AT1a-dependent GABA_A inhibition in the MnPO following chronic intermittent hypoxia

Abstract

Chronic intermittent hypoxia (CIH) is associated with diurnal hypertension, increased sympathetic nerve activity (SNA), and increases in circulating angiotensin II (ANG II). In rats, CIH increases angiotensin type 1 (AT1a) receptor expression in the median preoptic nucleus (MnPO), and pharmacological blockade or viral knockdown of this receptor prevents CIH-dependent increases in diurnal blood pressure. The current study investigates the role of AT1a receptor in modulating the activity of MnPO neurons following 7 days of CIH. Male Sprague-Dawley rats received MnPO injections of an adeno-associated virus with an shRNA against the AT1a receptor or a scrambled control. Rats were then exposed to CIH for 8 h a day for 7 days. In vitro, loose patch recordings of spontaneous action potential activity were made from labeled MnPO neurons in response to brief focal application of ANG II or the GABA_A receptor agonist muscimol. In addition, MnPO K-Cl cotransporter isoform 2 (KCC2) protein expression was assessed using Western blot. CIH impaired the duration but not the magnitude of ANG II-mediated excitation in the MnPO. Both CIH and AT1a knockdown also impaired GABA_A-mediated inhibition, and CIH with AT1a knockdown produced GABA_A-mediated excitation. Recordings using the ratiometric Cl⁻ indicator ClopHensorN showed CIH was associated with Cl⁻ efflux in MnPO neurons that was associated with decreased KCC2 phosphorylation. The combination of CIH and AT1a knockdown attenuated reduced KCC2 phosphorylation seen with CIH alone. The current study shows that CIH, through the activity of AT1a receptors, can impair GABA_A-mediated inhibition in the MnPO and contribute to sustained hypertension.

INTRODUCTION

Obstructive sleep apnea (OSA) is associated with elevated levels of circulating angiotensin II (ANG II) as well as increases in blood pressure (1). Chronic intermittent hypoxia (CIH), a condition that mimics the hypoxemia associated with OSA, is also associated with increases in blood pressure that persist even during periods of normoxia (2–4). The repeated bouts of hypoxia increase peripheral renin-angiotensin system activity, producing increases in circulating ANG II (1, 5). ANG II activates neurons in the subfornical organ (SFO), initiating a presumptive increase in the activity of the brain renin-angiotensin system (6, 7). The sustained hypertension associated with CIH has been shown to be dependent on angiotensin type 1 (AT1a) receptors in the SFO (8). Furthermore, AT1a receptors within the blood-brain barrier, specifically within the median preoptic nucleus (MnPO), have been shown to be necessary for CIH-induced diurnal hypertension (9).

The MnPO is involved in regulating cardiovascular function, as well as body fluid homeostasis, circadian rhythm, and thermoregulation (10, 11). ANG II signaling in the MnPO plays a particular role in the regulation of cardiovascular function. CIH is associated with an increase in ΔFosB staining in the MnPO, an activity-dependent transcription factor associated with neural adaptations (8, 12). Virally mediated expression of a dominant-negative construct against ΔFosB in MnPO prevented the sustained component of CIH hypertension that is expressed during normoxia (13). After 7 days of CIH, mRNA for angiotensin-converting enzyme 1, neuronal and endothelial nitric oxide synthase, and AT1a receptor in the MnPO was significantly increased, and these changes were blocked by dominant-negative inhibition of ΔFosB (9, 13). Moreover, blockade of AT1a receptors in the MnPO prevents the CIH-induced increase in ΔFosB as well as the persistent increase in blood pressure associated with CIH (9). More recently, the role of the MnPO in CIH hypertension was tested by injecting the paraventricular nucleus of the hypothalamus (PVN) with a retrograde adeno-associated viral vector expressing CRE and injecting the MnPO with an adeno-associated viral vector expressing a Cre-dependent caspase-3 (14). This approach effectively lesioned PVN-projecting MnPO neurons and blocked the development of the sustained component of CIH hypertension expressed during normoxia. Together these observations suggest that AT1a receptors expressed in MnPO participate in early changes in network function that are necessary for the development of sustained hypertension associated with CIH.

Our previous study demonstrated that ANG II can influence GABA_A-mediated inhibition in the MnPO by altering the phosphorylation status of the K⁺/Cl⁻ cotransporter KCC2 (15). It is currently unclear how CIH or chronic activation of AT1a influences MnPO excitability/inhibitory balance particularly with respect to the AT1a-mediated influence on GABA_A inhibition. The current studies addressed these questions using a viral knockdown approach to determine how CIH and AT1a receptors influence MnPO neurons. In these studies, loose patch recording and live-cell chloride imaging were used to test for changes in cellular responses to GABA_A receptor activation.

METHODS

Animals

Experiments were performed according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals (8th edition) and the University of North Texas Health Science Center Institutional Animal Care and Use Committee. Experiments used a total of 66 six-week-old (250–300 g) adult male Sprague-Dawley rats (Charles River Laboratory, Wilmington, MA). Rats were individually housed in temperature-controlled rooms with a 12-h light/dark cycle with the light phase lasting from 0700 to 1900 h. Plastic housing cages were connected to a closed air filtration system. Cages were filled with corn cob bedding and shredded paper for enrichment and were changed weekly. Rats had ad libitum access to water and standard rat chow (LabDiet, St. Louis, MO). Surgeries were performed using aseptic techniques, and postoperative infection was prevented by subcutaneous administration of procaine penicillin G (30,000 U). A nonsteroidal anti-inflammatory drug, carprofen (Rimadyl, 2 mg orally), was given before and after surgery for pain management. Rats were given 1 wk to acclimate to the vivarium before use in experiments. Viral vectors were then injected in the rats’ brain. One week following viral injections, some rats were implanted with radio telemetry devices. Two weeks following viral injections, rats were transferred to a CIH chamber for 5 days of baseline exposure followed by 7 days of CIH. At the end of the CIH protocol, rats were euthanized for in vitro studies.

