COX-1-derived PGE₂ and PGE₂ type 1 receptors are vital for angiotensin II-induced formation of reactive oxygen species and Ca²⁺ influx in the subfornical organ

Abstract

Regulation of blood pressure by angiotensin II (ANG II) is a process that involves the reactive oxygen species (ROS) and calcium. We have shown that ANG-II type 1 receptor (AT₁R) and prostaglandin E₂ (PGE₂) type 1 receptors (EP₁R) are required in the subfornical organ (SFO) for ROS-mediated hypertension induced by slow-pressor ANG-II infusion. However, the signaling pathway associated with this process remains unclear. We sought to determine mechanisms underlying the ANG II-induced ROS and calcium influx in mouse SFO cells. Ultrastructural studies showed that cyclooxygenase 1 (COX-1) codistributes with AT₁R in the SFO, indicating spatial proximity. Functional studies using SFO cells revealed that ANG II potentiated PGE₂ release, an effect dependent on AT₁R, phospholipase A₂ (PLA₂) and COX-1. Furthermore, both ANG II and PGE₂ increased ROS formation. While the increase in ROS initiated by ANG II, but not PGE₂, required the activation of the AT₁R/PLA₂/COX-1 pathway, both ANG II and PGE₂ were dependent on EP₁R and Nox2 as downstream effectors. Finally, ANG II potentiated voltage-gated L-type Ca²⁺ currents in SFO neurons via the same signaling pathway required for PGE₂ production. Blockade of EP₁R and Nox2-derived ROS inhibited ANG II and PGE₂-mediated Ca²⁺ currents. We propose a mechanism whereby ANG II increases COX-1-derived PGE₂ through the AT₁R/PLA₂ pathway, which promotes ROS production by EP₁R/Nox2 signaling in the SFO. ANG II-induced ROS are coupled with Ca²⁺ influx in SFO neurons, which may influence SFO-mediated sympathoexcitation. Our findings provide the first evidence of a spatial and functional framework that underlies ANG-II signaling in the SFO and reveal novel targets for antihypertensive therapies.

hypertension poses a significant challenge to public health and healthcare by increasing the risk for heart attack, stroke, and kidney failure. The “slow pressor” hypertension, a delayed increase in blood pressure in response to increased systemic angiotensin II (ANG II), is elicited through the subfornical organ (SFO), a small forebrain structure devoid of the blood-brain barrier (6, 22, 41, 49). SFO neurons express ANG-II type 1 receptors (AT₁Rs) and respond to circulating ANG II (11, 22) by sending signals through angiotenergic and glutamatergic projections directly to the paraventricular nucleus (PVN) (3, 11, 13) and other autonomic nuclei that control sympathetic outputs and blood pressure (31). The ANG II-induced upregulation of sympathoexcitation, followed by an elevation of blood pressure, is mediated through reactive oxygen species (ROS) (10, 42, 53, 69) and Ca²⁺ (15, 18, 33). It is well established that ANG II-mediated signals activate NADPH oxidase, leading to accumulation of ROS in the brain that are functionally coupled with hypertensive responses (67). This ANG II-mediated overproduction of ROS is also associated with the activation of voltage-gated L-type Ca²⁺ channels in central neurons (52, 57, 58, 70) and other cell types (2, 55, 62). Activation of L-type Ca²⁺ channels is critical to neurotransmission and long-term potentiation (LTP) in the central nervous system (60, 64).

Previous work has also revealed that the activation of AT₁R results in arachidonic acid (AA) release from membrane phospholipids by phospholipases A₂ (PLA₂) (54). AA is then metabolized to prostaglandin E₂ (PGE₂) by cyclooxygenase-1 (COX-1) or cyclooxygenase-2 (COX-2). Accumulating evidence suggests that COX-derived PGE₂ plays important roles in the regulation of cardiovascular dysfunctions, including ANG II-induced hypertension (4, 6, 21, 39, 40, 45, 56). The PGE₂ receptor subtype 1 (EP₁R) is expressed in the SFO, and studies using genetically modified mice have shown that it plays an important role in signaling the onset of slow-pressor ANG-II hypertension (6, 21). As a member of the G protein-coupled receptor family (17, 25), EP₁R is involved not only in ANG II-elicited ROS production in the SFO (6) but also in the regulation of Ca²⁺ influx through Ca²⁺ channels in neurons and neuroendocrine cells (4, 24, 25, 34). To better understand the complex mechanisms underlying hypertension, it is essential to determine if and how PGE₂, EP₁R, ROS, and Ca²⁺ are functionally linked to serve as key intermediate molecules in ANG II-mediated signaling in SFO neurons.

In this study, we sought to uncover the roles of key prostanoid-linked signals, including PLA₂, COX-1, COX-2, and EP₁R, in ANG II-elicited ROS production and voltage-gated Ca²⁺ influx. Using electron microscopic immunolabeling, we found that COX-1 was coexpressed with AT₁R in SFO neurons. Consistent with this finding, ANG II-triggered PGE₂ release from SFO cells was AT₁R and COX-1 dependent. In addition, ANG II induced ROS formation and L-type voltage-gated Ca²⁺ currents (L-VDCCs) via PLA₂, COX-1-derived PGE₂, and EP₁R. Finally, Nox2 was identified as the source of ANG II-induced, EP₁R-mediated ROS production in the SFO.

MATERIALS AND METHODS

Materials.

