Astrocyte Inositol Triphosphate Receptor Type 2 and Cytosolic Phospholipase A2 Alpha Regulate Arteriole Responses in Mouse Neocortical Brain Slices

Functional hyperemia of the cerebral vascular system matches regional blood flow to the metabolic demands of the brain. One current model of neurovascular control holds that glutamate released by neurons activates group I metabotropic glutamate receptors (mGluRs) on astrocytes, resulting in the production of diffusible messengers that act to regulate smooth muscle cells surrounding cerebral arterioles. The acute mouse brain slice is an experimental system in which changes in arteriole diameter can precisely measured with light microscopy. Stimulation of the brain slice triggers specific cellular responses that can be correlated to changes in arteriole diameter. Here we used inositol trisphosphate receptor type 2 (IP3R2) and cytosolic phospholipase A2 alpha (cPLA2α) deficient mice to determine if astrocyte mGluR activation coupled to IP3R2-mediated Ca2+ release and subsequent cPLA2α activation is required for arteriole regulation. We measured changes in astrocyte cytosolic free Ca2+ and arteriole diameters in response to mGluR agonist or electrical field stimulation in acute neocortical mouse brain slices maintained in 95% or 20% O2. Astrocyte Ca2+ and arteriole responses to mGluR activation were absent in IP3R2− /− slices. Astrocyte Ca2+ responses to mGluR activation were unchanged by deletion of cPLA2α but arteriole responses to either mGluR agonist or electrical stimulation were ablated. The valence of changes in arteriole diameter (dilation/constriction) was dependent upon both stimulus and O2 concentration. Neuron-derived NO and activation of the group I mGluRs are required for responses to electrical stimulation. These findings indicate that an mGluR/IP3R2/cPLA2α signaling cascade in astrocytes is required to transduce neuronal glutamate release into arteriole responses.


Introduction
Blood flow to the brain is precisely regulated to match regional perfusion with metabolic requirements. Local activation of neurons produces signals that increase regional cerebral blood flow (rCBF) in a process known as functional hyperemia. Conversely, increases in arterial perfusion pressure are countered by pressure-induced increases in myogenic tone to stabilize blood flow. Thus the cerebral vasculature is able to maintain appropriate rCBF by both vasodilation and vasoconstriction.
Recent experimental work has established a model of cerebral vascular regulation that has at its center an astrocyte-dependent signaling pathway (reviewed [1,2]). The cytoarchitecture of astrocytes makes such a model feasible because they form a physical bridge between neural synapses and vascular structures. Astrocyte processes envelop many glutamatergic synapses and these same astrocytes also send specialized foot processes that cover the blood vessels of the brain [3,4]. In one current model of neurovascular regulation, activation of excitatory neurons results in the presynaptic release of glutamate (and sometimes other neurotransmitters). Glutamate interacts with neuronal postsynaptic receptors but can also bind group I mGluRs of a nearby astrocyte. Early work supporting this model found that vascular responses were prevented by antagonists of group I metabotropic glutamate receptors (mGluR) and were triggered by agonistinduced activation of the mGluR [5].
Astrocyte mGluR activation is coupled to Gq and activates phospholipase C which hydrolyzes phosphatidylinositol 4,5-bisphosphate from cellular membranes to produce inositol 1,4,5trisphosphate (IP 3 ) and 1,2-diacylglycerol (DAG). IP 3 then binds a cognate receptor, the IP 3 R, on the cytosolic face of the endoplasmic reticulum. Within astrocytes of the neocortex the only form of IP 3 R expressed is the type 2 IP 3 R (IP 3 R2). IP 3 R2 binding opens a Ca 2+ channel within the receptor causing Ca 2+ mobilization from internal stores [6]. Consistent with this portion of the model, direct mechanical activation of astrocytes in cortical brain slices caused arteriolar dilation which was eliminated by the cell-permeant Ca 2+ chelator BAPTA/AM [5].
The phospholipases A 2 (PLA 2 s) are a family of enzymes that hydrolyze a free fatty acid from the sn-2 position of membrane glycerolphospholipids and are highly expressed in the brain [7]. Previous experiments have suggested that increases in astrocyte Ca 2+ can activate Ca 2+ -dependent PLA 2 and that a Ca 2+dependent PLA 2 is needed for cerebrovascular regulation [8].
Thus, the next step in the model is that PLA 2 releases arachidonic acid which is metabolized by cyclooxygenase enzymes to form prostaglandin (PG) H 2 and by epoxygenase enzymes to form epoxyeicosatrienoic acids (EETs). PGH 2 is rapidly metabolized by terminal synthase enzymes to any of the PGs. The PGs have demonstrated vascular effects that are mediated through prostaglandin and thromboxane receptors located on the extracellular surface of vascular smooth muscle cells (VSMC). It appears that cerebral metabolism is coupled to PGE 2 -dependent vasoregulation. A low oxygen tension in brain slices increases glycolysis which produces lactate which reduces PGE 2 uptake by astrocyte prostaglandin transporters [9]. Increased extracellular PGE 2 dilates cerebral arterioles while cyclooxygenase inhibitors prevent vasomotor response [9,10].
