₃ in astrocytic endfeet leads to diameter changes in the adjacent arteriole in brain slices and in vivo in both rats and mice.^{23,11,99–101} It should also be noted that retention of NVC upon IP₃R2 knockout is not a universal finding ⁹⁵ ; a recent study of IP₃R2-knockout mice using brain slices and in vivo preparations identified a diverse range of persistent astrocytic Ca²⁺ signaling events, localized mainly to fine processes and endfeet rather than somata. ¹⁰² Further recent studies have confirmed that astrocytic, fast Ca²⁺ transients that occur in the cell soma, fine processes, and endfeet reliably precede functional hyperemia in adult mice in vivo.^103,104 Thus, it appears that astrocytes from IP₃R2-knockout mice do continue to engage in Ca²⁺ signaling, perhaps due to compensatory upregulation of other IP₃R isoforms, likely on a spatial and temporal scale that previous studies were unable to detect. OPEN IN VIEWER

The astrocytic ER extends into the endfoot, as evidenced by the fact that both electrical field stimulation and direct photolysis of caged-IP₃ evoke Ca²⁺ release in endfoot processes adjacent to arterioles. ¹¹ Thus, it is thought that the presence of sufficiently high concentrations of IP₃ generated in response to neuronal activity evoke regenerative CICR, leading to a Ca²⁺ wave that propagates into the astrocytic endfoot. ¹¹ Arriving Ca²⁺ waves stimulate the production and/or release of numerous vasoactive molecules, which cause vasodilation and a resultant increase in local cerebral blood flow. Consistent with this, astrocytes in brain slices respond to electrical and pharmacological mGluR stimulation with an elevation in Ca²⁺.^11,105–107 These findings, which were obtained in juvenile animals, are compatible with an interesting recent study reporting that expression of G_q-coupled mGluRs is prominent in cortical and hippocampal astrocytes of young mice but is lost in adulthood. ¹⁰⁷ However, the implication of this latter study—that expression of these receptors is a transient, developmental phenomenon—must be weighed against functional evidence that mGluR-mediated Ca²⁺ signaling in astrocytes is known to occur in adult animals in vivo.^99,108 These seemingly irreconcilable observations suggest that the signaling cascade leading to astrocytic Ca²⁺ waves may be more complex than currently conceived—particularly in mature animals ¹⁰⁷ —and indicate that further work is needed to fully understand this phenomenon.

Given that the overlying SM acts as a barrier to immediate neuronal or astrocytic communication, it is unlikely that parenchymal arteriolar endothelium is directly stimulated during NVC. However, it is plausible that conducted signaling from capillary ECs (see below) could lead to parenchymal arteriole EC engagement. ¹⁰⁹ In ECs of small arteries of the periphery, Ca²⁺ release through IP₃Rs is well-established as an important signaling step leading to vasodilation. In particular, brief, high-amplitude, IP₃R-mediated Ca²⁺ signals localized specifically to MEP microdomains—termed Ca²⁺ pulsars—activate local IK channels, causing K⁺ efflux and hyperpolarization. ¹² This hyperpolarization can then be transmitted via gap junctions directly to the underlying SM; the released K⁺ may also amplify hyperpolarization by activating local K_IR channels and/or Na⁺/K⁺ ATPase pumps on the endothelium and/or SM ¹¹⁰ (Figure 7). Although currently unexplored, it is highly likely that this same signaling modality exists in parenchymal arteriole ECs, and if so, this could conceivably be engaged by conducted or blood-borne signals to further sculpt blood flow responses. OPEN IN VIEWER

