Adenosine inhibits paraventricular pre-sympathetic neurons through ATP-dependent potassium channels

Authors


Address correspondence and reprint requests to De-Pei Li, Department of Critical Care, Unit 110, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030, USA. E-mail: dpli@mdanderson.org

Abstract

J. Neurochem. (2010) 113, 530–542.

Abstract

Adenosine produces cardiovascular depressor effects in various brain regions. However, the cellular mechanisms underlying these effects remain unclear. The pre-sympathetic neurons in the hypothalamic paraventricular nucleus (PVN) play an important role in regulating arterial blood pressure and sympathetic outflow through projections to the spinal cord and brainstem. In this study, we performed whole-cell patch-clamp recordings on retrogradely labeled PVN neurons projecting to the intermediolateral cell column of the spinal cord in rats. Adenosine (10–100 μM) decreased the firing activity in a concentration-dependent manner, with a marked hyperpolarization in 12 of 26 neurons tested. Blockade of A1 receptors with the adenosine A1 receptor antagonist 8-cyclopentyl-1,3-dipropylxanthine or intracellular dialysis of guanosine 5′-O-(2-thodiphosphate) eliminated the inhibitory effect of adenosine on labeled PVN neurons. Immunocytochemical labeling revealed that A1 receptors were expressed on spinally projecting PVN neurons. Also, blocking ATP-dependent K+ (KATP) channels with 100 μM glibenclamide or 200 μM tolbutamide, but not the G protein-coupled inwardly rectifying K+ channels blocker tertiapin-Q, abolished the inhibitory effect of adenosine on the firing activity of PVN neurons. Furthermore, glibenclamide or tolbutamide significantly decreased the adenosine-induced outward currents in labeled neurons. The reversal potential of adenosine-induced currents was close to the K+ equilibrium potential. In addition, adenosine decreased the frequency of both spontaneous and miniature glutamatergic excitatory post-synaptic currents and GABAergic inhibitory post-synaptic currents in labeled neurons, and these effects were also blocked by 8-cyclopentyl-1,3-dipropylxanthine. Collectively, our findings suggest that adenosine inhibits the excitability of PVN pre-sympathetic neurons through A1 receptor-mediated opening of KATP channels.

Abbreviations used:
aCSF

artificial CSF

DPCPX

8-cyclopentyl-1,3-dipropylxanthine

GIRK

G protein-coupled inwardly rectifying K+

KATP

ATP-dependent K+ channels

mEPSCs

miniature EPSCs

mIPSCs

miniature IPSCs

PBS

phosphate-buffered saline

PVN

paraventricular nucleus

SCH 58621

2-(2-furanyl)-7-(2-phenylethyl)-7H-pyrazolo[4,3-e][1,2, 4]triazolo[1,5-c]pyrimidin-5-amine

sEPSCs

spontaneous excitatory post-synaptic currents

sIPSCs

spontaneous inhibitory post-synaptic currents

TBS

tris-buffered saline

TTX

tetrodotoxin

ZM-241385

4-(2-[7-amino-2-(2-furyl)[1,2,4]triazolo[2,3-a][1,3,5]triazin-5-ylamino]ethyl) phenol

Neurons in the paraventricular nucleus (PVN) of the hypothalamus are critically involved in the regulation of neuroendocrine, cardiovascular and other physiological functions (Swanson and Sawchenko 1983). The PVN is a heterogenous structure and contains interneurons and output neurons, which project to the sympathetic vasomotor neurons in the brainstem and sympathetic pre-ganglionic neurons in the intermediolateral cell column of the spinal cord (Ranson et al. 1998; Pyner and Coote 2000; Hardy 2001). Thus, stimulation of PVN neurons can directly and indirectly influence sympathetic outflow and blood pressure, especially during stress and certain types of hypertension (Ranson et al. 1998; de Wardener 2001; Allen 2002; Miyakubo et al. 2002).

Adenosine is released by neurons or glial cells, especially during hypoxia and ischemia (Mitchell et al. 1993; Latini et al. 1995). It is also generated by enzymatic cleavage of adenosine triphosphate (ATP) in the extracellular space (White and Hoehn 1991). Basal extracellular adenosine concentrations range from 40–460 nM (Ballarin et al. 1991; Mitchell et al. 1993; Latini and Pedata 2001) with possible higher local concentrations. Four adenosine receptor subtypes (A1, A2A, A2B, and A3) have been found and all are linked to G proteins (Fredholm et al. 2001). A1 adenosine receptors are coupled to Gi/o-proteins and widely distributed in the brain nuclei including the PVN (Fastbom et al. 1987; Fredholm et al. 2001). Because the PVN is involved in the control of vasomotor tone and intracerebroventricular injection of A1 receptor agonists decreases blood pressure, heart rate, and the plasma catecholamine level in anesthetized and conscious rats (Laborit et al. 1990; Stella et al. 1993), we determined the central effect of adenosine on spinally projecting PVN neurons. Activation of A1 receptors produces hyperpolarization in neurons (Ponzio and Hatton 2005) and inhibits release of neurotransmitters such as acetylcholine (Ginsborg and Hirst 1972), glutamate (Corradetti et al. 1984; Bagley et al. 1999), GABA (Chamberlin et al. 2003; Jeong et al. 2003), and serotonin (Feuerstein et al. 1985, 1988) in different brain regions. Furthermore, previous studies have shown that activation of A1 adenosine receptors can open ATP-dependent K+ (KATP) channels in myocytes and that this effect is mediated by pertussis toxin-sensitive Gi/o protein (Kirsch et al. 1990; Dart and Standen 1993). A1 receptor-mediated openning of KATP channels are also involved in the cardiac and cerebral ischemic pre-conditioning (Gross and Auchampach 1992; Heurteaux et al. 1995; Dunwiddie and Masino 2001). However, the signaling mechanisms involved in the central depressor effect of adenosine remain unclear.