While there is some experimental evidence to support this model of neurovascular regulation, many details remain unresolved. For example, experiments supporting the present model have relied on photolysis of caged IP 3 in astrocytes to trigger vascular responses [11]. However photo-activation of caged IP 3 releases supraphysiologic levels of IP 3 and does so in a way that may not represent the spatially-regulated release from subcellular compartments. In addition, the roles of PLA 2 s in neurovascular regulation have been principally examined by using pharmacologic inhibitors that are not specific for single PLA 2 isoforms and do not target specific cell types. Mammalian brain tissue expresses, and has enzymatic activity for all of, the major PLA 2 s including Ca 2+ -independent (iPLA2, GVIA), Ca 2+ -dependent secretory PLA 2 s (groups IIA, V and X) and the cytosolic PLA 2 s (cPLA 2 , GIV) [7]. The group IVA PLA 2 (cytosolic PLA 2 a, cPLA 2 a) is of particular interest because its translocation to specific cellular membranes is highly regulated and its enzymatic activity is enhanced by phospholipids that have arachidonate at the sn-2 position [12]. These pharmacologic inhibitors have effects on Figure 1. Astrocyte Ca 2+ responses to mGluR agonist application are attenuated IP 3 R2 2/2 slices. A. Neocortical brain slices from IP 3 R2 +/+ (left panel) and IP 3 R2 2/2 (right panel) mice were loaded with the Ca 2+ -sensitive fluorophore Rhod-2/AM and astrocytes were identified by dye uptake, morphology and location. Ca 2+ fluorescence was measured in the region of interest (green arrow) and is displayed at 3 time points in relation to 1S,3R-ACPD treatment: (a) before, (b) at peak response and (c) after. White outline indicates the region of magnification in C. Scale bar: 20 mm. B. Fluorescence intensity signals for the Ca 2+ fluorescence measured in the soma of the indicated astrocytes. Signals were corrected for background that was measured in an identical area immediately adjacent to the region of interest. Representative single traces of the Ca 2+ response in soma of IP 3 R2 +/ + astrocytes (left trace) and IP 3 R2 2/2 (right trace) are shown and the duration of 1S, 3R-ACPD application is indicated below the traces. C. Z-stack of 12 images encompassing 12 mm of depth in IP 3 R2 +/+ (left panel) and IP 3 R2 2/2 (right panel) brain slices. This demonstrates the ameboid shape of the astrocyte soma which extends a foot process near a neighboring arteriole. Scale Bar: 10 mm. D. Cumulative probability histograms of population responses are shown. Peak (left panel) and integrated (right panel) Ca 2+ responses of IP 3 R2 +/+ (open circles, 58 cells) and IP 3 R2 2/2 (filled circles, 63 cells) astrocytes with inset bar graphs indicating the mean 6 S.E.M. Nine slices were prepared from four mice for each genotype. doi:10.1371/journal.pone.0042194.g001 different forms of PLA 2 and may also have side-effects that are unrelated to PLA 2 blockade [13,14].
Here we have sought to determine the molecular constituents of astrocyte Ca 2+ signaling and PLA 2 activation in the cerebrovascular regulatory pathway. To do this we compared responses of cortical astrocytes and their neighboring arterioles in cortical brain slices derived from mice that were deficient in IP 3 R2 or cPLA 2 a.

Results
In one current model of neurovascular regulation it is postulated that activation of astrocyte mGluR by glutamate leads to activation of PLC which releases IP 3 . The free IP 3 binds to IP 3 receptors on the endoplasmic reticulum thus opening Ca 2+ channels and increasing intracellular Ca 2+ [8]. There are three IP 3 R isoforms but the type 2 receptor (IP 3 R2) appears to be the only form expressed in glial cells within the CNS [15,16]. Therefore we measured the astrocyte Ca 2+ and arteriole responses to stimulation of cortical brain slices from mice deficient in the type 2 IP 3 R (IP 3 R2 2/2 ) and their wild-type littermates (IP 3 R2 +/+ ).
We incubated acute cortical slices from IP 3 R2 +/+ and IP 3 R2 2/2 mice with a Ca 2+ sensitive fluorophore, Rhod-2/AM, in a manner that preferentially loads astrocytes [8]. After washout of unloaded dye and a period of equilibration in artificial CSF (ACSF) we introduced the metabotropic glutamate receptor agonist, 1S,3R-ACPD (50 mM) to the bath and examined the Ca 2+ responses with confocal fluorescence microscopy ( Figure 1A). Signals were measured in the soma of cortical astrocytes. These astrocytes were identified by their location, somatic morphology and the presence of a foot process extending from the soma ( Figure 1C). In the IP 3 R2 +/+ slices there were robust increases in the Ca 2+ signals in cells that were morphologically identified as astrocytes, while in the IP 3 R2 2/2 slices such Ca 2+ responses were absent ( Figure 1B and D) (IP 3 R2 +/+ , DF/F 0 = 302.4623.5%, Time Integrated F/F 0 = 78.8. 610.3. IP 3 R2 2/2 , no detectable Ca 2+ signal. P,0.001. n = 58 IP 3 R2 +/+ and 63 IP 3 R2 2/2 cells from 4 mice for each genotype).
Treatment with 1S,3R-ACPD initiates the vasoregulatory pathway at the point of astrocyte mGluR activation and thus bypasses upstream signaling in neurons. In order to examine the role of the astrocyte IP 3 signaling on neurovascular coupling following activation of neurons we performed electrical field stimulation in the cortical brain slices. We stimulated a cortical field with a bipolar electrode using 100 Hz for 200 ms repeated 48 times over a 4 min period and examined the arteriolar responses 200-300 mm from the electrode. We determined that this 100 Hz stimulation increased astrocyte somatic intracellular Ca 2+ consistent with their robust activation (Supporting Figure S3). In contrast to treatment with 1S,3R-ACPD, electrical field stimulation caused significant dilation in IP 3 R2 +/+ arterioles, which peaked 15 min after stimulation (5.4561.7%, n = 15). There was an insignificant constriction of 1.7060.92% in the IP 3 R2 2/2 slices ( Figure 3A) (n = 14; P,0.01 compared to IP 3 R2 +/+ between 5-30 minutes following stimulation). When the slices were pre-treated with U-46619 (50 nM), electrical stimulation caused a significant dilatory response in the IP 3 R2 +/+ arterioles when compared to the continuous constriction observed in the IP 3 R2 2/2 slices ( Figure 3B) (IP 3 R2 +/+ , 22.863.1% vs. IP 3 R2 2/2 , 222.265.0%; measured 15 minutes after electrical stimulation, expressed relative to diameter at start of stimulation; P,0.01).
In contrast to our results, a previous study in rat hippocampal slices found that electrical stimulation in a high O 2 environment caused arteriole constriction [9]. However, in a low O 2 environment, electrical stimulation caused arteriole dilation [9]. Therefore we wished to determine if O 2 tension alters the response to electrical stimulation of murine cortical slices. When electrical stimulation was delivered to IP 3 R2 +/+ and IP 3 R2 2/2 slices equilibrated in 20% O 2 the responses were identical to those observed in 95% O 2 ( Figure 3C, compare to 3A). To determine if electrical stimulation of neurons causes dilation by activation of the astrocyte mGluR, slices from IP 3 R2 +/+ were treated with the mGlu 5 antagonist, 2-Methyl-6-(phenylethynyl)pyridine hydrochloride (MPEP) (10 mM) and the mGlu 1 antagonist JNJ 16259685 (100 nM) before electrical stimulation. Blockade of these group I mGluRs prevented the arteriole dilation response to stimulation ( Figure 3C). Taken together these results show that, glutamatergic activation of astrocyte mGluRs results in IP 3 R2 receptor-mediated increases in astrocyte Ca 2+ that are necessary for either constriction or dilation of arterioles. The direction of the change in arteriole diameter depends on the pre-existing state of the vessel (U-46619 pre-treatment) and also the mode of activation (1S, 3R-ACPD vs. electrical stimulation).
