Epoxyeicosanoids as mediators of neurogenic vasodilation in cerebral vessels

Abstract

Epoxyeicosatrienoic acids (EETs) are potent vasodilators produced from arachidonic acid by cytochrome P-450 (CYP) epoxygenases and metabolized to vicinal diols by soluble epoxide hydrolase (sEH). In the brain, EETs are produced by astrocytes and the vascular endothelium and are involved in the control of cerebral blood flow (CBF). Recent evidence, however, suggests that epoxygenases and sEH are present in perivascular vasodilator nerve fibers innervating the cerebral surface vasculature. In the present study, we tested the hypothesis that EETs are nerve-derived relaxing factors in the cerebral circulation. We first traced these fibers by retrograde labeling in the rat to trigeminal ganglia (TG) and sphenopalatine ganglia (SPG). We then examined the expression of CYP epoxygenases and sEH in these ganglia. RT-PCR and Western blot analysis identified CYP2J3 and CYP2J4 epoxygenase isoforms and sEH in both TG and SPG, and immunofluorescence double labeling identified CYP2J and sEH immunoreactivity in neuronal cell bodies of both ganglia. To evaluate the functional role of EETs in neurogenic vasodilation, we elicited cortical hyperemia by electrically stimulating efferent cerebral perivascular nerve fibers and by chemically stimulating oral trigeminal fibers with capsaicin. Cortical blood flow responses were monitored by laser-Doppler flowmetry. Local administration to the cortical surface of the putative EET antagonist 14,15-epoxyeicosa-5(Z)-enoic acid (30 μmol/l) attenuated CBF responses to electrical and chemical stimulation. These results suggest that EETs are produced by perivascular nerves and play a role in neurogenic vasodilation of the cerebral vasculature. The findings have important implications to such clinical conditions as migraine, vasospasm after subarachnoid hemorrhage, and stroke.

epoxyeicosatrienoic acids (EETs) are potent vasodilators and important regulators of vascular function that are synthesized from arachidonic acid by cytochrome P-450 (CYP) epoxygenase enzymes of the CYP2C and CYP2J families (10, 24, 33). The enzyme soluble epoxide hydrolase (sEH) is the dominant route of EET degradation, catalyzing their metabolism to dihydroxyeicosatrienoic acids (9). In the brain, EETs are known to be released from astrocytes and the vascular endothelium (3, 18). They directly dilate cerebral arteries (4, 8) and exert regulatory control over several facets of vascular function, including angiogenesis (19, 34) and platelet adhesion (14). EETs are key regulators of cerebral blood flow (CBF), mediating neurovascular coupling in the intact brain (2, 20, 21). Finally, EETs play an important role in the blood flow response to cerebral ischemia. Inhibition of sEH results in a reduction of ischemic damage in a mouse model of experimental stroke and is associated with elevated cortical blood flow during vascular occlusion (35, 36).

Our group (16) recently reported the presence of sEH immunoreactivity in perivascular nerve fibers innervating the cortical surface vasculature including the middle cerebral artery (MCA) and basilar artery. This “extrinsic” perivascular innervation of the cerebral vasculature includes two populations of vasodilator nerves: parasympathetic fibers originating in the sphenopalatine ganglia (SPG) and otic ganglia and sensory afferents that project to the trigeminal ganglia (TG) (13). Immunofluorescence double labeling revealed that sEH immunoreactivity colocalized with markers for both the parasympathetic and sensory vasodilator fibers (16). Despite the presence of sEH in these nerves, it remains unclear whether EETs contribute to the vasomotor actions of these perivascular fibers.

In the present study, we proposed that neuron-derived EETs are involved in the regulation of CBF by these perivascular vasodilator fibers. To test this novel neurogenic hypothesis, we traced these perivascular nerve fibers to their source ganglia, identifying the expression of specific CYP epoxygenase isoforms and sEH within these neuronal populations. The functional involvement of nerve-derived EETs in the control of CBF was then examined with in vivo models of neurogenic cerebrovasodilation, including electrical stimulation of the efferent cerebral perivascular vasodilator nerve supply as well as chemical stimulation of sensory trigeminal fibers.

MATERIALS AND METHODS

The present experiments were conducted in accordance with National Institutes of Health guidelines for the care and use of animals in research under protocols approved by the Institutional Animal Care and Use Committee of Oregon Health and Science University.

Characterization of CYP epoxygenase and sEH expression in TG and SPG.

To determine the specific CYP epoxygenase isoforms expressed in TG and SPG, an RT-PCR screen was conducted against eight known rat epoxygenase isoforms from the CYP2C and CYP2J families (24) in addition to sEH using whole tissue from TG and SPG. Four male Wistar rats (250–300 g, Charles River) were deeply anesthetized with 5% isofluorane and transcardially perfused with cold heparinized saline (1 U/ml). Bilateral TG and SPG were dissected, and total tissue RNA was extracted per manufacturer's guidelines using a RNAqueous-4PCR kit (Ambion). Whole tissue poly-A RNA was reverse transcribed according to the manufacturer's protocol with a RETROscript reverse transcription kit (Ambion) using oligo(dT) primers. PCR (30 cycles with an annealing temperature of 59°C) was conducted using a GeneAmp PCR System 9700 (Applied Biosystems) with primers designed using online primer design software (Integrated DNA Technologies), with the specificity of the sequences being confirmed by a BLAST search. PCR products were run on a 2% agarose gel and visualized with 0.5 μg/ml ethidium bromide read on a Typhoon Trio multimodal scanner (Amersham). For each transcript, replicate PCRs were conducted from two separate animals (n = 2 biological replicates). Negative controls included the omission of PCR template material and whole cell (no reverse transcriptase) RNA in the place of cDNA. Positive controls included PCR upon liver cDNA, in which epoxygenase isoforms and sEH are known to be abundantly expressed.