All procedures were approved by the Animal Care and Use Committee at Weill Cornell Medical College. Adult (8–10 wk old) C57Bl/6 wild-type (WT) and transgenic mice, including AT₁R-enhanced green fluorescent protein (eGFP), EP₁R^−/−, COX-1^−/−, COX-2^−/−, and Nox2^−/−, were obtained from in-house colonies. ANG II, pronase, thermolysin, nifedipine, BAY K 8644, tetrodotoxin (TTX), ω-conotoxin (Ctx)-GVIA, and N-(p-amylcinnamoyl)anthranilic acid (ACA) were purchased from Sigma-Aldrich (St. Louis, MO). Mn(III)tetrakis(4-benzoic acid)porphyrin chloride (MnTBAP) was procured from EMD Biosciences. PGE₂, SC51089, SC560, and NS398 were obtained from Cayman Chemical (Ann Arbor, MI). The AT₁R antagonist losartan was a gift from the DuPont Merck Pharmaceutical (Deepwater, NJ). The NADPH oxidase polypeptide inhibitor gp91^phox docking sequence (gp9-ds) [H]GGGGCSTRIRRQL[NH₂] and the scrambled gp91-ds (s-gp91-ds) with a sequence of [H]GGGGCLRITRQSR[NH₂] were purchased from Bio-Synthesis (Lewisville, TX). Dihydroethidium (DHE) and calcein AM were purchased from Invitrogen (Carlsbad, CA). The COX-1 antibody (M-20, No. sc-1754, an affinity-purified goat antibody raised against the COOH-terminus of mouse COX-1) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA), and the GFP antibody (GFP-1020, a chicken antibody) was purchased from Aves Laboratories (Tigard, OR).

Electron microscopy.

Tissue was processed and analyzed as previously described (19, 43). In brief, three males of AT₁R-eGFP and WT age-matched mice were anesthetized with pentobarbital sodium, and their brains were fixed by perfusion with heparin-saline (1,000 U/ml), followed consecutively by 3.8% acrolein (Polysciences) in 2% paraformaldehyde in 0.1 mol/l phosphate buffer and then 2% paraformaldehyde in phosphate buffer. Coronal 40-μm vibratome sections were cut through the SFO and transferred to a storage solution for storage at −25°C. Random SFO sections were incubated for 48 h (24 h at room temperature, and 24 h at 4°C) in 0.1 mol/l Tris-buffered saline-0.1% bovine serum albumin solution containing either 1) a cocktail of the COX-1 antibody [1:150, for immunogold (ImG) labeling] and the GFP antibody [1:2,500, for immunoperoxidase (ImP) labeling] for double labeling or 2) the COX-1 antibody (1:2,000, for ImP labeling) for single labeling. Sections were then placed for 30 min in either biotinylated secondary donkey anti-chicken or anti-goat IgG (1:400, Jackson), and avidin-biotin peroxidase complex (ABC Elite Kit, Vector). Bound peroxidase was visualized with 3,3′-diaminobenzidine (Sigma-Aldrich) and hydrogen peroxide (H₂O₂). Tissue processed for dual immunolabeling was additionally incubated overnight in a 1-nm gold particle-conjugated donkey anti-goat secondary IgG (1:50, EMS), followed by postfixation in 2% glutaraldehyde (10 min) and silver intensification using an IntenSE-M Kit (7 min, Amersham). Sections were then postfixed in 2% osmium tetroxide and embedded in EM bed-812 (EMS). Ultrathin sections (70 nm) containing the SFO were cut and collected onto thin bar copper mesh grids (EMS), and counter stained with uranyl acetate and Renyold's lead citrate before examination on a Philips CM10 Technai electron microscope (FEI Company). To quantify COX-1 ImG labeling, undamaged regions of neuropil adjacent to the plastic/tissue interface (i.e., near to the tissue surface) in randomly selected grid square fields, in central portions of the SFO, were examined. In two grids from each AT₁R-eGFP animal, all ImG particles in such fields were categorized in terms of 1) the type of ultrastructural process they were associated with, in neurons (i.e., soma, dendrite, terminal, and axon), glia, and vascular glia (associated with fenestrated capillaries or vascular epithelial cells) and 2) whether that process also contained AT₁R-eGFP ImP labeling.

Enzymatic dissociation of cells from the SFO and the PVN tissues.

The brain was quickly removed and transferred to a chamber containing ice-cold sucrose-artificial cerebrospinal fluid composed of (in mmol/l) 26 NaHCO₃, 1 NaH₂PO₄, 3 KCl, 5 MgSO₄, 0.5 CaCl₂, 10 glucose, and 248 sucrose, oxygenated with 95% O₂-5% CO₂ (pH 7.4). Coronal section of 300 μm in thickness were obtained using Leica VT1000s vibratome (Leica Microsystems) and stored in a custom-designed chamber containing lactic acid-artificial cerebrospinal fluid (l-aCSF) composed of (in mmol/l): 124, NaCl; 26, NaHCO₃; 5, KCl; 1, NaH₂PO₄; 2, MgSO₄; 2, CaCl₂; 10, glucose; 4.5, lactic acid; oxygenated with 95% O₂ and 5% CO₂, pH 7.4. The location of the SFO is characterized by its dorsal connection with the hippocampal commissure, protruding into the third ventricle of the brain. The brain section containing the SFO was put under a dissection light microscope. The SFO tissue (∼450 μm in diameter) was quickly punched out and placed in oxygenated l-aCSF containing 0.01% thermolysin and 0.01% pronase for enzymatic digestion for 2 h at 33°C. The PVN region was identified using lateral ventricle, fornix, and optic tracts as landmarks. The PVN was punched out under dissection microscope and placed in oxygenated l-aCSF containing 0.01% thermolysin and 0.01% pronase for enzymatic digestion for 2 h at 33°C, followed by washing with l-aCSF and trituration of the digested tissue with only 0.3 ml l-aCSF to obtain cell suspension. The cell suspension was quickly plated on a 35-mm petri dish coated with poly-l-lysin for 5–8 min before starting perfusion with oxygenated l-aCSF. Calcein AM (1 μmol/l), a nonfluorescent and cell permeant compound that is hydrolyzed by intracellular esterases into the fluorescent anion calcein, was used to test the viability of isolated SFO cells (44) following enzymatic dissociation.

PGE₂ assay.

WT, COX-1^−/−, or COX-2^−/− mice were euthanized, and the SFO and PVN tissues were collected from mice (2 SFO tissues per biological sample). The dissociated SFO cells were obtained by enzymatic digestion, as described above. Each sample was pretreated with vehicle or ANG II (100 nmol/l) and various chemical modulators for 30 min before its supernatant was extracted for detection of PGE₂ concentration using an EIA kit (Cayman Chemical).

ROS detection.