Increases in astrocyte Ca 2+ are postulated to activate one or more forms of Ca 2+ -dependent PLA 2 and PLA 2 activity may be the rate limiting step in the generation of vasoactive eicosanoids [19]. Because of the unique biochemical and molecular properties of cPLA 2 a we explored arteriole responses in acute cortical brain slices derived from cPLA 2 a +/+ and cPLA 2 a 2/2 mice. Once again we selected arterioles based on the histology and diameter of the vessel and determined that there was no difference in the arteriole diameters between cPLA 2 a +/+ (11.661.3 mm, n = 29) and cPLA 2 a 2/2 neocortical slices (11.861.1 mm, n = 15). We repeated the 1S,3R-ACPD stimulation experiment in a 95% O 2 environment and found that in naïve cPLA 2 a +/+ slices, arterioles constricted 12.261.9% in response to 1S,3R-ACPD. In contrast arterioles in slices of cPLA 2 a 2/2 neocortex did not constrict in response to 1S, 3R-ACPD (1.661.3%; P,.001 compared to cPLA 2 a +/+ , Figure 4A). We postulated that cPLA 2 a serves to supply arachidonic acid and predicted that the response to exogenous PGE2 would be unaltered in cPLA 2 a 2/2 slices [5]. When 1S, 3R-ACPD was removed from the perfusate and replaced with 10 mM PGE2, arterioles of both the cPLA2a+/+ and cPLA 2 a 2/2 slices constricted identically (cPLA2a+/+, 35.165.2%; cPLA 2 a 2/2 , 34.466.6%; Figure 4A).
In a previous investigation the polarity of arteriole response to t-ACPD was dependent upon oxygen content of the slice media [9]. A low oxygen environment appeared to enhance glycolysis with increased lactate and PGE 2 levels and resulted in dilation of arterioles [9]. Therefore we equilibrated slices from cPLA 2 a +/+ and cPLA 2 a 2/2 mice in 20% O 2 and 5% CO 2 before and during treatment with 1S, 3R-ACPD. In low O 2 , pharmacological activation of the mGluR resulted in dilation of the cPLA 2 a +/+ arterioles while cPLA 2 a 2/2 arterioles remained unresponsive ( Figure 4C). Interestingly the 20% O 2 environment also reversed the polarity of the response to 10 mM PGE 2 , causing arteriole dilation in both genotypes ( Figure 4C). Taken together these results show that cPLA 2 a is required for the vascular responses to 1S,3R-ACPD stimulation of the mGluR. Importantly, cPLA 2 a reaction products and their metabolites can trigger either arteriole constriction or dilation, depending on the initial condition and the metabolic state of the vessel. . Arteriole responses to mGluR agonist application are eliminated in IP 3 R2 2/2 neocortical slices. A. Gradient contrast imaging was used to measure arteriole responses to treatment. Responses to 1S,3R-ACPD were quantified by defining 6 points (2 of the points are shown in this figure) at which to measure changes in arteriole diameter over time (described in Methods). The arteriole diameter is expressed as the average diameter of the points. Scale bar: 5 mm. B. Cortical slices from IP 3 R2 +/+ (open circle, n = 18) and IP 3 R2 2/2 (filled circle, n = 24) were treated with 1S,3R-ACPD followed by PGE 2 and arteriole responses were measured. C. During continuous application of U-46619 the IP 3 R2 +/+ and IP 3 R2 2/2 were treated with 1S,3R-ACPD while another group of IP 3 R2 +/+ slices were treated with vehicle instead of 1S,3R-ACPD (red inverted triangle, n = 7). Inset shows the complete experiment from the time of application of U-46619. The dashed white box indicates the expanded graph. Treatment with U-46619 (100 nM) constricted arterioles of both IP 3 R2 +/+ (n = 16) and IP 3 R2 2/2 slices (n = 17) to a similar extent. **, P,0.01 comparing IP 3 R2 +/+ to IP 3 R2 2/ 2 10 min following 1S, 3R-ACPD application. doi:10.1371/journal.pone.0042194.g002 Figure 3. Neocortical arterioles of IP 3 R2 2/2 slices do not respond to electrical field stimulation. A. Responses of arterioles in 95% O 2 following electrical field stimulation. Cortical slices from IP 3 R2 +/+ (open circle) and IP 3 R2 2/2 mice (closed circle) were treated with electrical field stimulation (EFS) of 100 Hz trains of 200 ms at 0.2 Hz for 4 minutes as indicated by the dark bar. Arteriole diameter was measured every 5 min during the experiment. IP 3 R2 +/+ , n = 15; IP 3 R2 2/2 n = 15. **, P,0.01, IP 3 R2 +/+ compared to IP 3 R2 2/2 . B. Pretreatment with U-46619 for 30 min was followed by electrical stimulation and arteriole diameters of IP 3 R2 2/2 were compared to IP 3 R2 +/+ . IP 3 R2 +/+ , n = 15; IP 3 R2 2/2 n = 14. **, P,0.01; ***, P,0.001. C. Blockade of Group I mGluR with MPEP and JNJ prevents arteriole responses to electrical stimulation in IP 3 R2 +/+ slices (inverted triangles) while a 20% O2 environment does not alter responses of naïve IP 3 R2 +/+ or IP 3 R2 2/2 arterioles. IP 3 R2 +/+ , n = 14; IP 3 R2 2/2 n = 12, IP 3 R2 +/+ with MPEP/JNJ, n = 9. **, P,0.01; ***, P,0.001. doi:10.1371/journal.pone.0042194.g003 . Arteriole responses to mGluR agonist application are eliminated in cPLA 2 a 2/2 neocortical slices. A. Bath application of 1S,3R-ACPD 50 mM to slices equilibrated with 95% O 2 induced constriction of arterioles in cortical slices of cPLA 2 a +/+ (empty square, n = 29), but not cPLA 2 a 2/2 mice (filled square, n = 15). Washout of 1S,3R-ACPD was followed by application of 10 mM PGE 2 which caused identical constriction of arterioles in both genotypes. B. In slices at equilibrium with 95% O 2 , arterioles were preconstricted with 100 nM U-If increased astrocyte Ca 2+ is directly coupled to activation of cPLA 2 a then we predicted that cPLA 2 a would be necessary for the arteriole response to electrical stimulation in cortical brain slices. We subjected cortical slices, in 95% O 2 , from cPLA 2 a+/+ and cPLA 2 a 2/2 mice to