Viral Vectors and Injections

Viral vectors used in these experiments were purchased from GENEDETECT (Auckland, New Zealand). The adeno-associated viruses (rAAV1/2) contained either a small hairpin RNA sequence targeted to the AT1a receptor (AT1a KD, GD1009-RV Rat AT1aR RNAi) or a scrambled (Scr, rAAV1/2-U6-SCR.shRNA-terminator-CAG-EGFP-WPRE-BGH-polyA) sequence. Viruses were used undiluted at a titer of 1.1 × 10¹² genomic particles/mL. Both viral constructs included a green fluorescent protein (GFP) reporter to allow verification of the injection sites. DIO-ClopHensorN was used undiluted at a titer of 1.0 × 10¹³ genomic particles/mL (UNC Vector Core), and Cre (AAV9-CMV-HI-eGFP-Cre-WPRE-SV40, Addgene) was used undiluted at a titer of 7.0 × 10¹² genomic particles/mL.

Microinjections

Rats were injected with viral vectors as previously described (15). Rats were anesthetized with 2% isoflurane, and their scalps were shaved and disinfected with alcohol and iodine. Each rat was placed in a stereotaxic head frame (David Kopf, Tujunga, CA). To ensure accurate injections, skulls were leveled between the two cranial suture landmarks bregma and lambda. The injector was angled 8° from medial to lateral, and the injection coordinates used for the MnPO were AP = 0 mm, ML = 0.9 mm, and DV = −6.7 mm from bregma (16). After a burr hole was drilled at the site of injection, a 30-gauge steel injector was lowered to the MnPO and a single 200–300 nL volume of AAV was delivered at a rate of 200 nL/min. The injector was connected to a Hamilton 5-µL syringe (No. 84851 Hamilton, Reno, NV) with calibrated polyethylene tubing used to determine the injection volume. The injector remained in place for 5 min to allow for absorption of AAV before being slowly withdrawn. Gel foam was packed into the burr hole, and absorbable antibiotic sutures were used to close the incision site and minimize postsurgical infection. In a separate group of rats, DIO-ClopHensorN was injected into the MnPO and a retrograde virus expressing Cre was injected into the PVN (bilateral injections; AP = −1.8, ML = ±0.4, DV = −7.6) as described (14). All the rats were given at least 2 wk to recover and to allow for the expression of viral constructs before the slice preparation.

Radio Telemetry Implantation

Rats received radio telemetry implants as previously described (8, 17). One week following microinjections, all rats were implanted with a Data Sciences International (DSI St. Paul, MN) TA11PA-C40 telemetry unit. Telemetry transmitters were used in conjunction with Dataquest A.R.T 4 receivers and data acquisition system. Prior to implantation, rats were anesthetized with 2% isoflurane and their abdomens shaved, cleaned, and sterilized with 100% ethanol and betadine. A midline incision was made in the abdomen from the xyphoid process to the start of the pelvis. The abdominal aorta was located, isolated, and punctured with a 23-gauge needle so that the tip of the telemetry could be inserted. Vetbond (3M, St. Paul, MN) and cellulose were used to close the blood vessel and adhere the telemetry unit to the abdominal aorta. The battery of the telemetry unit was secured to the abdominal muscle with prolene sutures and the incision site closed with absorbable Vicryl antibiotic sutures. Animals were allowed 7 days to recover following telemetry implantation. During the experiments, heart rate (HR), activity (ACT), and mean arterial pressure (MAP) were measured for 10 consecutive seconds every 10 min for 24 h.

Chronic Intermittent Hypoxia

Rats were exposed to 7 days of CIH as previously described (8, 17). All rats were transferred to the room containing the CIH chambers 1 wk following telemetry instrumentation and 2 wk following virus microinjections. Rats exposed to CIH were housed in an 8-in. × 9-in. cage that was placed inside of custom Plexiglas chambers. Rats were housed in these chambers for a 5-day baseline period before the start of the CIH protocol. The CIH protocol consisted of 6-min cycles; 3 min of 21% oxygen room air pumped in, 90 s of nitrogen pumped in to lower the chamber O₂ to 10% oxygen, and then 90 s of maintenance at 10% oxygen. This cycle repeated 10 times per hour, 8 h a day (0800–1600 h) for 7 consecutive days. Animals were exposed to room air for the remainder of the day. Normoxic controls were housed in the same room under similar conditions but were only exposed to room air.

Slice Preparation

After 7 days of CIH, rats were anesthetized with isoflurane (2%) and decapitated. A subset of rats (ClopHensorN injected) did not receive radiotelemetry implants, and instead, trunk blood was collected immediately following decapitation for the measurements of hematocrit and advanced oxidative protein products (AOPPs). Whole blood was collected in duplicate in heparinized capillary tubes and centrifuged at 6,500 g for 3 min, and hematocrit was measured as the percentage volume of red blood cells to total volume. Whole blood was also collected in heparinized tubes and centrifuged at 1,000 g for 15 min at 4°C. Plasma was collected for analysis of AOPP. Hematocrit and AOPPs were used as a proxy for CIH-induced blood pressure. Coronal slices (300 µm) containing the MnPO were cut using a Microslicer DTK Zero 1 (Ted Pella, Inc., Redding, CA) in ice-cold (0°C–1°C), oxygenated (95% O₂, 5% CO₂) cutting solution consisting of (in mM): 3.0 KCl, 1.0 MgCl₂·6H₂O, 2.0 CaCl₂, 2.0 MgSO₄, 1.25 NaH₂PO₄, 26 NaHCO₃, 10 d-glucose, and 206 sucrose (300 mosM, pH 7.4). Slices were incubated at room temperature in oxygenated (95% O₂, 5% CO₂) artificial cerebrospinal fluid (aCSF) containing (in mM): 126 NaCl, 3.0 KCl, 2.0 CaCl₂, 2.0 MgSO₄, 1.25 NaH₂PO₄, 26 NaHCO₃, and 10 d-glucose (300 mosM, pH 7.4) for a minimum of 1 h before recording.