TRPV4 channels

In addition to the well-studied IP₃-mediated waves described above, recent evidence indicates that spatially restricted astrocytic Ca²⁺ signals are also vital for the astrocyte to carry out its diverse roles, including the control of cerebral blood flow. Several studies have identified TRPV4 channels in astrocytes.^16,111 TRPV4 channels are enriched in astrocytic processes and endfeet adjacent to blood vessels and the glia limitans. ¹¹¹ Dunn et al. ¹⁶ recently studied the contribution of TRPV4 signaling to endfoot Ca²⁺ handling during NVC. This study identified spatially restricted TRPV4-mediated Ca²⁺-entry events that couple to endfoot IP₃Rs and then propagate throughout the endfoot as a Ca²⁺ wave ¹⁶ (Figure 6). Inhibition of TRPV4 prior to neuronal stimulation was shown to substantially reduce the magnitude of the evoked endfoot Ca²⁺ transient, suggesting that TRPV4 is an important contributor to endfoot Ca²⁺ entry during neurovascular signaling. Although these channels can be activated by EETs, Dunn and co-workers found no evidence for EET-mediated activation of TRPV4 during neuronal activity. Instead, the authors proposed that IP₃R-mediated Ca²⁺ release engages Ca²⁺-sensitive TRPV4 channels in the endfoot membrane, stimulating further Ca²⁺ entry into the cytosol to help propagate the wave. These data also raise the intriguing possibility of TRPV4 activation by local physical forces.^55,112 For example, mechanical activation of TRPV4 by diameter changes in blood vessels could lead to endfoot Ca²⁺ increases that then either feedback to modulate vessel diameter and blood flow ¹¹³ or propagate in retrograde fashion towards the astrocyte soma, where they might influence signaling to neurons.^16,114

In ECs of mesenteric arteries, Ca²⁺-entry events mediated by the gating of individual TRPV4 channels—termed ‘TRPV4 sparklets’—act as powerful regulators of vascular tone. ⁹⁰ Here, endothelial Ca²⁺ entry though TRPV4 evoked by synthetic agonists primarily activates IK channels leading to vasodilation. ⁹⁰ In the cerebral circulation, TRPV4 channels have been identified in the endothelium of the isolated middle cerebral (pial) arteries, where they contribute to Ca²⁺ entry in response to stimulation of purinergic G_qPCRs with UTP. ¹⁷ Similarly, Hamel and co-workers ¹⁸ recently confirmed the presence of functional TRPV4 channels in the endothelium of pressurized posterior cerebral (pial) arteries, showing that these channels link muscarinic receptor stimulation (with acetylcholine) to IK and SK channel activation. In the mesenteric endothelium, TRPV4 is co-localized with other signaling molecules at MEP sites, ⁹⁰ where cooperative gating among TRPV4 channels in a four-channel cluster is enhanced by localized concentrations of A-kinase anchoring protein (AKAP) 150. ¹¹⁵ Similarly, TRPV4 is also found at MEPs in cremaster arterioles. ¹¹⁶ This same clustered organization may also occur at MEPs in cerebral arteries/arterioles, a hypothesis supported by preliminary evidence. ¹¹⁷ To date, TRPV4 expression and function have not been explored in the parenchymal endothelium, but given its importance in mesenteric and pial ECs, this channel is a likely candidate for mediating Ca²⁺ entry into ECs in parenchymal arterioles as well.

K_Ca channels and K⁺ signaling: a mechanism for bi-directional control of parenchymal arteriole diameter

Among the key targets of endfoot Ca²⁺ increases are K_Ca channels. There is good evidence for the functional expression of BK channels in the endfoot membrane adjacent to underlying parenchymal arterioles.^22,118 Astrocytes also express IK channels on processes and endfeet, whereas evidence for the functional expression of SK channels in astrocytes is currently lacking. ²⁴ During neurovascular communication, IP₃-mediated Ca²⁺ waves arriving at the endfoot stimulate the opening of BK,^22,23 and possibly IK, ²⁴ channels. K⁺ ions then accumulate in the extracellular nano-space between the endfoot and SM.^22,23 This increase in local K⁺ concentration induces hyperpolarization and relaxation of the SM through activation of myocyte (and possibly endothelial ²⁶ ) K_IR channels ²² (Figure 7). In addition to being supported by in situ and in vivo experimental data,^22,23 a number of the features of this mechanism have been confirmed using computational modeling.^119–111