the same electrical stimulation protocol used previously. The cPLA 2 a+/+ arterioles had a small but significant dilation while the cPLA 2 a 2/2 arterioles constricted slightly in response to the electrical stimulation ( Figure 5A) (20 min after stimulation, cPLA 2 a+/+, 6.162.5%, n = 23 compared to cPLA 2 a 2/2 , 22.861.1%, n = 23, P,0.01). To determine if the vascular tone of the arteriole impacts the response to electrical stimulation we applied a lower dose of U-46619, 20 nM, to the slice perfusate. This treatment caused arteriole constriction that was the same in cPLA 2 a+/+ and cPLA 2 a 2/2 slices ( Figure 5B). When electrical stimulation was applied to U-46619-treated cPLA 2 a+/+ slices there was a highly significant dilation response when compared to arterioles that were treated with U-46619 but not electrically stimulated ( Figure 5B). This response became significant 10 min following electrical stimulation and continued through the 30 minute measurement period. In contrast there was no response of the cPLA 2 a 2/2 arterioles to electrical stimulation when compared to unstimulated slices ( Figure 5B) (20 minutes after stimulation, cPLA 2 a+/+, 22.664.3%, n = 23; cPLA 2 a 2/2 , 230.364.3%, n = 23; and cPLA 2 a+/+, without electrical stimulation, 221.563.5%, n = 9; all diameters relative to the diameter at time of the onset of electrical stimulation. P,0.01 for cPLA 2 a +/+ compared to cPLA 2 a +/+ without simulation). The results thus far show that 1S,3R-ACPD (isolated mGluR activation) causes constriction in naïve cPLA 2 a+/+ slices while electrical stimulation (which includes neuron activation) causes dilation. Neurons must therefore modulate vascular responses by trans-cellular activation of the astrocyte mGluR and a second mGluR-independent mechanism. This mGluR-independent mechanism must also be cPLA 2 adependent since arterioles in cPLA 2 a 2/2 slices also failed to respond to electrical stimulation.
We wished to understand the role of cPLA 2 a in arteriole regulation following electrical stimulation. Neuronal nitric oxide synthase (nNOS) is thought to be the only isoform that contributes to metabolic hyperemia [20]. In addition, neuronal nitric oxide (NO) has been postulated to modulate cerebrovascular responses by inhibition of arachidonic acid metabolism [21]. Therefore, we applied either 10 mM Nv-propyl-l-arginine (L-NPA), a specific nNOS antagonist, or vehicle to cPLA 2 a +/+ slices equilibrated in 20% O 2 . Slices that were treated with vehicle dilated in response to electrical stimulation just as they had in 95% O 2 ( Figure 6). In contrast L-NPA-treated and cPLA 2 a 2/2 slices had a small constrictive response ( Figure 6). Thus it appears that nNO plays a role in the generation of the cPLA 2 a-dependent production of a vasodilator compound.

Discussion
We have used mutant mice to identify IP 3 R2 and cPLA 2 a as essential elements for the transduction of neuronal activity into vascular responses in an acute neocortical brain slice model. We found that IP 3 R2 and cPLA 2 a are necessary for both constrictive and dilatory responses following activation of the astrocyte mGluR by 1S,3R-ACPD. Furthermore electrical stimulation of brain slices caused arteriole dilation which was dependent on IP 3 R2, cPLA 2 a and neuron-derived NO. The increase in astrocyte intracellular Ca 2+ following application of 1S,3R-ACPD requires IP 3 R2 but not cPLA 2 a. These results are the first demonstration that the effector of mGluR-evoked Ca 2+ response in astrocytes is the IP3R2 and that this receptor and cPLA 2 a are essential for cortical cerebrovascular regulation.
Many of the biochemical steps involved in the neuron-toastrocyte-to-vascular smooth muscle cell signal transduction process have been defined. The sufficiency of astrocyte Ca 2+ transients to initiate arteriole responses in brain slices was demonstrated by photo-uncaging of Ca 2+ within astrocytes [8] and photolysis of caged IP 3 in retinal Mueller cells which triggered both a Ca 2+ response in the Mueller cell and vasodilation [11]. Thus, previous work is supportive of the model but photolysis may have had effects on cells other than the targeted astrocytes. Previously, use of IP 3 R2 knockout mice demonstrated that IP 3 R2 is required for the Ca 2+ responses of hippocampal astrocytes to a combination of chemical G-protein coupled receptor activators [15]. Because IP 3 R2 is the only IP 3 R isoform that is expressed in astrocytes [15], we reasoned that if astrocyte IP 3 is an essential effector of vascular responses to astrocyte mGluR activation then the signal must be transduced through the IP 3 R2. This reasoning is supported by the finding that IP 3 R2 2/2 cortical slices failed to respond to mGluR stimulation (Figures 123). Furthermore pharmacologic blockade of group I mGluRs prevented vascular responses to electrical stimulation in IP 3 R2 +/+ slices and had no effect on IP 3 R2 2/2 slices ( Figure 3C). IP 3 R2 may also be expressed in cerebrovascular endothelial cells and it is possible that endothelial IP 3 R2 contributes to the arteriole response to mGluR stimulation [23]. However, it is unlikely that results in the IP 3 R2 2/2 slices can be attributed to loss of endothelial IP 3 R2 because both blockade of mGluRs in IP 3 R2 +/+ slices and cPLA 2 a deficiency eliminated the responses to electrical stimulation ( Figures 3C and 5). The role of endothelial IP 3 R2 in vascular responses can be explicitly tested using endothelial denuded cerebral arterioles in future studies [24]. These results also demonstrate that the diacylglycerol produced by PLC activation is not sufficient to regulate arteriole responses.