ClopHensorN Imaging

Slices containing the MnPO were transferred to a submersion recording chamber where they were superfused with oxygenated aCSF (31 ± 1°C, 1–2 mL/min). Slices were visualized using an upright epifluorescent microscope (BX50WI, Olympus) with differential interference contrast optics. MnPO neurons expressing both enhanced green fluorescent protein (eGFP) and tdTomato fluorescence were identified for ClopHensorN imaging. Image sets were captured every 3 s consisting of eGFP (445/500–550 nm Ex/Em) and tdTomato (556/580–653 nm Ex/Em). The ratio of eGFP to tdTomato fluorescence was measured in response to 10-s focal application of the GABA_A agonist muscimol (100 nM in aCSF) using a glass micropipette (1–2-μM tip) connected to a picopump (WPI, Sarasota, FL) set to 10 psi.

Electrophysiology

Slices containing the MnPO were transferred to a submersion recording chamber where they were superfused with oxygenated aCSF (31 ± 1°C, 1–2 mL/min). Slices were visualized using an upright epifluorescent microscope (BX50WI, Olympus) with differential interference contrast optics. Loose patch-clamp recordings were obtained using borosilicate glass micropipettes (1–3 MΩ) containing aCSF as the internal solution. Action potentials were amplified and digitized using Multiclamp 200B and Digidata 1440A, respectively (Molecular Devices, San Jose, CA). Signals were filtered at 2 kHz and digitized at 10 kHz. Recordings from MnPO neurons were made from eGFP-expressing neurons in slices prepared from rats injected with the shRNA against AT1aRs or the scrambled control construct.

Baseline action potential firing was recorded for 5 min. Drugs were focally applied for 30 s (10 psi) using a picopump (WPI, Sarasota, FL) with a patch pipette containing the drug placed 150–200 μm upstream of the recording electrode followed by an additional 10 min of recording. In some cases, focal application of muscimol (100 µM) occurred in the presence of a bath-applied KCC2 blocker (VU0240551, Tocris, 10 µM) or a Na-K-2Cl cotransporter isoform 1 (NKCC1) blocker (bumetanide, Tocris, 10 µM). In these instances, bath application of blockers occurred before baseline recordings, and these were maintained throughout the recording procedure. Compounds used were the AT1a agonist ANG II (Sigma-Aldrich, St. Louis, MO) and the GABA_A agonist muscimol (Sigma-Aldrich). ANG II was dissolved in aCSF. Stock solutions of muscimol (100 mM) were first dissolved in 0.05 M HCl and then diluted to a final concentration in aCSF (100 µM). VU0240511 and bumetanide were purchased from Tocris (Bristol, UK) and were first dissolved in DMSO (Sigma-Aldrich, St. Louis, MO) and then diluted to a final concentration of 10 µM (<0.1% DMSO) in aCSF for bath application. Parameters measured for action potential firing included spike count separated into 30-s bins.

Western Blot

Two weeks after microinjection of the AAV vectors, a subset of rats were exposed to either 7 days of CIH or 7 days of normoxia. At the end of the CIH protocol, rats were deeply anesthetized with Inactin (100 mg/kg ip, Sigma Aldrich) and then decapitated. Brains were removed and cut into 1-mm-thick coronal sections using a commercially available brain matrix. Punches containing the MnPO (2 or 3 per rat) were collected using a blunt-tip 23-gauge needle and stored at −80°C until protein isolation and Western blots were performed. Tissue was homogenized in a RIPA lysis buffer containing DTT, chelators, and protease inhibitors. Total lysate (20–25 µg) was separated via electrophoresis on 15% acrylamide sodium dodecyl sulfate (SDS) gel (Cat. No. 465-1084, Bio-Rad, Hercules, CA) in Tris-glycine buffer at 100 V for 1–2 h. Samples were then transferred to a PVDF membrane (Immobilon-P; EMD-Millipore) in Tris-glycine buffer (25 mM Tris, 192 mM glycine, 0.1% SDS; pH 8.3) containing 20% methanol (vol/vol) at 50 V for 2 h at 4°C. Membranes were blocked with 5% BSA in TBST (25 nM Tris base, 125 mM NaCl, 0.1% Tween 20) for 30 min at room temperature. Blots were incubated in primary antibodies for Ser940 pKCC2 (1:500 rabbit polyclonal, 612-401-E15 Rockland Immunochemicals Inc., Limerick, PA), KCC2 (1:300 rabbit polyclonal, 07–432 Millipore Sigma, Burlington, MA), pNKCC1 (1:1,000 rabbit polyclonal, AB5561 Millipore Sigma, Burlington, MA), NKCC1 (1:1,000 mouse monoclonal, ab98968 Abcam, Cambridge, UK), or β-actin (1:2,000 mouse monoclonal, A5441, Sigma-Aldrich, St. Louis, MO) overnight at 4°C. Blots were washed with TBST and incubated in species-appropriate secondary antibodies (1:2,000) for 2 h at room temperature. Proteins were detected by chemiluminescence using a Supersignal West Femto Maximum Sensitivity kit (Thermo Scientific) and imaged with a G-Box (Syngene, Frederick, MD). Membranes were then washed in TBST and stripped with Restore PLUS stripping buffer (Thermo Scientific, Waltham, MA) before being reprobed. Separate blots were used for the analysis of KCC2 and NKCC1 protein expression. Densitometry was analyzed using ImageJ, and quantification of protein was normalized to β-actin or presented as a ratio of phosphorylated protein to total protein.

Statistics

Data from loose patch recordings were analyzed for group and time-dependent effects using two-way repeated-measures ANOVA or three-way ANOVA followed by Holm–Sidak post hoc tests. Mann–Whitney U tests were used to make baseline and area under the curve (AUC) comparisons. Analyses were performed using SigmaPlot software (v. 12.0, Systat Software Inc., San Jose, CA).

