GABA is the main inhibitory neurotransmitter that participates in the regulation of cholinergic outflow to the airways. We have tested the hypothesis that a monosynaptic GABAergic circuit modulates the output of airway-related vagal preganglionic neurons (AVPNs) in the rostral nucleus ambiguus by using a dual-labeling electron microscopic method combining immunocytochemistry for glutamic acid decarboxylase (GAD) with retrograde tracing from the trachea. We also determined the effects of blockade of GABAA receptors on airway smooth muscle tone. The results showed that retrogradely labeled AVPNs received a significant GAD-immunoreactive (GAD-IR) terminal input. Out of a pooled total of 3,161 synaptic contacts with retrogradely labeled somatic and dendritic profiles, 20.2% were GAD-IR. GAD-IR terminals formed significantly more axosomatic synapses than axodendritic synapses (P < 0.02). A dense population of GABAergic synaptic contacts on AVPNs provides a morphological basis for potent physiological effects of GABA on the excitability of AVPNs. GAD-IR terminals formed exclusively symmetric synaptic specializations. GAD-IR terminals were significantly larger (P < 0.05) in both length and width than unlabeled terminals synapsing on AVPNs. Therefore, the structural characteristics of certain nerve terminals may be closely correlated with their function. Pharmacological blockade of GABAA receptors within the rostral nucleus ambiguus increased activity of putative AVPNs and airway smooth muscle tone. We conclude that a tonically active monosynaptic GABAergic circuit utilizing symmetric synapses regulates the discharge of AVPNs.
- airway-related vagal preganglionic neurons
- central control of airways
- synaptic neurotransmission
- tonic GABAergic inhibition
in humans, as in other mammals, the cholinergic outflow to the airways is centrally regulated. The motor component of the network innervating the airways arises from the nucleus ambiguus and from the dorsal motor nucleus of the vagus (21, 28, 32, 33, 43). Of these two groups of neurons, the airway vagal preganglionic neurons (AVPNs) within the rostral nucleus ambiguus (rNA) play a greater role in generating the cholinergic outflow to airway smooth muscle, secretory glands, and blood vessels (22-24, 51).
The activity of AVPNs and cholinergic outflow to the airways critically relies on afferent signals from the airways (10, 35, 38, 56, 60) and can be reduced or augmented by inputs from central nervous system cell groups that project to vagal preganglionic neurons (21, 25-27, 49, 51). It has been hypothesized that an imbalance between excitatory and inhibitory influences may result in overreactivity and increased cholinergic outflow to the airways, leading to elevation in bronchomotor tone, airway hypersecretion, and increased blood flow (27).
GABA, as the main inhibitory neurotransmitter, is essential for maintaining the overall balance between neuronal excitation and inhibition. We have found that AVPNs express the GABAA receptor subtype. Furthermore, when GABA or benzodiazepines are topically applied or endogenously released within the rNA region, airway smooth muscle relaxation results (24, 29, 30). The effects of GABA that have been observed could potentially be due either to local release of the neurotransmitter from classically defined synaptic contacts or to diffuse release into the extracellular space (volume transmission; see Ref. 1). The former mechanism would result in a precise localized regulation of synaptic transmission, whereas the latter would provide a mechanism for more tonic inhibitory regulation.
We hypothesized that cholinergic outflow to the airways is regulated mainly by direct axosomatic and/or axodendritic synapses between GABAergic terminals and AVPNs and to a lesser degree via extrasynaptic volume transmission. Hence, we anticipated that GABAergic synaptic unit, designed for precise discrimination of high-frequency signals and for fast off-switch of neuronal activity, is involved in regulation of AVPN excitability and exerts tonic inhibitory inputs to AVPNs. Determining GABAergic microcircuitry is of critical importance for better understanding of the kinetics of inhibitory regulation of AVPNs and the development of novel therapeutic strategies for the treatment of centrally mediated airway hyperresponsiveness.
Accordingly, in this study, we have used ultrastructural methods to test the hypothesis that GABA-containing nerve terminals make direct synaptic contacts with AVPNs. Furthermore, we have used localized injections of GABAA receptor antagonists into the rNA to demonstrate that a continuously active inhibitory GABAergic circuit modulates AVPNs, leading to withdrawal of cholinergic outflow to the airways and airway smooth muscle relaxation.
All experiments were reviewed and approved by the Institutional Animal Care and Use Committee of Case Western Reserve University in accordance with the guidelines established by the National Institutes of Health for the humane treatment of animals. The studies were performed on 20 male European ferrets, Mustella putorius furo (700-900 g). Male animals were utilized to minimize potential physiological changes due to hormonal effects associated with the reproductive cycle. Four ferrets were used for neuroanatomic experiments and 16 animals for physiological studies, which examined the effects of blockade of GABAA receptors on single-unit activity of presumptive AVPNs and airway smooth muscle tone.