Previous experiments have used non-specific inhibitors of various PLA 2 s to suggest that astrocyte Ca 2+ signaling activates PLA 2 and that this activity is required for cerebrovascular regulation [8,10]. For example inhibition of PLA 2 in intact mice by application of the drug MAFP to the cortical surface eliminated the response to photo-uncaging of Ca 2+ in astrocytes [25]. However, the particular molecular form(s) of PLA 2 needed for cerebrovascular regulation had not previously been determined. cPLA 2 a has been identified in astrocytes but other forms of PLA 2 are also expressed in astrocytes [26,27]. Our experiments show that the cPLA 2 a is the molecular species that is downstream from IP 3 R2 activation. We also examined the effect of cPLA 2 a expression on the astrocyte Ca 2+ response following 1S,3R-ACPD treatment. The results of the present Ca 2+ imaging experiments show that cPLA 2 a does not alter the IP 3 -mediated Ca 2+ response or impact Ca 2+ homeostasis in the astrocyte following efflux of Ca 2+ from the endoplasmic reticulum in the astrocyte. cPLA 2 a activity is regulated by phosphorylation and translocation to its membrane substrate. Translocation of cPLA 2 a to cellular membranes requires Ca 2+ binding to the C2 domain of the protein [28] while enzymatic activity is modulated by phosphorylation at sites that are not part of the C2 domain [29]. The local cellular Ca 2+ levels, lipid environment and the phosphorylation state of the protein play a role in determining the specific membrane compartment to which cPLA 2 a translocates [30]. When astrocyte Ca 2+ waves were initiated in brain slices by selective photo-uncaging of astrocyte Ca 2+ , peaks in endfoot Ca 2+ preceded those in astrocyte soma [8]. Our results demonstrate that mGluR activation causes Ca 2+ increases in the soma and the foot processes of astrocytes. Local elevation of endfoot Ca 2+ makes it possible that cPLA 2 a translocates to the cytosolic face of these membranes to hydrolyze arachidonic acid at this site. Localization of cPLA 2 a to the astrocyte perivascular endfoot could lead to increases in arachidonic acid concentration at the endfoot where it can be metabolized intracellularly or diffuse transcellularly into the neighboring vascular smooth muscle cell [8]. The membrane site of cPLA 2 a activity has the potential to profoundly impact the eicosanoids that are generated from arachidonic acid metabolism.
Our results are consistent with a model in which astrocyte cPLA 2 a generates the arachidonic acid that is metabolized within the astrocyte for production of prostaglandins and vasodilatory EETs. Astrocytes express COX-1 which metabolizes arachidonic acid to PGH 2 which is the precursor for all prostaglandins [25]. PGs, particularly PGE 2 , are implicated in the arteriole dilation Figure 6. Inhibition of nNOS with L-NPA prevents dilation of arterioles following electrical stimulation. cPLA 2 a +/+ slices were equilibrated in 20% O 2 and treated with ACSF (hollow square, n = 14; ) or ACSF with 10 mM L-NPA 60 min before electrical stimulation (filled red triangle, n = 9). cPLA 2 a 2/2 (filled square, n = 9) were treated with ACSF. Electrical stimulation was applied as indicated by the dark bar (electrical field stimulation, EFS). *, P,.05; **, P,.01. doi:10.1371/journal.pone.0042194.g006 Figure 7. Astrocyte Ca 2+ responses in neocortical slices to mGluR agonist application are not altered by absence of cPLA 2 a expression. Cortical brain slices from S100b-EGFP/cPLA 2 a +/+ (upper panel) and S100b-EGFP/cPLA 2 a 2/2 (lower panel) mice were loaded with the Ca 2+ -sensitive fluorophore Rhod-2/AM. Regions of interest representing astrocyte soma (white circles) and vascular foot processes (white box) were identified by EGFP expression and histologic location. Regions of interest representing background fluorescence for soma and endfeet are depicted by blue outlines. Ca 2+ fluorescence measured for the soma and endfeet are displayed at 3 times in relation to 1S,3R-ACPD treatment: (a) before, (b) at peak response and (c) after. Representative Ca 2+ measurements for soma and endfeet for each genotype are depicted in the right panel. The time of the 1S, 3R ACPD application is indicated by the black bar. The Ca 2+ responses of astrocyte populations are shown in B. Soma (+/+, n = 169; 2/2 , n = 166) and C. endfeet (+/+, n = 36; 2/2 , n = 33) and were measured as Ca 2+ peak amplitude, area under curve, half width, rise time, decay time or decay tau (as defined in Methods). The graphs show cumulative probability histograms analysis of the astrocyte populations by parameters response [9]. Cultured cortical astrocytes contain CYP450 epoxygenase activity that metabolizes arachidonic acid to vasodilatory EETs which can regulate cerebral blood flow by transcellular passage from the astrocyte foot process to the VSMC [31,32]. Cerebral vascular smooth muscle cells express CYP450 v-hydroxylase which metabolizes arachidonic acid to form 20-HETE [33]. The production of 20-HETE inhibits BK channels while increasing open probability of the L-type Ca 2+ channels of the VSMC, leading to VSMC and arteriole constriction [34]. When cPLA 2 a translocates to the vascular foot process arachidonic acid can diffuse into the VSMC where it is metabolized to vasconstrictive 20-HETE. In agreement with our results on 1S,3R-ACPD-treated, naïve brain slices, a non-specific PLA 2 inhibitor prevented the constrictive response following uncaging of Ca 2+ in mouse cortical brain slices [8]. Also consistent with our findings, treatment of rat brain slices with the combination of the thromboxane receptor agonist U-46619 and the non-specific PLA 2 inhibitor methyl arachidonyl fluorophosphanate (MAFP) significantly reduced the arteriole response to 1S,3R-ACPD treatment [10].
If the roles of IP 3 and cPLA 2 a in cerebrovascular regulation are simply to regulate the release of arachidonic acid in astrocytes for metabolism into both eicosanoid vasodilators and vasoconstrictors, what determines the response of the vessel? Our results provide some new insight into this process.