The magnitude of the neuronally evoked endfoot Ca²⁺ increase dictates relative BK channel activity and thus the concentration of K⁺ released into the nano-space between endfeet and arterioles. Notably, K⁺ can promote vasodilation or vasoconstriction depending on its concentration, inducing dilation at <20 mM through activation of K_IR channels, and constriction at higher concentrations through depolarization of the SM and activation of VDCCs. ²³ Accordingly, the degree of Ca²⁺ elevation in the astrocytic endfoot is a vital determinant of the polarity of the vascular response. This was demonstrated in experiments showing that endfoot Ca²⁺ levels could be manipulated by varying the intensity of electrical field stimulation or by carefully controlling uncaging of Ca²⁺ directly in the endfoot. Ca²⁺ levels less than ∼500 nM consistently caused dilation, whereas higher concentrations induced constriction. Both dilations and constrictions were sensitive to block by paxilline, supporting a model in which K⁺ release through endfoot BK channels mediates both vasodilation and vasoconstriction, depending on the intensity of the stimulus. ²⁴

In contrast to astrocytic endfeet, the parenchymal endothelium expresses IK and SK channels, which contribute to basal parenchymal arteriole tone. ¹⁰ SK and IK channels share a common topology, consisting of homomeric tetramers composed of four six-transmembrane α-subunits associated with intracellular Ca²⁺-binding calmodulin subunits at their C-terminal domains. ⁷⁹ This renders these channels exquisitely sensitive to changes in [Ca²⁺]_i. ¹²² Their vestigial voltage sensor, which contains fewer charged amino acids in the S4 transmembrane domain than that of K_V channels, means that SK and IK channels are voltage-insensitive. ¹²³ Hannah et al. ¹⁰ reported that blocking IK and SK channels in isolated pressurized parenchymal arterioles caused substantial vasoconstriction, and this same manipulation decreased resting cortical cerebral blood flow by 15% in vivo. Conversely, activation of these channels with NS309 maximally dilated isolated arterioles and greatly enhanced cortical cerebral blood flow, highlighting the role of these channels as powerful controllers of parenchymal arteriolar tone. ¹⁰ In contrast, isolated, pressurized middle cerebral arteries do not constrict to IK and SK channel inhibition, suggesting that these channels are not tonically active under basal conditions in larger surface arteries. ¹²⁴ In mesenteric arteries, the SK3 subtype has been identified at MEPs and EC-EC borders. ¹²⁵ The loss of apamin-sensitive K_Ca currents in mesenteric ECs from an SK3 gene suppression mouse confirms that this is the primary SK isoform for the regulation of vascular tone. ¹²⁶ SK2 is also present in ECs, but appears to be restricted to peri-nuclear regions of the cell, whereas SK1 is not present. ¹²⁷

Engagement of SK and IK channels is dependent on [Ca²⁺]_i; thus, the above observations suggest a level of tonic endothelial Ca²⁺ signaling in parenchymal arterioles in situ and in vivo. Extrapolating from mesenteric arteries, where IP₃-mediated Ca²⁺ pulsars activate IK channels within MEPs ¹² and thereby contribute to the control of vascular tone, it is possible to suggest that this same signaling architecture is present in the parenchymal circulation. If further studies establish that this is the case, K⁺ released through IK and SK channel activation by Ca²⁺ pulsars would be predicted to couple to local K_IR channel and Na⁺/K⁺ ATPase activation, which could act as an amplification mechanism to hyperpolarize the membrane and drive vasorelaxation ¹¹⁰ (Figure 7).