Intracellular ROS levels in dissociated SFO cells were assessed using DHE. SFO cells were dissociated as described above, moved to a petri dish mounted on Nikon diaphot 300 inverted microscope (Nikon, Japan), and incubated with DHE (2 μmol/l) for 30 min. The fluorescence intensity was measured before (vehicle) and after application of ANG II or PGE₂ using an ethidium bromide (DNA complex) filter (Chroma Technology, Bellows Falls, VT) with the spectrum of an excitation maximum 510 nm and an emission maximum 605 nm, which indicates the presence of superoxide (O₂⁻) and possibly other ROS such as H₂O₂, hydroxyl radical, and peroxynitrite (47). To validate ROS measurement using this system, we tested the capability of exogenous H₂O₂ to increase DHE signal in the SFO cells. The increase in DHE-derived fluorescence observed after treatment of SFO cells with H₂O₂ is attributed to intracellular stimulation of superoxide production by H₂O₂ (65). Addition of H₂O₂ (30 and 300 μmol/l) to the dissociated SFO cells did increase DHE signal in a dose-dependent manner (data not shown here). Data are expressed as DHE fluorescence intensity relative to control samples (Ft/Fo).

Whole cell-patch recording.

The dissociated SFO cells were placed in a 35-mm petri dish mounted on the Nikon inverted microscope and perfused with the oxygenated l-aCSF at room temperature (24–26°C). The whole cell-patch recording of isolated SFO neurons was performed using an Axopatch-200A patch-clamp amplifier (Molecular Devices) and a glass electrode pulled by a P-97 micropipette puller (Sutter Instruments) with tip resistance of 3–5 MΩ when filled with a pipette solution containing (in mmol/l) 145 Cs-gluconate, 17 CsCl, 5 EGTA, 1 MgCl₂, 10 HEPES, 2 ATP, and 0.2 cAMP (pH 7.3). Series resistance and membrane capacitance were compensated. Bipolar SFO neurons were visualized following dissociation. TTX-sensitive voltage-gated Na⁺ currents were recorded to confirm that the patched cell was a neuron. Two millimole/liter Ca²⁺ in l-aCSF was used as a charge carrier for the voltage-gated Ca²⁺ currents elicited by a 500-ms depolarization pulse from a holding potential at −60 mV to stepping potentials between −50 and +20 mV. To selectively record the long-lasting (L)-type Ca²⁺ current, the selective voltage-gated Na⁺ channel blocker TTX (1 μmol/l) and N-type Ca²⁺ channel blocker Ctx-GVIA (1 μmol/l) were added to the perfusion buffer in some patched SFO neurons. The amplitude of L-VDCC was measured at the end of 500-ms pulse (57, 58). The amplitudes of Ca²⁺ currents were analyzed using pClamp 8 (Molecular Devices, Sunnyvale, CA).

Data analysis.

Data are expressed as means ± SE. Multiple comparisons were evaluated by ANOVA, followed by post hoc analysis using Tukey's test with paired comparison. Differences were considered statistically significant at P < 0.05.

RESULTS

COX-1 is coexpressed with AT₁R in SFO neurons.

The SFO contains a network of diversified neurons, neuronal processes, and glial cells. To establish a structural framework from which to best unravel key functional interactions, we performed ultrastructural labeling of COX-1 and AT₁R within the SFO. Examination of central portions of the SFO revealed that COX-1 ImG labeling was common and was concentrated in postsynaptic neuronal processes and glial processes (Fig. 1). Quantification of this distribution pattern (Fig. 1D) demonstrated that 60% of the ImG particles were associated with either neuronal somata or dendrites, and 30% were associated with glial processes in the neuropil (total number of ImG particles examined, 1,072). In addition, 68% of the COX-1 ImG-labeled postsynaptic neuronal processes (i.e., somal and dendritic) also displayed AT₁R-eGFP ImP labeling. The results confirm that AT₁R and COX-1 are spatially linked and provide anatomical support for the theory that COX-1 may be associated with ANG II-elicited generation of PGE₂ in SFO neurons.

Fig. 1.

Cyclooxygenase-1 (COX-1) immunolabeling within the central subfornical organ (SFO) is highly concentrated within neuronal postsynaptic processes, most of which also display angiotensin-II (ANG II) type 1 receptor (AT₁R)-enhanced green fluorescent protein (eGFP) immunoperoxidase (ImP) labeling, and glial processes. A: AT₁R-eGFP-ImP-positive cell body (Soma) and dendrite (LDen), both near fenestrated capillaries (Fen Cap), contain COX-1-immunogold (ImG) particles (arrows). B: COX-1 ImG labeling (arrows) of a glial process (Glia) and an AT₁R-GFP-ImP-labeled dendrite (LDen). C: COX-1 ImP labeling (arrow) near a mitochondrion in an unlabeled dendritic process (Den). D: histogram of the percent distribution of all COX-1 ImG labeling by process type (Vasc glia, glial associated with fenestrated capillaries; Vasc epith, vascular epithelial cells), divided by whether the process did (dark grey) or did not (light grey) colabel for AT₁R-GFP. Scale bars = 0.5 μm.

ANG II-triggered endogenous PGE₂ release from dissociated SFO cells is dependent on AT₁R, PLA₂, and COX-1.

Following dissociation of SFO cells, calcein AM (1 μmol/l) was loaded to isolated cells to check viability. As shown in Fig. 2A, one example of calcein-labeling of dissociated SFO cells lasted 1 h, indicating viability of the in vitro model. In addition, the time course of calcein-labeled SFO cells also indicates that most isolated SFO cells, including neurons, remained viable following 50 min of calcein labeling (86.7 ± 4.7%, P > 0.05 vs. 0 min, cell number from each experiment = 48–54, number of experiments = 3) (Fig. 2B). Taken together, this proved the viability of the isolated SFO cells and neurons and allowed us to perform further functional studies in vitro.

Fig. 2.