Ultrastructural Studies of GABAergic Innervation of AVPNs
As previously described in detail (28, 42), cholera toxin β-subunit conjugated to horseradish peroxidase (CTB-HRP) was injected into the tracheal wall of four anesthetized ferrets (sodium pentobarbital, 50 mg/kg ip). Four days later, the animals were anesthetized as before, mechanically ventilated with 100% oxygen, and perfused through the left ventricle with 0.1 M sodium phosphate-buffered saline (PBS), pH 7.4, containing heparin (10,000 IU), followed by a mixture of 2.5% acrolein and 0.5% paraformaldehyde (in PBS). The brains were removed, stored in PBS, and then cut into 40-μm serial sections from the level of the spinomedullary junction to the rostral border of the pons by use of a vibratome. All incubations were done at room temperature unless otherwise noted.
To visualize cell bodies labeled with CTB-HRP, free floating Vibratome sections were processed with a modification of the tungstate-stabilized tetramethylbenzidine (TMB) method of Weinberg and van Eyck (63), as described in detail by Llewellyn-Smith and Minson (39). This method yields a crystalline reaction product that is easily detectable in the electron microscope. Briefly, medullary sections were preincubated for 20 min in a tungstate-TMB solution [0.05% ammonium paratungstate, 0.0025% TMB (free base), 0.04% ammonium chloride (NH4Cl), and 0.2% d-glucose in 0.1 M sodium phosphate buffer, pH 6.0 (PB6)]. After the preincubation, sections were incubated for 30 min in a fresh mixture of the same solution to which 100 μl of glucose oxidase is added. The sections were rinsed in PB6 and then incubated for 7-10 min in a cobalt chloride (CoCl)-diaminobenzidine (DAB) solution (100 mg DAB, 0.02% CoCl, 0.04% NH4Cl, 0.2% d-glucose, and 200 μl glucose oxidase in PB6) to stabilize and enhance the TMB reaction (minor variation of the method of Rye et al., Ref. 54). Sections were rinsed in PBS and then processed immunocytochemically for the presence of glutamatic acid decarboxylase (GAD).
Sections were washed for 30 min in 50% ethanol, incubated for 30 min in 1% bovine serum albumin (BSA) in PBS, and then incubated overnight (4°C) in rabbit anti-GAD-67 serum diluted 1:2,000 in PBS containing 0.1% BSA (PBS-BSA) (Chemicon Labs, Burlingame, CA). The next day, sections were rinsed with PBS-BSA, incubated for 1 h in biotinylated goat anti-rabbit serum diluted 1:200 in PBS-BSA (Vector Labs), rinsed with PBS-BSA, and then incubated for 1 h in avidin-biotin peroxidase complex (1:50 in PBS-BSA, Elite ABC kit, Vector Labs). Sections were rinsed in PBS and visualized with DAB (100 mg DAB, 0.04% NH4Cl, 0.2% d-glucose, 200 μl glucose oxidase in 0.1 M PBS). The result of this method appears as an amorphous reaction product in the electron microscope.
Sections containing CTB-HRP-labeled cells were processed for electron microscopic analysis. Briefly, these tissues were fixed for 1 h in 2% osmium tetroxide. Sections were rinsed in PBS, dehydrated through an ethanol series (30, 50, 70, 95, and 100%) and 100% propylene oxide, and then placed into a 50:50 epoxy resin-propylene oxide mixture and incubated overnight. The next day, the sections were incubated for 4 h in 100% resin (EMBed 812 kit, Electron Microscopy Sciences, Fort Washington, PA). In 100% resin, sections were flat embedded between two sheets of plastic (Aclar, Dupont) and cured for 48 h at 60°C. Section halves containing CTB-HRP-positive cells were identified and reembedded in Beem capsules for 48 h at 60°C. Ultrathin tissue sections (70-90 nm) were cut with a diamond knife (Delaware Diamond Knives, Wilmington, DE) on a Reichert Ultracut S ultramicrotome (Reichert, Leica Microsystems) and mounted onto 300-mesh copper grids (Electron Microscopy Sciences). Grids were stained with uranyl acetate and Reynolds lead citrate solutions with an Ultrastainer (Leica Microsystems) and examined in a JEOL JEM-1210 transmission electron microscope at 50 kV accelerating voltage.