Arachidonic acid metabolites

One early proposal ¹²⁸ was that neuronally evoked Ca²⁺ waves activate Ca²⁺-sensitive cytosolic phospholipase A₂ in astrocytes, mobilizing arachidonic acid (AA) from membrane phospholipid pools and leading to subsequent metabolism of AA to EETs by cytochrome P450 2C11 enzymes. The four possible EET regioisomers—5,6-, 8,9-, 11,12- and 14,15-EET—act as short-range signaling hormones, dilating cerebral arteries in situ ¹²⁹ and contributing to the control of cerebral blood flow in vivo. ¹³⁰

EETs are synthesized by astrocytes in culture, ¹³¹ and cytochrome P450 2C11 mRNA has been detected in perivascular astrocytes. ¹³⁰ Astrocyte-derived EETs could potentially act in either an autocrine or paracrine manner to open endfoot ¹³² or SM ¹³³ BK channels, respectively. In the case of endfoot BK channels, this would cause the release of K⁺, which could then stimulate SM K_IR channel activity. ¹³² If EETs diffuse to the SM, they could open BK channels to hyperpolarize the membrane directly. ¹³³ The mechanism of EET action at BK channels involves ADP-ribosylation of G_sα subunits, ¹³⁴ resulting in an increase in channel activity. ¹³⁴ Astrocyte-derived EETs could also promote vasodilation through actions on TRPV4 channels in the endothelium,^17,18 myocytes,^19,49 and/or endfeet. ¹⁶ Despite these possibilities, recent in vivo data argue against a direct contribution of endfoot-derived EETs to SM relaxation during NVC, as vasodilation evoked by uncaging Ca²⁺ directly in astrocytic endfeet in vivo was insensitive to cytochrome P450 blockade. ⁹⁹ These observations suggest that the mechanism underlying the contribution of astrocyte-derived EETs to NVC is more complex and indirect than was first envisioned.

AA liberated by PLA₂ can also serve as a substrate for the production of vasoactive prostaglandins (PGs). In this reaction, the dual peroxidase-cyclooxygenase actions of cyclooxygenase (COX) enzymes initially metabolize AA to PGG₂ followed by the rapid conversion of this intermediate to PGH₂. PGH₂ is metabolized by cytochrome P450 enzymes to a range of PG products. Of these, PGE₂ has been proposed to play a role in NVC. ¹⁰⁵ In this model, PGE₂ is liberated from the endfoot to act upon E-prostanoid (EP) receptors on SMCs. Notably, EP₄ receptors are G_s-coupled GPCRs, and stimulation of these receptors leads to activation of PKA, which can produce vasodilation. ¹³⁵ However, recent studies have raised questions about the astrocytic involvement in the vasodilatory response to COX products. One group reported that COX-1 and -2 are expressed in astrocytes in juvenile rat hippocampal-neocortical slices and showed that pharmacological blockade of COX activity inhibited in situ vasodilations evoked by mGluR agonists and astrocyte Ca²⁺ uncaging. ¹³⁶ However, another group found that COX-1 mRNA was expressed in only 10% of rat cortical astrocytes, whereas COX-2 was absent. ¹³⁷ This same group recently identified a sub-population of COX-2 expressing pyramidal neurons as the primary cell type able to synthesize PGE₂, which reportedly acts through G_s-coupled EP₂ and EP₄ receptors to produce hyperemia.^137,138 However, a fundamental issue is raised by a recent study showing that application of PGE₂ directly to isolated, pressurized parenchymal arterioles from both rat and mouse evoked constriction rather than dilation, through activation of EP₁ receptors, suggesting that PGE₂ is unlikely to act as a direct mediator of NVC on parenchymal arteriole smooth muscle. ¹³⁹ However, this does not rule out PGE₂-mediated vasodilatory effects through indirect actions on neurons ¹³⁹ or on the capillary endothelium.