The initial diameter of an arteriole is determined by the balance between signals for constriction and dilation on the VSMC. Previous studies conducted in 95% O 2 with rat brain slices suggested that the initial arteriole diameter determines the directionality and magnitude of changes in diameter following stimulation [10]. Arterioles in naïve slices are not pressurized and therefore lack intrinsic tone and are relatively dilated. Therefore, an arteriole in a naïve slice may be more responsive to constrictive stimuli [8,35]. In contrast, pretreatment with U-46619 causes vasoconstriction that favors dilation following activation of the astrocyte [10]. As we observed, absence of the IP 3 R2-cPLA 2 a pathway abolishes the vasomotor effects of mGluR activation regardless of the resting diameter of the arteriole. This indicates that this signaling pathway is required for elaboration of both vasoconstrictors and dilators.
The balance of the metabolism of arachidonic acid between CYP450 and cyclooxygenase enzymes appears to be essential for neurovascular regulation [36,37]. Post-synaptic neurons also release NO and NO has been implicated in determining the metabolic balance between synthesis of EETs and 20-HETE [11]. We electrically stimulated slices in order to evaluate the effect of neuron activation on the cPLA 2 a and IP 3 R2-dependent vascular regulatory pathways. Electrical stimulation resulted in astrocyte Ca 2+ responses (Supporting Figure S2) and does not directly activate VSMC [38]. In our study, electrical stimulation caused dilation in both the naïve and U-46619 preconstricted arterioles while it had no vasomotor effect on IP 3 R2 2/2 or cPLA 2 a 2/2 arterioles. This implies that electrical activation of neurons (as compared to 1S,3R-ACPD activation of astrocytes) either increases the relative amount of a cPLA 2 a-dependent vasodilator or decreases the amount a of cPLA 2 a-dependent vasoconstrictor that is produced by activation of the astrocyte IP 3 R2. A mechanism for neuronal regulation of the astrocyte cPLA 2 adependent response was first suggested by the finding that NO inhibits the synthesis of EETS and 20-HETE [39]. Indeed, in rat retinas NO appears to be a determinant in the polarity of lightinduced vascular responses with NO production favoring vasoconstriction because epoxygenanse activity (metabolizes arachidonic acid to EET, a vasodilator) is inhibited at lower NO concentrations than is V-hydroxylase (metabolizes arachidonic acid to 20-HETE, a vasoconstrictor) [11]. We found that inhibition of neuronal NOS with L-NPA pre-treatment prevented the electrical stimulation-induced dilation in cPLA 2 a +/+ slices. We hypothesize that electrical stimulation causes neuronal NO release which inhibits production of 20-HETE in the VSMC [40]. In contrast, when nNOS is inhibited by L-NPA the V-hydroxylase is no longer blocked by NO and production of 20-HETE causes arteriole constriction that opposes the dilating effects of EETs and PGs. In agreement with our results Liu et al. (2008) tested the effect of NOS expression on functional hyperemia in the whisker barrel cortex of rats and concluded that increased NO production following whisker stimulation may suppress 20-HETE synthesis [21]. Taken together, our data indicate that the vascular response to astrocyte Ca 2+ signaling depends upon both the pre-existing tone of the arteriole and the particular signaling pathways that a stimulus triggers.
Our experiments also demonstrate that regulation of cerebral arteriole responses to vasoactive eicosanoids is dependent upon levels of tissue oxygenation. We used PGE 2 to demonstrate that arteriole responses remained intact at the conclusion of each series of measurements. As expected, the response to PGE 2 was independent of the IP 3 R2 or cPLA 2 a expression in the slice because PGE 2 is a downstream metabolite of arachidonic acid. To our surprise, 10 mM PGE 2 bath application caused arteriole constriction in the 95% O 2 and dilation in the 20% O 2 environment. This result is identical to the responses seen with activation of the mGluR and is consistent with a model in which the response to PGE 2 is dependent on the metabolic and oxidative state of the region. In other studies, in rat brain slices, similar doses of PGE 2 were described as an arteriole vasodilator [5,9]. In one of these studies the slices were maintained in an ACSF solution with 2.8 mM glucose in contrast to the 20 mM glucose used in our ACSF [5]. This lower glucose concentration could favor glycolysis and thus dilation [9]. In the other study the response to PGE 2 in high O 2 was not evaluated while the low O 2 response was identical to our result [9]. Because arteriole responses to PGE 2 are largely determined by activation of the eicosanoid receptors of the VSMC and vascular endothelial cells we hypothesize that the regional metabolic state could alter the binding of PGE 2 on eicosanoid receptors [41]. For example, in renal interlobular arterioles PGE 2 causes constriction by binding the prostaglandin E type 3 (EP3) receptor [42] and in rat aortic rings concentrations of PGE 2 .1 mM cause constriction through activation of the thromboxane A 2 receptor [43,44]. We tested the possibility that constriction in 95% O 2 was related to the high concentration of PGE 2 used in the experiment by performing a dose response titration. In the high O 2 environment 100 nM PGE 2 still caused arteriole constriction and dilation was not observed at any concentration (not shown). Thus it is possible that the vascular responses to PGE 2 are regulated by metabolic state in which high O 2 favors binding to prostaglandin E receptors (EP) that cause constriction (EP1 and EP3) relative to those that cause dilation (EP2 and EP4). This hypothesis requires further testing because expression of thromboxane and EP receptors in the brain microvasculature has not been characterized and the dependence compared between cPLA 2 a +/+ (open circles and bars) and 2/2 (closed circles and bars) while inset bar graph shows the mean 6 S.E.M. for each parameter. There were no significant differences between the genotypes. doi:10.1371/journal.pone.0042194.g007 on oxidative state of specific prostaglandin binding on receptors has not been explored [45].
