In mammals, ventilation is peripherally controlled by the carotid body (CB), which receives afferent innervation from the petrosal ganglion and efferent innervation from neurons located along the glossopharyngeal nerve (GPN). GPN neurons give rise to the “efferent inhibitory” pathway via a plexus of neuronal nitric oxide (NO) synthase-positive fibers, believed to be responsible for CB chemoreceptor inhibition via NO release. Although NO is elevated during natural CB stimulation by hypoxia, the underlying mechanisms are unclear. We hypothesized that ATP, released by rat CB chemoreceptors (type 1 cells) and/or red blood cells during hypoxia, may directly activate GPN neurons and contribute to NO-mediated inhibition. Using combined electrophysiological, molecular, and confocal immunofluorescence techniques, we detected the expression of multiple P2X receptors in GPN neurons. These receptors involve at least four different purinergic subunits: P2X2 [and the splice variant P2X2(b)], P2X3, P2X4, and P2X7. Using a novel coculture preparation of CB type I cell clusters and GPN neurons, we tested the role of P2X signaling on CB function. In cocultures, fast application of ATP, or its synthetic analog 2′,3′-O-(4 benzoylbenzoyl)-ATP, caused type I cell hyperpolarization that was prevented in the presence of the NO scavenger 2-(4-carboxyphenyl)-4,4,5,5-tetramethyl-imidazoline-1-oxyl-3-oxide potassium. These data suggest that ATP released during hypoxic stress from CB chemoreceptors (and/or red blood cells) will cause GPN neuron depolarization mediated by multiple P2X receptors. Activation of this pathway will lead to calcium influx and efferent inhibition of CB chemoreceptors via NO synthesis and consequent release.
The peripheral control of ventilation is mediated by chemoreceptors located in the carotid body (CB) which receives afferent innervation from the carotid sinus nerve (CSN), a branch of the glossopharyngeal nerve (Gonzalez et al., 1994). Thus, in response to a fall in blood partial pressure of oxygen (PO2) (hypoxia), there is a compensatory reflex hyperventilation initiated by CB chemoreceptor cells, which release neurotransmitters, including ATP, that excite the CSN (Gonzalez et al., 1994; Nurse, 2005). In addition to this afferent pathway, there is a less well understood “efferent inhibitory” pathway that involves a plexus of neuronal nitric oxide synthase (nNOS)-positive nerve fibers that modulate CB function via release of NO (Prabhakar et al., 1993; Wang et al., 1993, 1995a,b; Grimes et al., 1994; Prabhakar, 1999). The origin of these nNOS-positive fibers includes two groups of autonomic neurons or “microganglia,” one located near the branch point of the CSN and glossopharyngeal (GPN) nerve and the other located more distally along the GPN nerve (Wang et al., 1993, 1995a,b; Campanucci et al., 2003; Campanucci and Nurse, 2005). Although there is good evidence that the CB is a source of both nNOS and endothelial NO synthase isoforms and that NO has an inhibitory effect on CB function (Wang et al., 1995a,b; Kline et al., 1998; Prabhakar, 1999), the physiological role of each isoform during hypoxic chemoexcitation is not well understood. We are particularly interested in the nNOS isoform because it is expressed in neurons that contribute to the efferent inhibitory pathway to the CB. We reported recently (Campanucci et al., 2003; Campanucci and Nurse, 2005) that these nNOS-positive autonomic GPN neurons are hypoxia sensitive, and it is known that NO is released in the CB during hypoxia (Fung et al., 2001). However, the mechanisms leading to NOS activation, and consequent NO release, are unclear and confounded by the requirement of molecular oxygen for enzyme activity (Prabhakar, 1999).
In the present study, we considered ATP stimulation as a mechanism of activation of the CB efferent inhibitory pathway. ATP, released from CB chemoreceptors as a key excitatory neurotransmitter (Zhang et al., 2000; Rong et al., 2003; Buttigieg and Nurse, 2004; Nurse, 2005) and/or from circulatory red blood cells (Ellsworth, 2000) during hypoxia, may play a central role in the depolarization of GPN neurons, leading to NO synthesis and release. Autonomic neurons are known to express a variety of excitatory ionotropic purinergic P2X receptors, which act as ATP-gated nonselective cation channels with significant permeability to calcium (Ralevic and Burnstock, 1998; Dunn et al., 2001; North, 2002), whose cytosolic levels must reach the 400–1000 nm range to facilitate NOS activation (Vincent, 1994; Fung et al., 2001). Here we show that GPN neurons expressed at least four different types of P2X receptors (P2X2, P2X3, P2X4, and P2X7). Moreover, using a novel GPN–CB coculture system, we demonstrate that activation of P2X receptors in GPN neurons leads to NO-dependent inhibition of CB chemoreceptors (type I cells) after exposure to either ATP or hypoxia.
Materials and Methods
Wistar rat pups [9–12 d old; Harlan (Madison, WI) or Charles River (Québec, Québec, Canada)] were first stunned by a blow to the head that rendered them unconscious and then killed by decapitation. The carotid bifurcation, surrounding ganglia, and attached nerves were removed under sterile conditions. All procedures for animal handling were performed according to the guidelines of the Canadian Council on Animal Care. A section of the GPN, extending from its intersection with the CSN to a region ∼5 mm distal to the intersection, was dissected bilaterally from each pup. Each GPN was then cut into two segments so as to separate the two populations (proximal and distal) of GPN neurons. The two segments were separately pooled, digested with enzyme [0.1% collagenase (Invitrogen, Burlington, Ontario, Canada), 0.1% trypsin, Sigma (St. Louis, MO), and 0.01% DNase (Invitrogen)] and mechanically dissociated to produce dispersed GPN neurons. The cell suspension was plated onto a thin layer of Matrigel (Collaborative Research, Bedford, MA) that was previously applied to the central wells of modified 35 mm tissue culture dishes. Cultures were grown at 37°C in a humidified atmosphere of 95% air–5% CO2, and used for either patch-clamp experiments or immunofluorescence studies ∼24–48 h after isolation. In some experiments, GPN neurons were cocultured with rat CB type I cells. To produce cocultures, an overlay of dissociated rat GPN neurons was added to a preexisting monolayer of CB type I cells prepared as described previously (Zhong et al., 1997; Nurse and Zhang, 1999; Zhang et al., 2000). Electrophysiological recordings from cocultures were usually performed ∼3 d after the neurons were plated.