For each animal, one 40-μm-thick tissue section that exhibited the best combination of morphological preservation and histochemical labeling was used for analysis. One ultrathin section from the center and each outside surface of each 40-μm-thick section were analyzed. This resulted in a total of three ultrathin sections separated from each other by no less than 18 μm. Ultrathin sections were examined at ×8,000 magnification for the presence of the tungstate-TMB-generated crystalline reaction product. The entire circumferences of all retrogradely labeled profiles were photographed at ×4,000-20,000 magnification. Synaptic contacts were characterized from photographs at ×10,000-50,000 magnification. Neuronal and dendritic profiles were identified according to previous ultrastructural descriptions (20, 42, 52). All terminals that made synaptic contact with CTB-HRP-labeled profiles were identified and categorized as “labeled” or “unlabeled” on the basis of the presence of GAD immunoreactivity in the terminal. These terminals were identified further as either “axosomatic” or “axodendritic” contacts. Terminals that were in close apposition to CTB-HRP-labeled profiles but did not display a clear synaptic contact in the plane of section were not counted. The total numbers of labeled and unlabeled terminals and axosomatic and axodendritic synaptic contacts were tabulated per animal and then summed. The size of 100 randomly selected GAD-IR and nonGAD-IR terminals, which either synapsed on AVPNs or were in the rNA but did not synapse on AVPNs, were measured in two dimensions (length and width).
The data are summarized as means ± SE. The data were analyzed with a Student's t-test or two one-way ANOVAs followed by post hoc Newman-Keuls tests. We considered differences significant at P < 0.05.
These experiments were designed to define the extent of the involvement of GABAA receptor activity on the regulation of cholinergic outflow to the airways. We used pressure ejection of nanoliter per picomole amounts of drug solutions of bicuculline (Bic, a competitive GABAA receptor antagonist; 1-10 mM; 20-60 nl). Because Bic may affect small-conductance calcium-activated potassium (SK) channels (12), we also used (+)β-hydrastine (Hyd, a specific GABAA receptor antagonist; 1-10 mM), which under in vivo conditions has minimal or no effects on apamin-sensitive SK channels of respiratory-related bulbospinal neurons (62).
Experiments were performed in ferrets that were initially anesthetized with thiopental (25 mg/kg ip) followed by α-chloralose (70 mg/kg ip) and then tracheotomized, and the carotid artery and femoral vein were cannulated. A tracheostomy tube was inserted through an incision placed in the caudal portion of the cervical trachea and connected to a Harvard ventilator. Animals were subsequently paralyzed (gallamine hydrochloride, 4 mg/kg iv) and mechanically ventilated with 100% oxygen at a constant volume of 7 ml/kg delivered at a frequency of 35-40 breaths/min. Body temperature was continuously monitored through an esophageal thermistor probe and maintained at 38.5-39.5°C by means of a heating pad. After instrumentation, ferrets were placed in a prone position in a stereotaxic apparatus and the effects of GABAA receptor blockade within the rNA region on single unit activity and tracheal smooth muscle tone were studied. Before blockade of GABAA receptors, all studied ferrets were pretreated with propranolol (1 mg/kg iv) to block β-adrenergic receptors and eventual peripheral effects of catecholamines on airway smooth muscle tone.
Extracellular unit recording. In seven ferrets, in vivo extracellular recordings were performed by using fine-tip (1.0-3 μm) tungsten microelectrodes (5-10 MΩ at 1 kHz; Fred Haer). As previously described (64), the microelectrode was attached to a glass micropipette with a diameter of 40 μm. The assembly was inserted unilaterally into the right rostral ventrolateral portion of the medulla oblongata, 3.0-3.6 mm rostral to the obex, 3.0-3.5 mm lateral to the midline, and 1-1.5 mm dorsal to the ventral medullary surface, by use of a dorsal approach to the brain stem, as previously described (30). Briefly, a portion of the bone and the dura overlying the right hemicerebellum were removed to permit advancement of the recording electrode and micropipette. In this region of the ferret medulla oblongata, our group has previously observed clustered AVPNs that project to the extrathoracic trachea (28).
Signals generated by these neurons were fed via high-impedance headstages to second-stage amplifiers (Grass P-511), filtered (0.3-3 kHz), and fed in parallel to an oscilloscope, audio monitor, and analog-to-digital computer-based acquisition system (PowerLab, AD Instruments) with a sampling rate of 4 kHz per channel for offline analyses. Neurons in this region were spontaneously active when ferrets were ventilated with a volume of 7 ml/kg delivered at a frequency of 35-40 breaths/min. Unit recordings were considered to emanate from putative AVPNs if firing rate was increased by orthodromic activation achieved through stimulation of pulmonary C fiber receptors and/or by lung deflation. Pulmonary C fiber receptors were stimulated by capsaicin (10 μg/kg) given as a bolus. In the open-chest ferret, lung deflation was induced by turning off the ventilator during the deflation phase for 5-10 s while the ferret was ventilated with oxygen.