Expression of CYP2J and sEH proteins in TG and SPG was confirmed by Western blot analysis. TG, SPG, and liver tissue were dissected from two saline-perfused rats and mechanically dissociated. Crude cellular extracts (microsomal and cytosolic fractions, 30 μg) were electrophoresed at 200 V for 105 min on SDS-10% polyacrylaminde gels and then transferred to polyvinylidene difluoride membranes for 120 min at 40 V. Membranes were washed 5% BSA for 1 h at room temperature and then immunoblotted with a rabbit polyclonal antibody raised against the peptide sequence FNPDHFLENGQFKKRE from human CYP2J2 (anti-CYP2J2pep4, 1:1,000), which cross-reacts with both rat CYP2J isoforms (CYP2J3 and CYP2J4). Membranes were also blotted with rabbit anti-sEH (1:500, Santa Cruz Biotechnology) polyclonal antibody. Secondary detection was conducted with Cy5-conjugated goat anti-rabbit secondary antibodies (Amersham) for 1 h at room temperature. Membranes were imaged with a Typhoon TRIO polymodal scanner (Amersham). Replicate blots (n = 3) were conducted on samples from two animals.

Cell type-specific expression of CYP2J and sEH proteins was determined by immunofluorescence double labeling of paraffin-embedded TG and SPG tissue. To localize CYP2J and sEH immunoreactivity in TG and SPG, thin ganglia slices (6 μm) were deparafinized, heated in 0.5 mmol/l sodium citrate buffer (pH 6.0) for antigen retrieval, and blocked with 3% normal donkey serum, 0.1% Triton X-100, and 1% BSA in PBS for 30 min at room temperature. Sections were incubated at 4°C overnight with one or two of the following primary antibodies: rabbit anti-CYP2J2_rec (1:100) (30), rabbit anti-sEH (1:100, Santa Cruz Biotechnology), mouse anti-neuronal nuclei (NeuN; 1:1,000), rabbit anti-fluorogold (1:5,000), sheep anti-neuronal nitric oxide (NO) synthase (nNOS; 1:1,000), guinea pig anti-VIP (1:1,000), guinea pig anti-substance P (SP; 1:000, Millipore), and goat anti-CGRP (1:1,000; Serotec). Secondary detection was conducted with donkey anti-rabbit Alexa fluor-488, donkey anti-mouse Alexa fluor-594, donkey anti-sheep Alexa fluor-594, goat anti-guinea pig Alexa fluor-594, and donkey anti-goat Alexa fluor-594 (1:800, Invitrogen) incubated at room temperature for 1 h. Sections were mounted with ProLong Antifade Gold with 4′,6-diamidino-2-phenylindole (Invitrogen), coverslipped, and observed by conventional fluorescence microscopy (Leica). Colocalization was quantified by counting positively labeled cells from six bilateral TG and SPG slices from each of two separate rats.

In cases where colabeling was used between two primary antibodies from rabbit hosts, labeling was conducted sequentially as follows: the first primary antibody was incubated as described above overnight at 4°C, this primary antibody was incubated with an excess of monovalent goat anti-rabbit Fab fragments (1:10, Jackson Immunoresearch) for 1 h at room temperature, and secondary detection of the Fab fragment was conducted with donkey anti-goat Alexa fluor-594 secondary antibody (1:800, Invitrogen) for 1 h at room temperature. The second primary antibody was then incubated for 1 h at room temperature, followed by secondary detection with donkey anti-rabbit Alexa fluor-488 antibody (1:800, Invitrogen).

Negative controls included both the omission of primary antibodies (n = 1) and preabsorption of rabbit anti-CYP2J2_rec and rabbit anti-sEH antibodies with a molar excess of recombinant human CYP2J2 and recombinant human sEH protein, respectively (n = 1). In addition, to confirm the specificity of the observed labeling, slices were incubated with separate anti-CYP2J and anti-sEH antibodies raised against distinct antigens. These included two antibodies raised against synthetic peptides from mouse CYP2J6 (QMEQNIMNRPLSVMQ, anti-CYP2J6pep1, n = 1) and CYP2J9 (GQARQPNLADRD, anti-CYP2J9pep2, n = 1) isoforms, which cross-react with rat CYP2J4 and CYP2J3 isoforms, respectively (31). Furthermore, an anti-sEH antibody raised against recombinant human sEH (n = 1) (6) and another antibody raised against a distinct peptide sequence of human sEH (DMKGYGESSAPPEIE, Cayman, n = 1) were used.

Determination of the origin of fibers projecting to the cerebral circulation.

Retrograde fluorogold nerve tracing was employed to confirm that nerve fibers originating in the TG or SPG project to the cerebral circulation. Animals were anesthetized with 5% isofluorane and placed in a stereotaxic frame. An incision was made in the midline of the scalp, and the skull was exposed. A small burr hole of 5 mm in diameter was drilled lateral of the saggital suture in the area of the MCA, with care being taken to leave the dura intact. The dura was incised with a corneal scalpel over the MCA, and 3 μl of 2% fluorogold was injected into the subdural space with a Hamilton syringe. After the injection, the syringe was left in place for 10 min to prevent spillage of the tracer from the subdural space. Sterile gel foam was placed over the burr hole, and the scalp was sutured. Animals survived for 8 days postinjection, after which they were killed. Perfusion fixation, dissection, and processing of tissue were as described above [n = 4 injected animals and 2 negative control (not injected) animals].

Determination of the role of EETs in neurogenic vasodilation in the cerebral circulation.

To define the role of EETs in the regulation of the cerebral surface vasculature by perivascular vasodilator nerves, we used as a model of neurogenic vasodilation: the electrical stimulation of efferent nerve fibers entering the ethmoidal foramen, as described by Ayajiki et al. (5). Briefly, the cerebral surface vasculature was exposed using an open cranial window over the MCA territory of the parietal cortex. Animals were anesthetized with α-chloralose and urethane (50 and 500 mg/kg ip, respectively). The femoral artery and vein were cannulated for the measurement of blood pressure, blood gas values, and delivery of drugs. Animals were tracheotomized, paralyzed with d-tubocurarine (1 mg/kg), and mechanically ventilated to control arterial blood gas values. The animal's head was fixed in a stereotaxic frame, the skull over the left parietal cortex in the MCA territory was exposed, a craniotomy was performed, and the dura was incised and reflected to provide direct access to the cortical surface vasculature. Cortical blood flow was monitored by laser-Doppler flowmetry (LDF; Moor Instruments), with the probe positioned over a cortical surface artery in the MCA territory. The cranial window was superfused with artificial cerebrospinal fluid (aCSF), and the ethmoidal nerve bundle containing parasympathetic efferent fibers from the SPG and sensory afferents projecting to the TG was isolated and stimulated with a fine bipolar stimulating electrode (1-ms pulse at 10 V) for 30 s. The resulting changes in CBF and mean arterial blood pressure (MABP) were monitored. A frequency-response curve (5, 10, and 20 Hz) was constructed, and the effect of the putative EET antagonist 14,15-epoxyeicosa-5(Z)-enoic acid (14,15-EEZE; 30 μmol/l, n = 8) (11), administered to the cortical surface, upon the neurogenic hyperemic response was determined.