Whole mounts of the intact petrosal/glossopharyngeal nerve/carotid sinus nerve/CB complex were carefully dissected from rat pups (9–10 d old), taking care not to cut the CSN. In some cases, before organ culture, the petrosal ganglion was removed and the remaining tissue was placed on the membrane of Transwell-COL culture plates (Costar, Cambridge, MA). The preparation was semi-submerged and maintained in organ culture for 12–24 h at 35°C. After incubation, the whole mount was processed for immunofluorescence as described below.
Nystatin perforated-patch whole-cell recording.
Nystatin perforated-patch, whole-cell recording was used in this study according to procedures described in detail previously (Zhong et al., 1997; Zhang et al., 2000). Patch pipettes were made from Corning 7052 glass (A-M Systems, Carlsborg, WA) or borosilicate glass (World Precision Instruments, Sarasota, FL) using a vertical puller (PP 83; Narishige, Tokyo, Japan) and were fire polished. Micropipettes had a resistance of 2–10 MΩ when filled with intracellular recording solution and formed gigaseals between 2 and 12 GΩ. In most experiments, ∼75% of the series resistance was compensated, and junction potentials were cancelled at the beginning of the experiment. The extracellular fluid was warmed before entering the recording chamber, in which the mean temperature was 34 ± 2°C over the time course of the experiments. Whole-cell currents or membrane potentials were recorded with the aid of an Axopatch 1D amplifier (Molecular Devices, Palo Alto, CA) equipped with a 1 GΩ head-stage feedback resistor and a Digidata 1200 analog-to-digital converter (Molecular Devices) and stored on a personal computer. Current- and voltage-clamp protocols, data acquisition, and analysis were performed using pClamp software (version 6.0; Molecular Devices) and Axotape (version 2.02; Molecular Devices). ATP-induced currents were sampled at 0.5–1 kHz and stored on a computer for analysis with pClamp (version 9.0; Molecular Devices).
Solutions and drugs.
Experiments were performed using extracellular fluid of the following composition (in mm): 110 NaCl, 5 KCl, 2 CaCl2, 1 MgCl2, 10 glucose, 12 sucrose, and 24 NaHCO3, at pH ∼7.4 (maintained by bubbling with 5% CO2). For nystatin perforated-patch recordings (Zhong et al., 1997; Zhang et al., 2000), the pipette solution contained the following: 115 mm potassium glutamate or gluconate, 25 mm KCl, 5 mm NaCl, 1 mm CaCl2, 10 mm HEPES, and 300 μg/ml nystatin, at pH 7.2. All solutions were filtered through a 0.2 μm Millipore (Bedford, MA) filter before use. During recordings, the culture was continuously perfused under gravity flow, and the fluid in the recording chamber was maintained at a constant level by suction. The following drugs were obtained from Sigma: tetrodotoxin; tetraethylammonium; ATP; α,β-methylene ATP (α,β-MeATP); 2′-3′-O-(4 benzoylbenzoyl)-ATP (BzATP); adenosine 5′-diphosphate-2′,3′-dialdehyde or oxidized-ATP (OxATP); ivermectin (IVM); brilliant blue G (BBG); 2-(4-carboxyphenyl)-4,4,5,5-tetramethyl-imidazoline-1-oxyl-3-oxide potassium (cPTIO); Nω-nitro-l-arginine methyl ester hydrochloride (l-NAME); and sodium nitroprusside (SNP). All agonists (ATP, BzATP, and α,β-MeATP) were applied by a “fast perfusion” system using double-barreled pipettes, whereas modulators (i.e., IVM) and antagonists (suramin, OxATP, and BBG) were applied to the bath by perfusion under gravity (Zhong et al., 1997). EC50 and IC50 values were obtained by the best fit of the data using the Hill function. Results are represented as means ± SEM. Current densities (in picoamperes per picofarad) were compared using paired Student’s t test. Membrane potentials before and after drug application were compared by repeated-measures ANOVA. Ratios were compared using a nonparametric test (Mann–Whitney). In all cases, the level of significance was set at p < 0.05.
Whole mounts of the freshly dissected GPN nerve or of organ culture explants (see above) were fixed in 4% paraformaldehyde overnight at 4°C. Cultures of dissociated GPN neurons were fixed for 15 min at room temperature (Campanucci et al., 2003). In experiments in which CB sections were immunostained, the animal was fixed by perfusion with PBS containing paraformaldehyde as described previously (Zhang et al., 2000). After washing in PBS (three times for 5 min), whole mounts were incubated for 72 h at 4°C with primary antibody; an overnight incubation was used for cultures. Primary antibodies used were as follows: polyclonal rabbit antibodies against rat P2X2, P2X4, and P2X7 (1:200; Alomone Labs, Jerusalem, Israel), guinea pig anti-rat P2X3 (1:500; Chemicon, Temecula, CA), polyclonal rabbit anti-rat nNOS (1:200; ImmunoStar, Hudson, WI), monoclonal mouse and polyclonal rabbit anti-rat tyrosine hydroxylase (TH) (1:1000; Chemicon), monoclonal mouse anti-rat neurofilament (NF) (68 kDa; 1:5; Boehringer Mannheim, Montreal, Quebec, Canada), and monoclonal mouse anti-rat GAP-43 (1:20,000). The primary antiserum was diluted in PBS containing 1% BSA and 0.5% Triton X-100. After incubation, the samples were washed in PBS (three times for 5 min) and incubated in the dark for 1 h at room temperature with the secondary antibody conjugated to Alexa 594 (1:500; Invitrogen), Texas red (1:50; Jackson ImmunoResearch, West Grove, PA) or FITC (1:50; Cappel, Malvern, PA). The secondary antiserum was diluted in PBS containing 1% BSA and 0.5% Triton X-100. Samples were covered with a photobleaching reagent (Vectashield; Vector Laboratories, Burlingame, CA) and viewed with a confocal microscope (Microradiance 2000; Bio-Rad, Mississauga, Ontario, Canada) equipped with argon (two lines, 488 and 514 nm) and helium/neon (543 nm) lasers. Tissue was scanned in optical sections separated by 1–2 μm at a rate of 166 lines per second, and scanning was controlled with the aid of Lasersharp software. Between 10 and 30 μm of tissue in each whole-mount preparation was routinely scanned in this manner. In control experiments, either preincubation with blocking peptide or omission of the primary antibody resulted in complete abolition of staining. Blocking peptides for P2X2, P2X4, P2X7 (Alomone Labs), and P2X3 (Chemicon) were preincubated with primary antibody (3 μg peptide/3 μg antibody) overnight at 4°C before application. Image processing and manipulation was performed using CorelDraw 9 (Corel, Ottawa, Ontario, Canada).