RESULTS

Telemetry Data

A total of 30 rats (Norm/Scr n = 6, Norm/AT1a KD n = 6, CIH/Scr n = 8, CIH/AT1a KD n = 10) were used in telemetry recordings and subsequent loose patch recordings with ANG II and muscimol. Mean arterial pressure was recorded from rats for the 5 days before, as well as throughout, the CIH protocol. There were no differences in baseline MAP among any of the treatment groups during either the light phase (0800–1600 h, coincident with CIH induction time) or the dark phase (1600–0800 h; Table 1). This was consistent with previous reports that MnPO AT1a receptor knockdown does not influence basal MAP (9). During CIH exposure, there was a significant difference between treatment groups during the light phase [one-way ANOVA, F(3,26) = 7.27, P < 0.01]. The increase in MAP of rats in the CIH/Scr group was significant as compared with Norm/Scr and Norm/AT1a KD rats (P < 0.01, Fig. 1A). In the CIH/AT1a KD-treated rats, the change in MAP associated with CIH was attenuated. Although the change in MAP of the CIH/AT1a KD group was greater than the change in the Norm/Scr group, this trend was not statistically significant (P = 0.06). In contrast, the changes in MAP of the CIH/AT1a KD group were significantly decreased as compared with those of the CIH/Scr group during the normoxic dark phase [one-way ANOVA, F(3,28) = 8.62, P < 0.001, Fig. 1B]. Furthermore, MAP from Norm/AT1a KD rats was reduced compared with that from CIH/Scr rats (P < 0.05) and was not different from Norm/Scr rats (P = 0.91). This suggests that knockdown of MnPO AT1aR blocked the persistent hypertension during the dark period in the CIH-treated rats but had no effect on MAP in normoxic rats.

Table 1.

Average MAP recorded from each group from 0700 to 1900 h (light) and from 1900 to 0700 h (dark) before the 7-day intermittent hypoxia protocol

	Norm/Scr	Norm/AT1a	CIH/Scr	CIH/AT1a
n	6	6	8	10
MAP, mmHg
Light	106.2 ± 0.9	107.9 ± 2.7	100.6 ± 2.0	105.2 ± 1.5
Dark	109.2 ± 1.22	109.5 ± 1.9	103.3 ± 1.5	105.6 ± 1.4

CIH/AT1a, chronic intermittent hypoxia/angiotensin type 1; CIH/Scr, chronic intermittent hypoxia/scrambled; MAP, mean arterial pressure; Norm/AT1a, Normoxic/angiotensin type 1; Norm/Scr, normoxic/scrambled.

Figure 1.

Exposure to 7 days of chronic intermittent hypoxia (CIH) was associated with an increase in mean arterial pressure (MAP; A) in scrambled (Scr) rats and a trend toward increased MAP in angiotensin type 1 (AT1a) KD rats. This increase in MAP persists during the normoxic dark phase (B) in CIH/Scr rats but not in CIH/AT1a rats. Normoxic/scrambled (Norm/Scr): n = 6 rats, Norm/AT1a: n = 6 rats, CIH/Scr: n = 8 rats, CIH/AT1a: n = 10 rats. *P < 0.05, **P < 0.01 compared with Norm/Scr; #P < 0.05 vs. CIH/AT1a using Holm-Sidak post hoc test.

CIH/AT1a KD Electrophysiology Data

The AT1a KD was effective in blocking the excitatory effects of ANG II in the MnPO (Fig. 2, A–C). The interaction of CIH and AT1a KD on the basal activity of MnPO neurons was investigated. There was a significant difference in the basal firing rate between the groups [one-way ANOVA, F(3, 89) = 3.52, P < 0.05]. The basal firing rate of neurons from CIH/AT1a KD-treated rats was significantly greater as compared with that of cells from Norm/Scr rats (Fig. 2D, CIH/AT1a KD 3.2 ± 0.4 Hz, n = 29; Norm/Scr 1.5 ± 0.3 Hz, n = 19; Holm–Sidak P < 0.05). There was also a trend toward an increase in basal firing rate of cells from rats in the Norm/AT1a KD group (3.1 ± 0.5 Hz, n = 21, Holm–Sidak, P = 0.057) when compared with the activity of neurons from Norm/Scr rats. The basal firing rate of neurons from CIH/Scr-treated rats (2.4 ± 0.4 Hz, n = 24) was intermediate between the AT1a KD and Norm/Scr groups.

Figure 2.

Effects of adeno-associated virus (AAV)-mediated angiotensin type 1 (AT1a) KD on firing rate and responses to ANG II. A: representative recordings of AT1a KD were effective in blocking the response of median preoptic nucleus (MnPO) neurons to exogenously applied ANG II. Associated firing rate histograms showing scrambled (Scr; B) neurons show an increase in firing rate in response to ANG II but AT1a KD neurons (C) do not. D: chronic intermittent hypoxia/angiotensin type 1 (CIH/AT1a) neurons exhibit a higher basal firing rate than normoxic/scrambled (Norm/Scr). Both Norm/AT1a and CIH/Scr also trend toward increased basal firing compared with Norm/Scr. In D, n = 19–29 per group and represents the number of neurons. *P < 0.05 vs. Norm/Scr. Analyses were conducted using Holm-Sidak post hoc test.

Consistent with previous reports (15), ANG II increased the spontaneous firing of MnPO neurons from CIH-treated rats injected with the AAV containing the control vector (Scr), whereas CIH-treated rats injected with the AT1a KD AAV were insensitive to ANG II application. There was a 350% increase in firing in Norm/Scr neurons (n = 9, Fig. 3B) and a 270% increase in firing in CIH/Scr neurons (n = 11, Fig. 3C) in response to ANG II. The attenuated fold increase in the activity of the cells from CIH/Scr rats was not significant when compared with the activity of cells from the Norm/Scr rats (P = 0.45) and may be related to the increased basal firing rate of neurons from CIH/Scr rats. The absolute firing rate increase between the Norm/Scr (3.6 ± 1.2 Hz) and CIH/Scr (3.7 ± 0.6 Hz) groups was not significantly different (P = 0.95). Focal application of ANG II did not significantly influence the activity of MnPO neurons from AT1a KD groups. The statistical analysis indicated a significant interaction between time and treatment on the firing rate of MnPO neurons [two-way ANOVA F(87,1259) = 2.40, P < 0.01, Fig. 3D]. The firing rate of Norm/Scr neurons was increased following ANG II application, and this increase persisted for 6 min. Neurons from CIH/Scr rats also responded to ANG II with an increase in firing; however, the duration of the response was reduced. The increase in firing in CIH/Scr neurons was only significantly elevated at 1 min following ANG II application. Neither the Norm/AT1aKD nor the CIH/AT1aKD neurons responded to ANG II application.