To confirm the specificity of the putative EET antagonist 14,15-EEZE in blocking EET-mediated signaling, we directly administered increasing concentrations of 14,15-EET (10 nmol/l--10 μmol/l) to the cortical surface and recorded the resulting increase in CBF in both the presence and absence of 14,15-EEZE (30 μmol/l, n = 4). In these experiments, the effect of 14,15-EEZE upon the CBF response to transient hypercapnia was likewise determined both before and after drug treatment. Mild transient hypercapnia was induced by ventilating the rat with a 5% CO₂-30% O₂-65% N₂ gas mixture for 3 min, during which the CBF response was monitored.

Chemical stimulation of oral trigeminal fibers.

A second, more “physiological” model of neurogenic cerebrovasodilation was evoked through stimulation of oral trigeminal nociceptor fibers, as previously described by Gottselig and Messlinger (12). Cumulative doses of the transient receptor potential vanilloid (TRPV)1 agonist capsaicin (300 μl; 10 μmol/l–1 mmol/l) were instilled orally at 10-min intervals, and the resulting cortical hyperemia was measured by LDF. To determine the role of EET signaling in the neurogenic regulation of the cortical vasculature, pharmacological inhibition of portions of the EET signaling system was used. In each case, capsaicin dose-response trials were repeated within the same animal in both the absence and presence of the pharmacological agent. To control for sensitization or desensitization of the capsaicin-evoked hyperemic response, half of the trials from each group began with the vehicle treatment arm followed by the drug treatment arm. In the other half of the trials, the order was reversed. The two treatment arms were separated by a 1-h period, during which the mouth was liberally irrigated with saline. Additionally, in a separate group of animals, capsaicin dose-response trials were repeated with the same 1-h period between arms to detect any measurable sensitization or desensitization of this response (n = 4).

The putative EET antagonist 14,15-EEZE (30 μmol/l, n = 8) or the CYP epoxygenase inhibitor N-methylsulfonyl-6-(2-propargyloxyphenyl)hexanamide (MS-PPOH; 20 μmol/l, n = 8) was perfused for 45 min individually into each of two groups of animals. The direct effects of these agents on CBF were determined. To confirm that the capsaicin-evoked CBF response was in fact mediated by cortical perivascular vasodilator nerves originating in the parasympathetic SPG or sensory TG, the efferent parasympathetic and afferent sensory nerve supply to the anterior circle of Willis was surgically cut as it entered the ethmoidal foramen. After 1 h of recovery time, the effect of nerve lesion upon the hyperemic response to oral trigeminal fiber stimulation was measured (n = 4).

Optical microangiography.

To confirm that the cortical hyperemic response to oral capsaicin administration was in fact the result of cerebral vasodilation rather than an associated elevation in perfusion pressure, in a limited study (n = 1) we used a novel optical coherence tomography-based imaging technique termed optical microangiography (OMAG) (28, 29), which is capable of resolving three-dimensional distribution of dynamic blood perfusion at the capillary level in vivo. Briefly, the cortex was illuminated at 1,310 nm, with the resulting backscattered and reference light being detected to produce spectral interferograms within the OMAG system. Volumetric imaging data were collected by scanning the probe beam through a 1,000 × 500 × 512-voxel cube, representing 2.5 × 2.5 × 2 mm³ (x-y-z) of tissue. For high-temporal-resolution scans, continuous x-z plane scans were performed on a selected segment of the MCA. Imaging of the left MCA was conducted through a thinned skull. Capsaicin (10 μmol/l) was administered orally as described above, and the resulting hyperemic response was imaged through a thinned skull, including changes in MCA diameter, MCA blood velocity, and MCA volume rate (flow).

Chemicals.

Recombinant human CYP2J2 was synthesized as previously described (30). Recombinant human sEH was purchased from the Proteintech Group. The composition of aCSF was as follows (in meq/l): 156.5 Na⁺, 2.95 K⁺, 2.50 Ca²⁺, 1.33 Mg²⁺, and 24.6 HCO₃⁻ with 66.5 mg/dl dextrose and 40.2 mg/dl urea. Capsaicin (vehicle: 0.1% ethanol in saline), d-tubocurarine, α-chloralose, and urethane were purchased from Sigma. 14,15-EET (vehicle: 0.1% ethanol in aCSF), 14,15-EEZE (vehicle: 0.1% ethanol in aCSF), and MS-PPOH (vehicle: 0.1% ethanol in aCSF) were purchased from Cayman Chemical.

Statistics.

All data are expressed as means ± SE. Variations among animals in MABP and LDF were normalized by reporting them as percent changes from the baseline value [100% × (final value − baseline value)/baseline value]. Data and statistical analysis were performed with GraphPad Prism 4 (Graphpad Software). Comparisons of data were made with paired or unpaired Student's t-test or with ANOVA followed by a post hoc Bonferroni procedure. P values of <0.05 were considered significant.

RESULTS

CYP2J is specifically expressed in TG and SPG neuronal tissue.

We sought to determine which CYP epoxygenase isoforms are present in TG and SPG, which house the cell bodies of sensory and parasympathetic fibers that innervate the cerebral surface vasculature (13). A RT-PCR screen using primers against eight known rat CYP epoxygenase isoforms in addition to sEH (Fig. 1B) identified that mRNA for two CYP2J isoforms, CYP2J3 and CYP2J4, and sEH were detectible in both TG and SPG tissue (Fig. 1A). Pronounced expression of all CYP epoxygenase transcripts was observed in the liver, the positive control tissue for the PCRs (24). All bands were of the predicted size. No PCR products were detected in either negative control group: the RT-PCR with no RNA template and the PCR with no reverse transcriptase.

Fig. 1.