Our work is largely consistent with previous findings using pharmacological inhibition of PLA 2 s and arachidonic acid metabolic enzymes. This is in contrast to a study in which a cPLA 2 a-deficient mouse had a normal in vivo circulatory response to sensory stimulation [46]. It has been proposed that in this in vivo study the compensatory mechanisms may overcome the permanent genetic loss of cPLA 2 a and maintain normal circulatory responses [38]. Our work demonstrates that this is not the case since there is no evidence of compensation in the vascular responses of our knock-out mice in the slice preparation. It is also important to note that the kinetics of the vascular response in the slice model are significantly slower than those measured in vivo. Our results are consistent with previously published work using similar slice conditions. For example, Gordon and colleagues found that in high O 2 concentration constriction was maximal 4 minutes after stimulation while dilation peaked 17 minutes after stimulation [9]. The slow kinetics of the slice model may be due to a number of factors. Vessel diameters are determined by the summation of constrictive and relaxing forces and the rates of change in diameter are likely due to the size of gradients in these forces. The lack of arteriole pressure in brain slice arterioles will decrease the dilation gradient and it is possible that contractile pressures are also reduced as the tissue of a slice can easily expand. Our results are also consistent with a pathway in which multiple sequential enzymatic steps are required to generate vasoactive compounds. It is possible that bath perfusion of the slices delays achieving the maximum concentration of arachidonic acid metabolites. Another possible reason for the difference between the in vivo and in slice models is the normal pH that we used to maintain slice health. The metabolic state of astrocytes impacts the polarity of vascular responses and regional acidosis appears to have a profound impact [9]. cPLA 2 a activity is sensitive to pH and it is possible that metabolic alterations that lower cellular pH could decrease cPLA 2 a activity within the astrocyte. Finally, it is likely that other regulatory pathways exist in vivo but not in slice. There are likely to be many other factors that influence cerebral vascular regulation.
The magnitude of the stimulated astrocyte Ca 2+ response may also determine whether an arteriole constricts or dilates [38]. We found that the presence or absence of cPLA 2 a had no apparent effect on the Ca 2+ response of astrocytes to 1S,3R-ACPD (Figure 7). Similarly the treatment of the slices with U-46619 did not cause any immediate Ca 2+ response in the astrocytes nor did it alter the relative Ca 2+ response of astrocytes in slices that were subsequently treated with 1S,3R-ACPD (not shown). This is consistent with an absence of thromboxane A 2 receptors in perivascular astrocytes [47]. Indeed while addition of U-46619 alters the resting tone and Ca 2+ responsiveness of the vascular smooth muscle cells of the arterioles [48] it appears to have little, if any effect upon the perivascular astrocytes. Therefore the changes in polarity of the arteriole response to 1S,3R-ACPD are unlikely to be due to changes in the concentration of Ca 2+ within the astrocyte. In this model, astrocyte cPLA 2 a generates arachidonic acid and the responses of the vascular system to this arachidonic acid release are determined by its metabolism and other factors.
While other investigations have used bath application of U-46619 to achieve an equilibrium arteriole diameter in brain slices [10] we were not able to replicate this result. Bath application with 100 nM U-46619 caused progressive arteriole constriction that did not equilibrate. In preliminary work we found that higher concentrations of U-46619 caused arteriole occlusion and prevented subsequent responses to chemical dilators and constric-tors. Application of lower concentrations of U-46619 slowed the constrictive response but did not achieve a stable arteriole diameter within the time frame of the experiments (not shown). In a model that continuously bath applies a dose of U-46619 that is ,5 fold above the EC 50 for thromboxane receptor occupancy it is not surprising that constriction progresses until the arteriole is completely constricted [49].
Neurons, smooth muscle, and endothelial cells also express cPLA 2 a [50][51][52] and because cPLA 2 a is globally deficient in the cPLA 2 a 2/2 mouse we must qualify our conclusions. While the results of this study are consistent with the model in which astrocyte cPLA 2 a is the generator of arachidonic acid mediators it remains possible that other cellular sources of cPLA 2 a are important in this signaling process. It will be necessary to create cell-line specific gene deletions to further test these possibilities.
We recognize that these experiments leave questions that can be answered by future investigations both in brain slices and in vivo.
Our results indicate that activation of the mGluR triggers Ca 2+ release through the IP 3 R2 receptor and that this increased Ca 2+ allows cPLA 2 a to release arachidonic acid which is metabolized to vasoactive metabolites. Other forms of PLA 2 can act synergistically with cPLA 2 a to amplify arachidonic acid release and lipid mediator generation [53] so while cPLA 2 a may be the first PLA 2 activity in the signaling pathway others may also be necessary. Synaptic activity and the resting tone of the VSMC influence the magnitude and direction of arteriole responses to stimulation and their precise interactions with cPLA 2 require further investigation. Importantly while cPLA 2 a blockade appears to be neuroprotective in excitotoxicity models the current results suggest that chemical inhibition of cPLA 2 a may significantly impair normal mechanisms of neurovascular regulation [54,55].

Slice Preparation and Imaging
Brains were removed from P20-35 mice after decapitation. Coronal slices of the somatosensory cortex (300 mm thick) were cut on a Leica VT1200S vibrating tissue slicer (Leica Biosystems, Richmond, IL) equipped with a sapphire blade in ice-cold cutting saline (in mM): 135 N-methyl-D-glucamine chloride (NMDG), 1 KCl, 1.2 KH 2 PO 4 , 0.5 CaCl 2 , 1.5 MgCl 2 , 24.2 Choline Bicarbonate, 13 glucose, adjusted to pH 7.4 and oxygenated with 95% O 2/ 5% CO 2 . Slices were then maintained in ACSF (in mM): 125 NaCl, 2.5 KCl, 1 NaH 2 PO 4 , 26.2 NaHCO 3 , 2.5 CaCl 2 , 1.3 MgCl 2 , 20 glucose (pH = 7.4), for at least 1 h at room temperature. For recording and imaging, slices were placed in a submerged chamber superfused with ACSF at a rate of 1-2 ml/ min at 34uC. A 10 minute period of baseline recording preceded brain slice stimulation. For experiments with 1S,3R-ACPD, this drug was bath applied at indicated concentrations for 10 min followed by a 10 min washout period. Following this, the responsiveness of the chosen arteriole was evaluated by the addition of PGE 2 . In experiments with U-46619 was added to the ACSF after the initial 10 min stabilization period at the indicated concentrations and applied continuously throughout the experiments. For electrical stimulation, a concentric bipolar electrode was placed 200-300 mm lateral to the arteriole of interest. The stimulation protocol consisted of 100 Hz monophasic pulse trains of duration 200 msec, with an intertrain interval of 5 sec for a total duration of 4 min.