KATP channels are widely expressed in the cytoplasmic membranes of neurons and play an important role in the regulation of cell excitability (Ashcroft and Gribble 1998), by virtue of their involvement in hyperpolarizing neurons. These channels open when the intracellular ATP to ADP ratio decreases and close when this ratio increases. Previous studies have shown that adenosine activates cardiac membrane KATP by reducing ATP blockade (Light et al. 1996; Hu et al. 1999). Activation of KATP channels by adenosine is mediated by A1 receptors in the hippocampus (Shan and Cheng 2000), brainstem (Mironov et al. 1999), and substantia nigra (Andoh et al. 2006). In this study, by using a combination of in vivo retrogradely labeling and in vitro whole-cell recording, we sought to examine the signaling mechanisms underlying the inhibitory effect of adenosine on the excitability of retrogradely identified spinally projecting PVN neurons. This study provides new evidence that the A1 receptors and post-synaptic KATP channels mediate primarily the inhibitory effect of adenosine on the excitability of PVN pre-sympathetic neurons.

Materials and methods

Retrograde labeling of spinally projecting PVN neurons

Experiments were carried out by using male Sprague–Dawley rats (6–8 weeks of age; Harlan Laboratories, Inc., Indianapolis, IN, USA). The surgical procedures and experimental protocols were approved by the Animal Care and Use Committee of the University of Texas M. D. Anderson Cancer Center and conformed to the National Institutes of Health guidelines on the ethical use of animals. Briefly, rats were anesthetized by 2–3% isoflurane in O2, and the spinal cord was exposed from the T1 to T4 vertebrae through dorsal laminectomy. A rhodamine-labeled fluorescent microsphere suspension (FluoSpheres, 0.04 μm; Molecular Probes, Eugene, OR, USA) was injected bilaterally into the intermediolateral cell column region of the spinal cord (500 μm from the midline and 500 μm below the surface of the spinal cord) in 3 or 4 separate injections (50 nL each) through a glass pipette. The microinjection of FluSpheres was done by using a microinjector (Nanojector II; Drummond Scientific Company, Broomall, PA, USA) and was monitored by a surgical microscope. After injection, the rats were recovered for 3–5 days to allow FluoSpheres being transported to the PVN (Yang et al. 2007; Li et al. 2008a,b). After injection, rats were treated prophylactically with an antibiotic (enrofloxacin 5 mg/kg, subcutaneously daily for 3 days) and an analgesic (buprenorphine 0.2–0.5 mg/kg, subcutaneously every 12 h for 2 days). The electrophysiological properties of the labeled neurons were not influenced by the rhodamine-labeled microspheres (Tseng et al. 1991).

Preparation of hypothalamic slices

Brain slices containing the PVN were prepared from the FluoSphere-injected rats, as described previously (Li et al. 2008a,b). Briefly, the rat was anesthetized with 2% isoflurane and quickly decapitated and the brain was quickly removed and placed in ice-cold artificial CSF (aCSF, saturated by a mixture of 95% O2 and 5% CO2). The aCSF solution contained (in mM) 124.0 NaCl, 3.0 KCl, 1.3 MgSO4, 2.4 CaCl2, 1.4 NaH2PO4, 10.0 glucose, and 26.0 NaHCO3. A tissue block containing the PVN was glued onto the stage of a vibrating microtome (Technical Products International, St Louis, MO, USA). Coronal slices (300 μm thick) were cut as described previously (Li et al. 2008a,b). The slices were then transferred to an incubation chamber containing aCSF continuously gassed with a mixture of 95% O2 and 5% CO2 at 34°C for at least 1 h before electrophysiological experiments.

Electrophysiological recordings

Whole-cell recordings were performed in labeled PVN neurons in the hypothalamic slices. A slice was placed in a recording chamber (Warner Instruments, Hamden, CT, USA) and was held to the bottom of the chamber by a nylon mesh attached to a U-shaped stainless steel weight. The recording chamber was continuously perfused (3 mL/min) with aCSF (saturated by 95% O2 and 5% CO2) at 34°C maintained by an in-line solution heater. The volume of the solution needed to fill the recording chamber was approximately 1.0 mL. It took approximately 1.5 min to completely exchange the solution inside the recording chamber. The labeled PVN neurons were identified by using an upright microscope (BX51WI; Olympus, Tokyo, Japan) with a combination of epifluorescence illumination and differential interference contrast optics. Each identified neuron was then visualized under differential interference contrast without further exposure to fluorescence stimulation.