In three additional ferrets, the recurrent laryngeal nerve of the right side was prepared and cut just caudal to the larynx. The tissue around the central end of the nerve was carefully removed for a length of 3 cm from the central cut end. The ferret was placed in a prone position with the head fixed in a stereotaxic apparatus, and then the dissected recurrent laryngeal nerve was isolated and placed on a bipolar hook electrode, made from a pair of silver wires with an interelectrode distance of 2 mm, and covered with a mixture of petroleum jelly and mineral oil. We attempted antidromic activation of single units within rNA by stimulating the recurrent laryngeal nerve with increasing frequencies (1-300 Hz) of electric current (0.1 μA-10 mA).
After identification of putative vagal preganglionic cells, and after recording of control levels of activity, the GABAA receptor blocker Bic was ejected from the puffer pipette (1-3 puffs of Bic; 20-60 nl; 0.2-0.6 nmol), by use of a Picospritzer III (Parker Hamilton Instruments). Changes were evaluated by employing a threshold-based software window discriminator to calculate frequency histograms for a given putative AVPN unit. Frequency changes for a given treatment were compared by use of Student's t-test, and significance was evaluated at P < 0.05.
Tracheal smooth muscle tone. The role of GABAergic inhibitory inputs in the regulation of airway smooth muscle tone was studied in a separate group of six ferrets. Tracheal smooth muscle tone was assessed indirectly by measuring the changes in pressure (in cmH2O) in a balloon placed in a bypassed rostral segment of the cervical trachea, as previously described (24). In bypassing the extrathoracic tracheal segment, care was taken not to damage the recurrent and superior laryngeal nerves and the plexus of ganglia on the posterior wall or to interrupt the blood supply. The balloon in the extrathoracic trachea was distended with 0.8-1.2 ml of saline.
Initial measurements were performed to ensure that the efferent transmission of cholinergic outflow to the airways was not affected by the surgery. This was achieved by demonstrating that the reflex responses that caused changes in tracheal smooth muscle tone were intact. To determine basal tracheal tone, the pressure in the balloon (Ptseg) was measured after withdrawal of cholinergic outflow to the airways induced by hyperoxic hypocapnia. The hyperoxic hypocapnia was produced by gradually increasing the rate of the ventilator to lower arterial CO2 and consequently to reduce the tracheal tone to ∼10 cmH2O. This value was considered to be basal tracheal tone and was close to that recorded after intravenous administration of atropine, as previously described (48). Only ferrets that, under hyperoxic and normocapnic conditions, responded to lung deflation of 10-s duration with an increase in tracheal smooth muscle tone >3 cmH2O were included in these studies. After the steady-state condition was achieved, the rate of the ventilator was slightly decreased to increase airway smooth muscle tone just above basal values. The arterial Pco2 caused by this procedure usually varies between 35 and 40 Torr. Once a steady state was reached, blockade of GABAA receptors was initiated, as described above. At the end of these experiments, 100 nl of 1% Fast Green dye was injected through the micropipette into the rNA region to permit histological identification of the location of the micropipette. Subsequently, the animal was perfused and the brain was removed, postfixed in formalin, and processed for identification of the injection site within the rNA. The areas with greatest dye density were considered to be the injection site. Figure 1 shows the region within the rostral ventrolateral medulla (i.e., the rNA) where interventions were made.
Recordings from physiological experiments were analyzed to determine airway smooth muscle tone responses to blockade of GABAA receptors. Average values of each variable are presented as means ± SE. Statistical comparisons were made by using the Student's t-test. The criterion for statistical significance was P < 0.05.
In the present study, we focused on the AVPNs that innervate the extrathoracic trachea and that are located within the rNA (Fig. 1). In the electron microscope, CTB-HRP labeled AVPN perikarya, proximal and distal dendrites, and myelinated axons were easily detected by the presence of a crystalline TMB-tungstate reaction product (Figs. 2, 3, 4). CTB-HRP-labeled axon terminals were not detected. AVPNs were relatively large with a round nucleus, prominent nucleolus and abundant mitochondria, rough endoplasmic reticulum, and few lysosomes (Fig. 2).