Total RNA was extracted from ∼18 short (2–3 mm) segments of the glossopharyngeal nerve containing the distal population of GPN neurons using Bio-Rad AURUM TOTAL RNA Mini kit. RNA was reverse transcribed using Bio-Rad iScript cDNA Synthesis kit or Superscript III, RNase H-Reverse Transcriptase (Invitrogen). Gene-specific primers were designed using GeneFisher (Giegerich et al., 1996) and synthesized by The Central Facility of the Institute for Molecular Biology and Biotechnology (MOBIX) (McMaster University, Hamilton, Ontario, Canada). The following primers were used and are listed as sequence amplified, 5′ to 3′ upstream primer sequence, 5′ to 3′ downstream primer sequence, and size of product(s): glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 5′ TTC ACC ACC ATG GAG AAG GC 3′, 5′ GGC ATG GAC TGT GGT CAT GA 3′, 230 bp; P2X2/P2X2(b), 5′AAG TTG GGC CAA ACC TTT G 3′, 5′ ACC CCA AAT ATT CTT TCC GG 3′, 600 and 400 bp; P2X3, 5′ TTG AGG GTA GGG GAT GTG GT 3′, 5′ GCT GAT AAT GGT GGG GAT GA 3′, 326 bp; P2X4, 5′ AGT CTG TGA AGA CCT GTG A 3′, 5′ GGA GAC ACA TTH TGT TCC A 3′, 408 bp; and P2X7, 5′ AGA CAA ACA AAG TCA CCC GG 3′, 5′ GGT ATA CAC CTG CCG GTC TGG 3′, 400 bp.
DNA was amplified in a single round of PCR using either TaqDNA Polymerase High Fidelity (Invitrogen) or SuperTaq Polymerase (Ambion, Austin, TX).
PCR was held at 94°C for 2 min and cycled 35 times. Each cycle consisted of 94°C for 30 s (denaturation), 52°C [P2X2 and P2X2(b)], 55°C (P2X3 and GAPDH), 58°C (P2X7), and 50°C (P2X4) for 30 s (annealing), and 72°C for 1 min (extension). This was followed by a 10 min final extension at 72°C. PCR products were visualized on an ethidium bromide-stained 2% agarose gel under ultraviolet illumination. The 1 kb plus ladder used was obtained from Invitrogen. The QIAquick Gel Extraction kit (Qiagen, Mississauga, Ontario, Canada) was used to extract PCR fragments from the agarose gel. The DNA sample was then sequenced (at MOBIX) using an ABI Prism automated Sequencer (with T7 polymerase). The sequencing results were analyzed by BLAST2 (basic local alignment search tool), an NIH computer software for identification of gene sequences. The sequences were matched to the Rattus norvegicus P2X2 (GenBank accession number NM_053656), P2X2(b) splice variant (GenBank accession number Y10473), P2X3 (GenBank accession number X91167), P2X4 (GenBank accession number X93565), and P2X7 (GenBank accession number X95882.1).
Rat pups (10–14 d old) were anesthetized by intraperitoneal administration of Somnotol (65 mg/kg), before perfusion via the aorta with PBS, followed by PBS containing 4% paraformaldehyde. Fixed animals were then perfused with filtered (0.2 μm Millipore filter) Evans blue dye (100 mg of dye per milliliter of 0.9% NaCl, adjusted pH to 7.4), so as to stain paraganglia attributable to the high permeability of their blood vessels to large dye molecules (McDonald and Blewett, 1981). The carotid bifurcation was then excised, and the tissue was mounted in fresh PBS. Images were captured with a Zeiss (Oberkochen, Germany) S16 upright microscope equipped with a Retiga monochrome 12-bit digital camera (QImaging, Burnaby, British Columbia, Canada), with the aid of Northern Eclipse software (Empix Imaging, Mississauga, Ontario, Canada).
ATP-evoked responses in GPN neurons
As described previously (Campanucci et al., 2003), GPN neurons are concentrated in two groups, one proximal and near the CSN bifurcation and the other located more distally along the GPN. Before ATP application, GPN neurons from both groups were tested for O2 sensitivity by perfusing the culture with a hypoxic solution (PO2 of ∼15 Torr) as described previously (Campanucci et al., 2003). Essentially all neurons (>90%) responded to hypoxia with membrane depolarization that was sometimes large enough to trigger action potentials (Fig. 1A). Rapid perfusion of ATP (5 μm) over such neurons caused membrane depolarization and action potential firing (Fig. 1B). Under voltage clamp at −60 mV, ATP evoked a dose-dependent inward current in GPN neurons with an apparent EC50 of ∼10.5 μm (n = 11) (Fig. 1C,D). These data were not well fitted by a simple Hill equation, suggesting that multiple receptors were involved. The reversal potential of the ATP-evoked current occurred near 0 mV (n = 7) (Fig. 1E,F), suggesting that it was carried by nonselective cation channels. In addition, the ATP-evoked current showed typical inward rectification (Fig. 1F), as described previously for currents carried by P2X purinergic receptors (Brake et al., 1994; Evans et al., 1996) (for review, see North, 2002).