Cell structure within cortical slices was visualized through a 40X water immersion objective with gradient contrast optics using a fixed-stage upright microscope equipped with a Zeiss Pascal confocal system with Argon ion (488 nm), and HeNe (543 nm) lasers. Arterioles in each slice were identified by their characteristic size (inner diameter of 5-20 mm) and the presence of a vascular smooth muscle layer. We selected arterioles from cortical layers 2-5 that could be observed for a minimum length of 200 mm and then measured changes in the inner diameter. For each arteriole a baseline image was obtained and we established between 5-10 reference lines across the arteriole lumen. These reference lines were spaced at ,5 mm intervals and were applied at the same axial location of the vessel for all subsequent radial measurements. At each time point one arteriole image was obtained and a second was obtained 30 seconds later. These images were digitally superimposed and radial measurements were taken from the resultant image. An investigator who was blinded to the experimental condition and genotype of the slice measured the arteriolar internal diameter for each time point.
After cutting, slices were incubated in ACSF saturated with 95% O 2 /5% CO 2 for 60 minutes. They were transferred and maintained in ACSF saturated with either 95% O 2 /5% CO 2 or 20% O 2 /5% CO 2 depending upon the experiment. For low O 2 experiments, slices were equilibrated in 20% O 2 saturated ACSF for at least 40 min before experiments. The switch from high O 2 to low O 2 caused a small constriction of vessels (1.5060.84%, P = 0.095, n = 17).
To eliminate NO generated by neurons 10 mM L-NPA, a highly selective nNOS inhibitor, was bath applied for 60 min before electrical stimulation. L-NPA had little effect on vessel tone (constriction:1.4661.93%, P = 0.47, n = 9). L-NPA was dissolved in water.
We used Rhod-2/AM as a cell-permeant indicator for Ca 2+ imaging experiments. It was dissolved in DMSO together with the detergent Pluronic F-127 and then diluted with HEPES-ACSF (in mM) (125 NaCl, 2.5 KCl, 1 NaH 2 PO 4 , 25 HEPES, 2.5 CaCl 2 , 1.3 MgCl 2 , 20 glucose, adjusted pH to 7.4) to a final concentration of 10 mM (final DMSO concentration: 0.23%). Because the Ca 2+ indicator Rhod-2/AM preferentially loads astrocytes [8] slices were incubated with Rhod-2/AM for 60-90 min at room temperature. Following loading, slices were maintained in ACSF. Astrocytes were selected for imaging on the basis of their uptake of Rhod-2 (or expression of EGFP), an ameboid-shaped cell body, a location in direct proximity to an arteriole, and the presence of a foot process in proximity to the arteriole. Rhod-2 was excited with 543 nm light while EGFP, marking a subpopulation of astrocytes of S100b-EGFP transgenic mice, was excited with 488 nm light. In this preparation 8263% of EGFP astrocytes were loaded with red Rhod-2/AM (335 of cells, 17 slices, 8 mice). Rhod-2 images were acquired at 1.3 Hz/frame and signals were expressed as DF/ F 0 = (F t -F 0 )/(F 0 -B 0 ), where F t is fluorescence intensity at any given time, F 0 is the average fluorescence intensity in the baseline period and B 0 is the average fluorescence intensity of background. Background values were taken from an adjacent region of interest (see Figure 7). For analysis of Ca 2+ transients the 10-90% rise time and 90-10% decay times were calculated. Group data were expressed as mean 6 SEM and compared by Student's t-test.

Mice
Mice were housed with a 12-hour diurnal light cycle and free access to food and water. All genetically altered mice used for experiments were produced by mating male and female mice that were heterozygous for the gene of interest. In these studies we used cPLA 2 a +/+ and cPLA 2 a 2/2 mice [54] that had been backcrossed on the BALB/C strain for .10 generations. Mice that were previously engineered to express a transgene for the EGFP protein under the control of the S100b promoter [22] were bred for greater than 6 generations with F1 progeny of BALB/c x cPLA 2 a 2/2 mating to create S100b-eGFP-cPLA 2 a +/2 mice. IP 3 R2 2/2 mice were originally supplied on the Swiss Webster background (Ju Chen, personal communication) and were mated with BALB/c WT mice to generate IP 3 R2 +/2 mice [56]. All genotyping was performed from tail samples on mice between the ages of 8-12 days and were analyzed by PCR using specific primer pairs.

Ethics
All studies were conducted with the approval of the Johns Hopkins University Animal Care and Use Committee under the protocol numbers MO07M135 and MO10M69. Performance of the studies was also in accordance with the guidelines of the National Institutes of Health and the National Research Council.

Data Analysis
Changes in Rhod-2 fluorescence were analyzed using IGOR Pro 6 (Wavemetrics, Inc. Portland, OR) and expressed as cumulative probability histograms and mean values 6 S.E.M. Arteriole diameters were measures using NIH Image J (NIH, Bethesda, MD) and expressed as mean values 6 S.E.M. Data between groups were compared by Student's t-test.  [54]. IP 3 type-2 receptor-deficient mice (IP 3 R2 2/2 ) were the gift of Ju Chen (University of California, San Diego, CA) [56]. Mice expressing the EGFP protein under the control of the S100b promoter were originally created in the laboratory of Legraverend (Institut de Génomique Fonctionnelle, Montpellier, France) [22] and were provided by Dwight E. Bergles (Johns Hopkins University, Baltimore, MD). Figure S1 Change in diameter of cortical arterioles upon sequential, combined exposure to U46619 and PGE 2 . Cortical brain slices from IP 3 R2 +/+ mice were at equilibrium with 95% O 2 , and treated with 100 nM U-46619 supplemented ACSF for 30 min. After 30 min ACSF was further supplemented with 10 mM PGE 2 for an additional 10 min. n = 9 arterioles. (TIF) Figure S2 Change in diameter of a single arteriole upon sequential, combined exposure to U46619 and 1S, 3R-ACPD. Time is expressed in minutes with the t = 0 set at the initiation of U46619 and the initial diameter at t = 210 minutes. Bars indicate the time of bath application of 100 nM U46619 or 50 mM 1S, 3R-ACPD. (TIF) Figure S3 Electrical stimulation evokes Ca 2+ transient in astrocytes of a cortical slice derived from an S100b-EGFP mouse. Slices were loaded with Rhod-2/AM and a concentric bipolar electrode was placed 200-300 mm from the region of interest. Rhod-2 fluorescence (red) in multiple astrocyte cell bodies that express EGFP (green) (circled in white; left panel) was measured after stimulation at 100 Hz for 200 ms (expanded black bar, right panel). The Ca 2+ fluorescence signals of individual astrocytes are plotted. Scale bar: 20 mM. (TIF)