The recording electrode was pulled from borosilicate capillaries (1.2 mm outer diameter, 0.68 mm inner diameter; World Precision Instruments, Sarasota, FL, USA) by using a micropipette puller (P-97; Sutter Instruments, Novato, CA, USA). The resistance of the pipette was 3–7 MΩ when it was filled with internal solution containing (in mM) 140.0 K+ gluconate, 2.0 MgCl2, 0.1 CaCl2, 10.0 HEPES, 1.1 EGTA, 0.3 Na2-GTP, and 2.0 Na2-ATP, adjusted to pH 7.25 with 1 M KOH (270–290 mosM). Signals were processed using a Multiclamp 700B amplifier (Molecular Devices, Foster City, CA, USA), filtered at 1–2 kHz and digitized at 20 kHz using Digidata 1440 (Molecular Devices). The spontaneous excitatory post-synaptic currents (sEPSCs) were recorded at a holding potential of −70 mV in the presence of a GABAA antagonist bicuculline (20 μM). The spontaneous inhibitory post-synaptic currents (sIPSCs) were recorded at a holding potential of 0 mV in the presence of an non-NMDA antagonist 6-cyano-7-nitroquinoxaline-2,3-dione (20 μM) (Li et al. 2008a). The miniature IPSCs (mIPSCs) or miniature EPSCs (mEPSCs) were recorded in the presence of 1 μM tetrodotoxin (TTX). In some experiments, a general G-protein inhibitor, GDP-β-S (1 mM), was added into the pipette solution to block the post-synaptic G-protein signaling. The spontaneous firing activity of labeled PVN neurons was recorded using the whole-cell current-clamp technique.

All the drugs were freshly prepared in aCSF before the experiments and delivered by using syringe pumps at final concentrations. We determined the effect of adenosine (1–100 μM), 8-cyclopentyl-1,3-dipropylxanthine (DPCPX, 1 μM), 4-(2-[7-amino-2-(2-furyl)[1,2,4]triazolo[2,3-a][1,3,5]triazin-5-ylamino]ethyl)phenol (ZM-241385, 1 μM), 2-(2-furanyl)-7-(2-phenylethyl)-7H-pyrazolo[4,3-e][1,2,4] triazolo[1,5-c]pyrimidin-5-amine (SCH 58621, 1 μM), glibenclamide (100 μM) and tolbutamide (200 μM) on firing activity in labeled PVN neurons. These drugs were purchased from Sigma (St Louis, MO, USA). TTX (1 μM) was purchased from Ascent Scientific Ltd. (Bristol, UK).

Immunocytochemical labeling of A1 receptors in spinally projecting neurons

Using specimens from three rats, we performed immunocytochemical labeling to determine if adenosine A1 receptors were expressed on spinally projecting PVN neurons. Under deep anesthesia induced by sodium pentobarbital (60 mg/kg, by intraperitoneal injection), rats were intracardially perfused with 200 mL of ice-cold normal saline containing 1000 units of heparin followed by 250 mL of fixative solution 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS; pH 7.4) and 200 mL of 10% sucrose in 0.1 M PBS (pH 7.4). The brain tissue containing the PVN was quickly removed and post-fixed for 2 h in the same fixative solution, and then cryoprotected in 30% sucrose in PBS for 48 h at 4°C. The tissue was then sectioned into 25-μm-thick slices in the coronal plane, with a freezing microtome; the sections were collected and allowed to float freely in 0.1 M phosphate buffer. Sections were then incubated with the primary antibody (rabbit anti-A1 adenosine receptor polyclonal antibody, dilution 1 : 100, Sigma) diluted in tris-buffered saline (TBS) containing 1% normal goat serum for 2 h at 22–25°C and overnight at 4°C. Subsequently, sections were rinsed in TBS and incubated with the secondary antibody (goat anti-rabbit immunoglobulin G conjugated to Alexa fluor 488 dilution 1 : 400; Molecular Probes) diluted in TBS containing 2% normal goat serum for 1.5 hr at 22–25°C. Then sections were rinsed in TBS for 40 min and mounted on slides, dried, and coverslipped. The negative control was created by replacing the primary antibody with non-immune serum from the same species. The sections were examined on a confocal microscope (Carl Zeiss, Jena, Germany) and the areas of interest were photo documented. In the higher-magnification images, the overlapping of the red and green color (yellow) indicated a co-localization of A1 receptor immunoreactivity and retrograde tracer in the PVN neurons, because the optical section thickness of a confocal image is thin enough to minimize the possibility of superimposition of stained neurons.

Data analysis

Data are presented as mean ± SEM. The discharge and membrane potentials were analyzed over a period of 3–5 min before, during, and after drug application. The junction potential was corrected off-line based on the composition of the internal and external solution used for the recordings. The firing rate, amplitude, and frequency of sEPSCs, sIPSCs, mEPSCs, and mIPSCs were analyzed off-line using a peak detection program (MiniAnalysis; Synaptosoft, Leonia, NJ, USA). Events were detected by setting a threshold above the noise level. The effect of drugs on the firing rate, adenosine-induced current, amplitude, and the frequency of sEPSCs and sIPSCs was analyzed by anova with the Dunn’s post hoc test. A p-value of < 0.05 was considered statistically significant.

Results

Whole-cell recordings were performed on 139 PVN cells (= 37 rats) labeled by FluoSpheres. The spinal cords were taken out and sectioned at 35 μm in thickness at the level of injection immediately after the rat was euthanized. The injection sites of FluoSpheres were identified under a microscope equipped with epifluorescence illumination. Data were excluded from analysis if the injection site was not located in the intermediolateral cell column of the spinal cord (Li et al. 2004).