GAD-IR, observed as an amorphous electron-dense DAB reaction product in the electron microscope, was detected in perikarya (Fig. 3A), unmyelinated axons (Fig. 3B), and nerve terminals (Figs. 2, 3, 4). GAD-IR terminals formed both axosomatic (Fig. 2) and axodendritic (Figs. 3 and 4) synaptic contacts with CTB-HRP-labeled perikarya and dendrites. GAD-IR terminals contained small round vesicles, numerous mitochondria (Figs. 3 and 4), and rarely a few dense-core vesicles (Fig. 3A). All GAD-IR terminals formed symmetric synapses (Figs. 2, 3, 4). Unlabeled terminals that synapsed on AVPNs also contained round vesicles, occasional dense-core vesicles, and mitochondria. These terminals formed both symmetric (Fig. 3A) and asymmetric synapses (Fig. 4). Out of a pooled total of 3,161 synaptic contacts with retrogradely labeled somatic and dendritic profiles, 639 (20.2%) were GADIR. The proportion of GAD-IR terminals that formed axosomatic synapses (101/331; 30.5 ± 0.003%) was significantly higher (P < 0.02) than the proportion that formed axodendritic synapses (538/2,830; 19 ± 0.004%).
The dimensions of 100 GAD-IR and 100 unlabeled terminals that contacted AVPNs and the dimensions of 100 GAD-IR and 100 unlabeled terminals found in the rNA but not contacting AVPNs were measured. The size of GAD-IR terminals that synapsed on AVPNs was 0.77 ± 0.03 × 1.48 ± 0.05 μm (n = 100). By comparison, the size of randomly selected GAD-IR terminals in the rNA was 0.89 ± 0.04 × 1.37 ± 0.06 μm (n = 100). The size of unlabeled terminals that synapsed on AVPNs was 0.66 ± 0.03 × 1.19 ± 0.06 μm(n = 100). By comparison, the size of randomly selected unlabeled terminals in the rNA was 0.67 ± 0.03 × 1.09 ± 0.05 μm (n = 100). Two ANOVAs were calculated. Significant F values were obtained for both length [F(3, 396) = 10.77, P < 0.0001] and width [F(3, 396) = 9.64, P < 0.0001] and further tested with Newman-Keuls post hoc tests. The dimensions of GAD-IR terminals that contacted AVPNs were significantly larger in both dimensions than unlabeled terminals that contacted AVPNs (P < 0.05). Randomly labeled GAD-IR terminals in the rNA were significantly smaller in length (P < 0.05) than GAD-IR terminals that contacted AVPNs. However, there was no statistically significant difference in the width of GAD-IR terminals in these two groups (P > 0.05). GAD-IR terminals in the rNA were significantly larger in both dimensions than unlabeled terminals in the rNA (P < 0.01). Finally, the dimensions of unlabeled terminals that contact AVPNs and unlabeled terminals in the rNA were not different in either dimension (P > 0.05).
Extracellular unit recordings. In seven ferrets, single-unit discharge was recorded from a total of 13 units within the right rNA region. Nine units were identified as presumptive AVPNs, on the basis of their response to stimulation of C-fiber receptors by capsaicin and/or rapidly adapting receptors by lung deflation, as exemplified in Fig. 5. The putative AVPN units exhibited low spontaneous activity that could be affected by changes in chemical drive. In two tested animals, at high levels of discharge, microinjection of GABA into the rNA region reduced their discharge by ∼60% (n = 2; data not shown). By contrast, blockade of GABAA receptors by local injections of Bic (0.6 nmol in 60 nl) into the same region increased unit discharge (n = 7). An example is shown in Fig. 6. As can be seen, in a control period, putative AVPNs exhibited relatively low spontaneous firing, which significantly increased after GABAA receptor blockade (Fig. 6B; P < 0.05).
In three additional ferrets, we observed that electrical stimulation of central end of recurrent laryngeal nerve induced an increase in unit frequency discharge. However, no single antidromic spikes were recorded. These three units also responded to lung deflation with elevation in unit discharge, as other presumptive AVPNs.
Tracheal smooth muscle tone. In six ferrets, (three spinalectomized at C8 level), microinjection of 20-60 nl saline into the right rNA had no effect on Ptseg. However, microinjection of equal volumes of the GABAA receptor antagonists Bic (0.6 nmol in 60 nl; n = 2) or Hyd (0.2-0.6 nmol in 20-60 nl; n = 4) into the right rNA increased tracheal tension. An example of the airway smooth muscle tone response to unilateral injection of 0.6 nmol of Hyd into the rNA region is presented in Fig. 7A. In this animal, i.e., blockade of GABAA receptors within the rNA caused tracheal pressure to increase from 12 to 21 cmH2O. On average (Fig. 7B), after administration of Bic or Hyd (14 trials in 6 animals), Ptseg increased by 8.8 ± 1.1 cmH2O (Fig. 7B, bar B). Differences between control and values after GABAA receptor blockade were statistically significant (P < 0.001).