A: one example of calcein-labeled isolated SFO cells 1 h after loading calcein AM (1 μmol/l). The white arrows indicate SFO neurons, whereas black arrows indicate red blood cells. Bar = 10 μm. B: time course of a group of calcein-labeled SFO cells is shown (P > 0.05 vs. 0 min, number of experiments = 3; cell number in each experiment = 48–54). C: histogram shows amounts of endogenous prostaglandin E₂ (PGE₂) released from the wild-type (WT) SFO cells in the presence of vehicle (Veh, N = 9), ANG II (100 nmol/l, N = 5), coapplied AT₁R antagonist losartan (Los, 3 μmol/l, N = 9) and the phospholipase A₂ (PLA₂) inhibitor N-(p-amylcinnamoyl)anthranilic acid (ACA; 1 μmol/l, N = 8) or endogenous PGE₂ released from COX-1^−/− (N = 6) and COX-2^−/− (N = 5) SFO cells in the presence of ANG II (100 nmol/l), respectively. D: histogram summarizes amounts of endogenous PGE₂ released from WT paraventricular nucleus cells in the presence of Veh (V; N = 4) and ANG II (100 nmol/l, N = 4). Each tissue sample was collected from 2 mice. *P < 0.05 vs. WT SFO vehicle; **P < 0.01 vs. WT SFO vehicle.

Considering that COX-1 was present in SFO neurons, often in the same cells as AT₁R, we sought to determine whether ANG II would elicit PGE₂ production from dissociated SFO cells and, if so, what pathway might be involved in the PGE₂ release (6, 29, 54). As shown in Fig. 2C, the basal amount of endogenous PGE₂ released from WT SFO cells following incubation in vehicle for 30 min was 1.69 ± 0.2 ng/mg (N = 9). Incubation in ANG II (100 nmol/l) for 30 min elicited a significant increase in endogenous PGE₂ release from WT SFO cells (+208 ± 38.5%, P < 0.01 vs. WT SFO vehicle, N = 5). This effect was blunted when the cells were preincubated with the AT₁R antagonist losartan (3 μmol/l, 58.8 ± 23.5%, P > 0.05 vs. WT SFO vehicle, N = 9) or the PLA₂ antagonist ACA (1 μmol/l, 88.2 ± 29.4%, P > 0.05 vs. WT SFO vehicle, N = 8). Furthermore, incubation of COX-1^−/− SFO cells in ANG II did not increase PGE₂ release (−64.5 ± 8.3%, P > 0.05 vs. WT SFO vehicle, N = 6). Incubation of COX-2^−/− SFO cells in ANG II partially but significantly inhibited PGE₂ release (+89.4 ± 12.4%, P < 0.05 vs. WT SFO vehicle, N = 5). As shown in Fig. 2D, the basal levels of endogenous PGE₂ released from WT PVN cells following incubation in vehicle and ANG II (100 nmol/l) were relatively low compared with WT SFO cells (vehicle, 0.75 ± 0.49 ng/mg N = 4; and ANG II, 0.57 ± 0.47 ng/mg; P > 0.05, N = 4). These results indicate that ANG II elicits endogenous PGE₂ release from SFO cells, and this PGE₂ production may be dependent on the conversion of AA by COX-1. Importantly, this provides mechanistic support for our previous finding that COX-1-dependent PGE₂ formation plays a key role in ANG II-induced hypertension in the SFO.

ANG II and PGE₂ induce increased ROS production in SFO cells.

It has been established that the SFO mediates systemic ANG II-dependent hypertension via increased production of ROS (6, 66, 68). Moreover, we have recently reported that a single dose (100 nmol/l) of ANG II causes a significant increase in ROS formation in WT SFO cells in vitro, a response that was absent in either EP₁R^−/− or COX-1^−/− SFO cells, but was intact in COX-2^−/− SFO cells (6). Using the EP₁R antagonist SC51089, we also confirmed that ANG II-induced ROS formation in WT SFO cells was via EP₁R. In addition, PGE₂ (100 nmol/l) elicited increases in ROS formation to a similar extent as ANG II (6). These data indicate that PGE_2, via EP₁R, serves as a downstream signal in ANG II-induced ROS formation. Using DHE as a fluorescence ROS indicator, we thus focused on the roles of other critical elements in ANG-II signaling pathway, including AT₁R, PLA₂, and NADPH oxidase, in ANG II- and PGE₂-induced ROS formation in SFO cells.

As shown in Fig. 3, both exogenous ANG II (N = 8–24 per individual dose) and PGE₂ (N = 6–18 per individual dose) increased the DHE signal in a dose-dependent manner in dissociated WT SFO cells. Coapplication of the AT₁R antagonist losartan (3 μmol/l) significantly inhibited the increase in DHE induced by ANG II, but not by PGE₂ (Fig. 3, B and C).

Fig. 3.

A: representative dihydroethidium (DHE) images of the dissociated WT SFO cells are shown in the presence of vehicle and 100 nmol/l ANG II. B and C: dose-response curves show effects by ANG II (N of cells = 8–24/each dose; and N of experiments = 5) and PGE₂ (N of cells = 6–18/each dose; and N of experiments = 4) on the relative intensity of DHE in the presence of Veh and the AT₁R antagonist Los (3 μmol/l) in WT SFO cells. *P < 0.05 vs. Veh; **P < 0.01 vs. Veh.

As shown in Fig. 4, ANG II (100 nmol/l) induced an increase in DHE signal (+25 ± 3%, P < 0.01 vs. vehicle, n = 9), a response that was blocked by the general PLA₂ inhibitor ACA (1 μmol/l), the COX-1 inhibitor SC560 (10 μmol/l), the NADPH oxidase peptide inhibitor gp91-ds (1 μmol/l) and in Nox2^−/− cells (Fig. 4A), but not by s-gp91-ds. PGE₂ (100 nmol/l) also induced a similar increase in ROS formation (+32 ± 8.2%, P < 0.01 vs. vehicle, n = 7) that was blocked by gp91-ds (1 μmol/l) and in Nox2^−/− cells. This effect, however, was not blocked by ACA (1 μmol/l), SC560 (10 μmol/l), or s-gp91-ds (Fig. 4B). Taken together, these results indicate that PLA₂-mediated PGE₂ production is downstream of AT₁R but upstream of COX-1 and that Nox2 is the source of ANG II and PGE₂-induced ROS formation in SFO cells.

Fig. 4.