Effect of adenosine on spontaneous firing activity of labeled PVN neurons

Spontaneous firing activity was recorded from the labeled neurons with resting membrane potentials of −50 mV or lower and with action potential overshoot greater than 10 mV. The majority of the labeled PVN neurons displayed spontaneous discharges (26 of 32 neurons, 81.1%), with an average firing rate of 1.3 ± 0.19 Hz. Bath application of 10, 50, and 100 μM adenosine decreased the firing activity in a concentration-dependent manner in 12 of 26 labeled PVN neurons (Fig. 1a and b). Adenosine (50 μM) decreased the firing activity from 0.84 ± 0.16 to 0.25 ± 0.6 Hz, while the adenosine-induced maximal reduction in the firing rate appeared at 100 μM (from 0.84 ± 0.16 to 0.2 ± 0.05 Hz). Also, the membrane potential was hyperpolarized by adenosine in these 12 cells (Fig. 1c). The firing rate of the remaining 14 neurons was not significantly different at 100 μM adenosine compared with the control value (1.96 ± 0.3 Hz versus 2.03 ± 0.29 Hz), nor was the membrane potential (−59.0 ± 0.9 mV versus −58.9 ± 1.0 mV).

Figure 1.

 Effects of adenosine on the firing activity and membrane potential in labeled PVN neurons. (a) Representative tracings show the spontaneous firing activity of a labeled PVN neuron during control, application of different doses of adenosine (10, 50, and 100 μM) and washout. (b, c) Summary data show the effects of adenosine (10, 50, and 100 μM) on the firing activity (b) and membrane potential (c) of labeled PVN neurons. Data are presented as mean ± SEM. *< 0.05, compared with control. Ado, adenosine.

Role of A1 receptors and G proteins in the inhibitory effect of adenosine

To determine the receptor subtypes that mediate the effect of adenosine on the firing activity, we used DPCPX (Sebastiao et al. 1990; Oliveira et al. 1991), a specific antagonist against A1 receptors, and ZM-241385 (Cunha et al. 1997), a selective antagonist against A2A receptors (Zocchi et al. 1996; Rebola et al. 2008). In a separate group of labeled PVN neurons, bath application of 100 μM adenosine decreased the firing activity from 0.99 ± 0.2 to 0.2 ± 0.1 Hz (= 8, < 0.05). Then, 1 μM DPCPX was added to the recording chamber. Repeated application of 100 μM adenosine failed to decrease the firing activity in the presence of 1 μM DPCPX (Fig. 2a–c). We also determined the effect of low concentrations (10 and 100 nM) of DPCPX on the adenosine-induced decrease in the firing activity in another eight PVN neurons. Low concentrations of DPCPX only slightly attenuated the effect of adenosine on the firing activity in these neurons (Fig. 2d). In contrast, adenosine still decreased the firing activity after blockade of A2A receptors with 1 μM ZM-241385 (= 7, Fig. 2e). In addition, 1 μM SCH58621, another selective A2A antagonist (Zocchi et al. 1996; Rebola et al. 2008), failed to block adenosine-induced decreases in the firing activity of PVN neurons (= 6, Fig. 2f). DPCPX, ZM-241385 or SCH58621 alone had no significant effect on the basal firing activity of labeled PVN neurons.

Figure 2.

 Blockade of A1 receptors eliminated the effect of adenosine on the firing activity of labeled PVN neurons. Frequency histogram (a) and raw traces (b) show the effect of 100 μM adenosine on the firing activity of a labeled PVN neuron in the absence and presence of 1 μM 8-cyclopentyl-1,3-dipropylxanthine (DPCPX). (c, d) Summary data show the effect of adenosine on the firing rate of labeled PVN neurons before and after blockade of A1 receptors with different concentrations of DPCPX. (e, f) Summary data show the effect of adenosine on the firing rate of labeled PVN neurons before and after blockade of A2A receptors with 1 μM ZM-241385 (e) or 1 μM SCH58621 (f). (g) Adenosine (100 μM) had no effect on the firing rate of labeled PVN neurons when GDP-β-S (1 mM) was included in the Internal solution. Data are presented as mean ± SEM. *< 0.05 compared with control. Ado, adenosine; ZM, ZM-241385.

We next determined if adenosine inhibits neuronal activity through a post-synaptic action. Because adenosine receptors are coupled to G proteins, we included 1 mM GDP-β-S in the internal recording solution to inhibit G-protein signaling. GDP-β-S is a non-hydrolyzable analog of GDP and competes with GTP for the binding site at the Gα subunit to inactivate GTPase at a concentration higher than the intracellular GTP level (about 25 μM) (Gilman 1987; Breitwieser and Szabo 1988; Ross 1989; Hepler and Gilman 1992). GDP-β-S also inhibits intracellular GTP synthesis from GDP (Breitwieser and Szabo 1988). Adenosine (100 μM) was administered to the recording chamber 15 min after whole-cell configuration was established. In all nine labeled PVN neurons tested, adenosine did not significantly change the firing activity (1.47 ± 0.25 Hz versus 1.48 ± 0.22 Hz, > 0.05, Fig. 2g).

Role of KATP channels in the effect of adenosine on labeled PVN neurons

Previous studies have shown that activation of A1 receptors opens KATP channels (Kirsch et al. 1990; Dart and Standen 1993). Thus, we determined the role of KATP channels in adenosine-induced decreases in the firing activity of labeled PVN neurons. The effect of adenosine on the firing activity was tested before and after bath application of the KATP channel blocker glibenclamide in 10 labeled PVN neurons. Initial application of adenosine (100 μM) decreased the firing activity from 0.90 ± 0.3 to 0.24 ± 0.16 Hz (< 0.05) in these 10 neurons. However, adenosine failed to change the firing activity in the presence of 100 μM glibenclamide (1.10 ± 0.16 Hz versus 1.17 ± 0.13 Hz, > 0.05, Fig. 3a–c). In addition, bath application of another KATP channel blocker, tolbutamide (200 μM) (Zhang et al. 2008) eliminated adenosine-induced decreases in the firing activity in another nine neurons (Fig. 3d). Application of glibenclamide (100 μM) or tolbutamide (200 μM) alone had no significant effect on the firing activity of labeled PVN neurons.