Liberation of AA from plasma membrane phospholipids by PLA₂ also occurs in the endothelium. In contrast to astrocytes, the endothelium has been clearly demonstrated to possess abundant COX-1 and PGI₂ synthase enzymes. ¹⁴⁰ PGI₂ derived from the activities of these enzymes is the primary vasodilatory PG liberated from the endothelium in response to a variety of stimuli, including agonists of muscarinic, bradykinin and purinergic receptors.^141,142 Similar to the actions of PGE₂ on EP receptors, PGI₂ activates G_s-coupled I prostanoid receptors on SMCs, evoking PKA-mediated vasodilation. ¹⁴⁰ Although, PGI₂ and its analogues do relax cerebral arteries and arterioles, ¹⁴³ preliminary evidence from mouse brain slices suggests that PGI₂ is not involved in signaling during NVC. ¹⁴⁴

The parenchymal endothelium is also a likely source of EETs, as ECs produce these molecules in a range of vascular beds.^145,146 Moreover, cerebral arteries and arterioles dilate in response to direct EET application.^129,147 However, to our knowledge, the endothelial production of these molecules has yet to be directly investigated in parenchymal arterioles. Nonetheless, it is reasonable to speculate that endothelial EETs could influence blood flow by interacting with a range of ion channels in the parenchymal NVU, such as myocyte ¹⁰ or astrocytic^22,132 BK channels or endothelial,^17,18 SM^49,19 and astrocytic ¹⁶ TRPV4 channels. EET-mediated Ca²⁺ influx through endothelial TRP channels ¹⁴⁸ could also contribute to endothelial SK and IK channel activation and membrane hyperpolarization.

Nitric oxide: NVC mediator or modulator?

Another commonly touted Ca²⁺-dependent NVC mediator is nitric oxide (NO). However, the potential contribution of NO is complex and its actual role remains uncertain. NO derived from both neuronal and endothelial sources has been implicated in the control of cerebral blood flow,^149–151 whereas it is generally accepted that astrocytes do not generate NO under normal conditions. NO activates guanylate cyclase, leading to the production of cyclic guanosine monophosphate (cGMP). ¹⁵² Cyclic GMP, in turn, is a substrate for PKG, which interacts with multiple downstream effectors. Considering ion channel targets alone, PKG activity can modulate the phosphorylation status of BK channels, leading to an increase in their activity,^153,154 and may act through similar mechanisms to inhibit VDCCs. ¹⁵⁵ Cyclic GMP can increase RyR-mediated Ca²⁺ spark frequency, ⁵⁷ and NO may also decrease IP₃-mediated Ca²⁺ release through the actions of cGMP and PKG. ¹⁵⁶ In the parenchymal NVU, NO from either eNOS or nNOS could interact with guanylate cyclase in both astrocytes ¹⁵⁷ and SM ¹⁵⁸ to increase the activity of BK channels in these respective cell types. Signaling to SM VDCCs or RyRs would also be predicted to promote vasodilation and increased blood flow. Moreover, because NO-cGMP-PKG signaling has been reported to inhibit TRPV4 channels (alone ¹⁵⁹ or when complexed with TRPC1 ¹⁶⁰ or TRPP2 ¹⁶¹ ), endothelial and endfoot Ca²⁺ signaling could also be affected.

Recent work has identified a contribution of eNOS signaling to resting tone in isolated pressurized middle cerebral arteries and parenchymal arterioles, ¹²⁴ but whether NO is actively recruited during NVC is controversial.^149,162,163 However, emerging evidence supports the concept of endothelial engagement during NVC through the release of parenchymal factors that increase eNOS activity.^164,165 A study by LeMaistre-Stobart and co-workers ¹⁶⁵ reported that astrocytic release of D-serine, in combination with glutamate, may activate endothelial N-methyl D-aspartate (NMDA) receptors, which conduct Na⁺ and Ca²⁺ into the cell. This could then stimulate eNOS, increasing the production of NO and causing vasodilation of arterioles through the actions of NO on the adjacent SMCs. ¹⁶⁵ In support of this model, this same group previously demonstrated that endothelial NMDA-receptor activation dilates isolated middle cerebral arteries and brain slice parenchymal arterioles through an eNOS-dependent mechanism. ¹⁶⁴ However, the functional expression of NMDA receptors in native parenchymal endothelium is yet to be confirmed.