A: histogram shows the relative intensity of DHE in the presence of 100 nmol/l ANG II (N = 9) and coapplied ACA (1 μmol/l, N = 17), SC560 (10 μmol/l, N = 26), gp91-ds-tat (1 μmol/l, gp91, N = 8), and scrambled gp91-ds (1 μmol/l, s-gp91, N = 22) in WT SFO cells and in Nox2^−/− SFO cells (N = 18) in the presence of ANG II (100 nmol/l). B: histogram summarizes the relative intensity of DHE in the presence of 100 nmol/l PGE₂ (N = 7) and coapplied ACA (1 μmol/l, N = 17), SC560 (10 μmol/l, N = 7), gp91-ds-tat (1 μmol/l, gp91, N = 9), and scrambled gp91-ds-tat (1 μmol/l, s-gp91, N = 20) in WT SFO cells and in Nox2^−/− SFO cells in the presence of PGE₂ (100 nmol/l) (N = 18). **P < 0.01 vs. Veh.

ANG II potentiates the L-VDCC in SFO neurons via AT₁R.

In addition to generation of ROS, the AT₁R is also involved in voltage-dependent Ca²⁺ influx in the central nervous system (58, 70). Since the L-VDCC is a key signal for the regulation of short- and long-term potentiation and neurotransmitter release (12, 16, 46, 60, 64), we next sought to determine whether the AT₁R-mediated signals were linked with the L-VDCC in SFO neurons. The VDCC in the mouse SFO neurons are composed of transient fast-inactivating and slowly inactivating components (59). Representative traces show that following application of TTX (1 μmol/l) to block voltage-gated Na⁺ currents and Ctx-GVIA (1 μmol/l) to inhibit the transient, fast inactivating N-type Ca²⁺ currents in SFO neurons, the L-VDCCs were well distinguished (Fig. 5A). Biophysically, the current-voltage curves of this L-VDCC show that it was maximally elicited by a depolarization pulse from the holding potential at −60 mV to the stepping potential at −10 mV (Fig. 5B). Pharmacologically, it was activated by ANG II (100 nmol/l), which is in agreement with previous observation in the SFO (61) and other central neurons (Wang et al. 2004). Such effect was attenuated by coapplied losartan (3 μmol/l) (Fig. 5, A and C). However, further coapplication of the L-type Ca²⁺ channel activator BAY K 8644 (2 μmol/l, P < 0.01 vs. vehicle, N = 16) restored increased currents, but the L-VDCC was completely inhibited by the L-type Ca²⁺ channel blocker nifedipine (2 μmol/l, P < 0.01 vs. vehicle, N = 6) (Fig. 5, A and C) and nonselective Ca²⁺ channel blocker Cd²⁺ (100 μmol/l, P < 0.01 vs. vehicle, N = 6) (12, 58) (Fig. 5C). Taken together, ANG II (100 nmol/l) triggered a potentiation of the L-VDCC in SFO neurons (+53.3 ± 5.8%, P < 0.01 vs. vehicle, N = 9), a response that was blocked by the AT₁R antagonist losartan (3 μmol/l; −15.7 ± 7.3%, P < 0.05 vs. vehicle, N = 9) (Fig. 5C). These results indicate that ANG II-elicited potentiation of the dihydropyridine-sensitive L-VDCC in SFO neurons is mediated by AT₁R.

Fig. 5.

A: representative traces shows L-type voltage-gated Ca²⁺ current (I_Ca; L-VDCC) recorded from a WT SFO neuron following applications of Veh, the Na⁺ channel blocker tetrodotoxin (TTX; 1 μmol/l), and N-type Ca²⁺ channel blocker ω-conotoxin (Ctx)-GVIA (1 μmol/l), ANG II (100 nmol/l), the AT₁R antagonist Los (10 μmol/l), the selective L-VDCC activator BAY K 8644 (BayK; 2 μmol/l), and the selective L-VDCC blocker nifedipine (Nifed; 2 μmol/l). SP, stepping potential; HP, holding potential. B: current-voltage curve of L-VDCC recorded from 4 WT SFO neurons in the presence of Veh, BayK (2 μmol/l), and Nifed (2 μmol/l). C: histogram summarizes percent changes in the amplitude of L-VDCC in the presence of Veh (N = 16), ANG II (10 μmol/l, N = 9), coapplied Los (3 μmol/l, N = 9), BayK (2 μmol/l, N = 16), Nifed (2 μmol/l, N = 6), and Cd²⁺ (100 μmol/l, N = 6). ##P < 0.01 vs. Veh, ANG II, Los, and BayK; *P < 0.05 vs. Veh; **P < 0.01 vs. Veh.

PLA₂, COX-1, and EP₁R are involved in ANG II-induced L-VDCC in SFO neurons.

Since the EP₁R is also involved in presynaptic Ca²⁺ influx (24) and glutamate release in the central nervous system (4), we used pharmacological and electrophysiological approaches to determine whether the PLA₂/COX-1/EP₁R signaling molecules that link to ANG II-induced ROS generation also play a role in the ANG II-mediated potentiation of L-VDCC in SFO neurons. As shown in Fig. 6, A and B, ANG II (100 nmol/l)-induced enhancement of L-VDCC (P < 0.01 vs. vehicle, N = 11) was abolished by pretreatment with the general PLA₂ inhibitor ACA (1 μmol/l, P > 0.05 vs. vehicle, N = 5), coapplication of the COX-1 inhibitor SC560 (10 μmol/l, P > 0.05 vs. vehicle, N = 6), and the EP₁R antagonist SC51089 (10 μmol/l, P > 0.05 vs. vehicle, N = 11) but were unaffected by coapplication of the COX-2 inhibitor NS398 (10 μmol/l, P < 0.01 vs. vehicle, N = 5). Moreover, ANG II-induced potentiation in L-type Ca²⁺ currents was abolished in EP₁R^−/− SFO neurons (P > 0.05 vs. vehicle, N = 7) (Fig. 6B). Figure 6C shows the time course of ANG II-induced enhancement in the amplitude of L-VDCC currents in one single isolated SFO neuron. The ANG II-enhanced L-VDCC was reversed by the coapplied EP₁R antagonist SC51089 (10 μmol/l) (Fig. 6C). These results establish that PLA₂, COX-1, and EP₁R act downstream of ANG II for potentiation of L-VDCC in SFO neurons.