Figure 3.

 Role of KATP channels in the effect of adenosine on the firing activity of labeled PVN neurons. (a, b) Frequency histogram (a) and raw traces (b) show the effect of 100 μM adenosine on the firing activity of a labeled PVN neuron before and after application of the KATP channel blocker glibenclamide (100 μM). (c, d), Summary data showing the effect of adenosine on the firing rate of labeled PVN neurons in the absence and presence of the KATP channel blocker glibenclamide (c) and tolbutamide (200 μM) (d). (e) Summary data show the effect of 10 μM baclofen on the firing activity of PVN neurons before and after blockade of KATP channel with 100 μM glibenclamide. Data are presented as mean ± SEM. *< 0.05 compared with control. Ado, adenosine; GBC, glibenclamide; TBM, tolbutamide.

To determine if KATP channels are also involved in the inhibitory effect of other G-protein coupled receptor agonists, we tested the effect of the GABAB agonist baclofen on the firing activity of labeled PVN neurons before and after blocking KATP channels. Initial application of baclofen (10 μM) decreased the firing activity from 1.13 ± 0.3 to 0.5 ± 0.1 Hz (< 0.05) in six of eight neurons. The firing activity returned to the baseline level 15–20 min after washout of baclofen. Baclofen still caused a similar decrease in the firing activity in the presence of 100 μM glibenclamide in these six neurons (from 1.1 ± 0.2 to 0.46 ± 0.2 Hz; > 0.05, Fig. 3e).

We further determined if KATP channels are involved in adenosine-induced currents in labeled PVN neurons. Membrane currents were recorded at a holding potential of −60 mV in the voltage-clamp mode. Bath application of adenosine (100 μM) induced an outward current with peak amplitude of 20.1 ± 5.3 pA in six PVN neurons (Fig. 4). The adenosine-induced currents were largely reduced by exposing these neurons to the KATP channel blocker glibenclamide (100 μM) (Fig. 4a and b). However, residual currents (6.4 ± 2.4 pA) were difficult to distinguish from the background. To determine if A1 receptors mediated adenosine-induced current, we used selective A1 receptor antagonist DPCPX. Adenosine-induced currents were only slightly reduced by low concentration of DPCPX (10 and 100 nM), but completely abolished by 1 μM DPCPX (Fig. 4c). In addition, to determine the adenosine-induced currents were G protein-coupled inwardly rectifying K+ (GIRK) currents, we used tertiapin-Q, a specific GIRK channel blocker (Jin and Lu 1998; Nakatsuka et al. 2008). Bath application of 1 μM tertiapin-Q had no significant effect on adenosine-induced currents in seven additional labeled PVN neurons (Fig. 4d).

Figure 4.

 Adenosine-induced outward currents in labeled PVN neurons. (a) Representative traces show currents elicited by application of 100 μM adenosine to a labeled PVN neuron before and after blockade of KATP channels with 100 μM glibenclamide. (b–d) Summary data show adenosine-induced currents elicited by application of 100 μM adenosine in labeled PVN neurons in the presence of the KATP channel blocker glibenclamide (b), different concentrations of A1 receptor antagonist 8-cyclopentyl-1,3-dipropylxanthine (DPCPX) (c), or the GIRK channel blocker 1 μM tertiapin-Q (d). Data are presented as mean ± SEM. *< 0.05 compared with adenosine alone. Ado, adenosine; GBC, glibenclamide; TRQ, tertiapin-Q.

To further delineate the ionic component involved in adenosine-induced currents, we performed voltage-clamp recordings under a ramp command pulse from −120 to −40 mV (ramp rate 80 mV/s). Membrane currents were obtained in response to the ramp command pulse before and during application of 100 μM adenosine. The current–voltage (I–V) relationships of the adenosine-induced currents were obtained by subtracting the current measured during the control experiments from the currents measured during adenosine application (Fig. 5). The reversal potential of adenosine-induced currents was determined as the intercept of the subtracted currents with the abscissa (0 pA line). The reversal potential of adenosine-induced currents was −95.5 ± 3.6 mV (= 9, after correction for a junction potential of 15.3 mV in our preparation). This reversal potential was close to the calculated K+ reversal potential of −97.7 mV (Zhang et al. 2008). Application of 100 μM glibenclamide completely abolished adenosine-induced currents (Fig. 5). To determine if G proteins were involved in the adenosine-induced currents, we included GDP-β-S (1 mM) in the internal recording solution. Adenosine induced a smaller current when we used GDP-β-S-containing solution than when we used the internal solution without GDP-β-s (= 6, Fig. 5).

Figure 5.