Cytochrome P-450 (CYP)2J isoforms are specifically expressed in sphenopalatine ganglia (SPG) and trigeminal ganglia (TG) tissue. A: RT-PCR screen of TG and SPG revealed CYP2J3, CYP2J4, and soluble epoxide hydrolase (sEH) mRNA expression. No expression of CYP2C isoforms was detected. All transcripts were detected in positive control liver tissue, whereas template-negative and reverse transcriptase-negative (−RT) controls produced no PCR products. B: primer sequences used for each transcript including National Center for Biotechnology Informatics Accession Numbers, forward (→) and reverse (←) primer sequences, primer positions, and predicted PCR product sizes (in bp). C: Western blot analysis of CYP2J expression in TG and SPG tissue. Immunoblot analysis with an antibody raised against human CYP2J2 (anti-CYP2J2pep4) resulted in two 50- to 60-kDa bands, corresponding to CYP2J3 and CYP2J4 isoforms. D: immunoblot analysis with an antibody raised against a peptide fragment of human sEH revealed a single 60-kDa band within both TG and SPG.

RT-PCR data were confirmed with Western blot analysis of TG and SPG tissue. Using a polyclonal antibody raised against a fragment of human CYP2J2 that reacts with both rat CYP2J3 and CYP2J4, two distinct bands were observed in the 50- to 60-kDa size range in the TG, SPG, and positive control liver tissue (Fig. 1C). These bands likely correspond to the CYP2J3 (58 kDa) and CYP2J3 (55 kDa) proteins identified by RT-PCR. Blotting with anti-sEH antibody revealed a single band in the 60-kDa size range in the TG, SPG, and positive control liver tissue. These data corroborate our previous immunofluorescence findings in the cerebral vasculature (16) and identify a specific family of CYP epoxygenases (CYP2Js) expressed within the neuronal tissue of TG and SPG.

CYP2J and sEH proteins are present in TG neurons.

We conducted immunofluorescence double labeling of TG slices with antibodies raised against recombinant human CYP2J2 [which cross-reacts with rat CYP2J3 and CYP2J4 (30)] and sEH to determine the cell type-specific expression of these proteins in TG. CYP2J2 immunoreactivity was broadly expressed throughout TG in cell bodies with neuronal morphology, including those of small, medium, and large diameters (Fig. 2A). Colabeling with the neuron-specific marker NeuN demonstrated near-complete overlap between CYP2J2 and NeuN immunoreactivity (Fig. 2, B, C, and M). Virtually every CYP2J2-positive cell was NeuN positive (97%), and, conversely, nearly all NeuN-positive cells were CYP2J2 positive (94%). Little CYP2J2 immunoreactivity was observed outside NeuN-labeled cell bodies, suggesting that CYP2J proteins are not significantly expressed in either Schwann or satellite cells. Where blood vessels were seen within the ganglia, endothelial CYP2J2 immunoreactivity was also observed (data not shown). To determine the cellular phenotype of TG neurons expressing CYP2J proteins, double labeling with antibodies against CGRP and SP was carried out. Expression of CGRP and SP was observed in a fraction of trigeminal neurons, of which virtually all were CYP2J2 positive (Fig. 3, A and B). Of the CYP2J2 positive TG cells, a corresponding fraction (22% and 16%) were immunoreactive for CGRP or SP, respectively (Fig. 2M).

Fig. 2.

CYP2J and sEH immunoreactivity localize to TG neurons. A: CYP2J2 immunoreactivity (green) was widely distributed throughout TG. B and C: double labeling against the neuronal marker neuronal nuclei (NeuN; red) indicated that CYP2J2 immunoreactivity (green) was restricted primarily to TG neurons (arrows). D: labeling with anti-CYP2J2_rec antibody after preabsorption with recombinant human CYP2J2_rec protein (+rp) revealed no apparent CYP2J2 immunoreactivity (green). E: sEH immunoreactivity (green) was widely distributed in TG. F and G: double labeling against NeuN (red) revealed sEH immunoreactivity (green) in most (arrows), but not all (white arrowheads), neurons. sEH immunoreactivity was also apparent in non-neuronal cells of satellite (outlined arrowheads) and Schwann cell morphology (not shown). H: labeling with anti-sEH antibody after preabsorption with recombinant human sEH protein (+rp) resulted in no sEH immunoreactivity (green). I–K: double labeling with antibodies against sEH and CYP2J2 demonstrated that sEH immunoreactivity (green) and CYP2J2 immunoreactivity (red) colocalized within the same TG neurons (arrows), whereas sEH immunoreactivity appeared to be additionally present in satellite (outlined arrowheads) and Schwann cells. L: omission of primary antibodies (−1° Ab) from the labeling protocol resulted in no nonspecific labeling. M: quantification of colocalization between CYP2J2 and sEH immunoreactivity, the neuronal marker NeuN, or the neuropeptides CGRP or substance P (SP). Values represent the percentage of CYP2J2- or sEH-immunoreactive cells also expressing NeuN, CGRP, or SP. Values in parentheses represent the percentage of labeled neurons (NeuN-positive cells) also expressing CYP2J2 or sEH immunoreactivity. DAPI, 4′,6-diamidino-2-phenylindole. Scale bars = 60 μm in A, D, E, H, and L and 30 μm in B, C, F, G, and I–K.

Fig. 3.

Characterization of the cellular phenotype of CYP2J2- and sEH-immunoreactive TG and SPG neurons. A fraction of CYP2J2-immunoreactive (green) neurons in TG were immunoreactive for the neuropeptides CGRP (red; A) and SP (red; B) (arrows), whereas virtually all CGRP- and SP-positive cells were CYP2J2 immunoreactive. Similar patterns of coexpression were observed in TG slices labeled against sEH (green) and CGRP (red; C) or SP (red; D). In SPG, most CYP2J2-immunoreactive (green) neurons colocalized with neuronal nitric oxide (NO) synthase (nNOS; red; E), whereas only a fraction of CYP2J2-immunoreactive (green) neurons were VIP positive (red; F). Similar patterns of coexpression were observed in SPG slices labeled against sEH (green) and nNOS (red; G) or VIP (red; H). Most sEH-immunoreactive neurons colocalized with nNOS, whereas only a fraction were VIP positive. Scale bars = 30 μm.