Fig. 6.

A, top: representative traces show L-VDCC recorded from an isolated WT SFO neuron following applications of Veh, ANG II (100 nmol/l), coapplied COX-2 inhibitor NS398 (10 μmol/l), and the PLA₂ inhibitor ACA (1 μmol/l). A, bottom: representative traces show L-VDCC recorded from another isolated WT SFO neuron following applications of Veh, ANG II (100 nmol/l), coapplied PGE₂ type 1 receptor (EP₁R) antagonist SC51089 (10 μmol/l), and L-VDCC blocker Cd²⁺ (100 μmol/l). B: histogram summarizes percent changes in the amplitude of L-VDCC in the presence of Veh, ANG II (100 nmol/l, N = 11), coapplied NS398 (10 μmol/l, N = 5), ACA (1 μmol/l, N = 5), SC560 (10 μmol/l, N = 6), SC51089 (10 μmol/l, N = 11) in WT SFO neurons, or L-VDCC in EP₁R^−/− (N = 7) SFO neurons in presence of ANG II (100 nmol/l). **P < 0.01 vs. Veh. C: time course of the amplitude of L-VDCC recorded from 1 single isolated SFO neuron before and after application of ANG II and coapplied EP₁R antagonist SC51089.

Nox2-derived ROS contribute to ANG II-induced L-VDCC in SFO neurons.

We next we sought to determine the molecular factor responsible for L-VDCC activation and its subcellular localization. Since ROS are known modulators of L-VDCC and they are downstream of the ANG-II signaling pathway, we were curious as to whether Nox2-derived ROS are involved in ANG II-induced L-type Ca²⁺ influx. We examined the effects of the ROS scavenger MnTBAP and Nox2^−/− SFO neurons on ANG II-elicited L-VDCC. As shown in Fig. 7A, ANG II-induced potentiation in L-VDCC in WT SFO neurons (P < 0.01 vs. vehicle, N = 6) was abolished by pretreatment with MnTBAP (100 μmol/l, P > 0.05 vs. vehicle, N = 6). This response was also abolished in Nox2^−/− SFO neurons (P > 0.05 vs. vehicle, N = 4), implying that Nox2-derived ROS are required for the ANG-II potentiation of L-VDCC in SFO neurons.

Fig. 7.

A: histogram summarizes percent changes in the amplitude of L-VDCC in the presence of Veh (N = 6), ANG II (100 nmol/l, N = 6), coapplied reactive oxygen species scavenger Mn(III)tetrakis(4-benzoic acid)porphyrin chloride (MnTBAP, TBAP; 100 μmol/l, N = 6) in WT SFO neurons, or L-VDCC in Nox2^−/− neurons (N = 4). **P < 0.01 vs. Veh. B: histogram shows percent changes in the amplitude of L-VDCC in the presence of vehicle (Veh, N = 16), PGE₂ (100 nmol/l, N = 16), coapplied AT₁R antagonist Los (10 μmol/l, N = 4), PLA₂ inhibitor ACA (1 μmol/l, N = 5), SC51089 (10 μmol/l, SC, N = 7) and MnTBAP (100 μmol/l, N = 5), or in EP₁R^−/− neurons (N = 5), and in Nox2^−/− SFO neurons (N = 5) in the presence of PGE₂ (100 nmol/l). **P < 0.01 vs. Veh.

PGE₂ potentiates L-VDCC via EP₁R and Nox2 in SFO neurons.

PGE₂ is known to induce an increase in Ca²⁺ influx in the central nervous system (4, 32). Therefore, we further analyzed whether PGE₂, as a downstream signal of ANG-II pathway, is capable of inducing L-VDCC in SFO neurons and, if so, whether it involves the activation of EP₁R/Nox2 pathway. We examined the effect of applying antagonists for AT₁R, PLA₂, and EP₁R or deleting EP₁R and Nox2 on PGE₂-induced L-VDCC in SFO neurons. As shown in Fig. 7B, PGE₂ (100 nmol/l) induced potentiation of L-VDCC in SFO neurons (55.6 ± 9.9%, P < 0.01 vs. vehicle, N = 16), a response that was unaffected by losartan (3 μmol/l, P < 0.01 vs. vehicle, N = 4) or ACA (1 μmol/l, P < 0.01 vs. vehicle, N = 5). However, coapplied SC51089 (10 μmol/l, SC, N = 7) or genetic deletion of EP₁R (N = 5) inhibited L-VDCC enhancement induced by PGE₂ (P > 0.05 vs. vehicle) (Fig. 7B). Moreover, coapplied MnTBAP (100 μmol/l) or genetic deletion of Nox2 (N = 5) also abolished the L-VDCC potentiated by PGE₂ (P > 0.05 vs. vehicle) (Fig. 7B). These results indicate that the EP₁R and Nox2 serve as critical signals linking PGE₂ to L-type Ca²⁺ channels in the SFO.

Paracrine regulation of L-VDCC in SFO neurons by endogenous PGE₂.

Our ultrastructural data show that COX-1 labeling is concentrated in postsynaptic neuronal processes in the SFO. Assuming that endogenous PGE₂ derived by postsynaptic COX-1 activates presynaptic EP₁R (24), which may be responsible for ANG II-induced potentiation of L-VDCC in isolated SFO neurons, we next sought to investigate whether locally generated PGE₂ derived from COX-1 in neighboring cells potentiates L-VDCC in the patched neuron. Adding SC560 (200 nmol/l) to the pipette buffer to inhibit intracellular COX-1 in the SFO neuron patched via the whole cell configuration (Fig. 8A), we found that exogenously applied ANG II (100 nmol/l) still increased L-VDCC (P < 0.01 vs. vehicle, n = 9). However, this response was blocked by extracellular coapplication of SC51089 (10 μmol/l, P > 0.05 vs. vehicle, n = 6) (Fig. 8B). Further coapplication of BAY K 8644 (2 μmol/l) reversed the inhibition of L-VDCC potentiation (P < 0.01 vs. vehicle, n = 4) in the same patched neuron, indicating that the L-type Ca²⁺ channels retained their functionality (Fig. 8B). These results indicate that endogenous PGE₂ derived from ANG II-activated COX-1 in neighboring SFO cells could exert paracrine action by potentiating L-VDCC through EP₁R. This assumption is further supported by our ultrastructural labeling studies that indicate extensive expression of COX-1 in SFO neuronal somata, dendrites, and glial cells.