 Voltage-current relationships of adenosine-induced currents in labeled PVN neurons. (a, b) Whole-cell membrane currents in response to ramping holding potentials from −120 to −40 mV (ramp 80 mV/s) before and after application of adenosine and adenosine plus glibenclamide (100 μM). Adenosine-induced currents were obtained by subtracting the currents measured during control from the current in the presence of adenosine plus glibencamide (b). Recordings were corrected by the junction potentials. The equilibrium potential of adenosine-induced currents was −96.6 mV. (c, d), Whole-cell currents recorded with Inclusion of 1 mM GDP-β-S in the internal solution in response to ramping holding potentials from −120 to −40 mV (ramp 80 mV/s) during application of adenosine and adenosine plus glibenclamide (c). Adenosine-induced currents were obtained by subtracting the currents measured during control from the current in the presence of adenosine plus glibencamide (d).

Localization of adenosine A1 receptors on labeled PVN neurons

The electrophysiological data strongly suggest that post-synaptic adenosine A1 receptors are expressed on spinally projecting PVN neurons. To further determine the spatial relationship between the adenosine A1 receptors and spinally projecting PVN neurons, we immunolabeled retrogradely labeled PVN neurons with a specific adenosine A1 antibody. All negative controls displayed no detectable staining. Most of the DiI-labeled cells (red) were positive for A1 receptor immunoreactivity in the PVN (Fig. 6). In the high magnification images, the color change (yellow) indicated that A1 receptors were expressed on DiI-labeled PVN neurons. In addition to expressing on the soma, puncta immunoreactive for A1 receptors were also noted in the PVN.

Figure 6.

 Confocal images show the presence of A1 receptors on PVN-spinal neurons. Low- and high-magnification confocal images show colocalization of A1 receptors (green) and DiI-labeled neurons (red) in the PVN. Images are in all cases single confocal optical sections. 3V, third ventricle.

Effects of adenosine on glutamatergic EPSCs and GABAergic IPSCs in labeled PVN neurons

To assess the effect of adenosine on glutamatergic and GABAergic synaptic inputs to labeled PVN neurons, we tested the effect of adenosine on glutamatergic sEPSCs and GABAergic sIPSCs after blockade of post-synaptic action of adenosine by including GDP-β-S (1 mM) in the internal pipette solution. To determine if the effects of adenosine on glutamate and GABA release are through synaptic terminal, we used TTX, a voltage-dependent Na+ channel blocker, to isolate the synaptic terminals by blocking impulse from soma of pre-synaptic neurons. Bath application of 100 μM adenosine significantly reduced the frequency from 3.4 ± 1.1 to 2.1 ± 0.8 Hz (= 8, < 0.05), but not the amplitude (20.9 ± 3.7 versus 20.7 ± 2.7 pA, > 0.05), of sEPSCs in these neurons (Fig. 7a–e). The sEPSCs were completely blocked by 20 μM 6-cyano-7-nitroquinoxaline-2,3-dione, a competitive AMPA/kainate receptor antagonist (Fig. 7a). Bath application of A1 receptor antagonist DPCPX (1 μM) eliminated the effect of adenosine on sEPSCs. In addition, adenosine decreased the frequency of mEPSCs from 3.1 ± 1.6 to 1.9 ± 0.5 Hz (= 7, < 0.05) without affecting the amplitude (21.5 ± 2.1 versus 20.7 ± 1.4 pA, > 0.05) of mEPSCs in the presence of 1 μM TTX (Fig. 7f and g).

Figure 7.

 Effect of adenosine on glutamatergic EPSCs in labeled PVN neurons. (a) Raw traces show sEPSCs during control and during bath application of 100 μM adenosine, adenosine plus 1 μM 8-cyclopentyl-1,3-dipropylxanthine (DPCPX), and 20 μM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) in a labeled PVN neuron. Cumulative plot analysis of sEPSCs of the same neuron in A showing the distribution of the interevent interval (b) and amplitude (c) during control, application of adenosine, and adenosine plus DPCPX. Summary data show the frequency (d) and amplitude (e) of sEPSCs during control and during application of adenosine and adenosine plus DPCPX. Summary data show the frequency (f) and amplitude (g) of mEPSCs in the presence of 1 μM TTX during control and application of adenosine and adenosine plus DPCPX.*< 0.05, compared with control. Ado, adenosine.

In a separate group of labeled PVN neurons, application of 100 μM adenosine significantly decreased the frequency of sIPSCs from 3.7 ± 1.0 to 2.2 ± 0.4 Hz (= 9, < 0.05) without changing the amplitude (41.9 ± 7.0 pA versus 42.9 ± 7.9 pA, p > 0.05) (Fig. 8d and e). Also, application of 100 μM of adenosine decreased the frequency of mIPSCs from 3.5 ± 0.9 to 2.2 ± 0.5 Hz (= 7, < 0.05) but failed to change the mIPSC amplitude (35.6 ± 5.1 pA vs. 34.9 ± 5.6 pA, p > 0.05) in these neurons (Fig. 8f and g). The sIPSCs and mIPSCs were completely blocked by 20 μM bicuculline, a GABAA receptor blocker (Fig. 8a). The A1 receptor antagonist DPCPX (1 μM) abolished the effect of adenosine on sIPSCs and mIPSCs (Fig. 8d and g).

Figure 8.

 Effect of adenosine on GABAeric IPSCs in labeled PVN neurons. (a) Representative recordings show sIPSCs during control and during bath application of 100 μM adenosine, adenosine plus 1 μM 8-cyclopentyl-1,3-dipropylxanthine (DPCPX), and 20 μM bicuculline in a labeled PVN neuron. (b, c) Cumulative plot analysis of sIPSCs of the same neuron in (a) showing the distribution of the interevent interval (b) and amplitude (c) during control, application of adenosine, and adenosine plus DPCPX. Summary data show the frequency (d) and amplitude (e) of sIPSCs during control and application of adenosine and adenosine plus DPCPX. Summary data show the frequency (f) and amplitude (g) of mIPSCs in the presence of 1 μM TTX during control and application of adenosine and adenosine plus DPCPX. *< 0.05, compared with control. Ado, adenosine; BIC, bicuculline.