The tracheal constrictor response to microinjection of GABAA receptor antagonists into rNA was abolished by prior microinjection of 100 nl of 2% lidocaine into the rNA. Data shown in Fig. 7B (bar C, n = 3) indicate changes in tracheal pressure from baseline due to Hyd after lidocaine has been administered. After lidocaine, GABAA receptor antagonists did not cause any significant change in tracheal smooth muscle tone (P > 0.05).
The response of the tracheal smooth muscle tone to GABAA receptor antagonists was completely abolished after blockade of muscarinic receptors by systemic administration of atropine methylnitrate (1 mg/kg iv), a drug that does not cross the blood barrier. After atropine, administration of effective concentrations of either GABAA receptor antagonist into the same rNA region had no effect on tracheal tone (Fig. 7B, bar D; n = 3; P > 0.05), indicating that effects of GABAA receptor antagonists on tracheal tone were mediated centrally via an increase in cholinergic outflow to the airways.
Changes in tracheal tone induced by microinjection of GABAA receptor antagonists were not associated with statistically significant changes in mean arterial pressure or heart rate in either spinalectomized or in nonspinalectomized animals (P > 0.05). Thus, for example, in nonspinalectomized animals, heart rate before Bic microinjection was 350 ± 15 beats/min, whereas heart rate after Bic was 306 ± 60 beats/min. Similarly, blood pressure before Bic was 109 ± 11 mmHg, whereas blood pressure after Bic was 123 ± 19 mmHg. In addition, in spinalectomized ferrets, no effect of GABAA receptor blockade was observed. It should be noted, however, that the small sample size used (three spinalectomized ferrets) may preclude the possibility of detecting significant changes in these parameters.
In the ferret, both GAD-IR nerve terminals and perikarya are found in the rNA. GAD-IR terminals formed exclusively symmetric synapses on AVPNs, as well as on unidentified neurons and their processes in the rNA. This morphological observation is consistent with the physiological data that show that GABA is a major inhibitory neurotransmitter in the ventrolateral medulla (5, 6, 18). Furthermore, the data indicate, for the first time, that inhibitory GABAergic synaptic transmission in the rNA occurs in large part through classically defined symmetric synaptic contacts, although a component utilizing “volume transmission” (1, 27, 55) cannot be excluded. In the rNA, unlabeled nerve terminals, by contrast, formed both symmetric and asymmetric synapses on AVPNs, as well as on unidentified neurons and their processes. These data suggest that other inhibitory neurotransmitters modulate the activity of neurons in the rNA, including AVPNs. Our laboratory has previously demonstrated, in fact, that a noradrenergic pathway originating from the locus coeruleus also inhibits the activity of AVPNs (27). The presence of asymmetric synapses in the rNA is consistent with the data that excitatory neurotransmitters also modulate the activity of AVPNs (22, 23, 42), as well as other neurons with different functions in the nucleus ambiguus (18, 19). GAD-IR terminals formed a significantly higher proportion of the axosomatic synapses than of the axodendritic synapses that were observed on AVPNs. These GABAergic axosomatic and axodendritic synapses on AVPNs provide a morphological basis for the potent physiological effects of GABA on airway functions that were observed in the present study. GAD-IR terminals were shown to be significantly larger in both length and width than unlabeled terminals synapsing on AVPNs in the rNA. GAD-IR terminals in the rNA that did not synapse on AVPNs were also shown to be significantly larger in both length and width than unlabeled terminals synapsing on unidentified neurons in the rNA. These data suggest that, as further comparative data become available, it may be possible to identify the neurotransmitter content of a particular population of nerve terminals in the rNA on the basis of the relative size of their terminals.
In the present study, we used antibodies against one of the isoforms of synthetic enzyme for GABA, i.e., glutamate decarboxylase isoform 67 (GAD67). A number of studies indicate that, in adult mammalian brain, GABA is synthesized mainly from decarboxylation of glutamic acid by two specialized enzymes (50), the glutamic acid decarboxylases (GADs). They are distinguished as GAD65 and GAD67 according to their molecular masses (65 and 67 kDa). These enzymes, which exhibit different affinities for their cofactor pyridoxal-5′-phosphate, are encoded by two different genes located on separate chromosomes. Both enzymes are often colocalized in the same GABAergic neurons but sometimes differ in their subcellular distribution (8, 15, 31, 34). GAD67, although also detected in axon terminals, shows a predilection for cell bodies whereas GAD65 tends to be associated with synaptic vesicles in nerve terminals (34). The GAD67 mRNA is more abundant than the GAD65 mRNA in many regions of rat brain (15). More detailed studies revealed, however, a more complex situation in which some neurons appeared to express more GAD65 mRNA (e.g., in some hypothalamic nuclei), some express more GAD67 mRNA (e.g., in cerebellar cortex), and some expressed similar amounts of the messages (e.g., in the reticular nucleus of the thalamus) (15-17, 31, 53). Recent results show that GAD is significantly degraded by proteolysis within hours after death and that GAD67 is degraded more rapidly than is GAD65 (41). Our ferrets were perfused with fixatives while ventilated with oxygen. Therefore, proteolytic processes were rapidly terminated, preserving the antigenicity of GAD.