Evidence for functional expression of P2X2–P2X3 heteromeric receptors
The kinetic profile of ATP-evoked inward currents in GPN neurons indicated a fast activation phase, followed by slow inactivation (Fig. 1C, see example traces). Such biphasic kinetics are characteristic of cells expressing P2X2 and P2X3 homomeric or P2X2–P2X3 heteromeric receptors (North, 2002). Consistent with the presence of P2X2–P2X3 heteromeric receptors (Radford et al., 1997), α,β-MeATP evoked slowly desensitizing inward currents (at −60 mV) with an EC50 of ∼3.2 μm (Fig. 2A,B). These data suggest that GPN neurons express P2X2–P2X3 heteromeric receptors, although the expression of P2X2 and P2X3 homomeric forms cannot be excluded.
In addition, our experiments with the nonselective purinergic antagonist suramin (North and Surprenant, 2000) indicated that multiple P2X receptors are expressed in a single GPN neuron. During application of 50 μm ATP in the presence of high doses of suramin (500–1000 μm), concentrations expected to block P2X2 and P2X3 homomeric and heteromeric receptors (North and Surprenant, 2000), we observed a residual inward current (Fig. 2C,D), whose amplitude was ∼20% of the total ATP-evoked inward current (Fig. 2D). These findings suggest that, in addition to contributions from P2X2 and P2X3 subunits, other P2X purinergic subunits are involved.
Evidence for functional expression of P2X7 homomeric receptors
Both P2X7 and P2X4 display low sensitivity to suramin when heterologously expressed in their homomeric forms (Ralevic and Burnstock, 1998; North and Surprenant, 2000; Khakh, 2001; Khakh et al., 2001). Therefore, either of these two subtypes could be present in GPN neurons. To test for functional expression of P2X7 receptors, we applied low micromolar concentrations of BzATP, an agonist with even greater potency for P2X7 receptors than ATP (North and Surprenant, 2000; North, 2002). BzATP had an EC50 of ∼1.9 μm (n = 4) (Fig. 3B). These data suggest that at least a population of purinergic receptors expressed in GPN neurons has moderate-to-high affinity for BzATP, and this is consistent with the expression of homomeric P2X7 receptors (North, 2002).
Because micromolar concentrations of BzATP also activate heterologously expressed rat P2X1, P2X2, and P2X3 homomeric, as well as rat P2X2–P2X3 heteromeric, receptors (Bianchi et al., 1999; North and Surprenant, 2000), it was necessary to confirm the presence of P2X7 receptors by other methods. We therefore used the antagonist BBG (0.1–100 nm) (Fig. 3C), which is reported to be a selective blocker of heterologously expressed rat P2X7 receptors in the nanomolar concentration range (Jiang et al., 2000). BBG at concentrations between 10 and 100 nm inhibited up to ∼40% of the BzATP-evoked inward current (n = 5) (Fig. 3D) in GPN neurons. Moreover, additional confirmation of the functional expression of P2X7 receptors was obtained after exposure to the irreversible P2X7-selective antagonist OxATP (Murgia et al., 1993). In these experiments, we first preincubated GPN neurons with 300 μm OxATP for 2 h at 37°C to block P2X7 receptors (Murgia et al., 1993) and then tested the sensitivity of the remaining ATP-evoked currents to suramin. Pretreatment with OxATP decreased the residual ATP-evoked current observed in the presence of high doses of suramin (Fig. 4A,B), such that the residual current significantly decreased from 0.31 ± 0.02 pA/pF (n = 11) in control cells to 0.15 ± 0.06 pA/pF (n = 5) in cells pretreated with OxATP (p < 0.001). Moreover, the half-maximal ATP concentration (EC50) for OxATP-pretreated cells was ∼2.5 μm, a value significantly lower than that for control cells (EC50 of ∼10.5 μm; p < 0.05), consistent with inhibition of P2X7 receptors. This is illustrated by the leftward shift of the dose–response relationship for cells pretreated with OxATP (Fig. 4C). Together, these data provide evidence for the functional expression of P2X7 receptors in GPN neurons and further imply that, when these receptors are chemically disabled, purinergic signaling is maintained at least in part by purinoceptors containing P2X2 and P2X3 subunits.
Functional evidence for the expression of P2X4 receptors in GPN neurons
In approximately one in three neurons that were preincubated with OxATP, a substantial ATP-evoked current persisted in the presence of high doses (500–1000 μm) of suramin (data not shown), suggesting the presence of other P2X receptor subtype(s) distinct from P2X7. One candidate, homomeric P2X4 receptors, is known for its relative insensitivity to suramin (North and Surprenant, 2000). To test whether GPN neurons also expressed functional P2X4 receptors, we used IVM, a channel modulator that has been shown to enhance specifically P2X4-mediated currents in heterologous expression systems (Khakh et al., 1999). IVM (2–10 μm) caused potentiation of the ATP-evoked current in 5 of 13 GPN neurons tested (Fig. 5A). For reasons that are not presently understood, this drug also had an inhibitory effect in 5 of 13 (Fig. 5B) of neurons tested. Last, IVM had no effect in 3 of 13 neurons tested, suggesting the lack of expression of functional P2X4 receptors (Fig. 5C).
Expression of P2X subunits in glossopharyngeal nerve using reverse transcriptase-PCR techniques
To corroborate the expression pattern of P2X receptors as revealed by immunofluorescence (see below) and electrophysiological studies, short segments of glossopharyngeal nerve containing the distal population of GPN neurons were analyzed by reverse transcriptase (RT)-PCR techniques using appropriate primers (see Materials and Methods). As illustrated in Figure 6, bands of the expected size indicated mRNA expression of P2X2 (or P2X2a), P2X2(b) splice variant, P2X3, P2X4, and P2X7. Sequence analysis confirmed that the amplified products were identical (95–98%) to the known sequences for each subunit (supplementary data, available at www.jneurosci.org as supplemental material).