Discussion

In this study, we determined cellular and ionic mechanisms involved in the inhibitory effect of adenosine on spinally projecting PVN neurons. We found that adenosine acted in a dose-dependent manner to decrease the firing activity and hyperpolarize the majority of spinally projecting PVN neurons. These effects were blocked by the A1 receptor antagonist DPCPX, post-synaptic dialysis of GDP-β-S, or KATP channel blockade. Furthermore, KATP channel blockers or DPCPX abolished adenosine-induced currents, and the reversal potential of adenosine-induced currents was close to the equilibrium potential for K+. In addition, adenosine decreased the frequency of glutamatergic EPSCs and GABAergic IPSCs in labeled PVN neurons. Immunocytochemical labeling also revealed that adenosine A1 receptors were expressed on spinally projecting PVN neurons. Therefore, our data provide new functional evidence adenosine reduces the excitability of PVN pre-sympathetic neurons through the A1 receptor-mediated opening of KATP channels.

An interesting finding of this study is that adenosine decreased the firing activity in a subpopulation of spinally projecting PVN neurons. The reason for the differential effect of adenosine on the PVN neurons remains unclear. It is likely that adenosine-responsive neurons express more adenosine A1 receptors. This notion is supported by the immunocytochemical staining that showed some retrogradely labeled spinally projecting PVN neurons expressed A1 receptors more densely than others. Although blockade of adenosine receptors had no effect on the basal firing activity of PVN neurons, we cannot rule out the possibility that the tonic endogenous activation of A1 receptors controls the firing activity of the PVN neurons in vivo because some neuronal connections could not be preserved in the thin hypothalamic slices in vitro. We found that the basal firing activity was lower in adenosine-sensitive neurons than in adenosine-insensitive neurons. It is unlikely that the adenosine-sensitive neurons were tonically influenced by endogenously generated adenosine in our preparation because blocking A1 receptors had no significant effect on the basal firing activity. The selectivity of DPCPX to A1 receptor at low concentrations has been determined in radioligand binding assay (Lohse et al. 1987; Fredholm et al. 2001). However, higher concentrations were used in the brain slice studies (Bagley et al. 1999; Ponzio and Hatton 2005). Because the antagonist concentration in the brain slices depends on the concentration of the agonist used, we found that 1 μM DPCPX was required to abolish the effect of 100 μM adenosine on the firing activity in the PVN neurons. Previous studies have shown that adenosine A2A receptors and its mRNA are expressed in the hypothalamus (Dixon et al. 1996; Ponzio and Hatton 2005; Ponzio et al. 2006). However, we found that A2A receptors were not involved in the effect of adenosine on the spinally projecting PVN neurons, because the highly selective A2A antagonist SCH58621 and ZM-241385 failed to block the adenosine-induced inhibition of PVN neurons. Although DPCPX is an A1 adenosine receptor antagonist at low concentrations, it can interacts with A2B adenosine receptors at higher concentrations (Fredholm et al. 2001). However, in situ hybridization revealed no A2B receptor mRNA in the PVN (Stehle et al. 1992). Therefore, it is unlikely that A2B receptors were involved in the inhibitory effect of adenosine on the PVN neurons.

The most salient finding of this study is that the post-synaptic KATP channels are essential for the inhibitory effect of adenosine on the firing activity of PVN neurons. Adenosine A1 receptor is linked to the pertussis toxin-sensitive Gi/o G protein signals. Activation of A1 receptor inhibits adenylyl cyclase, inhibits Ca2+ channels, and activates GIRK channels (Dunwiddie and Fredholm 1989; Dunwiddie and Masino 2001). In the present study, we found that the GIRK channel blocker had no effect on adenosine-induced currents in the spinally projecting PVN neurons. However, inhibition of the post-synaptic G protein signaling by including GDP-β-S in the pipette solution eliminated the adenosine-induced inhibitory effects and adenosine-induced currents, suggesting that activation of the A1 receptor activates KATP channels through G proteins. In fact, we found that blockade of KATP channels abolished adenosine-induced decreases in firing activity and adenosine-induced outward currents in spinally projecting PVN neurons. These data suggest that activation of KATP channels is critical for the post-synaptic inhibitory effect of adenosine on spinally projecting PVN neurons. We also determined that the reversal potential of the adenosine-induced current was close to the equilibrium potential of K+, providing further evidence that the ionic carrier of adenosine-induced currents is K+. Our data are consistent with those from previous studies showing that activation of A1 receptors activates KATP channels in neurons in other brain regions (Mironov et al. 1999; Shan and Cheng 2000; Andoh et al. 2006). We found that adenosine-induced currents were abolished by A1 receptor antagonist. These data suggest that A1 receptors mediate the action of adenosine on KATP channels. Furthermore, we determined if KATP channels were involved in the inhibitory effect induced by the GABAB receptor agonist baclofen. We found that blocking KATP channels had no effect on baclofen-induced decreases in the firing activity in PVN neurons. These data suggest that A1 receptors are preferentially coupled to KATP channels compared to other G-protein coupled receptors. However, it is not clear how adenosine A1 receptors are coupled to KATP channels in the PVN neurons. Previous studies have shown that KATP channels can be linked to G proteins. For instance, KATP channels are coupled to A1 receptors via G proteins and protein kinase C in cardiac myocytes (Kirsch et al. 1990; Hu et al. 1999). Conversely, stimulation of A1 adenosine receptors activates KATP channels through cyclic AMP signaling in inspiratory neurons in the brainstem (Mironov et al. 1999). Therefore, further studies are needed to establish the signaling pathways involved in activation of KATP channels by A1 adenosine receptors in PVN neurons.