TG expression of sEH was likewise interrogated by immunofluorescence double labeling. sEH immunoreactivity was observed throughout the ganglia, localizing to cell bodies of neuronal morphology (Fig. 2E). Compared with CYP2J2, sEH immunoreactivity was of a lower intensity, was less universally neuronal, and appeared to include cells of both Schwann and satellite cell morphology (Fig. 2, E–G and M). Localization of sEH immunoreactivity in CGRP- and SP-positive neurons also displayed a similar pattern to CYP2J2 immunoreactivity (Fig. 3, C and D); most CGRP- and SP-positive neurons were sEH positive, whereas only a portion of sEH-positive cell bodies exhibited immunoreactivity for these two neuropeptides (Fig. 2M).

To determine whether CYP2J and sEH proteins are expressed within the same individual neurons, colabeling of TG slices with antibodies against both CYP2J2 and sEH was conducted (Fig. 2, I–K). It was generally observed that CYP2J2 and sEH immunoreactivity were present in the same neurons. However, as noted above, expression of sEH appeared more irregular and not solely restricted to neuronal cell bodies.

When primary antibodies were omitted from the double-labeling protocol, no labeling of TG was observed (Fig. 2L). Likewise, when anti-CYP2J2_rec and anti-sEH antibodies were preabsorbed with a molar excess of recombinant human CYP2J2 or sEH, no CYP2J2 or sEH immunoreactivity was observed within TG (Fig. 2, D and H). To further confirm the specificity of the anti-CYP2J2_rec antibody, TG slices were labeled with two additional antibodies against distinct CYP2J antigens: one against a polypeptide sequence of mouse CYP2J6 that cross-reacts with rat CYP2J4 and another against a polypeptide sequence of mouse CYP2J9 that cross-reacts with rat CYP2J3 (31). Both of these antibodies displayed similar labeling patters to the anti-CYP2J2_rec antibody (data not shown), although minor differences in the cellular and subcellular distribution of CYP2J6 and CYP2J9 immunoreactivity were observed. To confirm the specificity of the anti-sEH antibody, labeling of TG slices was carried out with two additional anti-sEH antibodies raised against distinct antigens. Labeling with a rabbit polyclonal antibody against recombinant human sEH (6) or a rabbit polyclonal antibody against a distinct polypeptide sequence of human sEH (Cayman Chemical) resulted in immunoreactivity very similar to that of the Santa Cruz Biotechnology antibody (data not shown). Based on these findings, we conclude that the CYP2J2 and sEH immunoreactivity observed in TG neurons was specific.

CYP2J and sEH proteins are present in SPG neurons.

Expression of CYP2J and sEH was examined in parasympathetic SPG by immunofluorescence double labeling. As in TG, probing SPG with the anti-CYP2J2_rec antibody labeled primarily cells of neuronal morphology (Fig. 4A). Colabeling with the neuronal marker NeuN revealed that virtually all CYP2J2 immunoreactivity (97%) was restricted to neuronal cell bodies, whereas most neurons (91%) were CYP2J2 immunoreactive (Fig. 4, B, C, and G). Colabeling with antibodies against nNOS or VIP revealed that most (82%) CYP2J2-immunoreactive cells were also nNOS immunoreactive (Figs. 3E and 4G), whereas only a portion of them (25%) appeared to express the neuropeptide VIP (Figs. 3F and 4G).

Fig. 4.

CYP2J and sEH immunoreactivity localize to SPG neurons. A: CYP2J2 immunoreactivity (green) was distributed throughout SPG. B and C: double labeling against the neuronal marker NeuN (red) indicated that CYP2J2 immunoreactivity (green) was restricted primarily to SPG neurons. D: sEH immunoreactivity (green) in SPG was widely distributed throughout the ganglia. E and F: double labeling against NeuN (red) revealed sEH immunoreactivity in most neurons (arrows). Less intense sEH immunoreactivity (green) was also apparent in glial cells (outlined arrowheads). G: quantification of colocalization between CYP2J2 and sEH immunoreactivity, the neuronal markers NeuN or nNOS, or the neuropeptide VIP. Values represent percentages of CYP2J2- or sEH-immunoreactive cells also expressing NeuN, nNOS, or VIP. Values in parentheses represent percentages of labeled neurons (NeuN-positive cells) also expressing CYP2J2 or sEH immunoreactivity. Scale bars = 60 μm in A and D and 30 μm in B, C, E, and F.

Immunofluorescence labeling of sEH in SPG revealed that sEH was expressed throughout SPG, including dominant localization in cells with neuronal morphology as well as less intense immunoreactivity in apparent glial cells (Fig. 4D). Double-labeling experiments with the neuronal marker NeuN revealed that most (92%) sEH-immunoreactive cell bodies were labeled as neurons, whereas a smaller portion of the overall NeuN-positive neuronal pool (70%) were sEH positive (Fig. 4, E-G). As with CYP2J2 immunoreactivity in SPG, most (87%) sEH-immunoreactive cells were nNOS positive (Figs. 3G and 4G), whereas only a fraction (20%) appeared to express VIP (Figs. 3H and 4G).

These data in total argue that both EET synthetic CYP2J (likely both CYP2J3 and CYP2J4 isoforms) and EET-regulatory sEH proteins are present in neurons of the sensory TG and parasympathetic SPG.

Neurons innervating the MCA originate in TG and SPG.

Eight days after the subdural injection of the retrograde nerve tracer fluorogold around the left MCA, fluorogold immunoreactivity was detectible both in TG (Fig. 5A) and SPG (Fig. 5B). Labeling was generally present in the ipsilateral ganglia; however, a small number of labeled neurons was observed in contralateral TG. These data confirm that TG and SPG neurons project to the cerebral vasculature, including the MCA.

Fig. 5.

The middle cerebral artery (MCA) is innervated by neurons from TG and SPG. A: injection of the retrograde nerve tracer fluorogold (FG; green) around the MCA resulted in the detection of positively labeled neuronal cell bodies within TG. B: within SPG, a small subset of neurons were FG positive (green, arrows). Scale bars = 60 μm.

EET signaling mediates the regulation of CBF by extrinsic perivascular vasodilator nerves.

No treatment (excepting transient hypercapnia) had any significant effect on MABP, blood gas values, or cerebrovascular reactivity to mild hypercapnic ventilation (Table 1).

Table 1.