Figure 3.

Effects of angiotensin type 1 (AT1a) knockdown on ANG II responses of median preoptic nucleus (MnPO) neurons from rats exposed to 7 days of normoxia or chronic intermittent hypoxia (CIH). A: representative examples of electrophysiological responses of MnPO neurons from CIH rats injected with the control vector [scrambled (Scr)] or shRNA against the AT1a receptor (AT1a KD) to brief application of ANG II (100 nM). B: summary data as a percent of baseline activity showing that ANG II injections enhance spontaneous activity in Scr but not AT1a KD-injected rats exposed to normoxia [normoxic/angiotensin type 1 (Norm/AT1a) n = 10, Norm/Scr n = 9]. C: summary data showing that AT1a KD significantly reduces the responses to ANG II in MnPO neurons from rats exposed to CIH (CIH/AT1a, n = 12) as compared with neurons from rats exposed to CIH that were injected with the control virus (CIH/Scr, n = 11). Each n represents the number of neurons for the group. D: ANG II responses, analyzed as area under the curve (AUC) measurements, are blunted in CIH/Scr-treated and are absent in AT1a KD-injected rats regardless of treatment. *P < 0.05, **P < 0.01 vs. Norm/Scr. Analyses are Mann-Whitney U tests.

CIH/AT1a KD Muscimol Effects

To test the role of GABA_A-mediated inhibition in AT1a KD- and CIH-treated MnPO neurons, muscimol (100 µM) was focally applied to GFP-positive neurons. In CIH-treated rats, muscimol inhibits MnPO neurons in the Scr-injected but not AT1a KD-injected rats (Fig. 4A). Neurons from Norm/Scr rats (n = 10) exhibited a basal firing rate of 1.66 ± 0.4 Hz, which was reduced to 0.02 ± 0.01 Hz immediately following muscimol application. Focal application of muscimol induced a robust and transient inhibition in these neurons that persisted for 6 min (Fig. 4B). Neurons from Norm/AT1a KD rats (n = 10) showed a basal firing rate of 3.08 ± 0.8 Hz, which was reduced to 0.17 ± 0.1 Hz immediately following muscimol application. Muscimol-dependent inhibition was attenuated in MnPO neurons from Norm/AT1a KD rats as compared with neurons from Norm/Scr rats. After CIH, neurons from CIH/Scr rats (n = 13) demonstrated a basal firing rate of 2.46 ± 0.6 Hz that was decreased to 1.39 ± 0.7 Hz immediately after muscimol application. Neurons from CIH/AT1a KD rats (n = 15) showed a basal firing rate of 2.78 ± 0.6 Hz that was decreased to 0.58 ± 0.2 Hz immediately after muscimol application (Fig. 4C). However, the neurons from CIH/AT1a KD rats showed a delayed increase in firing rate that peaked at 3.27 ± 0.06 Hz 5 min after muscimol application. The increase in firing associated with muscimol in neurons from CIH/AT1a KD rats persisted for the duration of the 10-min recording. There was a time-dependent effect [repeated-measures two-way ANOVA, F(20,830) = 3.14, P < 0.001] but not a treatment-dependent effect [repeated-measures two-way ANOVA, F(3,42) = 0.85, P= 0.47, Fig. 4D]. However, there was a difference between groups when assessing the area under the curve [one-way ANOVA, F(3,45) = 2.99, P < 0.05]. The CIH/Scr, CIH/AT1a, and Norm/AT1a groups all showed a robust attenuation of muscimol inhibition compared with the Norm/Scr group (P < 0.01).

Figure 4.

Effects of intermittent hypoxia on GABA_A-mediated inhibition of median preoptic nucleus (MnPO) neuron. A: representative recordings demonstrating the effects of muscimol (100 µM) on the activity of MnPO neurons from rats exposed to chronic intermittent hypoxia (CIH) and injected with either the control vector [chronic intermittent hypoxia/scrambled (CIH/Scr)] or an adeno-associated virus (AAV) expressing shRNA against the angiotensin type 1 (AT1a) receptor (CIH/AT1a). B: GABA_A-mediated inhibition is blunted in normoxic AT1a KD [normoxic/scrambled (Norm/Scr) n = 10, Norm/AT1a n = 10]. C: GABA_A activation produces blunted inhibition in CIH/Scr and excitation in CIH/AT1a MnPO neurons (CIH/Scr n = 13, CIH/AT1a n = 15). Each n represents the number of neurons for that group. D: GABA_A-mediated area under the curve measures are blunted in Norm/AT1a and CIH/Scr MnPO neurons. GABA_A activation in CIH/AT1a MnPO neurons enhances area under the curve (AUC). **P < 0.01 vs. Norm/Scr. Analyses are Mann-Whitney U tests.

ClopHensorN Data

A total of six rats were used for ClopHensorN studies. Brain slices containing the MnPO were made after 7 days of CIH or normoxia from rats that had been injected with the Cre-dependent ratiometric Cl⁻ indicator ClopHensorN (Fig. 5, A and B). Fluorescent intensity was measured at 3-s intervals in response to focal application of the GABA_A agonist muscimol. In normoxic rats (n = 3), muscimol produced a reduction in ratiometric ClopHensorN signal in 34 of 123 recorded cells, indicating Cl⁻ influx (Fig. 5C). In addition, 25 of 123 recorded cells showed an increase in ratiometric ClopHensorN signal, suggesting Cl⁻ efflux in these neurons. The remaining 64 cells recorded did not exhibit a detectable change in the ratiometric signal following muscimol application. MnPO neurons expressing ClopHensorN from CIH-treated rats (n = 3) showed an increase in the ratiometric signal in 57 of 128 neurons after muscimol application (Fig. 5D), with the remaining 71 neurons showing no response. None of the ClopHensorN transfected neurons showed a decrease in the ratiometric signal, and the 70 remaining neurons showed no change in ratiometric signals in response to muscimol. There was a significant difference between normoxic and CIH-treated rats in the ClopHensorN response in neurons treated with muscimol (t test, P < 0.001, Fig. 5E).

Figure 5.