The present study demonstrates that GAD67-containing neurons innervate AVPNs that may play a substantial role in GABAergic control of AVPNs. However, the data do not address the question of whether GAD65 immunoreactive terminals also synapse on AVPNs in the ferret. As in other neuronal networks, the separate functional role of the two isoforms is unknown (58), but the present data and other observations indicate that both GAD65 and GAD67 may provide important reserve pools of GABA for regulation of inhibitory neurotransmission. This assumption is based on studies showing that a deficiency of either isoform of GAD can have significant physiological sequelae. For example, GAD67-defi-cient animals are born with a cleft palate and die within the first day of life, apparently from respiratory failure (3). By contrast, GAD65-deficient mice appear normal at birth, but the pyridoxal-5′-phosphate-inducible apoenzyme reservoir is significantly decreased, leading to increased susceptibility to epileptigenic stimuli (2). Hence, both isoforms of GAD appear to be physiologically important in the dynamic regulation of neural network excitability.
Physiological results of the present study indicate that the activity of AVPNs is under tonic GABAergic inhibition. Experimental reductions in the strength of this influence, using receptor antagonists administered into the rNA, produced an increase in the frequency of discharge of putative AVPNs associated with an increase in cholinergic outflow to the airways and an elevation of airway smooth muscle tone.
In the present study, single-unit recording experiments were designed to investigate whether putative AVPNs are under tonic GABAergic inhibitory influences. We employed orthodromic stimulation, as previously used to define gastric-related vagal motoneurons in the dorsal motor nucleus (45). We investigated single-unit responses to stimulation of bronchopulmonary sensory receptors, known to increase cholinergic outflow to the airways (10). These stimuli preferentially activate AVPNs. For example, stimulation of pulmonary C-fiber receptors excites AVPNs and consequently increases cholinergic outflow to the airways but inhibits discharge of inspiratory-related medullary neurons or sympathetic premotor neurons (10). Conceivably, our results do not obviate an effect of stimulation of bronchopulmonary sensory fibers on other, nonairway-related neurons in the rNA.
An indisputable way to identify vagal preganglionic cells controlling airways is to perform antidromic stimulation of peripheral branches of the vagus nerve that innervate the airways and to verify the projection by using the collision test (44). Although simple in concept, reliable antidromic identifi-cation of vagal preganglionic neurons poses numerous experimental challenges. This is because, in ferrets, AVPNs have slowly conducting, nonmyelinated axons (47). Furthermore, many airway-related vagal preganglionic neurons have a very low rate of spontaneous activity at low levels of respiratory drive. These neurons have a low axonal conduction velocity and require high-intensity current for antidromic activation.
The vagus nerve contains reciprocal pathways: afferent and efferent fibers. Hence, electrical stimulation of the vagus will activate both sensory (afferent) fibers and motor (efferent) axons (50). In three additional experiments, either electrical stimulation of the cut central end of the recurrent laryngeal nerve or lung deflection caused an increase in frequency of discharge in AVPNs. However, we were not able to record single antidromic responses, after recurrent laryngeal nerve stimulation probably due to collision with an increase in synaptic drive. The lack of initiation of a single antidromic spike by high-frequency stimulation of the vagus nerve or its branches does not preclude the unit from involvement in airway control, because activation of these neurons by excitatory amino acids, or by blockade of GABAA receptors, leads to an increase in cholinergic outflow to the airways.
The activity of AVPNs within the rNA that form an identifiable cell group is regulated through multiple distinct neuronal pathways, including a number of brain stem cell groups that project to AVPNs (21, 42, 49, 51). These afferents contain a wide variety of neurotransmitters or neuromodulators, including GABA, norepinephrine, and serotonin (27, 30, 25, 26).
In general, the effects of GABA are mediated by ionotropic GABAA and GABAC receptors and metabotropic GABAB receptors (4, 11, 14, 40). Recent studies reveal that the accumulation of ionotropic GABAA receptors at synapses is a highly regulated process facilitated by receptor-associated proteins and other cell-signaling molecules (36). This receptor subtype gates anionic Cl--permeable channels that are selectively blocked by the alkaloid Bic and are modulated by steroids, barbiturates, and benzodiazepines (4, 29, 40). To date, 18 GABAA receptor cDNAs have been cloned. In the adult brain, the most common ionotropic GABAA receptor is composed of α-, β-, and γ-subunits (4). Results of recent studies showed that AVPNs most abundantly express the β-subunit of a GABAA receptor (30). Expression of other subunits has not been examined.