Fig. 8.

A: schematic illustration of injection of the COX-1 inhibitor SC560 (200 nmol/l) into the patched SFO neuron via the patch electrode. B: histogram summarizes percent changes in the amplitude of L-VDCC in SFO neurons intracellularly loaded with SC560 plus extracellularly perfused Veh (N = 9), ANG II (100 nmol/l, N = 9), coapplied SC51089 (10 μmol/l, N = 6), BayK (2 μmol/l, N = 4), and Cd²⁺ (100 μmol/l, N = 4). **P < 0.01 vs. intracellular SC560 plus extracellular Veh; ##P < 0.01 vs. all other groups.

DISCUSSION

COX-derived prostanoids and other endogenous fatty acid metabolites are involved in a wide variety of biological processes (1, 8, 27, 45, 50) and have been implicated in blood pressure regulation (20, 45). The SFO (11, 14), PGE₂/EP₁R signaling (6, 7, 21), ROS (6, 66, 68), and voltage-gated Ca²⁺ channels (37, 52, 70) have each been independently proposed as important mediators in the regulation of hypertension. Here we provide the first in vitro evidence that all of these factors are transductionally linked in ANG II-mediated signaling in the SFO. We also demonstrate that COX-1 is highly concentrated in neuronal postsynaptic processes in the central SFO, particularly in the processes of cells that coexpress the AT₁R and that COX-1 is the major source for the ANG II-activated PGE₂ release in the SFO. Furthermore, ANG II-evoked PGE₂ and ROS formation in dissociated SFO cells is prevented by deletion of COX-1 and partially attenuated by deletion of COX-2. ANG II or PGE₂-potentiated ROS and Ca²⁺ currents are dependent on PLA₂, EP₁R, and ROS. These results support the notion that COX-1-derived PGE₂ and EP₁R signaling in the SFO are critical for slow-pressor ANG-II hypertension (6) and provide a detailed signaling pathway linking ANG II with ROS and L-VDCC, as summarized schematically in Fig. 9. As previously reported (53, 66), we speculate that this ANG II-induced intracellular increase in O₂⁻ may also lead to an increase in extracellular H₂O₂. Additionally, the present in vitro data also support the in vivo results recently published by our group (6).

Fig. 9.

A summary schematic overview highlights the signaling pathway proposed in the present study. Note: Nox2-containing NADPH oxidase is composed of 5 subunits, the membrane anchored catalytic subunits gp91^phox (Nox2) and p22^phox and the cytosolic subunits p47^phox, p40^phox, and p67^phox. It is speculated that the intracellular increase in superoxide (O₂⁻) leads to an increase in extracellular H₂O₂ (53, 66). P, phosphorylated; AA, arachidonic acid; PKC, protein kinase C; [Ca²⁺]_i, intracellular Ca²⁺ concentration.

Ultrastructural labeling of SFO neurons supports paracrine action of PGE₂.

The neuronal location of AT₁R in the SFO was illustrated by GFP labeling in postsynaptic somata and dendrites, many of which also contained COX-1. The colocalization of AT₁R and COX-1 in the SFO provides valuable ultrastructural support for the functional interaction seen between ANG II and COX-1 in dissociated SFO cells. In dual-labeled somatodendritic profiles, COX-1 ImG labeling was present in the cytoplasmic compartment of SFO neurons. Plasmalemmal labeling of COX-1 was also seen in small- and medium-sized dendrites receiving synaptic input from one or more axon terminals. Postsynaptic coexpression of COX-1 and AT₁R in SFO neurons indicates that COX-1 is likely to be associated with AT₁R-elicited generation of PGE₂ and that release of PGE₂ from postsynaptic sites may modulate EP₁ receptors at presynaptic sites (24). In addition to COX-1, our recent EM data indicate that COX-2 is also present in SFO neurons that express AT₁R (unpublished data).

It is notable that PGE₂ is also produced and released from glial cells (29). Our electron microscopic data demonstrate that processes of glial cells in the SFO express about 30% of the COX-1 ImG labeling, which might also express AT₁R. Furthermore, the fact that inhibition of COX-1 with intracellular SC-560 did not affect ANG II-induced, EP₁R-mediated L-type Ca²⁺ currents in whole cell-patched SFO neurons indicates that COX-1 activation leading to PGE₂ release is of exogenous origin. Thus these data raise the possibility that ANG II triggers PGE₂ release from the neighboring cells, including neurons or glial cells such as ependymal cells, microglia, and astrocytes, and PGE₂ exerts a paracrine action in the SFO.

ANG II, PLA₂, COX-1, and EP₁R-mediated ROS in the SFO.

COX-1, PGE₂, and NADPH oxidase-derived oxidative stress are involved in inflammatory responses (63) and EP₁Rs were recently linked to ANG-II hypertension (6, 7). However, if or how these key signaling molecules interacted with one another had not been established. In this study, we found that inhibition of PLA₂, COX-1, and EP₁R prevented ANG II-elicited ROS production in the SFO, raising the possibility that there is molecular cross talk between COX-1/EP₁R and Nox/ROS signaling in SFO neurons. Since the SFO is composed of neurons and nonneuronal glial cells such as ependymal cells, microglia, astrocytes, and vascular cells, ROS production detected in this in vitro study could come from dissociated SFO neurons or glial and vascular cells. However, further investigation is needed to identify different cell types and the exact mechanisms underlying EP₁R-mediated ROS formation via Nox2-NADPH oxidase in the SFO.

EP₁R and Ca²⁺ influx.