The immunocytochemical labeling revealed that, in addition to being expressed on the soma of spinally projecting PVN neurons, A1 receptors are also possibly expressed on the nerve terminals (indicated by puncta A1 receptor-immunoreactivities). These data suggest that pre-synaptic A1 receptors regulate neurotransmitter release in the PVN. In this regard, adenosine inhibits glutamatergic, but not inhibitory, synaptic inputs in the hippocampus (Yoon and Rothman 1991). Also, adenosine produces A1-mediated inhibition of both glutamateric and GABAergic synaptic inputs in the periaqueductal gray and supraoptic nucleus (Bagley et al. 1999; Oliet and Poulain 1999). It is possible that the different effects of adenosine on glutamatergic and GABAergic synaptic inputs is caused by the preferential distribution of A1 adenosine receptors on GABAergic and glutamateric terminals in different nuclei (Thompson et al. 1992). In this study, we found that adenosine decreased both inhibitory GABAergic and excitatory glutamatergic synaptic inputs to spinally projecting PVN neurons and that these effects were eliminated by the A1 receptor antagonist DPCPX. Furthermore, we found that TTX did not alter the inhibitory effect of adenosine on synaptic GABA and glutamate release, suggesting that adenosine A1 receptors are present at the pre-synaptic terminals. Although blocking A1 receptors with DPCPX eliminated adenosine-induced inhibition of glutamatergic and GABAergic synaptic inputs, DPCPX alone did not significantly alter the basal frequency and amplitude of sIPSCs and sEPSCs in labeled PVN neurons. Thus, it seems that pre-synaptic A1 receptors are not tonically active in regulating glutamate and GABA release in this slice preparation. The inhibitory effect of adenosine on the firing activity of PVN neurons may partly attribute to the reduction of glutamatergic synaptic inputs. On the other hand, the reduction in the GABAergic synaptic inputs may offset the adenosine-induced decreases in the firing activity. Because adenosine inhibited both excitatory and inhibitory synaptic inputs, these opposing pre-synaptic effects could have a small impact on the overall effect of adenosine on the firing activity. Our finding that inhibition of post-synaptic G proteins by intracellular dialysis of the G protein inhibitor GDP-β-S eliminated the effect of adenosine on the firing activity in the PVN neurons, however, suggests that adenosine inhibits firing activity mainly through a post-synaptic action (Fig. 9). The firing activity was higher in neurons recorded with GDP-β-S in the internal solution than that without GDP-β-S. It is likely that post-synaptic inhibitory signals coupled to G-protein coupled receptors are critical for regulating the firing activity of the PVN pre-sympathetic neurons. As blocking A1 receptors had no effect on the baseline firing activity, other inhibitory post-synaptic G-protein coupled receptors could be more important than A1 receptors in the physiological control of the excitability of PVN neurons. However, it is also possible that the higher firing rate in neurons recorded with GDP-β-S may be due to inclusion of adenosine-insensitive neurons with a higher firing rate.

Figure 9.

 Diagram showing the subcellular location of adenosine A1 receptors and the signaling mechanisms in the PVN.

Perspectives

The exact sources of adenosine released to the PVN pre-sympathetic neurons are not fully known. The major source of adenosine is dephosphorylation of 5′-AMP by 5′ nucleotidase in the cytoplasm and extracellular space (Latini et al. 1995). Most adenosine is probably transported into the extracellular space via bidirectional nucleoside transporters (Dunwiddie et al. 1997). A second source of adenosine is hydrolysis of S-adenosylhomocysteine by S-adenosylhomocysteine hydrolase (Latini et al. 1995). High levels of adenosine are also generated during tissue experiencing hypoxia and ischemia (Latini et al. 1995). Adenosine and adenosine receptor agonists could play a protective role in these conditions, mitigating the neurotoxicity by reducing glutamate release (Corradetti et al. 1984). More importantly, adenosine-induced opening of KATP is critically involved in the pre-conditioning protection during tissue ischemia (Heurteaux et al. 1995). Because blockade of A1 receptors had no effect on the basal firing activity of PVN neurons, it is possible that adenosine might be involved in the control of neuronal excitability in pathophysiological but not physiological conditions. However, because we used an in vitro brain slice preparation, it is difficult to assess the true contribution of endogenous adenosine levels to activation of A1 receptors in the PVN. In this study, we provide further evidence that, in addition to reducing glutamatergic synaptic inputs, adenosine initiates A1 receptor-mediated activation of post-synaptic KATP channels, which leads to hyperpolarization of PVN pre-sympathetic neurons. Thus, adenosine may be protective in reducing stress responses and excessive activation of the sympathetic nervous system during stroke.

Acknowledgements

This study was supported by the National Heart, Lung, and Blood Institute (grants HL60026 and HL77400) and the American Heart Association National Center (Scientist Development Grant 0635402N).

Ancillary