Cellular localization of P2X2, P2X3, P2X7, and P2X4 subunits in situ
Results from the functional, pharmacological, and expression studies discussed above led us to investigate the localization of P2X2, P2X3, P2X4, and P2X7 subunits in freshly dissected whole mounts of GPN nerve using immunofluorescence. As described previously (Campanucci et al., 2003), these neurons are concentrated in two groups, one proximal and near the CSN bifurcation and the other located more distally along the GPN. For both proximal (Fig. 7A–C) and distal (Fig. 7D–F) populations, there was colocalization of P2X2 and P2X3 subunits in the soma of GPN neurons as revealed by double-labeled immunofluorescence (n = 10). These data are consistent with the functional evidence for heteromeric P2X2–P2X3 receptors.
Because our electrophysiological data suggested the presence of P2X4 receptors in at least a subpopulation of GPN neurons, we used immunofluorescence to localize P2X4 subunits. Consistent with the data reported above, positive P2X4 immunofluorescence was localized in both populations of GPN neurons (Fig. 7G–L). However, the majority of NF/GAP-43-positive neurons observed along the GPN expressed immunoreactive P2X4 subunits (n = 10) (Fig. 7G–I). This contrasts with our electrophysiological data, which suggested that only ∼38% of GPN neurons expressed functional homomeric P2X4 receptors. In addition, P2X4 immunofluorescence colocalized with that of P2X3 in GPN cell bodies (n = 5) (Fig. 7J–L). These data confirm the expression of multiple P2X receptors in the same GPN neuron.
The distribution of P2X7 immunostaining was more complex. Whereas P2X7 subunits were localized in NF-positive cell bodies of the proximal GPN neurons (n = 7) (Fig. 8A–C), on the distal population, they appeared targeted to nerve endings surrounding the NF-positive cell bodies (n = 7) (Fig. 8D–F). Because both neuronal populations project to the CB (Wang et al., 1993, 1994; Campanucci et al., 2003), we also investigated the distribution of P2X7 immunoreactivity in tissue sections of the CB. We identified nerve terminals immunoreactive to NF and GAP-43 (n = 5) (Fig. 8G) in the CB; P2X7 punctate immunoreactivity colocalized with these nerve terminals in defined regions of the CB (Fig. 8H,I, arrows). Interestingly, punctate P2X7 immunostaining was seen near TH-positive clusters of CB chemoreceptor cells as illustrated in Figure 8J (n = 3). Furthermore, in CB cryosections, P2X7 immunoreactivity colocalized with that of P2X3 in nerve endings surrounding chemoreceptor cell clusters (data not shown), suggesting that GPN nerve terminals may also express multiple P2X receptor subunits, similar to the soma.
The CB receives both afferent innervation from the petrosal ganglion and efferent innervation from the GPN. Therefore, to determine whether this P2X7 immunostaining in the CB was solely of GPN origin, we performed “denervation” experiments. The small size of the GPN makes this extremely difficult in vivo, so we performed these experiments by transplanting the intact GPN, CSN, and CB to organ culture (incubated for 12 h at 37°C); in doing so, we effectively axotomized the afferent axons whose cell bodies are located in the petrosal ganglion. In these experiments, we observed a dramatic reduction in NF and GAP-43 staining of axons projecting to the CB. However, there were no significant differences in P2X7 puncta near clusters of CB chemoreceptor cells (n = 5; data not shown), suggesting that the GPN was at least a major source of P2X7 innervation in the CB. Equally important in these preparations, we observed several nNOS-positive nerve processes in close proximity to TH-positive CB chemoreceptor clusters (Fig. 8K), consistent with the proposal that local interactions may occur between GPN processes and CB chemoreceptor cells in situ (see below).
Localization of P2X7 immunofluorescence in cultured GPN neurons
Our immunocytochemical studies in situ indicated that P2X7 receptors are localized to nerve terminals rather than on the soma of the distal population (Fig. 8D–F); however, our electrophysiological results on freshly isolated neurons (≤24 h in culture) showed that both populations of GPN neurons expressed functional P2X7 receptors on their cell bodies. Our explanation for these results is that P2X7 receptors undergo a redistribution when neurons are isolated in culture and now target to the soma. To examine whether P2X7 undergo a redistribution, we immunolocalized P2X7 subunits in freshly-isolated GPN neurons (≤24 h old, i.e., conditions used for patch-clamp experiments) and on GPN neurons in culture for 48–72 h. Interestingly, in freshly isolated neurons, P2X7 immunofluorescence was concentrated on the cell bodies of both populations of GPN neurons (n = 10) (Fig. 9A–C). However, at 48–72 h after isolation, the P2X7 immunostaining was weaker in the cell bodies and was detected along nerve processes (n = 5) (Fig. 9D–F). These data suggest that, shortly after axotomy in vitro, P2X7 subunits undergo a redistribution and target to the soma of GPN neurons and that, over time in culture, the receptors are targeted, or transported away from the soma, to the nerve processes.
Effects of P2X receptor stimulation and hypoxia on NO signaling in cocultures of GPN neurons and carotid body chemoreceptors
During hypoxia, ATP is released from CB chemoreceptors (Zhang et al., 2000; Rong et al., 2003; Buttigieg and Nurse, 2004; Nurse, 2005) and/or from circulatory red blood cells (Ellsworth, 2000), suggesting that it may play a central role in the depolarization of GPN neurons, leading to activation of the CB efferent inhibitory pathway via release of NO (Wang et al., 1993, 1995a;b; Prabhakar, 1999). We hypothesized that stimulation of P2X receptors on GPN neurons leads to NOS activation and NO synthesis and release, which will cause hyperpolarization of CB chemoreceptors. To test this idea, we developed a novel coculture preparation of GPN neurons and CB receptor (type I) cell clusters. Thus, an overlay of dissociated rat GPN neurons was added to a preexisting monolayer of cultured CB type I cells, which lack P2X receptors at least at the ages (9–10 d) used in these studies (Zhang et al., 2000; Prasad et al., 2001) (but see Xu et al., 2005; He et al., 2006). GPN neurons, “juxtaposed” (JGPN) to type I cell clusters in coculture (Fig. 10A–C), were selected for rapid application of either ATP (5–10 μm) or BzATP (5 μm) while membrane potential was simultaneously monitored in a type I cell within the adjacent cluster. As reported previously (Zhang et al., 2000), 10–100 μm ATP had negligible effects on the resting potential of clustered type I cells cultured without neurons (mean ΔVm was 0.3 ± 0.1 mV; n = 30; p > 0.05) (Fig. 10D), and this was also the case for the P2X7 receptor agonist BzATP (n = 4) (Fig. 10E). However, in a few cases (n = 5; data not shown), ATP caused a small hyperpolarization (1.4 ± 0.5 mV; p < 0.05) (Xu et al., 2005).