A: ClopHensorN is a genetically encoded ratiometric chloride indicator consisting of both a [Cl⁻]-sensitive green fluorescent protein (GFP) and a [Cl⁻] insensitive tdTomato fluorophore. B: expression of Cre-dependent ClopHensorN in the median preoptic nucleus (MnPO) was induced by injections of a retrograde Cre virus in the paraventricular nucleus of the hypothalamus (PVN). Chloride influx (C) is observed as a reduction in the fluorescent ratio of GFP:tdTomato in MnPO neurons from normoxic-treated rats in response to muscimol application. Chloride efflux (D) is observed as an increase in the fluorescent ratio of GFP:tdTomato in MnPO neurons from chronic intermittent hypoxia (CIH)-treated rats in response to muscimol application. E: the peak response to muscimol is negative in normoxic neurons but positive in CIH neurons. Normoxic (Norm) n = 59 neurons from three rats, CIH n = 57 neurons from three rats, ***P < 0.001 by unpaired t test.

Western Blot

Based on the observed changes in muscimol responses, Western blot analysis was used to test for changes in the abundance of KCC2 and pKCC2 in MnPO samples collected from rats exposed to CIH and AT1a KD using Western blot analysis (Fig. 6A). A total of 23 rats (Norm/Scr n = 5, Norm/AT1a KD n = 6, CIH/Scr n = 6, CIH/AT1a KD n = 6) were used for Western blot analyses. There was a significant difference between groups in the expression of total KCC2 [one-way ANOVA, F(3,19) = 5.93, P < 0.001, Fig. 6B]. In normoxic rats, AT1a KD resulted in a slight, but not significant, increase in total KCC2 protein levels (when expressed as a ratio to β-actin). Scr-injected rats exposed to CIH, however, showed an increase in the total KCC2 (P < 0.01 compared with Norm/Scr). When CIH was combined with AT1a KD, the total KCC2 was reduced, though not significantly, compared with the CIH/Scr group and was comparable with both normoxic groups. However, there was no significant effect of treatment on pKCC2 expression [one-way ANOVA, F(3,19) = 2.03, P = 0.15, Fig. 6C]. There was also a significant difference between groups when protein levels were measured as a ratio between pKCC2 and total KCC2 [one-way ANOVA, F(3,19) = 9.15, P < 0.001, Fig. 6D]. AT1a KD decreased the pKCC2-to-total KCC2 ratio in the normoxic treated rats (P < 0.05). CIH alone also decreased the pKCC2-to-total KCC2 ratio when compared with the Norm/Scr rats (p < 0.001). AT1a KD in CIH-treated rats increased the ratio of pKCC2 to total KCC2 compared with in CIH/Scr (P < 0.05) rats but not to the level of Norm/Scr rats (P < 0.05). The pKCC2-to-total KCC2 ratio was similar between CIH/AT1a KD and Norm/AT1a KD groups.

Figure 6.

A: representative Western blots were used to assess the effects of normoxic/chronic intermittent hypoxia (Norm/CIH) and scrambled/angiotensin type 1 (Scr/AT1a) on expression of KCC2 and pKCC2 in median preoptic nucleus (MnPO) neurons. An increase in total KCC2:β-actin (B) was only observed in CIH/Scr. There was no change in pKCC2:β-actin (C) in any of the groups tested. There was a reduction in pKCC2:KCC2 (D) in all groups compared with Norm/Scr, and the reduction in pKCC2:KCC2 observed in CIH/Scr was attenuated in CIH/AT1a. Norm/Scr n = 5, Norm/AT1a n = 6, CIH/Scr n = 6, and CIH/AT1a n = 6. Each n represents the numbers of rats for each group. *P < 0.05, **P < 0.01, ***P < 0.001 vs. Norm/Scr; #P < 0.05 vs. CIH/Scr. Analyses used Holm-Sidak post hoc test.

Role of Cl⁻ Cotransporters in CIH

A pharmacological approach was used to further characterize the possible mechanisms associated with the CIH-dependent changes in GABA_A-mediated inhibition. A total of seven rats (AT1a KD n = 4, Scr n = 3) were used for functional tests of KCC2 and NKCC1. First, in MnPO neurons from rats exposed to CIH, bath application of the KCC2 blocker VU0240551 (Fig. 7, A and B) increased the inhibition associated with muscimol application when compared with bath application of vehicle [Scr/VU (n = 7 cells); Scr/VEH (n = 8 cells), P < 0.01, compared with the vehicle]. In addition, cells from rats in the CIH and AT1a KD groups also appear to have a more inhibitory response to muscimol in the presence of VU0240551, but this trend was not statistically significant (AT1a/VU n = 9 cells).

Figure 7.

Effects of KCC and NKCC inhibition on GABA_A-mediated inhibition of median preoptic nucleus (MnPO) neurons. A: representative examples of responses to muscimol (100 µM) in MnPO neurons from chronic intermittent hypoxia (CIH)-treated rats injected with scrambled (Scr) and angiotensin type 1 (AT1a) KD incubated with KCC2 inhibitor VU 0240551 (10 µM). B: KCC2 blockade recovers muscimol (100 µM) inhibition in CIH/Scr/VU and blocks muscimol excitation in CIH/AT1a/VU (CIH/Scr/VU n = 7, CIH/AT1a/VU n = 9). C: representative examples of responses to muscimol (100 µM) in MnPO neurons from CIH-treated rats injected with Scr and AT1a KD incubated with NKCC1 inhibitor bumetanide (10 µM). D: NKCC1 blockade recovers muscimol (100 µM) inhibition in CIH/Scr/Bumet and blocks muscimol excitation in CIH/AT1a/Bumet (CIH/Scr/Bumet n = 9, CIH/AT1a/Bumet n = 6). E: muscimol inhibition is attenuated in normoxic AT1a KD MnPO neurons [normoxic/scrambled (Norm/Scr) n = 10, Norm/AT1a n = 10]. F: GABA_A activation produces blunted inhibition in CIH/Scr and excitation in CIH/AT1a MnPO neurons (CIH/Scr n = 13, CIH/AT1a n = 15). Each n represents the numbers of neurons. G: summary of area under the curve (AUC) data. **P < 0.01 vs. CIH/Scr. Analyses used Holm-Sidak post hoc test.