In the present study, we used Bic and Hyd as GABAA receptor blockers. Under in vitro conditions, a number of observations cast doubt on the specificity of N-methyl derivatives of Bic on GABA synapses, particularly for in vitro studies at high concentrations (12). However, under in vivo conditions, Bic does not block SK channels, even at relatively high concentrations (62). Furthermore, previously we showed that prior topical application of Bic, in concentrations substantially lower than those needed to affect potassium conductances under in vitro conditions, significantly blocked the inhibitory effects of GABA or benzodiazepines on AVPNs and, consequently, on cholinergic outflow to the airways (24, 29). In the present study, to circumvent possible nonspecific effects of Bic in determining the extent of GABAA receptor involvement in the regulation of cholinergic activity, we also used Hyd, a potent and more selective competitive GABAA receptor antagonist (62), and similar effects were observed as with Bic. Thus our results indicate the presence of tonically active GABAA receptors and their role in the control of bronchomotor tone but do not exclude a potential role that GABAC, or GABAB receptors might play in the regulation of excitability of AVPNs.
The results of the present study support the hypothesis that a tonically active central GABAergic inhibitory microcircuit, which utilizes both axosomatic and axodendritic synapses on AVPNs, modulates cholinergic drive to the tracheobronchial system through GABAA receptors. However, our data do not exclude the possibility that, at some sites, GABA can activate extrasynaptic GABA receptors on AVPNs through volume (nonsynaptic) transmission (1, 55). Downregulation of GABAergic effects may result in a shift from inhibitory to excitatory influences, leading to a hyperexcitable state of the AVPNs and to airway hyperreactivity.
Physiological Relevance of GABAergic Innervation of the AVPNs
Previously, using conventional and transneural labeling techniques and ultrastructural, molecular, and physiological approaches, we have identified two major inhibitory cell groups that project to the AVPNs: i.e., brain stem norepinephrine- and serotonin-containing neurons (21, 25-27). Now, we have described a third system, a GABAergic inhibitory microcircuit that controls cholinergic outflow to the airways. In this study, we have observed a tonic GABAA-mediated inhibition of AVPNs.
Both axosomatic and axodendritic GABAergic innervation may exert phasic and tonic influences on the excitability of principal neurons. For example, phasic and tonic GABAergic signaling regulates activity of cerebellar granule cells (7) and GABAergic tonic signaling modulates the firing rate of hippocampal interneurons (57). Similarly, the activity and the discharge-frequency patterns of medullary respiratory premotoneurons are subject to potent tonic GABAergic gain modulation (9, 13, 46, 61, 62), and it seems that some GABAergic inputs may be functionally isolated from the soma/spike initiation zone, e.g., located on a dendritic shaft (61). Ultrastructural findings of the present study indicate that central GABAergic modulation of AVPNs is mediated in large part via classically defined inhibitory axosomatic synapses (20) and to a lesser degree through axodendritic synaptic transmission. The axodendritic synaptic GABAergic modulation of AVPNs is structurally distant from the soma/spike initiation zone (37, 59) and hence may exert less of an impact on the resting membrane potential of these neurons than axosomatic synapses close to the axon hillock. However, some of the differences in the apparent proportions of axodendritic vs. axosomatic GAD-IR synapses on AVPNs may also reflect the fact that very distal parts of dendrites of AVPNs are not as easily retrogradely labeled with CTB-HRP as their more proximal portions and their perikarya.
The functional relevance of the GABAergic microcircuit that has been presently demonstrated in the central regulation of cholinergic outflow to the airways is as yet not well understood. On the basis of these findings, we expect that abnormalities of GABAergic function caused by either genetic or acquired alterations of GAD activity, changes in binding to GABAA receptor sites, or ion channelopathies may result in a shift from inhibitory to excitatory neurotransmission on AVPNs. These changes may lead to a hyperexcitable state of AVPNs and to airway hyperreactivity. Hence, enhancement of GABA-induced neuronal inhibition may be useful in the treatment of disorders associated with an increased cholinergic outflow to the airways (i.e., obstructive bronchitis and bronchial asthma).
The authors thank Lee A. Watson for secretarial support.
This research was supported by grants from the National Institutes of Health (HL-50527 and 1U54NS39407 to M. A. Haxhiu and HL-51917 to V. J. Massari).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 the American Physiological Society