Interestingly, application of BzATP (5 μm) to JGPN neurons in coculture caused a strong hyperpolarization (∼8 mV) of CB type I cells within the adjacent cluster (Fig. 10G); the membrane potential hyperpolarized from a mean resting level of −44.7 ± 3.1 to −52.2 ± 3.3 mV (n = 8) during BzATP application. In two cases from similar cocultures (Fig. 10F), BzATP also caused inhibition of the spontaneous activity that is sometimes seen in type I cells within large clusters (Zhang and Nurse, 2000). Because BzATP tended to produce irreversible long-term changes in type I cell membrane potential observed at ≥30 min after original BzATP application, we tested the effects of the naturally occurring nucleotide ATP, which is itself released from CB type I cells during hypoxia (Zhang et al., 2000; Buttigieg and Nurse, 2004; Zhang and Nurse, 2004). Moreover, type I cells in situ are seen in close proximity to P2X7- and nNOS-positive terminals of GPN neurons (Fig. 8J,K). As shown in Figure 10H, ATP also caused a strong hyperpolarization (∼10 mV; n = 7) in cocultured type I cells without producing any long-term changes in membrane potential; the latter hyperpolarized from a mean resting level of −44.6 ± 3.0 to −54.5 ± 3.3 mV (n = 7) in the presence of ATP.
Next, we investigated whether NO release from JGPN neurons was responsible for the hyperpolarization observed in type I cells. To test this, we used several different approaches. First, we exposed isolated type I cell clusters, grown in the absence of GPN neurons, to the NO donor SNP (500 μm). As exemplified in Figure 11A, SNP mimicked the ATP-induced hyperpolarization of the type I cell observed in cocultures. For a group of nine cells, SNP hyperpolarized the membrane potential from a mean resting level of −47.3 ± 4.8 to −53.2 ± 6.2 mV; the mean ± SEM change (ΔVm) in membrane potential induced by 500 μm SNP is summarized in Figure 11C (n = 9). Second, we exposed cocultures to the NO scavenger cPTIO (Summers et al., 1999). As illustrated in Figure 11B, preincubation (>2 min) with 500–1000 μm cPTIO abolished irreversibly the ATP-induced hyperpolarization in cocultured type I cells. The time-dependent decrease in magnitude of the ATP-induced hyperpolarization in the presence of cPTIO is illustrated for several different coculture preparations in Figure 11C (n = 6; data represent mean ± SEM). Third, to further support the involvement of NO in the ATP-induced hyperpolarization in JGPN neurons, we exposed cocultures to the neuronal NOS inhibitor l-NAME (500 μm), which also inhibited the observed hyperpolarization (n = 3) (supplemental Fig. 2, available at www.jneurosci.org as supplemental material). These data suggest that activation of P2X receptors on GPN soma/terminals could lead to NO release, which in turn initiates efferent inhibition of type I cell function via membrane hyperpolarization.
In the intact carotid body, basal NO release is thought to exert a tonic efferent inhibition of chemoreceptor function even under normoxic conditions (Prabhakar, 1999). This raises the possibility that, in cocultures of JGPN and type I clusters, basal NO release might blunt the hypoxic response in type I cells, and this effect might even be exaggerated because GPN neurons can also be directly excited by hypoxia, leading to voltage-dependent Ca2+ entry and possibly NO release (Campanucci et al., 2003; Campanucci and Nurse, 2005) (Fig. 1A). As exemplified in Figure 12A, hypoxia (PO2 of ∼5 mmHg) induced membrane depolarization (or a receptor potential) in type 1 cells cultured without neurons, and this response was unaffected by the NO scavenger cPTIO, as expected because the source of NO production (i.e., GPN neurons) was absent. For a group of six cells from different cultures, the mean hypoxia-induced depolarization was 7.5 ± 0.9 and 7.3 ± 0.7 mV in the absence and presence of cPTIO, respectively (p > 0.05; n = 6). In cocultures, however, hypoxia was either ineffective (n = 8) or produced small (<2 mV) depolarizations (n = 3) in type I cells adjacent to GPN neurons. To test whether endogenous NO was active in the latter conditions, cocultures were exposed to 500 μm cPTIO for up to 8 min, and, interestingly, there was a gradual potentiation of the hypoxic response (n = 5). Data from a typical example are summarized in Figure 12B–E in which, after a weak initial hypoxic response in a type I cell (Fig. 12B), addition of 500 μm cPTIO caused an increase in basal spontaneous spike activity by 1 min exposure (Fig. 12C). At this stage, a hypoxia-induced depolarization was still detectable, but its magnitude progressively increased and was pronounced by 5 min exposure to 500 μm cPTIO (Fig. 12D). These effects were fully reversible after washout of cPTIO (Fig. 12E). Data from five similar experiments from different cocultures are summarized in Figure 12F. Together, these results strongly support a role of GPN neurons and/or their terminals in NO-mediated tonic inhibition of chemoreceptor cell function and in the blunting of the receptor response to hypoxia.
During hypoxia, potential sources of ATP that could stimulate P2X receptors on GPN neurons or their processes include ATP released from CB type I cells (Zhang et al., 2000) and from red blood cells in the circulation (Ellsworth, 2000). To confirm that the microenvironment around GPN neurons in situ has ready access even to large molecules in the circulation, we fixed and perfused rat pups with a physiological salt solution containing Evans blue dye (McDonald and Blewett, 1981). This procedure revealed the anatomical structure of blood vessels surrounding the area in which both populations of GPN neurons were concentrated (n = 4) (supplemental Fig. 3A,B, available at www.jneurosci.org as supplemental material). As described previously (McDonald and Blewett, 1981), these groups of neurons, or paraganglia, are closely related to tortuous blood vessels that seem to spiral around the area surrounding the GPN neurons (supplemental Fig. 3C, available at www.jneurosci.org as supplemental material). Deposits of Evans blue dye were observed at these locations (supplemental Fig. 3C, available at www.jneurosci.org as supplemental material), confirming the high permeability of these blood vessels to the dye. These data suggest that GPN paraganglia may have direct access to substances released into the blood stream, for example ATP, released from red blood cells during hypoxia (for review, see Ellsworth, 2000).