Bath application of the NKCC1 blocker, bumetanide (Fig. 7, C and D), also enhanced the inhibitory effects of muscimol focally applied to CIH/Scr treated MnPO neurons (P < 0.01). Bath application of bumetanide also changed the excitatory effects of muscimol on CIH/AT1a KD neurons to an inhibitory effect, though again, this was not statistically significant. AT1a KD reduced muscimol-mediated inhibition in MnPO neurons from normoxic treated (Fig. 7E) and CIH-treated (Fig. 7F) rats. In the rats that were treated with CIH, there was an interaction between the Cl⁻ cotransporter blocker and the viral injection [two-way ANOVA, F(2,53) = 3.66, P < 0.05, Fig. 7G].

DISCUSSION

The renin-angiotensin system contributes to the sustained hypertension observed in rats exposed to chronic intermittent hypoxia (8). In CIH, circulating ANG II activates neurons in circumventricular organs of the brain (18) and drives activity in the MnPO, which is, itself, sensitive to ANG II (9, 11). In the MnPO, ANG II has been shown to enhance the excitability of these neurons by reducing afterhyperpolarizations (19), decreasing leak potassium currents, or inducing calcium spikes (20). The MnPO projects to the PVN (11, 21–23), and this projection contributes to driving increases in sympathetic nerve activity (24). Lesions of PVN-projecting MnPO neurons prevent the sustained hypertensive response associated with CIH (14). The brain renin-angiotensin system also contributes to CIH hypertension. Infusion of AT1a antagonists in the cerebral ventricles (12) prevents CIH hypertension but does not indicate which populations of AT1a receptors mediate this effect. Losartan injection in the PVN (25) and virally mediated knockdown of either angiotensin-converting enzyme (26) or AT1a receptors (9, 15) in MnPO have all been shown to prevent CIH hypertension, suggesting that these regions contribute to CIH hypertension. Based on the tropism of the AAV vectors used and the lack of expression of the reporter genes in astrocytes in previous studies (8, 9, 15, 26), these effects appear to be mediated by MnPO neurons. After 7 days of CIH, there is an increase in the activity of MnPO neurons and reduced inhibition that may represent cellular changes that contribute to CIH hypertension (15). In the current study, we tested whether angiotensin receptors in MnPO contribute to changes in the activity of these neurons associated with CIH.

The current study suggests that both CIH and AT1a influence chloride transport through different mechanisms (Fig. 8). Although CIH alone showed a trend in the increased basal firing rate of MnPO neurons, AT1a KD alone was more effective in enhancing basal activity. The combination of CIH and AT1a KD showed the greatest increase in basal firing rate of MnPO neurons. Neurons injected with the Scr virus showed increases in activity in response to ANG II where the AT1a KD neurons were unresponsive to ANG II, showing that the shRNA against the AT1aR is effective in knocking down the function of AT1aRs and impairing the excitatory actions of ANG II in the MnPO. The ANG II-mediated increases in firing in the Scr neurons were slower to develop and more transient in the CIH-treated rats. Although the magnitude of the ANG II-mediated increase was not as robust in the CIH-treated animals, these cells had a higher baseline firing rate and reached the same peak firing rate as the normoxic controls. It is currently unclear what the mechanism and the functional consequence are for the more transient ANG II effect in the CIH/Scr group. It is possible that the increased transient nature of the ANG II effects in MnPO neurons from Scr/CIH rats is due to a depolarization block caused by persistent ANG II-mediated increases in excitability (27–29) or an inhibition of tetrodotoxin-sensitive sodium currents (30). It may also be possible that ANG II mediates increases in calcium spikes that facilitate an increase in calcium-dependent K⁺ currents (31) or that ANG II activates a negative feedback mechanism involving nitric oxide (32). These proposed mechanisms may function here to prevent excitotoxicity in MnPO neurons and deserve further investigation.

Figure 8.

During normoxia, basal angiotensin II receptor type 1 (AT1R) activation mediates KCC2 function, which maintains a low intracellular concentration of Cl⁻ and contributes to the inhibitory action of GABA_A receptor activation. During chronic intermittent hypoxia (CIH), increases in AT1R activation results in a functional shift in the activity of NKCC1 and KCC2 that favors NKCC1. This functional change results in a depolarizing shift in the Cl⁻ gradient that makes GABA_A receptor activation less inhibitory, or in some cases, excitatory.

The effects of muscimol on MnPO activity suggest that the more transient effects of ANG II in the Scr/CIH group are not due to increases in GABA_A-mediated inhibition. The data from this study suggest that GABA_A activation has the potential to contribute to the depolarization block and are consistent with previous reports showing the excitatory effects of ANG II on MnPO neurons (15) as well as the efficacy of the AT1a shRNA on blocking the effects of ANG II on MnPO neurons (15).

As shown previously, the AT1a KD was effective in reducing the GABA_A-mediated inhibition in MnPO neurons from normoxic rats (15). This effect was related to a reduction in pKCC2, suggesting a depolarizing shift in the Cl⁻ gradient. The current study showed that CIH produces similar changes in MnPO neurons. In addition, this study used CRE-dependent expression of a ratiometric Cl⁻ indicator ClopHensorN to show that CIH impairs GABA_A-mediated inhibition in MnPO neurons that project to the PVN. Similar to our electrophysiological studies, the results of the live imaging experiments indicated that CIH not only reduced the inhibitory effect of muscimol (many ClopHensorN transfected neurons did not respond to muscimol) but also may change the valence of GABA_A receptor activation such that GABA_A activation can facilitate chloride efflux and contribute to neuronal excitation. Our results also suggest that there may be a temporally dynamic shift in Cl⁻ flux of CIH-treated MnPO neurons characterized by a rapid and transient efflux of Cl⁻ immediately upon muscimol application followed by a prolonged influx of Cl⁻ (Fig. 5D). Data from our electrophysiology experiments suggest that this influx may not directly influence the activity of the MnPO neurons since such complex effects were not observed in the cell-attached recordings (Fig. 4). As mentioned earlier, the AAV approach used for the chloride imaging experiments allowed us to focus on a subset of MnPO neurons. It is possible that this delayed Cl⁻ response seen in some of the live cell imaging experiments is specific to PVN-projecting MnPO neurons. However, we did not conduct simultaneous extracellular recordings in the ClopHensorN experiments.