Effects of drug treatment upon physiological and cerebrovascular variables

Treatment Group	MABP, mmHg	pH	Arterial Pco2, mmHg	Arterial Po2, mmHg	Change in CBF, %	HCI, %	HCF, %
No treatment	94±6	7.40±0.08	35.8±4.6	154±22	3.5±1.1	18.0±4.3	30.3±13.4
Hypercapnia	90±6	7.36±0.12	49.6±8.7	127±15	18.0±4.3	NA	NA
14,15-EET	107±10	7.38±0.09	32.5±5.1	117±12	61.8±17.5	38.1±9.6	26.1±12.0
14,15-EEZE	103±9	7.39±0.06	29.7±8.3	150±19	−7.8±3.3	19.7±8.3	26.1±10.4
MS-PPOH	98±12	7.40±0.07	34.9±6.1	145±14	−5.4±10.1	19.6±1.7	22.6±2.5
Electrical stimulation	102±12	7.38±0.02	31.6±4.5	138±9	23.4±4.2	36.8±5.9	69.1±18.7
Capsaicin	98±14	7.41±0.05	31.1±6.4	161±36	25.2±4.1	18.6±9.4	30.3±7.2
Nerve lesion	95±7	7.41±0.04	34.1±9.2	114±7	1.5±7.7	25.9±11.2	37.1±10.1

Values are means ± SE. MABP, mean arterial blood pressure; CBF, cerebral blood flow, equal to [100% × (final value − initial value)/initial value] [percent change in laser-Doppler flowmetry (LDF)]; HCI, initial percent change LDF to transient hypercapnia; HCF, posttreatment percent change in LDF to transient hypercapnia; EET, 14,15-epoxyeicosatrienoic acid; 14,15-EEZE, 14,15-epoxyeicosa-5(Z)-enoic acid; MS-PPOH, N-methylsulfonyl-6-(2-propargyloxyphenyl)hexanamide; NA, not applicable.

Electrical stimulation of the efferent nerve supply to the MCA within the ethmoidal nerve resulted in a frequency-dependent increase in cortical blood flow with no concomitant changes in MABP (Fig. 6A). This hyperemia began immediately upon the initiation of the stimulus train, peaking within 20 s and returning immediately to baseline CBF values upon termination. Pharmacological blockade of EET signaling with the putative EET antagonist 14,15-EEZE (n = 6) significantly attenuated the hyperemic response to ethmoidal nerve stimulation, from a peak of 23 ± 4% at 20 Hz with vehicle treatment to 7 ± 1% after antagonist administration (P < 0.05, vehicle vs. 14,15-EEZE treatment by ANOVA; Fig. 6A). Administration of 14,15-EET to the cortical surface produced a concentration-dependent increase in CBF (peak value of 62% at 10 μmol/l 14,15-EET; Fig. 6B) that was blocked by pretreatment with the putative EET antagonist 14,15-EEZE (8% at 10 μmol/l 14,15-EET, P < 0.05, 14,15-EET vs. 14,15-EET + 14,15-EEZE by ANOVA, n = 4). 14,15-EEZE did not significantly alter the hyperemic response to hypercapnia (Fig. 6B). These data indicate that the effects of 14,15-EEZE upon CBF are specific to EET signaling pathways.

Fig. 6.

Epoxyeicosatrienoic acid (EET) signaling mediates the cerebrovascular response to electrical stimulation of extrinsic perivascular vasodilator fibers. A: electrical stimulation of the ethmoidal nerve (30 s, 5–20 Hz) produced a frequency-dependent increase in cortical blood flow. This response was significantly attenuated by pretreatment with the putative EET antagonist 14,15-epoxyeicosa-5(Z)-enoic acid (14,15-EEZE). *P < 0.05, vehicle vs. 14,15-EEZE treatment by ANOVA. n = 6. B: direct administration of 14,15-EET to the cortical surface produced a dose-dependent increase in cerebral blood flow (CBF). This effect was blocked by the putative EET antagonist 14,15-EEZE. *P < 0.05, vehicle vs. 14,15-EEZE treatment by ANOVA. n = 4. The CBF response to transient hypercapnia was not altered by treatment with 14,15-EEZE.

Stimulation of oral nociceptive fibers by instillation of the TRPV1 agonist capsaicin produced a dose-dependent cortical hyperemia peaking at 33 ± 3% at the highest concentration of 1 mmol/l (Fig. 7B). This response was slow to develop, peaking at 4–8 min after capsaicin instillation and persisting for the 10-min duration of the exposure. The capsaicin-evoked hyperemia was reversible, with CBF returning to basal levels after oral irrigation and a 1-h recovery period. Repeated instillation of capsaicin after recovery resulted in an unchanged hemodynamic response (Fig. 7B), ruling out sensitization or desensitization of this effector pathway under the present experimental conditions.

Fig. 7.

Oral trigeminal fiber stimulation evokes EET-mediated cortical hyperemia. A: schematic showing the oral trigeminal fiber stimulation paradigm. The transient receptor potential vanilloid (TRPV)1 agonist capsaicin (CAP; 300 μl, 10 μmol/l–1 mmol/l) was instilled orally, and the resulting cortical hyperemia was monitored by laser-Doppler flowmetry (LDF). This response is believed to involve efferent parasympathetic fibers from SPG (solid arrowheads) or afferent sensory fibers projecting to TG. TNC, trigeminal nucleus caudalis; SSN, superior salivary nucleus. B: oral capsaicin instillation produced a dose-dependent cortical hyperemia that did not change with repeated exposure (n = 4). Lesion of the ethmoidal nerve abolished this hyperemic response. *P < 0.01, first stimulation vs. nerve lesion by ANOVA. n = 4. C: local application of the putative EET antagonist 14,15-EEZE attenuated the hyperemic response to oral trigeminal fiber stimulation. *P < 0.05, vehicle vs. 14,15-EEZE treatment by ANOVA. n = 8. D: local administration of the CYP epoxygenase inhibitor N-methylsulfonyl-6-(2-propargyloxyphenyl)hexanamide (MS-PPOH) did not significantly alter the vascular response to oral capsaicin administration (n = 8).