Expression of multiple P2X receptors in GPN neurons
In this study, we characterized purinergic responses in nNOS-positive autonomic neurons that are embedded in the rat GPN and are proposed to mediate efferent inhibition of CB chemoreceptors via release of NO (Prabhakar et al., 1993; Wang et al., 1993, 1995a;b; Grimes et al., 1994; Höhler et al., 1994). These GPN neurons were shown previously to be hypoxia sensitive and are distributed in two distinct populations, i.e., a proximal one near the branch point of the GPN and CSN and a more distal one located farther along the GPN (Campanucci et al., 2003; Campanucci and Nurse, 2005). Our present findings indicate that both populations of GPN neurons express a variety of purinergic P2X receptors, and, additionally, we demonstrate in a novel coculture preparation that activation of these receptors by ATP could indeed lead to NO-mediated inhibition of CB chemoreceptors.
The ATP-evoked responses appeared to involve multiple P2X receptors. Electrophysiological and pharmacological data were consistent with the in situ immunolocalization studies showing the presence of purinergic P2X2, P2X3, P2X4, and P2X7 subunits in GPN neurons. Additionally, the RT-PCR approach using tissue containing the distal population of GPN neurons detected not only the full-length P2X2 isoform but also a splice variant, P2X2(b), which contains a 207 bp deletion in the C-terminal domain and is expressed in other autonomic neurons (Simon et al., 1997; Schädlich et al., 2001). This variant desensitizes more rapidly (approximately fivefold faster) during prolonged ATP application than the longer P2X2 isoform (Simon et al., 1997). The combined data suggested functional expression of at least heteromeric P2X2–P2X3, homomeric P2X4, and homomeric P2X7 receptors. The occurrence of heteromeric P2X2–P2X3 receptors was inferred from the observation that application of the ATP agonist α,β-MeATP activated inward currents with slow desensitization kinetics (Lewis et al., 1995; Radford et al., 1997) (for review, see Ralevic and Burnstock, 1998). These data, however, do not rule out the possibility that homomeric P2X2 and homomeric P2X3 receptors are also present.
Several pharmacological approaches indicated functional expression of P2X7 receptors in GPN neurons. Consistent with our electrophysiological data, P2X7 purinergic subunits were immunolocalized in both populations of GPN neurons in culture. Although P2X7 subunits were localized in the soma of proximal GPN neurons in situ, they were not detectable on the soma of the distal population, which appeared to be contacted by P2X7-immunopositive nerve endings. This raises the possibility that local circuits involving purinergic transmission may provide a means of cell-to-cell communication among GPN neurons and that P2X7 subunits are targeted to nerve endings as evidenced by the distribution of P2X7 immunostaining of isolated distal GPN neurons before and after process formation in culture. In this regard, it has been reported that P2X7 receptors are targeted to presynaptic terminals in both the CNS and PNS (Deuchars et al., 2001).
Our electrophysiological, immunofluorescence, and RT-PCR studies also uncovered a role for P2X4 receptors. In particular, IVM, a homomeric P2X4 potentiator (Khakh et al., 1999), enhanced ATP-evoked currents in ∼38% of the neurons, and this effect was likely mediated by P2X4 receptors. Surprisingly, our immunofluorescence data indicated that almost all GPN neurons were positive for P2X4 immunoreactivity. The reasons for this discrepancy are unknown, although P2X4 subunits may form a heteromultimeric variant with a different pharmacological profile than the homomeric form. Indeed, P2X4–P2X6 heteromeric receptors are potentiated by IVM (Khakh et al., 1999), but this does not explain the inhibition of the ATP-evoked response by IVM in ∼38% of the cells. An inhibitory effect of IVM has been reported in other native neuronal tissues, although the mechanisms are not understood (De Roo et al., 2003).
A comparison of fitted curves for the ATP dose–response curves for P2X receptors expressed in GPN neurons and for heterologously expressed P2X2–P2X3 heteromultimers (Lewis et al., 1995), homomeric P2X4 (Bo et al., 1995), and homomeric P2X7 (Chessell et al., 1998) receptors is shown in Figure 13. The EC50 for ATP on GPN neurons was close to that for P2X4 receptors, but the dose–response curve deviated from the P2X4 profile at lower and higher ATP concentrations (Fig. 13). The deviation is consistent with the additional presence of P2X2–P2X3 and P2X7 receptors, which should allow GPN neurons to be activated over a broad range of ATP concentrations given that P2X2–P2X3 heteromultimers possess an EC50 of ∼1 μm, whereas P2X4 and P2X7 homomeric receptors possess EC50 values of 10 and ≥100 μm, respectively.
Role of GPN neurons in carotid body chemoreceptor inhibition
In our immunofluorescence studies in situ, nNOS- and P2X7-positive nerve fibers, which appeared to originate from GPN neurons, were found in close proximity to CB chemoreceptor cells. These P2X7-positive fibers were also immunopositive for P2X3 subunits, suggesting that the terminal branches of GPN neurons may also express multiple P2X receptors similar to the soma. Although we did observe P2X7-positive immunoreactivity associated with blood vessels in the rat CB (Fig. 8H), this labeling did not colocalize with neuronal markers (Fig. 8G–I) and was likely attributable to the known expression of P2X7 subunits in smooth muscle and endothelium of blood vessels (for review, see Burnstock, 2002). To investigate directly whether P2X signaling might play a role in efferent CB chemoreceptor (type I cell) inhibition, we developed a novel coculture preparation consisting of type I cell clusters and juxtaposed GPN neurons. Interestingly, activation of P2X receptors on cocultured neurons by ATP or BzATP caused type I cell hyperpolarization that was prevented by preincubation with the NO scavenger cPTIO and mimicked by application of the NO donor SNP to type I cells cultured alone. These data suggest that NO was the “inhibitory signal” released on purinergic activation of cocultured GPN neurons. Additionally, the hypoxia-induced receptor potential recorded in type I cells was suppressed in similar cocultures, and this was apparently attributable to the release of NO from GPN neurons because it could be reversed by cPTIO. Because GPN neurons are also excited by hypoxia (Campanucci et al., 2003; Campanucci and Nurse, 2005), this stimulus may also lead to voltage-gated Ca2+ entry, NOS activation, and NO-mediated inhibition of the receptor potential in type I cells (Prabhakar, 1999). Additional interactions may occur between hypoxia and ATP in light of a recent report indicating an inhibitory effect of ATP, acting via P2Y receptors, on the hypoxia-induced [Ca2+]i elevation in isolated adult rat type I cells (Xu et al., 2005).