The role of Nox2-dependent ROS in the regulation of ANG II-potentiated L-type Ca²⁺ currents has been previously addressed in central neurons and other cell types in peripheral tissue (2, 23, 57, 58, 70). However, a link between PGE₂/EP₁R signaling and ANG II-potentiated L-VDCC in central neurons had not been established. Our data provide strong evidence that PGE₂ can potentiate L-VDCC through EP₁R. EP receptors are a family of G protein-coupled receptors that regulate intracellular free Ca²⁺, and PGE₂ is known to regulate Ca²⁺ influx (4, 5, 25–28, 32, 35, 36) through EP receptors in neurons and other cell types. EP₁R and EP₃R couple to G_q/11 protein and activate phospholipase C (38). However, contrary to potential Ca²⁺-mobilizing actions by EP₁R, the EP₃R-coupled regulatory effects on Ca²⁺ homeostasis in adrenal medulla chromaffin cells and melanotrophs are inhibitory (25, 36). Nevertheless, because of lack of validated antibodies against EP₁R and selective antagonists for other subtypes of EP receptors, the present study has not ruled in coexpression of AT₁R and EP₁R in the SFO or ruled out the possible involvement of other EP receptor subtypes in the regulation of ANG-II hypertension in this model.

Mechanisms underlying ROS-enhanced gating of L-type Ca²⁺ channels.

Our results demonstrate that EP₁R/ROS signaling is required for ANG II and PGE₂ potentiation of L-VDCC in SFO neurons, but EP₁R/ROS signaling is not necessary for direct gating induced by L-type channel activator BAY K 8644. This indicates that ROS has different binding sites from those of BAY K 8644. Recent reports suggest a coupled gating model of L-type isoform Ca_V1.2 channel activation by ROS in arterial smooth muscle (2, 37) that provides a novel mechanism for the regulation of Ca²⁺ influx and excitability. Using a TIRF imaging-based approach, a pathway has been characterized that functionally links subplasmalemmal generation of endogenous ROS that precede and colocalize with enhanced L-type Ca²⁺ channel activity via activation of PKCα in cerebral arterial smooth muscle (2). Further studies are needed to determine how ROS are functionally linked to L-VDCC potentiation in the SFO.

Potential contribution of L-type Ca²⁺ currents to SFO function.

Though the mechanisms underlying ANG II-induced sympathoexcitation via SFO neuronal projections to other autonomic nuclei remain elusive, it is likely that voltage-gated L-type Ca²⁺ influx is involved in presynaptic glutamate release and therefore synaptic plasticity, such as LTP (16, 46, 64). For LTP mediated by increased release of glutamate in the corticolateral nucleus of the amygdala, L-VDCCs are necessary for the persistent expression, and possibly the induction, of presynaptic cortico-LA LTP. Whereas L-VDCCs only weakly contribute to baseline release, they contribute significantly to increased synaptic transmission (16). On the other hand, the electrical stimulation-induced LTP of synapses onto the excitatory narrow-field vertical neurons of the sSC requires postsynaptic Ca²⁺ elevation by both NMDARs and L-type Ca²⁺ channels during induction but not maintenance (64). Therefore, L-type Ca²⁺ channels might be required for both induction and persistent expression of presynaptic LTP.

The role of SFO L-type voltage-dependent Ca²⁺ channels in ANG II-induced hypertension.

The functional link between ANG II and voltage-gated Ca²⁺ influx in central neurons, including sympathetic neurons, has been previously reported (15, 33, 61, 70). In vivo, it is evident that the SFO sends the ANG-II peptidergic efferent to the PVN neurons (13, 30, 31) and that the ANG II-induced hypertension mediated via the SFO is attenuated by intracerebroventricular administration of the L-VDCC blocker nifedipine (48). In vitro, L-VDCC plays an important role in the regulation of peptidergic neurotransmitter release from presynaptic axonal terminals of central neurons (59). Therefore, it can be speculated that AT₁R in SFO sensory neurons are activated by circulating ANG II, which increases Ca²⁺ influx mediated by L-VDCC through EP₁R and ROS pathways. Since increments in intracellular calcium concentration lead to ANG-II release from axonal terminals in the PVN-projecting SFO neurons (18), L-VDCC potentiation may cause excitation of presympathetic neurons in the PVN, leading to hypertension.

Perspective and significance.

The in vitro results described here, along with our recently published in vivo results (6), provide strong evidence that EP₁R-mediated ROS formation via Nox2-NADPH oxidase in the SFO may provide useful targets for the treatment of hypertension. SFO-targeted treatment with cytoplasmic superoxide dismutase has proven to be as effective in ameliorating ANG-II hypertension as chronic intracerebroventricular infusion of the AT₁R antagonist losartan (68). Peripheral administration of the AT₁R antagonist losartan may also target the AT₁R expressed in the SFO because of its lack of a normal blood-brain barrier (9). In addition, blocking the ANG-II/PGE2/L-VDCC pathway in the SFO might be beneficial for the treatment of hypertension (51).

In conclusion, the present study used diverse cutting-edge technologies, including electron microscopy, EIA, ROS imaging, patch recording, and null mutant mice to comprehensively determine that ANG II-mediated PLA₂ and PGE₂ signals are coupled with ROS formation and intracellular Ca²⁺ homeostasis in SFO neurons. Additionally, as PGE₂ is also released from glial cells, future studies should focus on the identification of possible distinct contributions by different cell types to ANG II-triggered PGE₂ release. Taken together, our findings are a critical step toward a more complete understanding of ANG-II signaling in the SFO and may provide novel targets for future antihypertensive therapy.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute Grants HL-096571, HL-063887, and HL-098351.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

G.W., P.S., J.P.P., C.I., and R.L.D. conception and design of research; G.W., P.S., J.R.P., J.A., J.P.P., J.M.M., J.F., P.Z., T.A.M., and R.L.D. performed experiments; G.W., P.S., J.R.P., J.A., J.P.P., J.F., T.A.M., V.M.P., C.I., and R.L.D. analyzed data; G.W., P.S., J.R.P., J.A., J.P.P., V.M.P., C.I., and R.L.D. interpreted results of experiments; G.W., P.S., J.P.P., V.M.P., and R.L.D. prepared figures; G.W., P.S., C.I., and R.L.D. drafted manuscript; G.W., P.S., J.P.P., C.I., and R.L.D. edited and revised manuscript; G.W., P.S., C.I., and R.L.D. approved final version of manuscript.