In some cases, a mild (<10 mmHg) pressor effect was observed in response to oral capsaicin instillation. This effect was transient, resolving within the first minute after oral trigeminal fiber stimulation, in contrast to the observed hyperemia, which peaked minutes later and persisted for an extended period of time. OMAG imaging of the rat MCA revealed that oral capsaicin (10 μmol/l) administration evoked a persistent increase in volume rate (flow), increasing to 108.5% of baseline levels within 8 min of stimulation (Fig. 8B). At the onset of nociceptor stimulation, a small (2.6%) and static elevation in in blood velocity was recorded that accounted for the immediate component of the MCA blood flow response (Fig. 8C). A 2.8% dilation of the MCA was then observed, which accounted for the majority of the stimulus-evoked blood flow response (Fig. 8D), following the fourth-power relationship between diameter and flow predicted by Poiseuille's law. Separate OMAG experiments carried out in mice demonstrated the same effects of capsaicin stimulation upon MCA diameter, blood velocity, and volume rate (unpublished observations). These findings confirm that the increase in MCA flow rate is attributable primarily to direct vasodilation rather than a change in perfusion pressure resulting from nociceptor activation.

Fig. 8.

Evaluation of the hemodynamic response to oral capsaicin administration by optical microangiography (OMAG). A: OMAG image of the rat cortical vasculature. The MCA was imaged through a cross-sectional two-dimensional (x-z plane) line scan (white line). B: oral nociceptor stimulation (10 μmol/l capsaicin) produced a long-lasting increase in MCA blood flow beginning 2 min after the onset of the stimulation and peaking at 108.5% of the baseline value. C: an immediate, small (2.6%), and static elevation in MCA blood velocity accompanied the onset of capsaicin instillation. This effect did not account for the MCA blood flow response. D: oral capsaicin stimulation produced a gradual and long-lasting increase in MCA diameter to 102.8% of baseline that corresponded both temporally and in terms of magnitude to the flow effect observed in A.

To confirm that the CBF response to oral trigeminal fiber stimulation was “neurogenic” in nature, we lesioned the parasympathetic and sensory nerve supply to the anterior cortical surface vasculature in the ethmoidal nerve. Nerve lesion did not alter MABP, blood gas values, resting CBF, or reactivity of the cortical vasculature to transient hypercapnia (Table 1). The CBF response to oral trigeminal fiber stimulation with capsaicin was abolished by nerve transection (P < 0.01, intact vs. nerve lesion by ANOVA, n = 4; Fig. 7B), demonstrating that this response required efferent transmission to the cerebral vasculature. These findings establish that the cortical hyperemia resulting from oral trigeminal fiber stimulation is due to the bona fide neurogenic dilation of cortical arteries.

Local blockade of the putative EET receptor with 14,15-EEZE (n = 8) attenuated the hyperemic response to oral trigeminal fiber stimulation, from a peak value of 33 ± 8% with 1 mmol/l capsaicin in the absence of inhibitor to 16 ± 3% in the presence of 14,15-EEZE (P < 0.05, vehicle vs. 14,15-EEZE treatment by ANOVA; Fig. 7C). In contrast, inhibition of CYP epoxygenases with MS-PPOH (n = 8) did not significantly alter the cortical hyperemic response to oral trigeminal fiber stimulation (Fig. 7D). These pharmacological data provide evidence for the functional involvement of EET signaling in the regulation of CBF by extrinsic perivascular vasodilator fibers.

DISCUSSION

In the present study, we identified in sensory TG and parasympathetic SPG the neuronal expression of CYP2J epoxygenases and sEH, two enzymes involved in EET synthesis and metabolism. We then demonstrated a functional role for EETs in the dilation of cerebral arteries by fibers originating in these ganglia. These findings support a neurogenic source of EETs in the cerebral circulation and a novel role for these neurogenic EETs in the regulation of CBF by extrinsic perivascular vasodilator fibers.

CBF is controlled in part by extrinsic perivascular nerve fibers innervating conduit arteries of the cerebral surface vasculature, including the MCA. This extrinsic innervation is composed of three populations of perivascular fibers: sympathetic vasoconstrictor fibers from the superior cervical ganglia, parasympathetic vasodilator fibers from SPG or otic ganglia, and vasodilator sensory afferents that project to TG (13). Our previous work (16) demonstrated the presence of the EET-metabolizing enzyme sEH in both of these vasodilator nerve populations, raising the possibility that these vasodilator lipids may contribute to neurogenic vasodilation by these fibers.

To test this hypothesis, we identified neuron-specific expression of epoxygenases and sEH in both TG and SPG. Such neuronal expression of epoxygenases and sEH is in agreement with previous studies (23, 25) describing neuronal localization of CYP epoxygenases and sEH in the brain. Furthermore, our finding that CYP2J is expressed in peripheral neurons is consistent with a previous study (32) demonstrating CYP2J expression in autonomic ganglia in the human and rat gut. Despite these previous reports of neuronal epoxygenase and sEH expression, the present study is the first to propose and provide evidence for a functional role of neurogenic EETs, specifically in the regulation of vasomotor tone.

To determine if EETs play a functional role in the neurogenic control of CBF, we stimulated vasodilation using two methods. The first method entailed direct electrical stimulation of efferent nerve fibers innervating the MCA (5), whereas the second method involved chemical stimulation of oral nociceptors with the TRPV1 agonist capsaicin, as previously described by Gottselig and Messlinger (12). This latter method involves the activation of a neuronal reflex pathway including TG and trigeminal and superior salivary nuclei in the brain stem and the recruitment of efferent parasympathetic fibers from SPG (Fig. 7A). While this method involves the indirect stimulation of parasympathetic vasodilator fibers, it may be considered more physiological than direct nerve stimulation. To confirm that chemical stimulation-evoked dilation is specifically mediated through nerve activation, we demonstrated that lesion of the ethmoidal nerve, which is composed of trigeminal afferents and efferent parasympathetic SPG fibers innervating the anterior circle of Willis, abolished the CBF response to oral capsaicin administration. We further confirmed that the hyperemic response to oral nociceptor stimulation was attributable to cerebrovasodilation rather than any effect of capsaicin administration on perfusion pressure and blood velocity.

We found that 14,15-EEZE alone slightly reduced basal CBF, suggesting that EET signaling contributes to the maintenance of resting CBF. Such a finding is in agreement with a previous study (1) in rats demonstrating that inhibition of CYP enzymes with miconazole reduced resting CBF. It is inconsistent with other recent findings in which 14,15-EEZE had no discernible effect on resting CBF (25). These disparate responses, in light of the mild effects of 14,15-EEZE on resting CBF in the present study, suggest that the contribution of EET signaling to resting CBF may be small in magnitude and sensitive to differences in experimental methodology.