Phosphorylation of KCC2, specifically at the S940 residue, is shown to be important in maintaining the membrane expression of the cotransporter (33). Because AT1a KD has been shown to reduce GABA_A-mediated inhibition of MnPO neurons from normoxic rats through a reduction in the expression of pKCC2 and subsequent function of KCC2 (responsible for Cl⁻ efflux), we tested KCC2 function in CIH rats. This study shows that the antagonist blockade of KCC2 increased the GABA_A-mediated inhibitory effects of muscimol in both the Scr/CIH and AT1a KD/CIH groups. Although previous studies showed that AT1a KD blocked KCC2 function in normoxic MnPO neurons (KCC2 blockade had no effect on GABA_A inhibition), the current study suggests that CIH activates mechanisms that contribute to KCC2 function independent of AT1a KD. It seems paradoxical that KCC2 inhibition would enhance GABA_A inhibition. One hypothesis is that the reduced function of KCC2 following CIH leads to a depolarizing shift in the Cl⁻ gradient but not a flip in the reversal potential and thus reduces the inhibitory role of GABA_A. Subsequent blockade of KCC2 function causes a further depolarizing shift in Cl⁻ gradient to the point where GABA_A activation is excitatory and thus contributes to depolarization block of MnPO neurons. Furthermore, it is unclear whether other G protein-coupled receptors that mediate KCC2 phosphorylation (33) are upregulated in the CIH condition that could compensate for a reduction in AT1a signaling.

The changes in KCC2 function associated with CIH do not occur in isolation and must also be considered in the context of NKCC1 function (responsible for Cl⁻ influx). Although attempts were made in this study to assess the expression of NKCC1 protein, we were unable to obtain reliable Western blot results. However, functional assessment of NKCC1 using pharmacological blockade enhanced GABA_A-mediated inhibition in both the Scr/CIH and AT1a KD/CIH groups. Evidence suggests that NKCC1 function is also increased (at least relative to KCC2 function), and this may contribute to the reduction in GABA_A-mediated inhibition. The functional data also suggest that whereas KCC2 function is mediated by AT1a receptor activation, the increased activity of NKCC1 is less dependent on AT1a receptors. Although ANG II activation of AT1a receptors has been shown to increase NKCC1 phosphorylation in smooth muscle (34, 35), it has not been demonstrated in neurons.

CIH produces a hypertensive response that is maintained throughout the 24-h cycle including periods of normoxia. The hypertension that persists during periods of normoxia can be blocked through the inhibition of AT1a receptors in the SFO and MnPO. Interestingly, AT1a KD is protective against the sustained hypertensive response associated with CIH, whereas it decreases GABA_A-mediated inhibition and increases spontaneous activity in vitro. In this and our previous studies, knockdown of AT1a receptors in MnPO did not influence basal blood pressure. This suggests that the changes in GABA_A inhibition and spontaneous activity associated with AT1a knockdown in MnPO are not translated into changes in sympathetic tone or blood pressure. The reduced ANG II-mediated excitation in the MnPO appears to prevent increased excitatory outflow to downstream nuclei that increase sympathetic outflow and blood pressure related to CIH. This reduction in ANG II-dependent excitation might mitigate changes in network function produced by CIH that increase blood pressure and alter negative feedback inhibition (GABA_A mediated) from baroreceptors.

Perspectives and Significance

The MnPO is involved in the regulation of body fluid homeostasis and cardiovascular function, and the interaction between AT1a and GABA_A receptors modulates the excitatory/inhibitory balance of MnPO neurons. The current study demonstrates an AT1a-dependent reduction of GABA_A-mediated inhibition in MnPO neurons following 7 days of CIH. AT1a receptor KD or CIH reduced GABA_A inhibition to similar levels. When AT1a KD was combined with CIH, GABA_A activation increased the excitability of MnPO neurons. A CIH-dependent change in GABA_A-mediated Cl⁻ flux was also demonstrated using the ratiometric Cl⁻ indicator ClopHensorN. The mechanisms underlying this shift in Cl⁻ flux involve functional changes in the Cl⁻ cotransporters KCC2 and NKCC1 causing dysregulation of the excitatory/inhibitory balance in MnPO neurons toward overall excitation. This increase in MnPO excitation facilitates the activation of sympathetic nerve activity and increases in blood pressure associated with CIH and OSA. Our previous work has shown that ΔFosB likely mediates some of the cellular adaptations that occur in the MnPO, which contributes to CIH hypertension (13), and that the AT1a receptor could be a downstream target of ΔFosB (9).

Our current working hypothesis is that posttranslational processing contributes to alterations of NKCC1 and KCC2 associated with CIH. Changes in NKCC1 and KCC2 mRNA have been observed in different models of epilepsy that are characterized by changes in GABAergic excitation (36). It is not clear if ΔFosB or related AP1 transcription factors influence KCC2 expression. Although some experiments demonstrate that fos expression is related to changes in KCC2 function, this may be due to the well-known relationship between fos and neural activity. It could be that ΔFosB plays a role in activity-dependent changes in NKCC1 and KCC2 functions associated with CIH indirectly by regulating the expression of the AT1a receptor or other parts of the signaling pathway that regulate chloride transporter expression and activity.

GRANTS

This study was supported by National Heart, Lung, and Blood Institute grants P01 HL088052 and R01 155977, and National Institute for Aging grant T32 AG020494.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

G.E.F., J.T.L., A.B.M., and J.T.C. conceived and designed research; G.E.F., J.T.L., and A.B.M. performed experiments; G.E.F. and J.T.L. analyzed data; G.E.F. and J.T.C. interpreted results of experiments; G.E.F. prepared figures; G.E.F. and J.T.C. drafted the manuscript; G.E.F., J.T.L., A.B.M., and J.T.C. edited and revised the manuscript; G.E.F., J.T.L., A.B.M., and J.T.C. approved the final version of manuscript.