A physiological model for the activation of GPN neurons
Our findings have led to a “working model” for the role of GPN neurons in the efferent inhibition of the rat CB via their activation by hypoxia and ATP. As summarized in supplemental Figure 1 (available at www.jneurosci.org as supplemental material), a reduction in blood PO2 (hypoxemia) causes at least four effects, as follows: (1) Inhibition of background [TASK-1 (TWIK-related acid-sensitive K+ channels)-like] K+ and large-conductance Ca2+ activated K+ channels in carotid body type I cells (for review, see López-Barneo et al., 2001), leading to or enhancing membrane depolarization, activation of L-type Ca2+ channels, and neurotransmitter (ATP and ACh) release (Zhang et al., 2000). These neurotransmitters act on petrosal sensory nerve endings via P2X2–P2X3 receptors (Prasad et al., 2001; Rong et al., 2003) and nicotinic ACh receptors, leading to a compensatory reflex increase in ventilation and restoration of blood PO2. (2) ATP, released from type I cells during hypoxia, may act on nearby GPN nerve terminals, which express multiple P2X receptors containing P2X2, P2X2(b), P2X3, P2X4, and P2X7 subunits. Activation of these P2X receptors causes Ca2+ influx and nNOS activation, leading to NO synthesis and release. NO actions on type I cells lead to inhibition of neurotransmitter release via membrane hyperpolarization (Silva and Lewis, 2002; this study) and/or direct inhibition of L-type Ca2+ channels (Summers et al., 1999). (3) Hypoxia causes red blood cells to release ATP (for review, see Ellsworth, 2000), which may directly stimulate GPN neurons via P2X receptors and endothelial cells via P2Y receptors (Ellsworth, 2000). Indeed, staining of whole mount of the GPN during perfusion with the large-molecular-weight Evans blue dye (supplemental Fig. 3, available at www.jneurosci.org as supplemental material) confirmed that GPN neurons were associated with an intricate network of fenestrated capillaries (McDonald and Blewett, 1981), thereby providing easy access to blood-borne substances such as ATP, present in millimolar quantities in erythrocytes (Miseta et al., 1993). (4) Hypoxia may also activate GPN neurons by inhibiting background K+ channels (Campanucci et al., 2003), leading to increased firing, voltage-gated Ca2+ entry, nNOS activation, and NO release. Thus, activation of GPN neurons by hypoxia and/or ATP, released during hypoxic stress, would lead to efferent inhibition of CB chemoreceptors, thereby providing a dual mechanism for negative feedback control of respiration via the same neuronal pathway.
This study was supported by Canadian Institutes for Health Research (CIHR) Grant MOP-57909 (C.A.N.). V.A.C. was supported by a scholarship from CIHR and a Dr. Sydney Segal research grant from The SIDS Foundation of Canada. We thank Stephen Brown for expert technical advice and unfailing assistance during the RT-PCR experiments, Dr. J. F. X. Jones for technical advice on Evans blue dye staining, and Dr. E. Cooper for his critical comments on this manuscript.
- Correspondence should be addressed to Dr. Verónica A. Campanucci at her present address: Department of Physiology, McGill University, McIntyre Medical Science Building, Room 1129, 3655 Promenade Sir William Osler, Montréal, Québec, Canada H3G 1Y6.
- Bianchi et al., 1999.↵
- Bo et al., 1995.↵
- Brake et al., 1994.↵
- Burnstock, 2002.↵
- Buttigieg and Nurse, 2004.↵
- Campanucci and Nurse, 2005.↵
- Campanucci et al., 2003.↵
- Chessell et al., 1998.↵
- De Roo et al., 2003.↵
- Deuchars et al., 2001.↵
- Dunn et al., 2001.↵
- Ellsworth, 2000.↵
- Evans et al., 1996.↵
- Fung et al., 2001.↵
- Giegerich et al., 1996.↵
- Gonzalez et al., 1994.↵
- Grimes et al., 1994.↵
- He et al., 2006.↵
- Höhler et al., 1994.↵
- Jiang et al., 2000.↵
- Khakh, 2001.↵
- Khakh et al., 1999.↵
- Khakh et al., 2001.↵
- Kline et al., 1998.↵
- Lewis et al., 1995.↵
- López-Barneo et al., 2001.↵
- McDonald and Blewett, 1981.↵
- Miseta et al., 1993.↵
- Murgia et al., 1993.↵
- North, 2002.↵
- North and Surprenant, 2000.↵
- Nurse, 2005.↵
- Nurse and Zhang, 1999.↵
- Prabhakar, 1999.↵
- Prabhakar et al., 1993.↵
- Prasad et al., 2001.↵
- Radford et al., 1997.↵
- Ralevic and Burnstock, 1998.↵
- Rong et al., 2003.↵
- Schädlich et al., 2001.↵
- Silva and Lewis, 2002.
- Simon et al., 1997.↵
- Summers et al., 1999.↵
- Vincent, 1994.↵
- Wang et al., 1993.↵
- Wang et al., 1994.↵
- Wang et al., 1995a.↵
- Wang et al., 1995b.↵
- Xu et al., 2005.↵
- Zhang and Nurse, 2000.↵
- Zhang and Nurse, 2004.↵
- Zhang et al., 2000.↵
- Zhong et al., 1997.↵