Within both models of neurogenic vasodilation, pretreatment of the cortical surface vasculature with the putative EET antagonist 14,15-EEZE markedly attenuated the hyperemic response to chemical or electrical stimulation. Because the CBF response to transient hypercapnia was not altered by 14,15-EEZE, it is unlikely that the small shift in basal vascular tone or a nonspecific effect on vascular reactivity accounts for the ability of 14,15-EEZE to inhibit neurogenic hyperemia. These findings suggest that EET signaling participates in the regulation of CBF by extrinsic perivascular vasodilator nerves.

In contrast to our results with the putative EET antagonist 14,15-EEZE, administration of the CYP epoxygenase inhibitor MS-PPOH to the cortical surface did not alter CBF responses to oral trigeminal fiber stimulation. This suggests that de novo synthesis of EETs is not required for the present neurogenic vasodilation and may reflect the existence of a latent membrane phospholipid-bound pool of EETs not depleted within the 45-min treatment window. Such a releasable pool of EETs has been described in the vascular endothelium (27) and brain astrocytes (26). Interestingly, although we report robust CYP2J immunoreactivity in TG and SPG soma, in unpublished results we have been unable to detect CYP2J immunoreactivity in perivascular neurons surrounding whole mount cerebral surface arteries. One explanation for these findings is that in perivascular nerves, EETs may be synthesized centrally, whereas release of these vasoactive compounds may occur peripherally from a latent membrane-bound pool.

We propose that the present findings support the hypothesis that vasodilator EETs represent novel nerve-derived relaxing factors in the cerebral circulation. One limitation to the present study is that the actual release of EETs from cerebral extrinsic perivascular nerve fibers was not measured in response to stimulation. Within the present in vivo context, such measurement would be exceedingly difficult, as methods for EET detection (such as liquid chromatography tandem mass spectroscopy) have limited sensitivity and do not allow for measurements in local microenvironments such as this neurovascular junction.

It is important to note that the cellular mechanisms underpinning the regulation of cerebral arteries by parasympathetic and sensory perivascular fibers have been studied in detail, and the vasomotor actions of these nerves have been largely attributed to such factors as NO and the neuropeptides VIP and CGRP (13). In a study (12) characterizing the CBF response to oral capsaicin administration, the hyperemic response to capsaicin was abolished by the administration of a VIP receptor antagonist to the dural surface. In other studies (5, 7, 15) investigating CBF responses to electrical stimulation of parasympathetic and sensory perivascular fibers, major contributions have been reported for NO and CGRP. Given the pronounced effect of 14,15-EEZE upon neurogenic hyperemia reported in the present study, it is likely that the EET signaling pathway interacts in some manner with these other vasomotor pathways (Fig. 9). In one potential mode of interaction, EETs may be released from nerves in response to activation, acting in an additive manner with other neurotransmitters to regulate vascular tone (Fig. 9A). Similarly, they may act to modulate the action of these neurotransmitters upon the vasculature. In unpublished results, we observed that 100 nmol/l 14,15-EET did not alter the EC₅₀ or maximal effect of exogenous VIP or CGRP administered directly to the cortical surface. This suggests that EETs do not act to potentiate the actions of these known mediators of neurogenic hyperemia.

Fig. 9.

Potential interaction between neurogenic EETs and other perivascular vasoactive neurotransmitters (NTs). The schematics show three possible modes of interaction between EETs and other known mediators of neurogenic vasodilation, including NO, VIP, or CGRP. A: EETs may act as a cotransmitter, released from perivascular nerve fibers in response to activation, acting on adjacent vascular smooth muscle (VSM) to produce vasorelaxation. EETs may be produced de novo by CYP epoxygenase enzymes (P450) or be released from a membrane phospholipid-bound pool. They may signal in an additive manner with other NTs or may act postjunctionally to fascilitate NT action upon the VSM. B: EETs may serve as an intracellular modulator, acting prejunctionally to modulate the release of vasoactive NTs to produce vasodilation. C: EETs may act as an endothelial mediator of other vasoactive NTs, which may act in part through a mechanism involving the endothelium (EC)-dependent production of EETs.

A second possibility is that neurogenic EETs are not released per se but rather act intracellularly to regulate the release of other neurotransmitters, such as VIP or CGRP (Fig. 9B). Such a role was recently observed for PGE₂ in the modulation of NO release from SPG neurons (17) and for cannabidiol in the modulation of CGRP release from dorsal root ganglia neurons (22). A third possibility is that one or more releasable factors from parasympathetic or sensory perivascular neurons act via a vasomotor pathway dependent on endothelial EET signaling (Fig. 9C). Each of these models could account for the present functional data involving neurogenic vasodilation. Resolving between these alternative explanations and specifically probing the occurrence and mechanisms of neurogenic EET release remain crucial to the conclusive evaluation of our hypothesis that EETs are novel nerve-derived relaxing factors in the cerebral vasculature. Given the difficulty in assessing these questions directly in vivo, experiments are currently underway to address them in a more appropriate in vitro context.

In summary, we provide evidence that extrinsic perivascular vasodilator nerves innervating the cerebral surface vasculature possess the biochemical machinery necessary for the synthesis and regulation of vasodilator EETs. We identify, for the first time, the functional involvement of EET signaling in the neurogenic regulation of CBF by these vasodilator fibers. These findings support the hypothesis that EETs represent novel nerve-derived relaxing factors in the cerebral circulation. A neurogenic source for EET release would have significant implications upon our understanding of the role that these important vasoactive lipids play in cardiovascular regulation. Their specific presence in the cerebral vasculature may have important clinical significance relative to neurovascular disorders such as migraine, vasospasm after subarachnoid hemorrhage, and stroke.

GRANTS

This work was supported by an Oregon Health and Science University Graduate Student Fellowship for the Neurobiology of Disease funded by Vertex Pharmaceuticals; National Institutes of Health (NIH) Grants F31-NS-060498 (to J. J. Iliff), RO1-NS-044313 and P01-NS-049210 (to N. J. Alkayed), 1-R01-HL-093140-01 (to R. Wang), and Z01-ES-025034 (to D. C. Zeldin); and the Intramural Research Program of the NIH (to D. C. Zeldin).